Social Work Research

Foster, Kirk, A

doi: 10.1093/swr/svz002pmid: N/A

Life, both personal and professional, is a series of transitions. Clients come and go. New colleagues are hired, and others move on to new opportunities. Loved ones are born into our lives and pass away from our lives. Executive directors change, and organizations mature. This is a certainty in all facets of our lives. Social work educators train students in effective termination techniques, thereby acknowledging in early professionalization efforts that transitions are omnipresent and should be handled deliberately and carefully. Transitions can be challenging. I am a creature of habit and enjoy the predictable in an unpredictable world. My wardrobe falls into various shades of blue and I hold onto physical things long past their usefulness simply because of their familiarity. Studies have shown that nostalgia brings comfort in stressful times (Batcho, 2013). As such, it is natural for us to look for stability in times of change and to embrace the familiar. Social Work Research has entered a transition period. We welcome a new editor-in-chief, Charlotte Lyn Bright from the University of Maryland School of Social Work, with this issue. This is our opportunity to look back and to look ahead—to note the journal’s accomplishments and to set out an agenda for the future. It has a new look and a new editor, but much of the rest remains the same. The mission of Social Work Research is to publish “exemplary research to advance the development of knowledge and inform social work practice.” It will continue to do this. This journal is home to rigorous and impactful research that shapes thinking about social work practice and future research endeavors. The editorial board encourages submissions to the journal that focus on translating research to practice and policy. In this time of transition, you will notice few differences—reviews will continue, staff will continue processing new submissions, and editorial decisions will be made. The average reader and author should experience no notable changes at this time. The journal accepted only 10 of 114 (8.8%) submissions in 2018 and published 23 articles. Social Work Research had a small publication backlog but will be current in 2019, and we expect to trend at more typical levels of acceptance. The editorial board encourages authors to consider the Grand Challenges for Social Work (GCSW) to shape their submissions. We also encourage submissions focused on social justice and social policy. Articles published in Social Work Research will not only build from strong theoretical foundations and research methods, but also make clear practice and policy implications that inform the profession. Social Work Research is home to high-quality, rigorous research that challenges our thinking and moves the profession in new directions. I expect nothing less as a member of the editorial board. As we look back, we thank James Herbert Williams for his service as editor-in-chief from 2013 to 2018. The journal grew positively under his leadership. The impact factor increased, and the number of manuscript submissions continued, on average, to grow each year. James Herbert sought to expand the international reach of Social Work Research by increasing the numbers of articles published from international scholars examining important issues around the globe. He pushed social work scholars to consider the GCSW (Williams, 2016) and their implications for research and practice. Social Work Research continued to publish high-quality, rigorous, and impactful research under his leadership. We live in this legacy and thank James Herbert for building a strong foundation for the next chapter in the life of this journal. And we pause to remember our friend, colleague, and mentor Matthew O. Howard, who passed away in December 2018. Matthew was the editor-in-chief of Social Work Research from 2009 to 2012. He was an intellectual powerhouse with a passion for promoting and conducting high-quality, methodologically rigorous, and impactful research. During his tenure, the journal’s impact factor rose as he sought to enhance the quality of manuscripts published. Matthew was instrumental in working with NASW Press staff on implementation of the journal’s electronic peer-review process. I recall having a conversation with him during this time; we discussed the meaningfulness of reviewer summary ratings and the best way to ensure rigorous and fair reviews. Matthew was a stickler for details, from commas in reports (and I wrote several with him) to research mechanics. He expected this same attention to detail and passion for methods and impact of all of us who worked with him and who reviewed for the journal. Matthew did this because, ultimately, he cared about the profession and the people served through our research. If we were not promoting evidence-based practice and if that evidence had not been thoroughly interrogated through peer review, then we were doing a disservice to the social work profession and its clients. It was this passion—to enhance knowledge that addresses human suffering—that made Matthew who he was. Matthew mentored many people through his decades of service to the social work academy. He touched the lives of countless social work faculty (many of us then-MSW or PhD students) through his time as a faculty member at Washington University in St. Louis, the University of Michigan, and the University of North Carolina at Chapel Hill, and his work on the boards of the Society for Social Work and Research and the American Academy of Social Work and Social Welfare. He instilled in me a commitment to good research and that same attention to detail. I remain a stickler for sampling methods because of my work managing his first two National Institutes of Health–funded studies in the early 2000s. Matthew’s impact on the profession reaches well beyond his over 200 publications, his journal editorships, and his innumerable presentations. His impact is and will remain most evident through those he mentored. The academy misses him already. Social Work Research is in a time of transition as James Herbert Williams passes the mantle of editor-in-chief to Charlotte Bright. As in all things, this is an opportunity for the journal to grow in new directions while embracing its mission and building on the good work done by previous editors-in-chief. As a member of the editorial board, I encourage you to continue viewing Social Work Research as the venue to publish your high-quality, methodologically rigorous, and impactful work. Kirk A. Foster, PhD, MSW, MDiv, is associate dean for Diversity, Equity, and Inclusion and associate professor, University of South Carolina College of Social Work. He is a member of the Social Work Research editorial board References Batcho , K. I. ( 2013 ). Nostalgia: Retreat or support in difficult times? American Journal of Psychology , 126 , 355 – 367 . Google Scholar Crossref Search ADS PubMed Williams , J. H. ( 2016 ). Grand challenges for social work: Research, practice, and education [Editorial] . Social Work Research , 40 , 67 – 70 . Google Scholar Crossref Search ADS PubMed © 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)

Gricus,, Michelle

doi: 10.1093/swr/svy023pmid: N/A

Abstract The Code of Ethics of the National Association of Social Workers makes several professional responsibilities clear; however, no explicit duty exists to restore dignity and reintegrate a social worker who has been reported to the licensing board for engaging in unethical or unprofessional behavior. This article, which reports the qualitative results of a mixed methods study, examines the personal and professional effects of disciplinary action on those social workers who have been sanctioned by a state board of social work. Through a symbolic interactionist lens, the author interviewed 18 social workers who had been disciplined by their licensing board and found that participants were often notably affected by the experience. The themes in the findings involved negative psychological and personal impacts, long-term vocational concerns, poor treatment by the board, and the importance of support to endure the experience. A notable social work implication includes generating clearer responsibilities for social workers around rehabilitation of those who have been disciplined. Headlines regarding social workers often focus on professional or personal misconduct (Reamer, 1995). Social workers are in the news for instances of identity theft, sexual contact with a vulnerable person, or lack of follow-through that resulted in a poor outcome. Such information regarding these violations is publicly accessible through state licensing boards and the National Association of Social Workers (NASW) and has been studied from several angles throughout the past 20 years (Boland-Prom, 2009; Daley & Doughty, 2006; Strom-Gottfried, 2000). The types of violations most frequently appearing across studies are boundary violations (which includes sexual contact with clients), “poor practice,” and improper treatment (Boland-Prom, 2009; Daley & Doughty, 2006; Reamer, 1995; Strom-Gottfried, 2000). Similar findings have been tracked across mental health disciplines (Garfinkel, Bagby, Waring, & Dorian, 1997; Neukrug, Milliken, & Walden, 2001). Because they are the most frequently reported complaint against mental health professionals, sexual violations between providers and their clients have been researched most heavily (Feldman-Summers & Jones, 1984; Gutheil & Brodsky, 2008; Lamb & Catanzaro, 1998; Luepker, 1999; Somer & Saadon, 1999). The research contains little about the impact of other types of violations on clients’ well-being and even less about the impact of a violation on the professional (Reamer, 2003). Because of the scarcity of this research, it is important to gain a better understanding of the effects that disciplinary action has on social workers. The purpose of this exploratory article was to identify the personal and professional effects of disciplinary action on social workers who have been sanctioned by a state board of social work to learn how to support this subset of the population while protecting the integrity of the profession. Violations Practice violations are reported to state licensing boards or directly to NASW, who determine the outcome of the social worker’s actions, though some violations likely go unreported. The number and types of violations by social workers have been reviewed at various intervals since 1956 (Berliner, 1989; McCann & Cutler, 1979; Reamer, 2003; Strom-Gottfried, 2000). Whereas these studies once focused on the violations reported to NASW, more recent studies have focused on state licensing board data available through state Web sites and the disciplinary records available through the Association of Social Work Boards (Boland-Prom, 2009; Daley & Doughty, 2006). Data collected over the past 15 years indicate similar patterns surrounding the nature of the violations. Strom-Gottfried’s (2000) often-cited research examined 781 total violations and found that the most frequently reported complaint involved boundary violations (n = 254, 32.5%), which includes a subset of 107 involving sexual relationships with a client or supervisee. Claims data from the NASW insurance division showed that 18.61% (n = 118) of malpractice suits were related to “incorrect treatment,” and 18.45% (n = 117) of them were related to “sexual impropriety” (Reamer, 1995). These two categories alone made up nearly 60% of the dollars paid to the claimant through lawsuits. Daley and Doughty (2006) reviewed complaints in Texas and found 22.3% (n = 310) classified as “poor practice” and 21.0% (n = 291) related to “boundary violations.” Boland-Prom (2009) found similar results involving boundaries in her review of 874 violations reviewed through 27 state licensing boards, indicating that dual relationships (both sexual and nonsexual) were the most sanctioned offense (n = 205, 23.4%). Complaints are often filed by a social worker’s colleagues rather than by a client or a person closely connected with a client (Berliner, 1989; McCann & Cutler, 1979; Strom-Gottfried, 2000). It is important to note that although many violations and lawsuits directly involve clients, not all do. Lack of compliance with licensing board regulations such as working without a license or a lapsed license represented 18.2% (n = 159) of all violations, failure to meet continuing education requirements accounted for 10.5% (n = 92) of offenses (Boland-Prom, 2009), and 4.2% (n = 33) were related to negative colleague or coworker interactions (Strom-Gottfried, 2000). The data collected in these studies did not include any information from the social workers themselves. Sanctions Consequences for disciplinary action vary by type and length of time between the reportable action and the resolution of the complaint. Lack of criteria, withdrawals of complaints, or complaints related to procedural errors are often not referred for hearings and thus do not result in sanctions in a majority of the cases filed (Strom-Gottfried, 2000). McCann and Cutler (1979) found that the most frequent reprimand was a private condemnation in 24.6% (n = 38) of complaints. Similarly, Boland-Prom (2009) indicated that a “reprimand or letter of admonishment” was the most serious consequence levied in 21.0% of complaints and a revocation of license or nonrenewal made up 12.1% of the cases. The average time for adjudication has also varied significantly, from several months to several years, which can result in significant legal expenses and lost productivity for the social worker and, often, the employing agency (Reamer, 1995). The impact of this timeline on clients and others involved in the disciplinary action is unknown. Correlational data between the nature of the offense and the type of disciplinary action were not found in the existing literature. Some violations, namely those related to neglect, theft, assault, or impaired driving, could also have significant legal consequences for the offending social worker. However, no consistent data were found to substantiate this. Scholarship available on the sanctions of disciplinary action was limited to aggregate data and did not involve the impact of the experience on the social workers involved. Unethical Behavior: Contributing Factors Unethical behavior or patterns of behavior by social workers can start out as intentional or unintentional instances of crossing the line, and boundary crossings can turn into boundary violations if left unchecked (Epstein, Simon, & Kay, 1992). Given the harm it causes clients and the consequences it has for the professional, professional–client sexual contact has been researched more than any other ethical violation. Between 2% and 4% of social workers in the United States reported engaging in sex with their clients (Berkman, Turner, Cooper, Polnerow, & Swartz, 2000; Jayaratne, Croxton, & Mattison, 1997), and 29.2% of complaints received by the NASW involved sexual activity with a client, the reports for which may have come from the clients themselves or from other sources (Berkman et al., 2000). Qualitative data on professionals who identify as therapists, which include disciplines other than social work, indicated several themes that can lead up to sexual contact with clients. These themes include exploitation of the client for the therapist’s own interests (Pope, Sonne, & Holroyd, 1993), role reversal in which the therapist begins sharing more about his or her personal life (Somer & Saadon, 1999), personal stress or transitions (Garfinkel et al., 1997), rationalization of the therapist’s actions (Gutheil & Brodsky, 2008) because the sexual contact was seen as a therapeutic intervention (Somer & Saadon, 1999), or a result of the client’s increased self-esteem (Berkman et al., 2000). Therapists who were themselves engaged in sexual contact “with their own therapists, supervisors, or educators at an earlier time have an increased probability of becoming offending therapists, supervisors, or educators themselves” (Lamb & Catanzaro, 1998, p. 498). Epstein et al.’s (1992) self-administered exploitation index has proven to be a useful tool in alerting professionals to many of these risk factors; however, it does not examine the impact the disciplinary action itself had on the professional. Impact on Clients Inevitably, clients and their loved ones who are the casualties of a social worker’s unethical or unprofessional behavior are affected by those experiences. Whereas little is known about the effects of other types of violations on clients (Reamer, 1995), sexual contact between clients and their social workers results in clients experiencing mistrust, anger, psychosomatic symptoms (Feldman-Summers & Jones, 1984; Gutheil & Brodsky, 2008), and notable mental health consequences, including posttraumatic stress disorder, depression and suicidal ideation, and alcohol and drug use (Luepker, 1999). In addition, 18% have been revictimized by other professionals they sought help from after the initial violation (Luepker, 1999). Impacts of Corrective Action across Disciplines The impact of disciplinary action on social workers does not appear in the literature specifically, but research in related disciplines reveals that disciplinary action stemming from unprofessional and unethical behavior can have important life-changing consequences for a practitioner (Coy, Lambert, & Miller, 2016). Marriage and family therapists reported feeling “powerless,” and did not feel supported to rehabilitate by the licensing board in any way during the process (Coy et al., 2016). Psychologists reported experiencing significant stress, anxiety, and depression because of licensing board investigations (Thomas, 2005). For those whose sanction included revocation or surrender of a license, a deep sense of grief accompanied the loss of professional identity (Coy et al., 2016). Lacking in the literature are studies focused on the impacts disciplinary action have on social workers (McAuliffe, 2005). One of the 13 standards in the NASW Code of Ethics specifically indicates that “social workers should defend and assist colleagues who are unjustly charged with unethical conduct” (NASW, 2017, “Unethical Conduct of Colleagues,” para. 5); however, there is no explicit definition of unethical conduct, nor is there an identified professional responsibility to restore dignity and eliminate maltreatment of a social worker who has been reported to the licensing board for engaging in unethical or unprofessional behavior. This absence of language may contribute to the “poignant sense of loneliness or isolation” experienced by social workers who grapple with ethical dilemmas in general (Holland & Kilpatrick, 1991) and may put them at risk of violating ethical principles or licensing standards (Council on Social Work Education, 2015; MacDonald et al., 2015). Method Participants To determine the impacts of disciplinary action on social workers, this study was conducted with a subset of social workers who were disciplined by the state licensing board in one midwestern state for engaging in unethical or unprofessional conduct. Although some information regarding licensed social workers is publicly available online, consistent contact information is not. As a result, I purchased contact information from that state’s social work board regarding licensees who had been disciplined by that board between the years of 2006 and 2015. This list contained 154 unique names. Procedures The mixed methods study incorporated both surveys and in-person interviews. Surveys were distributed via Qualtrics, online survey software, for those participants for whom e-mail addresses were known, and the U.S. mail for whom only mailing addresses were available. A total of 154 surveys were distributed, and 16 surveys were returned as undeliverable via e-mail or mail, making the total sample 138 participants. Participants who did not respond to the initial survey request were sent a second request two weeks later, and a total of 39 (28.2%) surveys were completed in a one-month period. The last question of the survey asked respondents for their contact information if they were interested in being interviewed to discuss their experience further. Interview data were expected to capture more in-depth perspectives than the survey alone could provide (Creswell, 2014). A total of 23 (58.97%) survey respondents indicated interest in participating in an interview and provided their preferred contact information. After receipt of this information, the researcher made two attempts to reach each respondent. Four respondents could not be reached, and one withdrew from the study before setting up an interview. Seventeen participants were interviewed face-to-face or over the telephone and one respondent provided responses to questions via e-mail. Each participant received a $25 gift card for their participation in the interview. This article focuses only on the qualitative strand of the study related to the psychological and vocational impacts of disciplinary action. Between May and July 2016, qualitative interviews were conducted in locations of the participants’ choice, which included private meeting rooms at coffee shops, libraries, and participants’ workplaces. Interviews lasted an average of 45 minutes. The interviews followed a semistructured procedure and included questions about the events leading up to the violation as well as the subsequent impacts of the violation. Questions and prompts included “Please tell me about how the whole experience affected your life” and “In what ways has this incident affected your view of the social work profession and your identity as a social worker?” Probing questions were used to help participants expand on their responses. Although information about each of the violations was publicly available online through the participant’s name, the aim of this study was to capture the voice and experience of the affected social workers and not to corroborate those personal accounts with public documentation. Thus, all the data used for this study came from the participants themselves. Inevitably, analyzing qualitative data through coding themes includes a degree of subjectivity (Saldaña, 2015). Segments of randomly selected transcript components were reviewed by an outside researcher and checked for validation. The results of the coding process were reviewed by an experienced qualitative researcher who assisted with identifying themes and subthemes from broad categories. Protection of Human Participants This study received institutional review board approval from a private midwestern university. Participants were e-mailed a copy of the consent form to review in advance of the interview. Before beginning the interview, I reviewed the consent form with the participant and asked questions to ensure clarity. Signatures were collected on paper or electronically before beginning the interview. This study incorporated a member checking component, in which I restated or summarized information heard from the participant to determine whether the summary accurately reflected the viewpoint of the participant (Harper & Cole, 2012). If accuracy and completeness were affirmed, the follow-up was complete. During the interview, the participants could decline to answer any question