Social Work Research

Bright, Charlotte, Lyn

doi: 10.1093/swr/svz021pmid: N/A

As I reflect on the three-month period since I last wrote an editorial, I am shocked by what the United States and the rest of the world have experienced. Mass shootings have claimed dozens of lives. A hurricane has decimated parts of the Bahamas, with the human and environmental toll still to be determined as I write these words. Catastrophic forest fires blaze in South America. Many factors contribute to disasters large and small. Climate change, global demand for products, and policies enacted at all levels can cause irreparable harm. Information flow has changed drastically in recent years, with opinions or falsehoods masquerading as facts and spreading faster than they can be checked and countered. Scapegoating, victim blaming, and “othering” are dispiriting dynamics of modern discourse (although perhaps this, at least, is nothing new). Although some of the causal factors perpetuating violence and disaster are unpredictable or outside anyone’s control, others can be identified and addressed. What is a social work scholar to do? Contributors to and readers of this journal are skilled at building and interpreting research evidence across a broad array of social problems, in areas ranging from intrapersonal development to international communication. Although we have a great deal to learn about intervention and service delivery from both a micro and a macro perspective, social work researchers have built a broad evidence base on topics relevant to social work and allied disciplines. Social work scholarship has grown dramatically in the history of our discipline, with social workers’ observations giving way to rigorous research designs and increasingly sophisticated interpretation of findings. Where social work and allied professions continue to struggle is in connecting relevant research to real-world practice, including addressing implementation and feasibility but also knowledge gaps (Palinkas & Soydan, 2011). More research is needed to effectively address social ills and promote human capacity. The problems our societies face are too complex for one-size-fits-all solutions, especially when we consider the role of context and culture (Eiroa-Orosa, 2018). By all means, I exhort social work researchers to do what they do best: ask meaningful questions, apply appropriate and flexible methods to answer those questions, and critically evaluate the applications of research findings with consideration for both internal and external validity. When we consider the immediate needs of our local and global communities, however, we may be compelled to do more. I teach a capstone seminar course in our PhD program, focusing on integrating theory and methods to inform a research agenda. We spend a great deal of class time talking through dissertation topics, samples, and methods. This spring, I added “develop a dissemination plan” to the course assignments. Students are now responsible for not only thinking through their proposed aims, theoretical perspective, and methods, but also describing how to reach an audience that stands to benefit from their findings. Students devise communication plans relevant to communities, consumers, practitioners, and policymakers. Their ideas have included social media posts and traditional media publications, art installations, legislative testimony, and presentations at practice conferences. In short, students apply both communication and advocacy skills to the task of translating research findings into meaningful, real-world applications. In my view, this skill set is a natural fit for social workers—who have explicit training in all these elements—but is underused with social work research. To move beyond simple consideration of our impact factor and promote actual impact in communities, we can rely on what we have learned throughout our training and experience. For those in practice, this may look like applying innovative research findings to various settings to assess feasibility and replicability. For those whose primary role is scholarship, impact may mean developing research partnerships with communities and organizations to meet the most pressing policy and practice needs of these groups, and to promote research-informed practice. In classroom teaching, we can facilitate discussions about how research connects to policy, practice, and advocacy. In administration, we can identify opportunities to bring together various constituencies and communities to promote uptake of research findings. All of us can bring our abilities to bear on getting useful information to people in a position to make use of it for the purpose of resolving social problems and promoting social and environmental justice. It is disheartening, to say the least, to see prominent examples of hatred and division in our world. Vitriol, fear, and disinformation abound. I encourage the Social Work Research community of scholars and readers to consider ways we can counteract damaging messages about the causes and consequences of violence and disaster. You are a wise, informed readership with an impressive capacity for building and disseminating knowledge. How can you use your skills to investigate and inform others about best practices in addressing human needs? The articles featured in this issue provide helpful food for thought. As you read cutting-edge research on such topics as intimate partner violence, juvenile justice, and neighborhood-level intervention, I hope you will consider how the authors’ findings contribute to the knowledge base on social work practice. Who needs to know this information, and how can we connect these research findings to the right audiences? As members of a profession based on a shared set of values, we are called to action when we see human and environmental suffering. As social work scholars, we routinely demonstrate the capacity for using research as a tool to improve policy and practice. The question is how to best use our knowledge alongside communication and advocacy skills to broaden our reach. In the (mis) information age, it is essential that our impact extend beyond academic audiences to those who can most benefit from and apply findings in the realms of policy and practice. When we convene for meetings and connect online, I look forward to engaging in rich discussions with many of you about how best to accomplish the mission of the journal and of the social work profession, harnessing the power of research to “enhance human well-being and help meet the basic human needs of all people, with particular attention to the needs and empowerment of people who are vulnerable, oppressed, and living in poverty” (National Association of Social Workers, 2017, p. 1). Charlotte Lyn Bright, PhD, MSW, is associate professor and associate dean for doctoral and postdoctoral education, School of Social Work, University of Maryland, 525 W. Redwood Street, Baltimore, MD 21201; e-mail: [email protected]. References Eiroa-Orosa , F. J. ( 2018 ). The sociocultural context of psychosocial interventions . Frontiers in Psychology, 9 , 1795 . Google Scholar Crossref Search ADS WorldCat National Association of Social Workers. ( 2017 ). Code of ethics of the National Association of Social Workers . Washington, DC : Author . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Palinkas , L. A. , & Soydan , H. ( 2011 ). Translation and implementation of evidence-based practice . New York : Oxford University Press . Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC © 2019 National Association of Social Workers This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Wachter,, Karin;Dalpe,, Jessica;Heffron, Laurie, Cook

doi: 10.1093/swr/svz008pmid: N/A

Abstract Despite robust evidence of the myriad consequences associated with intimate partner violence (IPV), social services in the United States may not adequately account for and respond to variations in how women resettling as refugees conceptualize service and support needs. With this study, the authors sought to develop a more nuanced understanding of needs as expressed by women resettling to the United States as refugees. Researchers conducted in-depth interviews and focus groups with refugee women who resettled to the United States (n = 35) and social services providers (n = 53), including those working in refugee resettlement or domestic violence agencies or community-based organizations. The research team used structural coding and thematic analysis to examine the data, through which they identified four themes: (1) living with enduring consequences of IPV, (2) “I just want them to help me,” (3) “I need you to talk to my husband,” and (4) “How will I pay the rent?” The study findings point to broader structural concerns shaping women’s resettlement experiences, as well as areas of incongruence between women’s conceptualizations of needs and established practice approaches. Implications for culturally competent and survivor-centered practice are discussed. In 2017, an estimated 65.5 million individuals were registered as forced migrants due to armed conflict, violence, or persecution (United Nations High Commissioner for Refugees [UNHCR], 2018). Approximately 50 percent of those registered as refugees worldwide are women and girls (UNHCR, 2018). Since enacting the Refugee Act in 1980, the United States has resettled 3 million people with refugee status at the recommendation of UNHCR (Connor & Krogstad, 2018). Research points to the scope, scale, and devastating consequences of violence against women in war and forced displacement (Abrahams et al., 2014; Stark & Ager, 2011; Vu et al., 2013). Violence against women, perpetrated by known and unknown people, escalates during armed conflict and in forced migration (Usta, Farver, & Zein, 2008). The instability, losses, and changes associated with armed conflict and forced migration can reinforce existing gender inequalities and foster conditions in which men use violence against female partners with relative impunity. Research indicates that immigrant and refugee communities in the United States face specific risks related to intimate partner violence (IPV) or domestic violence (DV) (Runner, Yoshihama, & Novick, 2009). Disparities related to language proficiency, economic and social resources, isolation, gender, employment, and legal status exacerbate risks and complicate access to services and support among immigrants and refugees in abusive relationships (Erez, Adelman, & Gregory, 2009; Lee & Hadeed, 2009; Menjívar & Salcido, 2002; Rodríguez, Valentine, Son, & Muhammad, 2009; Vidales, 2010). Experiences of IPV intersect with migration experiences and necessitate survivor-centered services that account for a host of needs (Rees & Pease, 2007); however, social services in the United States may not be set up to detect and respond to them. This article offers nuanced understandings of how women resettling to the United States as refugees conceptualize and express needs related to IPV and considers the implications for practice. Literature Review Effects of IPV An extensive literature details the myriad short- and long-term consequences of IPV and associated needs. Related physical and emotional traumas are strongly associated with poor physical, reproductive, and mental health outcomes throughout women’s lives (Campbell, 2002; Dutton et al., 2006; Ellsberg, Jansen, Heise, Watts, & Garcia-Moreno, 2008). Physical and sexual assault may result in chronic pain, severe injuries, brain trauma, gynecological complications, unwanted pregnancy, sexually transmitted infections, and death (Black, 2011; Breiding, Black, & Ryan, 2008). Women who have experienced IPV are more likely to have symptoms associated with depression, anxiety, and posttraumatic stress disorder (PTSD) and are more likely to develop high-risk behaviors, such as substance abuse and self-harm (Zinzow et al., 2010). The associated economic losses also increase women and children’s vulnerability over time. The increased annual health care costs for survivors of IPV can continue for years, even after the abuse has ended (Rivara et al., 2007). DV and Resettlement Social Services DV services in the United States originated from the grassroots battered women’s movement and was led by community members, many of whom were survivors themselves (Arnold & Ake, 2013; Davies, 2008; Mehrotra, Kimball, & Wahab, 2016). With the criminalization of IPV and the involvement of the state, DV services have evolved to be highly professionalized services (Goodman & Epstein, 2008; Mehrotra et al., 2016). DV services include shelter, advocacy, psychosocial support, legal representation and advice, and medical care. Social services agencies holding a cooperative agreement with the U.S. Department of State sponsor individuals and families approved for resettlement in collaboration with a nationwide network of affiliated offices, state governments, nonprofits, faith-based organizations, and community partners. Federal policies prioritize employment services, with the aim of ensuring that clients become economically self-sufficient within 120 to 180 days (Office of Refugee Resettlement, 2014). Case management forms the foundation of service provision for individuals and families resettling as refugees. However, time and mounting resource limitations exacerbated by anti-immigrant rhetoric and policies limit agencies from delivering more holistic services (Wachter & Snyder, 2018). Practice Approaches: Conceptual Considerations Research examining conceptualizations of needs defined by diverse clients requires an analysis of key practice approaches, especially cultural competence and survivor centeredness. What constitutes culturally competent practice is dynamic and contested (Bhuyan, Park, & Rundle, 2012; Dominelli, 2004; Furlong & Wight, 2011; Johnson & Munch, 2009). Cultural competence has often focused on the knowledge and skills social service practitioners require to provide culturally appropriate services (Danso, 2018). However, a growing understanding of entrenched structural, systemic, and interpersonal dynamics that foster disparities points to the conceptual evolution under way within the field of social work (Bhuyan et al., 2012). A critical understanding of culture foregrounds the intersections and reciprocity between identities and structural oppression and encourages resistance of essentialist, unidimensional, and fixed notions that reinforce human constructs of difference (Warrier, 2009). A critical approach to cultural competency focuses on lifelong processes that involve critical analysis, dialogue, self-reflection, and interrogation (Furlong & Wight, 2011; Johnson & Munch, 2009; Warrier, 2009). The survivor-centered approach is rooted in feminist theory, empowerment models, and a rights-based approach (Davies, 2008; Goodman & Epstein, 2008). The core of survivor centeredness is the act of giving voice and empowering women to make decisions about their lives and families (Davies, 2008; Goodman & Epstein, 2008). Although mainstream DV services originated from this model, they often are most effective for survivors for whom the anti-violence movement in the United States originated: white, middle-class, heterosexual women (Goodman & Epstein, 2008; Goodman et al., 2016). Recognizing that intersecting identities and social positions shape experiences, survivor-centered approaches respond to women in their entirety and amplify their voices to drive the terms of services, goals, and outcomes for themselves and their families (Davies, 2008; Goodman & Epstein, 2008; Goodman et al., 2016; Mills, 1999). Current Analysis Although a substantial literature provides robust evidence of myriad consequences associated with IPV, social services and practice approaches in the United States may not adequately account for or respond to variations in how women resettling as refugees conceptualize service and support needs. The aim of the current analysis was to examine conceptualizations of needs as expressed by women who resettled to the United States. The following research question guided the analysis: What services and support do women who have experienced IPV need and want, prior to and after resettling to the United States, and how do these findings inform approaches such as culturally competent and survivor-centered practices? Method This analysis drew from a broader project that studied help seeking related to experiences of IPV among women who resettled to the United States as refugees. In collaboration with an external researcher, a resettlement agency—the International Rescue Committee—implemented the study from 2016 to 2018 in a large urban metropolitan area in the southern region of the United States. Recruitment Agency staff recruited a purposive sample of participants categorized into three groups. Group A comprised women who, during routine screening or to agency staff, disclosed having experienced IPV, sexual assault, or both. Group B included women who had not disclosed during routine screening or to staff that they had experienced IPV or sexual assault. Group C comprised representatives from organizations serving refugees, immigrants, or survivors of IPV and other stakeholders at the local or state level. All group A and group B participants were agency clients who had come to the United States through the federal refugee resettlement program. Recruitment procedures for groups A and B included informing each woman that she would receive transportation to and from the meeting location and a VISA gift card. Agency staff did not recruit women known to be currently at threat of IPV, and they asked all women recruited as group A participants about any risks they might currently face. The team recruited group C participants from a wide range of service providers and community-based organizations in the metropolitan area where the study took place. All participants were over the age of 18 years. Participants Table 1 provides details pertaining to participant demographics (N = 88). The majority of group A and group B participants originated from the Central and East Africa region, and the remainder came from South and Southeast Asia and the Middle East. Factors precipitating migration among group A and group B participants, their refugee status, and their ultimate resettlement to the United States varied. The majority of women migrated because of war, and a small subset fled the threat of religious persecution. Table 1 Study Participants’ Demographics (N = 88) . Group A . Group B . Group C . Characteristic . (n = 10) . (n = 25) . (n = 53) . Region of origin South and Southeast Asia 0 8 4 Central and East Africa 8 17 10 West and North Africa 0 0 5 Eastern Europe 0 0 2 Middle East 2 0 1 Central America and Caribbean 0 0 4 North America 0 0 27 Gender Female 10 25 43 Male 10 Age (years) 18–29 2 8 16 30–39 3 7 24 ≥40 5 10 13 Years of formal education 0–5 5 15 0 6–12 5 9 7 4 years postsecondary (attained undergraduate degree) 0 1 27 ≥4 years postsecondary (attained graduate degree) 0 0 19 Marital status Married 2 15 Divorced/separated 3 1 Widowed 1 4 Single 2 5 Number of children 0 1 4 1–3 4 9 4–6 1 9 ≥7 3 3 Time in United Statesa <12 months 4 5 12–24 months 6 13 >2 years 0 4 Employment status Employed part- or full-time 6 7 53 Unemployed 4 18 0 . Group A . Group B . Group C . Characteristic . (n = 10) . (n = 25) . (n = 53) . Region of origin South and Southeast Asia 0 8 4 Central and East Africa 8 17 10 West and North Africa 0 0 5 Eastern Europe 0 0 2 Middle East 2 0 1 Central America and Caribbean 0 0 4 North America 0 0 27 Gender Female 10 25 43 Male 10 Age (years) 18–29 2 8 16 30–39 3 7 24 ≥40 5 10 13 Years of formal education 0–5 5 15 0 6–12 5 9 7 4 years postsecondary (attained undergraduate degree) 0 1 27 ≥4 years postsecondary (attained graduate degree) 0 0 19 Marital status Married 2 15 Divorced/separated 3 1 Widowed 1 4 Single 2 5 Number of children 0 1 4 1–3 4 9 4–6 1 9 ≥7 3 3 Time in United Statesa <12 months 4 5 12–24 months 6 13 >2 years 0 4 Employment status Employed part- or full-time 6 7 53 Unemployed 4 18 0 aMissing data for Group B (n = 22). Open in new tab Table 1 Study Participants’ Demographics (N = 88) . Group A . Group B . Group C . Characteristic . (n = 10) . (n = 25) . (n = 53) . Region of origin South and Southeast Asia 0 8 4 Central and East Africa 8 17 10 West and North Africa 0 0 5 Eastern Europe 0 0 2 Middle East 2 0 1 Central America and Caribbean 0 0 4 North America 0 0 27 Gender Female 10 25 43 Male 10 Age (years) 18–29 2 8 16 30–39 3 7 24 ≥40 5 10 13 Years of formal education 0–5 5 15 0 6–12 5 9 7 4 years postsecondary (attained undergraduate degree) 0 1 27 ≥4 years postsecondary (attained graduate degree) 0 0 19 Marital status Married 2 15 Divorced/separated 3 1 Widowed 1 4 Single 2 5 Number of children 0 1 4 1–3 4 9 4–6 1 9 ≥7 3 3 Time in United Statesa <12 months 4 5 12–24 months 6 13 >2 years 0 4 Employment status Employed part- or full-time 6 7 53 Unemployed 4 18 0 . Group A . Group B . Group C . Characteristic . (n = 10) . (n = 25) . (n = 53) . Region of origin South and Southeast Asia 0 8 4 Central and East Africa 8 17 10 West and North Africa 0 0 5 Eastern Europe 0 0 2 Middle East 2 0 1 Central America and Caribbean 0 0 4 North America 0 0 27 Gender Female 10 25 43 Male 10 Age (years) 18–29 2 8 16 30–39 3 7 24 ≥40 5 10 13 Years of formal education 0–5 5 15 0 6–12 5 9 7 4 years postsecondary (attained undergraduate degree) 0 1 27 ≥4 years postsecondary (attained graduate degree) 0 0 19 Marital status Married 2 15 Divorced/separated 3 1 Widowed 1 4 Single 2 5 Number of children 0 1 4 1–3 4 9 4–6 1 9 ≥7 3 3 Time in United Statesa <12 months 4 5 12–24 months 6 13 >2 years 0 4 Employment status Employed part- or full-time 6 7 53 Unemployed 4 18 0 aMissing data for Group B (n = 22). Open in new tab The age range among group A participants (n = 10) was between 20 and 53 years. Women described experiencing IPV prior to fleeing their countries of origin, during displacement in refugee camps, and in the United States. Participants disclosed forms of IPV that involved physical and sexual violence, intimidation and threats, controlling and degrading behaviors, emotional abuse, and economic abuse. With the exception of two participants, group A women described experiencing multiple and severe forms of IPV over time (estimated range: one through 30 years). Approximately one-third of group A participants had resettled with an abusive partner and had continued to experience one or more forms of IPV in the United States. Group B participants (n = 25) ranged in age from 18 to 69 years. More than half of group B participants had screened negative for IPV, and 11 participants had not participated in screening. The majority of group B participants were unemployed at the time of data collection and staying home to care for young children. Group C (n = 53) included 29 participants who currently worked in the field of refugee resettlement and 13 participants who were employed by a DV agency or community-based organization. The remaining 11 participants worked in other social service, health, legal, and political capacities that involved serving refugee or immigrant groups. The years of experience in their current field of work ranged from one to 24 years. Data Collection Procedures With the assistance of language interpreters as needed, two members of the research team conducted 10 individual interviews with group A participants, eight focus group discussions with group B participants, and five individual interviews and 16 group discussions with group C participants. Four group B focus groups involved four participants, one group involved three, and three groups involved two. One to four people participated in group C discussions. Group A interviews were 84 minutes long on average (range: 56–113 minutes), group B focus groups also averaged 84 minutes in length (range: 69–103 minutes), group C interviews averaged 68 minutes (range: 45–78 minutes), and group C discussions averaged 72 minutes (range: 42–109 minutes). Meetings with group A and group B participants took place in a private room at either the agency or a local church. Group C meetings took place at the agency or at participants’ workplaces. Interviews and group discussions began with the interviewer gathering nonidentifying demographic information, which varied by group. A semistructured interview guide examined various aspects of participants’ personal and professional experiences, based on the recruitment group. Individual interviews with group A women who had experienced IPV explored participants’ backgrounds, help seeking, and factors informing decision making in reference to seeking support. In focus group discussions with group B participants, the group explored how women’s lives have changed after resettlement, the challenges the women faced, and questions around IPV and help seeking. These focus groups also included a functional mapping activity to facilitate discussion on options for women experiencing IPV. In interviews and group discussions, group C participants were asked to describe the services and programs their organizations provide, how they relate to women’s needs pertaining to IPV, their understanding of support needs among refugee and immigrant groups, and how organizations can help to meet those needs. Audio recordings of all interviews were professionally transcribed and reviewed for accuracy. Protection of Human Subjects The University of Texas at Austin’s institutional review board reviewed and approved the study. The agency followed approved guidelines to recruit and protect the identity of research participants. Researchers provided participants with detailed information about the study, who in turn gave verbal consent to participate and allow researchers to audio record the discussion. Participants could stop the interview at any point or refuse to answer any question. All group A and group B participants received a $25 gift card and two round-trip tickets for the local public transportation system. All interpreters assisting with data collection signed confidentiality agreements. Data Analysis The lead researcher used structural coding to assess progress toward answering the research questions and guide ongoing data collection at the midway point of the project. When data collection was complete, the lead researcher applied the same structural codes to the remainder of the data to assist with its management. She then used a thematic approach to further examine the data and identify preliminary themes using codes developed a priori and inductively (Guest, MacQueen, & Namey, 2012). Reflection and discussion among three members of the research team allowed the analysis to be refocused from initial codes to second-cycle coding and to develop a broader level of abstraction (Saldaña, 2015). This process included the review and refinement of themes. The research team used a qualitative data analysis software program (NVivo, Version 11) to manage and code all data. The researchers followed procedures to establish rigor in qualitative research as explained by Creswell (2013). Members of the research team met throughout the analysis process to discuss and confirm research findings, and they documented all major activities and decisions in a detailed audit trail to establish transparency (Rodgers & Cowles, 1993). The team engaged in a reflexive process to reduce the possibility that their identities and preconceived ideas compromised the integrity of the analysis (Padgett, 2016). The study used the services of professionally trained language interpreters who were part-time employees of the resettlement agency, with one exception (a member of the research team who was a full-time employee of the agency assisted with interpretation for one group A interview to accommodate the participant’s schedule). All interpreters identified as female and originated from the same geographic regions as the participants. The interpreters participated in three hours of training related to the goals of qualitative research, the specific aims of the project, researcher and interpreter roles, the interview guide, data collection procedures, and expectations regarding confidentiality. Findings This section presents an analysis of four interrelated themes that provide insights into the support and service needs of women resettling in the United States. Names that appear are pseudonyms created for group A participants. Living with Enduring Consequences of IPV Participants provided insights into services and support needs primarily by describing the consequences of IPV on women’s health, mental health, and overall well-being. Women described difficulties managing stress, worry, depression, forgetfulness, insomnia, nightmares, heart palpitations, loss of appetite and weight, and intrusive thoughts. At least three women shared that they were living with acute physical health problems because of the violence they had endured. Women described suffering the consequences of the violence they had experienced in both past and present terms. As Evette explained, I don’t sleep since my first husband passed away and I married this one. All the stress that I’m having, I don’t get enough sleep. I felt like I wanted to take my life away [in the past]. That’s why it brought me all the disease I’m feeling now. Some symptoms appeared to have lessened with time, and others continued into the present day. Women explained how cyclical physical and emotional violence affected their minds and bodies, and some described the dramatic improvements they experienced since separating from their abusive spouse. Jeanne shared, Now when people see me, they say, “Eh, you are in good health [now]” . . . because I was very skinny. People thought that I was very old. . . . It was stress. When I see the way he was treating me, talking to me, I would sit and think a lot. I wouldn’t eat. So that’s how I lost my appetite and my body. Women who had separated from their abusive husbands also described a sense of relief, a feeling of peace, and the ability to sleep again. For some, however, the impact of violence on their physical and psychological well-being persisted over time and carried over into their resettlement experiences in the United States, even those now living on different continents from their ex-partners. Women described continuing to struggle with nightmares or keeping troubling thoughts and invasive memories of the violence at bay and having a fierce desire not to recall what they had endured. Service providers working with refugee and immigrant clients shared similar observations. One such provider shared, I’ve just heard some really horrific stories from survivors, women, who left—escaped, really—their husbands, and came to the U.S. PTSD is the focus of our work. Working through PTSD symptoms, so they can regain a sense of security, and regular daily functioning, not live in fear. I mean, it’s amazing that they know that there’s maybe an ocean between them and the abuser, but they’ll still wake up and fear that he’s at their door. Indeed, women described fear and anxiety as salient ramifications of the violence they had experienced. The sources, manifestations, and meanings of these fears varied among participants. One participant, for example, described palpable fear that she would one day have to go back to her husband’s bed where she had endured terrible abuse for years, even though the chances of this actualizing were low by her own account. Another woman struggled with the complexities of still being young, family and societal pressures to remarry, her own desire to remarry one day, and the intense distress caused by the idea of marrying again. Other fears women expressed related to their financial stability and ability to pay household bills. Finally, feelings of worthlessness and failure as a woman and wife expressed by some participants were a window into the pain and suffering women may carry within as part of the resettlement process. “I Just Want Them to Help Me” Although women described the consequences of the violence they had suffered in detail, they did not readily pinpoint specific support or service needs related to those consequences and expressed needs in broad terms, if at all. Mathilde stated, “I just want them to help me. Who can refuse to be helped?” Although Mathilde had received temporary shelter services since arriving in the United States, she did not specify any particular support or services she had desired in the past or needed now. She went on to express that those in a position to help her would know what assistance to provide, not her. Mathilde’s response and broad request for help reflected sentiments expressed by other participants. The need for people to turn to for advice and solace was evident for women in abusive relationships. Viviane explained the impact of her abusive partner preventing her from socializing with others in the refugee camp: “If you have problems, you need to go talk to people, so they give you advice so you feel better. But I didn’t have that.” Women indicated that they needed professional DV services and described “women’s rights people who actually get things done” and “groups of people who are really experts, who want to do this job, and do it like a job . . . going deep to solve problems.” Service providers shared examples of women coming to them to help ensure their safety and to seek advice, and they spoke in depth of the financial needs clients presented to cover housing and other household expenses. Otherwise, service providers shared similar reflections of their clientele. As one provider described, Most of the time, [our clients] don’t know what they want. A woman will call me and tell me everything. I’m like, “So what do you want to do?” And she’s like, “I don’t know what to do.” So you just give them options and what the options are, but most of the times they really—I mean, in fact, they can’t decide. This apparent indecision may have reflected that women chose to disclose their abuse to a person they trusted would know what the best course of action may be for them, or it may have been the result of feelings of powerlessness experienced when they were presented with options and expected to act autonomously in an entirely unfamiliar system. “I Need You to Talk to My Husband” Across groups, participants described women’s desire for a trusted family member, community leader, or service provider to intervene directly with the abusive partner. Women described seeking intervention from family members, community leaders, or an authority figure associated with an institution as a way to pursue the possibility of an abusive spouse changing his behaviors and to sanction their leaving their spouse if he did not. The process of women seeking intervention may thus represent both an effort to stay in the marriage as well as an attempt to build buy-in and support for their eventual separation should those efforts fail. Indeed, some women framed the ultimate decision to leave their abusive spouse as having been made by family or an outside authority after their help and involvement had been solicited. Service providers also observed how female clients experiencing IPV carefully considered whom in their network to approach for help. A provider reflected, “Involving the family or a community member seems to be something that the clients themselves have asked for or thought of as solutions.” Service providers indicated that when women disclosed IPV, they were less inclined to accept a referral and more interested in having that particular staff person talk with their husband. A provider associated with the resettlement service sector explained, “Maybe she needs you to call the husband, sit with them, advise them, and see who is at fault. . . . Maybe to see if the problem can be dealt with. That’s the culture. That’s why they report.” In a different group of resettlement providers, participants reported that women asked them to facilitate a conversation with their husband as a way of placing pressure on the spouse to change a problematic, abusive behavior. As one participant described, A lot of times, it starts with maybe controlling money and they [women] want help regaining control, or they ask for someone to talk to their partner about them having more financial equity, or freedom, or even any access to money . . . there’s a lot of threatening to leave. Another service provider explained how (presumably male) members of a refugee community perceived the resettlement agency’s response to IPV and their desire to be involved. [They say to me,] “You’re just making us divorce with our women. Why don’t you call us, sit with us, ask us what’s going on? Then we can explain. Maybe you can advise us. Maybe you get a suggestion rather than referring. Because other people don’t [know] our culture . . . So why do you send [refer] to other people who don’t know our culture? They will not solve our problem. They will add more problems on our hands.” Providers perceived requests from female clients to intervene directly with their male partners as a reflection of their desire to stay in the relationship and keep the family intact. One provider indicated, “Some women want somebody to tell the abuser to stop doing what he’s doing. They don’t want them to leave. They want their abuser to stop.” Participants explained that many survivors want to stay in the marriage out of concern for how the community will treat them if they were to divorce, as well as because of economic constraints. Others explained that clients sought information, emotional support, and someone who will listen, but they reiterated their perception that many women do not want to leave the marriage. A DV services provider serving immigrants explained, We have a lot of clients who actually stay. . . . Maybe there was a violent incident and they’re fed up, but they still may stay in the relationship. And so it doesn’t mean that everybody is always leaving. So sometimes, we’re still providing support while they’re still in that relationship. Despite the frequent requests to intervene directly with the abusive spouse or to mediate a conversation between partners, providers described what they felt was an inherent futility of such an intervention and indicated that they could not realistically agree to respond as requested. As one provider expressed, “Like he’s going to listen to me? Like I’m magic?” In the case of the refugee resettlement agencies, providers felt that serving both partners complicated their ability to intervene. Related to the expressed desire for direct intervention with husbands, there was also a resounding call across participant groups for longer-term educational initiatives that would explicitly target men and women. “How Will I Pay the Rent?” All participant groups highlighted financial assistance as a salient service need expressed by women after arriving in the United States, particularly for those experiencing IPV. The women’s comments made clear that the long-lasting consequences of IPV were interwoven with financial concerns, particularly in relation to housing. Women who separated from abusive partners after arriving in the United States struggled to become financially stable. Julie described her husband, who was physically and emotionally abusive, throwing her out of the apartment. She shared, “I worry because we came together as a family, and we were supposed to work together to give the children a better life. And now, I’m alone on my own, and sometimes I wonder if we’re going to make it.” Women described how the health and mental health consequences of the abuse they endured now intersected with financial concerns in the United States. Evette described her concerns about regaining both emotional and economic stability: “Even now, I don’t sleep well because I think about how other women are working and I’m not working. When am I going to feel well so that I can go to work like everybody else?” Providers reiterated women’s financial concerns in relation to the IPV the women were experiencing, as well as identifying them as an intrinsic component of their resettlement experience in the United States. One provider explained, “[Clients’] biggest problems are ‘how are we going to pay the rent and I can’t sleep at night because I keep having memories of the past.’ These are their biggest problems.” A DV services provider shared, “Number one, [clients say] ‘I need housing, housing, housing, housing.’ . . . There are also instances of ‘I don’t want to go into a shelter. I just need somebody to help—financial assistance for rent. For a deposit.’” Resettlement agency staff also expressed the need for an emergency fund for women experiencing IPV to allow them to address immediate needs in real time. As one such staff member explained, “We’ve had clients who are not able to go to work. She lost her job. Yes, the husband left the home, he was arrested and so forth. But now she has to be able to pay the rent.” Participants described the confluence of IPV (past and current) and resettlement and its compounding effect on their emotional well-being and economic vulnerability. Women who had resettled and were in abusive relationships faced the prospect of having to provide for themselves and multiple children in a foreign context in which they confronted new financial demands (rent, utilities, child care, and so on) and formal employment was the main option for generating income. Discussion The analysis highlights formidable IPV-related service and support needs of women who resettled to the United States. The findings reflect the breadth and depth of needs in line with the literature and thus serve as an important reminder of the vestiges of IPV women may carry across borders. Participants described women’s general needs and requests for someone to help them figure out what to do. Women regularly requested direct intervention with the abusive partner in an attempt to stop the abusive behaviors. Participants highlighted finances as a universal concern of women resettling to the United States, especially those in abusive relationships. The study findings point to broad structural forces shaping women’s experiences after they resettle in the United States, as well as areas of incongruence between women’s conceptualizations of needs and established practice approaches. These issues and their implications for practice and future research form the focus of the discussion here. Poverty, social marginalization, and disparities among refugees are factors that can complicate the experiences of women resettling to the United States. Poverty interferes with women’s ability to meet basic needs, forcing survivors to prioritize housing, food, and employment over safety and well-being (Postmus, Severson, Berry, & Yoo, 2009). The economic realities refugees face in the United States further exacerbate these tensions. Refugees resettling in the United States often arrive bereft of financial assets and with low levels of English proficiency, marketable skills, and formal employment experience, which contribute to the economic hardships they confront after arriving. As such, women and men resettling to the United States find themselves in low-paying jobs and frequently struggle to advance economically (Capps et al., 2015; Fix, Hooper, & Zong, 2017). A recent study revealed consistent pay disparities based on gender and region of origin among refugees in the United States, indicating that there is a need to ensure more equitable employment outcomes among those resettling (Minor & Cameo, 2018). Gender disparities related to education, employment, and mobility likely intensify women’s economic dependence on their male partners in resettlement. Furthermore, in contrast to how women may have generated income in the past by participating in informal markets, in the United States formal employment outside the home is the main avenue for income generation. Thus, some women may experience a loss of financial independence in resettlement. Past studies have highlighted the extent to which financial literacy, asset building, education assistance, and job readiness programming are an imperative for women living in poverty and experiencing IPV (Hahn & Postmus, 2014; Lindhorst, Meyers, & Casey, 2008). Effective policies and practices will prioritize financial assistance and emergency funds for survivors, with an emphasis on housing (Hahn & Postmus, 2014). The findings also bring into focus areas of incongruence between women’s conceptualizations of needs and mainstream practice approaches used by social services providers in the United States. As the findings illustrate, women may not explicitly express specific support needs beyond wanting help and may expect others to know how best to assist them. It is important to consider that some women may have never perceived having viable support or service options available to them and may never have had the opportunity to articulate needs for which they could receive support or services. Furthermore, varying degrees of trauma associated with forced migration, in addition to IPV, likely shape how women identify and express their needs. Temporal considerations further complicate expressions and detections of need. Women may continue to suffer the consequences of IPV they no longer experience, having separated from their abuser before resettling to the United States; may continue to experience ongoing IPV but under different conditions in resettlement; or may live with the threat of future violence, because they have resettled with an abusive partner. Needs expressed through this study may signify an appeal for providers to work with women resettling to the United States in ways that echo the roles families and communities may have played in the past and in different contexts (Wachter, Cook Heffron, & Dalpe, 2019). In the wake of social networks dismantled by war, forced migration, and resettlement, women may look to providers to engage with them as a trusted family or community member in ways that are familiar and of value to them, for example, by providing mediation, advice, or both. However, providers have traditionally hesitated to explore programming possibilities that challenge dominant service paradigms, possibly overlooking needs expressed by refugee clients. Indeed, women’s calls for providers to engage with abusive partners are in conflict with mainstream practices in the United States designed to prioritize responding to potentially lethal forms of abuse and individual survivors’ safety. Calls to broaden the spectrum of options available to women in abusive relationships highlight the extent to which women navigate multiple vulnerabilities—including IPV—and how manifestations of violence vary across abusive relationships (Davies, 2008; Goodman & Epstein, 2008). In other words, not every woman in an abusive relationship experiences physical violence or threats, and safety may not necessarily be women’s most pressing concern. The overriding focus of mainstream DV services, however, has been on safety, saving lives, and serving women who are ready to separate from their abusers (Kim, 2013; Mehrotra et al., 2016). Although safety and saving lives are critical, service options are nonetheless limited for women who do not prioritize safety or do not contemplate leaving their abusers. Furthermore, women who experience cumulative barriers to attaining help, as do women who resettle to the United States, will have inordinate difficulties accessing services that meet their perceived needs (Kennedy et al., 2012). The nuanced expressions of needs that emerged from this study reflect the complexity of women’s experiences at the intersection of forced migration and IPV. Yet they diverge from the construction of client and provider roles in the U.S. social services paradigm, in which clients express clearly defined needs that correspond to existing services, which providers deliver within time frames stipulated by state and federal governmental funding (Goodman et al., 2016). Rigid professional boundaries and power differentials between providers and clients further complicate the relationship between women’s needs and the services available to address those needs. Predetermined services drive assumptions of needs, obscuring the reality that any number of factors shape a woman’s need for services and that survivors’ experiences are subjective, varied, and context specific, necessitating nuanced assessments and responses rather than a standardized menu of services (Goodman & Epstein, 2008; Smyth, Goodman, & Glenn, 2006). Furthermore, the confluence of neoliberal policies with the criminalization of DV and professionalization of services has served to privilege efficiencies with specific consequences for socially and economically marginalized communities (Mehrotra et al., 2016). In a similar vein, neoliberal policies driving refugee resettlement programming and funding emphasize efficiencies and expediencies, possibly to the detriment of addressing more nuanced needs (Wachter & Gulbas, 2018). The analysis reveals implications for deepening the understanding of culturally competent and survivor-centered practice approaches underpinning programming responses, as they may fall short in acknowledging, understanding, and responding to the needs of women who resettle to the United States and likely the needs of other immigrant groups as well. Practices that predetermine women’s needs do not respond to varied experiences and perspectives through which women endure and heal from violence; as such, they are neither culturally responsive nor survivor centered. Furthermore, practices are not culturally responsive or survivor centered when practitioners feel inadequately equipped to recognize and respond to the nuances of experiences and needs and both parties lack reliable mechanisms for communication. Yet questions remain regarding how practitioners bring into balance culturally responsive and survivor-centered practices when clients’ requests are in tension with mainstream conceptualizations of professionalism, client and provider safety, and harm reduction. Efforts to carefully listen to and understand the needs women identify, taking past and present contexts into account, are important first steps. By carefully listening to and acknowledging the needs and requests for intervention expressed by women, practitioners not only build trust and rapport, but also create space for meaningful idea generation and exchanges of information, even when proposed interventions may not be possible within the traditional service mandate. In addition, recognizing and validating refugee and immigrant women’s desires for connection to community and the surrogate role service providers may play in the absence of trusted loved ones present an opportunity to bridge nontraditional and traditional requests and services women desire. The findings also raise questions regarding where the burden lies in responding to nuanced conceptualizations of needs such as those identified in this study. Is the burden on funders to understand nuanced needs and relevant responses, on agencies and practitioners to shift programmatic approaches and more nimbly configure their responses, or on resettling women to shift their own understanding of their needs to fit current social services approaches in the United States? Responsibility perhaps spans all parties involved, but to varying degrees. Arguably, the burden currently rests on the shoulders of women who do not fit mainstream conceptualizations of IPV survivors. It is essential that service providers make concerted efforts to gain skills and create time for assessing and understanding the needs women express in both explicit and subtle ways, while also developing grounded insights into how women conceptualize and communicate their needs. Researchers and providers together are responsible for communicating to funders and policymakers the nuanced needs they observe to ensure a better reflection of those needs in the services they provide. Future research should examine specific structural and systemic obstacles that hinder providers’ capacities to adapt and build on existing services to discern and meet diverse needs. Ongoing research must also account for how women may or may not express specific needs. The way forward requires macro-level changes to policy and funding for DV and refugee resettlement social services alike, specifically regarding the time allowed for women to access services, attain financial stability, and find their footing. Short- and long-term financial assistance, for housing in particular, are pivotal in reducing stressors and supporting decision-making processes complicated by factors associated with resettlement. In the wake of the drive for ever-increasing efficiencies in social services, practitioners must contend with the implications for practice and strive to bring into balance the push for professionalism and programmatic efficiency with survivor-centered and community-based programming priorities. Professional services must preserve and create additional space for client–provider interaction and community engagement with diverse groups to foster inclusive and innovative program development, drawing inspiration from innovations around the world (see Bhuyan, Osborne, Zahraei, & Tarshis, 2014; Michau, Horn, Bank, Dutt, & Zimmerman, 2015; Stern, Heise, & McLean, 2018). Developing approaches to safely address women’s requests for direct intervention with partners who use violence also requires consideration. Providers must strive to understand such requests within women’s frames of reference and respond in meaningful ways. The shape those responses might take requires additional understanding and engagement but would likely involve elements of psychoeducation; deeper analyses of women’s social support networks, past and present, as well as local and transnational; and risk assessment approaches that have been adapted for relevance. A robust response would also involve the development of strategies for engaging men who use violence that extend beyond the criminal justice system and court-mandated batterer intervention programs. Finally, it is important to consider these findings amid contemporary local and global contexts. The experiences of and needs expressed by women resettling in the United States must be viewed against the backdrop of increasingly anti-immigrant and antirefugee policies and sentiments. Recent shifts in policies, such as immigrant and refugee travel bans, decreased refugee resettlement, and changes in grounds for asylum in the United States, create a climate that likely hinders help seeking by members of immigrant communities, including those who resettle as refugees. Limitations A number of limitations are worthwhile to note. Generalizability is not an aim in qualitative research, and readers should situate the findings within those limitations. A larger sample of refugee participants would have allowed for greater variability in participants’ regions of origin and experiences and should factor into future research efforts to build on the current findings. Although research team members attempted to recruit four to six group B participants per focus group, scheduling was a challenge. In addition, notions of “needs” and “wants” as conceptualized by the study reflect individualistic constructs of self-determination and decision making that may not always be wholly relevant to clients originating from diverse contexts. Finally, language interpreters with a range of skills played a significant role in the study. Conclusion Nuanced needs reflect the complexity of the lives of women, profoundly affected by IPV, who fled their countries of origin and ultimately resettled to the United States as refugees. The study findings bring into focus ways in which social services may not adequately acknowledge, account for, and respond to variations in how women who resettle to the United States conceptualize and express needs. The analysis points to the need for programmatic responses that transcend mainstream conceptualizations of needs and services. Amid aims to develop culturally competent and survivor-centered practices, this study calls attention to the need for approaches to be more attuned and responsive to women’s needs and requests for support. Karin Wachter, PhD, MEd, is assistant professor, School of Social Work, Arizona State University, University Center, 411 N Central Avenue #800, Phoenix, AZ 85004; e-mail: [email protected]. Jessica Dalpe, LMSW, is technical adviser, U.S. Programs, Women’s Protection and Empowerment, International Rescue Committee, New York. Laurie Cook Heffron, PhD, LMSW, is assistant professor, School of Social Work, St. Edwards University, Austin, TX. References Abrahams , N. , Devries , K. , Watts , C. , Pallitto , C. , Petzold , M. , Shamu , S. , & García-Moreno , C. ( 2014 ). Worldwide prevalence of non-partner sexual violence: A systematic review . Lancet, 383 , 1648 – 1654 . Google Scholar Crossref Search ADS WorldCat Arnold , G. , & Ake , J. ( 2013 ). Reframing the narrative of the battered women’s movement . Violence Against Women, 19 , 557 – 578 . Google Scholar Crossref Search ADS WorldCat Bhuyan , R. , Osborne , B. , Zahraei , S. , & Tarshis , S. ( 2014 ). Unprotected, unrecognized: Canadian immigration policy and violence against women, 2008–2013 . Toronto, Ontario, Canada : University of Toronto, Migrant Mothers Project . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Bhuyan , R. , Park , Y. , & Rundle , A. ( 2012 ). Linking practitioners’ attitudes towards and basic knowledge of immigrants with their social work education . Social Work Education, 31 , 973 – 994 . Google Scholar Crossref Search ADS WorldCat Black , M. C. ( 2011 ). Intimate partner violence and adverse health consequences: Implications for clinicians . American Journal of Lifestyle Medicine, 5 , 428 – 439 . Google Scholar Crossref Search ADS WorldCat Breiding , M. J. , Black , M. C. , & Ryan , G. W. ( 2008 ). Chronic disease and health risk behaviors associated with intimate partner violence—18 US states/territories, 2005 . Annals of Epidemiology, 18 , 538 – 544 . Google Scholar Crossref Search ADS WorldCat Campbell , J. C. ( 2002 ). Health consequences of intimate partner violence . Lancet, 359 , 1331 – 1336 . Google Scholar Crossref Search ADS WorldCat Capps , R. , Newland , K. , Fratzke , S. , Groves , S. , Fix , M. , McHugh , M. , & Auclair , G. ( 2015 ). The integration outcomes of U.S. refugees: Successes and challenges . Retrieved from https://www.migrationpolicy.org/research/integration-outcomes-us-refugees-successes-and-challenges Connor , P. , & Krogstad , J. M. ( 2018 ). For the first time, U.S. resettles fewer refugees than the rest of the world. Retrieved from https://www.pewresearch.org/fact-tank/2018/07/05/for-the-first-time-u-s-resettles-fewer-refugees-than-the-rest-of-the-world/ Creswell , J. W. ( 2013 ). Qualitative inquiry and research design: Choosing among five approaches . Thousand Oaks, CA : Sage Publications . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Davies , J . ( 2008 ). When battered women stay . . . Advocacy beyond leaving (Building Comprehensive Solutions to Domestic Violence Publication 20). Retrieved from https://vawnet.org/sites/default/files/materials/files/2016-08/BCS20_Staying.pdf Danso , R. ( 2018 ). Cultural competence and cultural humility: A critical reflection on key cultural diversity concepts. Journal of Social Work, 18 , 410–430. doi:1468017316654341 Dominelli , L. ( 2004 ). Culturally competent social work: A way toward international anti-racist social work? In L. M. Gutiérrez , M. Zuniga , & D. Lum (Eds.), Education for multicultural social work practice: Critical viewpoints and future directions (pp. 281 – 294 ). Alexandria, VA : Council on Social Work Education . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Dutton , M. A. , Green , B. L. , Kaltman , S. I. , Roesch , D. M. , Zeffiro , T. A. , & Krause , E. D. ( 2006 ). Intimate partner violence, PTSD, and adverse health outcomes . Journal of Interpersonal Violence, 21 , 955 – 968 . Google Scholar Crossref Search ADS WorldCat Ellsberg , M. , Jansen , H. A. , Heise , L. , Watts , C. H. , & Garcia-Moreno , C. ( 2008 ). Intimate partner violence and women’s physical and mental health in the WHO multi-country study on women’s health and domestic violence: An observational study . Lancet, 371 , 1165 – 1172 . Google Scholar Crossref Search ADS WorldCat Erez , E. , Adelman , M. , & Gregory , C. ( 2009 ). Intersections of immigration and domestic violence: Voices of battered immigrant women . Feminist Criminology, 4 , 32 – 56 . Google Scholar Crossref Search ADS WorldCat Fix , M. , Hooper , K. , & Zong , J. ( 2017 ). How are refugees faring? Integration at U.S. and state levels . Retrieved from https://www.migrationpolicy.org/research/how-are-refugees-faring-integration-us-and-state-levels Furlong , M. , & Wight , J. ( 2011 ). Promoting “critical awareness” and critiquing “cultural competence”: Towards disrupting received professional knowledges . Australian Social Work, 64 , 38 – 54 . Google Scholar Crossref Search ADS WorldCat Goodman , L. A. , & Epstein , D. ( 2008 ). Listening to battered women: A survivor-centered approach to advocacy, mental health, and justice . Washington, DC : American Psychological Association . Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC Goodman , L. A. , Thomas , K. , Cattaneo , L. B. , Heimel , D. , Woulfe , J. , & Chong , S. K. ( 2016 ). Survivor-defined practice in domestic violence work: Measure development and preliminary evidence of link to empowerment . Journal of Interpersonal Violence, 31 , 163 – 185 . Google Scholar Crossref Search ADS WorldCat Guest , G. , MacQueen , K. M. , & Namey , E. E. ( 2012 ). Applied thematic analysis . London : Sage Publications . Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC Hahn , S. A. , & Postmus , J. L. ( 2014 ). Economic empowerment of impoverished IPV survivors: A review of best practice literature and implications for policy . Trauma, Violence, & Abuse, 15 , 79 – 93 . Google Scholar Crossref Search ADS WorldCat Johnson , Y. M. , & Munch , S. ( 2009 ). Fundamental contradictions in cultural competence . Social Work, 54 , 220 – 231 . Google Scholar Crossref Search ADS WorldCat Kennedy , A. C. , Adams , A. , Bybee , D. , Campbell , R. , Kubiak , S. P. , & Sullivan , C. ( 2012 ). A model of sexually and physically victimized women’s process of attaining effective formal help over time: The role of social location, context, and intervention . American Journal of Community Psychology, 50 , 217 – 228 . Google Scholar Crossref Search ADS WorldCat Kim , M. E. ( 2013 ). Challenging the pursuit of criminalisation in an era of mass incarceration: The limitations of social work responses to domestic violence in the USA . British Journal of Social Work, 43 , 1276 – 1293 . Google Scholar Crossref Search ADS WorldCat Lee , Y. S. , & Hadeed , L. ( 2009 ). Intimate partner violence among Asian immigrant communities: Health/mental health consequences, help-seeking behaviors, and service utilization . Trauma, Violence, & Abuse, 10 , 143 – 170 . Google Scholar Crossref Search ADS WorldCat Lindhorst , T. , Meyers , M. , & Casey , E. ( 2008 ). Screening for domestic violence in public welfare offices . Violence Against Women, 14 , 5 – 28 . Google Scholar Crossref Search ADS WorldCat Mehrotra , G. R. , Kimball , E. , & Wahab , S. ( 2016 ). The braid that binds us: The impact of neoliberalism, criminalization, and professionalization on domestic violence work . Affilia, 31 , 153 – 163 . doi: 10.1177/0886109916643871 Google Scholar Crossref Search ADS WorldCat Crossref Menjívar , C. , & Salcido , O. ( 2002 ). Immigrant women and domestic violence: Common experiences in different countries . Gender & Society, 16 , 898 – 920 . Google Scholar Crossref Search ADS WorldCat Michau , L. , Horn , J. , Bank , A. , Dutt , M. , & Zimmerman , C. ( 2015 ). Prevention of violence against women and girls: Lessons from practice . Lancet, 385 , 1672 – 1684 . Google Scholar Crossref Search ADS WorldCat Mills , L. G. ( 1999 ). Killing her softly: Intimate abuse and the violence of state intervention . Harvard Law Review, 113 , 550 – 613 . Google Scholar Crossref Search ADS WorldCat Minor , O. M. , & Cameo , M. ( 2018 ). A comparison of wages by gender and region of origin for newly arrived refugees in the USA . Journal of International Migration and Integration, 19 , 813 – 828 . Google Scholar Crossref Search ADS WorldCat Office of Refugee Resettlement . ( 2014 ). About the Voluntary Agencies Matching Grant Program. Retrieved from https://www.acf.hhs.gov/orr/programs/matching-grants/about Padgett , D. K. ( 2016 ). Qualitative methods in social work research (Vol. 36). Thousand Oaks, CA: Sage Publications. OpenURL Placeholder Text WorldCat Postmus , J. L. , Severson , M. , Berry , M. , & Yoo , J. A. ( 2009 ). Women’s experiences of violence and seeking help . Violence Against Women, 15 , 852 – 868 . Google Scholar Crossref Search ADS WorldCat Rees , S. , & Pease , B. ( 2007 ). Domestic violence in refugee families in Australia . Journal of Immigrant & Refugee Studies, 5 ( 2 ), 1 – 19 . Google Scholar Crossref Search ADS WorldCat Rivara , F. P. , Anderson , M. L. , Fishman , P. , Bonomi , A. E. , Reid , R. J. , Carrell , D. , & Thompson , R. S. ( 2007 ). Healthcare utilization and costs for women with a history of intimate partner violence . American Journal of Preventive Medicine, 32 , 89 – 96 . Google Scholar Crossref Search ADS WorldCat Rodgers , B. L. , & Cowles , K. V. ( 1993 ). The qualitative research audit trail: A complex collection of documentation . Research in Nursing & Health, 16 , 219 – 226 . Google Scholar Crossref Search ADS WorldCat Rodríguez , M. , Valentine , J. M. , Son , J. B. , & Muhammad , M. ( 2009 ). Intimate partner violence and barriers to mental health care for ethnically diverse populations of women . Trauma, Violence, & Abuse, 10 , 358 – 374 . Google Scholar Crossref Search ADS WorldCat Runner , M. , Yoshihama , M. , & Novick , S. ( 2009 ). Intimate partner violence in immigrant and refugee communities: Challenges, promising practices and recommendations. Retrieved from https://www.futureswithoutviolence.org/userfiles/file/ImmigrantWomen/IPV_Report_March_2009.pdf Saldaña , J. ( 2015 ). The coding manual for qualitative researchers (No. 14). Thousand Oaks, CA : Sage Publications . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Smyth , K. F. , Goodman , L. , & Glenn , C. ( 2006 ). The full-frame approach: A new response to marginalized women left behind by specialized services . American Journal of Orthopsychiatry, 76 , 489 – 502 . Google Scholar Crossref Search ADS WorldCat Stark , L. , & Ager , A. ( 2011 ). A systematic review of prevalence studies of gender-based violence in complex emergencies . Trauma, Violence, & Abuse, 12 , 127 – 134 . Google Scholar Crossref Search ADS WorldCat Stern , E. , Heise , L. , & McLean , L. ( 2018 ). Engaging opinion leaders in an IPV-prevention programme: Lessons from Indashyikirwa in Rwanda. Retrieved from https://www.whatworks.co.za/documents/publications/202-opinion-leaders-web/file United Nations High Commissioner for Refugees . ( 2018 ). Global trends: Forced displacement in 2017. Retrieved from https://www.refworld.org/docid/594aa38e0.html Usta , J. , Farver , J.A.M. , & Zein , L. ( 2008 ). Women, war, and violence: Surviving the experience . Journal of Women’s Health, 17 , 793 – 804 . Google Scholar Crossref Search ADS WorldCat Vidales , G. T. ( 2010 ). Arrested justice: The multifaceted plight of immigrant Latinas who faced domestic violence . Journal of Family Violence, 25 , 533 – 544 . Google Scholar Crossref Search ADS WorldCat Vu , A. , Adam , A. , Wirtz , A. , Pham , K. , Rubenstein , L. , Glass , N. , et al. ( 2013 ). The prevalence of sexual violence among female refugees in complex humanitarian emergencies: A systematic review and meta-analysis . PLOS Currents , 6 . doi: 10.1371/currents.dis.835f10 778fd80ae031aac12d3b533ca7 OpenURL Placeholder Text WorldCat Crossref Wachter , K. , Cook Heffron , L. , & Dalpe , J. ( 2019 ). “Back home you just go talk to the family”: The role of family among women who seek help for intimate partner violence pre- and post-resettlement to the United States . Journal of Interpersonal Violence. doi: 10.1177/0886260519835861 OpenURL Placeholder Text WorldCat Crossref Wachter , K. , & Gulbas , L. E. ( 2018 ). Social support under siege: An analysis of forced migration among women from the Democratic Republic of Congo . Social Science & Medicine, 208 , 107 – 116 . doi: 10.1016/j.socscimed.2018.04.056 Google Scholar Crossref Search ADS WorldCat Crossref Wachter , K. , & Snyder , S. ( 2018 ). Centering subjectivities: Theoretical considerations for practice with women in forced migration . Affilia, 33 , 395 – 409 . Google Scholar Crossref Search ADS WorldCat Warrier , S. ( 2009 ). Culture and cultural competency in addressing intimate partner violence. In C. Mitchell & D. Anglin (Eds.), Intimate partner violence: A health-based perspective (pp. 79 – 86 ). New York : Oxford University Press . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Zinzow , H. M. , Resnick , H. S. , Amstadter , A. B. , McCauley , J. L. , Ruggiero , K. J. , & Kilpatrick , D. G. ( 2010 ). Drug- or alcohol-facilitated, incapacitated, and forcible rape in relationship to mental health among a national sample of women . Journal of Interpersonal Violence, 25 , 2217 – 2236 . Google Scholar Crossref Search ADS WorldCat © 2019 National Association of Social Workers This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Stalker, Katie, Cotter

doi: 10.1093/swr/svz018pmid: N/A

Abstract Juvenile justice diversion programs, such as Teen Court (TC), represent an alternative to traditional juvenile justice responses to youth misbehavior and delinquency. However, although TC represents a potential strategy to address disproportionate minority contact, there is a dearth of research examining the extent to which TC programs are racially equitable. To address this gap, the current study examines racial disproportionality in a TC program in Arizona. Results indicated that in a diverse sample of youths involved in a TC program in Arizona, youths who identified as Latinx or American Indian were more likely to receive a severe consequence from the peer jury compared with their non-Latinx, white counterparts. Multiracial youths were less likely to receive a severe consequence compared with white youths. A hierarchical regression model indicated that offense-related variables explained the largest proportion of variance in number of consequence hours assigned. However, disparities for Latinx and American Indian youths compared with non-Latinx, white youths persisted after controlling for other demographics, type of offense, prior offenses, and additional charges. The results of the current study are the first to document racial disparity in the TC process. Disproportionate minority contact (DMC) refers to disproportionality in the number of minority youths who come into contact with the juvenile justice system (Office of Juvenile Justice and Delinquency Prevention [OJJDP], 2018b). The Juvenile Justice and Delinquency Prevention Act of 2002 (JJDPA) broadened the scope of DMC from “disproportionate minority confinement” to represent an expanded focus on disproportionality at each decision point in the juvenile justice system in addition to sentencing decisions (OJJDP, 2018b). DMC is a pervasive problem in the United States juvenile justice system. Indeed, according to OJJDP (2012), disproportionate minority representation is evident at nearly every contact point (for example, referral, diversion, confinement) in most jurisdictions. In 2002, amendments to JJDPA explicitly required states to address disproportionality throughout the juvenile justice process (Dawson-Edwards, Tewksbury, & Nelson, 2017). Furthermore, researchers and policymakers have called for alternatives to traditional juvenile justice responses to youth misbehavior and delinquency (Laub, 2014; Leiber & Rodriguez, 2011), such as juvenile justice diversion programs. Despite the potential for juvenile justice diversion programs to address DMC, little research has focused on racial equity within such programs. The current study focused on a popular juvenile justice diversion program, Teen Court (TC). TC programs provide an alternative to traditional juvenile justice system punishments. Rather than facing involvement with the juvenile justice system, nonchronic juvenile offenders who participate in TC are given consequences (for example, apology letter, workshops) by a youth jury made up of their peers (Butts, Buck, & Coggeshall, 2002). By providing an alternative process for determining consequences, TC programs represent an opportunity to address DMC. However, due to a lack of research, it is unclear whether the TC peer-derived consequences are racially equitable (that is, whether youths who participate in TC receive consequences of equal severity regardless of racial identification). Consequently, the aim of the current study was to examine whether the severity of TC peer-derived consequences varied based on racial identification in a TC program in Arizona. DMC in the Juvenile Justice System The relative rate index (RRI) is a measure used to compare rates of contact with the juvenile justice system among different groups of youths (OJJDP, 2018a). In 2015, national disparities were documented at each juncture in the juvenile justice system: referral to the juvenile justice system, diversion (youth given opportunity to avoid formal involvement with juvenile justice system), detention (youth placed in secure detention prior to hearing), petition (document filed with the court requesting transfer or adjudicatory hearing), adjudication (youth found to be delinquent), probation (youth found to be delinquent and placed on probation), placement (youth found to be delinquent and placed in secure confinement), and waiver (case sent to criminal court) (Puzzanchera & Hockenberry, 2017). The largest disparities existed in the rates of referral and detention. The national RRI for referrals to the juvenile justice system was 3.1 for black youths (that is, black youths were 3.1 times as likely to be referred to the juvenile justice system compared with their white counterparts) (Puzzanchera & Hockenberry, 2017). The national RRI for detention was 1.3 for black and American Indian/Alaskan Native youths and 1.5 for Latinx youths (Puzzanchera & Hockenberry, 2017). Several research studies have also documented DMC. In a review of studies examining racial disparities in the juvenile justice system, Crutchfield, Fernandes, and Martinez (2010) found that, at the referral stage, five studies showed strong to moderate disparities, two studies showed minimal disparities, and one study showed no disparity. The authors concluded that overall the results provide evidence of racial and ethnic disparities in the juvenile justice system. Other research studies have documented the persistence of DMC after controlling for self-reported delinquency (Crutchfield, Skinner, Haggerty, McGlynn, & Catalano, 2009) and other factors associated with the offense (Armstrong & Rodriguez, 2005; Evangelist, Ryan, Victor, Moore, & Perron, 2017; Leiber & Fox, 2005; Shook & Goodkind, 2009). Another study found that disparities in the juvenile justice system were cumulative such that harsher treatment for youths of color relative to their white counterparts at the front end of the juvenile justice system (that is, lower likelihood of being diverted from formal processing and higher likelihood of being detained before adjudication hearing) was associated with harsher treatment at the back end of the juvenile justice system (that is, higher likelihood of receiving an out-of-home placement) (Rodriguez, 2010). In addition to cumulative effects within the juvenile justice process itself, DMC has the potential to exacerbate racial inequities outside of the juvenile justice system that persist into adulthood. Involvement with the juvenile justice system during adolescence has negative consequences throughout the life course, marked by labeling, stigma, decreased opportunities, and decreased bonding with adults—a phenomenon described as cumulative disadvantage (Sampson & Laub, 1997). Together, these empirical studies document racial inequity across the juvenile justice system and highlight the long-term, negative consequences for youths of color, which underlines the need for interventions to address DMC. DMC in the Arizona Juvenile Justice System Although DMC is prevalent across the nation, state and local efforts are needed in addition to national efforts (Cabaniss, Frabutt, Kendrick, & Arbuckle, 2007; Kempf-Leonard, 2007). Thus, when examining DMC, it is important to consider state and local settings. For instance, one study in Missouri revealed that, compared with white youths, black youths were more likely to receive formal referral for juvenile justice processing, to experience pretrial detention, to be adjudicated, and to receive out-of-home placement in a large, urban county; however, there were no racial differences at these decision points in a smaller, rural county in the state (Ray & Alarid, 2004). Furthermore, another study that involved state-level examinations of DMC reported that overrepresentation tended to be higher in states where the proportion of minorities was lower (Leiber, 2002). In the context of the current study, it is important to consider the extent of DMC in Arizona. A report examining disparities in the Arizona juvenile justice system from 2013 to 2014 focused on nine decision points: referral (written document indicating a youth committed a delinquent act), direct file (prosecutor files case directly to adult court), deferment (charges filed to adult court due to the child’s age), diversion (youth given opportunity to avoid formal involvement with juvenile court), detention (youth placed in secure detention facility prior to adjudication), petition (filing of a written petition that child is delinquent and not eligible for diversion), adjudicated delinquent (youth is found delinquent by the court), probation (judge decides that youth will be placed on formal probation), and confinement (judge decides to place youth in secure facility) (Haight & Jarjoura, 2016). The authors summarized the following key findings: (a) Latinx youths were not overrepresented in referrals; however, black youths and American Indian youths were more likely to be referred to the juvenile justice system than their white counterparts and (b) Latinx and African American/black youths were overrepresented in the most severe punishments (that is, filings to adult court, placement in pre-adjudicatory detention, confinement) and underrepresented in diversion. Another study examined the comprehensive effects of race and ethnicity on diversion, petition, detention, adjudication, and out-of-home placement in Arizona in 2000 (Rodriguez, 2010). Results indicated that after controlling for other demographics, type of offense, prior referrals, type of county, and structural disadvantage, disparities existed for black, Latinx, and American Indian youths relative to white youths. However, there were nuances to these findings. Specifically, compared with their white counterparts, black youths were diverted from formal processing less often, experienced pre-adjudication detention more often, and experienced out-of-home placement more often. On the other hand, black youths were adjudicated less often than their white counterparts. Latinx and American Indian youths were less likely to be diverted from formal processing and more likely to be detained prior to adjudication compared with white youths. Together, studies on DMC for juveniles in Arizona suggest that disparities exist for Latinx, African American/black, and American Indian youths compared with their white counterparts at several decision points in the juvenile justice system. Addressing Disproportionality: TC as an Alternative Consequence The need for intervention to address DMC is clear; however, evaluation of existing intervention effectiveness to reduce disproportionality is minimal (Leiber & Rodriguez, 2011; Piquero, 2008). According to Leiber and Rodriguez (2011), direct services, such as diversion programs, represent one strategy to address DMC. Laub (2014) echoed the potential utility and need to evaluate diversion programs as alternative consequences to juvenile justice involvement. TC is one such diversion program that provides alternative consequences for youths who are involved with the juvenile justice system. Indeed, Cole and Heilig (2011) discussed the potential utility of TC programs in addressing the disproportionality inherent in the school-to-prison pipeline, a construct that highlights the connection between disproportionality in school discipline practices that leads to increased contact between youths of color and the juvenile justice system. TC models differ based on the roles of adults and youths in the courtroom; for example, the TC judge can be a youth or an adult, and youth attorneys may or may not be involved in the process (Butts et al., 2002). However, most often across TC programs, peer juries hear arguments from adolescent attorneys and are responsible for suggesting or determining appropriate constructive consequences (Butts et al., 2002). These consequences often include community service, restitution, letters of apology, TC “jury duty,” and educational workshops (Butts et al., 2002). The peer jury is tasked with hearing the case and assigning constructive consequences that give the respondent the opportunity to repair the harm that was caused by his or her actions and develop skills needed to avoid future misbehavior. If the respondent successfully completes the constructive consequences, his or her juvenile court case is closed; otherwise, the youth is referred back for traditional justice system processing (Butts et al., 2002). TC is a popular diversion program, with estimates of more than 1,800 programs in operation (Global Youth Justice, 2018). Despite its prevalence, research on the effectiveness of TC programs is minimal. Existing research tends to focus on recidivism as an outcome (for example, Cotter & Evans, 2018; Gase, Schooley, DeFosset, Stoll, & Kuo, 2016). Recent systematic reviews (Cotter & Evans, 2018; Gase et al., 2016) have examined studies on TC programs to date and concluded that differences in program components and research designs across studies make it difficult to compare results, thus additional studies are needed to draw conclusions about the impact of TC. Although study results are not directly comparable, Gase et al. (2016) reported that across 15 studies that reported recidivism results, four reported results favoring TC, one favored the traditional juvenile justice system, and 10 reported null results. Uncertainty surrounding program efficacy is not unique to TC—extant research on diversion programs in general is limited. Schwalbe, Gearing, MacKenzie, Brewer, and Ibrahim (2012) conducted a comprehensive meta-analysis of diversion programs for juvenile offenders and underlined the heterogeneous nature of diversion research. Results of the meta-analysis indicated that of the five types of diversion programs identified (that is, case management, individual treatment, family treatment, youth court, and restorative justice), only family treatment was associated with a statistically significant impact on recidivism (Schwalbe et al., 2012). Another recent study reported that the impact of TC may have implications beyond recidivism: Smokowski et al. (2017) found that, compared with a comparison group of youths who received a positive behavior intervention and a no-treatment comparison group, TC participants reported greater decreases in internalizing symptoms and parent–child conflict. Furthermore, relative to the no-treatment comparison group, TC participants reported greater improvements in a number of psychosocial functioning, social, and school indicators (that is, school satisfaction, association with delinquent friends, externalizing behavior, self-esteem, and violent behavior) (Smokowski et al., 2017). Although researchers have acknowledged the potential role of TC in addressing DMC, a review of the literature revealed a dearth of studies examining racial disproportionality in the TC program (that is, whether the peer-derived consequences are racially equitable). To my knowledge, only a single study has examined racial differences in consequence severity. Rasmussen and Diener (2005) did not find differences in sentence severity based on race in a sample of 38 TC participants. Despite encouraging findings, this study was limited in terms of statistical power due to sample size and the majority of the sample (68%) identified as white, which further limited the ability to examine racial differences. The authors also failed to control for other demographic factors (for example, age, gender) or type of offense in their models. Controlling for these variables is essential in that additional demographic variables and the severity of the offense committed likely influence the peer jury’s decision regarding appropriate consequences (Engen, Steen, & Bridges, 2002). In sum, although TC represents a potential intervention to address DMC, it is unclear whether TC peer-derived consequences are racially equitable. In other words, it is possible that we are diverting youths from an inequitable juvenile justice system into an inequitable diversion program. Identifying and addressing racial inequity is a key goal of social work practice. According to the American Academy of Social Work and Social Welfare (AASWSW) (2018), achieving “equal opportunity and justice” is one of 12 grand challenges for the profession. Therefore, given a general lack of previous research and considerable limitations in existing research, the current study seeks to address this research gap by examining racial disproportionality in a TC program. Theoretical Framework Macro-contextual theories of racial disparities in punishment suggest that characteristics of juvenile courts and communities influence whether racial disparities exist in a juvenile justice agency (Engen et al., 2002). Given that philosophy, structure, and procedure differ considerably across agencies, such characteristics are expected to predict whether racial disparities exist. In addition to these organizational characteristics, according to macro-contextual theorists, community characteristics influence formal social control, which has the potential to affect racial equity in formal juvenile punishment (Engen et al., 2002). For instance, large minority populations may lead to increased “social threat” among the white middle class that manifests in the form of intense punishment by formal institutions of social control (Sampson & Laub, 1993). In theory, TC programs differ considerably from traditional juvenile justice agencies in terms of both program philosophy or procedure and community involvement (the extent to which this is true in local practice remains questionable; see DeFosset, Schooley, Abrams, Kuo, & Gase, 2017, for a thorough case study on the theoretical underpinning of a TC). First, although TC is similar to the traditional juvenile justice system in that it relies on hierarchical decision making and procedural consistence, TC is based on a philosophy that TC participants are less likely to reoffend if they are given consequences by other youths, which is exemplified in that the TC hearing is a youth-controlled process (DeFosset et al., 2017). Second, the peer jury is often made up of both previous TC participants and youth volunteers from the surrounding community, which suggests that the youth jury is reflective of youths in the larger community. Therefore, given that the philosophy, process, and community involvement of TC differs from that of the traditional juvenile justice system, it is necessary to examine disproportionality in the TC process. Research Questions/Hypotheses Three research questions guide the current study: (1) What are the RRIs for peer-derived consequences assigned to Latinx, American Indian, and African American youths relative to their non-Latinx/white counterparts? (2) After controlling for offense-related variables, age, gender, and family income, are there significant differences in the number of peer-derived consequence hours assigned based on TC respondents’ race? (3) What is the relative impact of respondents’ demographics, offense-related variables, and respondents’ race on severity of consequences received? Overall, the proposed study addresses limitations of previous studies on racial equity in TC peer-derived consequences by analyzing a large, diverse sample of TC youths and controlling for type of offense and other relevant demographics, including age, gender, and family income. Method Setting The setting for the current study was a TC program in a large county in Arizona that serves both urban and rural areas. According to the U.S. Census Bureau (2017), approximately 37% of residents identified as Latinx; in terms of race, approximately 85% identified as white, 4% as black/African American, 4% American Indian/Alaskan Native, 3% as Asian, and 3% as two or more races. Overall, household income levels were lower than the national average (median household income was approximately $49,000 compared with the U.S. median income of $57,652) and poverty levels were higher (approximately 17% living in poverty compared with the national average of 12.3%) (U.S. Census Bureau, 2017). In the TC program, youths between the ages of 12 and 17 who have been charged with a misdemeanor offense can be referred to TC by their probation officer. These youths are referred to as respondents. TC respondents have typically been charged with their first, second, or third misdemeanor offense. Involvement with TC lasts approximately 30 days. A second group of youths involved in the TC process are the program volunteers. Youths in the community between the ages of 12 and 18 are invited to participate as volunteers. Volunteers serve a variety of roles including attorneys, bailiffs, clerks, and jurors. On a typical hearing day, respondents and parents arrive and complete an intake with a TC staff member. Then respondents meet with their peer attorney, who learns details of the incident. At the hearing, peer attorneys ask the respondent questions about the incident in front of the peer jury, which is made up of youths who have previously gone through TC and other youths from the community who volunteer to participate in the process. The respondent and parent of the respondent are also given the opportunity to speak to the peer jury. Next, the peer jury leaves the courtroom and deliberates to come to a consensus on the constructive consequences to give the respondent. Possible constructive consequences include jury duties (the respondent will serve on the jury at another hearing), letters of apology, journal writing, independent studies, homework help, peer mediation, and a variety of workshops (for example, shoplifting workshop, anger management, self-improvement, education-focused goals, substance abuse prevention). After the hearing, the respondent is required to complete all of the constructive consequences assigned by the peer jury. On average, the respondent completes all required consequences within 30 days. If the respondent successfully completes the TC program, their juvenile court case is closed and their arrest will not appear on their record. Data After receiving institutional review board approval, I obtained data from the TC program staff. The data file included all TC cases heard in the TC program from 2012 until 2016 (1,258 cases). The data set contained demographic variables, type of offense (for example, disorderly conduct, drug-related offense, assault), prior offenses, additional charges, and a variable indicating the number of hours of consequences assigned by the peer jury. The dependent variable for the current study was the number of hours of consequences assigned by the peer jury. The number of hours assigned was the best available measure of consequence severity. The peer jury is given leeway in determining consequences based on what they deem appropriate given the details of the case. It stands to reason that when the peer jury decides that a severe consequence is necessary, they assign more consequences that total a greater number of hours. The length of time required to complete each of the possible consequences (that is, workshops, letters of apology, jury duty) were summed by TC staff members to reflect the severity of consequences assigned to the respondent. For consequences in which the amount of time is not fixed (for example, letter writing), TC staff members estimated the approximate amount of time to complete the task. Demographics included ethnicity (Latinx or non-Latinx), race (African American, American Indian, multiracial, white, Asian, Pacific Islander), age, gender (male, female), and family income. Family income was measured as percentile categories based on the city in which the program was located: <30th percentile, 31st–50th percentile, 51st–80th percentile, and >80th percentile. Race, ethnicity, age, and gender were self-reported by youths. Family income was reported by parents and guardians. The type of offense was coded as a series of dummy variables and included assault, burglary, criminal damage, domestic violence, disorderly conduct, drug-related offense, alcohol-related offense, shoplifting, trespassing, or other. Trespassing was the reference group. Number of prior offenses included a count of the respondent’s number of prior arrests. When arrested, it is possible to be charged with multiple offenses. The variable additional offense was a count indicating the number of additional offenses the respondent was charged with at the time of the primary offense. Information on offense was provided in the referral to TC by probation officers. Data Analysis The administrative data included 1,258 TC cases. Due to small sample sizes for individuals identifying as Asian (n = 10) and Pacific Islander (n = 8), these cases were excluded from the analysis sample. Only four cases were missing data on variables of interest. Given the small proportion of missing data (0.3%), these cases were excluded from the analysis sample. The final analytic sample was 1,236. The first step of data analysis involved calculating RRIs. RRIs are the most common measure used to gauge racial disproportionality in juvenile justice processes (OJJDP, 2018a). RRIs are computed by calculating the rate of a given outcome for youths of color and dividing by the rate of the outcome for white youths. If the resulting RRI equals 1, then there is no evidence of disproportionality; that is, the likelihood of the outcome is equal for youths of color and their white counterparts. If, however, the RRI is greater than 1, this indicates that, compared with their white counterparts, youths of color are more likely to experience a given outcome. In the current study, the RRIs were calculated for receiving a “severe” consequence by the peer jury. A severe consequence was defined as receiving a total number of consequence hours above the median for the sample (19 or more hours). The second step of data analysis involved using regression modeling to determine the factors associated with the number of consequence hours assigned by the peer jury while controlling for other demographics and offense-related variables. The demographic and offense-related variables were selected based on previous literature examining disparities in the juvenile justice system (Armstrong & Rodriguez, 2005; Evangelist et al., 2017; Leiber & Fox, 2005; Rodriguez, 2010; Shook & Goodkind, 2009) and research on the factors that TC juries take into consideration (Engen et al., 2002). Hierarchical regression modeling was used to examine the relative impact of race, other demographics, and offense-related variables. Variables were entered into the model in three blocks. Results In terms of ethnicity, the majority of the sample identified as Latinx (59.8%). In terms of race, 73.3% identified as white, 15.3% as multiracial, 6.9% as African American/black, and 4.5% as American Indian (see Table 1). About 65% of the sample were male. The mean age of the sample was 15 years (SD = 1.50). The majority of the sample (49%) reported having a family income that was less than or equal to the 30th percentile for the city in which the program was located. Regarding offense-related characteristics, the majority of TC respondents were charged with a drug-related offense (39.5%), shoplifting (13.2%), or assault (11.6%). Approximately 30% of respondents were charged with one additional offense beyond the primary offense, and 14.2% had two additional charges. On average, respondents had 1.8 prior offenses at the time they were referred to the TC program. See Table 1 for additional sample descriptive statistics. Table 1 Sample Descriptive Statistics Characteristic . n (%) . M (SD) . Ethnicity Latinx 739 (59.8) Non-Latinx 497 (40.2) Race American Indian 56 (4.5) African American 85 (6.9) Multiracial 189 (15.3) White 906 (73.3) Gender (male) 809 (65.5) Age 15.0 (1.5) Offense Alcohol-related offense 87 (7.0) Assault 143 (11.6) Burglary/theft 54 (4.4) Criminal damage 51 (4.1) Criminal trespassing 46 (3.7) Domestic violence 48 (3.9) Disorderly conduct 94 (7.6) Drug-related offense 488 (39.5) Shoplifting 163 (13.2) Other 62 (5.0) Number of additional offenses 0 687 (55.6) 1 373 (30.2) 2 176 (14.2) Number of prior offenses 1.8 (1.2) Characteristic . n (%) . M (SD) . Ethnicity Latinx 739 (59.8) Non-Latinx 497 (40.2) Race American Indian 56 (4.5) African American 85 (6.9) Multiracial 189 (15.3) White 906 (73.3) Gender (male) 809 (65.5) Age 15.0 (1.5) Offense Alcohol-related offense 87 (7.0) Assault 143 (11.6) Burglary/theft 54 (4.4) Criminal damage 51 (4.1) Criminal trespassing 46 (3.7) Domestic violence 48 (3.9) Disorderly conduct 94 (7.6) Drug-related offense 488 (39.5) Shoplifting 163 (13.2) Other 62 (5.0) Number of additional offenses 0 687 (55.6) 1 373 (30.2) 2 176 (14.2) Number of prior offenses 1.8 (1.2) Open in new tab Table 1 Sample Descriptive Statistics Characteristic . n (%) . M (SD) . Ethnicity Latinx 739 (59.8) Non-Latinx 497 (40.2) Race American Indian 56 (4.5) African American 85 (6.9) Multiracial 189 (15.3) White 906 (73.3) Gender (male) 809 (65.5) Age 15.0 (1.5) Offense Alcohol-related offense 87 (7.0) Assault 143 (11.6) Burglary/theft 54 (4.4) Criminal damage 51 (4.1) Criminal trespassing 46 (3.7) Domestic violence 48 (3.9) Disorderly conduct 94 (7.6) Drug-related offense 488 (39.5) Shoplifting 163 (13.2) Other 62 (5.0) Number of additional offenses 0 687 (55.6) 1 373 (30.2) 2 176 (14.2) Number of prior offenses 1.8 (1.2) Characteristic . n (%) . M (SD) . Ethnicity Latinx 739 (59.8) Non-Latinx 497 (40.2) Race American Indian 56 (4.5) African American 85 (6.9) Multiracial 189 (15.3) White 906 (73.3) Gender (male) 809 (65.5) Age 15.0 (1.5) Offense Alcohol-related offense 87 (7.0) Assault 143 (11.6) Burglary/theft 54 (4.4) Criminal damage 51 (4.1) Criminal trespassing 46 (3.7) Domestic violence 48 (3.9) Disorderly conduct 94 (7.6) Drug-related offense 488 (39.5) Shoplifting 163 (13.2) Other 62 (5.0) Number of additional offenses 0 687 (55.6) 1 373 (30.2) 2 176 (14.2) Number of prior offenses 1.8 (1.2) Open in new tab The results of research question 1 (“What are the RRIs for peer-derived consequences assigned to Latinx, American Indian, and African American youths relative to their non-Latinx/white counterparts?”) are presented in Table 2. The RRI calculation indicated that Latinx youths were 1.19 times as likely to receive a severe consequence compared with non-Latinx youths. An RRI of 1.42 indicated that compared with their white counterparts, American Indian youths were 1.42 times as likely to receive a severe consequence. According to RRI calculations, multiracial youths and African American/black youths received fewer severe consequences compared with their white counterparts. Table 2 Differences in Receiving Severe Consequences by Race and Ethnicity . Total . Severe Consequences . . Characteristic . n . n (%) . RRI . Ethnicity Latinx 739 373 (50.5) 1.19 Non-Latinx 497 210 (42.3) Race American Indian 56 38 (67.9) 1.42 African American 85 37 (43.5) 0.91 Multiracial 189 74 (39.2) 0.82 White 906 434 (47.9) . Total . Severe Consequences . . Characteristic . n . n (%) . RRI . Ethnicity Latinx 739 373 (50.5) 1.19 Non-Latinx 497 210 (42.3) Race American Indian 56 38 (67.9) 1.42 African American 85 37 (43.5) 0.91 Multiracial 189 74 (39.2) 0.82 White 906 434 (47.9) Notes: RRI = relative rate index. A severe consequence was defined as receiving consequences that totaled 19 or more hours. Open in new tab Table 2 Differences in Receiving Severe Consequences by Race and Ethnicity . Total . Severe Consequences . . Characteristic . n . n (%) . RRI . Ethnicity Latinx 739 373 (50.5) 1.19 Non-Latinx 497 210 (42.3) Race American Indian 56 38 (67.9) 1.42 African American 85 37 (43.5) 0.91 Multiracial 189 74 (39.2) 0.82 White 906 434 (47.9) . Total . Severe Consequences . . Characteristic . n . n (%) . RRI . Ethnicity Latinx 739 373 (50.5) 1.19 Non-Latinx 497 210 (42.3) Race American Indian 56 38 (67.9) 1.42 African American 85 37 (43.5) 0.91 Multiracial 189 74 (39.2) 0.82 White 906 434 (47.9) Notes: RRI = relative rate index. A severe consequence was defined as receiving consequences that totaled 19 or more hours. Open in new tab Table 3 Predictors of Number of Hours of Peer-Derived Consequences (N = 1,236) . Model 1 . . Model 2 . . Model 3 . Predictor . B . SE . p . . B . SE . p . . B . SE . p . Male (female) 0.401 0.335 .232 0.229 0.328 .485 0.210 0.325 .518 Age 0.171 0.107 .109 –0.071 0.107 .509 –0.072 0.106 .497 Income –0.511 0.142 .000 –0.301 0.138 .029 –0.138 0.142 .333 Type of offense (trespassing) Assault 3.374 0.909 .000 3.166 0.901 .000 Burglary 3.718 1.077 .001 3.516 1.067 .001 Damage 0.674 1.082 .534 0.389 1.073 .717 Domestic violence 4.022 1.126 .000 3.851 1.113 .001 Disorderly conduct 2.583 0.962 .007 2.357 0.953 .014 Drug-related offense 4.034 0.843 .000 3.641 0.838 .000 Alcohol-related offense 3.866 0.986 .000 3.526 0.978 .000 Shoplifting 3.393 0.898 .000 3.229 0.890 .000 Other 2.628 1.042 .012 2.603 1.031 .012 Number additional offenses 1.229 0.229 .000 1.270 0.227 .000 Number prior offenses 0.817 0.139 .000 0.830 0.138 .000 Ethnicity (non-Latinx) Latinx 1.192 0.336 .000 Race (white) American Indian 2.601 0.740 .000 Multiracial –1.412 0.424 .001 African American/black 0.502 0.627 .423 Intercept 16.441 1.607 .000 14.236 1.748 .000 13.524 1.757 .000 R2 0.012 0.112 0.136 ∆R2 0.100 0.024 . Model 1 . . Model 2 . . Model 3 . Predictor . B . SE . p . . B . SE . p . . B . SE . p . Male (female) 0.401 0.335 .232 0.229 0.328 .485 0.210 0.325 .518 Age 0.171 0.107 .109 –0.071 0.107 .509 –0.072 0.106 .497 Income –0.511 0.142 .000 –0.301 0.138 .029 –0.138 0.142 .333 Type of offense (trespassing) Assault 3.374 0.909 .000 3.166 0.901 .000 Burglary 3.718 1.077 .001 3.516 1.067 .001 Damage 0.674 1.082 .534 0.389 1.073 .717 Domestic violence 4.022 1.126 .000 3.851 1.113 .001 Disorderly conduct 2.583 0.962 .007 2.357 0.953 .014 Drug-related offense 4.034 0.843 .000 3.641 0.838 .000 Alcohol-related offense 3.866 0.986 .000 3.526 0.978 .000 Shoplifting 3.393 0.898 .000 3.229 0.890 .000 Other 2.628 1.042 .012 2.603 1.031 .012 Number additional offenses 1.229 0.229 .000 1.270 0.227 .000 Number prior offenses 0.817 0.139 .000 0.830 0.138 .000 Ethnicity (non-Latinx) Latinx 1.192 0.336 .000 Race (white) American Indian 2.601 0.740 .000 Multiracial –1.412 0.424 .001 African American/black 0.502 0.627 .423 Intercept 16.441 1.607 .000 14.236 1.748 .000 13.524 1.757 .000 R2 0.012 0.112 0.136 ∆R2 0.100 0.024 Note: Reference groups for indicator variables are in parentheses. Open in new tab Table 3 Predictors of Number of Hours of Peer-Derived Consequences (N = 1,236) . Model 1 . . Model 2 . . Model 3 . Predictor . B . SE . p . . B . SE . p . . B . SE . p . Male (female) 0.401 0.335 .232 0.229 0.328 .485 0.210 0.325 .518 Age 0.171 0.107 .109 –0.071 0.107 .509 –0.072 0.106 .497 Income –0.511 0.142 .000 –0.301 0.138 .029 –0.138 0.142 .333 Type of offense (trespassing) Assault 3.374 0.909 .000 3.166 0.901 .000 Burglary 3.718 1.077 .001 3.516 1.067 .001 Damage 0.674 1.082 .534 0.389 1.073 .717 Domestic violence 4.022 1.126 .000 3.851 1.113 .001 Disorderly conduct 2.583 0.962 .007 2.357 0.953 .014 Drug-related offense 4.034 0.843 .000 3.641 0.838 .000 Alcohol-related offense 3.866 0.986 .000 3.526 0.978 .000 Shoplifting 3.393 0.898 .000 3.229 0.890 .000 Other 2.628 1.042 .012 2.603 1.031 .012 Number additional offenses 1.229 0.229 .000 1.270 0.227 .000 Number prior offenses 0.817 0.139 .000 0.830 0.138 .000 Ethnicity (non-Latinx) Latinx 1.192 0.336 .000 Race (white) American Indian 2.601 0.740 .000 Multiracial –1.412 0.424 .001 African American/black 0.502 0.627 .423 Intercept 16.441 1.607 .000 14.236 1.748 .000 13.524 1.757 .000 R2 0.012 0.112 0.136 ∆R2 0.100 0.024 . Model 1 . . Model 2 . . Model 3 . Predictor . B . SE . p . . B . SE . p . . B . SE . p . Male (female) 0.401 0.335 .232 0.229 0.328 .485 0.210 0.325 .518 Age 0.171 0.107 .109 –0.071 0.107 .509 –0.072 0.106 .497 Income –0.511 0.142 .000 –0.301 0.138 .029 –0.138 0.142 .333 Type of offense (trespassing) Assault 3.374 0.909 .000 3.166 0.901 .000 Burglary 3.718 1.077 .001 3.516 1.067 .001 Damage 0.674 1.082 .534 0.389 1.073 .717 Domestic violence 4.022 1.126 .000 3.851 1.113 .001 Disorderly conduct 2.583 0.962 .007 2.357 0.953 .014 Drug-related offense 4.034 0.843 .000 3.641 0.838 .000 Alcohol-related offense 3.866 0.986 .000 3.526 0.978 .000 Shoplifting 3.393 0.898 .000 3.229 0.890 .000 Other 2.628 1.042 .012 2.603 1.031 .012 Number additional offenses 1.229 0.229 .000 1.270 0.227 .000 Number prior offenses 0.817 0.139 .000 0.830 0.138 .000 Ethnicity (non-Latinx) Latinx 1.192 0.336 .000 Race (white) American Indian 2.601 0.740 .000 Multiracial –1.412 0.424 .001 African American/black 0.502 0.627 .423 Intercept 16.441 1.607 .000 14.236 1.748 .000 13.524 1.757 .000 R2 0.012 0.112 0.136 ∆R2 0.100 0.024 Note: Reference groups for indicator variables are in parentheses. Open in new tab Results from research question 2 (“After controlling for offense-related variables, age, gender, and family income, are there significant differences in the number of peer-derived consequence hours based on TC respondents’ race?”) are presented in Table 3. The final model indicated that, compared with youths charged with trespassing, those charged with eight out of the nine other offenses received a significantly greater number of hours of consequences. Consequences for respondents charged with criminal damage were not significantly different from those for respondents charged with trespassing. Domestic violence, drug-related charges, burglary, and alcohol-related charges were associated with the greatest additional number of hours. On average, respondents who were charged with domestic violence received an additional 3.9 hours of consequences, those charged with drug-related offenses received an additional 3.6 hours, and those charged with burglary or alcohol-related offenses received an additional 3.5 hours. Being charged with additional offenses and having prior offenses were significantly associated with receiving additional consequence hours. In terms of ethnicity, all else being equal, compared with their non-Latinx counterparts, Latinx youths received an additional 1.2 hours of consequences. Compared with their white counterparts, American Indian youths received an additional 2.6 hours of consequences. On the other hand, compared with their white counterparts, multiracial youths received 1.4 fewer hours of consequences. There was no statistically significant difference in number of consequence hours between African American/black youths and white youths. An examination of the change in R2 (∆R2) values provides insight into research question 3 (“What is the relative impact of respondents’ demographics, offense-related variables, and respondents’ race on severity of consequences received?”). The ∆R2 values denote the relative impact of variables included in each block. In block 1 (demographics other than race), gender, age, and income variables explained 1.2% of the variance in number of hours assigned. In block 2 (offense-related variables), type of offense, prior offenses, and additional charges explained an additional 10.0% of variance. In block 3, the TC respondents’ ethnicity and race explained an additional 2.4% of variance. Discussion Generally, study results suggested the presence of racial disproportionality that negatively affected American Indian and Latinx youths compared with white youths. However, youths identifying as multiracial received fewer peer-derived consequence hours compared with their white counterparts. Finally, although the strongest predictors of number of hours of consequences assigned were offense-related variables, race and ethnicity explained additional variance above and beyond other demographics and offense-related variables. These findings are discussed in the following sections. RRI RRIs for the current study ranged from 0.82 for multiracial youths (suggesting that, compared with white youths, multiracial youths were less likely to receive severe consequences) to 1.42 for American Indian youths (suggesting that, compared with white youths, American Indian youths were more likely to receive severe consequences). This is the first study to examine RRIs in a TC context. RRIs are commonly used to measure disproportionality in the juvenile justice system (OJJDP, 2018a). Given that TC is a juvenile justice diversion program that provides an alternative to traditional processing, RRIs can be used to inform our understanding of the nature of disproportionality in TC programs and, ultimately, to gauge progress of interventions aimed at decreasing any inequity. The current study represents an important first step in demonstrating the utility of RRIs in a TC context. American Indian youths in particular were the most overrepresented group receiving severe consequences, followed by Hispanic youths. Compared with their white counterparts, multiracial and African American/black youths were less likely to receive severe consequences. A comparison of the patterns of RRIs for the Arizona juvenile justice system and the TC sample is useful in providing some context for the RRIs calculated in the current study. However, it is important to note that direct comparisons are not possible given that decision points differ between the juvenile justice system (for example, diversion, referral, detention) and the TC process (peer-derived consequence hours). In other words, although previous studies have presented RRIs for diversion (whether or not youths were diverted from traditional justice system processing), the current study focused on the severity of consequences for a group of youths who were diverted from traditional processing to the TC program. Haight and Jarjoura’s (2016) examination of RRIs in the Arizona juvenile justice system indicated disparities for Latinx, African American/black, and American Indian youths compared with their white counterparts. The pattern of results for the TC RRIs in the current study differ in that African American/black youths were less likely to receive severe consequences compared with their white counterparts. In addition, although the current study found that multiracial youths were less likely to receive severe peer-derived consequences, juvenile justice data do not capture multiracial youths. These results are an important first step in using RRI calculations within TC programs to examine patterns in disproportionality. However, the RRI results must be interpreted with caution given that a major limitation of RRIs is that they do not give an indication of statistical significance (Piquero, 2008). Instead, regression models (such as the hierarchical model used to address research question 2 in the current study) can be used to examine statistical significance. Overall, additional research examining RRIs in other TC samples is critical to understand the nature of disproportionality in peer-derived consequences. Researchers should seek to compare patterns of disproportionality in TC programs to disproportionality in the local juvenile justice system. Relative Impact of Other Demographics, Offense-Related Variables, and Race Current study findings indicated that disparities for Latinx and American Indian youths compared with white youths persisted after controlling for other demographics, type of offense, prior offenses, and additional charges. This extends findings documenting the persistence of racial disproportionality after controlling for other factors in the juvenile justice system both in Arizona (Rodriguez, 2010) and elsewhere (Armstrong & Rodriguez, 2005; Evangelist et al., 2017; Leiber & Fox, 2005; Shook & Goodkind, 2009). The current study is the first to document disproportionality in the TC model. For African American/black youths, compared with their white counterparts, there were no statistically significant differences in the number of hours of consequences received after controlling for other demographics and offense-related variables. This finding differs from previous studies of the Arizona juvenile justice system in which racial disproportionality was documented for African American/black youths relative to their white counterparts (Haight & Jarjoura, 2016; Rodriguez, 2010). According to macro-contextual theories, agency characteristics have the potential to affect racial equity in juvenile punishment (Engen et al., 2002). Therefore, it is possible that the current study results are due to nuances in the context of the TC program. The importance of context in issues of disproportionality is demonstrated by extant research on disproportionality in school discipline. Results of a study on school context and exclusionary discipline suggested that the proportion of black student enrollment in a school increases one’s risk of out-of-school suspension beyond the influence of individual demographics or behavior) (Skiba et al., 2014). Given that demographics of the larger context affect disproportionality, it is possible that the racial makeup of the peer juries played a role in the severity of consequences given to TC participants. Future studies should use multilevel modeling to simultaneously examine the impact of TC respondent demographics and the demographics of peer juries on the severity of peer-derived consequences. Examination of the ∆R2 values between the three regression blocks revealed that the offense-related variables explained the largest amount of variance in number of constructive consequence hours assigned, followed by respondent’s race and ethnicity. Offense-related variables explained approximately four times the amount of variance compared with race. This finding suggests that factors associated with the offense are mostly driving the peer jury’s selection of appropriate consequences. However, any degree of racial bias in a program designed to promote justice is troubling. Consequently, the fact that race and ethnicity explained variance above and beyond these relevant factors indicates the need for intervention. Researchers should consider developing and evaluating racial bias interventions for peer jurors in a TC context. Jury members could receive a racial bias training component as part of regular jury training. Extant research suggests that implicit racial bias is malleable (for a review, see Dasgupta, 2013). Moreover, existing research has shown that implicit bias affects jurors’ decisions in the adult courtroom (Kang et al., 2012) and in assessments of juvenile offenders (Bridges & Steen, 1998). Therefore, an intervention targeting implicit racial bias among TC jurors may serve to decrease disproportionality in peer-derived consequences. Additional research is also needed to examine other factors that are associated with the peer jury decision-making process. Indeed, macro-contextual theories suggest that process influences racial equity (Engen et al., 2002); thus examinations of the decision-making process of the TC peer jury could yield particularly valuable insights (Huizinga, Thornberry, Knight, & Lovegrove, 2007). For instance, qualitative analysis of peer jury deliberation sessions could identify other factors that peer jurors take into consideration when determining peer-derived consequences. In one novel study, researchers observed TC hearing and jury deliberation sessions and compared the percentage of jurors presented with different types of evidence with the percentage of jurors on juries in which each piece of information was discussed during jury deliberation sessions (Greene & Weber, 2008). They found that the average percentage of jurors who were presented with relevant evidence was 55% and the mean percentage of jurors on juries in which the evidence was discussed in deliberation was 29%. When peer jurors were asked the extent to which they believed different types of evidence were important, evidence related to the offense itself (that is, physical injuries, property damage) were rated as most important and age and gender of the respondent were rated as least important. Other considerations included whether the respondent expressed remorse, family difficulties, school difficulties, and peer attorneys’ recommendations. Greene and Weber (2008) did not, however, examine racial and ethnic disparities in their study. Additional studies could extend the current work by examining both the peer jury deliberations and how these decisions translate into disparities for youths of color relative to their white counterparts. Limitations The results of the current study must be understood in light of study limitations. First, the use of secondary administrative data presented both strengths and limitations. The administrative data were comprehensive in that information on each TC hearing from 2012 through 2016 was included and there was very little missing data. However, the secondary administrative data did not include other potentially relevant factors that could have influenced peer jury decisions, such as those found in the study conducted by Greene and Weber (2008) (for example, teen attorney’s recommendations, family difficulties, school difficulties), as well as contextual differences (for example, neighborhood and school factors). Although peer jury deliberation observations could have provided additional information on how the deliberation process contributes to racial and ethnic disparities, observations were not feasible for the current study. In addition, data on the racial and ethnic composition of the juries were not available. Future studies should consider whether peer jury decisions differ based on the racial composition of the jury. The external validity of the current study is limited because the sample was based on a single TC program in Arizona. Study findings may not translate to other TC programs, particularly those in regions with different racial and ethnic demographics or in programs with different program processes. Due to low sample sizes, youths who identified as Asian or Pacific Islander were excluded from the study, which limits generalizability to these youths. Conclusion and Implications The aim of the current study was to examine racial disproportionality in a sample of TC respondents from a TC program in Arizona. Study results indicated that youths who identified as Latinx or American Indian were more likely to receive a severe consequence from the TC peer jury compared with their non-Latinx, white counterparts. On the other hand, compared with their white counterparts, multiracial youths were less likely to receive a severe consequence. For African American/black youths, there was not a statistically significant difference in the number of peer-derived consequence hours compared with white youths after controlling for other demographics and offense-related variables. This study is the first to document racial and ethnic disparities in peer-derived consequences in a TC program. Findings highlight that although TC programs may benefit youths by avoiding involvement in the juvenile justice system, the TC process is not free from racial bias. Additional research and information on DMC in TC programs are needed so that researchers can begin to create interventions and programs to help combat racial bias in TC programs. Indeed, researchers should seek to replicate study findings using samples from other TC programs. Future studies should also seek to expand on the current study by examining the impact of contextual variables such as peer jury demographics on assigned consequences. TC programs vary widely in their processes and consequences (Cotter & Evans, 2018), and additional studies could provide insight into processes that may contribute to disparities. Research examining disproportionality within other types of diversion programs such as family group conferencing or victim–offender mediation is also needed. Toward the grand challenge of ensuring “equal opportunity and justice” (AASWSW, 2018), social workers can play a key role by critically examining youth justice program processes to identify possible disparities. Comparing the current study findings with existing literature is challenging given the lack of research on racial disparities in the TC process. However, comparisons to research examining disparities in the formal juvenile justice system provide some context. Current study findings suggesting disparities for Latinx and American Indian youths compared with white youths are consistent with previous research examining disproportionality in the juvenile justice system in Arizona (Haight & Jarjoura, 2016; Rodriguez, 2010). The finding that African American/black youths did not receive statistically significantly more consequence hours compared with their white counterparts differs from existing juvenile justice research in Arizona, which has documented disparities for African American/black youths (Haight & Jarjoura, 2016; Rodriguez, 2010). Additional research on racial disproportionality in peer-derived consequences across different contexts will allow for a deeper understanding of potential bias when youths are responsible for ensuring justice for their peers. The current study also examined the relative impact of other demographics, offense-related variables, and race. Offense-related variables explained the largest amount of variance in the number of peer-derived consequence hours assigned, suggesting that youth juries do primarily consider context of the offense. Nonetheless, the fact that disparities for Latinx and American Indian youths persisted after controlling for other demographics, type of offense, prior offenses, and additional charges suggests the presence of racial bias within the TC process. Social workers are often involved in advocating for alternatives to juvenile involvement in the justice system. However, current study results caution against making assumptions that alternative programs are unbiased. Instead, social workers working within juvenile justice and diversion agencies can lead efforts to assess racial disproportionality within their programs. RRIs provide a straightforward method for examining disparities. If racial disparities exist, social workers can implement and evaluate interventions targeting racial bias. Interventions addressing implicit bias among peer juries may be effective in decreasing disparities. At the policy level, several states have legislation guiding the operation of TC programs (see Heward, 2006). Social workers can advocate for legislative requirements to examine racial disproportionality in TC consequences, which can be used to strengthen the programs and help ensure an unbiased TC trial and consequences for all respondents. References American Academy of Social Work and Social Welfare . ( 2018 ). Grand challenges for social work. Retrieved from http://grandchallengesforsocialwork.org/grand-&break;challenges-initiative/ Armstrong , G. S. , & Rodriguez , N. ( 2005 ). Effects of individual and contextual characteristics on preadjudication detention of juvenile delinquents . Justice Quarterly, 22 , 521 – 539 . Google Scholar Crossref Search ADS WorldCat Bridges , G. S. , & Steen , S. ( 1998 ). Racial disparities in official assessments of juvenile offenders: Attributional stereotypes as mediating mechanisms . American Sociological Review, 63 , 554 – 570 . Google Scholar Crossref Search ADS WorldCat Butts , J. , Buck , J. , & Coggeshall , M. ( 2002 ). The impact of Teen Court on young offenders . Washington, DC : Urban Institute . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Cabaniss , E. R. , Frabutt , J. M. , Kendrick , M. H. , & Arbuckle , M. B. ( 2007 ). Reducing disproportionate minority contact in the juvenile justice system: Promising practices . Aggression and Violent Behavior, 12 , 393 – 401 . Google Scholar Crossref Search ADS WorldCat Cole , H. A. , & Heilig , J. V. ( 2011 ). Developing a school-based youth court: A potential alternative to the school to prison pipeline . Journal of Law and Education, 40 , 1 – 18 . OpenURL Placeholder Text WorldCat Cotter , K. L. , & Evans , C.B.R. ( 2018 ). A systematic review of teen court evaluation studies: A focus on evaluation design characteristics and program components and processes . Adolescent Research Review, 3 , 425 – 447 . doi: 10.1007/s40894-017-0056-1 Google Scholar Crossref Search ADS WorldCat Crossref Crutchfield , R. D. , Fernandes , A. , & Martinez , J. ( 2010 ). Racial and ethnic disparity and criminal justice: How much is too much . Journal of Criminal Law and Criminology, 100 , 903 – 932 . OpenURL Placeholder Text WorldCat Crutchfield , R. D. , Skinner , M. L. , Haggerty , K. P. , McGlynn , A. , & Catalano , R. F. ( 2009 ). Racial disparities in early criminal justice involvement . Race and Social Problems, 1 , 218 – 230 . Google Scholar Crossref Search ADS WorldCat Dasgupta , N. ( 2013 ). Implicit attitudes and beliefs adapt to situations: A decade of research on the malleability of implicit prejudice, stereotypes, and the self-concept . Advances in Experimental Social Psychology, 47 , 233 – 279 . OpenURL Placeholder Text WorldCat Dawson-Edwards , C. , Tewksbury , R. , & Nelson , N. T. ( 2017 ). The causes and pervasiveness of DMC: Stakeholder perceptions of disproportionate minority contact in the juvenile justice system . Race and Justice . Advance online publication. doi: 10.1177/2153368717735365 Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC Crossref DeFosset , A. R. , Schooley , T. S. , Abrams , L. S. , Kuo , T. , & Gase , L. N. ( 2017 ). Describing theoretical underpinnings in juvenile justice diversion: A case study explicating Teen Court program theory to guide research and practice . Children and Youth Services Review, 73 , 419 – 429 . doi: 10.1016/j.childyouth.2017.01.005 Google Scholar Crossref Search ADS WorldCat Crossref Engen , R. L. , Steen , S. , & Bridges , G. S. ( 2002 ). Racial disparities in the punishment of youth: A theoretical and empirical assessment of the literature . Social Problems, 49 , 194 – 220 . Google Scholar Crossref Search ADS WorldCat Evangelist , M. , Ryan , J. P. , Victor , B. G. , Moore , A. , & Perron , B. E. ( 2017 ). Disparities in adjudication in the juvenile justice system: An examination of race, gender, and age . Social Work Research, 41 , 199 – 211 . Google Scholar Crossref Search ADS WorldCat Gase , L. N. , Schooley , T. , DeFosset , A. , Stoll , M. A. , & Kuo , T. ( 2016 ). The impact of Teen Courts on youth outcomes: A systematic review . Adolescent Research Review, 1 , 51 – 67 . doi: 10.1007/s40894-015-0012-x Google Scholar Crossref Search ADS WorldCat Crossref Global Youth Justice . ( 2018 ). Background and history. Retrieved from https://www.globalyouthjustice.org/about/background/ Greene , E. , & Weber , K. ( 2008 ). Teen Court jurors’ sentencing decisions . Criminal Justice Review, 33 , 361 – 378 . Google Scholar Crossref Search ADS WorldCat Haight , K. , & Jarjoura , G. R. ( 2016 ). An examination of ethnic disparities in Arizona’s juvenile justice system: Final technical report . Indianapolis : American Institutes for Research . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Heward , M. E. ( 2006 ). An update on Teen Court legislation. Retrieved from https://www.ncjrs.gov/pdffiles1/ojjdp/237390.pdf Huizinga , D. , Thornberry , T. , Knight , K. , & Lovegrove , P. ( 2007 ). Disproportionate minority contact in the juvenile justice system: A study of differential minority arrest/referral to court in three cities . Washington, DC : U.S. Department of Justice . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Juvenile Justice and Delinquency Prevention Act, 42 U.S.C. § 5601 ( 2002 ). Kang , J. , Bennett , M. , Carbado , D. , Casey , P. , Dasgupta , N. , Faigman , D. , et al. ( 2012 ). Implicit bias in the courtroom . UCLA Law Review, 59 , 1124 – 1186 . OpenURL Placeholder Text WorldCat Kempf-Leonard , K. ( 2007 ). Minority youths and juvenile justice: Disproportionate minority contact after nearly 20 years of reform efforts . Youth Violence and Juvenile Justice, 5 , 71 – 87 . Google Scholar Crossref Search ADS WorldCat Laub , J. H. ( 2014 ). Understanding inequality and the justice system response: Charting a new way forward [ William T. Grant Foundation Inequality Paper ]. New York : William T. Grant Foundation . Leiber , M. J. ( 2002 ). Disproportionate minority confinement (DMC) of youth: An analysis of state and federal efforts to address the issue . Crime & Delinquency, 48 , 3 – 45 . Google Scholar Crossref Search ADS WorldCat Leiber , M. J. , & Fox , K. C. ( 2005 ). Race and the impact of detention on juvenile justice decision making . Crime & Delinquency, 51 , 470 – 497 . Google Scholar Crossref Search ADS WorldCat Leiber , M. , & Rodriguez , N. ( 2011 ). The implementation of the disproportionate minority confinement/contact (DMC) mandate: A failure or success? Race and Justice, 1 , 103 – 124 . Google Scholar Crossref Search ADS WorldCat Office of Juvenile Justice and Delinquency Prevention. ( 2012, November ). Disproportionate minority contact . OJJDP in Focus. Retrieved from https://www.ojjdp.gov/pubs/239457.pdf OpenURL Placeholder Text WorldCat Office of Juvenile Justice and Delinquency Prevention . ( 2018a ). National disproportionate minority contact databook. Retrieved from https://www.ojjdp.gov/ojstatbb/dmcdb/asp/whatis.asp Office of Juvenile Justice and Delinquency Prevention . ( 2018b ). Program summary: Disproportionate minority contact. Retrieved from https://www.ojjdp.gov/programs/ProgSummary.asp?pi=18 Piquero , A. R. ( 2008 ). Disproportionate minority contact . Future of Children, 18 , 59 – 79 . Google Scholar Crossref Search ADS WorldCat Puzzanchera , C. , & Hockenberry , S. ( 2017 ). National disproportionate minority contact databook. Office of Juvenile Justice and Delinquency Prevention . Retrieved from http://www.ojjdp.gov/ojstatbb/dmcdb/ OpenURL Placeholder Text WorldCat Rasmussen , A. , & Diener , C. L. ( 2005 ). A prospective longitudinal study of Teen Court’s impact on offending youths’ behavior . Juvenile and Family Court Journal, 56 ( 1 ), 17 – 32 . Google Scholar Crossref Search ADS WorldCat Ray , K.E.B. , & Alarid , L. F. ( 2004 ). Examining racial disparity of male property offenders in the Missouri juvenile justice system . Youth Violence and Juvenile Justice, 2 , 107 – 128 . Google Scholar Crossref Search ADS WorldCat Rodriguez , N. ( 2010 ). The cumulative effect of race and ethnicity in juvenile court outcomes and why preadjudication matters . Journal of Research in Crime and Delinquency, 47 , 391 – 413 . Google Scholar Crossref Search ADS WorldCat Sampson , R. J. , & Laub , J. H. ( 1993 ). Structural variations in juvenile court processing: Inequality, the underclass, and social control . Law & Society Review, 27 , 285 – 312 . Google Scholar Crossref Search ADS WorldCat Sampson , R. J. , & Laub , J. H. ( 1997 ). A life-course theory of cumulative disadvantage and stability of delinquency. In T. Thornberry (Ed.), Advances in criminological theory (pp. 138 – 162 ). New Brunswick, NJ : Transaction Publishers . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Schwalbe , C. S. , Gearing , R. E. , MacKenzie , M. J. , Brewer , K. B. , & Ibrahim , R. ( 2012 ). A meta-analysis of experimental studies of diversion programs for juvenile offenders . Clinical Psychology Review, 32 , 26 – 33 . Google Scholar Crossref Search ADS WorldCat Shook , J. J. , & Goodkind , S. A. ( 2009 ). Racial disproportionality in juvenile justice: The interaction of race and geography in pretrial detention for violent and serious offenses . Race and Social Problems, 1 , 257 – 266 . Google Scholar Crossref Search ADS WorldCat Skiba , R. , Chung , C.-G. , Trachok , M. , Baker , T. L. , Sheya , A. , & Hughes , R. L. ( 2014 ). Parsing disciplinary disproportionality: Contributions of infraction, student, and school characteristics to out-of-school suspension and expulsion . American Educational Research Journal, 51 , 640 – 670 . doi: 10.3102/0002831214541670 Google Scholar Crossref Search ADS WorldCat Crossref Smokowski , P. R. , Rose , R. A. , Evans , C.B.R. , Barbee , J. , Cotter , K. L. , & Bower , M. ( 2017 ). The impact of Teen Court on rural adolescents: Improved social relationships, psychological functioning, and school experiences . Journal of Primary Prevention, 38 , 447 – 464 . doi: 10.1007/s10935-017-0470-y Google Scholar Crossref Search ADS WorldCat Crossref U.S. Census Bureau . ( 2017 ). United States Census Bureau QuickFacts. Retrieved from https://www.census.gov/quickfacts/fact/table/US/PST045217 © 2019 National Association of Social Workers This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Brisson,, Daniel;Lechuga Peña,, Stephanie;Mattocks,, Nicole;Plassmeyer,, Mark;McCune,, Sarah