or terminate the interview at any time. Data Analysis All interviews were audiotaped and transcribed verbatim by me or a professional transcription service (I checked for accuracy each transcript provided by the transcription service). Data were analyzed thematically using the software program MAXQDA 12, selected for its ability to organize qualitative data. In vivo coding, which involves selecting meaningful words and phrases from qualitative data, was selected in the first round of coding. This approach is considered appropriate for honoring the voice of ostracized participants (Saldaña, 2015). Then, I organized the data by categories and again into themes and subthemes. The interview participants consisted of 12 women (67%) and six men (33%). Seven (39%) reported being between the ages of 25 and 34, one (5%) between the ages of 35 and 44, nine (50%) were between 45 and 54, and one (5%) participant declined to answer. Participants earned their most recent social work degree between 1982 and 2016, with 13.5 years in practice on average. During the time of the interviews, five (28%) were licensed at the BSW level, five (28%) at the MSW level, six (33%) were licensed as independent clinical social workers, and two (11%) were no longer licensed as social workers. The types of disciplinary action and the sanctions levied by their state board of social work for those actions, as reported by the participants themselves, are found in Table 1 and Figure 1, respectively. Three participants (15%) received their disciplinary action in the six months prior to completing the survey, four (20%) in the previous seven to 11 months, four (20%) in the past one or two years, six (30%) between three and five years prior, and two (10%) in the past six to 10 years. Table 1: Nature of Violation Leading Up to Disciplinary Action Nature of Violation % n Basic practice (for example, records, consent, confidentiality) 16 3 Dual relationships and boundary violations 26 5 License-related problems (for example, continuing education units, practicing without a license) 53 10 Personal 5 1 Total 100 19a Nature of Violation % n Basic practice (for example, records, consent, confidentiality) 16 3 Dual relationships and boundary violations 26 5 License-related problems (for example, continuing education units, practicing without a license) 53 10 Personal 5 1 Total 100 19a aParticipants could indicate more than one type of violation. Table 1: Nature of Violation Leading Up to Disciplinary Action Nature of Violation % n Basic practice (for example, records, consent, confidentiality) 16 3 Dual relationships and boundary violations 26 5 License-related problems (for example, continuing education units, practicing without a license) 53 10 Personal 5 1 Total 100 19a Nature of Violation % n Basic practice (for example, records, consent, confidentiality) 16 3 Dual relationships and boundary violations 26 5 License-related problems (for example, continuing education units, practicing without a license) 53 10 Personal 5 1 Total 100 19a aParticipants could indicate more than one type of violation. Figure 1: View largeDownload slide Types of Sanctions from the Board of Social Work Figure 1: View largeDownload slide Types of Sanctions from the Board of Social Work Results Thematic analysis identified six major themes connected to the personal and professional effects of disciplinary action on this study’s participants. They are described in the following sections. Negative Effects of Disciplinary Action on Personal Life Most participants described negative psychological and personal impacts of the experience. These impacts began as soon as the first contact was made with the board of social work and sometimes extended a decade into the future. Several participants connected the experience of disciplinary action with feelings of being watched. Across interviews, participants indicated that the experience created a sense of hypervigilance in their professional and personal lives. Expressions such as “a hammer could drop on my head,” “walking on eggshells,” and “waiting for the other shoe to drop” were common. Another participant reported that the experience caused “this kind of overwhelming panic and fear about everything that I say and do.” And another participant stated, “We become so afraid of doing harm that we’re not doing any good either.” For another participant, the effects permeated my whole ideas of self-worth and self-esteem, and I find myself just dealing with the lasting effects of doubt with decisions I make and who I can kind of trust and where I can get help. . . . These mistakes that are going to follow me forever. [The feelings] sometimes take over my life. Participants also identified feelings of embarrassment, shame, fear, anxiety, and shock. One participant described her anxiety in this way: “I would just have all of a sudden these panic attacks. . . . I left that hearing, and I [said], ‘I’m a really horrible person.’” Another participant said, “It was one of the most painful things I think I’ve ever been through, really hard.” Another participant indicated that the feelings are just as strong as they were when the incident occurred: The stress . . . hasn’t let up. I’ve just had to learn to . . . put it in a box and just put it on the shelf . . . and just say, this is what I have control of . . . and this is what I don’t. The negative personal effects extended to long-term physical impacts as well. Several participants explained notable periods of sleep problems, including nightmares and sleeplessness. One participant who reported having continuing stomach problems said, “There’s not a doubt in my mind that it’s probably taken some time off my life.” The financial impact of the disciplinary action was also shown to create hardships. Attorney fees, fines from the board of social work, and costs of evaluations were mentioned. One participant reported filing bankruptcy because of the expenses incurred during the disciplinary process and the loss of income during that time. Long-Term Effects of Licensure Violations on Vocational Life When considering the effects the disciplinary action has had on their lives, a few participants reported that the experience was “no big deal” and “not the end of the world.” However, many participants indicated the long-term vocational effects they experienced in addition to the personal and psychological impacts of disciplinary action. Scarlet Letter By far the most often identified consequence across participants, regardless of nature of the violation, was the permanent scarlet letter of having their name listed on the board of social work Web site. In the state where the study’s sample was derived, all board actions are subject to listing, regardless of the length of time since or the severity of the violation. The experience was described in this way by one participant: “Of all of the social workers in the state, I’m the bad one”; however, that opinion was felt across participants’ responses. “There’s this bad mark against me. It’s not that it’s wrong or unfair. It’s that it’s never going to go away”; “I think there should be a statute of limitations”; “My name goes on the ‘bad social worker list’ forever.” One participant said that continuing to publicly post the disciplinary action is misaligned: I’m still doing the work so clearly I might have did [sic] something back in the day, but I’ve clearly come through and I’m OK now and you’re still letting me do the work. I don’t feel the need for it to come up 15, 20 years later. Another participant agreed, seeing the violation of practicing without a license as not warranting the permanent sanction: “It is just humiliating that they can publicly post my name as being disciplined when it was a simple mistake and lack of education.” Similarly, another participant felt trapped: “There’s no way of outgrowing or earning your way out by good behavior for this reprimand. It stays on my record.” Another participant compared the stigma of this list with the registry of people with criminal sex offenses. Three participants lost their jobs as a direct consequence of boundary violations against a client or a coworker, and they indicated the difficulty of finding and keeping employment connected to the “big black mark” associated with their social work license. One participant stated, “[One supervisor] wanted to hire me and ultimately somebody further up the ladder wouldn’t let her,” and another had a similar experience: “I went to two interviews . . . but when I had to tell them, then it was all shut down.” Several other participants worried about the ways in which the sanction would affect future hiring decisions. For example, one participant stated, “Every time I fill anything out, I have to say, ‘Yes, I’ve had a disciplinary action,’ and it looks horrible . . . a big fat yes, you don’t look any further, you have no idea.” Another wondered, “Is this going to ruin my career? . . . What does that mean for my life?” Concealing and Revealing Participants expressed different viewpoints regarding whether to disclose information about the disciplinary action. Shame and embarrassment seemed to drive several participants’ decisions regarding “the dark secret.” One participant expressed some hesitancy about sharing the information: “I wasn’t reluctant, necessarily, to disclose that to my employer, but to have to open up about that after being in a new position for a period of time, it was embarrassing.” Another participant saw disclosure as a necessary component of surviving the experience: “Because I was not willing to walk alone feeling ashamed as everyone at every single agency I consulted with, I went directly to the person in charge and told them.” Others chose not to disclose information about the disciplinary action in any professional capacity, finding the disclosure “irrelevant” or unnecessary. “I don’t know if my boss there, my immediate supervisor ever actually knew or not,” one participant said. Several shared the sentiment of the participant who stated, “I didn’t tell them what had happened because I didn’t have to.” Change for the Better Some participants stated that some good came out of the experience. These benefits have improved participants’ practice and changed agency policy. About practice improvements, one participant reported, “I’ve learned a lot about how to be professional and how to carry myself in that professional role”; another stated, “I do better record keeping than I’ve ever done in my entire life.” Another participant reported, “I’ve become aware that I probably was not as vigilant as I thought . . . I’ve become much more vigilant in my practice.” Another participant saw improvements by accessing supervision and consultation more frequently. Mirrors and Magnifying Glasses When discussing the incident that led up to the disciplinary action, participants reported viewpoints that indicated they had made a mistake and should face some consequence of that error. Other participants saw the situation as “unfair” or unjustified, which highlighted participants’ beliefs about who was responsible for the disciplinary action. These attributions were also extended to the ways in which a participant saw his or her violation in relation to those of others. I Made a Mistake Some participants identified a sense of personal responsibility for the disciplinary action. These participants stated that they were “guilty” of making an error, and “I learned my lesson.” Several noted that they took “full responsibility” for their actions. One participant stated, “I made this mess-up, and I am going to be an adult and take the consequences.” Looking back on the situation, one participant stated, “I didn’t realize how proactive I needed to be to know all of the rules and regulations.” Another participant continued to stand by the decision that led to the disciplinary action, which involved a boundary concern that “saved [the client’s] life,” though understood that the decision was outside the norm of typical social work practice. Some participants acknowledged that the situation could be attributed to an error on their part; however, the circumstances required some understanding and “leeway” from the board of social work, adding that the board “should be looking at individual circumstances.” It’s Not My Fault Several participants felt that the incident was a result of the actions of another. The responsibility was shifted externally, extending to the board of social work, university professors, employers, and clients. Several participants whose disciplinary action involved becoming licensed shared the perspective that “no one told me.” Another participant, whose violation involved boundaries with a client, attributed the situation to that person, indicating that the client took no responsibility: “The lies [the client] told me . . . through the course of therapy were well thought out and [the client] defended them vigorously.” This Is Not as Bad as You Think Regardless of whether the participant looked in the mirror regarding the licensing violation or examined the external environment with a magnifying glass, participants seemed to find it helpful to lessen the impact of their violation through comparison with others who had broken professional or societal rules. During the interactions with the board of social work, one participant said it felt “like I was being investigated for a murder.” How the board viewed the violation and how the participant viewed the violation generated the most comparison responses. One participant defended the violation: “It seemed from my perspective I was not maliciously, intentionally trying to beat the system or try to practice or try to be sneaky and not paying dues or whatever.” Another participant indicated, “It’s one thing if you’re sexually abusing a client, or if you’re embezzling money. I mean, if you’re doing something bad, this is a problem. But, I just didn’t feel like mine was on that same level.” Other comparisons included “doctors who write faulty prescriptions,” “hidden money from the IRS,” “abusing a child client,” “falsifying records,” bribery, and “purposefully harmed a client.” Shame on You Many participants saw their interactions with the board of social work as directly related to the way they interpreted the impact of the disciplinary action. Whereas some indicated that they were treated with respect by the board and were provided adequate information to make informed decisions about their disciplinary action, many more participants felt belittled and shamed during the experience. These participants described their interactions with the board as “intimidating” and “not sympathetic, not empathetic, not caring and considerate.” One participant stated, “I didn’t feel safe. I didn’t feel comfortable” talking to the board. Others indicated that the board “slapped me on the wrist” and “was shaking its finger at me” and they hoped that the board “would have just listened.” Another said, “I didn’t need to be treated that way, shamed and yelled at.” Presumption of Guilt Compacting their negative feelings about the interactions with the board, participants also perceived the experience lacking a presumption of innocence and feeling like “a criminal” stating that “it’s not you’re innocent until proven guilty. It’s you’re guilty, and you need to prove your innocence.” Other participants’ responses aligned, stating that “they had decided everything before I even walked in. I was way guilty before I even sat down” and “[The board said] ‘This is what will happen. If you want to appeal this, you have the right to. But this is what our findings are.’ Why would I waste that money or time?” No Credit for Good Work Another effect of the interactions with the board of social work was that participants, specifically those with long social work careers, felt that the time spent practicing social work before the violation went unacknowledged in discussing the violation with the board during the process. Overwhelmingly, the belief was that “none of it mattered.” One participant with 25 years of experience said, “I’d had a completely blameless record, it was perfect. I’d done a lot of really good things. . . . None of it counted. It was never addressed. It was never discussed.” Another participant similarly stated, “It didn’t matter that [my employer] stood behind me. It didn’t matter that I had a spotless 19-year history with [my employer].” This experience caused a substantial amount of grief for another participant: “I’m grieving the fact that for all the years of service that I did for the people that I have helped, for . . . acknowledgments that I have received . . . it doesn’t matter.” Change of Heart The process of disciplinary action had such a powerful impact on some of the participants that their entire view of, pride in, and responsibility to the profession of social work were irreparably altered. Some participants indicated feelings of abandonment, and in turn responded by (or are considering) leaving the field: “I am completely moving away from social work”; “I’m selling [my social work practice]. I’m done. It’s not worth the risk”; “If this doesn’t work out for me I will not probably be pursuing a position in social work.” Although still practicing currently as a licensed social worker, one participant indicated feeling like an outsider: “It just feels like a world that I am never going to be a part of.” Some participants were disheartened by the experience and indicated regret or questioned their decision to go into social work as a career. One participant said, “My takeaway from this whole experience is that I’m sorry that I became a clinical social worker. I really am.” Another reported, “I am extremely ambivalent about the profession and probably will retire feeling like that.” Several others indicated that they were once proud to be social workers, but no longer felt that way. Importance of Support to Survive the Process During their interviews, participants were asked to identify the kinds of support and reactions they received from personal and professional relationships during and after the disciplinary process. Participants identified these relationships as both helpful and hurtful at various stages. Participants who felt supported indicated that those relationships made an important difference in the way the experience affected them in the long term. Several participants felt “very supported,” and the people in their lives were “encouraging.” One participant whose violation was related to practicing without a license felt that support, and also issued a warning to those supporters: I had quite a bit of support from people who were social workers who were completely in disbelief also. . . . A few of them, sadly, are going to end up going through the same process. . . . I just told them, “This is probably going to happen to you.” One participant’s employer paid the fine levied by the board of social work for the violation, and several participants’ supervisors wrote letters to the board or accompanied the participant to board hearings. Another participant indicated that the personal support was life-saving: The people who were closest to me . . . would really go out of their way to make sure that I was doing OK emotionally . . . that I had a roof over my head, that I had food in my cabinets, and that I was able to keep doing what I needed to do to keep working and to keep my head above water and to be able to keep taking care of my kids. Other participants experienced negative reactions from personal and professional relationships, which participants identified as exacerbating the impact of the experience. They reported feeling “judged,” “betrayed,” and “humiliated” by the people in their lives. One participant said this of her employer: “I felt like I was hung out to dry. They basically said, ‘This is your problem. You deal with it.’” For those participants who experienced serious financial hardship and job loss, personal relationships seemed to be affected even more negatively. One participant stated, “The event that led to the disciplinary action actually prompted the end of my marriage, and so it really turned my whole life right upside down.” Another participant indicated that a sibling was the one responsible for reporting the incident to the board of social work, which resulted in the permanent loss of license and severed ties with the family. The shame and embarrassment of the violation resulted in several participants choosing not to (or waiting to) access support: “Talking about it right now is basically the first time I’ve openly spoken about it, which I know is not probably the best, and it’s one of the reasons why I still deal with it so physically and emotionally.” This same participant went on to say that fear drove her not to disclose: Because I was so fragile. If I heard someone close to me say something like, “Oh my God, that is really bad,” I maybe wouldn’t have been able to take that, and so that’s why I withheld so much and probably played it off [to] myself like it wasn’t that big of a deal. Another participant reported that lack of support felt isolating. “I felt kind of alone . . . I didn’t have any help or support from anyone. . . . It was a very lonely, frustrating place to be.” Discussion This study provides new information about an aspect of the profession that has not been previously examined in social work literature—the impacts of disciplinary action on social workers. The study has uncovered several findings in 18 interviews with participants who were disciplined by one state’s board of social work. The findings revealed that disciplinary action created substantial personal, psychological, and vocational impacts on the social workers involved. Symbolic interactionism theory provides a helpful framework for understanding the ways in which identity is changed when disciplinary action is introduced (J. Forte, personal communication, October 27, 2015). In his research, Forte (2003) found that symbolic interactionism can “contribute to our understanding of the social labeling process and the impact of labels on a member’s self-image and self-esteem” (p. 922). The theory pays close attention to groupings, including those who are “in” and those who are “out” based on certain actions or characteristics (J. Forte, personal communication, October 27, 2015). Disciplinary action changes the relationship of the social worker with the larger group of social workers, and the focus of the group also shifts to helping the social worker do the work needed to regain status (J. Forte, personal communication, October 27, 2015). Many participants seemed to indicate that the shift did not, and has not, happened for them. For example, the most harmful consequence noted by participants is the permanency of their names appearing on a public online database, regardless of the type of violation and length of time since it was resolved. In a data-driven world so highly attuned to gathering information before deciding (for example, Yelp, Angie’s List, RateMyProfessor.com), it would seem that participants’ concerns about the way their future employment could be affected by this “scarlet letter” are well founded. Several participants were so discouraged by the experience that their viewpoint of the field was described in absolute and permanent terms as forever changed. Differences appeared across the