doi: 10.1093/swr/svz020pmid: N/A

Abstract The objective of this study was to ascertain whether participation in the Your Family, Your Neighborhood (YFYN) intervention, an intervention for families living in low-income neighborhoods, leads to improved perceptions of neighborhood social cohesion and informal neighborhood social control. Fifty-two families in three low-income, urban neighborhoods participated in the manualized YFYN intervention. In this quasi-experimental study treatment families (n = 37) in two low-income neighborhoods received YFYN and control families (n = 15) from one separate low-income neighborhood did not. Families receiving YFYN attended 10 two-hour skills-based curriculum sessions during which they gathered for a community dinner and participated in parent- and child-specific skills-based groups. Treatment families reported increases in both neighborhood social cohesion and informal neighborhood social control after receiving YFYN. However, families receiving YFYN did not experience statistically significant improvements in perceptions of neighborhood social cohesion or informal neighborhood social control compared with nontreatment families. In conclusion, the delivery of YFYN in low-income neighborhoods may improve perceptions of neighborhood social cohesion. Further testing, with randomization and a larger sample, should be conducted to provide a more robust understanding of the impact of YFYN. Individual poverty and living in a low-income neighborhood erect major barriers to the health and well-being of families (Jargowsky, 2013). Forty-seven million people in the United States live in poverty (U.S. Census Bureau, 2016), and these individuals and families struggle to find housing in thriving neighborhoods (Aurand et al., 2018). Only about 25% of those eligible for a housing subsidy receive one (Fischer & Sard, 2016). A housing subsidy adds real income to a low-income family’s household budget; however, even with a housing subsidy many individuals and families struggling with poverty are clustered in communities of concentrated disadvantage (Jargowsky, 2013). Estimates suggest that over 50% of low-income families live in a neighborhood of concentrated disadvantage (Jargowsky, 2013). Living in a neighborhood of concentrated disadvantage has been linked to a number of detrimental outcomes (Brooks-Gunn, Duncan, & Aber, 1997) including a lower life expectancy (Swain, 2016), a greater risk of attending poorly performing schools (Evans, 2004), and an increased exposure to violence (Morenoff, Sampson, & Raudenbush, 2001). Figure 1 Open in new tabDownload slide Neighborhood Mediators in the Your Family, Your Neighborhood Conceptual Framework Figure 1 Open in new tabDownload slide Neighborhood Mediators in the Your Family, Your Neighborhood Conceptual Framework There are a number of interventions (Aha Process Inc., 2017) and benefit programs, such as food stamps and housing subsidies, that are designed to assist families living in poverty. These interventions and programs typically address a single poverty-related challenge but do not tackle the complex, ecological barriers families encounter while living in a low-income neighborhood (Berger & Font, 2015). In contrast, the Your Family, Your Neighborhood (YFYN) intervention uses a prevention-oriented approach to address individual, family, school, and neighborhood barriers encountered by families living in neighborhoods of concentrated disadvantage. YFYN is a dual-generation, manualized intervention delivered in participant families’ neighborhoods over the course of 10 weekly sessions. The skills-based curriculum focuses on fostering family health and well-being, strengthening the parent–child bond, promoting parental attachment to schools, improving children’s academic outcomes, and establishing cohesive community relationships. This article offers the results of a quasi-experimental study, which tested the effects of YFYN on perceptions of neighborhood social cohesion and informal neighborhood social control. Perceptions of neighborhood social cohesion and informal neighborhood social control have strong support in the extant literature as key mediating mechanisms for health and well-being. YFYN, with its community-based delivery model and focus on relationships in a neighborhood cohort setting, has an explicit focus on improving perceptions of neighborhood social cohesion and informal neighborhood social control. Concentrated Disadvantage as an Ecological Barrier Neighborhoods of concentrated disadvantage are complex ecological contexts that present many community- and family-level challenges. Challenges in these ecological contexts include crime (Morenoff et al., 2001), poorly performing schools (Evans, 2004; Fram, Miller-Cribbs, & Van Horn, 2007), and poor health (Marmot & Wilkinson, 2006). Other challenges include inadequately nutritious food options (Raja, Ma, & Yadav, 2008), limited health care access (Kirby & Kaneda, 2005), and compromised employment opportunities (Elliott, 1999). Neighborhoods of concentrated disadvantage are often associated with lower levels of social cohesion, community trust, and informal social control (that is, the belief that one has power or control over their environment) (Brisson & Walker, 2019). Ecological systems theory provides a lens through which to view the challenges that low-income families living in neighborhoods of concentrated disadvantage encounter and the interventions that may address them (Bronfenbrenner, 1979). The theory suggests that individual outcomes result from interactions across ecological systems, including individual, family, connecting, organizational, and institutional systems (Bronfenbrenner, 1986). Considering the influence of ecological systems, interventions designed to address poverty within low-income neighborhoods must look beyond individual and family-level challenges to school, neighborhood, and institutional systems. Despite the many challenges they face, families have the power to build relationships (that is, social cohesion) and take action (that is, informal social control), which can be important mediators to influence the ecological context of a neighborhood (Sampson, Raudenbush, & Earls, 1997). Building on these theories, Figure 1 is a conceptual framework that shows how the YFYN intervention positions neighborhood social cohesion and informal neighborhood social control as key mediators for family health and well-being. Interventions That Use an Ecological Framework Social workers, community psychologists, prevention scientists, public health professionals, and other actors in similar fields have worked for many decades to identify and implement practice approaches that provide low-income families living in neighborhoods of concentrated disadvantage with the resources and support necessary to address the community- and family-level challenges described earlier (Bryan & Davis, 1990). These interventions target the systemic factors that influence individual-level outcomes (Bronfenbrenner, 1979; McLeroy, Bibeau, Steckler, & Glanz, 1988). The extant literature is replete with evidence from rigorous evaluations of community-level, public health interventions designed to prevent negative health and social outcomes, such as HIV/AIDS (Kalichman et al., 2014; Martinez, Roth, Kelle, Downs, & Rhodes, 2014; Woelk et al.,2016), obesity (Cheadle, Rauzon, & Schwartz, 2014), diabetes (Ahmad & Tsang, 2013; Albright & Gregg, 2013), incidents of violence (Morrel-Samuels, Zimmerman, & Reischl, 2013), and drug use (Rhew, Brown, Hawkins, & Briney, 2013). Government agencies, foundations, and other funding entities have invested substantially in the development, implementation, and evaluation of large-scale, community-level initiatives. Several of these investments have been in initiatives designed to prevent obesity (Cheadle et al., 2014). One such initiative, Shape Up Somerville, used a participatory approach to influence all aspects of the community, from educating parents, medical professionals, and local restaurants, to revising after-school program curricula and expanding local, environmental policies (Economos et al., 2007). The initiative was successful in significantly reducing the body mass index of children in first, second, and third grade, one year following program implementation (Cheadle et al., 2014). Kaiser Permanente’s Healthy Eating, Active, Living–Community Health Initiative (also known as HEAL-CHI) emphasized policy and environmental changes for preventing obesity in three neighborhoods in northern California. Policy changes were made at the organizational level to increase healthy food and physical activity in schools and at worksites. Changes were also made to urban-planning policies to improve the built environment for physical activity promotion. This initiative was effective in significantly increasing daily physical activity from 61% to 67% for youths in four of the nine “high-dose” strategy sites, where a combination of after-school programming, physical activities, and a new physical education curriculum were implemented (Cheadle et al., 2014). The Communities That Care (CTC) program uses a comprehensive approach to mobilize communities to transform their prevention systems to reduce youth problem behaviors such as drug use and delinquent actions (Rhew et al., 2013). CTC provides consultation, training, and evidence-based tools to all stakeholders invested in promoting positive youth development in their community (Hawkins & Catalano, 2005). To ensure that efforts are locally relevant and sustainable, all new programs are developed collaboratively by community stakeholders, and are implemented with support from the CTC coalition (Hawkins et al., 2009). Evidence from the Community Youth Development Study, a randomized trial of CTC, demonstrated a significant reduction in incidences of drug use and delinquent behavior among a panel of 4,407 students in grades 5 through 8 across seven states (Hawkins et al., 2009). Another CTC study evaluated the program’s effectiveness at changing community norms around youths’ drug use and antisocial behavior by surveying 928 community leaders across 24 towns, half of which were assigned to the intervention and half to the control condition. Findings revealed that 1.5 years after funding ended CTC communities reported higher levels of adoption of evidence-based approaches to prevention and more significant increases in community norms opposing adolescent drug use, compared with control communities (Rhew et al., 2013). Another comprehensive community-level intervention that has proven to be effective is the Michigan Youth Violence Prevention Center (MI-YVPC). MI-YVPC, located in Flint, Michigan, offers six programs designed to strengthen relationships, increase social cohesion, and improve the physical aspects of the environment. Through these programs, adolescents are connected to positive role models and community engagement activities to reduce injury and assault among youths (Morrel-Samuels et al., 2013). A recent evaluation of MI-YVPC compared counts of reported assault offenses and injuries between the intervention area and a matched comparison area over the six years prior to and 30 months post-intervention (Heinze et al., 2015). Findings revealed that the overall numbers of assaults and assault injuries per month were lower in the intervention area post-intervention, suggesting that MI-YVPC is successful in reducing youth violence (Heinze et al., 2015). The YFYN intervention is also an intervention that builds on social ecological theory. YFYN is different than other community interventions as it has an explicit focus on the key mediators of neighborhood social cohesion and informal neighborhood social control to improve family well-being outcomes. Neighborhood social cohesion is the shared trust, common values and connections in a neighborhood, and informal social control is the feeling that one can take action to affect the neighborhood. YFYN follows prevention strategies identified by Nation et al. (2003), such as using varied teaching methods, following a guiding theory, and providing opportunities for positive relationships. Despite the evidence showing the important role of these two neighborhood variables (Sampson, 2001), there are, to our knowledge, no community-based interventions that have improved both neighborhood social cohesion and informal neighborhood social control. Summary Neighborhood concentrated disadvantage is a potential obstacle to family health and well-being. Although some community-level interventions have demonstrated positive effects using an ecological lens, more work can be done. This study tests how an innovative intervention, YFYN, affects the development of neighborhood social cohesion and informal neighborhood social control, two key social processes that have proven to mediate the negative impact of concentrated disadvantage (Sampson, 2001). The following research questions are addressed: (a) Does YFYN affect perceptions of neighborhood social cohesion? (b) Does YFYN affect perceptions of informal neighborhood socialcontrol? Method The YFYN intervention study was conducted in three low-income neighborhoods with a high percentage of affordable housing in a city in the western United States. Families in two neighborhoods received the YFYN intervention, and families in one neighborhood did not. The neighborhood where families did not receive YFYN was purposefully chosen because of its similarities to intervention neighborhoods. All neighborhoods had a high percentage of affordable housing, and the nontreatment neighborhood is adjacent to one of the intervention neighborhoods—but separated by a major boulevard and public park. YFYN was delivered at community-based sites immediately after school or in the evenings to nine treatment cohorts consisting of three to 10 families each. Families were eligible for YFYN if they resided in the neighborhood where the intervention was delivered and if they had at least one child between the ages of seven and 12. Families were recruited for the intervention by YFYN staff and staff at the community-based sites where the intervention was being delivered. A range of community-based sites hosted YFYN and included two schools, one community center, and one subsidized housing complex. For two of the nine treatment cohorts YFYN was delivered in Spanish to monolingual Spanish-speaking families. The YFYN intervention consists of 10 two-hour curriculum sessions. Curriculum sessions that were delivered after school began with a snack. Following the snack, parents met in a parent-only group, and children met in a child-only group for 45 minutes. After the parent and child groups, the families shared a group meal provided by YFYN, and parents and children gathered to share what they had learned during the session. Each group was led by a trained facilitator. Substantive curriculum content for the parent and child groups were matched. For example, in one session, the parent group discussion is focused on the parents’ hopes and dreams for their child’s education, and the child group activity is focused on the children’s hopes and dreams for their own education. The following content topics were included in the 10-week curriculum: week 1: introductions and the parent--child bond; week 2: your family and education; week 3: emotional communication; week 4: systemic oppression in education; week 5: promoting positive youth development; week 6: promoting leadership in my family, school, and community; week 7: neighborhood strengths; week 8: planning a community celebration; week 9: planning a community celebration; and week 10: celebrating your family and your neighborhood. The structure and the content of the YFYN curriculum are designed to increase opportunities for neighborhood engagement and facilitate critical neighborhood conversations. Through increased engagement and critical conversations, participation in YFYN leads to a common understanding of neighborhood issues and shared neighborhood values and norms. The structure of the YFYN curriculum offers two distinct opportunities for the development of shared neighborhood norms and values. First, the common meal provides an informal opportunity for families to meet and build relationships. Sharing a meal is a commonly understood community bonding activity. The facilitated parent group offers a second structural opportunity to develop shared understanding of neighborhood issues and values. The facilitated parent group sessions ask parents to think critically about themselves, their children, their schools, their health, and their neighborhood. These critical discussions are designed to bond neighborhood parents. In addition to the curriculum structure, the curriculum content—specifically the final four curriculum sessions—is designed to engage parents and their children in a critical dialogue about their neighborhood. In the final four facilitated group sessions, parents and their children are asked to reflect and share about the issues they feel are important in the neighborhood. Parents and children are given the opportunity and informal support to take a leadership role on issues in the neighborhood. Then, the YFYN group is asked to plan a neighborhood celebration. Planning and carrying out the neighborhood celebration provides families with skill building on neighborhood planning and an action step of engaging in a dialogue with neighbors to achieve a positive outcome (the neighborhood celebration). Study Design The study used a quasi-experimental design to compare changes in treatment families’ perceptions of neighborhood social cohesion and informal neighborhood control from pre- to posttest with changes in control families’ perceptions from pre- to posttest. The YFYN survey consists of previously validated items and takes 30 minutes to complete. For treatment families the survey was administered before the first YFYN session and at the conclusion of the 10th and final YFYN session. For nontreatment families the pretest survey was administered at community events; families were then contacted approximately 10 weeks later to complete the posttest. Families received a $50 gift card after completing the posttest. Figure 2 is a decision tree for inclusion in the study. Figure 2 Open in new tabDownload slide Your Family, Your Neighborhood Participant Flowchart in the Quasi-Experimental Study Figure 2 Open in new tabDownload slide Your Family, Your Neighborhood Participant Flowchart in the Quasi-Experimental Study Sample A total of 52 families were included in the study: 37 families completed the YFYN intervention and 15 families were in the nontreatment group. Recruitment of nontreatment group families for the study proved difficult and is the reason for differences in sample sizes between the two groups. Table 1 provides descriptive statistics for the full sample, the treatment group, and the comparison group. As indicated in Table 1, almost all the respondents were women who had lived in their respective neighborhoods for an average of 5.1 years. Respondents were on average 36 years old, and the average age of their participating child was 8.8 years. Seventy-four percent of respondents were born in the United States, and respondents represent various races and ethnicities. Forty-one percent reported cohabiting with a partner. Descriptive statistics for the treatment and comparison groups showed that the groups were comparable, and bivariate analyses indicated no statistical differences between groups. (P values for differences between cohabitating partners were assessed using Yates continuity correction to address the small sample size. The p value for differences between the treatment and control group was .08.) Table 1 Sample Descriptive Statistics: Means and Standard Deviations or Percentages Full Sample (N = 52) Treatment Group (n = 37) Comparison Group (n = 15) Characteristic M (SD) % M (SD) % M (SD) % Participant age 36.0 (8.5) 34.9 (8.0) 38.9 (9.5) Participant gender (female) 96 97 93 Years in current residence 5.1 (6.3) 4.7 (7.3) 5.7 (4.1) Born in the United States (yes) 74 69 87 White 14 11 20 Black 22 22 20 Latino 45 47 40 Education completed (high school education or more) 57 56 60 Cohabiting with partner (yes) 41 52 20 Child’s age (years) 8.8 (2.2) 8.6 (2.2) 9.3 (2.1) Child’s gender (female) 54 57 46 Neighborhood social cohesion (pre) 2.44 (0.72) 2.5 (0.68) 2.31 (0.82) Neighborhood social cohesion (post) 2.60 (0.68) 2.69 (0.66) 2.39 (0.71) Informal neighborhood social control (pre) 2.72 (1.17) 2.82 (1.27) 2.48 (0.90) Informal neighborhood social control (post) 2.97 (1.26) 3.09 (1.31) 2.61 (1.16) Full Sample (N = 52) Treatment Group (n = 37) Comparison Group (n = 15) Characteristic M (SD) % M (SD) % M (SD) % Participant age 36.0 (8.5) 34.9 (8.0) 38.9 (9.5) Participant gender (female) 96 97 93 Years in current residence 5.1 (6.3) 4.7 (7.3) 5.7 (4.1) Born in the United States (yes) 74 69 87 White 14 11 20 Black 22 22 20 Latino 45 47 40 Education completed (high school education or more) 57 56 60 Cohabiting with partner (yes) 41 52 20 Child’s age (years) 8.8 (2.2) 8.6 (2.2) 9.3 (2.1) Child’s gender (female) 54 57 46 Neighborhood social cohesion (pre) 2.44 (0.72) 2.5 (0.68) 2.31 (0.82) Neighborhood social cohesion (post) 2.60 (0.68) 2.69 (0.66) 2.39 (0.71) Informal neighborhood social control (pre) 2.72 (1.17) 2.82 (1.27) 2.48 (0.90) Informal neighborhood social control (post) 2.97 (1.26) 3.09 (1.31) 2.61 (1.16) Open in new tab Table 1 Sample Descriptive Statistics: Means and Standard Deviations or Percentages Full Sample (N = 52) Treatment Group (n = 37) Comparison Group (n = 15) Characteristic M (SD) % M (SD) % M (SD) % Participant age 36.0 (8.5) 34.9 (8.0) 38.9 (9.5) Participant gender (female) 96 97 93 Years in current residence 5.1 (6.3) 4.7 (7.3) 5.7 (4.1) Born in the United States (yes) 74 69 87 White 14 11 20 Black 22 22 20 Latino 45 47 40 Education completed (high school education or more) 57 56 60 Cohabiting with partner (yes) 41 52 20 Child’s age (years) 8.8 (2.2) 8.6 (2.2) 9.3 (2.1) Child’s gender (female) 54 57 46 Neighborhood social cohesion (pre) 2.44 (0.72) 2.5 (0.68) 2.31 (0.82) Neighborhood social cohesion (post) 2.60 (0.68) 2.69 (0.66) 2.39 (0.71) Informal neighborhood social control (pre) 2.72 (1.17) 2.82 (1.27) 2.48 (0.90) Informal neighborhood social control (post) 2.97 (1.26) 3.09 (1.31) 2.61 (1.16) Full Sample (N = 52) Treatment Group (n = 37) Comparison Group (n = 15) Characteristic M (SD) % M (SD) % M (SD) % Participant age 36.0 (8.5) 34.9 (8.0) 38.9 (9.5) Participant gender (female) 96 97 93 Years in current residence 5.1 (6.3) 4.7 (7.3) 5.7 (4.1) Born in the United States (yes) 74 69 87 White 14 11 20 Black 22 22 20 Latino 45 47 40 Education completed (high school education or more) 57 56 60 Cohabiting with partner (yes) 41 52 20 Child’s age (years) 8.8 (2.2) 8.6 (2.2) 9.3 (2.1) Child’s gender (female) 54 57 46 Neighborhood social cohesion (pre) 2.44 (0.72) 2.5 (0.68) 2.31 (0.82) Neighborhood social cohesion (post) 2.60 (0.68) 2.69 (0.66) 2.39 (0.71) Informal neighborhood social control (pre) 2.72 (1.17) 2.82 (1.27) 2.48 (0.90) Informal neighborhood social control (post) 2.97 (1.26) 3.09 (1.31) 2.61 (1.16) Open in new tab Table 2 Assessment of Your Family, Your Neighborhood Treatment Effects on Neighborhood Social Cohesion and Informal Neighborhood Social Control t Test Regression of Posttest Scores Pre--Posttest Analysis of Variance of Changes Controlling for Pretest M Difference M Change M Change Unstandardized B (SE), Outcome Variable t Value Treatment Control F Standardized B Neighborhood social cohesion 0.18*, 2.09 0.18 0.08 0.36 0.16 (0.15), 0.11 Informal neighborhood social control 0.26, 1.5 0.26 0.13 0.16 0.24 (0.31), 0.09 t Test Regression of Posttest Scores Pre--Posttest Analysis of Variance of Changes Controlling for Pretest M Difference M Change M Change Unstandardized B (SE), Outcome Variable t Value Treatment Control F Standardized B Neighborhood social cohesion 0.18*, 2.09 0.18 0.08 0.36 0.16 (0.15), 0.11 Informal neighborhood social control 0.26, 1.5 0.26 0.13 0.16 0.24 (0.31), 0.09 *p < .05. Open in new tab Table 2 Assessment of Your Family, Your Neighborhood Treatment Effects on Neighborhood Social Cohesion and Informal Neighborhood Social Control t Test Regression of Posttest Scores Pre--Posttest Analysis of Variance of Changes Controlling for Pretest M Difference M Change M Change Unstandardized B (SE), Outcome Variable t Value Treatment Control F Standardized B Neighborhood social cohesion 0.18*, 2.09 0.18 0.08 0.36 0.16 (0.15), 0.11 Informal neighborhood social control 0.26, 1.5 0.26 0.13 0.16 0.24 (0.31), 0.09 t Test Regression of Posttest Scores Pre--Posttest Analysis of Variance of Changes Controlling for Pretest M Difference M Change M Change Unstandardized B (SE), Outcome Variable t Value Treatment Control F Standardized B Neighborhood social cohesion 0.18*, 2.09 0.18 0.08 0.36 0.16 (0.15), 0.11 Informal neighborhood social control 0.26, 1.5 0.26 0.13 0.16 0.24 (0.31), 0.09 *p < .05. Open in new tab Figure 3 Open in new tabDownload slide Comparing Change Scores of Neighborhood Social Cohesion Control for Participants Receiving Your Family, Your Neighborhood Intervention and Participants in a Control Condition Figure 3 Open in new tabDownload slide Comparing Change Scores of Neighborhood Social Cohesion Control for Participants Receiving Your Family, Your Neighborhood Intervention and Participants in a Control Condition Figure 4 Open in new tabDownload slide Comparing Change Scores of Informal Neighborhood Social Control for Participants Receiving Your Family, Your Neighborhood Intervention and Participants in a Control Condition Figure 4 Open in new tabDownload slide Comparing Change Scores of Informal Neighborhood Social Control for Participants Receiving Your Family, Your Neighborhood Intervention and Participants in a Control Condition Measures Perceptions of neighborhood social cohesion and informal neighborhood social control are the study outcomes. Measures for both outcomes were taken from the Project on Human Development in Chicago Neighborhoods, and they are considered standard measures for these concepts (Sampson et al., 1997). The neighborhood social cohesion measure consists of four items: (1) “This neighborhood is a good place to raise children”; (2) “People around here are willing to help neighbors”; (3) “This is a close-knit neighborhood”; and (4) “People in this neighborhood can be trusted.” These items are measured using a Likert-type scale, and response options range from 1 = strongly disagree to 4 = strongly agree. The informal neighborhood social control measure consists of five items: (1) “How likely is it that your neighbors would do something about children who were skipping school and hanging out on a street corner?” (2) “How likely is it that your neighbors would do something about children who were spray painting graffiti on a local building?” (3) “How likely is that your neighbors would do something about children who were showing disrespect to an adult?” (4) “How likely is it that your neighbors would do something about a fight that broke out in front of their house?” and (5) “How likely is it that your neighbors would do something if the fire station closest to their house was threatened with budget cuts?” These items are assessed using a Likert-type scale, with the following response options: 1 = very unlikely; 2 = somewhat unlikely; 3 = a 50–50 chance; 4 = somewhat likely; and 5 = very likely. The informal neighborhood social control score is the average of the five items. Descriptive statistics for perceived neighborhood social cohesion and informal neighborhood social control can be found in Table 1. As Table 1 illustrates, the mean neighborhood social cohesion score at pretest was 2.44 (SD = 0.72) and the mean informal neighborhood social control score at pretest was 2.72 (SD = 1.17). Neighborhood social cohesion and informal neighborhood social control scores increased from pre- to posttest for both treatment and control groups. Analysis T tests, using Levene’s unequal variances test, were used to assess changes in neighborhood social cohesion and informal neighborhood social control scores from pre- to posttest for the treatment group. Change scores were then created for the treatment and control groups’ scores on the outcome measures. Analysis of variance (ANOVA) was used to assess the differences in change scores between treatment and control groups. Then, holding pretest scores constant, regression analysis was used to assess differences in the treatment and control groups’ posttest scores. Results Table 2 outlines the full set of results. T tests of pre–post neighborhood social cohesion scores for treatment families were statistically significant: t(37) = 2.09, p < .05. Families reported a 0.18 point increase in perceptions of neighborhood social cohesion from pretest to posttest. The results of the ANOVA, which compared differences in neighborhood social cohesion change scores for treatment and control families were not significant. Although not significant, changes in the neighborhood social cohesion score were greater for treatment families (.18) than for control families (.08). Figure 3 depicts the change in neighborhood social cohesion for treatment and control families. The results of posttest neighborhood social cohesion scores using regression analysis did not indicate a statistically significant difference between treatment and control families. However, regression analysis parameter estimates trend in the direction of YFYN treatment families having higher posttest neighborhood social cohesion scores compared with control families while holding pretest scores constant. Results did not provide evidence that participation in YFYN led to improved perceptions of informal neighborhood social control. Table 2 outlines the full set of results for this outcome. The results of each statistical test indicate that informal neighborhood social control improved after completing YFYN. However, none of the results were statistically significant. Figure 4 depicts the increased change in informal neighborhood social control for treatment families compared with comparison families. Discussion The study is an important test of an innovative intervention designed to build neighborhood social cohesion and informal neighborhood social control. Neighborhood social cohesion and informal neighborhood social control are critical mediators for individual and family health and well-being. We used three different models to test the impact of YFYN on each of these two neighborhood mechanisms. Results of the impact of YFYN on neighborhood social cohesion are mixed. Results fail to show that YFYN has a significant impact on perceptions of informal neighborhood social control. Despite the mixed results, the findings are important as there are no tested interventions available that have been able to show an improvement on perceived neighborhood social cohesion. Neighborhood social cohesion is a key program component in many community-based practice approaches, and it has been shown to mediate crime, safety, health, and economic well-being. The availability of a 10-week manualized intervention that can potentially build neighborhood social cohesion will be of great interest to social work and public health practitioners working in low-income neighborhoods. However, considering the mixed results YFYN should be tested in other practice settings using rigorous research designs. YFYN was not associated with improved informal neighborhood social control, an important community-level mediator, particularly for neighborhood safety. Although the differences in informal neighborhood social control from pre- to posttest for treatment families were larger than differences in neighborhood social cohesion, the differences were not statistically significant. Comparisons between the treatment and control groups revealed that treatment families experienced larger gains in informal neighborhood social control. However, these gains were not statistically significant. Further testing of YFYN’s relationship to informal social control with a larger sample is necessary. Limitations The major limitation of this study is statistical power. One potential explanation for the lack of statistical significance on key variables is the relatively small sample size. It is likely that the sample size was not adequate to reveal all potentially statistically significant relationships. Effect size estimates from this study provide the information necessary to conduct a power analysis for future YFYN research designs. Power analysis should be used in future tests of YFYN to ensure that an adequate sample is available to detect statistically significant relationships between participation in YFYN and perceptions of neighborhood social cohesion and informal neighborhood social control. The sample size also prevented the researchers from assessing implementation effects (for example, whether main effects differed by partner site, race or ethnicity, or participants’ spoken language) and from testing dosage, as measured by the number of sessions that families attended. Despite the modest sample size, the researchers were able to detect some treatment effects. Statistical modeling was also limited to bivariate analysis due to the small sample. One regression model was run but did not include covariates. Future research on YFYN should include larger samples that can account for covarying conditions such as race and ethnicity, age, years in the neighborhood, and others. The quasi-experimental design is another limitation. The study used a control group composed of families that are similar to treatment families on key characteristics (see Table 1). However, selection bias cannot be eliminated as a possible explanation for treatment effects because families were not randomly assigned to treatment and control groups. Future testing of YFYN should prioritize an experimental design that relies on probability sampling in the treatment condition. Also, at this early stage of testing, there is no way of knowing which elements of the intervention can be attributed to changes in neighborhood social cohesion from pretest to posttest. The curriculum is the main element of the intervention. However, a shared meal with family and community members, respite for parents away from their children during the adult group, and skill building in a community setting at the end of each session are also aspects of the intervention that may increase perceptions of neighborhood social cohesion. Future testing of YFYN should attempt to discern which aspects of the intervention contribute to changes in perceived neighborhood social cohesion. Implications for Practice, Policy, and Research This study has several important implications for social work practice, policy, and research. Two practice-oriented implications are particularly salient. Stakeholders engaged with YFYN from health care, early childhood, housing, child welfare, and school sectors understand the positive mediating effects that neighborhood social cohesion has on family outcomes. They have, however, required guidance on how to cultivate this important resource. Practitioners in low-income neighborhoods around the country are working to build social cohesion in low-income neighborhoods. These practitioners will benefit from the availability of a manualized intervention that is proven to improve neighborhood social cohesion. As a curriculum-based practice intervention, YFYN is potentially an important asset for practitioners who mobilize community resources to improve the well-being of families living in low-income neighborhoods. Moreover, many interventions administered in low-income neighborhoods only address a single issue. For example, an intervention may only focus on school readiness, early childhood health, bullying, or family attachment. Although these interventions are valuable, the single-issue approach fails to directly address relationship building across the neighborhood that can lead to sustainable solutionsto for families navigating the complex community- and family-level challenges that often characterize low-income neighborhoods. Results from this study demonstrate the importance of taking an ecological approach to prevention intervention. Indeed, this approach seems particularly important when practicing in low-income neighborhoods, where challenges are often complex and involve multiple systems. Results from the study also have important implications for policy. YFYN was developed from an assessment that affordable housing communities often lack family-level services. Policy largely drives affordable housing practices, and current policies focus on mixed-income housing, housing choice vouchers, and low-income housing tax-credit developments. Each of these housing policies can benefit from embedded services for low-income and market-rate residents alike. YFYN can easily be attached to these current affordable housing policies. Families in low-income neighborhoods would benefit if affordable housing providers can attach YFYN to their community-based service approach. Results from the study also have several implications for research. First, although the results are promising, testing the effects of YFYN with a cluster-randomized design with adequate power will be critical. In addition, future research needs to consider the importance of partner sites and external validity. YFYN has been delivered at various partner sites, including subsidized housing developments, community centers, and schools. The intervention would benefit greatly from a directed approach to assessing effects across different partner sites. Considering how unique neighborhoods are, tests of YFYN can assess variations in dosage to address presenting issues of specific neighborhoods. More specifically, additional curriculum sessions can be designed and either added or removed from the 10-week YFYN delivery depending on neighborhood-specific challenges and strengths. Other community interventions, such as CTC, have used a similar modular approach. In addition to testing YFYN with a cluster-randomized trial, continued intervention testing should focus on effects related to implementation partners. It is likely that the theoretical mechanisms for change will have a different impact depending on factors specific to partner sites. Delivery of YFYN at schools, for example, may compromise intervention effects if participating families have low school attachment due to issues related to trust. Conversely, delivery at mixed-income housing developments may enhance intervention effects as the YFYN mediating mechanisms address challenges, such as micro-segregation, that have been identified in the extant literature (Chaskin & Joseph, 2015). In conclusion, this early-phase study of YFYN offers some encouraging results. Practitioners and administrators working on developing neighborhood relationships may have a particular interest in YFYN. Although more testing needs to be done, in the future YFYN may be an important approach for building neighborhood social cohesion and informal neighborhood social control. Daniel Brisson, PhD, MSW, is professor and executive director, Burnes Center on Poverty and Homelessness, Graduate School of Social Work, University of Denver, 2148 South High Street, Denver, CO 80208-7100; e-mail: [email protected]. Stephanie Lechuga Peña, PhD, MSW, is assistant professor, School of Social Work, Arizona State University, Phoenix. Nicole Mattocks, MSW, is a PhD candidate and research assistant, School of Social Work, University of Maryland, Baltimore. Mark Plassmeyer, PhD, MSW, is assistant professor, School of Social Work, University of Arkansas, Fayetteville. Sarah McCune, MSW, is a graduate student, Graduate School of Professional Psychology, University of Denver. References Aha Process Inc . ( 2017 ). [Home page]. Retrieved from http://www.ahaprocess.com Ahmad , F. S. , & Tsang , T. ( 2013 ). Diabetes prevention, health information technology, and meaningful use: Challenges and opportunities . American Journal of Preventive Medicine, 44 , 357 – 363 . Google Scholar Crossref Search ADS WorldCat Albright , A. L. , & Gregg , E. W. ( 2013 ). Preventing type 2 diabetes in communities across the U.S.: The National Diabetes Prevention Program . American Journal of Preventive Medicine, 44 , 346 – 351 . Google Scholar Crossref Search ADS WorldCat Aurand , A. , Emmanuel , D. , Yentel , D. , Errico , E. , Gaby-Biegel , J. , & Kerr , E. ( 2018 ). Out of reach: The high cost of housing . Washington, DC : National Low-Income Housing Coalition . Google Preview WorldCat COPAC Berger , L. M. , & Font , S. A. ( 2015 ). The role of the family and family-centered programs and policies . Future of Children, 25 ( 1 ), 155 – 176 . Google Scholar Crossref Search ADS PubMed WorldCat Brisson , D. , & Walker , L. A. ( 2019 ). A description of social cohesion in low to moderate-income neighborhoods . Unpublished manuscript . Google Preview WorldCat COPAC Bronfenbrenner , U. ( 1979 ). The ecology of human development: Experiments in nature and design . Cambridge, MA : Harvard University Press . Google Preview WorldCat COPAC Bronfenbrenner , U. ( 1986 ). Ecology of the family as a context for human development: Research perspectives . Developmental Psychology, 22 ( 6 ), 723 – 742 . Google Scholar Crossref Search ADS WorldCat Brooks-Gunn , J. , Duncan , G. J. , & Aber , J. L. ( 1997 ). Neighborhood poverty: Context and consequences for children . New York : Russell Sage Foundation . Google Preview WorldCat COPAC Bryan , M.L.M. , & Davis , A. F. (Eds.). ( 1990 ). One hundred years at Hull-House . Bloomington : Indiana University Press . Google Preview WorldCat COPAC Chaskin , R. J. , & Joseph , M. L. ( 2015 ). The promise and perils of mixed-income public housing transformation . Chicago : University of Chicago Press . Google Preview WorldCat COPAC Cheadle , A. , Rauzon , S. , & Schwartz , P. M. ( 2014 ). Community-level obesity prevention initiatives: Impact and lessons learned . National Civic Review, 103 ( 1 ), 35 – 39 . Google Scholar Crossref Search ADS WorldCat Economos , C. D. , Hyatt , R. R. , Goldberg , J. P. , Must , A. , Naumova , E. N. , Collins , J. J. , & Nelson , M. E. ( 2007 ). A community intervention reduces BMI z-score in children: Shape Up Somerville first year results . Obesity, 15 , 1325 – 1336 . Google Scholar Crossref Search ADS PubMed WorldCat Elliott , J. R. ( 1999 ). Social isolation and labor market insulation: Network and neighborhood effects on less-educated urban workers . Sociological Quarterly, 40 , 199 – 216 . doi:10.1111/j.1533-8525.1999.tb00545.x Google Scholar Crossref Search ADS WorldCat Evans , G. W. ( 2004 ). The environment of childhood poverty . American Psychologist, 59 , 77 – 92 . doi:10.1037/0003-066X.59.2.77 Google Scholar Crossref Search ADS WorldCat Fischer , W. , & Sard , B. ( 2016 ). Chart book: Federal housing spending is poorly matched to need. Retrieved from http://www.cbpp.org/research/housing/chart-book-federal-housing-spending-is-poorly-matched-to-need Fram , M. S. , Miller-Cribbs , J. , & Van Horn , L. ( 2007 ). Poverty, race, and the contexts of achievement: Examining educational experiences of children in the U. S. South . Social Work, 52 , 309 – 319 . doi:10.1093/sw/52.4.309 Google Scholar Crossref Search ADS PubMed WorldCat Hawkins , J. D. , & Catalano , R. F. ( 2005 ). Investing in your community’s youth: An introduction to the Communities That Care system. Retrieved from http://www.communitiesthatcare.net/userfiles/files/Investing-in-Your-Community-Youth.pdf Hawkins , J. D. , Oesterle , S. , Brown , E. C. , Arthur , M. W. , Abbot , R. D. , Fagan , A. A. , & Catalano , R. F. ( 2009 ). Results of a type 2 translational research trial to prevent adolescent drug use and delinquency: A test of Communities That Care . Archives of Pediatric Adolescent Medicine, 163 , 789 – 798 . Google Scholar Crossref Search ADS WorldCat Heinze , J. E. , Reischl , T. M. , Bai , M. , Roche , J. S. , Morrel-Samuels , S. , Cunningham , R. M. , & Zimmerman , M. A. ( 2015 ). A comprehensive prevention approach to reducing assault offenses and assault injuries among youth . Prevention Science, 17 , 168 – 176 . WorldCat Jargowsky , P. A. ( 2013 ). Concentration of poverty in the new millennium: Changes in prevalence, composition, and location of high poverty neighborhoods . New Brunswick, NJ: Rutgers : University Press . Google Preview WorldCat COPAC Kalichman , S. C. , Simbayi , L. C. , Cain , D. , Carey , K. B. , Carey , M. P. , Eaton , L. , et al. ( 2014 ). Randomized community-level HIV prevention intervention trial for men who drink in South African alcohol-serving venues . European Journal of Public Health, 24 , 833 – 839 . Google Scholar Crossref Search ADS PubMed WorldCat Kirby , J. B. , & Kaneda , T. ( 2005 ). Neighborhood socioeconomic disadvantage and access to health care . Journal of Health and Social Behavior, 46 ( 1 ), 15 – 31 . doi:10.1177/002214650504600103 Google Scholar Crossref Search ADS PubMed WorldCat Marmot , M. , & Wilkinson , R. G. (Eds.). ( 2006 ). Social determinants of health (2nd ed.). Oxford, UK : Oxford University Press . Google Preview WorldCat COPAC Martinez , O. , Roth , A. M. , Kelle , G. , Downs , M. , & Rhodes , S. D. ( 2014 ). Adaptation and implementation of HoMBReS: A community-level, evidence-based HIV behavioral intervention for heterosexual Latino men in the midwestern United States . AIDS Education & Prevention, 26 , 68 – 80 . Google Scholar Crossref Search ADS WorldCat McLeroy , K. R. , Bibeau , D. , Steckler , A. , & Glanz , K. ( 1988 ). An ecological perspective in health promotion programs . Health Education Quarterly, 15 , 351 – 377 . Google Scholar Crossref Search ADS PubMed WorldCat Morenoff , J. D. , Sampson , R. J. , & Raudenbush , S. ( 2001 ). Neighborhood inequality, collective efficacy, and the spatial dynamics of urban violence . Criminology, 39 , 517 – 560 . Google Scholar Crossref Search ADS WorldCat Morrel-Samuels , S. , Zimmerman , M. A. , & Reischl , T. M. ( 2013 ). Creating safe and healthy futures: Michigan Youth Violence Prevention Center . Reclaiming Children and Youth, 22 ( 3 ), 31 – 36 . WorldCat Nation , M. , Crusto , C. , Wandersman , A. , Kumpfer , K. L. , Seybolt , D. , Morissey-Kane , E. , & Davino , K. ( 2003 ). What works in prevention: Principles of effective prevention programs . American Psychologist, 58 , 449 – 456 . Google Scholar Crossref Search ADS PubMed WorldCat Raja , S. , Ma , C. , & Yadav , P. ( 2008 ). Beyond food deserts: Measuring and mapping racial disparities in neighborhood food environments . Journal of Planning Education and Research, 27 , 469 – 482 . doi:10.1177/0739456X08317461 Google Scholar Crossref Search ADS WorldCat Rhew , I. C. , Brown , E. C. , Hawkins , D. , & Briney , J. S. ( 2013 ). Sustained effects of the Communities That Care system on prevention service system transformation . American Journal of Public Health, 103 , 529 – 535 . Google Scholar Crossref Search ADS PubMed WorldCat Sampson , R. J. ( 2001 ). How do communities undergird or undermine human development? Relevant contexts and social mechanisms. In A. Booth & A. C. Crouter (Eds.), Does it take a village?: Community effects on children, adolescents, and families (pp. 3 – 30 ). Mahwah, NJ : Lawrence Erlbaum . Google Preview WorldCat COPAC Sampson , R. J. , Raudenbush , S. W. , & Earls , F. ( 1997 ). Neighborhoods and violent crime: A multilevel study of collective efficacy . Science, 227 ( 5328 ), 918 – 924 . Google Scholar Crossref Search ADS WorldCat Swain , G. R. ( 2016 ). How does economic and social disadvantage affect health? Focus, 33 ( 1 ), 1 – 6 . WorldCat U.S. Census Bureau. ( 2016 ). American Fact Finder . Washington, DC : Author . Google Preview WorldCat COPAC Woelk, G. B., Kieffer, M. P., Walker, D., Mpofu, D., Machekano, R., & the Project ACCLAIM Study Group . ( 2016 ). Evaluating the effectiveness of selected community-level interventions on key maternal, child health, and prevention of mother-to-child transmission of HIV outcomes in three countries (the ACCLAIM Project): A study protocol for a randomized controlled trial . Trials, 17 ( 88 ), 1 – 16 . PubMed WorldCat © 2019 National Association of Social Workers This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Rose, Roderick, A;Bowen, Natasha, K

doi: 10.1093/swr/svz017pmid: N/A

Abstract Nonrandomized evaluation designs are an important part of social work research because randomization is not always feasible in social work settings. Although randomly assigned groups are assumed to be equivalent, nonrandomly assigned groups are not. In nonrandomized settings, designs with multiple waves are ideal, but two-wave designs are still widely used. A common method for estimating a treatment effect in nonrandom two-wave designs is the pretest–posttest model. However, depending on relationships among participants and the method of assignment to treatment groups, researchers should consider a difference-in-differences approach to testing treatment effects. Authors describe and compare the pretest–posttest and difference-in-differences approaches and assumptions and offer guidelines, developed from a literature review, about the conditions under which each model is likely to be best. Authors also demonstrate the decision-making process and application of the methods in an evaluation of an elementary school intervention program. Randomized controlled designs are preferred by researchers for establishing causality in intervention research. In a simple randomized controlled trial (RCT), a pool of study participants is assigned to treatment and control conditions according to a random process. It is a credible assumption that as a result of the random process, on average, the two groups are balanced on both measured and unobserved characteristics, allowing for the unbiased estimation of causal treatment effects. However, randomized assignment is not always possible in practice settings. It may be unethical, illegal, or impractical to deny or delay treatment, or it may not be possible to identify members for a control group. For example, withholding a domestic violence intervention might be unethical; randomization of Medicaid benefits to study access to care would be illegal. It is impractical to randomize in studies of mediators, retrospective studies, and studies using secondary data. In some situations, participants may refuse to be randomized. In the absence of random assignment, the equivalence of treatment groups cannot be assumed because the factors that led to assignment may also be associated with the outcome. For example, if students are assigned to reading remediation based on a test score, and will be retested later, the correlation in the two test scores may explain part of the correlation between the remediation and posttest. In the most common variation of two-wave nonrandomized design, a posttreatment dependent variable (posttest) is regressed on the treatment, pretreatment values of the outcome (pretest), and observed covariates (for example, Hoefer & Bryant, 2017; Lee, 2018). Most social work researchers refer to this as a pretest–posttest (PP) regression. In some literature it is referred to as analysis of covariance (ANCOVA) regression (for example, van Breukelen, 2013). Less common to social work research is difference-in-differences (DD), an econometric approach to estimating change. In two-wave designs, DD is equivalent to the change score model, which has been discussed previously in this journal (Gillespie & Streeter, 1994) and has been examined closely (for example, van Breukelen, 2013). We believe that an update for social work researchers is needed given the continued reliance on the two-wave design and PP, the evolving views of experts, and the unfamiliarity with DD among social work researchers. The appropriate conditions for using DD discussed here are consonant with those of change score analysis. More generally, DD will be valuable to social work researchers because it can be extended to designs with multiple waves and conditions, among other numerous variations, although an extensive discussion of these approaches is beyond the scope of this article. Because researchers cannot always implement studies with randomized or repeated measures designs, they will continue to use two-wave nonrandomized designs. Therefore, it is important for them to understand the conditions under which it is appropriate to use DD as opposed to the more common PP approach. The relevant factor guiding the choice of PP or DD is the amount of change from pretest to posttest that each group would have experienced in the absence of treatment, which is unobserved in the treatment group. Thus, the choice of PP or DD in the two-wave design relies on an untestable assumption. Nevertheless, the researcher can base this assumption on available information about the design and the sample. Regression to the mean, the tendency for extreme scores in one period to be preceded by or followed by less extreme scores, factors into this assumption. Our objective is to describe and compare the PP and DD approaches using the two-wave nonequivalent comparison group design (Shadish, Cook, & Campbell, 2002). In this design, a treatment group receives an intervention and an untreated comparison group does not. We assume that there is a true but unknown treatment effect for the intervention. Our goal is to estimate the effect with data. Further on, we use PP data on end-of-grade (EOG) reading achievement from a school intervention panel study to illustrate the PP and DD approaches. In closing, we offer a brief set of guidelines to aid in selection of PP or DD. The Secular Change Assumption In a nonrandomized design, we expect that a comparison on the posttest only will lead to a biased estimate because the treatment and comparison groups are not expected to be equivalent at pretest. It is intuitive that including a pretest in the model to adjust for pretreatment differences provides a fairer comparison of the two conditions. In addition, by including the pretest we might be able to control for some unobserved confounders. Although the process by which the effect of an unmeasured confounder is mediated by the pretest cannot be directly examined, the inclusion of pretest scores for this purpose is common (Morgan & Winship, 2007). Regardless of the purpose, the amount of the pretest that must be adjusted for is an important factor to consider. As we explain in the next section, the PP and DD models each make different, very specific, adjustments. A criterion known as secular change, the amount that the two groups would have changed from pretest to posttest in the absence of treatment (Senn, 2006), is available for deciding which adjustment, and thus which method, is more appropriate. Although we are carrying out these adjustments using the pretest, it is the untreated posttest difference between groups that we are adjusting for. For the treatment group, the untreated posttest is an unobservable condition, because the observed posttest score includes the treatment effect. However, the unobserved untreated posttest in the treatment group can be connected to the observed pretest by an assumption about the amount of secular change in the treatment group. The comparison group experiences a secular change as well, but because the comparison group does not receive the treatment, we treat it as equivalent to the observed trend. The secular change assumption and how it enables us to correctly or incorrectly estimate the treatment effect is illustrated in Figure 1. The observed change in treatment group scores is a sum of its secular change and the treatment effect. Treatment and comparison groups have parallel secular trends if we assume that treatment secular 1 is correct; they have nonparallel (in this case, converging) secular trends if we assume that treatment secular 2 is correct. Unlike in this illustration, secular change cannot be observed directly in the treatment group, and it cannot be indirectly estimated in a two-wave design; doing so would require multiple pretests, not just the one pretest available in the two-wave design. And of course, we never know the true treatment effect. Therefore, we must make an untestable assumption about the value of secular change in the treatment group to make an adjustment to the pretest and estimate a treatment effect. As noted, secular change connects the unobserved posttest difference (in untreated scores) to the pretest difference. It is represented as the rate at which the difference between groups persists from pre- to posttest, or in other words the fraction of the pretest difference that should be subtracted from the observed posttest difference in estimating the treatment effect, which we label “α” in accordance with Morgan and Winship (2007): $$\begin{eqnarray}&d = \textrm{Observed\ posttreatment}\qquad& \nonumber\\[-3pt] &\mathrm{difference}-\alpha \mathrm{(pretreatment\ difference)}&\quad\\[-29pt] \nonumber\end{eqnarray}$$ (1.1) Equation 1.1 says the estimated treatment effect d is the difference between groups at posttest minus a fraction of their difference at pretest. Each of the different secular trends in Figure 1 implies a different fraction. For example, assuming a –0.5 SD pretreatment difference and –0.1 posttreatment difference as Figure 1, we would have to remove 80% of the pretreatment difference (α = .8), –0.4 SD, to obtain a true effect of 0.3. Secular Change in PP and DD Quantifying secular change is not pertinent to randomized designs, because in RCTs the pretreatment difference has an expectation of zero and thus α can take on any value without affecting the finding. The assumptions regarding secular change in PP and DD are incompatible with each other in a nonrandomized design in which the expected pretest difference is nonzero. The assumption about secular change, and the corresponding fraction of the pretest to adjust for, is the deciding factor in the choice of PP or DD in nonrandomized designs. In this section we describe how secular change manifests in each method. PP Model In a PP model, the dependent variable is regressed on treatment assignment, the pretest, and possibly covariates. The PP model estimates the association between the treatment and the posttest, controlling for, among other factors, the pretest. Ignoring covariates, a typical PP regression is as follows: $$\begin{equation}{Y}_{1i}={a}_{PP}+{b}_Y{Y}_{0i}+{d}_{PP}{M}_i+{e}_i\end{equation}$$ (2.1) In equation 2.1, |${Y}_{1i}$| = posttest for student i; |${Y}_{0i}$| = pretest for student i; |${a}_{PP}$| is the intercept; |${M}_i$| gives treatment assignment, with |$M$| = 1 assigned to treatment and |$M$| = 0 assigned to the comparison condition; |${b}_Y$| is the coefficient of the pretest; and |${d}_{PP}$| is the coefficient representing the PP treatment effect. Figure 1 Open in new tabDownload slide Hypothetical Example of Total Change from Pretest to Posttest. Including Change Due to Treatment and Potential Secular Changes for the Treatment Condition. Notes: Tx = treatment. Assume that the true effect d = 0.3 SD. The highest line (square) illustrates the comparison group experiencing an observed secular change of 0.2 SD, increasing from 0.7 to 0.9 from pretest to posttest. The figure also illustrates three different scenarios for the treatment group. The Tx observed line (diamond) represents the observed pretest to posttest change. Tx secular 1 (circle) illustrates a hypothetical rate of secular change for the treatment group that is the same as in the comparison group (the two lines are parallel; the treatment group changes from 0.2 to 0.4, a change of 0.2 SD). The Tx secular 2 line (triangle), alternatively, illustrates a rate of secular change in the treatment group that is greater than the rate of comparison group change; that is, the trend in the treatment group’s scores brings them closer to the comparison group’s scores, with an increase of 0.3 SD (going from 0.2 to 0.5) instead of 0.2 SD from pretest to posttest. If the true effect is 0.3 SD, then assuming secular 1 would actually overestimate this effect by one-third (0.4 SD). The correct assumption to estimate the treatment effect accurately is secular 2. Figure 1 Open in new tabDownload slide Hypothetical Example of Total Change from Pretest to Posttest. Including Change Due to Treatment and Potential Secular Changes for the Treatment Condition. Notes: Tx = treatment. Assume that the true effect d = 0.3 SD. The highest line (square) illustrates the comparison group experiencing an observed secular change of 0.2 SD, increasing from 0.7 to 0.9 from pretest to posttest. The figure also illustrates three different scenarios for the treatment group. The Tx observed line (diamond) represents the observed pretest to posttest change. Tx secular 1 (circle) illustrates a hypothetical rate of secular change for the treatment group that is the same as in the comparison group (the two lines are parallel; the treatment group changes from 0.2 to 0.4, a change of 0.2 SD). The Tx secular 2 line (triangle), alternatively, illustrates a rate of secular change in the treatment group that is greater than the rate of comparison group change; that is, the trend in the treatment group’s scores brings them closer to the comparison group’s scores, with an increase of 0.3 SD (going from 0.2 to 0.5) instead of 0.2 SD from pretest to posttest. If the true effect is 0.3 SD, then assuming secular 1 would actually overestimate this effect by one-third (0.4 SD). The correct assumption to estimate the treatment effect accurately is secular 2. Properties of the PP Estimator According to van Breukelen (2013), the coefficient |${d}_{PP}$| is equal to $$\begin{equation}{d}_{PP} = {\overline{Y}}_{11}- {\overline{Y}}_{10}-{b}_Y\big({\overline{Y}}_{01}- {\overline{Y}}_{00}\big)\end{equation}$$ (2.2) Here, |${\overline{Y}}_{11}$| = sample mean posttest for treatment; |${\overline{Y}}_{10}$| = sample mean posttest for comparison; |${\overline{Y}}_{01}$| = sample mean pretest for treatment; and |${\overline{Y}}_{00}$| = sample mean pretest for comparison. Thus, the PP estimate of the treatment effect is the difference in mean outcomes at posttest (⁠|${\overline{Y}}_{11}-{\overline{Y}}_{10})$|⁠, minus the pretest regression coefficient (⁠|${b}_Y$|⁠) for |${Y}_{0i}$| (from equation 2.1) times the difference in mean pretests (⁠|${\overline{Y}}_{01}-{\overline{Y}}_{00}$|⁠). Equation 2.2 is a restatement of equation 1.1, with α = |${b}_Y$|⁠. Because |${b}_Y$| is typically between 0 and 1, |${d}_{PP}$| does not adjust for the full difference between treatment and comparison groups at pretest (⁠|${\overline{Y}}_{01}-{\overline{Y}}_{00}$|⁠), but only a fraction |${b}_Y$| of the difference. The finding that α = |${b}_Y$| in the PP is crucial. If the measures are standardized, then the coefficient |${b}_Y$| in equations 2.1 and 2.2 is equal to the within-group correlation between pretest and posttest, that is, the correlation in each group after adjusting for treatment assignment (Morgan & Winship, 2007). If the measures are not standardized, it is a function of this correlation. This coefficient is usually less than 1, signifying that the predicted values within each group regress to a group mean (regression to the mean = 1 – correlation; Campbell & Kenny, 1999). Scores in the treated group regress toward the treatment group mean at posttest, and scores in the comparison group regress toward the comparison group mean at posttest. If α—the rate at which the pretest difference persists from pretest to posttest—is required to be equal to this value for the PP model to be unbiased, then both groups must be regressing toward a common mean at a rate given by 1 -- |${b}_Y$|⁠, making the secular trends nonparallel. In Figure 1, the assumption is represented by the secular 2 line. DD Model In a two-wave design with two groups, a DD analysis is a comparison of changes: We estimate the change between pretest (⁠|${\overline{Y}}_{01}$|⁠) and posttest (⁠|${\overline{Y}}_{11}$|⁠) in the group receiving treatment and subtract from this the change between the pretest (⁠|${\overline{Y}}_{00}$|⁠) and posttest (⁠|${\overline{Y}}_{10}$|⁠) in the comparison group: $$\begin{equation}{d}_{DD} = {\overline{Y}}_{11}-{\overline{Y}}_{01}-\big({\overline{Y}}_{10}-{\overline{Y}}_{00}\big)\end{equation}$$ (3.1) The resulting difference score |${d}_{DD}$| represents the difference between the two changes. This difference score can be estimated directly by applying equation 3.1 to the pre- and posttest values in each group. According to Imbens and Wooldridge (2009), a regression approach uses |$T$| = time point (0 = pretest, 1 = posttest), |${M}_i$| = treatment assignment, and an interaction of these two variables (⁠|$T{M}_i$|⁠) as follows: $$\begin{equation}{Y}_{ti}={a}_{DD}+{g}_TT+{g}_M{M}_i+{d}_{DD}T{M}_i+{r}_{ti}\end{equation}$$ (3.2) In this equation, |${Y}_{ti}$| is the outcome for person i at time t; |${a}_{DD}$| is a constant; |${g}_T$| is the coefficient for change over time T within the comparison group; |${g}_M$| is the coefficient capturing differences between the treatment and comparison groups at pretest; and |${r}_{ti}$| is an error term. The treatment effect is estimated by |${d}_{DD}$|⁠, the coefficient on the interaction between time and treatment, representing the difference between the treatment arms at posttest. Properties of the DD Estimator To understand the situations in which it is appropriate to use the DD model instead of the PP, we must examine the properties of the DD estimator. First, we rearrange the middle two terms from equation 3.1, |${\overline{Y}}_{01}$| and |${\overline{Y}}_{10}$|⁠, as follows: $$\begin{equation}{d}_{DD} = {\overline{Y}}_{11}-{\overline{Y}}_{10}-\big({\overline{Y}}_{01}-{\overline{Y}}_{00}\big)\end{equation}$$ (3.3) This renders |${d}_{DD}$| the DD counterpart to equation 2.2, but with the implication that α = 1. Because α = 1, the full difference at pretest is subtracted from the posttest estimate. Thus, with DD, we are assuming that scores for the two groups experience the same secular change and do not regress toward each other in the absence of treatment. There will still be within-group regression toward a common mean, but in the absence of treatment, the means would always remain the same distance apart. This assumption about secular change in the DD model is inconsistent with the secular change assumption in the PP model. Therefore, at best only one of them can be true. Referring back to Figure 1, secular 1 is the correct secular trend for DD. The secular trend assumption for DD is sometimes referred to as the “parallel trends” assumption, though this may be confusing if the reader does not understand that it refers to secular rather than observed trends. Informing Selection of PP or DD in Two-Wave Designs To summarize the prior discussion, the PP and DD models make different assumptions about secular change in the treatment and comparison groups. Whether either assumption is correct comes down to whether it is more likely that the two groups regress toward a common mean at the same rate that scores for individuals within each group regress toward the group’s mean (PP), or whether it is more likely that scores for the groups on average maintain the difference observed at pretest (DD) regardless of within-group regression to the mean. What does our intuition suggest, and what does the extant research say? Factors Favoring the PP Model Given that regression of both groups to a common mean is the deciding factor, we would be better off choosing PP in designs in which the groups differ primarily on time-varying or transient characteristics but are otherwise similar, suggesting their differences are to a certain extent transitory and may revert. Time varying in this case pertains to any characteristic that may change from wave to wave during the study period, such as a score on a test or number of absences. Factors that vary in general over time but are fixed for most youths during the study period, such as neighborhood or school of enrollment, should not be viewed as time varying for this purpose. The research on PP and change score modeling supports this intuition. Van Breukelen (2013) characterized PP as a design that is appropriate in situations in which there is one group prior to pretest, which is then subsequently nonrandomly assigned to two groups after measuring a transient characteristic such as a pretest: $$\begin{eqnarray}&\mathrm{One\ group} \rightarrow \mathrm{Transient\ characteristic} \rightarrow& \nonumber\\ &\mathrm{Assignment\ to\ treatment} \rightarrow \mathrm{PP}&\end{eqnarray}$$ (4.1) Because they are constituted prior to pretest, members of the initial group may maintain their similarities despite assignment to different treatment arms, and as a result, the two groups may naturally regress toward each other. Two types of pretest-based assignments are worth singling out: assignment based on the pretest and situations where the pretest causes the posttest. Assignment Based on Pretest It is possible that assignment is not just correlated with the pretest, but also based on the pretest itself. The most extreme form of these types of designs is the regression discontinuity design (Shadish et al., 2002), in which assignment is determined strictly by a single covariate (for example, the pretest) (Allison, 1990). Even if assignment is not completely determined by the pretest, we often see designs in social work research in which participants are assigned nonrandomly to the treatment because they have greater need for services, as evidenced by worse outcomes on the pretest. Finally, even if the pretest is not used to determine assignment and a correlated but unmeasured time-varying factor from the same time point is, PP may be better (Allison, 1990), although there may be some omitted variable bias because controlling for the pretest may not address all pretreatment differences. Pretest Causes Posttest In some circumstances the pretest may have an independent causal effect on the posttest (Allison, 1990). For example, students’ reading comprehension may increase in the pretest period, which may lead to better reading comprehension in later periods. In these scenarios, the PP model will usually be a better choice. Factors Favoring the DD Model Alternatively, assignment based on time-invariant or fixed characteristics, such as school membership and birth cohort, suggests that the two groups are not members of a single group nonrandomly assigned to treatment arms, but rather are two (or more) separate groups by their nature (van Breukelen, 2013). Our intuition suggests that there should be no “pull” causing scores for the two groups to regress toward each other naturally. In these cases, DD is the better approach: $$\begin{eqnarray}&\mathrm{Fixed\ characteristic} \rightarrow \mathrm{Multiple\ groups} \rightarrow& \nonumber\\ &\mathrm{Assignment\ to\ treatment} \rightarrow DD&\end{eqnarray}$$ (4.2) Comparing 4.2 to 4.1, we see that the characteristic is determined before group membership in 4.2 and after membership in 4.1. Group membership by school—that is, from assignment of schools to treatment conditions—is one such time-invariant factor that is often used in school research. Example: Evaluation of a Model of Assessment and Prevention Students in low-performing schools in low-performing school districts often have little opportunity to learn and achieve at their potential. Chronically low-performing schools typically have inadequate instructional and student support resources (for example, human, financial, and material resources), and the schools most in need of supports may be least capable of organizing to exploit an influx of supports and resources. Efforts to increase student proficiency rates by incentivizing mandates (for example, testing, trainings, and monitoring) have failed to change the educational prospects of millions of school children (Ryan & Deci, 2009). As an alternative, the model of assessment and prevention (MAP) (Bowen, Thompson, & Powers, 2012) presented in this study emphasized the cultivation of competence and autonomy among school personnel for implementing an evidence-based process. Elementary School Success Profile MAP: Description The example comes from a one-year efficacy study of the Elementary School Success Profile MAP in a large school district encompassing urban and rural areas in a southeastern state (Bowen et al., 2012). MAP is a guided sequence of assessment, decision-making, and intervention steps coupled with financial resources to be used in elementary schools, that are hypothesized to affect learning (and test scores). In each treatment school, the process included (a) formation of a four- to six-person implementation team that met five times over the course of the school year, (b) collection of social environmental data from a random sample of low-achieving students in grades 3 to 5, (c) use of assessment data by teams to identify intervention targets for all low-achieving students, (d) selection and implementation of appropriate intervention strategies from an online database, and (e) a posttest social environmental assessment. Each school team was selected by the principal and included a school-specific combination of social workers, counselors, school leaders, teachers, and intervention specialists. Teams identified two or three intervention targets based on assessment results. They then examined an extensive online database describing empirically based practices and selected appropriate interventions related to their targets. Each school team was given about $2,000 for intervention expenses, and each implemented three evidence-based programs or promising practices. Intervention targets chosen by teams included attendance, social skills, social behavior, learning behavior, and educational involvement of parents. Interventions targeted all low-achieving students. Study Design The design of the study was based on a three-stage assignment process (see Figure 2). First, of 28 elementary schools in the district, 10 with the lowest average EOG scores in the 2006–2007 school year were chosen purposively for the school-level RCT. (The 10 schools remained the lowest performing schools in 2007–2008.) Second, five matched pairs of the low-performing schools were then created based on demographic characteristics that were most predictive of school-level performance according to data from prior years. One school from each pair was then randomly assigned to the treatment condition and the other to the control condition. Third, in the student-level design, third- through fifth-grade students were purposively assigned to the treatment based on their early fall 2008 benchmarks, which were assumed to be highly associated with their 2007–2008 EOG scores. MAP was offered in 2008–2009. We used 2007–2008 and 2008–2009 EOG reading scores obtained from the North Carolina Education Research Data Center as our pretest and posttest outcome data. The institutional review boards of both authors’ universities approved the study. Figure 2 Open in new tabDownload slide Flowchart of Sample Figure 2 Open in new tabDownload slide Flowchart of Sample The multiple stages of assignment and the within-school and between-school aspects of the intervention provide a good way to frame the comparison of PP and DD. We break down the design into three different assignment rules. The school-level nonrandom assignment is a design comparing the 10 low-performing schools in the RCT with the 18 high-performing schools. The school-level random design consisted of a comparison between the five low-performing schools randomly assigned to MAP and the five low-performing schools randomly assigned to the control. The student-level nonrandom design consisted of a comparison between low-performing youths and high-performing youths within the five schools selected for MAP treatment. We use all three designs to illustrate how PP and DD behave under real-world conditions. Per the preceding discussion on how to choose the appropriate method given a design, we argue the following: First, time-varying factors could include the value of the pretest; any variable based on this pretest score including a designation as low performing; or classroom groupings, which will likely change from year to year, across grade-level advancements. School membership would not constitute a time-varying characteristic under this definition. Therefore, we hypothesize that (a) in the school-level nonrandom design, DD is the better choice because it reflects preassignment groupings; (b) in the school-level randomized design, randomization will ensure that PP and DD will produce equivalent estimates; and (c) in the student-level nonrandom design, PP is the better choice because it reflects assignment after pretest performance. It is taken as a given based on their respective designs that the nonrandomized designs are subjected to selection influences. Sample and Measures Our analysis used data from all 2007–2008 third- and fourth-grade students in the 28 elementary schools in the district (n = 4,772). We focused on grades 3 and 4 at pretest to ensure students were still in elementary school at wave 2 (2008–2009). Just over half (52%) were enrolled in grade 3 in 2007–2008 and 48% in grade 4; nearly half were female; 54% were African American; 21% were Hispanic; 19% were white; and about 4% were Native American or multiracial. In the school-level nonrandom design, 1,410 students (30%) were in the 10 low-performing schools selected for “treatment” and 3,362 students (70%) were in the 18 comparison schools. In the school-level randomized design, there were 663 students (47%) in the treatment schools and 747 students (53%) in the comparison schools. In the student-level nonrandom design, 405 students (61%) were assigned to treatment and 258 students (39%) were assigned to the comparison. The same group of students was sampled at each wave. EOG reading scores were used to evaluate the intervention. At pretest the mean score in all schools was 338 (SD = 12.2; range = 308 to 370); at posttest the mean score was 346 (SD = 10.3; range = 310 to 374); 747 students (15%) were missing an EOG reading score at pretest (wave 0), as were 344 at posttest (7%; wave 1). Missing values on the dependent variable do not contribute to parameter bias if the values are missing at random (Allison, 2002). Given that this is a demonstration and not a true test of MAP, we did not pursue further missing data amelioration strategies beyond full information maximum likelihood. A treatment assignment variable was created for each design (1 = treated, 0 = comparison or control). A wave or time period variable was created for each DD model (0 = pretest, 1 = posttest). Analysis Best Model Choice In typical two-wave scenarios, no further data would be available to determine whether the correct model has been chosen. However, in this demonstration pre-pretest 2006–2007 data (M = 333, SD = 10.3, range = 307 to 362) were available to show the pretreatment trend in both groups, giving us the only objective standard for the correct model choice. These three-wave trends provide some insight into whether treatment groups regress to a common mean, by showing their pretreatment trends. As noted previously, regression to the mean is signified by nonparallel secular trends. This allows us to validate whether our assumption about secular change was correct and whether the best approach was chosen for each design. We calculated the average score in each treatment group in each of the three waves and graphed them to show whether the trends before treatment were parallel or not. Converging or diverging trends from the pre-pretest wave to pretest were indicative of regression to the mean for a single group of participants assigned to different conditions based on transient characteristics (for example, 4.1) and suggested that PP is the better choice. Parallel trends over this period supported the assumption of two different groups at assignment (4.2) and that the DD was the most appropriate analysis choice. PP versus DD Regression models were run in accordance with the PP and DD models shown in equations 2.1 and 3.2, respectively, to estimate the treatment effect under the assumptions of each method. Appropriate multilevel methods were used for each design given the nesting of time and children in schools. For the ANCOVA models in all designs, a two-level model of students nested in schools was used. For the DD models in all designs, a three-level model of repeated measures taken on students nested in schools was used. To maintain a focus on the change analysis, no covariates were included in any models. Although researchers typically include more covariates than the pretest, their omission should not be relevant to this demonstration. We address the issue of covariates further in the Discussion section. Figure 3 Open in new tabDownload slide Trends in Designs 1, 2, and 3 Including Two Pretests Using End-of-Grade (EOG) Reading Figure 3 Open in new tabDownload slide Trends in Designs 1, 2, and 3 Including Two Pretests Using End-of-Grade (EOG) Reading Results Figure 3 contains three graphs (3.1–3.3) of the average EOG reading scores in each treatment group at each of three time points, showing whether regression to the mean between groups is present prior to treatment, the key factor determining whether secular change is parallel (favoring DD) or nonparallel (favoring PP). Graph 3.1 (school-level nonrandom design) shows that the scores for the two groups are parallel between waves –1 and 0, which supports our hypothesis that the DD should be more appropriate in this design. Graph 3.2 (school-level random design) shows that the two groups are not parallel between waves –1 and 0, but they are virtually identical at wave 0, which suggests that randomization was successful and groups were equivalent at pretest. This implies that either design should be unbiased. Graph 3.3 (student-level nonrandom design) supports the PP model because regression of both groups to a common mean is present: The pretest scores that were used to assign participants to treatment conditions were preceded by less extreme scores in wave –1 (Campbell & Kenny, 1999). Table 1 presents results of PP and DD analyses in each of the designs. The school-level nonrandom design shows a negative and nonsignificant treatment effect in the PP analysis (⁠|${d}_{PP}$| = −0.73), but a positive and significant effect in the DD analysis (⁠|${d}_{DD}$| = 0.78). The rules described in the previous section, supported by the trend in Figure 3, graph 3.1, suggest that we should use DD in the school-level nonrandom design. The school-level randomized design shows that both PP and DD have similar positive but not statistically significant treatment effects (⁠|${d}_{PP}$| = 0.28 and |${d}_{DD}$| = 0.33, respectively). The rules cited previously, and Figure 3, graph 3.2, suggest that either design would be correct. The student-level nonrandom design shows a negative and nonsignificant effect in the PP (⁠|${d}_{PP}$| = –0.99), but a very large and significant effect in the DD (⁠|${d}_{DD}$| = 5.74). The rules cited previously, and Figure 3, graph 3.3, support the use of PP in this design. Table 1 Regression Results Under Each Design, End-of-Grade Reading . Beta . SE . t Value . Design 1: Nonrandom School Assignment PP model Intercept (⁠|${a}_{PP}$|⁠) 109.30 2.56 42.71 Pretest (⁠|${b}_Y$|⁠) 0.70 0.01 93.67 MAP (⁠|${d}_{PP}$|⁠) –0.73 0.38 –1.91 DD model Intercept (⁠|${a}_{DD}$|⁠) 339.52 0.81 418.80 Time (⁠|${g}_T$|⁠) 8.07 0.12 66.59 MAP (⁠|${g}_M$|⁠) –5.25 1.37 –3.84 Time-by-MAP (⁠|${d}_{DD}$|⁠) 0.78 0.23 3.47 Design 2: Random School Assignment PP model Intercept (⁠|${a}_{PP}$|⁠) 103.30 4.78 21.59 Pretest (⁠|${b}_Y$|⁠) 0.72 0.01 50.26 MAP (⁠|${d}_{PP}$|⁠) 0.28 0.51 0.56 DD model Intercept (⁠|${a}_{DD}$|⁠) 334.08 1.14 292.77 Time (⁠|${g}_T$|⁠) 8.70 0.25 34.58 MAP (⁠|${g}_M$|⁠) 0.37 1.62 0.23 Time-by-MAP (⁠|${d}_{DD}$|⁠) 0.33 0.37 0.89 Design 3: Nonrandom Student Assignment PP model Intercept (⁠|${a}_{PP}$|⁠) 107.04 11.26 9.51 Pretest (⁠|${b}_Y$|⁠) 0.71 0.03 22.04 MAP (⁠|${d}_{PP}$|⁠) –0.99 0.84 –1.17 DD model Intercept (⁠|${a}_{DD}$|⁠) 345.80 1.09 315.97 Time (⁠|${g}_T$|⁠) 4.51 0.51 8.84 MAP (⁠|${g}_M$|⁠) –16.08 0.80 –20.08 Time-by-MAP (⁠|${d}_{DD}$|⁠) 5.74 0.60 9.62 . Beta . SE . t Value . Design 1: Nonrandom School Assignment PP model Intercept (⁠|${a}_{PP}$|⁠) 109.30 2.56 42.71 Pretest (⁠|${b}_Y$|⁠) 0.70 0.01 93.67 MAP (⁠|${d}_{PP}$|⁠) –0.73 0.38 –1.91 DD model Intercept (⁠|${a}_{DD}$|⁠) 339.52 0.81 418.80 Time (⁠|${g}_T$|⁠) 8.07 0.12 66.59 MAP (⁠|${g}_M$|⁠) –5.25 1.37 –3.84 Time-by-MAP (⁠|${d}_{DD}$|⁠) 0.78 0.23 3.47 Design 2: Random School Assignment PP model Intercept (⁠|${a}_{PP}$|⁠) 103.30 4.78 21.59 Pretest (⁠|${b}_Y$|⁠) 0.72 0.01 50.26 MAP (⁠|${d}_{PP}$|⁠) 0.28 0.51 0.56 DD model Intercept (⁠|${a}_{DD}$|⁠) 334.08 1.14 292.77 Time (⁠|${g}_T$|⁠) 8.70 0.25 34.58 MAP (⁠|${g}_M$|⁠) 0.37 1.62 0.23 Time-by-MAP (⁠|${d}_{DD}$|⁠) 0.33 0.37 0.89 Design 3: Nonrandom Student Assignment PP model Intercept (⁠|${a}_{PP}$|⁠) 107.04 11.26 9.51 Pretest (⁠|${b}_Y$|⁠) 0.71 0.03 22.04 MAP (⁠|${d}_{PP}$|⁠) –0.99 0.84 –1.17 DD model Intercept (⁠|${a}_{DD}$|⁠) 345.80 1.09 315.97 Time (⁠|${g}_T$|⁠) 4.51 0.51 8.84 MAP (⁠|${g}_M$|⁠) –16.08 0.80 –20.08 Time-by-MAP (⁠|${d}_{DD}$|⁠) 5.74 0.60 9.62 Notes: Cutoff for Significance: |t|> 1.96. PP = pretest–posttest; DD = difference-in-differences; MAP = model of assessment and prevention. Open in new tab Table 1 Regression Results Under Each Design, End-of-Grade Reading . Beta . SE . t Value . Design 1: Nonrandom School Assignment PP model Intercept (⁠|${a}_{PP}$|⁠) 109.30 2.56 42.71 Pretest (⁠|${b}_Y$|⁠) 0.70 0.01 93.67 MAP (⁠|${d}_{PP}$|⁠) –0.73 0.38 –1.91 DD model Intercept (⁠|${a}_{DD}$|⁠) 339.52 0.81 418.80 Time (⁠|${g}_T$|⁠) 8.07 0.12 66.59 MAP (⁠|${g}_M$|⁠) –5.25 1.37 –3.84 Time-by-MAP (⁠|${d}_{DD}$|⁠) 0.78 0.23 3.47 Design 2: Random School Assignment PP model Intercept (⁠|${a}_{PP}$|⁠) 103.30 4.78 21.59 Pretest (⁠|${b}_Y$|⁠) 0.72 0.01 50.26 MAP (⁠|${d}_{PP}$|⁠) 0.28 0.51 0.56 DD model Intercept (⁠|${a}_{DD}$|⁠) 334.08 1.14 292.77 Time (⁠|${g}_T$|⁠) 8.70 0.25 34.58 MAP (⁠|${g}_M$|⁠) 0.37 1.62 0.23 Time-by-MAP (⁠|${d}_{DD}$|⁠) 0.33 0.37 0.89 Design 3: Nonrandom Student Assignment PP model Intercept (⁠|${a}_{PP}$|⁠) 107.04 11.26 9.51 Pretest (⁠|${b}_Y$|⁠) 0.71 0.03 22.04 MAP (⁠|${d}_{PP}$|⁠) –0.99 0.84 –1.17 DD model Intercept (⁠|${a}_{DD}$|⁠) 345.80 1.09 315.97 Time (⁠|${g}_T$|⁠) 4.51 0.51 8.84 MAP (⁠|${g}_M$|⁠) –16.08 0.80 –20.08 Time-by-MAP (⁠|${d}_{DD}$|⁠) 5.74 0.60 9.62 . Beta . SE . t Value . Design 1: Nonrandom School Assignment PP model Intercept (⁠|${a}_{PP}$|⁠) 109.30 2.56 42.71 Pretest (⁠|${b}_Y$|⁠) 0.70 0.01 93.67 MAP (⁠|${d}_{PP}$|⁠) –0.73 0.38 –1.91 DD model Intercept (⁠|${a}_{DD}$|⁠) 339.52 0.81 418.80 Time (⁠|${g}_T$|⁠) 8.07 0.12 66.59 MAP (⁠|${g}_M$|⁠) –5.25 1.37 –3.84 Time-by-MAP (⁠|${d}_{DD}$|⁠) 0.78 0.23 3.47 Design 2: Random School Assignment PP model Intercept (⁠|${a}_{PP}$|⁠) 103.30 4.78 21.59 Pretest (⁠|${b}_Y$|⁠) 0.72 0.01 50.26 MAP (⁠|${d}_{PP}$|⁠) 0.28 0.51 0.56 DD model Intercept (⁠|${a}_{DD}$|⁠) 334.08 1.14 292.77 Time (⁠|${g}_T$|⁠) 8.70 0.25 34.58 MAP (⁠|${g}_M$|⁠) 0.37 1.62 0.23 Time-by-MAP (⁠|${d}_{DD}$|⁠) 0.33 0.37 0.89 Design 3: Nonrandom Student Assignment PP model Intercept (⁠|${a}_{PP}$|⁠) 107.04 11.26 9.51 Pretest (⁠|${b}_Y$|⁠) 0.71 0.03 22.04 MAP (⁠|${d}_{PP}$|⁠) –0.99 0.84 –1.17 DD model Intercept (⁠|${a}_{DD}$|⁠) 345.80 1.09 315.97 Time (⁠|${g}_T$|⁠) 4.51 0.51 8.84 MAP (⁠|${g}_M$|⁠) –16.08 0.80 –20.08 Time-by-MAP (⁠|${d}_{DD}$|⁠) 5.74 0.60 9.62 Notes: Cutoff for Significance: |t|> 1.96. PP = pretest–posttest; DD = difference-in-differences; MAP = model of assessment and prevention. Open in new tab The school-level and student-level nonrandomized designs would be biased if the wrong model were chosen. Bias in the school-level randomized design would be minimal, given that the estimates only differ by 0.05. Although in the student-level design students were not assigned directly on the basis of the pretest, they were assigned on the basis of a follow-up fall benchmark test, which is highly correlated with the pretest. The nonsignificant findings in the school-level designs (random and nonrandom) are interpreted in the context of relatively small sample sizes for schools (n = 28 in the nonrandomized design and n = 10 in the randomized design). Discussion Social work researchers should use randomized designs and repeated measures designs when possible. However, we acknowledge that the nonequivalent comparison group two-wave design is not going away any time soon. In this design the choice of whether to use PP or DD depends on assumptions made about secular change in the treatment and comparison groups. Both PP and DD have very specific assumptions about α (the fraction of pretest difference that should be subtracted from the posttest difference). As Morgan and Winship (2007) noted, the fact that α is not required to be 1 in PP as it is in DD does not mean it can take on any other value. Campbell and Kenny (1999) noted it is likely that the PP model over- or underadjusts to the extent that the α = |${b}_Y$| assumption, required by the method, is not reflected in the data. In a demonstration using real data, PP and DD approaches led to similar conclusions in the school-level randomized design. However, in nonrandomized designs, such as purposive school and student assignments to conditions, the methods led to drastically different conclusions, signified by the different magnitudes, signs, and t-tests for the PP and DD models in the nonrandom school and student designs. We chose to set aside the issue of covariates for the demonstration, but they may be relevant in the typical research setting. First, covariates in PP can address confounding and reduce expected pretest differences between groups, which could reduce the impact of making the wrong assumption about α. In DD, such variables are unnecessary because the treatment assignment variable absorbs all group differences at pretest. Instead, in DD (including in a two-group, two-wave, design), time-varying covariates that also vary between participants can be included to help improve the validity of the DD assumption about α (Angrist & Pischke, 2009). The following guidelines are suggested primarily by the preceding review of the literature. First, the PP model is likely to be best if assignment is known to be based on a time-varying characteristic including, but not limited to, the pretest or an unmeasured correlate of the pretest. Second, the DD model is likely to be best if assignment is based on fixed-group assignment or assignment on the basis of fixed personal or group characteristics. Whether characteristics are perceived as fixed or time varying should be determined with respect to the study period or the period of time in which participants will be followed. Third, if the researcher believes that the nonrandomly assigned treatment arms are actually members of one group that exhibited extreme scores at pretest, then the PP model is likely to be better. Fourth, if the pretest causes the posttest, the PP model should be used. Fifth, the availability of covariates (a pretest covariate explaining selection for the PP model, or a time and person-varying covariate supporting parallel secular trends for the DD model) might sway the choice in one direction and may even help to make estimates from these methods more similar. Meeting multiple criteria without contradictions should strengthen the argument in favor of the chosen method. These guidelines are broad generalities and may not apply in all nonexperimental contexts. Furthermore, the literature provides little guidance on what to do in studies meeting contradictory criteria or for which the assignment mechanism is unknown. The choice is not clear, for example, in the presence of self-selection, if the factors behind self-selection are not known (van Breukelen, 2013). In self-selected designs, it may be necessary to look for correlates of selection that might suggest either PP or DD. Results may only be trustworthy if both methods lead to similar conclusions or have the same sign and statistical significance (Allison, 1990; van Breukelen, 2013). Although this may seem unlikely, it may be possible if covariates are available to explain assignment or support parallel secular trends. Although randomized designs and designs with multiple pretreatment waves are better, the two-wave design remains a common approach. The guidelines presented here can help researchers choose the most appropriate analysis strategy. Roderick A. Rose, PhD, is assistant professor, School of Social Work, University of Maryland, 525 West Redwood Street, Baltimore, MD 21201; e-mail: [email protected]. Natasha K. Bowen, PhD, is professor, College of Social Work, Ohio State University, Columbus. References Allison , P. D. ( 1990 ). Change scores as dependent variables in regression analysis . Sociological Methodology, 20 , 93 – 114 . Google Scholar Crossref Search ADS WorldCat Allison , P. D. ( 2002 ). Missing data series: Quantitative applications in the social sciences 136 . Thousand Oaks, CA : Sage Publications . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Angrist , J. D. , & Pischke , J.-S. ( 2009 ). Mostly harmless econometrics: An empiricist’s companion . Princeton, NJ : Princeton University Press . Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC Bowen , N. K. , Thompson , A. M. , & Powers , J. D. ( 2012 ). A quasi-experimental test of the Elementary School Success Profile Model of Assessment and Prevention. Journal of the Society for Social Work and Research, 3 , 178 – 196 . doi: 10.5243/jsswr.2012.12 Campbell , D. T. , & Kenny , D. A. ( 1999 ). A primer on regression artifact . New York : Guilford Press . Gillespie , D. F. , & Streeter , C. L. ( 1994 ). Fitting regression models to research questions for analyzing change in nonexperimental research . Social Work Research, 18 , 239 – 245 . Google Scholar Crossref Search ADS WorldCat Hoefer , R. , & Bryant , D. ( 2017 ). A quantitative evaluation of the multi-disciplinary approach to prevention services (MAPS) program to protect children and strengthen families . Journal of Social Service Research, 43 , 459 – 469 . doi: 10.1080/01488376.2017.1295009 Google Scholar Crossref Search ADS WorldCat Crossref Imbens , G. W. , & Wooldridge , J. M. ( 2009 ). Recent developments in the econometrics of program evaluation . Journal of Economic Literature, 47 , 5 – 86 . Google Scholar Crossref Search ADS WorldCat Lee , C. D. ( 2018 ). Social work with groups’ practice ethics and standards: Student confidence and competence . Research on Social Work Practice, 28 , 475 – 481 . Google Scholar Crossref Search ADS WorldCat Morgan , S. L. , & Winship , C. ( 2007 ). Counterfactuals and causal inference: Methods and principles for social research . New York : Cambridge University Press . Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC Ryan , R. M. , & Deci , E. L. ( 2009 ). Promoting self-determined school engagement: Motivation, learning, and well-being. In A. Wigfield & K. R. Wentzel (Eds.), Handbook of motivation at school (pp. 171 – 196 ). New York : Routledge . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Senn , S. ( 2006 ). Change from baseline and analysis of covariance revisited . Statistics in Medicine, 25 , 4334 – 4344 . Google Scholar Crossref Search ADS WorldCat Shadish , W. R. , Cook , T. D. , & Campbell , D. T. ( 2002 ). Experimental and quasi-experimental designs for generalized causal inference . New York : Houghton Mifflin . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC van Breukelen , G.J.P. ( 2013 ). ANCOVA versus CHANGE from baseline in non-randomized studies: The difference . Multivariate Behavioral Research, 48 , 895 – 922 . Google Scholar Crossref Search ADS WorldCat © 2019 National Association of Social Workers This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Greif, Geoffrey, L;Woolley, Michael, E