data regarding the attributable source of the violation and the ways in which other people played a role in their experience. Overall, these findings seem to corroborate those found in studies across disciplines regarding unethical and unprofessional behavior. The experience of disciplinary action in other fields has similarly deleterious effects regarding professional pride (Coy et al., 2016), external locus of control (Gutheil & Brodsky, 2008), psychological effects, and hypervigilance (Thomas, 2005). As in another study, participants in this study identified specific personal stressors—serious intimate partner problems, recent death of a loved one, depression—as precursors to professional violations (Garfinkel et al., 1997). Implications for Social Work Study results have implications for social work practice and social work education in several ways. First, the lack of a statute of limitations on the board action Web site may suggest that social workers do not feel an obligation to rehabilitate and reinstate the good name of a social worker who has taken the necessary steps to resolve a disciplinary action as Forte suggests (personal communication, October 27, 2015). Second, it is important to note the effect that continuing education has as both an ongoing license requirement and a sanction of disciplinary action. Only one participant’s disciplinary action was related to insufficient continuing education activities to maintain a license; therefore, all other participants completed some form of ethics training as well as dozens of additional hours in continuing education. It is possible that the influence of these activities is overstated in preventing unethical or unprofessional conduct (Mascari & Webber, 2006). Last, participants’ beliefs that their work responsibilities were not understood by the licensing board or that their violation was not more harmful may indicate that job descriptions and expectations of the profession are perceived to be incompatible, thereby creating dilemmas in decision-making priorities. Limitations This is the first social work study to give a voice to a group of licensees affected by disciplinary action. Although only one state’s information was included in this study, limiting the sample size of the potential interview participants, the study can be replicated for a larger area. The overall response rate to the survey was low, thus it is possible that the viewpoints of the sample are not representative of the population of people who were disciplined by a board of social work. The interview volunteers may have been more considerably affected by the experience of disciplinary action than those who did not participate. To focus on the impacts of the experience, I relied on self-report for information regarding the violation and subsequent disciplinary action and did not cross-check participants’ stories with publicly accessible information. Those participants who self-identified as black or African American were overrepresented in this sample, and both Asian Americans and Hispanic or Latino Americans were underrepresented. Further study into potential implicit bias may be important. Data on marital status, religious affiliation, and sexual orientation were not collected. Conclusion This study provides a new framework for identifying the subjective experience of disciplinary action and the profession’s responsibility to disciplined colleagues. Future studies may allow for comparisons across states regarding the ways in which disciplinary action affects social workers. In addition, studies focusing on the practice ramifications for this group of social workers who feel unsupported by their board and their profession are needed. A more in-depth look at the causes of distress associated with disciplinary action may also be helpful. Further research should evaluate the responsibility of the profession to restore and rehabilitate those who have been disciplined. Recent headlines are responsible for bringing news to the public about social workers’ unprofessional and unethical conduct, but few stories exist that explore the impact that situation had on these social workers. This work begins to fill that gap in the social work knowledge base. An exploration of these personal and professional impacts of disciplinary action sheds light into this dark and sometimes lonely corner of social work practice. Michelle Gricus, DSW, MSW, is assistant professor of social work, Department of Sociology and Social Work, Hood College, 401 Rosemont Avenue, Frederick, MD 21701; e-mail: [email protected] References Berkman , C. S. , Turner , S. G. , Cooper , M. , Polnerow , D. , & Swartz , M. ( 2000 ). Sexual contact with clients: Assessment of social workers’ attitudes and educational preparation . Social Work , 45 , 223 – 235 . doi:10.1093/sw/45.3.223 Google Scholar Crossref Search ADS PubMed Berliner , A. K. ( 1989 ). Misconduct in social work practice [Briefly Stated] . Social Work , 34 , 69 – 72 . doi:10.1093/sw/34.1.69 Boland-Prom , K. W. ( 2009 ). Results from a national study of social workers sanctioned by state licensing boards . Social Work , 54 , 351 – 360 . doi:10.1093/sw/54.4.351 Google Scholar Crossref Search ADS PubMed Council on Social Work Education . ( 2015 ). Educational policy and accreditation standards. Retrieved from http://www.cswe.org/File.aspx?id=13780 Coy , J. S. , Lambert , J. E. , & Miller , M. M. ( 2016 ). Stories of the accused: A phenomenological inquiry of MFTs and accusations of unprofessional conduct . Journal of Marital and Family Therapy , 42 ( 1 ), 139 – 152 . doi:10.1111/jmft.12109 Google Scholar Crossref Search ADS PubMed Creswell , J. W. ( 2014 ). Research design: Qualitative, quantitative, and mixed methods approaches ( 4th ed. ). Thousand Oaks, CA : Sage Publications . Daley , M. R. , & Doughty , M. O. ( 2006 ). Unethical social work: Comparing licensing and NASW perspectives . Arete , 30 ( 2 ), 35 – 50 . Epstein , R. S. , Simon , R. I. , & Kay , G. G. ( 1992 ). Assessing boundary violations in psychotherapy: Survey results with the exploitation index . Bulletin of the Menninger Clinic , 56 ( 2 ), 150 – 166 . Google Scholar PubMed Feldman-Summers , S. , & Jones , G. ( 1984 ). Psychological impacts of sexual contact between therapists or other health care practitioners and their clients . Journal of Consulting and Clinical Psychology , 52 , 1054 – 1061 . doi:10.1037/0022-006X.52.6.1054 Google Scholar Crossref Search ADS PubMed Forte , J. A. ( 2003 ). Applied symbolic interactionism: Meanings, memberships, and social work. In L. T. Reynolds & N. J. Herman-Kinney (Eds.), Handbook of symbolic interactionism (pp. 915 – 936 ). Portland : Ringgold . Garfinkel , P. E. , Bagby , M. E. , Waring , E. M. , & Dorian , B. ( 1997 ). Boundary violations and personality traits among psychiatrists . Canadian Journal of Psychiatry , 42 , 758 – 762 . doi:10.1177/070674379704200709 Google Scholar Crossref Search ADS PubMed Gutheil , T. G. , & Brodsky , A. ( 2008 ). Preventing boundary violations in clinical practice . New York : Guilford Press . Harper , M. , & Cole , P. ( 2012 ). Member checking: Can benefits be gained similar to group therapy? Qualitative Report , 17 , 510 – 517 . Retrieved from http://nsuworks.nova.edu/tqr/vol17/iss2/1 Holland , T. P. , & Kilpatrick , A. C. ( 1991 ). Ethical issues in social work: Toward a grounded theory of professional ethics . Social Work , 36 , 138 – 144 . doi:10.1093/sw/36.2.138 Jayaratne , S. , Croxton , T. , & Mattison , D. ( 1997 ). Social work professional standards: An exploratory study . Social Work , 42 , 187 – 199 . doi:10.1093/sw/42.2.187 Google Scholar Crossref Search ADS Lamb , D. H. , & Catanzaro , S. J. ( 1998 ). Sexual and nonsexual boundary violations involving psychologists, clients, supervisees, and students: Implications for professional practice . Professional Psychology: Research and Practice , 29 , 498 – 503 . doi:10.1037/0735-7028.29.5.498 Google Scholar Crossref Search ADS Luepker , E. T. ( 1999 ). Effects of practitioners’ sexual misconduct: A follow-up study . Journal of the American Academy of Psychiatry and the Law , 27 ( 1 ), 51 – 63 . Google Scholar PubMed MacDonald , K. , Sciolla , A. F. , Folsom , D. , Bazzo , D. , Searles , C. , Moutier , C. , et al. ( 2015 ). Individual risk factors for physician boundary violations: The role of attachment style, childhood trauma and maladaptive beliefs . General Hospital Psychiatry , 37 , 489 – 496 . doi:10.1016/j.genhosppsych.2015.06.002 Google Scholar Crossref Search ADS PubMed Mascari , J. B. , & Webber , J. M. ( 2006 ). Salting the slippery slope: What licensing violations tell us about preventing dangerous ethical situations. In G. R. Walz , J. Bleuer , & R. K. Yep (Eds.), VISTAS: Compelling perspectives on counseling, 2006 (pp. 165 – 168 ). Alexandria, VA : American Counseling Association . McAuliffe , D. ( 2005 ). “I’m still standing”: Impacts and consequences of ethical dilemmas for social workers in direct practice . Journal of Social Work Values and Ethics , 2 ( 1 ), 1–10. McCann , C. W. , & Cutler , J. P. ( 1979 ). Ethics and the alleged unethical . Social Work , 24 , 5 – 8 . doi:10.1093/sw/24.1.5 Google Scholar PubMed National Association of Social Workers . ( 2017 ). Code of ethics of the National Association of Social Workers. Retrieved from https://www.socialworkers.org/pubs/code/code.asp Neukrug , E. , Milliken , T. , & Walden , S. ( 2001 ). Ethical complaints made against credentialed counselors: An updated survey of state licensing boards . Counselor Education and Supervision , 41 ( 1 ), 57 – 70 . doi:10.1002/j.1556-6978.2001.tb01268.x Google Scholar Crossref Search ADS Pope , K. S. , Sonne , J. L. , & Holroyd , J. ( 1993 ). Sexual feelings in psychotherapy: Explorations for therapists and therapists-in-training . Washington, DC : American Psychological Association . Reamer , F. G. ( 1995 ). Malpractice claims against social workers: First facts . Social Work , 40 , 595 – 601 . doi:10.1093/sw/40.5.595 Reamer , F. G. ( 2003 ). Boundary issues in social work: Managing dual relationships . Social Work , 48 , 121 – 133 . doi:10.1093/sw/48.1.121 Google Scholar Crossref Search ADS PubMed Saldaña , J. ( 2015 ). The coding manual for qualitative researchers . Thousand Oaks, CA : Sage Publications . Somer , E. , & Saadon , M. ( 1999 ). Therapist-client sex: Clients’ retrospective reports . Professional Psychology: Research and Practice , 30 , 504 – 509 . doi:10.1037/0735-7028.30.5.504 Google Scholar Crossref Search ADS Strom-Gottfried , K. ( 2000 ). Ensuring ethical practice: An examination of NASW code violations, 1986–97 . Social Work , 45 , 251 – 261 . doi:10.1093/sw/45.3.251 Google Scholar Crossref Search ADS PubMed Thomas , J. T. ( 2005 ). Licensing board complaints: Minimizing the impact on the psychologist’s defense and clinical practice . Professional Psychology , 36 , 426 – 433 . doi:10.1037/0735-7028.36.4.426 Google Scholar Crossref Search ADS © 2018 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)

McNeil, Robyn, J;McCarthy,, Maria;Dunt,, David;Thompson,, Kate;Kosola,, Silja;Orme,, Lisa;Drew,, Sarah;Sawyer,, Susan

doi: 10.1093/swr/svy027pmid: N/A

Abstract This study examined the financial impact of cancer and the use of income support in adolescents and young adults (AYAs) with cancer and their parent caregivers. As part of a national Australian study exploring the psychosocial impacts of cancer, 196 AYAs ages 15 to 25 years, six to 24 months from diagnosis, and 204 parent caregivers from 18 cancer sites were surveyed. Logistic regression and chi-square analyses were conducted to assess the influence of clinical and sociodemographic variables on financial status. Qualitative responses were coded, and key themes were identified using thematic analysis. The findings indicate that more than half of AYAs and parents reported financial issues as a consequence of AYA cancer. Financial issues resulted from direct medical costs, associated costs from treatment, and indirect costs from loss of income. AYAs and parents reported that it was important for them to receive income support, both during and after cancer treatment. However, large proportions of those who reported needing income support had difficulty accessing it. AYAs and their families are substantially financially disadvantaged by cancer, many for a prolonged time. Patient- and family-centered assessments and interventions are required to reduce the financial burden of AYA cancer. The diagnosis of cancer during adolescence and young adulthood signals the potential for major disruption of normal developmental trajectories (Grinyer, 2007; Sawyer et al., 2012; Zebrack, 2011). At the time when most adolescents and young adults (AYAs) are engaged in education and training, exploring and establishing career choices, and have yet to gain financial independence, cancer treatment has the potential to profoundly affect education and employment pathways (Grinyer, 2007; Thompson, Palmer, & Dyson, 2009; Zebrack, 2011) with ramifications for health (Patton et al., 2016) and emotional and economic well-being (D’Agostino, Penney, & Zebrack, 2011; Hall et al., 2012; Yabroff et al., 2016). The extent of the financial burden from cancer is increasingly recognized in adults, with three broad categories of costs: (1) the “financial toxicity” from the direct costs of cancer and its treatment (Zafar et al., 2013), even for those with medical insurance (Longo, Fitch, Deber, & Williams, 2006; Markman & Luce, 2010; Zafar et al., 2013); (2) treatment-related out-of-pocket expenses (for example: transport to attend treatment, food, accommodation, vehicle parking); and (3) indirect costs including loss of income from being unable to work (Aaronson et al., 2014; Guy et al., 2013; Kim, 2007; Longo et al., 2006). The financial burden associated with inability or partial return to work is not simply a feature of the acute treatment phase, but has been shown to last up to 10 years after diagnosis (Bloom, 2002; Mehnert, 2011; Paalman et al., 2016). In contrast to the financial impacts on adult cancer patients, the pediatric literature indicates that it is families of children with cancer who experience significant financial burden during and after treatment, and includes direct and indirect costs. Families of children living in rural and regional areas have been found to incur higher overall out-of-pocket costs for cancer treatment, particularly around transport and accommodation (Cohn, Goodenough, Foreman, & Suneson, 2003). In Australia, where around a third of new cancers arise in patients from rural, regional, and remote regions, the distances patients must travel to access care at cancer centers makes this issue particularly pertinent (Australian Insititute of Health and Welfare & Australian Association of Cancer Registries, 2007). Direct and indirect costs are compounded by the loss of family income when a parent gives up work to care for her or his child with cancer (Eiser & Upton, 2007; Heath, Lintuuran, Rigguto, Tikotlian, & McCarthy, 2006; Wakefield, McLoone, Evans, Ellis, & Cohn, 2014). These impacts are likely to be compounded for families with fewer financial reserves to draw on. Few studies have examined the financial experiences and impacts of AYAs with cancer (D’Agostino et al., 2011). Studies from the United States have focused on medical costs and insurance issues related to the U.S. health care system (Freyer & Barr, 2007; Zebrack et al., 2014) for 15- to 39-year-old individuals. The breadth of this age span comprises AYAs with a diversity of education and employment needs, including those who are just embarking on their careers with few individual financial reserves as well as those with established careers and potentially greater financial security (D’Agostino et al., 2011; Geue et al., 2014; Hall et al., 2012). In addition, the U.S. health care system differs significantly from Australia’s; Australia has a universal health care system that provides comprehensive cancer care in public health services, which theoretically should minimize medical costs associated with cancer treatment. This system includes a parallel model of health care involving patients purchasing private health care for services. Until recently, it has been unclear how the use of the private model or a combination of public/private services affects financial outcomes from cancer treatment, although some research indicates that direct medical costs can be significant for adult cancer patients using private health care (Cohn et al., 2003; Gordon et al., 2017). The United Kingdom (UK) and Australia have developed specialist AYA cancer services for young people between ages 13 (UK) and 15 years (Australia) to the mid-twenties (Osborn, Little, Bowering, & Orme, 2013; Teenage Cancer Trust, 2015), a period of formative transitions within education, from education to employment, and from financial dependence on family toward relative independence. Disruption during these critical years can be appreciated to have a different salience for AYAs than for older adults with more established employment track records and greater financial assets. This narrower age span also includes the years in which many parents continue to support AYAs physically, emotionally, and financially (D’Agostino et al., 2011; Wakefield, McLoone, Butow, Lenthen, & Cohn, 2013). This raises questions about to what extent families might be financially affected by the experience of cancer in their AYA children. We hypothesized that the financial impact of cancer on families would be similar to the impact in pediatric studies and greater for those in regional, rural, and remote areas as well as in families of lower income or with less financial reserves (Heath et al., 2006). This study aimed to examine the financial impact of cancer for AYAs ages 15 to 25 years and their parent caregivers in Australia, including whether clinical and sociodemographic factors identified in extant literature were associated with these outcomes. A second aim was to examine AYA and parent caregiver experiences in terms of (a) their need for income support and (b) the challenges associated with accessing this support. Method The Youth Friendly Cancer Care project is a four-stage sequential strategy of inquiry undertaken to determine the degree to which Australian cancer services are meeting the needs of AYAs and their parents. We used data from stage three, a nationally representative survey of AYAs and their parents, of which detailed methods have been reported (Sawyer et al., 2016). Australia has both universal health care and a social support system, both of which would be hypothesized to buffer families from the financial costs of cancer. The context of the Australian health care system and its approach to income support is briefly summarized in the appendix. Figure 1: View largeDownload slide Degree of Financial Difficulties for Adolescents and Young Adults (AYAs) and Parents Figure 1: View largeDownload slide Degree of Financial Difficulties for Adolescents and Young Adults (AYAs) and Parents Participants Eligible AYAs were (a) 15 to 25 years old with a cancer diagnosis (including relapsed or second cancers) between September 2010 and December 2012 and (b) six to 24 months from diagnosis. Exclusion criteria were (a) inability to complete the survey due to poor literacy in English, cognitive deficit, or being too unwell; and (b) stage 1 and 2 melanomas (as these involve brief surgical treatment only). Parent caregivers were identified if they were nominated by the AYA or were listed as the nominated primary parent in hospital databases. Procedure Twenty-one hospitals providing AYA cancer care across Australia were approached, of which 17 (12 adult and five pediatric hospitals) agreed to participate, together with one charitable AYA cancer organization, CanTeen. Ethics and governance approvals were obtained from each site. Potentially eligible participants were identified by local staff using clinical databases and mailed a survey package. Parent contact details were not available at seven adult hospitals, or from the CanTeen database. In these instances, packages were mailed to the AYA with the request to forward the survey to a nominated parent. Response rates were conservatively estimated to be 25.7% and 27.3% for AYAs and parents, respectively (Sawyer et al., 2016). Measures The AYA survey comprised a 70-item self-administered questionnaire that included a combination of validated psychosocial measures and study-specific items that were developed from the literature and our earlier qualitative analysis of AYA and parent interviews (Sawyer et al., 2016). Survey items relating to financial burden examined two key areas: (1) the financial impact of cancer and (2) use of income support. The surveys included space for open-ended commentary. These questions were replicated in the self-administered parent surveys to similarly explore the financial impact on parents. Financial Burden Two single-item questions from the Psychosocial Assessment Tool (Pai et al., 2008) were used to assess the financial costs experienced by participants: “Did you experience any financial difficulties as a consequence of your cancer diagnosis and treatment?” and “In what areas have you experienced financial difficulties while you have been receiving treatment?” For the former, the response options were no; yes, some financial problems; yes, many financial problems; yes, it’s hard to meet our basic needs; and I don’t know. For the latter, response options were phone/utility bills, rent/mortgage, buying food, vehicle related (upkeep/petrol/insurance), medical expenses, parking at the main cancer treatment center, having to pay for television while an inpatient at the main cancer treatment center, and other. Income Support Two single-item questions with Likert response scales were used to explore the need for, and challenges associated with, accessing income support both during and after treatment: “Was it important for you to receive income support from the government?” and “Did you experience difficulties/challenges getting access to income support from the government?” Response options for the former were yes, very important; yes, somewhat important; no, not important; and not applicable. For the latter, response options were yes, very important; yes, somewhat important; no, not important; and not applicable. All response options were dichotomized to yes and no for analysis. Not applicable options were omitted. Educational and Work Impact AYAs were asked a single-item question that was used as a proxy indicator of current work and financial capacity: “At the current time, have you been able to get back on track with work plans and activities?” Response options were yes, no, to some extent, and not applicable. Demographic and Clinical Variables Several sociodemographic and clinical variables were hypothesized from the adult and pediatric literature to be associated with financial difficulties and the need for income support. For AYAs these variables include older age at diagnosis, being unemployed, living outside the family home, and living in a regional or rural area. For parents these variables include a younger AYA age at diagnosis, AYA living in the family home, living in a regional or rural area, a blood cancer diagnosis, and length of stay in hospital. In the absence of family income assessment, the