doi: 10.1093/swr/svz016pmid: N/A

Abstract Each marriage is the beginning of multiple intergenerational in-law relationships. Drawing on a survey of 351 women about their relationships with their mothers-in-law (MILs), this article reports on the impact of family triangles and boundary ambiguity on the quality of the relationship. In a three-block regression analysis, predicting a seven-item relationship quality scale, the first block included six demographic variables. The second and third blocks included three survey items each of reported relationship triangles and boundary ambiguity in a woman’s interactions with her MIL. Results revealed that although the couple having children predicted worse relationship quality in the demographic block, that result was no longer significant once family triangle measures were entered. All three measures of boundary ambiguity were also significant. The three measures of boundary ambiguity eclipsed two of the three family triangles when entered in the third block, leaving only the daughter-in-law feeling comfortable going directly to the MIL about important issues still significant. When a couple marries, a new family is formed that is interconnected with the two existing families—each of whom is changed by the addition of new members—leading to the creation of three new families (Morr Serewicz & Canary, 2008). The new in-law relationships that are formed hold the potential for great joy, yet may also generate significant familial strain, with the potential to negatively affect the couple and multiple generations in all three families (Merrill, 2007). With more than 2 million couples marrying annually in the United States (Centers for Disease Control and Prevention, National Center for Health Statistics, 2015), advancing our understanding of in-law relationships can inform practice focused on improving interactions between couples, parents-in-law, and grandchildren. Women are often considered central in family relationships. For example, research has found they frequently play the role of kin keeper in the family, demonstrating a greater investment than men in intergenerational relationships (Rittenour, 2012; Santos & Levitt, 2007), sibling relationships (Greif & Woolley, 2016; Mikkelson, 2014), and child rearing (Medved, 2014). Women also often devote more time to caretaking of the elderly (Chesley & Poppie, 2009) and other family members (Caputo, Pavalko, & Hardy, 2016). With the multiple roles often occupied by mothers and mothers-in-law (MILs), the daughter-in-law (DIL) and her spouse (we studied heterosexual marriages of DILs) have the challenge of finding an identity for their newly formed family unit and a place for themselves within each of their now expanded families of origin. Clinicians have supported the need for a newly married couple to build a boundary around their relationship so that it can develop while also, in a culturally appropriate way, including the newly formed in-law relationships and preventing unhelpful interference from both sets of parents (Silverstein, 1990). For a DIL, our focus here, that means building a satisfactory relationship with her parents-in-law—and perhaps the MIL in particular, given her often central role in the family—to both make life easier for her spouse and herself and to increase the likelihood of having a satisfactory marriage (Bryant, Conger, & Meehan, 2001). This often requires a balancing act in the potential triangle formed between the DIL, her spouse, and her MIL (Morr Serewicz, 2008)—a balancing act that may also include other members of the family system, most notably her father-in-law (FIL). How these roles are expected to be shared between generations and between men and women may look different from one cultural group to the next. For example, societal oppression may require certain family groups to form closer relationships with greater collectivity (Smith & Landor, 2018). Striving for independence may be seen differently than it is among nonoppressed groups. The purpose of this article, which draws on a survey of 351 women married to men and with a MIL, is to describe the factors related to a DIL reporting a positive relationship with her MIL. Background Unless the DIL, her husband, and her new MIL have been in communication about their relational and interactional expectations for each other, this potential lack of clarity might lead to boundary ambiguity (see, for example, Turner, Young, & Black, 2006). Boundary ambiguity has been used in research to understand the shifting relationships in divorce, remarriage, and stepfamily situations (Carroll, Olson, & Buckmiller, 2007). Boundary ambiguity can be defined as “not knowing who is in or out of your family or relationship. . . . A high degree of boundary ambiguity becomes a risk factor, which predicts depression, somatic symptoms, and family conflict” (Boss, 2006, p. 12). Examples of when family conflict may occur if there is a lack of clarity between a DIL, her spouse, and MIL include who makes social plans for the couple, who will care for the DIL’s spouse (the MIL’s son or daughter) if he or she has a serious disease, and what role a parent-in-law might have in the couple’s child rearing. These relationships and communications may be even more ambiguous if there are differences in each of their cultural or experiential backgrounds. For example, research has indicated that such differences could come to the fore if the marriage is interracial and family members are not comfortable with each other (Campbell & Herman, 2015) or if the couple grew up in families from different socioeconomic classes (Ross, 1995). Speaking specifically about the demands from family on the marital relationship, Minuchin (1974) wrote, “The capacity for complementary accommodation between spouses requires freedom from interference by in-laws and children, and sometimes by the extrafamilial. The development of skills for negotiating with peers, learned among siblings, requires noninterference from parents”(p. 54). Thus, as a newlywed couple attempts to build multiple new relationships, including with their families of origin, a DIL may struggle to form a boundary around her relationship with her spouse. If she and her spouse have a child, a new set of difficulties can arise for the couple. When a DIL does not know where she stands with her MIL, she is likely to feel uncomfortable with her. Boundary ambiguity can be conceptually linked to triangulation: Visher and Visher (1982) observed that “unproductive triangles involve three individuals in a struggle so that clear dyadic relationships are not possible” (p. 348). When boundaries are unclear, cross-generation (for example, MIL, son, and DIL) coalitions may form that lead to triangulation (Horsley, 1996). Such triangulation can develop in the women’s relationship and may have the potential to negatively affect all three relationships if the DIL and her spouse fail to maintain appropriate boundaries with the spouse’s mother (and with his father) and thus allow interference in their marital dyad. Despite the negative societal tropes and ubiquitous jokes about struggles with MILs, research has revealed that many women consider their MIL to be both close and an important source of support to them (Rittenour, 2012; Santos & Levitt, 2007; Serovich & Price, 1994). Serovich and Price (1994) drew a sample of 618 children-in-law from the National Survey of Parents and Households. The participants reported a great deal of satisfaction with in-laws; DILs were equally satisfied with their relationship with their MIL and with their FIL. Santos and Levitt (2007) surveyed 170 children-in-law (120 were DILs) and asked them to place their parents-in-law in a network diagram indicating how close and important they were. A majority (58%) placed their MIL in a position that indicated they were considered a significant part of their network (49% rated their FIL similarly). Rittenour (2012)—using a sample of 624 DILs from Listserv groups and Web sites that focused on family and relationship issues—found that DILs were more satisfied with their relationships with their MIL when the MIL was open about family struggles and issues (such as problems tied to past family relationships) and when this was accompanied by generally supportive communi-cation. Inclusive gestures demonstrated by the MIL (that is, sharing personal information) conveyed acceptance of the DIL and promoted greater trust in their relationship (Rittenour, 2012). Morr Serewicz and Canary (2008) also found that positive disclosures by parents-in-law are related to closeness. Sharing of personal information may be interpreted as drawing an inclusive boundary around the DIL and MIL relationship. For DILs, being shown respect and having shared values and goals were the most frequently cited reasons for getting along with their MIL, among a rural sample of 55 DILs (Marotz-Baden & Cowan, 1987). Respect for boundaries around the marital relationship (that is, lack of interference by the MIL) was also cited as a reason. Furthermore, if a DIL was struggling with her MIL, she was most likely to turn to the other side of that triangle and seek advice from her husband. Both the amount of time spent together and the number of years married have also been related to closeness in a DIL–MIL relationship. Santos and Levitt (2007) found that more frequent contact and a longer marriage predicted a better relationship. Fowler and Rittenour (2017) asked married college students to complete a survey about their relationship with their parents-in-law. Students who were not married could approach others they knew to complete the survey. The majority of the 179 respondents were DILs. Of these DILs, the majority reported on their MIL. Taking a life span approach, Fowler and Rittenour found that the frequency of interaction with a parent-in-law was positively correlated with measures of closeness. Contrary to Santos and Levitt, however, the number of years married was negatively correlated with in-law relationships. Fowler and Rittenour’s findings were explained by the strains in the relationships that emerge from the emotional toll of caring for aging parents and parents-in-law. The data were not analyzed by either gender of the respondents or of the in-law being described, so it is difficult to know the extent to which responses from sons-in-law or feelings about FILs influenced the findings. Turner et al. (2006) helped to further refine our understanding of DIL–MIL relationships. Out of 23 DILs who participated in focus groups and individual interviews, many described feelings of discomfort and being eager to please their MIL while trying to maintain autonomy and independence (a form of boundary ambiguity). Many felt closer to their MIL before the marriage only to feel shut out or disregarded by her postnuptially and over the years, which illustrates how boundaries in the relationship can shift over time. Other research has also shown how boundaries are dynamic between generations. Fischer (1983) found that a shift occurred between the DIL and the MIL with the birth of a child such that the DIL gravitated away from her MIL and toward her own mother for support and assistance. This finding can be contrasted with Santos and Levitt’s (2007) study, in which children-in-law were asked to place the MIL in the diagram indicating closeness. Those respondents with children of their own indicated greater closeness with their MIL than those respondents who did not have children. Silverstein (1990), Horsley (1996), and Meyerstein (1996)—all clinicians—focused on boundaries in their work with families struggling with in-law issues. Silverstein suggested that children-in-law who were nurtured well and were allowed autonomy by their parents when young are more capable of forming nurturing relationships with their parents-in-law and can avoid boundary issues. Horsley, in a book dedicated to in-law relationships, drew on Murray Bowen’s theory to help explain how family members differentiate from their family emotional system and thus learn to clarify boundaries. Meyerstein suggested ways of working with either parents-in-law or children-in-law. One of her case examples centered on a DIL who believed that her husband was more closely tied to his parents than to her. Meyerstein’s intervention focused, in part, on drawing a boundary around the younger couple and improving their communication. This resulted in the husband’s increased differentiation from his parents and led to greater clarification of the intergenerational boundaries. As the research indicates, the DIL–MIL relationship is not easily characterized. Whereas many DILs find the relationship close and satisfying, others struggle and pull away as their parents-in-law age or at the birth of their child. However, feeling included by the MIL and sharing similar values can help the DIL feel closer to the MIL, as can frequent contact. When boundaries between the MIL and the DIL are ambiguous, it can be harder for DILs and MILs to achieve closeness. Building on this prior research, we use both the concept of boundary ambiguity and the dynamics found within triangulated relationships in families—linked to a myriad of mainstream family therapy theories—to explore the correlates of the quality surrounding the DIL–MIL relationship. We also use these concepts to suggest, based on our findings, how to work with this complex family relationship. Method Data analyzed for this study were collected with a 114-item quantitative survey that we had developed, piloted in live interviews with the help of MSW students enrolled in an advanced-level research course, and then formatted in the Qualtrics platform (see also Woolley & Greif, 2019). Qualtrics is an online survey platform in which researchers may design, format, and administer their own e-mail surveys. Qualtrics has an infrastructure to administer surveys for social science researchers to a sample of respondents who meet screening criteria for inclusion in a study. Supported by a grant from the University of Maryland School of Social Work, we hired Qualtrics to administer this survey in 2017. Initial screening questions assured that respondents met the study criteria, including the following: (a) being female, (b) being married to a man, (c) having a living MIL, and (d) having or having had interpersonal interactions with that MIL. Respondents who participate on the Qualtrics Panel are compensated, sometimes with gift certificates. For the current study, we focused on the DIL–MIL relationships in heterosexual marriages. Other research we are conducting is exploring gay and lesbian couples’ in-law relationships and other in-law relationships (Greif & Woolley, in press). Sample Qualtrics made the survey available to their sample frame of potential respondents on July 20, 2017. Over the next 12 days, 1,042 potential DIL respondents initiated the survey, with the targeted sample of at least 250 achieved on August 1, 2017. The four screening questions screened out 691 of those 1,042 as not meeting all four study eligibility criteria. The final sample included 351 DILs with a mean age of 39.6 years (SD = 10.7) and an age range from 18 to 69. The MILs’ mean age was 67.4 years (SD = 11.1) and their ages ranged from 43 to 94. The mean length of time these DILs had been married into the family was 12.7 years (SD = 10.4). Of the 351 DILs, 250 (71%) reported being European American or white, 35 (10%) reported being Asian, 23 (6%) Latina or Hispanic, 14 (4%) black or African American, three (1%) American Indian or Alaska Native, 1 (<1%) Native Hawaiian or Pacific Islander, three (1%) multiple races or ethnicities, and 22 (7%) another race. Respondents had an average of 15 years (SD = 2.5) of education (three years beyond a high school degree) with a range of six to 22 years. In terms of combined household income, 54 (15%) endorsed $40,000 or below, 103 (29%) between $41,000 and $80,000, 104 (30%) $81,000 to $120,000, 49 (14%) $121,000 to $160,000, and 41 (12%) above $160,000. Analytic Strategy The central goal of the current study was to assess the impact of family relationship triangles and the presence of boundary ambiguity on the quality of the relationship between a DIL and her MIL. The current regression analysis therefore included 11 individual survey item questions used as independent variables predicting a seven-survey-item latent variable scale measure of the quality of the respondent’s relationship with her MIL. Those 11 independent variables were organized into three blocks. The first block measured sociodemographics. The second block comprised three individual survey item variables assessing triangulated family relationship interactions including the DIL, MIL, and one other family member. The final block included three survey item variables in which the DIL reported on relationship indicators of boundary ambiguity with the MIL. The survey was set up to require respondents to provide a response to all questions. This resulted in no missing responses in the data. However, some questions had “Does not apply” as a response option when the nature of the question suggested the need for such a response choice. These responses were coded for the current analysis as missing values with an available-case analysis (pairwise deletion) missing values strategy used in the regression model. Measures The outcome variable was the mean of a set of seven survey items, whereas all model independent variable predictors were single survey items. With the exception of one dichotomous item (yes, no) and several sociodemographic survey questions, such as length of time married or years of education, all other survey items had the same five-point Likert scale response set ranging from 0 = strongly disagree to 4 = strongly agree. Dependent Variable Seven survey items, five measuring positive aspects of the daughter’s reports of her relationship with her MIL and two measuring negative aspects (reverse coded for the analyses), were used as indicators of a latent variable assessing relationship quality. The seven items, which all used the five-point response scale detailed in the previous section, included (1) “My mother-in-law and I have a close relationship”; (2) “Overall, I admire my mother-in-law”; (3) “I can ask my mother-in-law for advice”; (4) “I enjoy spending time with my mother-in-law”; (5) “I trust my mother-in-law”; (6) “I avoid my mother-in-law”; and (7) “I have problematic conflicts with my mother-in-law.” This scale had strong internal consistency reliability (Cronbach’s α = .93). An exploratory factor analysis (EFA) (maximum likelihood extraction) with oblique rotation (Promax) yielded a one-factor solution that captured 74.8% of the variance across the seven items with factor loadings ranging from .64 (problematic conflicts) to .94 (enjoy spending time). Together, the internal consistency reliability and EFA results indicate a seven-item latent measure with strong psychometric properties. Independent Variables There were 11 independent variables included in the current analysis. Five variables measured demographic characteristics of the respondent and her marriage. Race or ethnicity was dummy coded with black, Asian, Latino or Hispanic, and “other race or ethnicity” (which included Native American or Alaska Native, Native Hawaiian or other Pacific Islander, multiple races or ethnicities, and another race or ethnicity) entered in the model, and white respondents were the reference group. Time married was the years the DIL had been married to her husband. Years of education was self-reported, with 12 being defined as high school completion and 16 as college completion. DIL household income range was measured by the following item: In what range is your combined household income? Response options included 1 = up to $40,000; 2 = $41,000 to $80,000; 3 = $81,000 to $120,000; 4 = $121,000 to $160,000; and 5 = above $160,000. Whether the respondent has children was solicited with the question: “Do you have any children from this marriage?” (1 = yes, 0 = no). Three survey items each measured factors related to family relationship triangles and boundary ambiguity on the previously mentioned five-point response scale. The first of the three measuring family triangles was MIL closeness, which was assessed with DIL’s agreement to the survey item “My mother-in-law is closer to another child-in-law than me.” Second triangle variable, talking directly, was measured with the survey item “I usually speak with my mother-in-law directly about important matters between us rather than going to my husband first.” The third triangle measure, hindrance, was assessed by agreement with the following survey item: “My marriage feels hindered by my mother-in-law.” The first of the three variables measuring boundary ambiguity was MIL withholding, assessed by DIL’s agreement with the following statement: “My mother-in-law is withholding from me.” DIL anxiety was elicited with the survey item “My mother-in-law increases my anxiety or makes me nervous.” The final boundary ambiguity variable, different parenting philosophies, was measured as DIL’s agreement with the following statement: “My mother-in-law and I have different parenting philosophies.” Results Table 1 details descriptive statistics for 12 study variables in the current analysis (race or ethnicity was detailed in the sample description). Relationship quality, the dependent variable, revealed that on average across the seven survey items, DILs reported being in between neutral and agreeing with those items (M = 2.55). Respondents were, on average, about 40 years old, had completed three years of college with household incomes in the middle range, and were likely to have children from the current marriage. Across the family triangles and boundary ambiguity survey items, the least endorsed was marriage hindrance (M = 1.09) by the MIL; the most endorsed was having a different parenting philosophy. Table 2 presents correlations between study variables. Of note, relationship quality was correlated with all triangle and ambiguity measures and moderate to strong correlations were found among the six triangle and boundary ambiguity measures. Table 1 Descriptive Statistics of Study Variables Variable . Range . M . SD . Skewness . Kurtosis . Relationship quality 0–4 2.55 1.05 –0.67 –0.30 Age 18–69 39.55 10.69 0.54 –0.49 Time married (in years) 0–45 12.65 10.36 0.94 0.17 Years of education 6–22 15.30 2.54 –0.68 2.10 Income range 1–5 2.77 1.21 0.32 –0.75 Children 0–1 0.62 0.49 –0.51 –1.75 MIL closer 0–4 1.73 1.43 0.25 –1.28 Talk directly 0–4 1.81 1.33 0.06 –1.21 Relationship hindered 0–4 1.09 1.20 0.96 –0.04 MIL withholding 0–4 1.32 1.14 0.62 –0.39 DIL anxiety 0–4 1.51 1.31 0.49 –0.94 Different parenting philosophies 0–4 2.40 1.18 –0.32 –0.85 Variable . Range . M . SD . Skewness . Kurtosis . Relationship quality 0–4 2.55 1.05 –0.67 –0.30 Age 18–69 39.55 10.69 0.54 –0.49 Time married (in years) 0–45 12.65 10.36 0.94 0.17 Years of education 6–22 15.30 2.54 –0.68 2.10 Income range 1–5 2.77 1.21 0.32 –0.75 Children 0–1 0.62 0.49 –0.51 –1.75 MIL closer 0–4 1.73 1.43 0.25 –1.28 Talk directly 0–4 1.81 1.33 0.06 –1.21 Relationship hindered 0–4 1.09 1.20 0.96 –0.04 MIL withholding 0–4 1.32 1.14 0.62 –0.39 DIL anxiety 0–4 1.51 1.31 0.49 –0.94 Different parenting philosophies 0–4 2.40 1.18 –0.32 –0.85 Note: Sample included 351 daughters-in-law (DILs) reporting about their relationships with their mothers-in-law (MILs). Open in new tab Table 1 Descriptive Statistics of Study Variables Variable . Range . M . SD . Skewness . Kurtosis . Relationship quality 0–4 2.55 1.05 –0.67 –0.30 Age 18–69 39.55 10.69 0.54 –0.49 Time married (in years) 0–45 12.65 10.36 0.94 0.17 Years of education 6–22 15.30 2.54 –0.68 2.10 Income range 1–5 2.77 1.21 0.32 –0.75 Children 0–1 0.62 0.49 –0.51 –1.75 MIL closer 0–4 1.73 1.43 0.25 –1.28 Talk directly 0–4 1.81 1.33 0.06 –1.21 Relationship hindered 0–4 1.09 1.20 0.96 –0.04 MIL withholding 0–4 1.32 1.14 0.62 –0.39 DIL anxiety 0–4 1.51 1.31 0.49 –0.94 Different parenting philosophies 0–4 2.40 1.18 –0.32 –0.85 Variable . Range . M . SD . Skewness . Kurtosis . Relationship quality 0–4 2.55 1.05 –0.67 –0.30 Age 18–69 39.55 10.69 0.54 –0.49 Time married (in years) 0–45 12.65 10.36 0.94 0.17 Years of education 6–22 15.30 2.54 –0.68 2.10 Income range 1–5 2.77 1.21 0.32 –0.75 Children 0–1 0.62 0.49 –0.51 –1.75 MIL closer 0–4 1.73 1.43 0.25 –1.28 Talk directly 0–4 1.81 1.33 0.06 –1.21 Relationship hindered 0–4 1.09 1.20 0.96 –0.04 MIL withholding 0–4 1.32 1.14 0.62 –0.39 DIL anxiety 0–4 1.51 1.31 0.49 –0.94 Different parenting philosophies 0–4 2.40 1.18 –0.32 –0.85 Note: Sample included 351 daughters-in-law (DILs) reporting about their relationships with their mothers-in-law (MILs). Open in new tab Table 2 Correlation Matrix of Study Variables Variable . Relation- ship Quality . Age\end . Time Married\end . Years of Educa- tion\end . Income Range\end . Children\end . MIL Close- ness\end . Talking Directly\end . Relation- ship Hindrance\end . MIL With- holding\end . DIL Anxiety\end . Age .01 Time married –.02 .82* Years of education –.07 –.02 –.07 Income range –.05 .16** .07 .33** Children –.13* .11* .30* –.02 .10 MIL closeness –.46** .06 .10 .00 .04 .11 Talking directly .41** .12* .09 –.10 –.04 .03 –.23** Relationship hindrance –.50** –.05 –.04 –.04 .04 .07 –.31** –.24** MIL withholding –.54** –.07 –.03 –.01 –.04 .12* .45** –.18** .48** DIL anxiety –.67** –.12* –.12* .00 –.04 .02 .35** –.28** .60** .56** Different parenting philosophies –.58** .04 –.13* .06 –.06 –.05 .39** –.39** .35** .37** .48** Variable . Relation- ship Quality . Age\end . Time Married\end . Years of Educa- tion\end . Income Range\end . Children\end . MIL Close- ness\end . Talking Directly\end . Relation- ship Hindrance\end . MIL With- holding\end . DIL Anxiety\end . Age .01 Time married –.02 .82* Years of education –.07 –.02 –.07 Income range –.05 .16** .07 .33** Children –.13* .11* .30* –.02 .10 MIL closeness –.46** .06 .10 .00 .04 .11 Talking directly .41** .12* .09 –.10 –.04 .03 –.23** Relationship hindrance –.50** –.05 –.04 –.04 .04 .07 –.31** –.24** MIL withholding –.54** –.07 –.03 –.01 –.04 .12* .45** –.18** .48** DIL anxiety –.67** –.12* –.12* .00 –.04 .02 .35** –.28** .60** .56** Different parenting philosophies –.58** .04 –.13* .06 –.06 –.05 .39** –.39** .35** .37** .48** Notes: MIL = mother-in-law; DIL = daughter-in-law. *p < .05. **p < .01. Open in new tab Table 2 Correlation Matrix of Study Variables Variable . Relation- ship Quality . Age\end . Time Married\end . Years of Educa- tion\end . Income Range\end . Children\end . MIL Close- ness\end . Talking Directly\end . Relation- ship Hindrance\end . MIL With- holding\end . DIL Anxiety\end . Age .01 Time married –.02 .82* Years of education –.07 –.02 –.07 Income range –.05 .16** .07 .33** Children –.13* .11* .30* –.02 .10 MIL closeness –.46** .06 .10 .00 .04 .11 Talking directly .41** .12* .09 –.10 –.04 .03 –.23** Relationship hindrance –.50** –.05 –.04 –.04 .04 .07 –.31** –.24** MIL withholding –.54** –.07 –.03 –.01 –.04 .12* .45** –.18** .48** DIL anxiety –.67** –.12* –.12* .00 –.04 .02 .35** –.28** .60** .56** Different parenting philosophies –.58** .04 –.13* .06 –.06 –.05 .39** –.39** .35** .37** .48** Variable . Relation- ship Quality . Age\end . Time Married\end . Years of Educa- tion\end . Income Range\end . Children\end . MIL Close- ness\end . Talking Directly\end . Relation- ship Hindrance\end . MIL With- holding\end . DIL Anxiety\end . Age .01 Time married –.02 .82* Years of education –.07 –.02 –.07 Income range –.05 .16** .07 .33** Children –.13* .11* .30* –.02 .10 MIL closeness –.46** .06 .10 .00 .04 .11 Talking directly .41** .12* .09 –.10 –.04 .03 –.23** Relationship hindrance –.50** –.05 –.04 –.04 .04 .07 –.31** –.24** MIL withholding –.54** –.07 –.03 –.01 –.04 .12* .45** –.18** .48** DIL anxiety –.67** –.12* –.12* .00 –.04 .02 .35** –.28** .60** .56** Different parenting philosophies –.58** .04 –.13* .06 –.06 –.05 .39** –.39** .35** .37** .48** Notes: MIL = mother-in-law; DIL = daughter-in-law. *p < .05. **p < .01. Open in new tab The results from the regression model are presented in Table 3. Results revealed that although the couple having children predicted worse relationship quality in the demographic block, the result was no longer significant once family triangle measures were entered. In addition, the sociodemographic variables only accounted for 3% of variance in relationship quality. When the three family triangle measures were entered into the model, 38% of the variance in the outcome was captured. All three of those family triangle variables were significant in block 2. When the DIL reported that her MIL was closer to another child-in-law, the DIL also reported a drop in relationship quality on average of about one-third of a step on the response scale (β = –0.31). Similarly, when the DIL reported that her marriage was hindered by her MIL, the DIL also reported about one-third of a step drop (β = –0.31) in relationship quality. However, when the DIL reported that she speaks directly with her MIL about something important (as opposed to speaking first with her husband), the relationship quality was higher by about a quarter of a step in the response scale (β = 0.23). Table 3 Regression Model Predicting DIL Relationship Quality with MIL . Block 1: DemographicsR2 = .03 . . Block 2: Family TrianglesR2= .38 . . Block 3: Relationship AmbiguityR2= .59 . Variable . B . SE . |$\boldsymbol{\beta}$| . t . . B . SE . |$\boldsymbol{\beta}$| . t . . B . SE . |${\boldsymbol\beta}$| . t . Constant 3.14** 0.46 6.78 3.34** 0.44 7.61 3.95** 0.41 9.59 Black 0.26 0.29 0.05 0.89 0.32 0.26 0.06 1.24 0.06 0.24 0.01 0.24 Asian 0.04 0.20 0.01 0.20 0.11 0.19 0.03 0.59 0.20 0.17 0.06 1.17 Hispanic –0.05 0.26 –0.01 –0.19 0.16 0.22 0.04 0.75 –0.10 0.20 –0.02 –0.53 Other race or ethnicity –0.22 0.21 –0.06 –1.06 0.70 0.18 0.02 0.39 0.28 0.16 0.08 1.73 Age 0.00 0.01 0.18 0.18 0.00 0.01 –0.02 –0.16 0.00 0.01 –0.09 –0.01 Time married 0.00 0.01 0.01 0.07 0.00 0.01 0.03 0.27 –0.01 0.01 –0.09 –0.98 Years of education –0.03 0.02 –0.06 –1.08 –0.01 0.01 –0.03 –0.60 0.01 0.02 0.03 0.50 Income range –0.02 0.05 –0.02 –0.39 –0.03 0.04 –0.03 –0.56 –0.08 0.04 –0.10 –1.99 Children –0.32** 0.13 –0.15 –2.51 –0.18 0.12 –0.09 –1.59 –0.12 0.12 –0.05 –0.99 MIL closeness –0.22** 0.04 –0.31 –5.68 –0.07 0.04 –0.11 –1.92 Talking directly 0.17** 0.04 0.23 4.36 0.10** 0.04 0.14 2.75 Relationship hindrance –0.27** 0.05 –0.31 –5.76 0.04 0.05 0.05 0.82 MIL withholding –0.19** 0.06 –0.21 –3.19 DIL anxiety –0.28** 0.05 –0.36 –5.61 Different parenting philosophies –0.24** 0.05 –0.28 –5.12 . Block 1: DemographicsR2 = .03 . . Block 2: Family TrianglesR2= .38 . . Block 3: Relationship AmbiguityR2= .59 . Variable . B . SE . |$\boldsymbol{\beta}$| . t . . B . SE . |$\boldsymbol{\beta}$| . t . . B . SE . |${\boldsymbol\beta}$| . t . Constant 3.14** 0.46 6.78 3.34** 0.44 7.61 3.95** 0.41 9.59 Black 0.26 0.29 0.05 0.89 0.32 0.26 0.06 1.24 0.06 0.24 0.01 0.24 Asian 0.04 0.20 0.01 0.20 0.11 0.19 0.03 0.59 0.20 0.17 0.06 1.17 Hispanic –0.05 0.26 –0.01 –0.19 0.16 0.22 0.04 0.75 –0.10 0.20 –0.02 –0.53 Other race or ethnicity –0.22 0.21 –0.06 –1.06 0.70 0.18 0.02 0.39 0.28 0.16 0.08 1.73 Age 0.00 0.01 0.18 0.18 0.00 0.01 –0.02 –0.16 0.00 0.01 –0.09 –0.01 Time married 0.00 0.01 0.01 0.07 0.00 0.01 0.03 0.27 –0.01 0.01 –0.09 –0.98 Years of education –0.03 0.02 –0.06 –1.08 –0.01 0.01 –0.03 –0.60 0.01 0.02 0.03 0.50 Income range –0.02 0.05 –0.02 –0.39 –0.03 0.04 –0.03 –0.56 –0.08 0.04 –0.10 –1.99 Children –0.32** 0.13 –0.15 –2.51 –0.18 0.12 –0.09 –1.59 –0.12 0.12 –0.05 –0.99 MIL closeness –0.22** 0.04 –0.31 –5.68 –0.07 0.04 –0.11 –1.92 Talking directly 0.17** 0.04 0.23 4.36 0.10** 0.04 0.14 2.75 Relationship hindrance –0.27** 0.05 –0.31 –5.76 0.04 0.05 0.05 0.82 MIL withholding –0.19** 0.06 –0.21 –3.19 DIL anxiety –0.28** 0.05 –0.36 –5.61 Different parenting philosophies –0.24** 0.05 –0.28 –5.12 Notes: DIL = daughter-in-law; MIL = mother-in-law. Sample included 351 women surveyed about their relationship with their MIL. Outcome variable was the mean of seven survey items assessing both positive and negative qualities of the DIL–MIL relationship. *p < .05. **p < .01. Open in new tab Table 3 Regression Model Predicting DIL Relationship Quality with MIL . Block 1: DemographicsR2 = .03 . . Block 2: Family TrianglesR2= .38 . . Block 3: Relationship AmbiguityR2= .59 . Variable . B . SE . |$\boldsymbol{\beta}$| . t . . B . SE . |$\boldsymbol{\beta}$| . t . . B . SE . |${\boldsymbol\beta}$| . t . Constant 3.14** 0.46 6.78 3.34** 0.44 7.61 3.95** 0.41 9.59 Black 0.26 0.29 0.05 0.89 0.32 0.26 0.06 1.24 0.06 0.24 0.01 0.24 Asian 0.04 0.20 0.01 0.20 0.11 0.19 0.03 0.59 0.20 0.17 0.06 1.17 Hispanic –0.05 0.26 –0.01 –0.19 0.16 0.22 0.04 0.75 –0.10 0.20 –0.02 –0.53 Other race or ethnicity –0.22 0.21 –0.06 –1.06 0.70 0.18 0.02 0.39 0.28 0.16 0.08 1.73 Age 0.00 0.01 0.18 0.18 0.00 0.01 –0.02 –0.16 0.00 0.01 –0.09 –0.01 Time married 0.00 0.01 0.01 0.07 0.00 0.01 0.03 0.27 –0.01 0.01 –0.09 –0.98 Years of education –0.03 0.02 –0.06 –1.08 –0.01 0.01 –0.03 –0.60 0.01 0.02 0.03 0.50 Income range –0.02 0.05 –0.02 –0.39 –0.03 0.04 –0.03 –0.56 –0.08 0.04 –0.10 –1.99 Children –0.32** 0.13 –0.15 –2.51 –0.18 0.12 –0.09 –1.59 –0.12 0.12 –0.05 –0.99 MIL closeness –0.22** 0.04 –0.31 –5.68 –0.07 0.04 –0.11 –1.92 Talking directly 0.17** 0.04 0.23 4.36 0.10** 0.04 0.14 2.75 Relationship hindrance –0.27** 0.05 –0.31 –5.76 0.04 0.05 0.05 0.82 MIL withholding –0.19** 0.06 –0.21 –3.19 DIL anxiety –0.28** 0.05 –0.36 –5.61 Different parenting philosophies –0.24** 0.05 –0.28 –5.12 . Block 1: DemographicsR2 = .03 . . Block 2: Family TrianglesR2= .38 . . Block 3: Relationship AmbiguityR2= .59 . Variable . B . SE . |$\boldsymbol{\beta}$| . t . . B . SE . |$\boldsymbol{\beta}$| . t . . B . SE . |${\boldsymbol\beta}$| . t . Constant 3.14** 0.46 6.78 3.34** 0.44 7.61 3.95** 0.41 9.59 Black 0.26 0.29 0.05 0.89 0.32 0.26 0.06 1.24 0.06 0.24 0.01 0.24 Asian 0.04 0.20 0.01 0.20 0.11 0.19 0.03 0.59 0.20 0.17 0.06 1.17 Hispanic –0.05 0.26 –0.01 –0.19 0.16 0.22 0.04 0.75 –0.10 0.20 –0.02 –0.53 Other race or ethnicity –0.22 0.21 –0.06 –1.06 0.70 0.18 0.02 0.39 0.28 0.16 0.08 1.73 Age 0.00 0.01 0.18 0.18 0.00 0.01 –0.02 –0.16 0.00 0.01 –0.09 –0.01 Time married 0.00 0.01 0.01 0.07 0.00 0.01 0.03 0.27 –0.01 0.01 –0.09 –0.98 Years of education –0.03 0.02 –0.06 –1.08 –0.01 0.01 –0.03 –0.60 0.01 0.02 0.03 0.50 Income range –0.02 0.05 –0.02 –0.39 –0.03 0.04 –0.03 –0.56 –0.08 0.04 –0.10 –1.99 Children –0.32** 0.13 –0.15 –2.51 –0.18 0.12 –0.09 –1.59 –0.12 0.12 –0.05 –0.99 MIL closeness –0.22** 0.04 –0.31 –5.68 –0.07 0.04 –0.11 –1.92 Talking directly 0.17** 0.04 0.23 4.36 0.10** 0.04 0.14 2.75 Relationship hindrance –0.27** 0.05 –0.31 –5.76 0.04 0.05 0.05 0.82 MIL withholding –0.19** 0.06 –0.21 –3.19 DIL anxiety –0.28** 0.05 –0.36 –5.61 Different parenting philosophies –0.24** 0.05 –0.28 –5.12 Notes: DIL = daughter-in-law; MIL = mother-in-law. Sample included 351 women surveyed about their relationship with their MIL. Outcome variable was the mean of seven survey items assessing both positive and negative qualities of the DIL–MIL relationship. *p < .05. **p < .01. Open in new tab When the three measures of boundary ambiguity were added to the model, the model captured 59% of the variance in relationship quality. Furthermore, two of the three previously significant predictor measures of family triangles were no longer significant, leaving only the positive triangle measure—speaking directly to the MIL about important matters. The three measures of boundary ambiguity were all significant. When the DIL reported the MIL as withholding, that predicted a fifth of a step in the response scale drop in relationship quality (β = –0.21). When the DIL agreed that her MIL made her anxious or nervous, that predicted a little more than a third of a step drop in relationship quality (β = –0.36). Finally, when the DIL endorsed having a different parenting philosophy than her MIL, that was associated with more than a quarter of a step drop in the response scale in relationship quality (β = –0.28). Discussion DILs in our sample reported, on average, a positive relationship with their MILs. The relationship quality scale had a mean of 2.55 (with 2 = neutral and 3 = agree). Serovich and Price (1994) and Santos and Levitt (2007) also reported close relationships. The responses indicate that a minority may be in the neutral range, neither especially close nor distant, or may be struggling with their relationship with their MIL. These “neutrals” may be DILs who have made peace with or figured out how to manage the relationship. Some DILs, for example, manage problems in their relationships by ignoring them (Marotz-Baden & Cowan, 1987). Others make the best of the relationship to keep their spouse happy (Turner et al., 2006). Of course, not every family relationship has to be in a more definite domain, and neutrals may not feel this is problematic unless the DIL specifically wants to be closer to her MIL, but is unable to achieve that. Those who feel that their relationship quality is not good and are struggling with it may be the ones who are most in need of service from social workers. The six variables that were measures of triangulation and boundary ambiguity were strong predictors of relationship quality, whereas the demographic variables were not. The first of the triangle variables, the perception that the MIL was closer with another child-in-law, could be interpreted as predictive for a number of reasons: the DIL might believe she is competing with another child-in-law; feeling not as well accepted as another child-in-law could resurrect feelings the DIL might have from her own experiences as being a less favored sibling (see, for example, Greif & Woolley, 2016); or the DIL’s husband may not be as close with his mother as one of the husband’s married siblings. According to Dunn and Plomin (1991), feelings of parental favoritism occur in more than half of all families. In this scenario, the DIL and her husband, as a unit, might not be as close as another sibling and spouse, so the DIL does not feel as close. Finally, it could be that the DIL does not feel as close with the MIL because of her own interactions with the MIL. If the DIL feels the MIL is hindering the DIL’s marital relationship (the second triangle variable), she is likely to report lower relationship quality. In-law interference has been cautioned against by Meyerstein (1996), Minuchin (1974), and Silverstein (1990). Minuchin and Silverstein warned that couples need to have autonomy from their families of origin, and Meyerstein offered a case where the husband is too close to his mother, placing the DIL at a distance in the triangle. The third triangle variable, speaking directly with the MIL about important matters between them, reinforces the notion that when a third party is not pulled in, the relationship is better. Not being part of a triangle is a positive form of interaction in relation to relationship quality in our sample. Rittenour (2012) found greater closeness between DILs and MILs when the MIL is open about family issues and when communication is supportive (see also Morr Serewicz & Canary, 2008, in relation to supportive communication). In high-quality DIL–MIL relationships, women seem to be able to engage in open communication without including the DIL’s spouse. The three variables related to boundary ambiguity all deal with dyadic relationships, though from a family systems perspective we acknowledge that everyone is influenced by other family members. The DIL reporting that the MIL is withholding and that the MIL makes her feel anxious can be seen as the MIL not including the DIL and thus making her feel unsure about her position in the family. When the boundaries between family members are ambiguous, Boss (2006) suggested, conflict can occur. When the MIL is perceived as withholding, it could be a sign that she feels emotionally unavailable to the DIL. As Turner et al. (2006) noted, DILs often try to please their MILs yet feel discomfort. When someone is withholding, they are harder to read and any attempts to please them may feel awkward and fall short. The third boundary ambiguity variable, concerning differing parenting philosophies, was also an important predictor of relationship quality. In our sample, the couple having children was related to a worse relationship quality, which is consistent with some of the research on the DIL–MIL relationship post-birth (see, for example, Fischer, 1983), but not with other research (see, for example, Santos and Levitt, 2007) where the birth brought the two women closer. The importance that parenting philosophies play in predicting relationship quality is consistent with the central parenting role that women play in many families (Medved, 2014; Rittenour, 2012; Santos & Levitt, 2007). For example, when philosophies differ, the DIL and MIL may be unclear of what their roles are in relation to each other when it comes to child rearing and assisting with child rearing. The MIL may be giving unwanted advice that is not consistent with the DIL’s thinking, which may challenge the role that each woman feels she is supposed to fulfill. The inclusion of the three boundary ambiguity variables in the model in block 3 rendered the two negative family triangle variables no longer significant. One way to interpret this is that a family triangle may include boundary ambiguity elements, which may be key to the triangle dynamics that negatively affect relationship quality. Supporting this hypothesis are the moderate to high correlations among the triangle and ambiguity variables, and that the boundary ambiguity variables are stronger predictors of relationship quality once in the model with the triangle measures. Furthermore, the one triangle variable that revealed less triangulation stayed a significant predictor in the model after the boundary ambiguity variables were entered. If boundary ambiguity in the dyads of a family triangle is a key element of problematic triangles, it seems a measure of direct open communication across one of those dyads would be an indicator of lower levels of ambiguity in that dyad and that such a variable would stay significant despite entering variables measuring boundary ambiguity into the model. Implications Our findings underline the systemic and interconnected nature of the relationship between the DIL, the MIL, and other family members. Social workers engaged in treatment must consider not only the two women, but also the spouse and child of the MIL, other siblings of the spouse and their spouses, the FIL (or female spouse), and the children and grandchildren who are key to the impact of differing parenting philosophies. While we explore dyadic and triadic relationships, it is important to remember that each person is related to others in the family. Boss (2006), when treating boundary ambiguity, suggested listening to the conversational voice of the family and observing their interactions when they are together. Hearing the voice of the family can also mean being aware of the family’s culture, which includes how they assimilate a new child-in-law into the family, how independent the couple should be from the older generation, and what the role of the grandmother (and grandfather) should be. The patriarchal versus matriarchal nature of the family should also be considered in helping to understand how to best help the family function if struggles arise between the DIL and the MIL. For the families in our predominantly white sample, the ability to speak directly with a MIL is one sign of a positive relationship that would diminish triangulation. DILs and MILs can both be encouraged to improve communication between them by directly addressing differences in expectations around inclusion in family events (thereby also addressing the feeling that the DIL is not as accepted as another child-in-law) and around parenting expectations. This might include an acknowledgment of disagreements in parenting philosophies and an attempt to achieve a clearer delineation of what each sees the other’s role as being in supportive parenting and grandparenting. An understanding that both women may have brought different expectations to their in-law roles based on their past experiences in their families of origin may be helpful here. When parenting philosophies differ and the MIL is perceived as withholding, the DIL may wonder what her role is within the boundaries of the family. The DIL may have to reach out to the MIL and try to include her in ways that may change the boundaries in the relationship. Both women may be operating consciously or unconsciously to perpetuate a narrative in which parenting differences cause both the DIL and the MIL to distance themselves from one another and thereby remain unclear where they stand with each other. The DIL and her spouse may need help in drawing a boundary around their relationship (see, for example, Minuchin, 1974; Silverstein, 1990). The boundary marking could include, as suggested, open discussions about differences in child-rearing approaches between the two generations. An open discussion then may help family members understand to what extent they are invited into the family processes. The FIL’s role should also be considered in how he is interacting with the MIL in her relationship with the DIL and how he is establishing his own relationship with the DIL. Spending time together, finding common interests, and opening direct lines of communication, potentially with both parents-in-law, can all work to improve relationship quality and reduce the chances of triangulation. Those actions may help to define the boundaries more clearly, which our findings strongly support need attention. Ultimately, the social worker should inquire about what closeness means for family members, how that closeness may enhance or hinder other family relationships, and how the family members want to interact as a family. The conversation must be conducted in a way that is both culturally sensitive and cognizant of the differing family histories of the DIL and the MIL, which may be driving their interactions. Limitations The current findings need to be interpreted in light of the limitations of this study. The first limitation is that all data were self-reports from DILs and therefore subject to the limitations of individual perceptions and interpretations of social relationships. The second limitation is similar in that data were collected from one family member who reported about her thoughts and feelings, and the perceived feelings of two other members of that family (the respondent’s MIL and husband) who did not provide any of their own data. This may perpetuate a narrative of “blaming” the MIL that is not reflective of the unique role that she often occupies in the family. Third, although the sample size was sufficient for the analyses reported and the effect size is substantial (R2 = .59), this is a cross-sectional sample of DILs gleaned from the survey sample frame established by Qualtrics’s online survey platform. Because of this, caution should be taken in generalizing our findings. However, even with these limitations, we believe our findings are consistent with other scholarly conclusions in this area of limited research and that these findings lend additional aspects of particular relevance to clinicians. Future Research Longitudinal research into the relationship trajectories between parents(-in-law) and adult children(-in-law) across the life span and within the multiple triangular in-law relationships may provide more clues about when families are most at risk for developing problems. In addition, conducting research on a more racially diverse sample than the current sample could aid in understanding differences between races but also in exploring issues that arise with interracial and interethnic marriages, which our data were not able to discern. Finally, concurrently collecting data from multiple members of a family about these understudied family relationships would provide insights not available when analyzing data from one respondent in a family. With adults living longer and with interracial and interethnic marriage on the rise, gaining an up-to-date understanding of these vitally important family relationships can go a long way toward sustaining healthy marriages and promoting well-functioning intergenerational relationships. Geoffrey L. Greif, PhD, is professor and Michael E. Woolley, PhD, is associate professor, University of Maryland School of Social Work, Baltimore. Address correspondence to Geoffrey L. Greif, University of Maryland School of Social Work, 525 West Redwood Street, Baltimore, MD 21201; e-mail: [email protected]. References Boss , P. ( 2006 ). Loss, trauma, and resilience: Therapeutic work with ambiguous loss . New York : W. W. Norton . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Bryant , C. M. , Conger , R. D. , & Meehan , J. M. ( 2001 ). The influence of in-laws on change in marital success . Journal of Marriage and Family, 63 , 614 – 626 . Google Scholar Crossref Search ADS WorldCat Campbell , M. E. , & Herman , M. R. ( 2015 ). Both personal and public: Measuring interethnic marriage attitudes . Journal of Social Issues, 71 , 712 – 732 . Google Scholar Crossref Search ADS WorldCat Caputo , J. , Pavalko , E. K. , & Hardy , M. A. ( 2016 ). The long-term effects of caregiving on women’s health and mortality . Journal of Marriage and Family, 78 , 1382 – 1398 . Google Scholar Crossref Search ADS WorldCat Carroll , J. S. , Olson , C. D. , & Buckmiller , N. ( 2007 ). Family boundary ambiguity: A 30-year review of theory, research, and measurement . Family Relations, 56 , 210 – 230 . Google Scholar Crossref Search ADS WorldCat Centers for Disease Control and Prevention, National Center for Health Statistics. ( 2015 ). National Marriage and Divorce Rate Trends, November 23 . Atlanta : Author . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Chesley , N. , & Poppie , K. ( 2009 ). Assisting parents and in-laws: Gender, type of assistance, and couples’ employment . Journal of Marriage and Family, 71 , 247 – 262 . Google Scholar Crossref Search ADS WorldCat Dunn , J. , & Plomin , R. ( 1991 ). Why are siblings so different? The significance of differences in sibling experiences within the family . Family Process, 30 , 271 – 283 . Google Scholar Crossref Search ADS WorldCat Fischer , L. R. ( 1983 ). Mothers and mothers-in-law . Journal of Marriage and the Family, 45 , 187 – 192 . doi: 10.2307/351307 Google Scholar Crossref Search ADS WorldCat Fowler , C. , & Rittenour , C. ( 2017 ). A life-span approach to children-in-law’s perception of parent-in-law communication . Journal of Family Communication, 17 , 254 – 272 . Google Scholar Crossref Search ADS WorldCat Greif , G. L. , & Woolley , M. ( 2016 ). Adult sibling relationships . New York : Columbia University Press . Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC Greif , G. L. , & Woolley , M. E. (in press). In-law relationships . New York : Oxford University Press . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Horsley , G. C. ( 1996 ). In-laws: A guide to extended family therapy . New York : John Wiley & Sons . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Marotz-Baden , R. , & Cowan , D. ( 1987 ). Mothers-in-law and daughters-in-law: The effect of proximity on conflict and stress . Family Relations, 36 , 385 – 390 . Google Scholar Crossref Search ADS WorldCat Medved , C. E. ( 2014 ). Fathering, caregiving, and masculinity: Stay-at-home fathers and family communication. In K. Floyd & M. T. Morman (Eds.), Widening the family circle: New research on family communication (2nd ed., pp. 115 – 132 ). Los Angeles : Sage Publications . Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC Merrill , D. M. ( 2007 ). Mothers-in-law and daughters-in-law: Understanding the relationship and what makes them friends or foe . Westport, CT : Praeger . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Meyerstein , I. ( 1996 ). A systemic approach to in-law dilemmas . Journal of Marital & Family Therapy, 22 , 469 – 480 . Google Scholar Crossref Search ADS WorldCat Mikkelson , A. C. ( 2014 ). Adult sibling relationships. In K. Floyd & M. T. Morman (Eds.), Widening the family circle: New research on family communication (2nd ed., pp. 19 – 34 ). Los Angeles : Sage Publications . Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC Minuchin , S. ( 1974 ). Families and family therapy . Cambridge, MA : Harvard University Press . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Morr Serewicz , M. C. ( 2008 ). Toward a triangular theory of the communication and relationships of in-laws: Proposal and social relations analysis of relational satisfaction and private disclosure in in-law triads . Journal of Family Communication, 8 , 264 – 292 . Google Scholar Crossref Search ADS WorldCat Morr Serewicz , M. C. , & Canary , D. J. ( 2008 ). Assessments of disclosure from the in-laws: Links among disclosure topics, family privacy orientations, and relational quality . Journal of Social and Personal Relationships, 25 , 333 – 357 . Google Scholar Crossref Search ADS WorldCat Rittenour , C. ( 2012 ). Daughter-in-law standards for mother-in-law communication: Associations with daughter-in-law perceptions of relational satisfaction and shared family identity . Journal of Family Communication, 12 , 93 – 110 . Google Scholar Crossref Search ADS WorldCat Ross , J. L. ( 1995 ). Social class tensions within families . American Journal of Family Therapy, 23 , 338 – 350 . Google Scholar Crossref Search ADS WorldCat Santos , J. D. , & Levitt , M. J. ( 2007 ). Intergenerational relations with in-laws in the context of the social convoy: Theoretical and practical implications . Journal of Social Issues, 63 , 827 – 843 . Google Scholar Crossref Search ADS WorldCat Serovich , J. M. , & Price , S. J. ( 1994 ). In-law relationships: A role theory perspective . International Journal of Sociology of the Family, 24 , 127 – 146 . OpenURL Placeholder Text WorldCat Silverstein , J. L. ( 1990 ). The problem with in-laws . Journal of Family Therapy, 14 , 399 – 412 . Google Scholar Crossref Search ADS WorldCat Smith , S. M. , & Landor , A. M. ( 2018 ). Toward a better understanding of African American families: Development of the sociocultural family stress model . Journal of Family Theory & Review, 10 , 434 – 450 . Google Scholar Crossref Search ADS WorldCat Turner , M. , Young , C. R. , & Black , K. I. ( 2006 ). Daughters-in-law and mothers-in-law seeking their place within the family: A qualitative study of differing viewpoints . Family Relations, 55 , 588 – 600 . Google Scholar Crossref Search ADS WorldCat Visher , J. S. , & Visher , E. B. ( 1982 ). Stepfamilies and stepparenting. In F. Walsh (Ed.), Normal family processes (pp. 331 – 353 ). New York : Guilford Press . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Woolley , M. E. , & Greif , G. L. ( 2019 ). Mother-in-law reports of closeness to daughter-in-law: The determinant triangle with the son and husband . Social Work, 64 , 73 – 81 . Google Scholar Crossref Search ADS WorldCat © 2019 National Association of Social Workers This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

King McLaughlin,, Jessica;Greenfield, Jennifer, C;Hasche,, Leslie;, De Fries, Carson