research team used parent education attainment as a proxy indicator of income. Statistical Analysis Quantitative analyses were conducted using Stata (Version 13) (StataCorp, 2013). Demographic and clinical variables, financial burden, and income support for AYAs and parents were characterized using descriptive analyses. Several sociodemographic and clinical variables were dichotomized to simplify data analysis. Differences between groups were analyzed by chi-square analyses; significance level (α) was less than or equal to 5% (0.05). Logistic regression analyses were conducted to test associations between sociodemographic and clinical characteristics. Results are reported in odds ratios (OR) with 95% confidence intervals (CI). Multicollinearity between independent variables in the regression models was assessed using variance inflation factor and all found to be acceptable at less than 1.8 (O’Brien, 2007). Qualitative responses to open-ended questions were coded using open and axial coding to summarize the text (Rice & Ezzy, 1999). Key themes were identified using inductive thematic analysis (Braun & Clarke, 2006). Results The study sample consisted of 196 AYAs and 204 parents. The mean age at diagnosis was 19.9 years and the mean time since diagnosis was 19 months (SD = 8.18). Fifty percent of AYAs were studying part- or full-time, 47% were working part- or full-time, and 11% were unemployed at the time of the survey. AYA demographic and clinical details are shown in Table 1. Parent participants were predominantly mothers (89%). At the time of the survey, 70% of parents were working either part- or full-time. Demographic details of parent carers are shown in Table 2. Table 1: Adolescents and Young Adults Sociodemographic and Clinical Characteristics (N = 196) Characteristic M (SD), Range n (%) Age (years) at survey  Dichotomizeda 21.6 (3.1), 15–27  15–24 150 (79)  25+ 41 (21) Age (years) at diagnosis (n = 194) 19.9 (3.2), 15–26 Age (years) group at diagnosis (n = 194)  15–19 87 (45)  20–25 107 (55) Time (months) since diagnosis (n = 183) 19.06 (8.18), 6–33 Treatment setting  Adult 168 (86)  Pediatric 27 (14) Cancer type (n = 193)  Malignant hematological cancer 60 (31)  Hodgkin’s lymphoma 48 (25)  Sarcoma 29 (15)  Brain tumor 17 (9)  Germ cell tumor 14 (7)  Melanoma 7 (4)  Thyroid tumor 5 (3)  Other 13 (7) Gender  Male 99 (51)  Female 97 (49) Employment/education status at surveyb Full-time Part-time  High school student 29 6 (18%)  Tertiary student 47 15 (32%)  Working 48 44 (47%)  Unemployed 21 (11%)  Homemaker/family caregiver 2 (1%)  Other 10 (5%) Geographic location (n = 193)  Major metropolitan city 123 (64)  Regional city 44 (23)  Rural area 26 (13) Resides with (n = 230)b  Parents 141 (72)  Partner 23 (12)  Other family 20 (10)  Boyfriend/girlfriend 18 (9)  Friends 15 (8)  Other 13 (7) Characteristic M (SD), Range n (%) Age (years) at survey  Dichotomizeda 21.6 (3.1), 15–27  15–24 150 (79)  25+ 41 (21) Age (years) at diagnosis (n = 194) 19.9 (3.2), 15–26 Age (years) group at diagnosis (n = 194)  15–19 87 (45)  20–25 107 (55) Time (months) since diagnosis (n = 183) 19.06 (8.18), 6–33 Treatment setting  Adult 168 (86)  Pediatric 27 (14) Cancer type (n = 193)  Malignant hematological cancer 60 (31)  Hodgkin’s lymphoma 48 (25)  Sarcoma 29 (15)  Brain tumor 17 (9)  Germ cell tumor 14 (7)  Melanoma 7 (4)  Thyroid tumor 5 (3)  Other 13 (7) Gender  Male 99 (51)  Female 97 (49) Employment/education status at surveyb Full-time Part-time  High school student 29 6 (18%)  Tertiary student 47 15 (32%)  Working 48 44 (47%)  Unemployed 21 (11%)  Homemaker/family caregiver 2 (1%)  Other 10 (5%) Geographic location (n = 193)  Major metropolitan city 123 (64)  Regional city 44 (23)  Rural area 26 (13) Resides with (n = 230)b  Parents 141 (72)  Partner 23 (12)  Other family 20 (10)  Boyfriend/girlfriend 18 (9)  Friends 15 (8)  Other 13 (7) Note: Percentages use total number of responses as denominator; otherwise N = 196 is used as the denominator. aDichotomized for comparison to the National Housing Supply Council data (2013). bTotals are for number of responses due to a “tick all that apply” question type. Table 1: Adolescents and Young Adults Sociodemographic and Clinical Characteristics (N = 196) Characteristic M (SD), Range n (%) Age (years) at survey  Dichotomizeda 21.6 (3.1), 15–27  15–24 150 (79)  25+ 41 (21) Age (years) at diagnosis (n = 194) 19.9 (3.2), 15–26 Age (years) group at diagnosis (n = 194)  15–19 87 (45)  20–25 107 (55) Time (months) since diagnosis (n = 183) 19.06 (8.18), 6–33 Treatment setting  Adult 168 (86)  Pediatric 27 (14) Cancer type (n = 193)  Malignant hematological cancer 60 (31)  Hodgkin’s lymphoma 48 (25)  Sarcoma 29 (15)  Brain tumor 17 (9)  Germ cell tumor 14 (7)  Melanoma 7 (4)  Thyroid tumor 5 (3)  Other 13 (7) Gender  Male 99 (51)  Female 97 (49) Employment/education status at surveyb Full-time Part-time  High school student 29 6 (18%)  Tertiary student 47 15 (32%)  Working 48 44 (47%)  Unemployed 21 (11%)  Homemaker/family caregiver 2 (1%)  Other 10 (5%) Geographic location (n = 193)  Major metropolitan city 123 (64)  Regional city 44 (23)  Rural area 26 (13) Resides with (n = 230)b  Parents 141 (72)  Partner 23 (12)  Other family 20 (10)  Boyfriend/girlfriend 18 (9)  Friends 15 (8)  Other 13 (7) Characteristic M (SD), Range n (%) Age (years) at survey  Dichotomizeda 21.6 (3.1), 15–27  15–24 150 (79)  25+ 41 (21) Age (years) at diagnosis (n = 194) 19.9 (3.2), 15–26 Age (years) group at diagnosis (n = 194)  15–19 87 (45)  20–25 107 (55) Time (months) since diagnosis (n = 183) 19.06 (8.18), 6–33 Treatment setting  Adult 168 (86)  Pediatric 27 (14) Cancer type (n = 193)  Malignant hematological cancer 60 (31)  Hodgkin’s lymphoma 48 (25)  Sarcoma 29 (15)  Brain tumor 17 (9)  Germ cell tumor 14 (7)  Melanoma 7 (4)  Thyroid tumor 5 (3)  Other 13 (7) Gender  Male 99 (51)  Female 97 (49) Employment/education status at surveyb Full-time Part-time  High school student 29 6 (18%)  Tertiary student 47 15 (32%)  Working 48 44 (47%)  Unemployed 21 (11%)  Homemaker/family caregiver 2 (1%)  Other 10 (5%) Geographic location (n = 193)  Major metropolitan city 123 (64)  Regional city 44 (23)  Rural area 26 (13) Resides with (n = 230)b  Parents 141 (72)  Partner 23 (12)  Other family 20 (10)  Boyfriend/girlfriend 18 (9)  Friends 15 (8)  Other 13 (7) Note: Percentages use total number of responses as denominator; otherwise N = 196 is used as the denominator. aDichotomized for comparison to the National Housing Supply Council data (2013). bTotals are for number of responses due to a “tick all that apply” question type. Table 2: Parent Sociodemographic Characteristics (N = 204) Characteristic n (%) of Parents Relationship to AYA with cancer (n = 203)  Mother 180 (89)  Father 19 (9)  Stepmother 1 (0.5)  Stepfather 2 (1)  Female guardian 1 (0.5) Country of birth (n = 200)  Australia 138 (69)  Other 62 (31) Education level (n = 201)  Left school before completing year 10 14 (7)  Year 10 or equivalent 31 (15)  Year 12 or equivalent 35 (17)  Certificate or diploma 59 (29)  Bachelor or higher degree 62 (31) Employment/education status at survey (%) Full-time Part-time  Working 78 65 (70)  Unemployed 4 (2)  Homemaker/family caregiver 37 7 (22)  Other 9 4 (6) Geographic location (n = 200)  Major metropolitan city 119 (60)  Regional city 49 (25)  Rural area 28 (14)  Remote or very remote area 4 (2) Number of children in family (n = 198)  M (SD), range 2.76 (1.13), 1–7 Relationship status (n = 200)  No partner 23 (12)  De facto partner 12 (6)  Married (first marriage) 124 (62)  Separated 7 (4)  Divorced 16 (8%)  Remarried 18 (9) Who do you live with most of the time? (n = 294)a  Child/children 154 (75)  Extended family 2 (1)  Partner/de facto/spouse 127 (62)  Alone 7 (3)  Other 4 (2) Number of children living at home (n = 129)  M (SD), range 2.12 (1.00), 1–7 Characteristic n (%) of Parents Relationship to AYA with cancer (n = 203)  Mother 180 (89)  Father 19 (9)  Stepmother 1 (0.5)  Stepfather 2 (1)  Female guardian 1 (0.5) Country of birth (n = 200)  Australia 138 (69)  Other 62 (31) Education level (n = 201)  Left school before completing year 10 14 (7)  Year 10 or equivalent 31 (15)  Year 12 or equivalent 35 (17)  Certificate or diploma 59 (29)  Bachelor or higher degree 62 (31) Employment/education status at survey (%) Full-time Part-time  Working 78 65 (70)  Unemployed 4 (2)  Homemaker/family caregiver 37 7 (22)  Other 9 4 (6) Geographic location (n = 200)  Major metropolitan city 119 (60)  Regional city 49 (25)  Rural area 28 (14)  Remote or very remote area 4 (2) Number of children in family (n = 198)  M (SD), range 2.76 (1.13), 1–7 Relationship status (n = 200)  No partner 23 (12)  De facto partner 12 (6)  Married (first marriage) 124 (62)  Separated 7 (4)  Divorced 16 (8%)  Remarried 18 (9) Who do you live with most of the time? (n = 294)a  Child/children 154 (75)  Extended family 2 (1)  Partner/de facto/spouse 127 (62)  Alone 7 (3)  Other 4 (2) Number of children living at home (n = 129)  M (SD), range 2.12 (1.00), 1–7 Notes: AYA = adolescents and young adults. Percentages use total number of responses as denominator; otherwise n = 204 is used as the denominator. aTotals are for number of responses due to a “tick all that apply” question type. Table 2: Parent Sociodemographic Characteristics (N = 204) Characteristic n (%) of Parents Relationship to AYA with cancer (n = 203)  Mother 180 (89)  Father 19 (9)  Stepmother 1 (0.5)  Stepfather 2 (1)  Female guardian 1 (0.5) Country of birth (n = 200)  Australia 138 (69)  Other 62 (31) Education level (n = 201)  Left school before completing year 10 14 (7)  Year 10 or equivalent 31 (15)  Year 12 or equivalent 35 (17)  Certificate or diploma 59 (29)  Bachelor or higher degree 62 (31) Employment/education status at survey (%) Full-time Part-time  Working 78 65 (70)  Unemployed 4 (2)  Homemaker/family caregiver 37 7 (22)  Other 9 4 (6) Geographic location (n = 200)  Major metropolitan city 119 (60)  Regional city 49 (25)  Rural area 28 (14)  Remote or very remote area 4 (2) Number of children in family (n = 198)  M (SD), range 2.76 (1.13), 1–7 Relationship status (n = 200)  No partner 23 (12)  De facto partner 12 (6)  Married (first marriage) 124 (62)  Separated 7 (4)  Divorced 16 (8%)  Remarried 18 (9) Who do you live with most of the time? (n = 294)a  Child/children 154 (75)  Extended family 2 (1)  Partner/de facto/spouse 127 (62)  Alone 7 (3)  Other 4 (2) Number of children living at home (n = 129)  M (SD), range 2.12 (1.00), 1–7 Characteristic n (%) of Parents Relationship to AYA with cancer (n = 203)  Mother 180 (89)  Father 19 (9)  Stepmother 1 (0.5)  Stepfather 2 (1)  Female guardian 1 (0.5) Country of birth (n = 200)  Australia 138 (69)  Other 62 (31) Education level (n = 201)  Left school before completing year 10 14 (7)  Year 10 or equivalent 31 (15)  Year 12 or equivalent 35 (17)  Certificate or diploma 59 (29)  Bachelor or higher degree 62 (31) Employment/education status at survey (%) Full-time Part-time  Working 78 65 (70)  Unemployed 4 (2)  Homemaker/family caregiver 37 7 (22)  Other 9 4 (6) Geographic location (n = 200)  Major metropolitan city 119 (60)  Regional city 49 (25)  Rural area 28 (14)  Remote or very remote area 4 (2) Number of children in family (n = 198)  M (SD), range 2.76 (1.13), 1–7 Relationship status (n = 200)  No partner 23 (12)  De facto partner 12 (6)  Married (first marriage) 124 (62)  Separated 7 (4)  Divorced 16 (8%)  Remarried 18 (9) Who do you live with most of the time? (n = 294)a  Child/children 154 (75)  Extended family 2 (1)  Partner/de facto/spouse 127 (62)  Alone 7 (3)  Other 4 (2) Number of children living at home (n = 129)  M (SD), range 2.12 (1.00), 1–7 Notes: AYA = adolescents and young adults. Percentages use total number of responses as denominator; otherwise n = 204 is used as the denominator. aTotals are for number of responses due to a “tick all that apply” question type. Financial Burden Forty-five percent of AYAs (n = 191) reported they had been able to “get back on track” with work plans, 30% were back on track to some extent, whereas 15% reported they had not been able to get back on track. Comments indicated that many were unable to function at their previous capacity due to fatigue, frequency of medical appointments, or due to having changed career paths. More than half of AYAs (57%) reported financial issues as a consequence of their cancer diagnosis and treatment (see Figure 1). Of those who reported financial issues, almost two-thirds (63%) reported that they lived with their parents. More 20- to 25-year-olds reported financial issues (64%) than 15- to 19-year-olds (47%), and an older age (20 to 25 years) at diagnosis was associated with increased likelihood of financial issues (OR = 1.98, CI [1.06, 3.67], p = .031). There was a reduced likelihood of having financial issues if the AYA was living in the family home (OR = 0.5, CI [0.25, 0.98], p = .044) in regression analyses (see Table 3). Table 3: Logistic Regression Analysis of Sociodemographic and Clinical Setting and Treatment Variables on Financial Impact of Cancer on Adolescents and Young Adults and Parents Variable Adolescents and Young Adults Parents OR 95% CI p OR 95% CI p Sociodemographic variables  Gender   Male 0.61 [0.33, 1.11] .11 1.24 [0.61, 2.55] .56   (female)  Geography   Regional/rural 1.06 [0.57, 2.00] .85 2.03 [1.13, 3.71] .02   (Metropolitan)  Living arrangement   Family home 0.5 [0.25, 0.98] .04 2.90 [1.0, 8.5] .05   (Living outside family home)  Employment status   Unemployed 2.09 [0.87, 5.06] .10   (Employed)  Education level Completed year 10 or equivalent 0.78 [0.11, 2.08] .32 Completed year 12 or equivalent 0.34 [0.08, 1.45] .14 Certificate or diploma 0.53 [0.13, 2.14] .37 Bachelor or higher degree (<year 10) 0.31 [0.78, 1.24] .10 Clinical setting and treatment variables  Age at diagnosis Age at Diagnosisa   20–25 years 1.98 [1.06, 3.67] .03 15–19 years 2.46 [1.18, 5.12] .02   (15–19 years) (20–25 years)  Length of hospital stay   Long stay 1.32 [0.71, 2.43] .38 2.04 [1.12, 3.7] .02   (Short stay)  On/off treatment   On treatment 1.05 [0.48, 2.31] .91 1.58 [0.8, 3.11] .19   (Off treatment)  Treatment setting   Adult 0.89 [0.34, 2.3] .81 0.63 [0.29, 1.4] .24   (Pediatric)  Cancer type   Blood cancer 0.99 [0.54, 1.82] .98 1.96 [0.94, 4.1] .07   (Other) Variable Adolescents and Young Adults Parents OR 95% CI p OR 95% CI p Sociodemographic variables  Gender   Male 0.61 [0.33, 1.11] .11 1.24 [0.61, 2.55] .56   (female)  Geography   Regional/rural 1.06 [0.57, 2.00] .85 2.03 [1.13, 3.71] .02   (Metropolitan)  Living arrangement   Family home 0.5 [0.25, 0.98] .04 2.90 [1.0, 8.5] .05   (Living outside family home)  Employment status   Unemployed 2.09 [0.87, 5.06] .10   (Employed)  Education level Completed year 10 or equivalent 0.78 [0.11, 2.08] .32 Completed year 12 or equivalent 0.34 [0.08, 1.45] .14 Certificate or diploma 0.53 [0.13, 2.14] .37 Bachelor or higher degree (<year 10) 0.31 [0.78, 1.24] .10 Clinical setting and treatment variables  Age at diagnosis Age at Diagnosisa   20–25 years 1.98 [1.06, 3.67] .03 15–19 years 2.46 [1.18, 5.12] .02   (15–19 years) (20–25 years)  Length of hospital stay   Long stay 1.32 [0.71, 2.43] .38 2.04 [1.12, 3.7] .02   (Short stay)  On/off treatment   On treatment 1.05 [0.48, 2.31] .91 1.58 [0.8, 3.11] .19   (Off treatment)  Treatment setting   Adult 0.89 [0.34, 2.3] .81 0.63 [0.29, 1.4] .24   (Pediatric)  Cancer type   Blood cancer 0.99 [0.54, 1.82] .98 1.96 [0.94, 4.1] .07   (Other) Notes: Reference categories are in parentheses. Boldface used to highlight a p value of less than .05. OR = odds ratio; CI = confidence interval. aParent age at diagnosis has older age group as reference. Table 3: Logistic Regression Analysis of Sociodemographic and Clinical Setting and Treatment Variables on Financial Impact of Cancer on Adolescents and Young Adults and Parents Variable Adolescents and Young Adults Parents OR 95% CI p OR 95% CI p Sociodemographic variables  Gender   Male 0.61 [0.33, 1.11] .11 1.24 [0.61, 2.55] .56   (female)  Geography   Regional/rural 1.06 [0.57, 2.00] .85 2.03 [1.13, 3.71] .02   (Metropolitan)  Living arrangement   Family home 0.5 [0.25, 0.98] .04 2.90 [1.0, 8.5] .05   (Living outside family home)  Employment status   Unemployed 2.09 [0.87, 5.06] .10   (Employed)  Education level Completed year 10 or equivalent 0.78 [0.11, 2.08] .32 Completed year 12 or equivalent 0.34 [0.08, 1.45] .14 Certificate or diploma 0.53 [0.13, 2.14] .37 Bachelor or higher degree (<year 10) 0.31 [0.78, 1.24] .10 Clinical setting and treatment variables  Age at diagnosis Age at Diagnosisa   20–25 years 1.98 [1.06, 3.67] .03 15–19 years 2.46 [1.18, 5.12] .02   (15–19 years) (20–25 years)  Length of hospital stay   Long stay 1.32 [0.71, 2.43] .38 2.04 [1.12, 3.7] .02   (Short stay)  On/off treatment   On treatment 1.05 [0.48, 2.31] .91 1.58 [0.8, 3.11] .19   (Off treatment)  Treatment setting   Adult 0.89 [0.34, 2.3] .81 0.63 [0.29, 1.4] .24   (Pediatric)  Cancer type   Blood cancer 0.99 [0.54, 1.82] .98 1.96 [0.94, 4.1] .07   (Other) Variable Adolescents and Young Adults Parents OR 95% CI p OR 95% CI p Sociodemographic variables  Gender   Male 0.61 [0.33, 1.11] .11 1.24 [0.61, 2.55] .56   (female)  Geography   Regional/rural 1.06 [0.57, 2.00] .85 2.03 [1.13, 3.71] .02   (Metropolitan)  Living arrangement   Family home 0.5 [0.25, 0.98] .04 2.90 [1.0, 8.5] .05   (Living outside family home)  Employment status   Unemployed 2.09 [0.87, 5.06] .10   (Employed)  Education level Completed year 10 or equivalent 0.78 [0.11, 2.08] .32 Completed year 12 or equivalent 0.34 [0.08, 1.45] .14 Certificate or diploma 0.53 [0.13, 2.14] .37 Bachelor or higher degree (<year 10) 0.31 [0.78, 1.24] .10 Clinical setting and treatment variables  Age at diagnosis Age at Diagnosisa   20–25 years 1.98 [1.06, 3.67] .03 15–19 years 2.46 [1.18, 5.12] .02   (15–19 years) (20–25 years)  Length of hospital stay   Long stay 1.32 [0.71, 2.43] .38 2.04 [1.12, 3.7] .02   (Short stay)  On/off treatment   On treatment 1.05 [0.48, 2.31] .91 1.58 [0.8, 3.11] .19   (Off treatment)  Treatment setting   Adult 0.89 [0.34, 2.3] .81 0.63 [0.29, 1.4] .24   (Pediatric)  Cancer type   Blood cancer 0.99 [0.54, 1.82] .98 1.96 [0.94, 4.1] .07   (Other) Notes: Reference categories are in parentheses. Boldface used to highlight a p value of less than .05. OR = odds ratio; CI = confidence interval. aParent age at diagnosis has older age group as reference. Figure 2: View largeDownload slide Areas of Financial Burden for Adolescents and Young Adults (AYAs) and Parents Figure 2: View largeDownload slide Areas of Financial Burden for Adolescents and Young Adults (AYAs) and Parents Financial challenges for AYAs related to both ongoing costs of living and additional costs incurred from cancer treatment including vehicle-related costs (49%), medically related costs (44%), utilities (40%), and vehicle parking at the treatment center (38%) (see Figure 2). Their comments indicated that financial burden was also from loss of income: “I was unable to work for one whole year. I went back to work only because I needed the money.” Almost two-thirds (62%) of parents reported financial issues as a consequence of their child’s cancer (see Figure 1). Logistic regressions indicated that parents who lived in a regional or rural area, parents of 15- to 19-year-olds, and parents of AYAs whose inpatient stay was longer than one month were more likely to experience financial issues as a result of their child’s cancer. Parents whose child was living in the family home were more likely to experience financial difficulties, although this finding was not statistically significant (OR = 2.90, CI [1.0, 8.5], p = .050) (see Table 3). For parents, areas of financial burden were similar to those described by AYAs: Vehicle costs (40%), utilities (39%), medical expenses (29%), and mortgage repayments (27%) were each reported by substantial numbers of parents (see Figure 2). Vehicle parking at the treatment center was the leading area of financial burden for parents (58%). Many comments related to the cost of transportation to the cancer center. One parent said, “The main cost was the toll road. We clocked up $1,800 for the year. Normal year $50 to $60.” Other comments acknowledged the impact of financially supporting their child who was unable to do so himself or herself, such as, “Supporting him financially while he was off work for six months. He was living at home, not paying his usual board, and I was buying a lot of healthy food for him.” However, the majority of parent commentary related to the impact of the direct loss of parent income: “High medical expenses and my loss of income has placed tremendous strain on our family’s financial resources.” This was particularly salient when parents were self-employed: “I am self-employed and was unable to work for a substantial amount of time so I could be with my son.” Analysis of comments suggested that many parents who were more able to manage financially had flexibility in their workplace (for example, possibility of extended leave), supportive family or friends, and the safety net of accumulated savings. One parent stated, I had a lot of support from my parents-in-law and my boss. My father-in-law helped with transport to and from [the] hospital on many occasions. My boss gave me flexibility with my work hours, and I was also able to work from the hospital. Income Support Sixty percent of AYAs reported it was important for them to receive income support during treatment and 48% reported it was important after treatment. Of those AYAs who needed income support during treatment, 77% also reported needing income support after treatment. The need for income support for AYAs during treatment was significantly associated with older age at diagnosis and being unemployed (see Table 4). Table 4: Adolescents and Young Adults (AYA) and Parent Logistic Regression—Need for Income Support during Treatment AYA Parent Variable OR 95% CI p OR 95% CI p Sociodemographic variables  Gender   Male 0.83 [0.47, 1.49] .54 1.17 [0.56, 2.42] .68   (Female)  Geography   Regional/rural 1.34 [0.73, 2.48] .34 2.40 [1.32, 4.37] <.01   (Metropolitan)  Living arrangement   Family home 0.67 [0.35, 1.29] .23 3.4 [0.91, 12.42] .07   (Living outside family home)  Employment status   Unemployed 3.29 [1.28, 8.45] .01   (Employed)  Education level Completed year 10 or equivalent 0.62 [0.17, 2.20] .46 Completed year 12 or equivalent 0.67 [0.19, 2.34] .53 Certificate or diploma 0.58 [0.18, 1.90] .37 Bachelor or higher degree (< year 10) 0.20 [0.06, 0.69] .01 Clinical setting and treatment variables  Age (years) at diagnosis Age of AYA at diagnosis   20–25 2.22 [1.23, 4.01] <.01 15–19 yearsa 4.9 [2.2, 10.96] <.001   (15–19 years) (20–25 years)  Length of hospital stay   Long stay 1.24 [0.69, 2.23] .48 3.8 [2.08, 7.11] <.001   (Short stay)  On/off treatment   On treatment 1.06 [0.50, 2.23] .88 1.27 [0.66, 2.45] .48   (Off treatment)  Treatment setting   Adult 1.19 [0.51, 2.78] .69 7.8 [3.3, 18.42] <.001   (Pediatric)  Cancer type   Blood cancer 1.13 [0.63, 2.03] .68 2.3 [1.06, 4.97] .04   (Other) AYA Parent Variable OR 95% CI p OR 95% CI p Sociodemographic variables  Gender   Male 0.83 [0.47, 1.49] .54 1.17 [0.56, 2.42] .68   (Female)  Geography   Regional/rural 1.34 [0.73, 2.48] .34 2.40 [1.32, 4.37] <.01   (Metropolitan)  Living arrangement   Family home 0.67 [0.35, 1.29] .23 3.4 [0.91, 12.42] .07   (Living outside family home)  Employment status   Unemployed 3.29 [1.28, 8.45] .01   (Employed)  Education level Completed year 10 or equivalent 0.62 [0.17, 2.20] .46 Completed year 12 or equivalent 0.67 [0.19, 2.34] .53 Certificate or diploma 0.58 [0.18, 1.90] .37 Bachelor or higher degree (< year 10) 0.20 [0.06, 0.69] .01 Clinical setting and treatment variables  Age (years) at diagnosis Age of AYA at diagnosis   20–25 2.22 [1.23, 4.01] <.01 15–19 yearsa 4.9 [2.2, 10.96] <.001   (15–19 years) (20–25 years)  Length of hospital stay   Long stay 1.24 [0.69, 2.23] .48 3.8 [2.08, 7.11] <.001   (Short stay)  On/off treatment   On treatment 1.06 [0.50, 2.23] .88 1.27 [0.66, 2.45] .48   (Off treatment)  Treatment setting   Adult 1.19 [0.51, 2.78] .69 7.8 [3.3, 18.42] <.001   (Pediatric)  Cancer type   Blood cancer 1.13 [0.63, 2.03] .68 2.3 [1.06, 4.97] .04   (Other) Notes: Reference categories are in parentheses. Bold used to highlight p values of less than .05. OR = odds ratio; CI = confidence interval. aParent age at diagnosis has older age group as reference. Table 4: Adolescents and Young Adults (AYA) and Parent Logistic Regression—Need for Income Support during Treatment AYA Parent Variable OR 95% CI p OR 95% CI p Sociodemographic variables  Gender   Male 0.83 [0.47, 1.49] .54 1.17 [0.56, 2.42] .68   (Female)  Geography   Regional/rural 1.34 [0.73, 2.48] .34 2.40 [1.32, 4.37] <.01   (Metropolitan)  Living arrangement   Family home 0.67 [0.35, 1.29] .23 3.4 [0.91, 12.42] .07   (Living outside family home)  Employment status   Unemployed 3.29 [1.28, 8.45] .01   (Employed)  Education level Completed year 10 or equivalent 0.62 [0.17, 2.20] .46 Completed year 12 or equivalent 0.67 [0.19, 2.34] .53 Certificate or diploma 0.58 [0.18, 1.90] .37 Bachelor or higher degree (< year 10) 0.20 [0.06, 0.69] .01 Clinical setting and treatment variables  Age (years) at diagnosis Age of AYA at diagnosis   20–25 2.22 [1.23, 4.01] <.01 15–19 yearsa 4.9 [2.2, 10.96] <.001   (15–19 years) (20–25 years)  Length of hospital stay   Long stay 1.24 [0.69, 2.23] .48 3.8 [2.08, 7.11] <.001   (Short stay)  On/off treatment   On treatment 1.06 [0.50, 2.23] .88 1.27 [0.66, 2.45] .48   (Off treatment)  Treatment setting   Adult 1.19 [0.51, 2.78] .69 7.8 [3.3, 18.42] <.001   (Pediatric)  Cancer type   Blood cancer 1.13 [0.63, 2.03] .68 2.3 [1.06, 4.97] .04   (Other) AYA Parent Variable OR 95% CI p OR 95% CI p Sociodemographic variables  Gender   Male 0.83 [0.47, 1.49] .54 1.17 [0.56, 2.42] .68   (Female)  Geography   Regional/rural 1.34 [0.73, 2.48] .34 2.40 [1.32, 4.37] <.01   (Metropolitan)  Living arrangement   Family home 0.67 [0.35, 1.29] .23 3.4 [0.91, 12.42] .07   (Living outside family home)  Employment status   Unemployed 3.29 [1.28, 8.45] .01   (Employed)  Education level Completed year 10 or equivalent 0.62 [0.17, 2.20] .46 Completed year 12 or equivalent 0.67 [0.19, 2.34] .53 Certificate or diploma 0.58 [0.18, 1.90] .37 Bachelor or higher degree (< year 10) 0.20 [0.06, 0.69] .01 Clinical setting and treatment variables  Age (years) at diagnosis Age of AYA at diagnosis   20–25 2.22 [1.23, 4.01] <.01 15–19 yearsa 4.9 [2.2, 10.96] <.001   (15–19 years) (20–25 years)  Length of hospital stay   Long stay 1.24 [0.69, 2.23] .48 3.8 [2.08, 7.11] <.001   (Short stay)  On/off treatment   On treatment 1.06 [0.50, 2.23] .88 1.27 [0.66, 2.45] .48   (Off treatment)  Treatment setting   Adult 1.19 [0.51, 2.78] .69 7.8 [3.3, 18.42] <.001   (Pediatric)  Cancer type   Blood cancer 1.13 [0.63, 2.03] .68 2.3 [1.06, 4.97] .04   (Other) Notes: Reference categories are in parentheses. Bold used to highlight p values of less than .05. OR = odds ratio; CI = confidence interval. aParent age at diagnosis has older age group as reference. Being unemployed was also associated with AYAs needing income support after treatment. AYAs who indicated they did not need government income support reported financial assistance from other sources, including preexisting employment structures, income protection, parents, and personal savings (see Table 5). Table 5: Adolescents and Young Adults and Parent Logistic Regression—Need for Income Support after Treatment Variable Adolescents and Young Adults Parent OR 95% CI p OR 95% CI p Sociodemographic variables  Gender   Male   (Female) 0.64 [0.36, 1.14] .13 0.89 [0.38, 2.07] .79  Geography   Metropolitan   (Regional/rural) 1.34 [0.74, 2.44] .34 2.31 [1.18, 4.5] .01  Living arrangement   Family home 0.79 [0.42, 1.49] .47 1.54 [0.41, 5.85] .53   (Living outside family home)  Employment status   Unemployed 3.22 [1.40, 7.44] .01   (Employed) Clinical setting and treatment variables  Age at diagnosis Age at Diagnosis   20–25 years 1.14 [0.64, 2.04] .65 15–19 yearsa 3.97 [1.53, 10.28] <.01   (15–19 years) (20–25 years)  Length of hospital stay   Long stay 1.50 [0.84, 2.68] .17 2.30 [1.18, 4.51] .02   (Short stay)  Treatment setting   Adult 1.17 [0.50, 2.73] .72 3.31 [1.52, 7.2] <.01   (Pediatric)  Cancer type   Blood cancer 0.60 [0.33, 1.06] .08 1.50 [0.63, 3.61] .36   (Other) Variable Adolescents and Young Adults Parent OR 95% CI p OR 95% CI p Sociodemographic variables  Gender   Male   (Female) 0.64 [0.36, 1.14] .13 0.89 [0.38, 2.07] .79  Geography   Metropolitan   (Regional/rural) 1.34 [0.74, 2.44] .34 2.31 [1.18, 4.5] .01  Living arrangement   Family home 0.79 [0.42, 1.49] .47 1.54 [0.41, 5.85] .53   (Living outside family home)  Employment status   Unemployed 3.22 [1.40, 7.44] .01   (Employed) Clinical setting and treatment variables  Age at diagnosis Age at Diagnosis   20–25 years 1.14 [0.64, 2.04] .65 15–19 yearsa 3.97 [1.53, 10.28] <.01   (15–19 years) (20–25 years)  Length of hospital stay   Long stay 1.50 [0.84, 2.68] .17 2.30 [1.18, 4.51] .02   (Short stay)  Treatment setting   Adult 1.17 [0.50, 2.73] .72 3.31 [1.52, 7.2] <.01   (Pediatric)  Cancer type   Blood cancer 0.60 [0.33, 1.06] .08 1.50 [0.63, 3.61] .36   (Other) Notes: Reference categories are in parentheses. Bold used to highlight p values of less than .05. OR = odds ratio; CI = confidence interval. aParent age at diagnosis has older age group as reference. Table 5: Adolescents and Young Adults and Parent Logistic Regression—Need for Income Support after Treatment Variable Adolescents and Young Adults Parent OR 95% CI p OR 95% CI p Sociodemographic variables  Gender   Male   (Female) 0.64 [0.36, 1.14] .13 0.89 [0.38, 2.07] .79  Geography   Metropolitan   (Regional/rural) 1.34 [0.74, 2.44] .34 2.31 [1.18, 4.5] .01  Living arrangement   Family home 0.79 [0.42, 1.49] .47 1.54 [0.41, 5.85] .53   (Living outside family home)  Employment status   Unemployed 3.22 [1.40, 7.44] .01   (Employed) Clinical setting and treatment variables  Age at diagnosis Age at Diagnosis   20–25 years 1.14 [0.64, 2.04] .65 15–19 yearsa 3.97 [1.53, 10.28] <.01   (15–19 years) (20–25 years)  Length of hospital stay   Long stay 1.50 [0.84, 2.68] .17 2.30 [1.18, 4.51] .02   (Short stay)  Treatment setting   Adult 1.17 [0.50, 2.73] .72 3.31 [1.52, 7.2] <.01   (Pediatric)  Cancer type   Blood cancer 0.60 [0.33, 1.06] .08 1.50 [0.63, 3.61] .36   (Other) Variable Adolescents and Young Adults Parent OR 95% CI p OR 95% CI p Sociodemographic variables  Gender   Male   (Female) 0.64 [0.36, 1.14] .13 0.89 [0.38, 2.07] .79  Geography   Metropolitan   (Regional/rural) 1.34 [0.74, 2.44] .34 2.31 [1.18, 4.5] .01  Living arrangement   Family home 0.79 [0.42, 1.49] .47 1.54 [0.41, 5.85] .53   (Living outside family home)  Employment status   Unemployed 3.22 [1.40, 7.44] .01   (Employed) Clinical setting and treatment variables  Age at diagnosis Age at Diagnosis   20–25 years 1.14 [0.64, 2.04] .65 15–19 yearsa 3.97 [1.53, 10.28] <.01   (15–19 years) (20–25 years)  Length of hospital stay   Long stay 1.50 [0.84, 2.68] .17 2.30 [1.18, 4.51] .02   (Short stay)  Treatment setting   Adult 1.17 [0.50, 2.73] .72 3.31 [1.52, 7.2] <.01   (Pediatric)  Cancer type   Blood cancer 0.60 [0.33, 1.06] .08 1.50 [0.63, 3.61] .36   (Other) Notes: Reference categories are in parentheses. Bold used to highlight p values of less than .05. OR = odds ratio; CI = confidence interval. aParent age at diagnosis has older age group as reference. Thirty-eight percent of parents reported that it was important to receive income support for themselves during their child’s treatment. Income support was significantly associated with parents who lived in a rural or regional area, whose child was younger (15 to 19 years) at diagnosis, had an inpatient stay of one month or longer, was treated in a pediatric setting, and had a blood cancer compared with other types of cancer. Parents who had a university education level had a reduced likelihood of requiring income support during treatment (see Table 4). After treatment, 26% of parents reported it was important for them to receive income support. Sixty-eight percent of parents who reported needing income support during treatment also reported needing income support after treatment. In regression analyses, income support after treatment was significantly associated with parents who lived in a rural or regional area, whose child was younger at diagnosis, whose child required a longer inpatient stay, and whose child was treated in a pediatric setting (see Table 5). Education attainment level data were not analysed for parents after treatment due to low numbers. Difficulties Accessing Income Support Fifty-two percent of AYAs and 32% of parents reported difficulties accessing income support during treatment, whereas 37% of AYAs and 22% of parents reported difficulties after treatment. Seventy-four percent of AYAs who reported needing income support during treatment had difficulty accessing it [χ2(1, N = 191) = 54.08, p < .001]; 67% of AYAs who reported needing income support after treatment also reported difficulties [χ2(1, N = 188) = 64.00, p < .001]. Similarly, of parents who reported needing income support, 70% reported challenges accessing it during their child’s cancer treatment [χ2(1, N = 194) = 86.68, p < .001], and 62% of parents who reported needing income support after treatment reported challenges accessing it [χ2(1, N = 181) = 70.62, p < .001]. Qualitative analysis of AYA and parent comments in relation to income support revealed prominent issues related to the eligibility criteria of Centrelink (see the appendix for description). Many described confusion around eligibility as the diagnosis of cancer did not fit well with the criteria for any income support scheme. One AYA said, “Centrelink was very difficult to organize as there was no specific payment for my circumstances, there was a lot of time involved to get Centrelink payments.” Others were classified as eligible for an income support scheme that appeared inappropriate due to their health circumstances. One AYA said, Centrelink are keeping me on a Newstart [job seeking allowance] and making me regularly submit medical certificates to be exempt from job-seeking requirements. Won’t grant me disability [income support] and recognize my study as it is only part-time and I’m not well enough to work or study full-time. These problems continued after treatment, especially when the AYA was unable to work but was no longer classified as being on active treatment. One said, “During treatment I received income from Centrelink, but as soon as my treatment stopped (even though I could not work) Centrelink cut me off telling me to get a job, which was not possible.” Several comments made by AYAs who had received government income support suggested it was not sufficient to cover their basic needs. One said, “$400 fortnight [from Centrelink] was barely enough to cover basic needs such as rent and petrol.” Other AYAs reported they were ineligible for government assistance due to the strictness of eligibility requirements, such as them having some financial savings. One AYA reported, “Centrelink would not support me with income, due to bank account savings.” A consistent theme within AYA and parent comments was the bureaucratic challenge of engaging with Centrelink. This included the extent of paperwork required, delays in processing applications, requests to present in person that were inappropriate for health reasons, and delays receiving financial assistance once deemed eligible. One AYA stated, “We were given wrong information, sent on wild goose chases and no support. Documents needed three to four times. Even phoned on the day of the operation wanting more paperwork that had already been given three times.” Many parents, including parents of 20- to 25-year-olds, commented on the extent to which they were required to help their child access income support: “Long, drawn-out process, my daughter decided she didn’t have energy to persevere, so I started taking over on her behalf.” Many parents made similar comments about the challenges accessing income support for themselves as carers, predominately due to carer eligibility criteria. One wrote, “Centrelink withdrew their payment/support when my son turned 16. I felt that was unfair as I still had expenses with hospital appointments, etc.” Another said, “Because of my income we were not eligible for Centrelink help, even though my wife had to give up work for six months to care for our daughter.” Other comments suggested lack of knowledge about what income assistance may have been available: “We did not receive any benefits from Centrelink—we were never made aware it was available.” Discussion This cohort study of Australian AYAs with cancer and their parent carers shows that considerable financial impacts were experienced by young people and their families. More than half of the sample of both AYAs and parents reported financial difficulties during cancer treatment, and two-thirds of AYAs and one-third of parents reported it was important to receive income support during treatment. It is notable that over two-thirds of AYAs and parent carers who reported needing financial support described difficulty accessing it during treatment, with bureaucratic challenges commonly experienced around accessing government financial support. Although the financial impacts of cancer have been previously shown for cohorts of older adults and younger children, few studies have articulated the particular challenges for 15- to 25-year-olds. To our knowledge, this is the first study to outline the extent of financial impact on parent carers of AYAs with cancer. Financial issues and income support remained pertinent for many AYAs after treatment, with almost half of those who were off treatment reporting they were either only partially or not back on track with work. Given that survival rates from cancer for AYAs are quite high (Australian Institute of Health and Welfare, 2011), and the potential of protracted chronic health effects, there appears to be a need for ongoing income support for at least some. This is reflective of the adult cancer literature in which the duration for which income support is required is increasingly appreciated (D’Agostino et al., 2011; Paalman et al., 2016; Wakefield et al., 2013; Yabroff et al., 2016). A notable finding of our study is that almost three-quarters of AYAs were living with their parents at the time they completed the survey. We are unable to identify whether, and if so, what proportion of young people moved home following the diagnosis of cancer or specifically due to financial challenges. Our overall proportion of 15- to 24-year-olds (79%) living at home is considerably higher than the Australian average of 65.7%. More specifically, a higher proportion of young people age 25 years and older lived at home in our sample (21%) than in the Australian population (12%) (National Housing Supply Council, 2013). It may be anticipated that AYAs with cancer would choose to live with family during and after cancer treatment for physical and emotional support. That almost two-thirds of AYAs who reported financial difficulties were living with their parents suggests that AYAs may also move home or continue to live at home for financial support. Regression data indicate that living in the family home is more likely to provide a protective effect on AYA financial issues, which potentially supports the notion that financial impacts may be a contributing factor to the high numbers of AYAs living with their families. Despite Australia’s national universal health care system, almost one-half of AYAs (44%) and almost one-third of parents (29%) reported financial challenges due to direct costs of medical expenses. These results may reflect the growing trend of increased medical expenses, higher medical and pharmaceutical co-payments (Gordon et al., 2017), and the extent of gap payments for both public and private patients that have been reported for Australian adults with cancer (Gordon et al., 2017). Further research is required to fully comprehend the extent of direct medical costs associated with cancer that are borne by Australian AYAs and their families. High indirect costs of transportation, including vehicle parking, were a substantial financial burden for many parents, consistent with previous research on families of children and adults with cancer (Brooks, Wilson, & Amir, 2011; Cohn et al., 2003; Heath et al., 2006; Stommel, Given, & Given, 1993). Together with the costs of accommodation, this likely explains why a regional or rural location was associated with parent financial issues and is particularly relevant for Australia given the large distances that many families have to travel for cancer treatment (Daniel et al., 2013; Fluchel et al., 2014; McGrath et al., 1999). Although families from regional or rural areas in Australia can access special initiatives to alleviate travel costs, such as sponsored accommodation and travel assistance, it is not known to what extent these schemes were used by this cohort. What is known is that family accommodation support is heavily used and not always available (Cohn et al., 2003; Daniel et al., 2013). Finally, more than one-quarter of parents reported needing income support both during and after treatment. During treatment, income support was significantly associated with a blood cancer diagnosis. Younger age at diagnosis, pediatric treatment setting, and prolonged admission were also significant and are factors that likely reflect the intensive treatment of blood cancers such as leukemia in younger populations. Many of these same factors were associated with the need for ongoing income support. Consistent with the pediatric literature (Daniel et al., 2013; Fluchel et al., 2014), this reinforces how long income support may be required for some AYA carers. Our data suggest that AYAs with cancer and their families are multiply disadvantaged financially. They experience significant financial expenses due to cancer, which for many families is compounded by loss of income of the AYA and a parent. In addition to many AYAs being ineligible for financial assistance, most parents in this cohort were also ineligible for any significant government assistance as carers. This raises questions about the appropriateness of current eligibility criteria for income support for AYAs with cancer as well as for their parent carers. There was confusion about which of the different income support schemes AYAs with cancer were eligible. Furthermore, many of those deemed eligible were placed on schemes that appeared inappropriate given their health circumstances. These data suggest that a more systematic approach is required for the assessment of the financial needs of AYAs with cancer who have diverse and changing health needs. This would ideally recognize the extent and duration of health needs; the high costs of cancer care; the availability of family carers; and for many, a significant delay in their ability to return to their former level of study or work. A similar approach could be used for the financial assessment of parent carers. There are several limitations of this study. Being cross-sectional, it precludes interpretation of causality. A longitudinal repeated measures design would enable more dynamic assessment of the financial impacts of cancer over time. Families with limited English literacy, those with cognitive impacts, and those too unwell to participate were excluded. These groups would be expected to experience financial issues, suggesting that the data presented here are conservative and may well be an underestimate of AYAs and their families who are financially affected by cancer. Although the relatively low response rate means that caution must be exercised when extrapolating findings to other populations, especially to those with different systems of health financing, insurance, and financial support, this response rate is consistent with other studies of this age group (Clinton-McHarg, Carey, Sanson-Fisher, & Tracey, 2011; Drew, Duncan, & Sawyer, 2010). Our study did not comprehensively explore the extent of out-of-pocket medical expenses, the impact of private health insurance, household income, or the role of charitable organizations in lessening the financial burden for AYAs and their families. Further research is required to fully appreciate the financial impact of cancer on Australian AYAs and their families and how this might best be alleviated. In conclusion, this study has identified that cancer in AYAs is a cause of significant financial impact in both young people themselves and their parent carers, not only during treatment but well into survivorship. A key finding is the extent to which 15- to 25-year-old AYAs rely on their parents for financial support. These findings suggest that, like the Psychosocial Standards of Care outlined for pediatric cancer (Pelletier & Bona, 2015), policy around financial support for AYAs with cancer must also extend to address the financial impacts on families to alleviate the substantial financial burdens that accrue from AYAs’ cancer experience. Any review of government policy related to income support should consider the introduction of a medium-term disability component or extended sickness benefit for those people with complex illnesses, including AYAs with cancer and their caregivers. Robyn J. McNeil, MPH, is project manager, Centre for Adolescent Health and Maria McCarthy, PhD, is senior research officer, Social and Mental Health Aspects of Illness, Murdoch Children’s Research Institute, Parkville, Victoria, Australia. David Dunt, PhD, is professor emeritus for health policy, School of Population and Global Health, University of Melbourne, Australia. Kate Thompson, MASW, is program manager, OnTrac at Peter Mac Victorian Adolescent and Young Adult Cancer Service, Melbourne, Victoria, Australia. Silja Kosola, MD, PhD, is postdoctoral researcher, Pediatric Research Centre, Helsinki Children’s Hospital. Lisa Orme, MD, PhD, is medical director, OnTrac at Peter Mac Victorian Adolescent and young Adult Cancer Service, Melbourne, Victoria, Australia. Sarah Drew, PhD, (deceased) was research fellow, Centre for Adolescent Health, Department of Pediatrics, University of Melbourne, Australia. Susan Sawyer, MD, PhD, is professor and director, Centre for Adolescent Health, Victoria, Australia. Address correspondence to Robyn J. McNeil, Centre for Adolescent Health, Murdoch Children’s Research Institute, 50 Flemington Road, Parkville, VIC 3052 Australia; e-mail: [email protected]. This project was funded by a Cancer Australia grant (APP1010977) in association with Beyond Blue and CanTeen. Additional funding was gratefully received from Redkite, OnTrac at Peter Mac Victorian Youth Cancer Service, the Victorian Department of Health (Cancer Strategy and Planning, Department of Health), and the Royal Children’s Hospital Foundation. The Murdoch Children’s Research Institute is supported by the Victorian government’s Operational Infrastructure Support Program. The authors wish to thank the young people and their families who participated in this study, as well as the many staff responsible for data collection at each site. The authors are indebted to Dr. Sarah Drew, whose passion for improving the quality of care provided to adolesents and young adults with cancer and their families initiated this research but whose own cancer journey led to her premature death. References Aaronson , N. K. , Mattioli , V. , Minton , O. , Weis , J. , Johansen , C. , Dalton , S. O. , et al. ( 2014 ). Beyond treatment: Psychosocial and behavioural issues in cancer survivorship research and practice . European Journal of Cancer Supplements , 12 ( 1 ), 54 – 64 . Google Scholar Crossref Search ADS PubMed Australian Institute of Health and Welfare . ( 2011 ). Cancer in adolescents and young adults in Australia (Cancer series no. 62, Cat. no. CAN59). Canberra : Author . Australian Insititute of Health and Welfare & Australian Association of Cancer Registries . ( 2007 ). Cancer in Australia: An overview, 2006 (Cancer series no. 32, Cat. no. CAN32). Canberra : Australian Institute of Health and Welfare . Bloom , J. R. ( 2002 ). Surviving and thriving? Psycho‐Oncology , 11 ( 2 ), 89 – 92 . Google Scholar Crossref Search ADS PubMed Braun , V. , & Clarke , V. ( 2006 ). Using thematic analysis in psychology . Qualitative Research in Psychology , 3 ( 2 ), 77 – 101 . Google Scholar Crossref Search ADS Brooks , J. , Wilson , K. , & Amir , Z. ( 2011 ). Additional financial costs borne by cancer patients: A narrative review . European Journal of Oncology Nursing , 15 , 302 – 310 . doi:10.1016/j.ejon.2010.10.005 Google Scholar Crossref Search ADS PubMed Clinton-McHarg , T. , Carey , M. , Sanson-Fisher , R. , & Tracey , E. ( 2011 ). Recruitment of representative samples for low incidence cancer populations: Do registries deliver? BMC Medical Research Methodology , 11 ( 1 ), 1 – 9 . doi:10/1186/1471-2288-11-5 Google Scholar Crossref Search ADS PubMed Cohn , R. J. , Goodenough , B. , Foreman , T. , & Suneson , J. ( 2003 ). Hidden financial costs in treatment for childhood cancer: An Australian study of lifestyle implications for families absorbing out-of-pocket expenses . Journal of Pediatric Hematolology/Oncology , 25 , 854 – 863 . Google Scholar Crossref Search ADS D’Agostino , N. M. , Penney , A. , & Zebrack , B. ( 2011 ). Providing developmentally appropriate psychosocial care to adolescent and young adult cancer survivors . Cancer , 117 ( Suppl. 