doi: 10.1093/swr/svz019pmid: N/A

Abstract Young adult caregivers (YACs) of older adults are an often-overlooked subset of the caregiver population, though they make up more than a quarter of all caregivers. Because of their stage in life and their economic and work status, YACs (ages 21 to 40) are likely to face different caregiving challenges than other age cohorts of caregivers. Using the life course perspective and role conflict theory as foundational frameworks, this article compares the resources and strains of YACs with those of their middle-age caregiver (MAC) (ages 41 to 60) and older adult caregiver (OAC) (ages 61 and older) counterparts. Authors used data from a cross-sectional pilot study of caregivers recruited across one western state through community agencies. Through multivariate regression analysis, findings indicated that YACs reported more financial strain than MACs and OACs, despite being more likely to be employed. In contrast, YACs reported greater positive feelings toward caregiving than both MACs and OACs. These findings remained while controlling for employment status, education, and hours per week spent caregiving. Although YACs may find great value in caregiving, they may also be in more financially precarious situations. The article concludes with recommendations for caregiver support programs to reach YACs in the workplace. Despite growing interest in understanding the role of informal caregiving as an important aspect of long-term care in the United States, relatively little is known about the experiences of young adult caregivers (YACs), with the vast majority of existing research focusing on middle-age caregivers (MACs) and older caregivers (OACs). However, research has shown that millennials (defined as those born between 1980 and 2000 [Taylor, 2014]) make up a quarter of the caregiver population (AARP & ReACT, 2017). Furthermore, the Pew Research Center (2015) found that 86% of those ages 18 to 29 felt that it was their obligation to provide financial support for a parent. The Associated Press and NORC Center for Public Affairs Research (2018) reported that 35% of adults ages 18 to 40 expected to become caregivers for an older adult in the next five years, thus implying that those young adults who are not currently caregivers may very well become caregivers in the near future (Shabo, 2016). These younger adults who serve or are preparing to serve as caregivers for older adults may experience caregiving in meaningfully different ways than their older counterparts, largely because they begin the caregiving experience at a different time in their own life course, are often at the beginning of their working careers, and may be at an earlier stage in raising families of their own. As responsibility for caregiving falls increasingly to this group, it is important to understand how caregiving may affect them differently. This article describes a pilot study in which aspects of the caregiving experience among younger caregivers are compared with the experiences of those at other stages of life. Theoretical Framework: The Life Course Perspective This research and the measures selected for this pilot study are guided by the theoretical framework of the life course perspective, as conceptualized by Glen Elder, Jr. (1998, 2001). This perspective posits that an individual’s life is shaped by personal, societal, generational, and developmental forces (Elder, 2001), and also by an individual’s life stage. This perspective considers the influences that relationships and life events have on an individual’s future trajectory and the roles and responsibilities that they must fulfill. When considering YACs, the life course perspective is particularly applicable, as it accounts for the influences of familial processes, such as caregiving, and their effects on the individual (Fruhauf, Jarrott, & Allen, 2006), and also the impact of social relationships on the life course (Hendricks, 2012). The timing of certain events occurring in a person’s life shapes the life course (Elder, 2001). Fruhauf et al. (2006) referred to the timing of when certain life events occur as “on-time” and “off-time”—that is, they may occur at an expected point in the life course (for instance, entering the workforce when an individual is in their early 20s), or they may occur at an unexpected point in time (such as adopting an infant in later life). Caregiving as a young adult could be viewed as an off-time event, as caregiving can interfere with career development and job prospects (Fruhauf et al., 2006) and relationship development and finding a life partner (Dellmann-Jenkins, Blankemeyer, & Pinkard, 2000). In the life course perspective, birth cohorts are an important consideration, as those born during the same time periods experience similar historical events and societal changes and fluctuations. Thus, shared life stages are inevitable among those born during similar years (Elder, 2001). The present study specifically focuses on the “early adulthood” life stage (Levinson, 1986). DellmannJenkins et al. (2000) asserted “that the majority of life-shaping decisions are made between the years of 18 and 40” (p. 178). Major life events in early adulthood are often related to establishing independence from family of origin, such as moving out of family home, meeting life partner, and starting a career. Early adulthood may be characterized as flexible and having a lower propensity to have child care and job-related responsibilities, compared with other life stages. Yet early adulthood is also described as the most stressful life stage as individuals must learn to live independently on modest financial resources and limited “real world” experience without the buffering effects of parents and other adults who previously served in protector roles during their upbringing (Dellmann-Jenkins, Blankemeyer, & Pinkard, 2001; Levinson, 1986). Furthermore, the common assumption that this life stage has fewer obligations may be inaccurate if a young adult is a caregiver. Social relationships and interactions, such as caregiving, “make a profound difference in the life course” (Hendricks, 2012, p. 229). Consequences of Caregiving among Younger Adults This article focuses on three aspects of the caregiving experience: role conflict, financial strain, and positive aspects of caregiving. Although all three aspects are ubiquitous among caregivers, they affect YACs in a particular way. Role Conflict In the context of the discussion of role conflict that arises for informal caregivers, it is helpful to define the word “role” and what the caregiver role entails. Per role theory, “roles are socially agreed on functions and behaviors” (Landry-Meyer & Newman, 2004, p. 1006) that are performed in accordance with societal norms. Miller and Kaufman (1996) defined the caregiver role as “primarily a home-based activity” and caregivers as “those who provide assistance with at least one ADL [activity of daily living] or IADL [instrumental activity of daily living] limitation” (p. 191). Tetz et al. (2006) expounded on this definition by stating that caregivers “view the factors in the family care situation and take into account the activities of others (i.e. care receiver, other family members, health care professionals” (p. 252). Role conflict is defined as the dissonance that can occur when the obligations of an individual’s multiple social roles overlap and compete for time and attention (Zuba & Schneider, 2013). Biddle and Thomas (1966), two of the initial researchers of role theory, stated that “conflict may create confusion, anxiety, and ambivalence for the individual” (p. 273). Existing research suggests that YACs face significant challenges with balancing caregiving time and romantic and social relationships. In qualitative studies, YACs frequently mention that their caregiving duties cut into the time that they can spend with their partners and friends (Dellmann-Jenkins et al., 2000, 2001; Fruhauf et al., 2006). Trying to balance many roles and relationships can cause role conflict. A recent study by the Associated Press and NORC Center for Public Affairs Research (2018) found that YACs felt that they had to forgo a variety of social activities and time spent with loved ones so that they could provide care. YACs may feel that they are not able to put as much effort or time into their relationships, which may feel consequential, given that their peers may be focusing on building romantic relationships and spending time with friends. This may cause greater role conflict and exacerbate feelings of burden and strain for a YAC in a manner that does not affect OACs. In their research, Montgomery, Gonyea, and Hooyman (1985) found a negative correlation between caregiver strain and age. These findings were mirrored in the Associated Press and NORC (2018) report on caregivers; they found that although younger caregivers provided fewer hours of care every week, on average, than OACs, they were more likely than OACs to report that caregiving was “very or extremely stressful” or “moderately stressful.” Financial Strain Caregivers provide a vital and financially significant contribution not only to their families, but to society as a whole. Estimates of the current financial value of caregiving is approximately $522 billion per year and growing, yet most caregivers receive little or no formal compensation for the work that they do (Timmerman, 2018). It is not surprising that many caregivers find themselves in financial straits, especially because caregiving may require time off from work, while also prompting extra expenses on behalf of the loved one (Colorado Health Institute, 2017; Lee & Zurlo, 2014). Financial hardship and strain may be dependent on a caregiver’s age and stage in life, as older adults are more likely to have savings and other assets, and also access to stable income and health insurance through Social Security, retirement accounts, and Medicare than those who are younger (Taylor, 2014). In addition, the responsibilities of caregiving may interfere with career development and job prospects, which can further inhibit the ability of a YAC to achieve financial stability (Fruhauf et al., 2006). Although these dynamics may seem logical, very little evidence has examined the extent to which caregiving affects YACs differently than their older counterparts, and thus further exploration of the financial impacts of caregiving on YACs is needed. Positive Aspects of Caregiving Despite the challenges of caregiving, there are some areas of benefit. Fruhauf et al. (2006), Dellmann-Jenkins et al. (2000), and Dellman-Jenkins and Brittain (2003) found in their qualitative interviews with YACs that the caregivers derived a sense of purpose and satisfaction from caregiving. Quantitative results from Dellmann-Jenkins et al.’s (2001) research showed that half of the young adult participants reported that they had grown closer with the care recipient because of caregiving. A quarter of the participants in their study also reported that they had acquired an “enhanced self-concept” from caregiving and had more empathy toward older adults. Similarly, grandchild caregivers expressed a variety of positive benefits of caregiving in Fruhauf et al.’s (2006) research. Through caregiving, the grandchild caregivers were able to learn more about their families and themselves, gain a greater understanding of the aging process, and learn how to have more patience (Fruhauf et al., 2006). Grandchildren also reported that they were able to spend more time with their grandparents, which enabled them to garner wisdom, knowledge, and advice from them (Fruhauf et al., 2006). Caregivers reported that they felt less “egocentric” as a result of caregiving (Fruhauf et al., 2006). Although Day’s (2015) literature review focuses on younger caregivers (ages 18 to 25) than those who are the focus of this research, she cited research that caregiving can give caregivers “a sense of security” and enable caregivers to feel like they are taking an active role in creating “positive outcomes for the care-recipient” (p. 860). Hypotheses Research on YACs of older adults is insufficient. To our knowledge, no research yet exists that compares YACs with MACs and OACs. As such, we will be addressing an important gap in the literature by looking specifically at YACs and their unique experience in caregiving, specifically as it relates to overall caregiver strain, financial strain, and the positive aspects of caregiving. We hypothesize that (a) YACs will experience greater overall strain from caregiving than MACs or OACs, (b) YACs will report greater financial strain than MACs or OACs, and (c) YACs will report higher levels of positive aspects of caregiving than MACs or OACs. Findings from YACs will be contrasted with data on MACs and OACs to provide further insight into the challenges and benefits experienced by this subpopulation of caregivers. Design and Method Survey Recruitment Recruitment flyers and online ads were shared with a network of community agencies in Colorado that included churches, libraries, employee assistance programs, senior and recreation centers, human services departments, Area Agencies on Aging, and Veterans Service offices, to recruit a convenience sample of caregivers for our study. Study staff also attended community fairs across the state to network with additional service providers and recruit participants directly. Recruitment materials were offered in both English and Spanish. The Spanish materials were translated using a forward and backward translation process to help increase cultural responsiveness (Harkness, Pennell, & Schoua-Glusberg, 2004). Those interested in participating in the study could directly access the survey link provided on recruitment materials or contact study staff, including an available Spanish-speaking staff member, by e-mail or telephone. Between September 2016 and June 2017, participants completed the online survey. The eligibility criteria of the survey required all respondents to be current self-identified caregivers for someone age 60 or over within the state. Respondents were given a $20 gift card to a local grocery store in return for completing the survey. The study was approved by University of Denver’s institutional review board. We had a total of 95 respondents who completed the survey, with 10.52% (n = 10) completing it via phone interview. Although many efforts were made to reach out to the local Latinx community and create materials in Spanish that were culturally responsive, no respondents completed the survey in Spanish. Measures In this study, three dependent variables were measured: overall caregiver strain, caregiver financial strain, and positive aspects of caregiving. To measure caregiver strain, we used the Modified Caregiver Strain Index (Thornton & Travis, 2003), a 13-question validated scale that asks about types of strain caregivers might experience. These questions also measured the degree of strain for each item using a three-point scale (none, sometimes, often). Two of these questions were dropped from the scale because of the similarity and strong correlation with other variables in the study. The first of these questions that was dropped was adjustments in work hours, which had a large number of missing responses but correlated strongly with employment status. The second of these questions was experience of financial strain, which strongly correlated with the separate financial strain score in the survey. The remaining 11 items from the caregiver strain scale were summed to form a composite caregiver strain score (α = .89). Caregiver financial strain was measured using three items derived from Coley and Chase-Lansdale (2000): (1) “How often have you decided not to buy something you really needed for yourself or your loved one because you felt you couldn’t afford it?” Response options ranged between 0 = not at all and 4 = very often. (2) “In the last three months, how often have you had trouble paying bills?” Response options ranged between 0 = not at all and 4 = very often. (3) “Which of these four statements best describes your ability to get along on your income?” Response options ranged between 0 = I always have money left over and 4 = I can’t make ends meet. The results of these three items were added together to create a composite caregiver financial strain score, with higher scores indicating more financial strain (α = .83). To determine the degree to which caregivers experienced positive aspects of caregiving, caregivers were asked nine questions about how caregiving made them feel (for example, “Providing help made me feel strong and confident,” “Providing help made me feel good about myself,” and “Providing help strengthened my relationships with others”). A five-point scale (disagree a lot, disagree a little, neither agree or disagree, agree a little, agree a lot) was used to measure the degree of these feelings (Tarlow et al., 2004). Higher scores indicated greater levels of positive feelings (α = .94). The key predictor variable in this study was age group. Participants were grouped together by age into three different categories: those between the ages of 21 and 40 years were placed in the YAC group, those ages 41 to 60 years in the MAC group, and those ages 61 through 80 years in the OAC group. Data on several caregiver characteristics were collected to be used as potential control variables, including demographic information such as gender; race and ethnicity, which was constructed by combining responses to questions about race (black or African American, Asian, Pacific Islander, American Indian/Alaskan Native, white or Caucasian, other, and multiracial) and Latinx origin (yes/no); level of completed education (no schooling, first through eighth grades, some high school, high school graduate, some college, associate’s degree, four-year college degree, postgraduate, or other [trade, specialized]); work status (working full-time, working part-time, working two or more part-time jobs, student, military, retired, homemaker or caregiver, unemployed and looking for work, or out of work and not looking for work); federal poverty level state (above or below); and marital status (married, separated, divorced, widowed, never married, or domestic partnership). Additional control variables related to caregiving were used, including time spent caregiving each week (one to eight hours, nine to 20 hours, 21 to 40 hours, or 41 or more hours), length of time they had been providing care (less than six months, six months to one year, one to four years, five to nine years, 10 years or more), cognitive impairment of care recipient (yes or no), and distance from caregiver to care recipient (living in the same household, less than 20 minutes away, 20 minutes to one hour away, more than two hours away). Analysis We used IBM SPSS Statistics version 25 to analyze the data. First, we conducted a univariate analysis to determine frequencies, central tendencies, and normality of distributions. In terms of missing data, most items had 2% missing data (n = 2). However, the three caregiver financial strain measures each were missing five cases (5%). Listwise deletion was used to omit cases with missing data in each analysis because of the low percentage of missing data and the limited sample size. Bivariate analyses examined relationships between age group and the caregivers’ sociodemographics, caregiving experience, work experience, caregiver strain, extent of financial strain, and positive aspects of caregiving. Simultaneous multivariate regression was used to test models regarding the age group differences in terms of each dependent variable: (1) caregiver strain, (2) financial strain, and (3) positive aspects of caregiving. Each model included covariates for work status, education, and hours spent caregiving. The selection of the specific control variables was data driven. When conducting multivariate analyses, education, work status, and hours per week spent caregiving were selected as the control variables based on the results of bivariate analyses. A Bonferroni correction was applied to the analyses to limit the chance of encountering a Type I error. Results As shown in Table 1, the 95 respondents were categorized into three age groups for the purposes of this analysis: YACs ages 21 to 40 years (n = 12), MACs ages 41 to 60 years (n = 34), and OACs ages 61 to 80 years (n = 49). All of the YACs were women (n = 12); however, the YAC sample was relatively diverse in terms of race and ethnicity with 17% identifying as white or Caucasian (n = 2), 17% as black or African American (n = 2), 25% as Asian or Pacific Islander (n = 3), 17% as American Indian or Alaskan Native (n = 2), and 17% as other (n = 2). Eighty-three percent of YAC respondents were currently employed at least part-time (n = 10), with 75% of the YACs living below the federal poverty level (n = 9). Seventy-five percent of YACs reported that they spend between one and 20 hours a week providing care, while 17% spend between 21 and 40 hours a week providing care (n = 2), and 8% spend 41 or more hours a week providing care (n = 1). Eighty-three percent of YACs reported that they had been providing care for between six months to four years (n = 10). Table 1 Sample Description by Age Group Ages 21—40 Ages 41—60 Ages 60 and Older Variable n (%) n (%) n (%) p Gender (n = 93) .289  Male 0 (0) 2 (5.82) 6 (12.77)  Female 12 (100) 32 (94.12) 41 (87.23) Race* (n = 94) <.001  White or Caucasian 2 (16.67) 25 (73.53) 44 (89.80)  Black or African American 2 (16.67) 2 (5.88) 2 (4.08)  Asian 2 (16.67) 0 (0) 1 (2.04)  Pacific Islander 1 (8.33) 0 (0) 0 (0)  American Indian or Alaskan Native 2 (16.67) 0 (0) 0 (0)  Other 2 (16.67) 3 (8.82) 1 (2.04)  Multiracial 0 (0) 4 (11.76) 1 (2.04) Education level (n = 95) .140  No school 0 (0) 0 (0) 1 (2.04)  First through eighth grade 0 (0) 1 (2.94) 1 (2.04)  Some high school 1 (8.33) 1 (2.94) 0 (0)  High school graduate 2 (16.67) 1 (2.94) 3 (6.12)  Some college 2 (16.67) 10 (29.41) 8 (16.33)  Associate’s degree 0 (0) 3 (8.82) 4 (8.16)  Four-year college degree 3 (25) 9 (26.47) 11 (22.45)  Postgraduate 2 (16.67) 8 (23.53) 21 (42.86)  Other (trade, specialized) 2 (16.67) 1 (2.94) 0 (0) Employed* (n = 93) .035  Yes 10 (83.33) 19 (55.88) 18 (36.73)  No 2 (16.67) 15 (44.12) 29 (59.18) Marital status* (n = 93) <.001  Married 4 (33.33) 29 (57.58) 35 (72.92)  Separated 0 (0) 0 (0) 1 (2.08)  Divorced 1 (8.33) 10 (30.30) 6 (12.50)  Widowed 0 (0) 1 (3.03) 4 (8.33)  Never married 7 (58.33) 2 (6.06) 2 (4.17)  Domestic partner 0 (0) 1 (3.03) 0 (0) Average caregiving hours each week (n = 89) .086  20 hours or less 9 (75.00) 133 (38.23) 22 (48.90)  More than 20 hours 3 (25.00) 19 (55.88) 23 (51.11) Care recipient living location (n = 91) .140  In caregiver’s home 4 (33.33) 18 (52.94) 29 (64.44)  Not in caregiver’s home 8 (66.67) 16 (47.06) 16 (35.56) Length of time providing care (n = 94) .065  Less than 6 months 1 (8.33) 1 (3.03) 0 (0)  6 months to 1 year 3 (25) 0 (0) 4 (8.16)  1–4 years 7 (58.33) 17 (51.51) 24 (48.98)  5–9 years 1 (8.33) 8 (24.24) 10 (20.41)  10 years or more 0 (0) 7 (21.21) 11 (22.45) Care recipient relationship to caregiver* (n = 94) <.001  Parent 0 (0) 24 (70.59) 14 (29.17)  Spouse/partner 1 (8.33) 2 (5.88) 25 (52.08)  Parent-in-law 0 (0) 2 (5.88) 0 (0)  Grandparent/grandparent-in-law 6 (50) 0 (0) 0 (0)  Uncle/aunt 0 (0) 0 (0) 1 (2.08)  Sibling 0 (0) 1 (2.94) 3 (6.25)  Other relative 1 (8.33) 0 (0) 1 (2.08)  Nonrelative 4 (33.33) 5 (14.71) 4 (8.33) Economic status (n = 95) .540  Above FPL 3 (25.00) 22 (64.71) 37 (75.51)  Below FPL 9 (75.00) 12 (35.29) 12 (24.49) Ages 21—40 Ages 41—60 Ages 60 and Older Variable n (%) n (%) n (%) p Gender (n = 93) .289  Male 0 (0) 2 (5.82) 6 (12.77)  Female 12 (100) 32 (94.12) 41 (87.23) Race* (n = 94) <.001  White or Caucasian 2 (16.67) 25 (73.53) 44 (89.80)  Black or African American 2 (16.67) 2 (5.88) 2 (4.08)  Asian 2 (16.67) 0 (0) 1 (2.04)  Pacific Islander 1 (8.33) 0 (0) 0 (0)  American Indian or Alaskan Native 2 (16.67) 0 (0) 0 (0)  Other 2 (16.67) 3 (8.82) 1 (2.04)  Multiracial 0 (0) 4 (11.76) 1 (2.04) Education level (n = 95) .140  No school 0 (0) 0 (0) 1 (2.04)  First through eighth grade 0 (0) 1 (2.94) 1 (2.04)  Some high school 1 (8.33) 1 (2.94) 0 (0)  High school graduate 2 (16.67) 1 (2.94) 3 (6.12)  Some college 2 (16.67) 10 (29.41) 8 (16.33)  Associate’s degree 0 (0) 3 (8.82) 4 (8.16)  Four-year college degree 3 (25) 9 (26.47) 11 (22.45)  Postgraduate 2 (16.67) 8 (23.53) 21 (42.86)  Other (trade, specialized) 2 (16.67) 1 (2.94) 0 (0) Employed* (n = 93) .035  Yes 10 (83.33) 19 (55.88) 18 (36.73)  No 2 (16.67) 15 (44.12) 29 (59.18) Marital status* (n = 93) <.001  Married 4 (33.33) 29 (57.58) 35 (72.92)  Separated 0 (0) 0 (0) 1 (2.08)  Divorced 1 (8.33) 10 (30.30) 6 (12.50)  Widowed 0 (0) 1 (3.03) 4 (8.33)  Never married 7 (58.33) 2 (6.06) 2 (4.17)  Domestic partner 0 (0) 1 (3.03) 0 (0) Average caregiving hours each week (n = 89) .086  20 hours or less 9 (75.00) 133 (38.23) 22 (48.90)  More than 20 hours 3 (25.00) 19 (55.88) 23 (51.11) Care recipient living location (n = 91) .140  In caregiver’s home 4 (33.33) 18 (52.94) 29 (64.44)  Not in caregiver’s home 8 (66.67) 16 (47.06) 16 (35.56) Length of time providing care (n = 94) .065  Less than 6 months 1 (8.33) 1 (3.03) 0 (0)  6 months to 1 year 3 (25) 0 (0) 4 (8.16)  1–4 years 7 (58.33) 17 (51.51) 24 (48.98)  5–9 years 1 (8.33) 8 (24.24) 10 (20.41)  10 years or more 0 (0) 7 (21.21) 11 (22.45) Care recipient relationship to caregiver* (n = 94) <.001  Parent 0 (0) 24 (70.59) 14 (29.17)  Spouse/partner 1 (8.33) 2 (5.88) 25 (52.08)  Parent-in-law 0 (0) 2 (5.88) 0 (0)  Grandparent/grandparent-in-law 6 (50) 0 (0) 0 (0)  Uncle/aunt 0 (0) 0 (0) 1 (2.08)  Sibling 0 (0) 1 (2.94) 3 (6.25)  Other relative 1 (8.33) 0 (0) 1 (2.08)  Nonrelative 4 (33.33) 5 (14.71) 4 (8.33) Economic status (n = 95) .540  Above FPL 3 (25.00) 22 (64.71) 37 (75.51)  Below FPL 9 (75.00) 12 (35.29) 12 (24.49) Note: FPL = federal poverty level. *p < .05. Open in new tab Table 1 Sample Description by Age Group Ages 21—40 Ages 41—60 Ages 60 and Older Variable n (%) n (%) n (%) p Gender (n = 93) .289  Male 0 (0) 2 (5.82) 6 (12.77)  Female 12 (100) 32 (94.12) 41 (87.23) Race* (n = 94) <.001  White or Caucasian 2 (16.67) 25 (73.53) 44 (89.80)  Black or African American 2 (16.67) 2 (5.88) 2 (4.08)  Asian 2 (16.67) 0 (0) 1 (2.04)  Pacific Islander 1 (8.33) 0 (0) 0 (0)  American Indian or Alaskan Native 2 (16.67) 0 (0) 0 (0)  Other 2 (16.67) 3 (8.82) 1 (2.04)  Multiracial 0 (0) 4 (11.76) 1 (2.04) Education level (n = 95) .140  No school 0 (0) 0 (0) 1 (2.04)  First through eighth grade 0 (0) 1 (2.94) 1 (2.04)  Some high school 1 (8.33) 1 (2.94) 0 (0)  High school graduate 2 (16.67) 1 (2.94) 3 (6.12)  Some college 2 (16.67) 10 (29.41) 8 (16.33)  Associate’s degree 0 (0) 3 (8.82) 4 (8.16)  Four-year college degree 3 (25) 9 (26.47) 11 (22.45)  Postgraduate 2 (16.67) 8 (23.53) 21 (42.86)  Other (trade, specialized) 2 (16.67) 1 (2.94) 0 (0) Employed* (n = 93) .035  Yes 10 (83.33) 19 (55.88) 18 (36.73)  No 2 (16.67) 15 (44.12) 29 (59.18) Marital status* (n = 93) <.001  Married 4 (33.33) 29 (57.58) 35 (72.92)  Separated 0 (0) 0 (0) 1 (2.08)  Divorced 1 (8.33) 10 (30.30) 6 (12.50)  Widowed 0 (0) 1 (3.03) 4 (8.33)  Never married 7 (58.33) 2 (6.06) 2 (4.17)  Domestic partner 0 (0) 1 (3.03) 0 (0) Average caregiving hours each week (n = 89) .086  20 hours or less 9 (75.00) 133 (38.23) 22 (48.90)  More than 20 hours 3 (25.00) 19 (55.88) 23 (51.11) Care recipient living location (n = 91) .140  In caregiver’s home 4 (33.33) 18 (52.94) 29 (64.44)  Not in caregiver’s home 8 (66.67) 16 (47.06) 16 (35.56) Length of time providing care (n = 94) .065  Less than 6 months 1 (8.33) 1 (3.03) 0 (0)  6 months to 1 year 3 (25) 0 (0) 4 (8.16)  1–4 years 7 (58.33) 17 (51.51) 24 (48.98)  5–9 years 1 (8.33) 8 (24.24) 10 (20.41)  10 years or more 0 (0) 7 (21.21) 11 (22.45) Care recipient relationship to caregiver* (n = 94) <.001  Parent 0 (0) 24 (70.59) 14 (29.17)  Spouse/partner 1 (8.33) 2 (5.88) 25 (52.08)  Parent-in-law 0 (0) 2 (5.88) 0 (0)  Grandparent/grandparent-in-law 6 (50) 0 (0) 0 (0)  Uncle/aunt 0 (0) 0 (0) 1 (2.08)  Sibling 0 (0) 1 (2.94) 3 (6.25)  Other relative 1 (8.33) 0 (0) 1 (2.08)  Nonrelative 4 (33.33) 5 (14.71) 4 (8.33) Economic status (n = 95) .540  Above FPL 3 (25.00) 22 (64.71) 37 (75.51)  Below FPL 9 (75.00) 12 (35.29) 12 (24.49) Ages 21—40 Ages 41—60 Ages 60 and Older Variable n (%) n (%) n (%) p Gender (n = 93) .289  Male 0 (0) 2 (5.82) 6 (12.77)  Female 12 (100) 32 (94.12) 41 (87.23) Race* (n = 94) <.001  White or Caucasian 2 (16.67) 25 (73.53) 44 (89.80)  Black or African American 2 (16.67) 2 (5.88) 2 (4.08)  Asian 2 (16.67) 0 (0) 1 (2.04)  Pacific Islander 1 (8.33) 0 (0) 0 (0)  American Indian or Alaskan Native 2 (16.67) 0 (0) 0 (0)  Other 2 (16.67) 3 (8.82) 1 (2.04)  Multiracial 0 (0) 4 (11.76) 1 (2.04) Education level (n = 95) .140  No school 0 (0) 0 (0) 1 (2.04)  First through eighth grade 0 (0) 1 (2.94) 1 (2.04)  Some high school 1 (8.33) 1 (2.94) 0 (0)  High school graduate 2 (16.67) 1 (2.94) 3 (6.12)  Some college 2 (16.67) 10 (29.41) 8 (16.33)  Associate’s degree 0 (0) 3 (8.82) 4 (8.16)  Four-year college degree 3 (25) 9 (26.47) 11 (22.45)  Postgraduate 2 (16.67) 8 (23.53) 21 (42.86)  Other (trade, specialized) 2 (16.67) 1 (2.94) 0 (0) Employed* (n = 93) .035  Yes 10 (83.33) 19 (55.88) 18 (36.73)  No 2 (16.67) 15 (44.12) 29 (59.18) Marital status* (n = 93) <.001  Married 4 (33.33) 29 (57.58) 35 (72.92)  Separated 0 (0) 0 (0) 1 (2.08)  Divorced 1 (8.33) 10 (30.30) 6 (12.50)  Widowed 0 (0) 1 (3.03) 4 (8.33)  Never married 7 (58.33) 2 (6.06) 2 (4.17)  Domestic partner 0 (0) 1 (3.03) 0 (0) Average caregiving hours each week (n = 89) .086  20 hours or less 9 (75.00) 133 (38.23) 22 (48.90)  More than 20 hours 3 (25.00) 19 (55.88) 23 (51.11) Care recipient living location (n = 91) .140  In caregiver’s home 4 (33.33) 18 (52.94) 29 (64.44)  Not in caregiver’s home 8 (66.67) 16 (47.06) 16 (35.56) Length of time providing care (n = 94) .065  Less than 6 months 1 (8.33) 1 (3.03) 0 (0)  6 months to 1 year 3 (25) 0 (0) 4 (8.16)  1–4 years 7 (58.33) 17 (51.51) 24 (48.98)  5–9 years 1 (8.33) 8 (24.24) 10 (20.41)  10 years or more 0 (0) 7 (21.21) 11 (22.45) Care recipient relationship to caregiver* (n = 94) <.001  Parent 0 (0) 24 (70.59) 14 (29.17)  Spouse/partner 1 (8.33) 2 (5.88) 25 (52.08)  Parent-in-law 0 (0) 2 (5.88) 0 (0)  Grandparent/grandparent-in-law 6 (50) 0 (0) 0 (0)  Uncle/aunt 0 (0) 0 (0) 1 (2.08)  Sibling 0 (0) 1 (2.94) 3 (6.25)  Other relative 1 (8.33) 0 (0) 1 (2.08)  Nonrelative 4 (33.33) 5 (14.71) 4 (8.33) Economic status (n = 95) .540  Above FPL 3 (25.00) 22 (64.71) 37 (75.51)  Below FPL 9 (75.00) 12 (35.29) 12 (24.49) Note: FPL = federal poverty level. *p < .05. Open in new tab Table 2 Correlations Total Hours Caregiver Positive Total Spent per Not Able OACs MACs YACs Working Aspects Care Average to Get Y = 1 Y = 1 Y = 1 Y = 1 of Care- giver Week By on N = 0 N = 0 N = 0 N = 0 giving Strain Helping Income OACs Pearson correlation 1 –.771** –.392** .133 –.424** –.239 –.005 –.399** Y = 1 Sig. (2-tailed) .000 .000 .197 .000 .061 .964 .000 N = 0 n 95 95 95 91 62 89 85 MACs Pearson correlation 1 –.284** –.107 .191 .207 .104 .241* Y = 1 Sig. (2-tailed) .005 .304 .069 .106 .332 .026 N = 0 n 95 95 91 62 89 85 YACs Pearson correlation 1 –.047 .368** .046 –.139 .237* Y = 1 Sig. (2-tailed) .652 .000 .724 .193 .029 N = 0 n 95 91 62 89 85 Working Pearson correlation 1 –.021 –.079 –.211* –.215* Y = 1 Sig. (2-tailed) .846 .540 .047 .048 N = 0 N 91 62 89 85 Total positive Pearson correlation 1 –.440** –.225* .116 aspects of Sig. (2-tailed) .000 .039 .303 caregiving n 59 85 81 Total caregiver Pearson correlation 1 .384** .299* strain Sig. (2-tailed) .003 .025 n 59 56 Hours spent Pearson correlation 1 .180 per average Sig. (2-tailed) .111 week helping n 80 Caregiver not Pearson correlation 1 able to get Sig. (2-tailed) by on income n Total Hours Caregiver Positive Total Spent per Not Able OACs MACs YACs Working Aspects Care Average to Get Y = 1 Y = 1 Y = 1 Y = 1 of Care- giver Week By on N = 0 N = 0 N = 0 N = 0 giving Strain Helping Income OACs Pearson correlation 1 –.771** –.392** .133 –.424** –.239 –.005 –.399** Y = 1 Sig. (2-tailed) .000 .000 .197 .000 .061 .964 .000 N = 0 n 95 95 95 91 62 89 85 MACs Pearson correlation 1 –.284** –.107 .191 .207 .104 .241* Y = 1 Sig. (2-tailed) .005 .304 .069 .106 .332 .026 N = 0 n 95 95 91 62 89 85 YACs Pearson correlation 1 –.047 .368** .046 –.139 .237* Y = 1 Sig. (2-tailed) .652 .000 .724 .193 .029 N = 0 n 95 91 62 89 85 Working Pearson correlation 1 –.021 –.079 –.211* –.215* Y = 1 Sig. (2-tailed) .846 .540 .047 .048 N = 0 N 91 62 89 85 Total positive Pearson correlation 1 –.440** –.225* .116 aspects of Sig. (2-tailed) .000 .039 .303 caregiving n 59 85 81 Total caregiver Pearson correlation 1 .384** .299* strain Sig. (2-tailed) .003 .025 n 59 56 Hours spent Pearson correlation 1 .180 per average Sig. (2-tailed) .111 week helping n 80 Caregiver not Pearson correlation 1 able to get Sig. (2-tailed) by on income n Notes: OACs = older adult caregivers (ages 61 and older); MACs = middle-age caregivers (ages 41 to 60); YACs = young adult caregivers (ages 21 to 40). *p < .05. **p < .01. Open in new tab Table 2 Correlations Total Hours Caregiver Positive Total Spent per Not Able OACs MACs YACs Working Aspects Care Average to Get Y = 1 Y = 1 Y = 1 Y = 1 of Care- giver Week By on N = 0 N = 0 N = 0 N = 0 giving Strain Helping Income OACs Pearson correlation 1 –.771** –.392** .133 –.424** –.239 –.005 –.399** Y = 1 Sig. (2-tailed) .000 .000 .197 .000 .061 .964 .000 N = 0 n 95 95 95 91 62 89 85 MACs Pearson correlation 1 –.284** –.107 .191 .207 .104 .241* Y = 1 Sig. (2-tailed) .005 .304 .069 .106 .332 .026 N = 0 n 95 95 91 62 89 85 YACs Pearson correlation 1 –.047 .368** .046 –.139 .237* Y = 1 Sig. (2-tailed) .652 .000 .724 .193 .029 N = 0 n 95 91 62 89 85 Working Pearson correlation 1 –.021 –.079 –.211* –.215* Y = 1 Sig. (2-tailed) .846 .540 .047 .048 N = 0 N 91 62 89 85 Total positive Pearson correlation 1 –.440** –.225* .116 aspects of Sig. (2-tailed) .000 .039 .303 caregiving n 59 85 81 Total caregiver Pearson correlation 1 .384** .299* strain Sig. (2-tailed) .003 .025 n 59 56 Hours spent Pearson correlation 1 .180 per average Sig. (2-tailed) .111 week helping n 80 Caregiver not Pearson correlation 1 able to get Sig. (2-tailed) by on income n Total Hours Caregiver Positive Total Spent per Not Able OACs MACs YACs Working Aspects Care Average to Get Y = 1 Y = 1 Y = 1 Y = 1 of Care- giver Week By on N = 0 N = 0 N = 0 N = 0 giving Strain Helping Income OACs Pearson correlation 1 –.771** –.392** .133 –.424** –.239 –.005 –.399** Y = 1 Sig. (2-tailed) .000 .000 .197 .000 .061 .964 .000 N = 0 n 95 95 95 91 62 89 85 MACs Pearson correlation 1 –.284** –.107 .191 .207 .104 .241* Y = 1 Sig. (2-tailed) .005 .304 .069 .106 .332 .026 N = 0 n 95 95 91 62 89 85 YACs Pearson correlation 1 –.047 .368** .046 –.139 .237* Y = 1 Sig. (2-tailed) .652 .000 .724 .193 .029 N = 0 n 95 91 62 89 85 Working Pearson correlation 1 –.021 –.079 –.211* –.215* Y = 1 Sig. (2-tailed) .846 .540 .047 .048 N = 0 N 91 62 89 85 Total positive Pearson correlation 1 –.440** –.225* .116 aspects of Sig. (2-tailed) .000 .039 .303 caregiving n 59 85 81 Total caregiver Pearson correlation 1 .384** .299* strain Sig. (2-tailed) .003 .025 n 59 56 Hours spent Pearson correlation 1 .180 per average Sig. (2-tailed) .111 week helping n 80 Caregiver not Pearson correlation 1 able to get Sig. (2-tailed) by on income n Notes: OACs = older adult caregivers (ages 61 and older); MACs = middle-age caregivers (ages 41 to 60); YACs = young adult caregivers (ages 21 to 40). *p < .05. **p < .01. Open in new tab The majority (94%) of MACs were women (n = 32), and they were less diverse than YACs in terms of race and ethnicity, with 74% who identified as white or Caucasian (n = 25), 20% as other/multicultural (n = 7), and 6% as black or African American (n = 2). Fifty-six percent of MACs were employed on at least a part-time basis (n = 19), with 35% living below the federal poverty level (n = 12). Thirty-eight percent of MACs reported providing between one and 20 hours of care per week (n = 13), 29% between 21 and 40 hours of care per week (n = 10), and 26% spending 41 or more hours a week providing care (n = 9). Fifty-two percent of MACs had been providing care for between one and four years (n = 17), 24% had been providing care for between five and nine years (n = 8), and 21% had been providing care for 10 years or more (n = 7). OACs identified primarily (87%) as female (n =41), with 90% of OACs identifying as white or Caucasian (n = 44), 4% as black or African American (n = 2), 4% as other/multiracial (n = 2), and 2% as Asian (n = 1). A minority of the OACs (37%) reported that they were employed on at least a part-time basis (n = 18), with 24% of OACs living below the federal poverty level (n = 12). Forty-nine percent of OACs reported spending between one and 20 hours per week providing care (n = 22), 29% reported spending between 21 and 40 hours per week providing care (n = 13), and 22% reported spending 41 hours or more per week providing care (n = 10). Fifty-seven percent of OACs reported that they had spent four years or less providing care (n = 28), 20% had spent between five and nine years providing care (n = 10), and 22% had spent 10 years or more providing care (n = 11). Demographically, YACs only significantly differed from MACs and OACs in employment (83% were employed versus 56% of MACs and 37% of OACs) and were significantly more likely to have never been married (58% versus 6% of MACs and 4% of OACs). All other differences between groups were nonsignificant. Table 2 provides a correlation matrix of the variables included in the regression models. Table 3 includes the findings for each regression model. First, when financial strain was examined as a dependent variable, YACs reported higher mean financial strain scores (M = 9.25, SD = 1.658) than both MACs (M = 8.531, SD = 2.514) and OACs (M = 6.634, SD = 2.557), despite being more likely to be employed [F(1, 4) = 0.702, p < .001], though scores were only significantly different for YACs and OACs and MACs and OACs. These findings remained significant while controlling for employment status, education, and hours per week spent caregiving. In addition, YAC caregivers also reported more overall caregiver strain than OACs. YACs’ caregiver strain scores (M = 14.4, SD = 6.720) were not significantly different from MACs’ (M = 15.4, SD = 6.843), but OAC scores (M = 11.852, SD = 6.62) were significantly lower than both YACs’ and MACs’ [F(1, 4) = 3.786, p = .009]; there were not significant differences between YACs and MACs on level of overall caregiver strain. Table 3 Multivariate Regression Results That Compare Caregiver Age Groups on Positive Aspects of Caregiving, Financial Strain, and Caregiver Strain Total Positive Aspects of Caregiver Able to Get By on Caregiving Income Caregiver Strain Variable B SE t Sig. B SE t Sig. B SE t Sig. Age groups 5.935** 1.282 4.630 .000 1.345** 0.376 3.583 .001 1.302 0.888 1.465 .148 Working (yes) 0.001 0.061 0.021 .984 –0.019 0.018 –1.069 .289 –0.557 1.312 –0.424 .673 Hours spent per week providing care –1.900* 0.821 –2.314 .023 0.361 0.248 1.455 .150 1.947** 0.600 3.245 .002 Total Positive Aspects of Caregiver Able to Get By on Caregiving Income Caregiver Strain Variable B SE t Sig. B SE t Sig. B SE t Sig. Age groups 5.935** 1.282 4.630 .000 1.345** 0.376 3.583 .001 1.302 0.888 1.465 .148 Working (yes) 0.001 0.061 0.021 .984 –0.019 0.018 –1.069 .289 –0.557 1.312 –0.424 .673 Hours spent per week providing care –1.900* 0.821 –2.314 .023 0.361 0.248 1.455 .150 1.947** 0.600 3.245 .002 *p < .05. **p < .01. Open in new tab Table 3 Multivariate Regression Results That Compare Caregiver Age Groups on Positive Aspects of Caregiving, Financial Strain, and Caregiver Strain Total Positive Aspects of Caregiver Able to Get By on Caregiving Income Caregiver Strain Variable B SE t Sig. B SE t Sig. B SE t Sig. Age groups 5.935** 1.282 4.630 .000 1.345** 0.376 3.583 .001 1.302 0.888 1.465 .148 Working (yes) 0.001 0.061 0.021 .984 –0.019 0.018 –1.069 .289 –0.557 1.312 –0.424 .673 Hours spent per week providing care –1.900* 0.821 –2.314 .023 0.361 0.248 1.455 .150 1.947** 0.600 3.245 .002 Total Positive Aspects of Caregiver Able to Get By on Caregiving Income Caregiver Strain Variable B SE t Sig. B SE t Sig. B SE t Sig. Age groups 5.935** 1.282 4.630 .000 1.345** 0.376 3.583 .001 1.302 0.888 1.465 .148 Working (yes) 0.001 0.061 0.021 .984 –0.019 0.018 –1.069 .289 –0.557 1.312 –0.424 .673 Hours spent per week providing care –1.900* 0.821 –2.314 .023 0.361 0.248 1.455 .150 1.947** 0.600 3.245 .002 *p < .05. **p < .01. Open in new tab Finally, when the dependent variable “positive feelings about caregiving” was examined, YACs reported greater positive feelings (M = 48.182, SD = 4.936) than both MACs (M = 33.939, SD = 9.552) and OACs (M = 27.426, SD = 8.846) [F(1, 4) = 8.212, p < .001]. The findings for financial strain, caregiver strain, and positive aspects of caregiving remained significant when controlling for employment status, education, and hours per week spent caregiving. Implications The average millennial caregiver is an individual who lives in a household earning less than the median national household income, despite working almost full-time (National Alliance for Caregiving & AARP Public Policy Institute, 2015). Based on this small pilot study, our results suggest a similar profile of a YAC: someone who, in spite of working while caregiving, is struggling to get by financially. More research is needed to further test potential explanations for these results, as mentioned later in this text. The financial struggles of YACs could be attributed to their inability to achieve a higher level of education, possibly because caregiving duties interfere with education (Fruhauf et al., 2006), which thus inhibits YACs’ earning potential. Another possibility is that caregiving has interfered with YACs’ ability to work at all, thus hurting their income, which existing research has found in abundance (see, for example, MetLife Mature Market Institute, 2011; Pavalko & Henderson, 2006; Shabo, 2016). The results show that YACs also experience other strains from caregiving at a level similar to their middle-age counterparts, though at a lower level than OACs, even when characteristics such as employment status and the hours per week spent caregiving are controlled for. This may occur because the caregiving responsibilities come at an “off time” in the life course and are relatively unexpected at the caregiver’s developmental stage. As a result, younger caregivers are more acutely aware of the disruptions and discomforts that come with caregiving. At the same time, YACs are also more likely to report positive experiences related to their caregiving experience. The life course perspective may lend some insight into this finding, in that younger caregivers experience caregiving at an unexpected time in their life trajectory, yet they unexpectedly may feel fulfilled by the role in which they find themselves—offering care to a loved one in need may be a way that younger adults who are still forming a sense of identity find purpose and meaning in their lives. More research on this aspect of the young adult caregiving is needed so that strategies for maximizing these benefits can be developed. Finally, although YACs may find great value in caregiving, they may also be in more financially precarious situations. Many millennials feel greater burdens from student loans and other debts than individuals in prior cohorts, in part because of the global economic recession, which struck in 2007–2008 (Taylor, 2014). Paying for daily expenses, paying down debts, and managing the costs or a decrease in income that YACs may experience as a result of caregiving may be placing greater burdens on this cohort of caregivers than on MACs and OACs. Of particular concern is the potential for caregivers to forgo promotions and raises as a result of the time pressures of caregiving—these are costs that may permanently alter their income and wealth trajectories and make it harder for millennial caregivers to pay off student loans, afford mortgages, and prepare for their own child-rearing and retirement needs. Much more research is needed to understand the long-term financial implications of caregiving for older adults among this cohort and to develop policies and other strategies to foster long-term financial security for millennials who balance work with caregiving. Research has shown that YACs support the idea of employer-provided long-term care insurance (Associated Press & NORC Center for Public Affairs Research, 2018) and flexibility in work schedule, hours, and location (Sloan Center on Aging & Work at Boston College, 2013). In addition, workplaces could provide paid family leave that is inclusive of older adult caregiving responsibilities. Caregiver support programs, like the National Family Caregiver Support Program, may want to expand efforts to reach YACs by offering social programs that reimburse caregivers. The National Family Caregivers Support Program, established in the 2000 reauthorization of the Older Americans Act, provides resources and grants to states to support caregivers over age 18 caring for individuals over 60 years of age in their homes (Administration for Community Living, 2017), but this program’s funding varies by state and does not focus specifically on special populations of caregivers, like YACs. Furthermore, publicly available services, like respite, and funding for caregiving activities may be obscure and difficult to obtain through complicated forms and waivers. YACs, who may be juggling a career and parenting in addition to caregiving, may not know about such programs, not know how to gain access to these programs, or have the time or energy available to enroll in them. More efforts are needed within aging services to reach younger caregivers and to adapt these programs to their cohort-specific needs. Limitations Despite the intriguing findings from this study, there are several limitations to acknowledge. First, the data were drawn from a small convenience sample of caregivers in one state. As the overall sample size was small, the sample size of YACs was also small. Although the study was theoretically based, the small sample size limited the ability to perform between-group analyses by age and prevented a true theoretically driven analysis. However, there was adequate power to run multivariate analyses with the three variables that we used. In addition, the sampling strategy was useful for testing the validity of the survey instrument and feasibility of the study but limits the generalizability of the study findings. Furthermore, no data were collected that would reflect the care recipient’s age. In the event that this pilot study is expanded to capture data on more caregivers in Colorado, it would be advantageous to include a survey question about the care recipient’s age so that relationships between caregivers’ and care recipients’ ages could be observed. Fortunately, several findings from the demographic analysis suggest that this sample was similar in key ways to those included in nationally representative studies of caregivers. Therefore, we believe the findings to be noteworthy and should prompt further investigation into the experiences of YACs. Conclusion YACs make up a substantial and important segment of the caregiver population. Our findings are consistent with the limited research on YACs and offer additional support to the need to focus on the experiences of this growing population of caregivers. Unfortunately, the concerns and issues that YACs face are often unrecognized and unaddressed in policy and supportive programs for caregivers. As our population ages, YACs will gain more importance as caregivers. It is crucial to expand knowledge of this important subpopulation of caregivers now so that policies can be developed and implemented to ensure that they can maintain their financial security and overall well-being as they age. Jessica King McLaughlin, MSW, is third-year PhD student; Jennifer C. Greenfield, PhD, MSW, is associate professor; and Leslie Hasche, PhD, MSW, is associate professor and associate dean for academic affairs, Graduate School of Social Work, University of Denver. Carson De Fries, MSW, is intergenerational program coordinator, Knoebel Institute for Healthy Aging, University of Denver. Address correspondence to Jessica King McLaughlin, Graduate School of Social Work, University of Denver, 2148 South High Street, Fort Collins, CO 80521; e-mail: [email protected]. References AARP & ReACT. ( 2017 ). Supporting working caregivers: Case studies of promising practices. Retrieved from https://respectcaregivers.org/wp-content/uploads/2017/05/AARP-ReAct-MASTER-web.pdf Administration for Community Living. ( 2017 ). National Family Caregiver Support Program. Retrieved from https://www.acl.gov/programs/support-caregivers/national-family-caregiver-support-program# Associated Press & NORC Center for Public Affairs Research. ( 2018, May ). Younger adults’ experiences and views on long-term care [Issue Brief]. Retrieved from https://www.longtermcarepoll.org/wp-content/uploads/2018/05/APNORC_LTC_2018_brief.pdf Biddle , B. J. , & Thomas , E. J. (Eds.). ( 1966 ). Role theory: Concepts and research . New York : Wiley . Google Preview WorldCat COPAC Coley , R. L. , & Chase-Lansdale , P. L. ( 2000 ). Welfare receipt, financial strain, and African-American adolescent functioning . Social Service Review, 74 , 380 – 404 . Google Scholar Crossref Search ADS WorldCat Colorado Health Institute. ( 2017 ). Putting a price on informal caregiving in Colorado: Research for the Strategic Action Planning Group on Aging . Retrieved from https://www.colorado.gov/pacific/sites/default/files/SAPGA%20Caregiving%20Report.pdf Day , C. ( 2015 ). Young adult carers: A literature review informing the re-conceptualisation of young adult caregiving in Australia . Journal of Youth Studies, 18 , 855 – 866 . Google Scholar Crossref Search ADS WorldCat Dellmann-Jenkins , M. , Blankemeyer , M. , & Pinkard , O. ( 2000 ). Young adult children and grandchildren in primary caregiver roles to older relatives and their service needs . Family Relations, 49 , 177 – 186 . doi:10.1111/j.1741-3729.2000.00177.x Google Scholar Crossref Search ADS WorldCat Dellmann-Jenkins , M. , Blankemeyer , M. , & Pinkard , O. ( 2001 ). Incorporating the elder caregiving role into the developmental tasks of young adulthood . International Journal of Aging and Human Development, 52 ( 1 ), 1 – 18 . doi:10.2190/FGQA-65FU-JGNT-6C9J Google Scholar Crossref Search ADS WorldCat Dellmann-Jenkins , M. , & Brittain , L. ( 2003 ). Young adults’ attitudes toward filial responsibility and actual assistance to elderly family members . Journal of Applied Gerontology, 22 , 214 – 229 . Google Scholar Crossref Search ADS WorldCat Elder , G. H., J. ( 1998 ). The life course as developmental theory . Child Development, 69 ( 1 ), 1 – 12 . Google Scholar Crossref Search ADS WorldCat Elder , G. ( 2001 ). Families, social change, and individual lives . Marriage and Family Review, 31 ( 1–2 ), 177 – 192 . WorldCat Fruhauf , C. A. , Jarrott , S. E. , & Allen , K. R. ( 2006 ). Grandchildren’s perceptions of caring for grandparents . Journal of Family Issues, 27 , 887 – 911 . doi:10.1177/0192513X05286019 Google Scholar Crossref Search ADS WorldCat Harkness , J. A. , Pennell , B.-E. , & Schoua-Glusberg , A. ( 2004 ). Survey questionnaire translation and assessment. In S. Presser , J. M. Rothgeb , M. P. Couper , J. T. Lessler , E. Martin , J. Martin , & E. Singer (Eds.), Methods for testing and evaluating survey questionnaires (pp. 453 – 474 ). New York : Wiley . Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Hendricks , J. ( 2012 ). Considering life course concepts . Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 67B ( 2 ), 226 – 231 . Google Scholar Crossref Search ADS WorldCat Landry-Meyer , L. , & Newman , B. M. ( 2004 ). An exploration of the grandparent caregiver role . Journal of Family Issues, 25 , 1005 – 1025 . Google Scholar Crossref Search ADS WorldCat Lee , Y. , & Zurlo , K. A. ( 2014 ). Spousal caregiving and financial strain among middle-aged and older adults . International Journal of Aging & Human Development, 79 , 302 – 321 . doi:10.1177/0091415015574181 Google Scholar Crossref Search ADS WorldCat Levinson , D. J. ( 1986 ). A conception of adult development . American Psychologist, 41 , 3 – 13 . doi:10.1037/0003-066X.41.1.3 Google Scholar Crossref Search ADS WorldCat MetLife Mature Market Institute. ( 2011 ). The MetLife Study of Caregiving Costs to Working Caregivers: Double jeopardy for baby boomers caring for their parents . Retrieved from https://www.caregiving.org/wp-content/uploads/2011/06/mmi-caregiving-costs-working-caregivers.pdf Miller , B. , & Kaufman , J. E. ( 1996 ). Beyond gender stereotypes: Spouse caregivers of persons with dementia . Journal of Aging Studies, 10 , 189 – 204 . Google Scholar Crossref Search ADS WorldCat Montgomery , R. , Gonyea , J. , & Hooyman , N. ( 1985 ). Caregiving and the experience of subjective and objective burden . Family Relations, 34 ( 1 ), 19 – 26 . Google Scholar Crossref Search ADS WorldCat National Alliance for Caregiving & AARP Public Policy Institute . ( 2015, August ). Caregiver profile: The millennial caregiver. Caregiving in the U.S. Retrieved from https://www.aarp.org/content/dam/aarp/ppi/2015/Caregiving%20in%20the%20US%202015_Millennial_CG%20Profile.pdf Pavalko , E. K. , & Henderson , K. A. ( 2006 ). Combining care work and paid work: Do workplace policies make a difference? Research on Aging, 28 , 359 – 374 . doi:10.1177/0164027505285848 Google Scholar Crossref Search ADS WorldCat Pew Research Center. ( 2015 ). Family support in graying societies: How Americans, Germans and Italians are coping with an aging population. Retrieved from http://assets.pewresearch.org/wp-content/uploads/sites/3/2015/05/2015-05-21_family-support-relations_FINAL.pdf Shabo , V. ( 2016 ). Advances in workplace protections for family caregivers . Journal of the American Society on Aging, 39 , 89 – 93 . WorldCat Sloan Center on Aging & Work at Boston College. ( 2013 ). Schedule flexibility in the workplace: Employee access and use (Fact sheet 34A). Chestnut Hill, MA : Author . Google Preview WorldCat COPAC Tarlow , B. J. , Wisniewski , S. R. , Belle , S. H. , Rubert , M. , Ory , M. G. , & Gallagher-Thompson , D. ( 2004 ). Positive aspects of caregiving: Contributions of the REACH project to the development of a new measure for Alzheimer’s caregiving . Research on Aging, 26 , 429 – 453 . Google Scholar Crossref Search ADS WorldCat Taylor , P. ( 2014 ). The next America: Boomers, millennials, and the looming generational showdown . New York : Public Affairs . Google Preview WorldCat COPAC Tetz , K. , Archbold , P. , Stewart , B. , Messecar , D. , Hornbrook , M. , & Lucas , S. ( 2006 ). How frail elders evaluate their caregiver’s role enactment: A scale to measure affection, skill, and attentiveness . Journal of Family Nursing, 12 , 251 – 275 . Google Scholar Crossref Search ADS WorldCat Thornton , M. , & Travis , S. ( 2003 ). Analysis of the reliability of the modified caregiver strain index . Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 58 ( 2 ), S127 – S132 . Google Scholar Crossref Search ADS WorldCat Timmerman , S. ( 2018, January 22 ). Families and finances [Webinar]. In Global Caregiving Summit . Retrieved from http://awbcaregiving.com/day-3/3-family/sandra-timmermann/ Zuba , M. , & Schneider , U. ( 2013 ). What helps working informal caregivers? The role of workplace characteristics in balancing work and adult-care responsibilities . Journal of Family and Economic Issues, 34 , 460 – 469 . Google Scholar Crossref Search ADS WorldCat © 2019 National Association of Social Workers This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

DeVylder,, Jordan;Fedina,, Lisa;Jun,, Hyun-Jin

doi: 10.1093/swr/svz015pmid: N/A

The issue of neighborhood gentrification and its impact on U.S. cities continues to be a controversial area of debate. Gentrification is generally defined as the growth of affluence and its associated changes to local infrastructure, housing costs, and displacement of long-term residents, who are typically of lower socioeconomic status (SES) (Barton, 2016; Freeman, 2005; Newman & Wyly, 2006). Neighborhood change and gentrification are key social exposures that may exert a significant and broadly variable array of effects on individuals, ranging from involuntary displacement of marginalized residents to economic prosperity for investors or property owners (Freeman, 2005). Although causes of gentrification have been debated, scholars have generally agreed on the importance of the dynamics of neighborhood change and the historical disinvestment in communities that create conditions opportune for reinvestment and gentrification (Beauregard, 1986; Freeman & Braconi, 2004; Rose, 1984). Furthermore, theorists suggest that displacement is largely behind changing neighborhood dynamics (for example, demographics) that result from gentrification (Freeman & Braconi, 2004; Hartman, 1979) and that such changes to public spaces can affect both health and health equity (Mehdipanah et al., 2015), with a growing body of literature confirming that gentrification may lead to negative health and mental health outcomes (Atkinson, 2000; Curtis, Cave, & Coutts, 2002; Lim et al., 2017). Research on community and neighborhood factors, including gentrification, often depends on objective measures using population-level data and public records that are aggregated across communities or geographic areas (Anguelovski, 2015; Hwang & Sampson, 2014; Levy, Comey, & Padilla, 2007; Prince, 2016). Yet a critical question in social work is how such community factors affect vulnerable individuals within communities, who may then in turn benefit from social, economic, and health services. When aggregate measures are used (for example, neighborhood income, income disparities, walkability) (Galster, Andersson, Musterd, & Kauppinen, 2008; King et al., 2011; Sallis et al., 2011), we assume uniformity or a “mean effect” in terms of how community factors affect individuals, even though there is likely substantial variance in how such neighborhood variables affect individuals. In their recent review of existing measures for neighborhood gentrification and displacement, Ohmer, Coulton, Freedman, Sobeck, and Booth (2018) highlighted common quantitative approaches to assessing neighborhood gentrification, which include the use of census data (for example, median income, rental inflation) and technology (for example, Google Street View) to observe indicators of gentrification in neighborhoods (for example, rehabbed buildings, lack of disorder and decay). Although it is valuable to understand objective indicators of community and neighborhood-level factors, it is also important to understand how these factors are perceived by individual residents within those communities, and subjective measures of residents’ perceptions may be as important as objective measures in understanding neighborhood characteristics (Centers for Disease Control and Prevention, 2013; Ohmer et al., 2018). Scholars have previously called for research on gentrification to go beyond the use of census-based measures and avoid dichotomous assessments of gentrification (gentrified or not), which limits the ability to understand nuance in the gentrifying process (Hwang, 2016), including its impacts on residents. Furthermore, prior research suggests that neighborhood perceptions affect residents’ health even after controlling for individual-level factors such as depression and SES, suggesting that neighborhood perceptions uniquely affect health apart from these potential confounders (Weden, Carpiano, & Robert, 2008) and objective measures of gentrification. To the best of our knowledge, there is no existing quantitative measure of individuals’ perceived gentrification, apart from other related but distinct perceived neighborhood constructs (for example, neighborhood quality, collective efficacy, civic engagement, and so on) (Ohmer et al., 2018). To address this need, we developed a novel self-report measure—the Neighborhood Change and Gentrification Scale (NCGS). The aims of this study are to present NCGS and to identify underlying factors using principal components analysis (PCA). Method Study Sample and Procedures The second Survey of Police-Public Encounters is a cross-sectional, general population survey administered in March 2017 via Qualtrics online survey software to English-speaking adults age 18 and older in Baltimore City and New York City (DeVylder et al., 2018). A quota sampling methodology was used to recruit study respondents using Qualtrics Panels—an online service that retains survey panels consisting of several million U.S. residents who consent to participating in periodic survey research. Demographically representative samples (that is, ±10% of 2010 census distributions for age, sex, and race and ethnicity in each city) were recruited based on geographic household data. Participants were not asked to complete the survey if limits had been met for their demographic group within the boundaries of each city. Participants received a monetary incentive at rates determined by Qualtrics for completing the survey. The study was approved by the sponsoring university’s institutional review board (IRB). An IRB waiver for written consent was granted, and participants were provided with information on the purpose of the study. Informed consent was obtained by agreeing to proceed to complete the survey online. Of 1,221 adults who agreed to participate in the survey, 221 were excluded due to incorrectly responding to attention checks and discontinued participation before completion, which resulted in a final sample of 1,000 (81.9% of consenting respondents). Measure NCGS items asked respondents to self-report whether they agreed with a set of statements describing changes that they may have experienced in their neighborhood over the past several years. NCGS includes 10 items with a five-point Likert response set (1 = strongly disagree, 2 = disagree, 3 = neutral, 4 = agree, 5 = strongly agree) that were developed based on prior research on gentrification and its potential impacts on residents. Four items were developed based on prior research using census-based and technology-related measures of neighborhood gentrification characteristics: (1) recent increased renovation activity, (2) increased amenities and services, (3) decreased crime rates, and (4) increased influx of affluent and nonminority residents (Freeman, 2005; Hwang & Sampson, 2014; Lees, Slater, & Wyly, 2013; Sallis et al., 2011). An additional six items were informed by prior qualitative and quantitative research on self-reported experiences with gentrification and displacement: experiences of harassment by landlords; feelings of being “pushed out,” “unwelcome,” or “out of place”; and disruption of social networks and supports due to recent neighborhood changes (Freeman, 2005; Levy et al., 2007; Newman & Wyly, 2006; Prince, 2016; Valli, 2015). We attempted to develop a concise set of items, based primarily on this empirically derived set of constructs identified in these prior studies, which would capture the breadth of these aspects of gentrification (both positive and negative). We then relied on empirical tests of the measure (as described later) to confirm or disconfirm the validity of those items, while also identifying potential subscales (that is, factor analysis) or redundant items (that is, tests of internal consistency). Factor Analysis A PCA was conducted using SPSS Version 24 to reveal the underlying structure of NCGS. Data were initially screened for missing data and univariate outliers across each item of NCGS. There were no out-of-range values or missing data across items, and the sample size needed to conduct a factor analysis was sufficient given the number of survey items. Factorability of the 10 items was assessed through several preliminary tests. First, Pearson’s correlations revealed that each item was correlated with at least one other item at .3. Second, Kaiser-Meyer-Olkin measure for sampling adequacy was excellent at .82 (above the recommended .6), and Bartlett’s test of sphericity was significant [χ2(45, N = 1,000) = 2,889.74, p < .001] (Worthington & Whittaker, 2006). Communalities were above .3 for each item, suggesting common variance across all items. Based on Kaiser’s criterion, all factors with eigenvalues greater than 1 were retained. Table 1 Description of Sample Characteristics . New York City (n = 550) . Baltimore (n = 450) . Total Sample (N = 1,000) . Characteristic . n . % . n . % . n . % . Gender  Male 211 38.4 183 40.7 394 39.4  Female 337 61.3 263 58.4 600 60.0  Transgender/nonconforming 2 0.4 4 0.9 6 0.6 Age group  18–24 97 17.6 64 14.2 161 16.1  25–44 223 40.5 251 55.8 474 47.4  45–64 174 31.6 114 25.3 288 28.8  65+ 56 10.2 21 4.7 77 7.7 Race and ethnicity  White, non-Latino 204 37.1 135 30.0 339 33.9  Black, non-Latino 144 26.2 246 54.7 390 39.0  Latino 143 26.0 35 7.8 178 17.8  Other 59 10.7 34 7.6 93 9.3 Sexual orientation  Heterosexual 498 90.5 405 90.0 903 90.3  Gay/lesbian 17 3.1 11 2.4 28 2.8  Bisexual 20 3.6 24 5.3 44 4.4  Not specified 15 2.7 10 2.2 25 2.5 Annual household income ($)  <20,000 97 17.6 112 24.9 209 20.9  20,000–39,999 102 18.5 109 24.2 211 21.1  40,000–59,999 100 18.2 98 21.8 198 19.8  60,000–79,999 90 16.4 48 10.7 138 13.8  80,000–99,999 58 10.5 30 6.7 88 8.8  >100,000 103 18.7 53 11.8 158 15.8 Education  < High school 18 3.3 20 4.4 38 3.8  High school/GED 121 22.0 143 31.8 264 26.4  Some college/tech 152 27.6 135 30.0 287 28.7  College graduate 180 32.7 99 22.0 279 27.9  Grad/professional 79 14.4 53 11.8 132 13.2 . New York City (n = 550) . Baltimore (n = 450) . Total Sample (N = 1,000) . Characteristic . n . % . n . % . n . % . Gender  Male 211 38.4 183 40.7 394 39.4  Female 337 61.3 263 58.4 600 60.0  Transgender/nonconforming 2 0.4 4 0.9 6 0.6 Age group  18–24 97 17.6 64 14.2 161 16.1  25–44 223 40.5 251 55.8 474 47.4  45–64 174 31.6 114 25.3 288 28.8  65+ 56 10.2 21 4.7 77 7.7 Race and ethnicity  White, non-Latino 204 37.1 135 30.0 339 33.9  Black, non-Latino 144 26.2 246 54.7 390 39.0  Latino 143 26.0 35 7.8 178 17.8  Other 59 10.7 34 7.6 93 9.3 Sexual orientation  Heterosexual 498 90.5 405 90.0 903 90.3  Gay/lesbian 17 3.1 11 2.4 28 2.8  Bisexual 20 3.6 24 5.3 44 4.4  Not specified 15 2.7 10 2.2 25 2.5 Annual household income ($)  <20,000 97 17.6 112 24.9 209 20.9  20,000–39,999 102 18.5 109 24.2 211 21.1  40,000–59,999 100 18.2 98 21.8 198 19.8  60,000–79,999 90 16.4 48 10.7 138 13.8  80,000–99,999 58 10.5 30 6.7 88 8.8  >100,000 103 18.7 53 11.8 158 15.8 Education  < High school 18 3.3 20 4.4 38 3.8  High school/GED 121 22.0 143 31.8 264 26.4  Some college/tech 152 27.6 135 30.0 287 28.7  College graduate 180 32.7 99 22.0 279 27.9  Grad/professional 79 14.4 53 11.8 132 13.2 Open in new tab Table 1 Description of Sample Characteristics . New York City (n = 550) . Baltimore (n = 450) . Total Sample (N = 1,000) . Characteristic . n . % . n . % . n . % . Gender  Male 211 38.4 183 40.7 394 39.4  Female 337 61.3 263 58.4 600 60.0  Transgender/nonconforming 2 0.4 4 0.9 6 0.6 Age group  18–24 97 17.6 64 14.2 161 16.1  25–44 223 40.5 251 55.8 474 47.4  45–64 174 31.6 114 25.3 288 28.8  65+ 56 10.2 21 4.7 77 7.7 Race and ethnicity  White, non-Latino 204 37.1 135 30.0 339 33.9  Black, non-Latino 144 26.2 246 54.7 390 39.0  Latino 143 26.0 35 7.8 178 17.8  Other 59 10.7 34 7.6 93 9.3 Sexual orientation  Heterosexual 498 90.5 405 90.0 903 90.3  Gay/lesbian 17 3.1 11 2.4 28 2.8  Bisexual 20 3.6 24 5.3 44 4.4  Not specified 15 2.7 10 2.2 25 2.5 Annual household income ($)  <20,000 97 17.6 112 24.9 209 20.9  20,000–39,999 102 18.5 109 24.2 211 21.1  40,000–59,999 100 18.2 98 21.8 198 19.8  60,000–79,999 90 16.4 48 10.7 138 13.8  80,000–99,999 58 10.5 30 6.7 88 8.8  >100,000 103 18.7 53 11.8 158 15.8 Education  < High school 18 3.3 20 4.4 38 3.8  High school/GED 121 22.0 143 31.8 264 26.4  Some college/tech 152 27.6 135 30.0 287 28.7  College graduate 180 32.7 99 22.0 279 27.9  Grad/professional 79 14.4 53 11.8 132 13.2 . New York City (n = 550) . Baltimore (n = 450) . Total Sample (N = 1,000) . Characteristic . n . % . n . % . n . % . Gender  Male 211 38.4 183 40.7 394 39.4  Female 337 61.3 263 58.4 600 60.0  Transgender/nonconforming 2 0.4 4 0.9 6 0.6 Age group  18–24 97 17.6 64 14.2 161 16.1  25–44 223 40.5 251 55.8 474 47.4  45–64 174 31.6 114 25.3 288 28.8  65+ 56 10.2 21 4.7 77 7.7 Race and ethnicity  White, non-Latino 204 37.1 135 30.0 339 33.9  Black, non-Latino 144 26.2 246 54.7 390 39.0  Latino 143 26.0 35 7.8 178 17.8  Other 59 10.7 34 7.6 93 9.3 Sexual orientation  Heterosexual 498 90.5 405 90.0 903 90.3  Gay/lesbian 17 3.1 11 2.4 28 2.8  Bisexual 20 3.6 24 5.3 44 4.4  Not specified 15 2.7 10 2.2 25 2.5 Annual household income ($)  <20,000 97 17.6 112 24.9 209 20.9  20,000–39,999 102 18.5 109 24.2 211 21.1  40,000–59,999 100 18.2 98 21.8 198 19.8  60,000–79,999 90 16.4 48 10.7 138 13.8  80,000–99,999 58 10.5 30 6.7 88 8.8  >100,000 103 18.7 53 11.8 158 15.8 Education  < High school 18 3.3 20 4.4 38 3.8  High school/GED 121 22.0 143 31.8 264 26.4  Some college/tech 152 27.6 135 30.0 287 28.7  College graduate 180 32.7 99 22.0 279 27.9  Grad/professional 79 14.4 53 11.8 132 13.2 Open in new tab Results Participant demographics are described in Table 1. Initial eigenvalues indicated that the first three factors explained 35.14%, 18.19%, and 9.20% of the variance, respectively. The remaining (fourth through 10th) factors each had eigenvalues over 2 and collectively accounted for 37.54% of the variance. Oblimin and varimax rotation techniques of the factor loading matrix were used, which yielded similar results for a two-factor solution. Final PCA of all 10 items with oblimin rotation that provided the best-defined factor structure produced two components with eigenvalues greater than 1. Oblimin rotation of factor loadings provided the best-defined factor structure and therefore was used in the final two-factor solution, which explained 62.53% of the variance. All 10 items had a primary factor loading of at least .4 or above and no cross-loading of .3 or above. The following items loaded onto the first factor, neighborhood disruption: “I have feared being ‘pushed out’ of my neighborhood”; “I have seen a disruption of local community ties and social networks”; “I have experienced or heard of others being harassed by their landlords to vacate an apartment”; “I have felt increasingly ‘out of place’ in my neighborhood”; “I worry about feeling ‘unwelcome’ in my neighborhood”; and “I have observed changes to the sense of ‘community’ in the neighborhood” (items 3, 5–9). Composite scores were assessed for items in the first scale, indicating that higher scores suggest more negative perceptions of recent changes to the participant’s neighborhood. Results suggest excellent internal consistency in this sample for first factor items (Cronbach’s alpha = .83). Table 2 Factor Loadings and Communalities from Principal Components Analysis for 10 Items in the Neighborhood Change and Gentrification Scale (N = 1,000) Item . Factor 1 Neighborhood Disruption . Factor 2 Neighborhood Gentrification . Communality . I have experienced improved access to neighborhood amenities and city services. .78 .59 I have seen an influx of affluent or non- minority residents moving into the neighborhood. .68 .52 I have feared being ‘‘pushed out’’ of my neighborhood. .74 .56 Crime has decreased in my neighborhood. .64 .40 I have seen a disruption of local community ties and social networks. .66 .45 I have experienced or heard of others being harassed by their landlords to vacate an apartment. .69 .49 I have felt increasingly ‘‘out of place’’ in my neighborhood. .86 .72 I worry about feeling ‘‘unwelcome’’ in my neighborhood. .84 .68 I have observed changes to the sense of “community” in the neighborhood. .61 .46 I have observed a lot of renovation activity in the neighborhood. .62 .45 Item . Factor 1 Neighborhood Disruption . Factor 2 Neighborhood Gentrification . Communality . I have experienced improved access to neighborhood amenities and city services. .78 .59 I have seen an influx of affluent or non- minority residents moving into the neighborhood. .68 .52 I have feared being ‘‘pushed out’’ of my neighborhood. .74 .56 Crime has decreased in my neighborhood. .64 .40 I have seen a disruption of local community ties and social networks. .66 .45 I have experienced or heard of others being harassed by their landlords to vacate an apartment. .69 .49 I have felt increasingly ‘‘out of place’’ in my neighborhood. .86 .72 I worry about feeling ‘‘unwelcome’’ in my neighborhood. .84 .68 I have observed changes to the sense of “community” in the neighborhood. .61 .46 I have observed a lot of renovation activity in the neighborhood. .62 .45 Open in new tab Table 2 Factor Loadings and Communalities from Principal Components Analysis for 10 Items in the Neighborhood Change and Gentrification Scale (N = 1,000) Item . Factor 1 Neighborhood Disruption . Factor 2 Neighborhood Gentrification . Communality . I have experienced improved access to neighborhood amenities and city services. .78 .59 I have seen an influx of affluent or non- minority residents moving into the neighborhood. .68 .52 I have feared being ‘‘pushed out’’ of my neighborhood. .74 .56 Crime has decreased in my neighborhood. .64 .40 I have seen a disruption of local community ties and social networks. .66 .45 I have experienced or heard of others being harassed by their landlords to vacate an apartment. .69 .49 I have felt increasingly ‘‘out of place’’ in my neighborhood. .86 .72 I worry about feeling ‘‘unwelcome’’ in my neighborhood. .84 .68 I have observed changes to the sense of “community” in the neighborhood. .61 .46 I have observed a lot of renovation activity in the neighborhood. .62 .45 Item . Factor 1 Neighborhood Disruption . Factor 2 Neighborhood Gentrification . Communality . I have experienced improved access to neighborhood amenities and city services. .78 .59 I have seen an influx of affluent or non- minority residents moving into the neighborhood. .68 .52 I have feared being ‘‘pushed out’’ of my neighborhood. .74 .56 Crime has decreased in my neighborhood. .64 .40 I have seen a disruption of local community ties and social networks. .66 .45 I have experienced or heard of others being harassed by their landlords to vacate an apartment. .69 .49 I have felt increasingly ‘‘out of place’’ in my neighborhood. .86 .72 I worry about feeling ‘‘unwelcome’’ in my neighborhood. .84 .68 I have observed changes to the sense of “community” in the neighborhood. .61 .46 I have observed a lot of renovation activity in the neighborhood. .62 .45 Open in new tab Table 3 Descriptive Statistics for Neighborhood Change and Gentrification Scale Subscales (N = 1,000) Subscale . Number of Items . M (SD) . Skewness . Kurtosis . Cronbach’s Alpha . Neighborhood Disruption 6 2.61 (0.88) .23 –.39 .83 Neighborhood Gentrification 4 3.12 (0.78) –.16 .23 .64 Subscale . Number of Items . M (SD) . Skewness . Kurtosis . Cronbach’s Alpha . Neighborhood Disruption 6 2.61 (0.88) .23 –.39 .83 Neighborhood Gentrification 4 3.12 (0.78) –.16 .23 .64 Open in new tab Table 3 Descriptive Statistics for Neighborhood Change and Gentrification Scale Subscales (N = 1,000) Subscale . Number of Items . M (SD) . Skewness . Kurtosis . Cronbach’s Alpha . Neighborhood Disruption 6 2.61 (0.88) .23 –.39 .83 Neighborhood Gentrification 4 3.12 (0.78) –.16 .23 .64 Subscale . Number of Items . M (SD) . Skewness . Kurtosis . Cronbach’s Alpha . Neighborhood Disruption 6 2.61 (0.88) .23 –.39 .83 Neighborhood Gentrification 4 3.12 (0.78) –.16 .23 .64 Open in new tab The following items loaded onto the second factor, neighborhood gentrification: “I have experienced improved access to neighborhood amenities and city services”; “I have seen an influx of affluent or nonminority residents moving into the neighborhood”; “Crime has decreased in my neighborhood”; and “I have observed a lot of renovation activity in the neighborhood” (items 1, 2, 4, and 10) (see Table 2 for factor loading matrix results). Composite scores were assessed for items in the second subscale, indicating that higher scores suggest more positive perceptions of recent changes to the participant’s neighborhood. Results suggest acceptable internal consistency in this sample for second factor items (Cronbach’s alpha = .64) (see Table 3 for subscale descriptives). Discussion This report introduces a novel measure, which, to the best of our knowledge, is the first quantitative self-report assessment of neighborhood change and perceived gentrification. The present study sought to examine the underlying structure of NCGS using PCA, followed by oblimin rotation. We found that a two-factor solution accounted for nearly two-thirds of the variance and that both resultant factors had adequate internal consistency as separate subscales. The first factor, neighborhood disruption, composed of six items, was uniquely correlated with negative perceptions of neighborhood changes and displacement; the second factor, neighborhood gentrification, included four items and was uniquely correlated with the perception of recent influx of money and resources in the neighborhood. The Neighborhood Disruption subscale assesses connectedness of one’s experiences with their neighborhood, particularly in terms of how that connectedness has subjectively changed in the recent past. All items are coded such that higher scores indicate a greater sense of disconnection or separation. The Neighborhood Gentrification subscale includes several items that can be considered positive changes (for example, less crime, greater financial resources), but tend to be indicators of impending displacement for lower-income residents, particularly people of color (Freeman, 2005; Newman & Wyly, 2006). This is a preliminary study of a novel measure and, as such, has some limitations. The survey sample was recruited using quota sampling rather than probability sampling methods, and therefore may not be representative of Baltimore and New York City, or other U.S. cities. This may affect the generalizability of our results, which should be replicated in a probability sample when such data become available. We also did not include residents of suburban or rural areas, who may have experiences of community change that are qualitatively different from those of their urban counterparts. Likewise, children and adolescents were not included in our sample, although they may be particularly affected by both displacement of their families and influxes in resources. Furthermore, the internal consistency of the Neighborhood Gentrification subscale was low but in the acceptable range; this may be partly due to the low number of items in this subscale, which can bias Cronbach’s alpha toward lower scores (Emons, Sijtsma, & Meijer, 2007). Translations of NCGS into other languages (for example, Spanish, Chinese) would increase the potential applicability of the measure in U.S. cities, particularly in neighborhoods with large immigrant populations most affected by gentrification (Hwang, 2015). This study provides evidence for the utility of a new self-report measure to assess perceived neighborhood change. Whereas empirical knowledge on neighborhood gentrification has largely relied on population-level and aggregate measures, NCGS may be useful in future research to advance a more nuanced understanding of neighborhood gentrification and the variability of its impact on individuals. Social work research studies in urban areas should consider including measures of perceived gentrification, such as NCGS, to enhance our understanding of how neighborhood change and gentrification distinctively and collectively influences health, mental health, and service utilization and access. Jordan DeVylder, PhD, is associate professor, Graduate School of Social Service, Fordham University, 113 W. 60th Street, 7th floor, New York, NY 10023; e-mail: [email protected]. Lisa Fedina, PhD, is assistant professor, School of Social Work, University of Michigan, Ann Arbor. Hyun-Jin Jun, PhD, is postdoctoral fellow, School of Social Work, University of Maryland, Baltimore. References Anguelovski , I. ( 2015 ). Healthy food stores, greenlining and food gentrification: Contesting new forms of privilege, displacement and locally unwanted land uses in racially mixed neighborhoods . International Journal of Urban and Regional Research, 39 , 1209 – 1230 . Google Scholar Crossref Search ADS WorldCat Atkinson , R. ( 2000 ). The hidden costs of gentrification: Displacement in Central London . Journal of Housing and the Built Environment, 15 , 307 – 326 . Google Scholar Crossref Search ADS WorldCat Barton , M. A. ( 2016 ). An exploration of the importance of the strategy used to identify gentrification . Urban Studies, 53 , 92 – 111 . doi: 10.1177/0042098014561723 Google Scholar Crossref Search ADS WorldCat Crossref Beauregard , R. ( 1986 ). The chaos and complexity of gentrification. In N. Smith & P. Williams (Eds.), Gentrification of the city (pp. 35 – 55 ). Boston : Allen and Unwin . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Centers for Disease Control and Prevention. ( 2013 ). Health effects of gentrification . Retrieved from http://www.cdc.gov/healthyplaces/healthtopics/gentrification.htm Curtis , S. , Cave , B. , & Coutts , A. ( 2002 ). Is urban regeneration good for health? Perceptions and theories of the health impacts of urban change . Environment and Planning C: Government and Policy, 20 , 517 – 534 . Google Scholar Crossref Search ADS WorldCat DeVylder , J. E. , Jun , H. , Fedina , L. , Coleman , D. , Anglin , D. , Cogburn , C. , et al. ( 2018 ). Association of exposure to police violence with prevalence of mental health symptoms among urban residents in the United States . JAMA Network Open . doi: 10.1001/jamanetworkopen.2018.4945 OpenURL Placeholder Text WorldCat Crossref Emons , W. H. , Sijtsma , K. , & Meijer , R. R. ( 2007 ). On the consistency of individual classification using short scales . Psychological Methods, 12 ( 1 ), 105 – 120 . Google Scholar Crossref Search ADS WorldCat Freeman , L. ( 2005 ). Displacement or succession?: Residential mobility in gentrifying neighborhoods . Urban Affairs Review, 40 , 463 – 491 . Google Scholar Crossref Search ADS WorldCat Freeman , L. , & Braconi , F. ( 2004 ). Gentrification and displacement: New York City in the 1990s . Journal of the American Planning Association, 70 ( 1 ), 39 – 52 . Google Scholar Crossref Search ADS WorldCat Galster , G. , Andersson , R. , Musterd , S. , & Kauppinen , T. M. ( 2008 ). Does neighborhood income mix affect earnings of adults? New evidence from Sweden . Journal of Urban Economics, 63 , 858 – 870 . Google Scholar Crossref Search ADS WorldCat Hartman , C. ( 1979 ). Comment on neighborhood revitalization and displacement: A review of the evidence . Journal of the American Planning Association, 45 , 488 – 492 . Google Scholar Crossref Search ADS WorldCat Hwang , J. ( 2015 ). Gentrification in changing cities: Immigration, new diversity, and racial inequality in neighborhood renewal . Annals of the American Academy of Political and Social Science, 660 ( 1 ), 319 – 340 . Google Scholar Crossref Search ADS WorldCat Hwang , J. ( 2016 ). While some things change, some things stay the same: Reflections on the study of gentrification . City & Community, 15 , 226 – 230 . Google Scholar Crossref Search ADS WorldCat Hwang , J. , & Sampson , R. J. ( 2014 ). Divergent pathways of gentrification: Racial inequality and the social order of renewal in Chicago neighborhoods . American Sociological Review, 79 , 726 – 751 . Google Scholar Crossref Search ADS WorldCat King , A. C. , Sallis , J. F. , Frank , L. D. , Saelens , B. E. , Cain , K. , Conway , T. L. , et al. ( 2011 ). Aging in neighborhoods differing in walkability and income: Associations with physical activity and obesity in older adults . Social Science & Medicine, 73 , 1525 – 1533 . Google Scholar Crossref Search ADS WorldCat Lees , L. , Slater , T. , & Wyly , E. ( 2013 ). Gentrification . New York : Routledge . Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC Levy , D. K. , Comey , J. , & Padilla , S. ( 2007 ). In the face of gentrification: Case studies of local efforts to mitigate displacement . Journal of Affordable Housing & Community Development Law, 16 , 238 – 315 . OpenURL Placeholder Text WorldCat Lim , S. , Chan , P. Y. , Walters , S. , Culp , G. , Huynh , M. , & Gould , L. H. ( 2017 ). Impact of residential displacement on healthcare access and mental health among original residents of gentrifying neighborhoods in New York City . PLOS One, 12 ( 12 ). Retrieved from https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0190139 OpenURL Placeholder Text WorldCat Mehdipanah , R. , Manzano , A. , Borrell , C. , Malmusi , D. , Rodriguez-Sanz , M. , Greenhalgh , J. , et al. ( 2015 ). Exploring complex causal pathways between urban renewal, health and health inequality using a theory-driven realist approach . Social Science & Medicine, 124 , 266 – 274 . Google Scholar Crossref Search ADS WorldCat Newman , K. , & Wyly , E. K. ( 2006 ). The right to stay put, revisited: Gentrification and resistance to displacement in New York City . Urban Studies, 43 ( 1 ), 23 – 57 . Google Scholar Crossref Search ADS WorldCat Ohmer , M. L. , Coulton , C. , Freedman , D. A. , Sobeck , J. L. , & Booth , J. ( 2018 ). Measures for community and neighborhood research . Thousand Oaks, CA : Sage Publications . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Prince , S. ( 2016 ). African Americans and gentrification in Washington, DC: Race, class and social justice in the nation’s capital . New York : Routledge . Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC Rose , D. ( 1984 ). Rethinking gentrification: Beyond the uneven development of Marxist urban theory . Environment and Planning D: Society and Space, 2 , 47 – 74 . Google Scholar Crossref Search ADS WorldCat Sallis , J. F. , Slymen , D. J. , Conway , T. L. , Frank , L. D. , Saelens , B. E. , Cain , K. , & Chapman , J. E. ( 2011 ). Income disparities in perceived neighborhood built and social environment attributes . Health & Place, 17 , 1274 – 1283 . Google Scholar Crossref Search ADS WorldCat Valli , C. ( 2015 ). A sense of displacement: Long-time residents’ feelings of displacement in gentrifying Bushwick, New York . International Journal of Urban and Regional Research, 39 , 1191 – 1208 . Google Scholar Crossref Search ADS WorldCat Weden , M. M. , Carpiano , R. M. , & Robert , S. A. ( 2008 ). Subjective and objective neighborhood characteristics and adult health . Social Science & Medicine, 66 , 1256 – 1270 . Google Scholar Crossref Search ADS WorldCat Worthington , R. L. , & Whittaker , T. A. ( 2006 ). Scale development research: A content analysis and recommendations for best practices . Counseling Psychologist, 34 , 806 – 838 . Google Scholar Crossref Search ADS WorldCat © 2019 National Association of Social Workers This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

doi: 10.1093/swr/svz022pmid: N/A

In this index, the following abbreviations are used: Mar. for March, Sept. for September, Dec. for December, E for Editorial, and RN for Research Notes. SUBJECT AND TITLE INDEX Academic–practitioner relationships building knowledge to support human service organizational and management practice, June 115–127 Adolescents and young adults financial challenges of cancer for parent caregivers and, Mar. 17–30 phenomenological exploration of student mental health and college success, Sept. 145–156 young adult caregiver strain and benefits, Dec. 269–278 See alsoYouth violence; Youths Adversity psychotropic medication use in children and role of, June 81–89 African Americans Spirituality, employment hope, and grit: modeling the relationship among underemployed urban, Mar. 43–52 American Indians importance of biological parent coparenting in stepfamilies among, Sept. 168–180 Body image toward an understanding of racial and ethnic diversity in, June 69–80 Building Knowledge to Support Human Service Organizational and Management Practice: An Agenda to Address the Research-to-Practice Gap, by Bowen McBeath, Jennifer Mosley, Karen Hopkins, Erick Guerrero, Michael Austin, and John Trop man, June 115–127 Bullying statewide study of school-based victimization, discriminatory bullying, and weapon victimization by student homelessness status, Sept. 181–194 Cancer and financial challenges for adolescents and young adults and their parent caregivers, Mar. 17–30 Capacity building building knowledge to support human service organizational and management practice, June 115–127 reflections and new attention to long-standing themes in social work research (E), Sept. 131–132 Caregiving needs financial challenges of cancer for parent caregivers and adolescents and young adults, Mar. 17–30 mental health among older adults with, Sept. 157–167 young adult caregiver strain and benefits, Dec. 269–278 See alsoParenting Children prevalence and correlates of sex trafficking among homeless and runaway youths presenting for shelter services, June 91–99 role of adversity and psychotropic medication use in, June 81–89 See alsoAdolescents and young adults Cognitive Enhancement Therapy Improves Social Relationships Quality of Life among Individuals with Schizophrenia Misusing Substances (RN), by Jessica A. Wojtalik and Shaun M. Eack, Mar. 59–63 College success phenomenological exploration of student mental health and, Sept. 145–156 Community effects of the Your Family, Your Neighborhood intervention on neighborhood social processes, Dec. 235–246 filmed simulation to train peer researchers in community-based participatory research (RN), Sept. 195–199 Conceptualizations of Domestic Violence–Related Needs among Women Who Resettled to the United States as Refugees, by Karin Wachter, Jessica Dalpe, and Laurie Cook Heffron, Dec. 207–219 Depression importance of biological parent coparenting in American Indian stepfamilies, Sept. 168–180 parenting resilience factor for fathers with, June 101–112 Difference-in-Differences as an Alternative to Pretest–Posttest Regression for Social Work Intervention Evaluation and Research,by Roderick A. Rose and Natasha K. Bowen, Dec. 247–258 Disciplinary action impact on social workers, Mar. 5–15 Disproportionality in Juvenile Justice Diversion: An Examination of Teen Court Peer-Derived Consequences, by Katie Cotter Stalker, Dec. 221–233 Domestic violence. SeeIntimate partner violence Econometrics difference-in-differences as an alternative to pretest–posttest regression for social work intervention evaluation and research, Dec. 247–258 Effects of the Your Family, Your Neighborhood Intervention on Neighborhood Social Processes, by Daniel Brisson, Stephanie Lechuga Peña, Nicole Mattocks, Mark Plassmeyer, and Sarah McCune,Dec. 235–246 Ethics impact of disciplinary action on social workers, Mar. 5–15 Ethnicity. SeeRace and ethnicity An Examination of the Sufficiency of Small Qualitative Samples(RN), by Diane S. Young and Erin A. Casey, Mar. 53–58 Families effects of the Your Family, Your Neighborhood intervention on neighborhood social processes, Dec. 235–246 evidence from the National Transgender DiscriminationSurvey, Sept. 133–144 importance of biological parent coparenting in AmericanIndian stepfamilies, Sept. 168–180 role of social support as a moderator of housing instability insingle mother and two-parent households, Mar. 31–42 women and their mothers-in-law: triangles, ambiguity, and relationship quality, Dec. 259–268 Fathers parenting resilience factor for depressed, June 101–112 Filmed Simulation to Train Peer Researchers in Community-Based Participatory Research (RN), by Andrew David Eaton, Sept. 195–199 Financial Challenges of Cancer for Adolescents and Young Adults and Their Parent Caregivers, by Robyn J. McNeil, Maria McCarthy, David Dunt, Kate Thompson, Silja Kosola, Lisa Orme, Sarah Drew, and Susan Sawyer, Mar. 17–30 Gentrification Neighborhood Change and Gentrification Scale (RN), Dec. 280–285 Homelessness prevalence and correlates of sex trafficking among homeless and runaway youths presenting for shelter services, June 91–99 statewide study of school-based victimization, discriminatory bullying, and weapon victimization by student homelessness status, Sept. 181–194 Housing instability role of social support as a moderator of housing instability in single mother and two-parent households, Mar. 31–42 Human service organizations building knowledge to support management practice and, June 115–127 The Importance of Biological Parent Coparenting in an American Indian Stepfamily Context, by Kaitlin P. Ward and Gordon E. Limb, Sept. 168–180 In Their Own Words: A Phenomenological Exploration of StudentMental Health and Success in College, by Megan CallahanSherman, Sept. 145–156 Increasing the Impact of Social Work Scholarship in an Age of (Mis)Information (E), by Charlotte Lyn Bright, Dec. 205–206 In-laws triangles, ambiguity, and relationship quality for women and their, Dec. 259–268 Intimate partner violence women who resettled to the United States as refugees and conceptualizations of domestic violence–related needs, Dec. 207–219 Introduction to the New Editor-in-Chief (E), by Charlotte LynBright, June 67–68 Juvenile justice examination of teen court peer-derived consequences, Dec.221–233 Licensing impact of disciplinary action on social workers, Mar. 5–15 Management practice building knowledge to support human service organizationaland, June 115–127 Mental health cognitive enhancement therapy’s effects on individuals withschizophrenia (RN), Mar. 59–63 financial challenges of cancer for adolescents and youngadults and their parent caregivers, Mar. 17–30 importance of biological parent coparenting in AmericanIndian stepfamilies, Sept. 168–180 parenting resilience factor for fathers with depression, June101–112 phenomenological exploration of success in college and, Sept.145–156 psychotropic medication use in children and role of adversity, June 81–89 Mental Health among Older Adults with Caregiving Needs: The Roleof Social Networks, by Fengyan Tang, Heejung Jang,Elizabeth A. Mulvaney, Jane Seoyoon Lee, Donald Musa, andScott Beach, Sept. 157–167 Misinformation increasing the impact of social work scholarship in an ageof (E), Dec. 205–206 Mothers, single role of social support as a moderator of housing instability insingle mother and two-parent households, Mar. 31–42 Native Americans. SeeAmerican Indians The Neighborhood Change and Gentrification Scale: Factor Analysisof a Novel Self-Report Measure (RN), by Jordan DeVylder,Lisa Fedina, and Hyun-Jin Jun, Dec. 280–285 “Of All the Social Workers . . . I’m the Bad One”: Impact ofDisciplinary Action on Social Workers, by Michelle Gricus,Mar. 5–15 Older adults with caregiving needs and their mental health, Sept. 157–167 Parenting evidence from the National Transgender Discrimination Survey, Sept. 133–144 importance of biological parent coparenting in AmericanIndian stepfamilies, Sept. 168–180 role of social support as a moderator of housing instability insingle mother and two-parent households, Mar. 31–42 Paternal Self-Efficacy: A Parenting Resilience Factor for Fathers withDepression, by Mark Herrick Trahan and Kevin Shafer,June 101–112 Poverty effects of the Your Family, Your Neighborhood interventionon neighborhood social processes, Dec. 235–246 role of social support as a moderator of housing instability insingle mother and two-parent households, Mar. 31–42 Pretest–posttest regression social work intervention evaluation and research anddifference-in-differences as an alternative to, Dec.247–258 Prevalence and Correlates of Sex Trafficking among Homeless andRunaway Youths Presenting for Shelter Services, by JohannaK. P. Greeson, Daniel Treglia, Debra Schilling Wolfe, andSarah Wasch, June 91–99 Psychotropic Medication Use in Children: What Role Does ChildAdversity Play? by Héctor Ernesto Alcalá, Masako Horino, and Jorge Delva, June 81–89 Qualitative research examination of the sufficiency of small qualitative samples(RN), Mar. 53–58 Race and ethnicity toward an understanding of body image among women ofdiversity, June 69–80 See alsoAfrican Americans Racial equity disproportionality in juvenile justice diversion, Dec. 221–233 Refugees conceptualizations of domestic violence–related needsamong women who resettled to the United States as,Dec. 207–219 Relationship ambiguity women and their mothers-in-law: triangles and, Dec.259–268 Research training filmed simulation to train peer researchers in community-based participatory research (RN), Sept. 195–199 The Role of Social Support as a Moderator of Housing Instability in Single Mother and Two-Parent Households, by Stacia Martin-West, Mar. 31–42 Sanctions impact of disciplinary action on social workers, Mar. 5–15 Schizophrenia cognitive enhancement therapy’s effects on individuals with (RN), Mar. 59–63 Scholarship age of (mis)information and increasing the impact of (E), Dec.205–206 Schools statewide study of school-based victimization, discriminatorybullying, and weapon victimization by student homelessness status, Sept. 181–194 See alsoCollege success Sex trafficking homeless and runaway youths presenting for shelter servicesand prevalence and correlates of, June 91–99 Spirituality, Employment Hope, and Grit: Modeling the Relationship among Underemployed Urban African Americans, by DavidR. Hodge, Philip Young P. Hong, and Sangmi Choi,Mar. 43–52 A Statewide Study of School-Based Victimization, DiscriminatoryBullying, and Weapon Victimization by Student HomelessnessStatus, by Hadass Moore, Ron Avi Astor, and Rami Benbenishty, Sept. 181–194 Substance abuse cognitive enhancement therapy’s effects on individuals with (RN), Mar. 59–63 Survivor-centered approach conceptualizations of domestic violence–related needsamong women who resettled to the United States asrefugees, Dec. 207–219 Toward an Understanding of Racial and Ethnic Diversity in BodyImage among Women, by Virginia Ramseyer Winter,Laura King Danforth, Antoinette Landor, and DaniellePevehouse-Pfeiffer, June 69–80 Transfeminine Spectrum Parenting: Evidence from the National Transgender Discrimination Survey, by N. Eugene Walls, ShannaK. Kattari, Stephanie Rachel Speer, and M. Killian Kinney,Sept. 133–144 Transitions: Looking Back and Looking Ahead (E), by Kirk A. Foster,Mar. 3–4 Unemployment spirituality, employment hope, and grit: modeling the relationship among underemployed urban African Americans,Mar. 43–52 Violence disproportionality in juvenile justice diversion, Dec. 221–233 statewide study of school-based victimization, discriminatorybullying, and weapon victimization by student homelessness status, Sept. 181–194 women who resettled to the United States as refugees andconceptualizations of domestic violence–related needs,Dec. 207–219 Women conceptualizations of domestic violence–related needsamong women who resettled to the United States asrefugees, Dec. 207–219 role of social support as a moderator of housing instability insingle mother and two-parent households, Mar. 31–42 toward an understanding of racial and ethnic diversity in bodyimage among, June 69–80 Women and Their Mothers-in-Law: Triangles, Ambiguity, and Relationship Quality, by Geoffrey L. Greif and Michael E. Woolley,Dec. 259–268 Working Together Works Better: Reflections and New Attention toLong-Standing Themes in Social Work Research (E), by CharlotteLyn Bright, Sept. 131–132 Young Adult Caregiver Strain and Benefits, by Jessica KingMcLaughlin, Jennifer C. Greenfield, Leslie Hasche, andCarson De Fries, Dec. 269–278 Young adults. SeeAdolescents and young adults Your Family, Your Neighborhood intervention and effects on neighborhood social processes, Dec. 235–246 Youth violence disproportionality in juvenile justice diversion, Dec. 221–233 Youths prevalence and correlates of sex trafficking among homelessand runaway youths presenting for shelter services, June91–99 statewide study of school-based victimization, discriminatorybullying, and weapon victimization by student homelessness status, Sept. 181–194 AUTHOR INDEX Alcalá, Héctor Ernesto and Masako Horino and Jorge Delva, Psychotropic MedicationUse in Children: What Role Does Child Adversity Play? June81–89 Astor, Ron Avi. See Moore, Hadass Austin, Michael. See McBeath, Bowen Beach, Scott. See Tang, Fengyan Benbenishty, Rami. See Moore, Hadass Bowen, Natasha K. See Rose, Roderick A. Bright, Charlotte Lyn Increasing the Impact of Social Work Scholarship in an Age of(Mis)Information (E), Dec. 205–206 Introduction to the New Editor-in-Chief (E), June 67–68 Working Together Works Better: Reflections and New Attention toLong-Standing Themes in Social Work Research (E), Sept.131–132 Brisson, Daniel and Stephanie Lechuga Peña, Nicole Mattocks, Mark Plassmeyer, and Sarah McCune, Effects of the Your Family, Your Neighborhood Intervention on Neighborhood Social Processes,Dec. 235–246 Casey, Erin A. See Young, Diane S. Choi, Sangmi. See Hodge, David R. Dalpe, Jessica. See Wachter, Karin Danforth, Laura King. See Winter, Virginia Ramseyer De Fries, Carson. See McLaughlin, Jessica King Delva, Jorge. See Alcalá, Héctor Ernesto DeVylder, Jordan and Lisa Fedina and Hyun-Jin Jun, The Neighborhood Changeand Gentrification Scale: Factor Analysis of a Novel Self-ReportMeasure (RN), Dec. 280–285 Drew, Sarah. See McNeil, Robyn J. Dunt, David. See McNeil, Robyn J. Eack, Shaun M. See Wojtalik, Jessica A. Eaton, Andrew David Filmed Simulation to Train Peer Researchers in Community-BasedParticipatory Research (RN), Sept. 195–199 Fedina, Lisa. See DeVylder, Jordan Foster, Kirk A. Transitions: Looking Back and Looking Ahead (E), Mar. 3–4 Greenfield, Jennifer C. See McLaughlin, Jessica King Greeson, Johanna K. P. and Daniel Treglia, Debra Schilling Wolfe, and Sarah Wasch, Prevalence and Correlates of Sex Trafficking among Homeless andRunaway Youths Presenting for Shelter Services, June 91–99 Greif, Geoffrey L. and Michael E. Woolley, Women and Their Mothers-in-Law:Triangles, Ambiguity, and Relationship Quality, Dec. 259–268 Gricus, Michelle “Of All the Social Workers . . . I’m the Bad One”: Impact ofDisciplinary Action on Social Workers, Mar. 5–15 Guerrero, Erick. See McBeath, Bowen Hasche, Leslie. See McLaughlin, Jessica King Heffron, Laurie Cook. See Wachter, Karin Hodge, David R. and Philip Young P. Hong and Sangmi Choi, Spirituality,Employment Hope, and Grit: Modeling the Relationship amongUnderemployed Urban African Americans, Mar. 43–52 Hong, Philip Young P. See Hodge, David R. Hopkins, Karen. See McBeath, Bowen Horino, Masako. See Alcalá, Héctor Ernesto Jang, Heejung. See Tang, Fengyan Jun, Hyun-Jin. See DeVylder, Jordan Kattari, Shanna K. See Walls, N. Eugene Kinney, M. Killian. See Walls, N. Eugene Kosola, Silja. See McNeil, Robyn J. Landor, Antoinette. See Winter, Virginia Ramseyer Lee, Jane Seoyoon. See Tang, Fengyan Limb, Gordon E. See Ward, Kaitlin P. Martin-West, Stacia The Role of Social Support as a Moderator of Housing Instability inSingle Mother and Two-Parent Households, Mar. 31–42 Mattocks, Nicole. See Brisson, Daniel McBeath, Bowen and Jennifer Mosley, Karen Hopkins, Erick Guerrero, Michael Austin, and John Tropman, Building Knowledge to SupportHuman Service Organizational and Management Practice: AnAgenda to Address the Research-to-Practice Gap, June115–127 McCarthy, Maria. See McNeil, Robyn J. McCune, Sarah. See Brisson, Daniel McLaughlin, Jessica King and Jennifer C. Greenfield, Leslie Hasche, and Carson DeFries, Young Adult Caregiver Strain and Benefits, Dec269–278 McNeil, Robyn J. and Maria McCarthy, David Dunt, Kate Thompson, SiljaKosola, Lisa Orme, Sarah Drew, and Susan Sawyer,Financial Challenges of Cancer for Adolescents and Young Adultsand Their Parent Caregivers, Mar. 17–30 Moore, Hadass and Ron Avi Astor and Rami Benbenishty, A Statewide Studyof School-Based Victimization, Discriminatory Bullying, andWeapon Victimization by Student Homelessness Status, Sept.181–194 Mosley, Jennifer. See McBeath, Bowen Mulvaney, Elizabeth A. See Tang, Fengyan Musa, Donald. See Tang, Fengyan Orme, Lisa. See McNeil, Robyn J. Peña, Stephanie Lechuga. See Brisson, Daniel Pevehouse-Pfeiffer, Danielle. See Winter, Virginia Ramseyer Plassmeyer, Mark. See Brisson, Daniel Rose, Roderick A. and Natasha K. Bowen, Difference-in-Differences as an Alternative to Pretest–Posttest Regression for Social Work InterventionEvaluation and Research, Dec. 247–258 Sawyer, Susan. See McNeil, Robyn J. Shafer, Kevin. See Trahan, Mark Herrick Sherman, Megan Callahan In Their Own Words: A Phenomenological Exploration of StudentMental Health and Success in College, Sept. 145–156 Speer, Stephanie Rachel. See Walls, N. Eugene Stalker, Katie Cotter Disproportionality in Juvenile Justice Diversion: An Examinationof Teen Court Peer-Derived Consequences, Dec. 221–233 Tang, Fengyan and Heejung Jang, Elizabeth A. Mulvaney, Jane Seoyoon Lee, Donald Musa, and Scott Beach, Mental Health among Older Adults with Caregiving Needs: The Role of Social Networks, Sept. 157–167 Thompson, Kate. See McNeil, Robyn J. Trahan, Mark Herrick and Kevin Shafer, Paternal Self-Efficacy: A Parenting Resilience Factor for Fathers with Depression, June 101–112 Treglia, Daniel. See Greeson, Johanna K. P. Tropman, John. See McBeath, Bowen Wachter, Karin and Jessica Dalpe and Laurie Cook Heffron, Conceptualizations of Domestic Violence–Related Needs among Women Who Resettled to the United States as Refugees, Dec. 207–219 Walls, N. Eugene and Shanna K. Kattari, Stephanie Rachel Speer, and M. Killian Kinney, Transfeminine Spectrum Parenting: Evidencefrom the National Transgender Discrimination Survey, Sept.133–144 Ward, Kaitlin P. and Gordon E. Limb, The Importance of Biological ParentCoparenting in an American Indian Stepfamily Context, Sept.168–180 Wasch, Sarah. See Greeson, Johanna K. P. Winter, Virginia Ramseyer and Laura King Danforth, Antoinette Landor, and DaniellePevehouse-Pfeiffer, Toward an Understanding of Racial andEthnic Diversity in Body Image among Women, June 69–80 Wojtalik, Jessica A. and Shaun M. Eack, Cognitive Enhancement Therapy ImprovesSocial Relationships Quality of Life among Individuals withSchizophrenia Misusing Substances (RN), Mar. 59–63 Wolfe, Debra Schilling. See Greeson, Johanna K. P. Woolley, Michael E. See Greif, Geoffrey L. Young, Diane S. and Erin A. Casey, An Examination of the Sufficiency of SmallQualitative Samples (RN), Mar. 53–58 © 2019 National Association of Social Workers This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)