10 ), 2329 – 2334 . Google Scholar Crossref Search ADS PubMed Daniel , G. , Wakefield , C. , Ryan , B. , Fleming , C. , Levett , N. , & Cohn , R. ( 2013 ). Accommodation in pediatric oncology: Parental experiences, preferences and unmet needs . Rural and Remote Health , 13 ( 2 ), 1 – 9 . Drew , S. E. , Duncan , R. E. , & Sawyer , S. M. ( 2010 ). Visual storytelling: A beneficial but challenging method for health research with young people . Qualitative Health Research , 20 , 1677 – 1688 . Google Scholar Crossref Search ADS PubMed Eiser , C. , & Upton , P. ( 2007 ). Costs of caring for a child with cancer: A questionnaire survey . Child: Care, Health and Development , 33 , 455 – 459 . Google Scholar Crossref Search ADS PubMed Fluchel , M. N. , Kirchhoff , A. C. , Bodson , J. , Sweeney , C. , Edwards , S. L. , Ding , Q. , et al. ( 2014 ). Geography and the burden of care in pediatric cancers . Pediatric Blood Cancer , 61 ( 11 ), 1918 – 1924 . doi:10.1002/pbc.25170 Google Scholar Crossref Search ADS PubMed Freyer , D. R. , & Barr , R. D. ( 2007 ). Making ends meet: Financial issues from the perspectives of patients and their healthcare team. In A. W. Bleyer & R. D. Barr (Eds.), Cancer in adolescents and young adults (pp. 489 – 500 ). New York : Springer . Geue , K. , Sender , A. , Schmidt , R. , Richter , D. , Hinz , A. , Schulte , T. , et al. ( 2014 ). Gender-specific quality of life after cancer in young adulthood: A comparison with the general population . Quality of Life Research , 23 , 1377 – 1386 . Google Scholar Crossref Search ADS PubMed Gordon , L. , Walker , S. , Mervin , M. , Lowe , A. , Smith , D. , Gardiner , R. , & Chambers , S. ( 2017 ). Financial toxicity: A potential side effect of prostate cancer treatment among Australian men . European Journal of Cancer Care , 26 ( 1 ), e12392 . doi:10.1111/ecc.12392 Google Scholar Crossref Search ADS Grinyer , A. ( 2007 ). The biographical impact of teenage and adolescent cancer . Chronic Illness , 3 , 265 – 277 . Google Scholar Crossref Search ADS PubMed Guy , G. P. , Ekwueme , D. U. , Yabroff , K. R. , Dowling , E. C. , Li , C. , Rodriguez , J. L. , et al. ( 2013 ). Economic burden of cancer survivorship among adults in the United States . Journal of Clinical Oncology , 31 , 3749 – 3757 . doi:10.1200/JCO.2013.49.1241 Google Scholar Crossref Search ADS PubMed Hall , A. E. , Boyes , A. W. , Bowman , J. , Walsh , R. A. , James , E. L. , & Girgis , A. ( 2012 ). Young adult cancer survivors’ psychosocial well-being: A cross-sectional study assessing quality of life, unmet needs, and health behaviors . Supportive Care in Cancer , 20 , 1333 – 1341 . Google Scholar Crossref Search ADS PubMed Heath , J. A. , Lintuuran , R. M. , Rigguto , G. , Tikotlian , N. , & McCarthy , M. ( 2006 ). Childhood cancer: Its impact and financial costs for Australian families . Pediatric Hematology and Oncology , 23 , 439 – 448 . Google Scholar Crossref Search ADS PubMed Kim , P. ( 2007 ). Cost of cancer care: The patient perspective . Journal of Clinical Oncology , 25 , 228 – 232 . Google Scholar Crossref Search ADS PubMed Longo , C. J. , Fitch , M. , Deber , R. B. , & Williams , A. P. ( 2006 ). Financial and family burden associated with cancer treatment in Ontario, Canada . Supportive Care in Cancer , 14 , 1077 – 1085 . Google Scholar Crossref Search ADS PubMed Markman , M. , & Luce , R. ( 2010 ). Impact of the cost of cancer treatment: An Internet-based survey . Journal of Oncology Practice , 6 ( 2 ), 69 – 73 . Google Scholar Crossref Search ADS PubMed McGrath , P. , Patterson , C. , Yates , P. , Treloar , S. , Oldenburg , B. , & Loos , C. ( 1999 ). A study of postdiagnosis breast cancer concerns for women living in rural and remote Queensland. Part I: Personal concerns . Australian Journal of Rural Health , 7 ( 1 ), 34 – 42 . Google Scholar Crossref Search ADS PubMed Mehnert , A. ( 2011 ). Employment and work-related issues in cancer survivors . Critical Reviews in Oncology/Hematology , 77 ( 2 ), 109 – 130 . Google Scholar Crossref Search ADS PubMed National Housing Supply Council . ( 2013 ). Housing supply and affordability issues 2012–13. Retrieved from https://treasury.gov.au/publication/housing-supply-and-affordability-issues-2012-13/ O’Brien , R. ( 2007 ). A caution regarding rules of thumb for variance inflation factors . Quality & Quantity , 41 , 673 – 690 . doi:10.1007/s11135-006-9018-6 Google Scholar Crossref Search ADS Osborn , M. , Little , C. , Bowering , S. , & Orme , L. ( 2013 ). Youth cancer services in Australia: Development and implementation (International perspectives on AYAO, Part 3) . Journal of Adolescent and Young Adult Oncology , 2 ( 3 ), 118 – 124 . Google Scholar Crossref Search ADS PubMed Paalman , C. , van Leeuwen , F. , Aaronson , N. , de Boer , A. , van de Poll-Franse , L. , Oldenburg , H. , & Schaapveld , M. ( 2016 ). Employment and social benefits up to 10 years after breast cancer diagnosis: A population-based study . British Journal of Cancer , 114 ( 1 ), 81 – 87 . Google Scholar Crossref Search ADS PubMed Pai , A. L. , Patiño-Fernández , A. M. , McSherry , M. , Beele , D. , Alderfer , M. A. , Reilly , A. T. , et al. ( 2008 ). The Psychosocial Assessment Tool (PAT2.0): Psychometric properties of a screener for psychosocial distress in families of children newly diagnosed with cancer . Journal of Pediatric Psychology , 33 ( 1 ), 50 – 62 . Google Scholar Crossref Search ADS PubMed Patton , G. C. , Sawyer , S. M. , Santelli , J. S. , Ross , D. A. , Afifi , R. , Allen , N. B. , et al. ( 2016 ). Our future: A Lancet commission on adolescent health and wellbeing . Lancet , 387 ( 10036 ), 2423 – 2478 . Google Scholar Crossref Search ADS PubMed Pelletier , W. , & Bona , K. ( 2015 ). Assessment of financial burden as a standard of care in pediatric oncology . Pediatric Blood & Cancer , 62 ( Suppl. 5 ), S619 – S631 . Google Scholar Crossref Search ADS PubMed Rice , P. L. , & Ezzy , D. ( 1999 ). Qualitative research methods: A health focus. Melbourne, Australia : Oxford University Press . Sawyer , S. M. , Afifi , R. A. , Bearinger , L. H. , Blakemore , S.-J. , Dick , B. , Ezeh , A. C. , & Patton , G. C. ( 2012 ). Adolescence: A foundation for future health . Lancet , 379 ( 9826 ), 1630 – 1640 . Google Scholar Crossref Search ADS PubMed Sawyer , S. M. , McCarthy , M. C. , Dunt , D. , McNeil , R. , Thompson , K. , Orme , L. , & Drew , S. E. ( 2016 ). Fulfilling the vision of youth-friendly cancer care: A study protocol . Journal of Adolescent and Young Adult Oncology , 5 , 267 – 277 . Google Scholar Crossref Search ADS PubMed StataCorp . ( 2013 ). Stata (Version 13) [Computer software]. College Station, TX : Author . Stommel , M. , Given , C. W. , & Given , B. A. ( 1993 ). The cost of cancer home care to families . Cancer , 71 , 1867 – 1874 . Google Scholar Crossref Search ADS PubMed Teenage Cancer Trust . ( 2015 ). About us. Retrieved from https://www.teenagecancertrust.org/about-us/ Thompson , K. , Palmer , S. , & Dyson , G. ( 2009 ). Adolescents & young adults: Issues in transition from active therapy into follow-up care . European Journal of Oncology Nursing , 13 ( 3 ), 207 – 212 . Google Scholar Crossref Search ADS PubMed Wakefield , C. , McLoone , J. , Butow , P. , Lenthen , K. , & Cohn , R. ( 2013 ). Support after the completion of cancer treatment: Perspectives of Australian adolescents and their families . European Journal of Cancer Care , 22 , 530 – 539 . Google Scholar Crossref Search ADS PubMed Wakefield , C. , McLoone , J. K. , Evans , N. T. , Ellis , S. J. , & Cohn , R. J. ( 2014 ). It’s more than dollars and cents: The impact of childhood cancer on parents’ occupational and financial health . Journal of Psychosocial Oncology , 32 , 602 – 621 . Google Scholar Crossref Search ADS PubMed Yabroff , K. R. , Dowling , E. C. , Guy , G. P. Jr. , Banegas , M. P. , Davidoff , A. , Han , X. , et al. ( 2016 ). Financial hardship associated with cancer in the United States: Findings from a population-based sample of adult cancer survivors . Journal of Clinical Oncology , 34 , 259 – 267 . doi:10.1200/jco.2015.62.0468 Google Scholar Crossref Search ADS PubMed Zafar , S. Y. , Peppercorn , J. M. , Schrag , D. , Taylor , D. H. , Goetzinger , A. M. , Zhong , X. , & Abernethy , A. P. ( 2013 ). The financial toxicity of cancer treatment: A pilot study assessing out-of-pocket expenses and the insured cancer patient’s experience . Oncologist , 18 , 381 – 390 . Google Scholar Crossref Search ADS PubMed Zebrack , B. ( 2011 ). Psychological, social, and behavioral issues for young adults with cancer . Cancer , 117 ( Suppl. 10 ), 2289 – 2294 . Google Scholar Crossref Search ADS PubMed Zebrack , B. , Kent , E. E. , Keegan , T. H. , Kato , I. , Smith , A. W. , & AYA HOPE Study Collaborative Group . ( 2014 ). “Cancer sucks,” and other ponderings by adolescent and young adult cancer survivors . Journal of Psychosocial Oncology , 32 ( 1 ), 1 – 15 . Google Scholar Crossref Search ADS PubMed © 2018 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)

Martin-West,, Stacia

doi: 10.1093/swr/svy028pmid: N/A

Abstract With a dwindling supply of affordable housing and limited public dollars to help families afford rising rents, lower-income households and those headed by a single mother often pay rent late and face the threat of eviction. Social support systems have long been associated with helping families weather financial instability; yet, no research has explored whether different types and levels of social support are associated with housing stability for different family structures. The 2008 Survey of Income and Program Participation and both additive and multiplicative interactions were used to test for moderating effects of social support on a late rent payment for two-parent households (n = 1,106) and those headed by single mothers (n = 813). Compared with two-parent households, single mothers with little to no perceived social support from family experienced a 33% increase in the relative risk of a late rent payment. Those who indicated little to no social support from other community resources experienced a 68% greater risk of a late rent payment compared with two-parent families. Findings are discussed in the context of a potential economic climate that may exacerbate inequality and will require social workers to cultivate and maintain informal support networks that can buffer financial insecurity. Although national estimates point to overall reductions in poverty, single mothers’ rates of poverty remain constant—approximately one in three households headed by a single mother fall below the federal poverty line (National Women’s Law Center, 2016). These households frequently contend with precarious financial circumstances, including income insufficiency and volatility, limited opportunity to draw on social networks for financial support, lack of access to financial services, and limited liquid assets (Casey, 2011; Financial Industry Regulatory Authority [FINRA] Foundation, 2016; Goldberg, 2014; Lusardi, Schneider, & Tufano, 2011; Sherraden, 2013). In the event of an unforeseen emergency or sudden drop in household income, the majority of families headed by a single mother often cannot draw on existing assets and must make difficult choices to stretch already limited financial resources. For some, this means relying on family, friends, and community resources for support (Henly, Danziger, & Offer, 2005; Mills & Zhang, 2013). It may also mean prioritizing which bills are paid on time and which ones can wait without considerable consequence (Edin & Shaefer, 2015). In 2014, lower-income households, which disproportionately comprise single-mother families, made up nearly 50% of the households using over one-third of their monthly income for rent (Steffen et al., 2015). When that rent money is reprioritized for an overdue utility payment or other financial emergency, single mothers, particularly black and Latina mothers, are at a high risk of eviction (Desmond, 2012b). The financial lives of lower-income single mothers are delicate balancing acts that require deft negotiations of relationships, resources, and institutions. Single mothers have historically struggled more than others to make ends meet, and renewed efforts to cut already insufficient federal programs may translate to even fewer public resources for vulnerable families and introduce new complications to the balancing act. Beginning with welfare reform in the late 1990s and extending through the most recent figures available in 2015, the number of families in poverty receiving Temporary Assistance for Needy Families has steadily declined—from 68 families of 100 in 1996 to 31 families of 100 in 2015 (Floyd, Pavetti, & Schott, 2017). In the absence of those resources, single mothers and lower-income families may be more reliant on the informal support systems found in their communities. It will be the role of social workers to determine if and how existing sources of support can be engaged or augmented to help families weather these difficult times (Williams, 2017). As such, the purpose of this article is to determine if perceived sources of instrumental social support may moderate housing instability in both two-parent households and those headed by a single mother. Implications for social work policy and practice in the context of public welfare retrenchment are discussed. Literature Review Housing Affordability and Instability Single mothers cope with high rent burdens, substandard housing, and unstable neighborhoods (Desmond, 2014; National Low Income Housing Coalition [NLIHC], 2013; Steffen et al., 2015). Even among families who participate in affordable housing programs, single mothers are at the greatest risk for being late on or missing a rent payment (Brisson & Covert, 2015). When compared with two-parent families and households headed by a single father, they are also at the greatest risk of being late on a rent payment (West, 2016). Lower-income black women, especially those with children, appear to be at the greatest risk for eviction according to a regional study of eviction records (Desmond, 2014, 2015). The shortage of affordable housing, inadequate income, and cuts to public welfare programs have affected single mothers’ abilities to avoid housing instability. In terms of affordable housing, supply has failed to keep pace with need. In 2014, for every 100 households earning 30% or below of the area median income, there were only 31 available rental homes or apartments that met affordability guidelines (NLIHC, 2015). Of the extremely low-income households in need of affordable housing that would use only 30% of their monthly income, 59% were either single women or women with children (NLIHC, 2013). Another metric for assessing extant need in housing research is worst case housing need (WCHN), defined as very low-income renters with incomes below 50 percent of the Area Median Income (AMI) who do not receive government housing assistance and who either paid more than half of their income for rent or lived in severely inadequate conditions, or who faced both of these challenges. (Steffen et al., 2015, p.vii) The percentage of renters who faced a WCHN increased by 43.5% from 2007 to 2011, although in 2014, the rate declined by 9% to 8.7 million renters. Families with children accounted for 40.3% of households with WCHNs in 2014 (Steffen et al., 2015). Given that single mothers made up a disproportionate share of households experiencing poverty, many families facing worst case needs were likely headed by single mothers. Inadequate incomes exacerbated by cuts to income-based public welfare programs have made it difficult for single mothers to cover housing expenses in unsubsidized units. The national average hourly wage required to afford a two-bedroom rental unit while only spending 30% of monthly income is $22.10 (Aurand et al., 2018). A single mother with one dependent child would need to earn a gross income of approximately $40,000 annually to cover housing expenses not in excess of 30% of her income. In comparison with 2015 figures, when single mothers had median earnings of $37,797 per year, there was a gap of nearly $3,000 between their median annual income and the amount of annual income required to secure housing without facing a high rent burden (Proctor, Semega, & Kollar, 2016). Household Economic Instability Declining real wages and income volatility have also led to increased income instability among households headed by single mothers. For example, 20% of households with children in the lowest income quintile of the Survey of Income and Program Participation (SIPP) across three waves lost at least 50% of their income in a year (Acs, Loprest, & Nichols, 2009). Of those, only about one-half were able to make a full income recovery in the next year. Welfare reform further exacerbated income volatility among lower-income households. Rates of income volatility among single mothers increased 60% between the passage of welfare reform in 1996 and follow-up measures in 2004 (Bollinger & Ziliak, 2007). Households headed by single mothers experience income shocks that require them to draw on existing resources to cover household expenses; yet some simply do not have the existing liquid assets or even access to a savings account to cover those expenses (FINRA Foundation, 2016; Lusardi et al., 2011). Having ready access to liquid assets helps mitigate adverse economic events for lower-income families who are more likely to experience an income shock (McKernan, Ratcliffe, & Vinopal, 2009; Mills & Amick, 2010). Yet, when single mothers are able to accumulate some liquid assets, they are more likely than other groups to have to deplete them and reenter asset poverty (Leonard & Di, 2014). Those lower-income households without liquid assets were more likely than others to report food instability, difficulty paying bills, and general deprivation after experiencing an income shock (McKernan et al., 2009). To withstand these somewhat frequent and substantial fluctuations in income, single mothers may turn to family, friends, or other sources of community support. Social Support Lower-income households may rely on social support to withstand material hardship associated with the inability to cover costs related to housing, utilities, medical care, and food (Mills & Zhang, 2013). As de Souza Briggs (1998) pointed out, social support and social leverage are two forms of social capital. Social support refers to close social ties with friends, family, neighbors, and social services organizations that help families to navigate the day-to-day struggle of living in poverty. In particular, lower-income mothers may use social support to respond to the hectic demands of both work and parenthood in the context of poverty (Henly & Lyons, 2000; Stack, 1974). However, as Dominguez and Watkins (2003) indicated, these close ties can be problematic and tenuous. Social support may be “used up,” particularly when a mother has multiple, complex needs and cannot reciprocate the help she receives from friends or family (Desmond, 2012a; Harknett & Hartnett, 2011; Kalil & Ryan, 2010; Letiecq, Anderson, & Koblinsky, 1998). In these instances, mothers may turn to social services agencies to receive more reliable assistance that, due to its institutionalized nature, is less emotionally demanding and has little to no expectation of reciprocity (Dominguez & Watkins, 2003). The perception of social support in lower-income communities may be overstated when compared with actual receipt of that support, perhaps because the people who make up social networks tend to be economically similar; thus a lower-income household is likely to ask a similarly situated lower-income household for help (Harknett, 2006; Meadows, 2009). Other social ties that may be made up of acquaintances or others outside of a close-knit family or group of friends are forms of social leverage (de Souza Briggs, 1998). By introducing new employment opportunities or sharing information about community resources, social leverage support may relate to greater employment and earnings and reduced material hardship (Harknett, 2006; Henly et al., 2005; Mills & Zhang, 2013). Because social leverage and social support often work in tandem, Dominguez and Watkins (2003) suggested that the social responsibilities ascribed to motherhood and female gender may preclude the ability to use social leverage for economic mobility. The measurement and operationalization of social support matters in relation to how its impacts can be interpreted. For example, Henly et al. (2005) measured perceived social support on three domains: emotional, instrumental, and financial. Harknett (2006) as well as Harknett and Hartnett (2011) measured social support as perceived material or instrumental—the ability to rely on someone for help with transportation, child care, or money; and emotional support—whether there is someone you can depend on to talk about your feelings. Each of these authors noted the nearly unavoidable endogeneity and the problematic counterfactual of research on the outcomes of social support. Greater receipt of social support may be a function of greater need or greater need may be related to greater receipt of social support. Harknett (2006) and Henly et al. (2005) ultimately contended that measurement of perceived, as opposed to received, social support may minimize these measurement issues. Four key studies lay the quantitative groundwork for exploring how social support may be related to housing stability in different households. First, single mothers are more likely to be late on housing payments compared with single fathers and two-parent households (West, 2016). Second, research using a nationally representative sample of households suggests later hardship avoidance when social support networks are present, though household type was not controlled (Mills & Zhang, 2013). Third, stronger material and emotional support as well as cohabiting with a spouse or partner were associated with greater income and employment stability in a sample of lower-income single mothers from the 1990s (Harknett, 2006). Research using a representative birth cohort sample of 20 American cities found that poor single mothers, those with mental or physical health problems, and those with greater need for help with multiple children tend to have weaker support systems (Harknett & Hartnett, 2011). If social support is financially protective for single mothers, but poor single mothers with complex needs tend to have less of that support, do those single mothers without social support end up facing greater material hardship than a two-parent family? To address this remaining question, this study tests an effect moderation model of different types of social support on a late rent payment for two-parent versus single-mother households. Method Does perceived level of instrumental social support moderate the relative risk (RR) of a late rent payment for different household types? Data This study used data from waves 4 and 6 of the 2008 SIPP. The SIPP data set, collected by the U.S. Census Bureau, contains information on a multitude of social and economic measures including liquid and illiquid assets, measures of material hardship, social support, and housing variables, making it ideal to answer the proposed research question. To adjust for sampling error, and to produce a nationally representative sample that accounts for the sample restriction to both waves 4 and 6, the data were weighted at wave 6 (WPFINWGT), the last wave of data used in the models (Shaefer, 2013). Demographic, income, and some housing and asset variables were extracted from the core files of the initial wave 4 observation period, collected August through November 2009. Housing instability and social support variables were extracted from the wave 6 core file collected April through July 2011. Beginning with the wave 4 Core Module, the wave 4 Topical Module was linked into a wide-file format based on a generated unique identifier. The wave 6 Core Module and wave 6 Topical Module were then linked in a wide-file format onto the wave 4 file. Only cases that completed the wave 4 Core, wave 4 Topical, wave 6 Core and wave 6 Topical Modules were retained in the sample. No missing data were detected in the sample because all missing values in SIPP were imputed using the hot deck imputation method before being released to the public (U.S. Census Bureau, 2008). More information related to SIPP sampling and design is accessible in the technical documentation (https://www.census.gov/programs-surveys/sipp/data/2008-panel.html). Sample Only those respondents who were surveyed at wave 4 and wave 6 and did not have a change in household type for the duration of the observation period were included in the sample. The sample was then filtered to the following inclusion criteria: respondent over age 18, child living in the household under 18 years of age, and respondent rented their primary residence. Single-father households were removed from the sample due to relatively small representation and model parsimony (n = 173). This resulted in a final analytic sample of 1,106 two-parent families and 813 families headed by a single mother. Measures This research tested a moderating relationship of perceived instrumental social support on the outcome of a late rent payment for different household types. Several demographic variables were included as controls. Dependent Variable Late rent payment was the dependent variable. Respondents were asked if they had not paid the full amount of rent at any time in the past year. Responses were coded as 0 = no and 1 = yes. Independent Variables of Interest Only those households with dependent children under 18 years were included in the sample. Household relationships were traced back to the household reference person, that is, the survey respondent. Households that included two adults where one adult was the reference person and the other indicated a partnership or marriage to the respondent adult were coded as 0 = two-parent family. Only one adult from two-parent households was the respondent. Households that contained a single woman with a dependent child or children were coded as 1 = household headed by a single mother. Three variables measuring social support were derived from questions related to expected level of social support from family, friends, or other community resources. Responses were collapsed into 0 = all or most of the help I need or 1 = no help or very little of the help I need. Control Variables Self-identified race was coded as 0 = white alone, 1 = black alone, 2 = Asian, and 3 = Native Hawaiian/Pacific Islander/other. Respondents were coded as either 0 = non-Hispanic/Latino(a) or 1 = Hispanic/Latino(a), depending on self-identification. Thus, respondents could identify as both black (race) and Hispanic/Latino(a) (ethnicity). Respondents were asked to indicate the highest level of education completed, ranging from less than first grade to a doctoral degree. These levels were condensed into the following attributes: 0 = less than high school, 1 = high school diploma or equivalent, 2 = some college, 3 = diploma, certificate or associate’s degree, 4 = college degree, and 5 = graduate or professional degree. Age was a continuous variable collected at the time of the wave 4 Core Module interview. Respondents were also asked whether they had a mental or physical disability that prevented them from working all or part of the time. Responses were coded as 0 = no and 1 = yes. The survey asked about the number of dependent children under 18 years of age in the household. To establish receipt of means-tested cash benefits, respondents were asked, “Did someone in the household receive means-tested cash benefits?” Responses were coded as 0 = no receipt of cash benefits and 1 = receipt of cash benefits. Monthly income, a continuous variable (THTOTINC), was the total amount of monthly income for all individuals over 15 years old in the household. Due to considerable skewness and kurtosis, this variable was log transformed for analysis. Respondents were asked to report the amount of money held in a checking, savings, money market, or CD account. Due to nonnormality, this continuous variable was collapsed into the following categories: 0 = less than $100, 1 = $101 to $500, and 2 = greater than $500. Used frequently in housing research, housing cost burden typically refers to paying 30% or more of total monthly income in rent. Severe housing cost burden refers to paying 50% or more of monthly income in rent (Joint Center for Housing Studies, 2017). Here, the variable is separated into three categories calculated from the percentage of monthly rent (THOMEAMT) divided by monthly income (THTOTINC). This was coded as 0 = less than or equal to 30% of income toward rent, 1 = 31% to 49% of income toward rent, and 2 = greater than or equal to 50% or more of income toward rent. The variable for debt was drawn from the question, “[Do you] have any debts in your own name, such as credit card bills, loans from a financial institution, or educational loans?” Responses were coded as 0 = no and 1 = yes. Data Analysis Data were extracted from the U.S. Census Bureau data repository using R. Sample descriptives and binary logistic regressions for each model were conducted in Stata SE version 14 (2015). First, a model testing main effects of household type and different types of social support were regressed on the late rent payment variable. Next, moderating effect analyses were conducted. Tests were completed using both multiplicative and additive interactions among variables of interest (VanderWeele & Knol, 2014; Wang & Tran, 2017). Finally, relative excess risk due to interaction (RERI) with confidence interval (CI) estimations were calculated to determine additive interaction effects (Hosmer & Lemeshow, 1992). Results Descriptives As shown in Table 1, the majority of two-parent households identified as white (76%), and nearly half of single-mother households identified as women of color. Seventeen percent of two-parent households did not have a high school diploma compared with 23% of single mothers. Single mothers also tended to be younger with an average age of 39, compared with an average age of 42 for two-parent households. Single mothers also had on average one more child than two-parent households. Single-mother households had considerably higher rates of reported mental and physical disabilities than did two-parent families. One-quarter of single mothers reported receiving household cash benefits in the prior month, compared with 8% of two-parent families. The average monthly income for two-parent households was $4,519 compared with $2,251 for single mothers. Table 1: Descriptive Statistics and Frequencies of Measures of Weighted Renter Sample (N = 1,919) Covariates % Two-Parent Family Households (n = 1,106) % Female Heads of Family Households (n = 813) Race and ethnicity  White 76 55  Black 12 38  Asian 8 2  Hawaiian or Pacific Islander 4 5  Hispanic or Latino/a 24 22 Education level  Less than high school 17 23  Some college 28 28  Diploma/certificate/associate 17 25  College degree 15 6  Graduate or professional degree 9 3 Mental or physical disability 10 15 Age M (SD) 42 (15) 39 (12) Number of children 1 (1) 2 (1) Mean monthly income $4,519 ($4,116) $2,251 ($1,989) Received household benefits 8 25 Late rent payment 11 19 Liquid asset category  <$100 69 88  $101 to $500 7 4  >$500 23 8 Debt (yes/no) 20 40 Housing cost burden  ≤30% 72 66  31% to 49% 15 13  ≥50% 13 21 Expected social support  All or most from family 66 67  All or most from friends 67 62  All or most from community 57 43 Covariates % Two-Parent Family Households (n = 1,106) % Female Heads of Family Households (n = 813) Race and ethnicity  White 76 55  Black 12 38  Asian 8 2  Hawaiian or Pacific Islander 4 5  Hispanic or Latino/a 24 22 Education level  Less than high school 17 23  Some college 28 28  Diploma/certificate/associate 17 25  College degree 15 6  Graduate or professional degree 9 3 Mental or physical disability 10 15 Age M (SD) 42 (15) 39 (12) Number of children 1 (1) 2 (1) Mean monthly income $4,519 ($4,116) $2,251 ($1,989) Received household benefits 8 25 Late rent payment 11 19 Liquid asset category  <$100 69 88  $101 to $500 7 4  >$500 23 8 Debt (yes/no) 20 40 Housing cost burden  ≤30% 72 66  31% to 49% 15 13  ≥50% 13 21 Expected social support  All or most from family 66 67  All or most from friends 67 62  All or most from community 57 43 Source: Survey of Income and Program Participation: 2008; wave 4 Core and Topical Module, wave 6 Core and Topical Module; weighted using population weights at wave 6. Table 1: Descriptive Statistics and Frequencies of Measures of Weighted Renter Sample (N = 1,919) Covariates % Two-Parent Family Households (n = 1,106) % Female Heads of Family Households (n = 813) Race and ethnicity  White 76 55  Black 12 38  Asian 8 2  Hawaiian or Pacific Islander 4 5  Hispanic or Latino/a 24 22 Education level  Less than high school 17 23  Some college 28 28  Diploma/certificate/associate 17 25  College degree 15 6  Graduate or professional degree 9 3 Mental or physical disability 10 15 Age M (SD) 42 (15) 39 (12) Number of children 1 (1) 2 (1) Mean monthly income $4,519 ($4,116) $2,251 ($1,989) Received household benefits 8 25 Late rent payment 11 19 Liquid asset category  <$100 69 88  $101 to $500 7 4  >$500 23 8 Debt (yes/no) 20 40 Housing cost burden  ≤30% 72 66  31% to 49% 15 13  ≥50% 13 21 Expected social support  All or most from family 66 67  All or most from friends 67 62  All or most from community 57 43 Covariates % Two-Parent Family Households (n = 1,106) % Female Heads of Family Households (n = 813) Race and ethnicity  White 76 55  Black 12 38  Asian 8 2  Hawaiian or Pacific Islander 4 5  Hispanic or Latino/a 24 22 Education level  Less than high school 17 23  Some college 28 28  Diploma/certificate/associate 17 25  College degree 15 6  Graduate or professional degree 9 3 Mental or physical disability 10 15 Age M (SD) 42 (15) 39 (12) Number of children 1 (1) 2 (1) Mean monthly income $4,519 ($4,116) $2,251 ($1,989) Received household benefits 8 25 Late rent payment 11 19 Liquid asset category  <$100 69 88  $101 to $500 7 4  >$500 23 8 Debt (yes/no) 20 40 Housing cost burden  ≤30% 72 66  31% to 49% 15 13  ≥50% 13 21 Expected social support  All or most from family 66 67  All or most from friends 67 62  All or most from community 57 43 Source: Survey of Income and Program Participation: 2008; wave 4 Core and Topical Module, wave 6 Core and Topical Module; weighted using population weights at wave 6. Single mothers more frequently reported not paying the full amount of rent on time (19%) compared with two-parent households (11%). Liquid assets were relatively low across the sample, though two-parent households more frequently reported having savings in excess of $500 (23%) than did single mothers (8%). Single mothers also reported they had debt in their own name two times more often than did two-parent families. Single mothers reported more severe housing cost burden than did two-parent households. Twenty-one percent of single mothers reported paying 50% or more of their monthly income toward rent, compared with 13% of two-parent households. Social support from family, friends, or the community was relatively balanced among single mothers and two-parent families. Of note, single mothers reported less confidence in the ability to call on other sources of community support than did two-parent households. Multivariate Logistic Regression Table 2 presents the findings of the variables of interest (household type and social support) regressed on the late rent payment variable while controlling for covariates. The model was statistically significant (p = .001; Nagelkerke pseudo R2 = .118). Lack of each type of social support was significantly associated with making a late rent payment. Little to no support from family approached significance with a 32% increase in odds of paying rent late (odds ratio [OR] = 1.32; p < .10). Households that reported little to no expected support from friends were 92% more likely to report a late rent payment (OR = 1.92; p = .000). Finally, renters who expected little to no social support from other community resources were 141% more likely to also report a late rent payment (OR = 2.41; p = .000). Compared with two-parent families, single mothers’ reported expectations of social support and late rent payments approached significance (OR = 1.32, p < .10). Other covariates that were strongly associated with not paying the full amount of rent on time in the prior 12 months included lower educational attainment, holding debt in one’s own name, and having more children under 18 years of age in the household. Table 2: Logistic Regression Testing Main Effects of Household Type and Types of Social Support on Late Rent Payments (N = 1,919) Late Rent Payment (Model 1) Covariate β SE Odds Ratio Race/ethnicity (white non-Latino[a])  Black −.023 (.182) NS  Asian −.459 (.415) NS  Hawaiian/Pacific Islander/other .103 (.319) NS  Latino/a −.005 (.192) NS Education (less than high school)  High school or equivalent .236 (.209) NS  Some college −.384 (.275) NS  Diploma/certificate/associate’s .245 (.227) NS  College degree −.511 (.345) NS  Graduate or professional degree −.168** (.631) 0.185 Age .000 (.007) NS Number of children .142* (.060) 1.15 Monthly income—log −.088 (.083) NS Mental or physical disability .236 (.213) NS Received government benefits (no) −.138 (.198) NS Single mother (two-parent family) .278† (.165) 1.32 Debt (no) .422** (.160) 1.52 Liquid asset category (<$100)  $101 to $500 −.158 (.344) NS  >$500 −.339 (.260) NS Housing cost burden (≤30%)  31% to 49% .312 .197 NS  ≥50% .308 .222 NS Expected social support (all or most)  Little or none from family .280† (.164) 1.32  Little or none from friends .652*** (.167) 1.92  Little or none from community .879*** (.181) 2.41 Constant −2.56 (.768) p = .001 Pseudo (Nagelkerke) R2 .118 Late Rent Payment (Model 1) Covariate β SE Odds Ratio Race/ethnicity (white non-Latino[a])  Black −.023 (.182) NS  Asian −.459 (.415) NS  Hawaiian/Pacific Islander/other .103 (.319) NS  Latino/a −.005 (.192) NS Education (less than high school)  High school or equivalent .236 (.209) NS  Some college −.384 (.275) NS  Diploma/certificate/associate’s .245 (.227) NS  College degree −.511 (.345) NS  Graduate or professional degree −.168** (.631) 0.185 Age .000 (.007) NS Number of children .142* (.060) 1.15 Monthly income—log −.088 (.083) NS Mental or physical disability .236 (.213) NS Received government benefits (no) −.138 (.198) NS Single mother (two-parent family) .278† (.165) 1.32 Debt (no) .422** (.160) 1.52 Liquid asset category (<$100)  $101 to $500 −.158 (.344) NS  >$500 −.339 (.260) NS Housing cost burden (≤30%)  31% to 49% .312 .197 NS  ≥50% .308 .222 NS Expected social support (all or most)  Little or none from family .280† (.164) 1.32  Little or none from friends .652*** (.167) 1.92  Little or none from community .879*** (.181) 2.41 Constant −2.56 (.768) p = .001 Pseudo (Nagelkerke) R2 .118 Notes: Reference categories in parentheses; NS = not significant. Source: Survey of Income and Program Participation: 2008; wave 4 Core and Topical Module, wave 6 Core and Topical Modules; weighted using population weights at wave 6. †p < .10. *p < .05. **p < .01. ***p < .001. Table 2: Logistic Regression Testing Main Effects of Household Type and Types of Social Support on Late Rent Payments (N = 1,919) Late Rent Payment (Model 1) Covariate β SE Odds Ratio Race/ethnicity (white non-Latino[a])  Black −.023 (.182) NS  Asian −.459 (.415) NS  Hawaiian/Pacific Islander/other .103 (.319) NS  Latino/a −.005 (.192) NS Education (less than high school)  High school or equivalent .236 (.209) NS  Some college −.384 (.275) NS  Diploma/certificate/associate’s .245 (.227) NS  College degree −.511 (.345) NS  Graduate or professional degree −.168** (.631) 0.185 Age .000 (.007) NS Number of children .142* (.060) 1.15 Monthly income—log −.088 (.083) NS Mental or physical disability .236 (.213) NS Received government benefits (no) −.138 (.198) NS Single mother (two-parent family) .278† (.165) 1.32 Debt (no) .422** (.160) 1.52 Liquid asset category (<$100)  $101 to $500 −.158 (.344) NS  >$500 −.339 (.260) NS Housing cost burden (≤30%)  31% to 49% .312 .197 NS  ≥50% .308 .222 NS Expected social support (all or most)  Little or none from family .280† (.164) 1.32  Little or none from friends .652*** (.167) 1.92  Little or none from community .879*** (.181) 2.41 Constant −2.56 (.768) p = .001 Pseudo (Nagelkerke) R2 .118 Late Rent Payment (Model 1) Covariate β SE Odds Ratio Race/ethnicity (white non-Latino[a])  Black −.023 (.182) NS  Asian −.459 (.415) NS  Hawaiian/Pacific Islander/other .103 (.319) NS  Latino/a −.005 (.192) NS Education (less than high school)  High school or equivalent .236 (.209) NS  Some college −.384 (.275) NS  Diploma/certificate/associate’s .245 (.227) NS  College degree −.511 (.345) NS  Graduate or professional degree −.168** (.631) 0.185 Age .000 (.007) NS Number of children .142* (.060) 1.15 Monthly income—log −.088 (.083) NS Mental or physical disability .236 (.213) NS Received government benefits (no) −.138 (.198) NS Single mother (two-parent family) .278† (.165) 1.32 Debt (no) .422** (.160) 1.52 Liquid asset category (<$100)  $101 to $500 −.158 (.344) NS  >$500 −.339 (.260) NS Housing cost burden (≤30%)  31% to 49% .312 .197 NS  ≥50% .308 .222 NS Expected social support (all or most)  Little or none from family .280† (.164) 1.32  Little or none from friends .652*** (.167) 1.92  Little or none from community .879*** (.181) 2.41 Constant −2.56 (.768) p = .001 Pseudo (Nagelkerke) R2 .118 Notes: Reference categories in parentheses; NS = not significant. Source: Survey of Income and Program Participation: 2008; wave 4 Core and Topical Module, wave 6 Core and Topical Modules; weighted using population weights at wave 6. †p < .10. *p < .05. **p < .01. ***p < .001. Interaction Effects Tables 3 through 5 present the moderation effect of expected social support on the odds of a late rent payment within strata of household type. Table 3: Effect Moderation of Level of Expected Family Support on Family Composition Strata Some or All Expected Family Support Little or No Expected Family Support No Family Support versus Family Support within Family Composition Strata Variable RR [95% CI] RR [95% CI] RR [95% CI] Two parent 1 2.24 [1.44, 3.50]*** 2.39 [1.59, 3.60]*** Single mother 0.936 [0.608, 1.44] 3.55 [2.40, 5.27]*** 3.78 [2.51, 5.74]*** Multiplicative interaction 1.48 [0.844, 2.61], p = .171 Additive interaction (RERI) 1.33 [0.000, 2.07], p = .05 Some or All Expected Family Support Little or No Expected Family Support No Family Support versus Family Support within Family Composition Strata Variable RR [95% CI] RR [95% CI] RR [95% CI] Two parent 1 2.24 [1.44, 3.50]*** 2.39 [1.59, 3.60]*** Single mother 0.936 [0.608, 1.44] 3.55 [2.40, 5.27]*** 3.78 [2.51, 5.74]*** Multiplicative interaction 1.48 [0.844, 2.61], p = .171 Additive interaction (RERI) 1.33 [0.000, 2.07], p = .05 Notes: RR = relative risk; CI = confidence interval; RERI = relative excess risk due to interaction. ***p < .001. Table 3: Effect Moderation of Level of Expected Family Support on Family Composition Strata Some or All Expected Family Support Little or No Expected Family Support No Family Support versus Family Support within Family Composition Strata Variable RR [95% CI] RR [95% CI] RR [95% CI] Two parent 1 2.24 [1.44, 3.50]*** 2.39 [1.59, 3.60]*** Single mother 0.936 [0.608, 1.44] 3.55 [2.40, 5.27]*** 3.78 [2.51, 5.74]*** Multiplicative interaction 1.48 [0.844, 2.61], p = .171 Additive interaction (RERI) 1.33 [0.000, 2.07], p = .05 Some or All Expected Family Support Little or No Expected Family Support No Family Support versus Family Support within Family Composition Strata Variable RR [95% CI] RR [95% CI] RR [95% CI] Two parent 1 2.24 [1.44, 3.50]*** 2.39 [1.59, 3.60]*** Single mother 0.936 [0.608, 1.44] 3.55 [2.40, 5.27]*** 3.78 [2.51, 5.74]*** Multiplicative interaction 1.48 [0.844, 2.61], p = .171 Additive interaction (RERI) 1.33 [0.000, 2.07], p = .05 Notes: RR = relative risk; CI = confidence interval; RERI = relative excess risk due to interaction. ***p < .001. Table 4: Effect Moderation of Level of Expected Friend Support on Family Composition Strata Some or All Expected Friend Support Little or No Expected Friend Support No Friend Support versus Friend Support within Family Composition Strata Variable RR [95% CI] RR [95% CI] RR [95% CI] Two parent 1 2.24 [1.43, 3.50]*** 2.39 [1.59, 3.60]*** Single mother 0.936 [0.608, 1.44] 3.55 [2.40, 5.27] 1.48 [0.843, 2.61] Multiplicative interaction 1.48, (0.843, 2.61) Additive interaction (RERI) 1.33 (0.000, 2.67)* Some or All Expected Friend Support Little or No Expected Friend Support No Friend Support versus Friend Support within Family Composition Strata Variable RR [95% CI] RR [95% CI] RR [95% CI] Two parent 1 2.24 [1.43, 3.50]*** 2.39 [1.59, 3.60]*** Single mother 0.936 [0.608, 1.44] 3.55 [2.40, 5.27] 1.48 [0.843, 2.61] Multiplicative interaction 1.48, (0.843, 2.61) Additive interaction (RERI) 1.33 (0.000, 2.67)* Notes: RR = relative risk; CI = confidence interval; RERI = relative risk due to interaction. *p < .05. ***p < .001. Table 4: Effect Moderation of Level of Expected Friend Support on Family Composition Strata Some or All Expected Friend Support Little or No Expected Friend Support No Friend Support versus Friend Support within Family Composition Strata Variable RR [95% CI] RR [95% CI] RR [95% CI] Two parent 1 2.24 [1.43, 3.50]*** 2.39 [1.59, 3.60]*** Single mother 0.936 [0.608, 1.44] 3.55 [2.40, 5.27] 1.48 [0.843, 2.61] Multiplicative interaction 1.48, (0.843, 2.61) Additive interaction (RERI) 1.33 (0.000, 2.67)* Some or All Expected Friend Support Little or No Expected Friend Support No Friend Support versus Friend Support within Family Composition Strata Variable RR [95% CI] RR [95% CI] RR [95% CI] Two parent 1 2.24 [1.43, 3.50]*** 2.39 [1.59, 3.60]*** Single mother 0.936 [0.608, 1.44] 3.55 [2.40, 5.27] 1.48 [0.843, 2.61] Multiplicative interaction 1.48, (0.843, 2.61) Additive interaction (RERI) 1.33 (0.000, 2.67)* Notes: RR = relative risk; CI = confidence interval; RERI = relative risk due to interaction. *p < .05. ***p < .001. Table 5: Effect Moderation of Level of Expected Other Support on Family Composition Strata Some or All Expected Other Support Little or No Expected Other Support No Other Support versus Other Support within Family Composition Strata Variable RR [95% CI] RR [95% CI] RR [95% CI] Two parent 1 3.24 (1.90, 5.56)*** 2.12 [1.36, 3.31]** Single mother 1.53 (0.843, 2.81) 5.42 (3.29, 8.93)*** 3.54 [2.25, 5.58] *** Multiplicative interaction 2.59 (1.31, 4.50), p = .006 Additive interaction (RERI) 1.76 = 7 (0.683, 2.85), p = .001 Some or All Expected Other Support Little or No Expected Other Support No Other Support versus Other Support within Family Composition Strata Variable RR [95% CI] RR [95% CI] RR [95% CI] Two parent 1 3.24 (1.90, 5.56)*** 2.12 [1.36, 3.31]** Single mother 1.53 (0.843, 2.81) 5.42 (3.29, 8.93)*** 3.54 [2.25, 5.58] *** Multiplicative interaction 2.59 (1.31, 4.50), p = .006 Additive interaction (RERI) 1.76 = 7 (0.683, 2.85), p = .001 Notes: RR = relative risk; CI = confidence interval; RERI = relative risk due to interaction. **p < .01. ***p < .001. Table 5: Effect Moderation of Level of Expected Other Support on Family Composition Strata Some or All Expected Other Support Little or No Expected Other Support No Other Support versus Other Support within Family Composition Strata Variable RR [95% CI] RR [95% CI] RR [95% CI] Two parent 1 3.24 (1.90, 5.56)*** 2.12 [1.36, 3.31]** Single mother 1.53 (0.843, 2.81) 5.42 (3.29, 8.93)*** 3.54 [2.25, 5.58] *** Multiplicative interaction 2.59 (1.31, 4.50), p = .006 Additive interaction (RERI) 1.76 = 7 (0.683, 2.85), p = .001 Some or All Expected Other Support Little or No Expected Other Support No Other Support versus Other Support within Family Composition Strata Variable RR [95% CI] RR [95% CI] RR [95% CI] Two parent 1 3.24 (1.90, 5.56)*** 2.12 [1.36, 3.31]** Single mother 1.53 (0.843, 2.81) 5.42 (3.29, 8.93)*** 3.54 [2.25, 5.58] *** Multiplicative interaction 2.59 (1.31, 4.50), p = .006 Additive interaction (RERI) 1.76 = 7 (0.683, 2.85), p = .001 Notes: RR = relative risk; CI = confidence interval; RERI = relative risk due to interaction. **p < .01. ***p < .001. Expected Support of Family In Table 3, a two-parent family that expects some or all social support from family in a time of need is the reference group. In comparison with the reference group, there is a 139% increase in RR of being late on a rent payment for two-parent families who expect little to no social support (RR = 2.39 [CI = 1.59, 3.60], p < .001). For a single mother with little to no expected social support, there is a 278% increased RR of being late on a rent payment compared with the reference group (RR = 3.78 [CI = 2.51, 5.74], p < .001). The RERI measures the degree to which interaction term exceeds the impact of the two factors independently while keeping covariates constant. Here, the RERI equation is significant and indicates a 33% increase in RR of a late rent payment for single mothers without the social support of family (RERI = 1.33 [CI = .000, 2.07], p = .05). Expected Support of Friends Table 4 shows the effect moderation of friend support on family composition for the late rent payment outcome. Again, the two-parent family who can expect some or all support from friends is the reference group. In comparison with that reference group, a two-parent family who cannot turn to friends in a time of need faces a 139% in RR of a late rent payment (RR = 2.39 [CI = 1.59, 3.60], p < .001). Comparatively, single mothers’ RR of being late on a rent payment was not significant. Expected Support of Other Community Resources Two-parent households that expect little to no social support from other community resources (see Table 5) reported a 112% increased RR of a late rent payment in the last year when compared with those two-parent families who reportedly can expect some or all support from community resources in a time of need (RR = 2.12 [CI = 1.36, 3.31], p < .01). In addition, single mothers without the support of other community resources reportedly experience a 254% increase in RR of a late housing payment compared with the reference group (RR = 3.54 [CI = 2.25, 5.58], p < .001). The RERI score for this model is also significant, as the excess risk due to effect moderation is 76% (RERI = 1.76 [CI = .683, 2.85), p = .001]). Discussion The purpose of this research was to determine if a late rent payment was moderated by perceived instrumental social support for different household types. Data from two waves of SIPP were used; both logistic regression and moderation effect testing were conducted with a nationally representative sample of single mothers and heads of two-parent households (N = 1,919). The moderator of instrumental social support was tested on three different levels: perceived support from family, friends, and other community supports. The hypothesis was that social support would moderate the relationship of household type to a late rent payment. In other words, single mothers who perceived little to no social support would be more likely to report a late rent payment in the prior year than a two-parent family with any level of social support or a single mother who did perceive that some or all social support she needed would be provided. Results suggest that perceived instrumental social support may be an important protective factor for single mothers to maintain housing stability. According to RERI values, the interaction of household type and level of different social support left single mothers 33% more likely to have been late on a rent payment in the past year when they expected to receive little to no social support from their families. If they felt they could expect little to no help from other people or social services in the community, single mothers were 76% more likely to have been late on a rent payment in the past year. These findings provide new knowledge regarding the importance of social support for maintaining housing stability. From prior empirical work, we know that single mothers are more likely to report late rent payment than two-parent families or those headed by a single father (West, 2016). Social support is related to actual and perceived hardship avoidance (Henly et al., 2005; Mills & Zhang, 2013). Social support is related to reduced welfare receipt and improved employment and earnings among single mothers (Harknett, 2006), and lower-income single mothers with complex needs tend to have less perceived social support (Harknett & Hartnett, 2011). Now there is empirical evidence that suggests a significant relationship between social support and housing stability. Qualitative work on social capital in lower-resource communities adds important context to these findings. Of note, this analysis suggests that whereas the instrumental social support of family or other community resources is associated with reduced risk of late rent payments, the instrumental support of friends appears to have no significant relationship. Some mothers may effectively leverage relationships among friends or fictive kin to help make ends meet; these relationships tend to be weaker than familial ones and provide a specific resource (for example, vehicle repair, child care) that frequently requires reciprocity (McLanahan, Wedemeyer, & Adelberg, 1981). Desmond (2012a) has characterized some social ties, especially those in the context of extreme material deprivation, as “disposable.” In these relationships, individuals in extreme poverty strike up intense and sudden relationships to collectively afford housing or food. Although the majority are short-lived, many relationships do endure the ebb and flow of making ends meet for years. As such, there may be a few plausible explanations for why two-parent families who have friends to call on in a time of need tend to not miss rent payments, but there seems to be no similar effect for single mothers. First, two-parent households likely have more material or instrumental resources to exchange in these relationships, thus maintaining them through reciprocity. Second, the friends and extended networks of two-parent households are likely of similar socioeconomic background and have more financial resources than do single mothers and their friends or extended networks (Granovetter, 1973). Finally, single mothers in the sample have lower incomes and lower indicators of social capital than do two-parent families. Their relationships with friends and extended networks may be more appropriately considered disposable ties instead of simply weak ones, suggesting that there may be periods when close, reliable friends are absent from their lives (Desmond, 2012a; McLanahan et al., 1981). As Dominguez and Watkins (2003) explained, leveraging close family ties is often a first line of defense for negotiating the competing demands of the market economy and motherhood. Mothers may call on a sister or an aunt to help with child care or an overdue bill; these relationships, though often reliable and durable, may become tenuous due to role expectations and violations, as well as the emotional labor and discord frequent in intimate family relationships. Struggling families may altogether avoid asking better-resourced family members for help due to embarrassment or the anticipation of judgment (Desmond, 2012a). In these cases, social services agencies may be the preferred lifeline to help make ends meet; this may help explain the dramatic differences in RR of a late rent payment when single mothers reportedly receive little to no help from social services or other community support. Specifically, the RR of being late on a rent payment for a single mother who indicates little to no support from a social services agency is 76% compared with a similarly situated two-parent household. This RR is over double the amount of RR of a late rent payment when a single mother cannot count on her family for help, suggesting that not receiving help from what may be the last line of defense is related to substantial consequences for housing stability. It is important to note that the protective relationship between social support and housing stability may be functioning through the indirect material value of instrumental support rather than direct financial help from friends, family, or the community. There is little possibility of receiving financial support from family and friends for a lower-resourced family, as family and friends are also likely struggling to make ends meet (Henly et al., 2005; Meadows, 2009). These same social networks, however, may be able to offer instrumental support that has indirect material value. For example, one item on the survey asked mothers whether they could turn to family, friends, or other people in the community if they were sick. They could likely also turn to one of those supportive neighbors to watch a sick child to avoid missing work. The capacity to leverage support in a time of need essentially prevents the income lost from a missed workday or the unexpected expense of having to pay an at-home babysitter. It translates to more cash on hand for a single mother, which may then translate to enough cash on hand each month to pay bills on time. The ability to draw on these sources of support has been and continues to be essential for the survival and financial well-being of single mothers and their children (Henly, 2002; Stack, 1974). By understanding how instrumental social support operates as a financially protective factor for families, social workers may help leverage existing support systems and help create spaces to nurture new relationships between families, their neighbors, and their communities. Geens and Vandenbroeck (2014) noted that there is little research on social support and parenting in social work literature that does not suffer from either sorting clients into “risk groups” or focusing primarily on parental health. Furthermore, there is a concerning dearth of research on creating or preserving social support systems among the clients we serve. This lack of information on the “how” of building and maintaining informal social support systems that are, perhaps, place-based instead of based on a group facing a perceived risk, represents an important opportunity for social work research and practice. Limitations These findings and implications should be interpreted with consideration of several limitations. First, selection bias, which occurred as only respondents who completed waves 4 and 6 and maintained their housing type for the duration were filtered into the subsample for analyses, which likely compromised the effects of complex random sampling used by SIPP (U.S. Census Bureau, 2008). Those individuals who were evicted or moved into another housing type from wave 4 to wave 6 likely would have reported higher rates of missed rent payments; thus, it is important to note that we can only interpret these findings as conservative and applicable only to those who maintained their rental housing for the observation period. Second, two variables, housing cost burden and liquid assets, were once continuous and were transformed to categorical variables for analyses. In the logistic regression that already had a binary outcome variable, variance in the model is decreased and there is potential for an increase in a Type I error (Austin & Brunner, 2004). In other words, the observed relationship between increased risk of a late rent payment and social support tested in the models may have been a false positive. Third, the inability to control for variables that have demonstrated relationships to the outcome variable (experiences of interpersonal violence and receipt of a housing voucher) raise concern of endogeneity (Stone & Rose, 2011). Finally, the direction of relationships between the variables could not have been determined due to the design of the 2008 SIPP. All demographic and financial variables were measured at wave 4. Nine months later, when responding to wave 6, participants were asked if they had missed a rent or mortgage payment in the prior 12 months. As such, a missed rent payment could have occurred in the three months before collection of the financial instability variables. This complication of data collection, as well as the nature of the research design, impedes the ability to establish whether the independent variables preceded or followed the dependent variable tested. Conclusion Single mothers continue to face considerable economic inequality and may be disproportionately affected by the pervasive and growing lack of affordable housing (Steffen et al., 2015). Considered in tandem with a political environment that has the potential to exacerbate income and asset insufficiency for already struggling households, it is reasonable to expect that single mothers’ inabilities to meet the basic needs for their families will only worsen (Floyd et al., 2017; Williams, 2017). Prior research indicated that single mothers’ social support systems may help mitigate the day-to-day financial struggles of parenthood. However, there was no empirical evidence of how social support may relate to housing instability for single mothers and whether this relationship was more or less pronounced in comparison with two-parent households. When compared with similarly situated two-parent families, single mothers who report little to no perceived social support from family or other community resources also reported significantly higher RR of a late housing payment in the past year. Though these findings should be considered in light of the previously mentioned limitations, they present notable implications for social work practice and research. Social workers frequently help lower-income clients cope with fragile household balance sheets and are increasingly assisting clients to navigate financial institutions and resources to build security and wealth (Despard & Chowa, 2010). Threats to funding for programs that build financial stability implore social workers to not only find solutions to mitigate material deprivation in the policy sphere, but also do so in the near-term through helping to build or maintain social connections that are indispensable for everyday survival. Stacia Martin-West, PhD, is assistant professor, College of Social Work, University of Tennessee, 193 Polk Avenue, Nashville, TN; e-mail: [email protected] References Acs , G. , Loprest , P. , & Nichols , A. ( 2009 ). Risk and recovery: Documenting the changing risks to family incomes. Retrieved from http://www.urban.org/UploadedPDF/411890_risk_and_recovery.pdf Aurand , A. , Emmanuel , D. , Yentel , D. , Errico , E. , Gaby-Biegel , J. , & Kerr , E. ( 2018 ). Out of reach. Retrieved from http://nlihc.org/sites/default/files/oor/OOR_2018.pdf Austin , P. C. , & Brunner , L. J. ( 2004 ). Inflation of the type I error rate when a continuous confounding variable is categorized in logistic regression analyses . Statistics in Medicine, 23 , 1159 – 1178 . Google Scholar Crossref Search ADS Bollinger , C. , & Ziliak , J. ( 2007 ). Welfare reform and the level, composition, and volatility of income. Retrieved from http://www.ukcpr.org/SeminarSeries/Ziliak-Bollinger_WelfareReformandVolatility_032007.pdf Brisson , D. , & Covert , J. ( 2015 ). Housing instability risk among subsidized housing recipients: Characteristics associated with late or nonpayment of rent . Social Work Research, 39 , 119 – 128 . Google Scholar Crossref Search ADS Casey , T. ( 2011 ). Single mothers since 2000: Falling further down. Retrieved from http://www.legalmomentum.org/sites/default/files/reports/single-mothers-since-2000.pdf de Souza Briggs , X. ( 1998 ). Brown kids in white suburbs: Housing mobility and the many faces of social capital . Housing Policy Debate, 9 , 177 – 221 . Google Scholar Crossref Search ADS Desmond , M. ( 2012 a). Disposable ties and the urban poor . American Journal of Sociology, 117 , 1295 – 1335 . Google Scholar Crossref Search ADS Desmond , M. ( 2012 b). Eviction and the reproduction of urban poverty 1 . American Journal of Sociology, 118 , 88 – 133 . Google Scholar Crossref Search ADS Desmond , M. ( 2014 ). Poor black women are evicted at alarming rates, setting off a chain of hardship . Chicago : MacArthur Foundation . Desmond , M. ( 2015 ). Unaffordable America: Poverty, housing, and eviction . Fast Focus, 22 , 1 – 6 . Despard , M. , & Chowa , G. ( 2010 ). Social workers’ interest in building individuals’ financial capabilities . Journal of Financial Therapy, 1 ( 1 ), 8 . doi:10.4148/jft.v1i1.257 Google Scholar Crossref Search ADS Dominguez , S. , & Watkins , C. ( 2003 ). Creating networks for survival and mobility: Social capital among African-American and Latin-American low-income mothers . Social Problems, 50 ( 1 ), 111 – 135 . Google Scholar Crossref Search ADS Edin , K. J. , & Shaefer , H. L. ( 2015 ). $2.00 a day: Living on almost nothing in America . New York : Houghton Mifflin Harcourt . Financial Industry Regulatory Authority (FINRA) Foundation . ( 2016 ). Financial capability in the United States 2016. Retrieved from http://www.usfinancialcapability.org/downloads/NFCS_2015_Report_Natl_Findings.pdf Floyd , I. , Pavetti , L. , & Schott , L. ( 2017 ). TANF reaching few poor families. Retrieved from https://www.cbpp.org/research/family-income-support/tanf-continues-to-weaken-as-a-safety-net Geens , N. , & Vandenbroeck , M. ( 2014 ). The (ab)sense of a concept of social support in parenting research: A social work perspective . Child & Family Social Work, 19 , 491 – 500 . Google Scholar Crossref Search ADS Goldberg , E. ( 2014 ). These most vulnerable groups are suffering the most because of income inequality. Huffington Post. Retrieved from http://www.huffingtonpost.com/2014/01/29/sotu-income-inequality_n_4682553.html Granovetter , M. ( 1973 ). The strength of weak ties . American Journal of Sociology, 78 , 1360 – 1380 . Google Scholar Crossref Search ADS Harknett , K. ( 2006 ). The relationship between private safety nets and economic outcomes among single mothers . Journal of Marriage and Family, 68 , 172 – 191 . Google Scholar Crossref Search ADS Harknett , K. S. , & Hartnett , C. S. ( 2011 ). Who lacks support and why? An examination of mothers’ personal safety nets . Journal of Marriage and Family, 73 , 861 – 875 . Google Scholar Crossref Search ADS Henly , J. ( 2002 ). Informal support networks and the maintenance of low-wage jobs. In F. Munger (Ed.), Laboring below the line: The new ethnography of poverty, low-wage work, and survival in the global economy (pp. 179 – 203 ). New York : Russell Sage Foundation . Henly , J. R. , Danziger , S. K. , & Offer , S. ( 2005 ). The contribution of social support to the material well-being of low-income families . Journal of Marriage and Family, 67 , 122 – 140 . Google Scholar Crossref Search ADS Henly , J. R. , & Lyons , S. ( 2000 ). The negotiation of child care and employment demands among low-income parents . Journal of Social Issues, 56 , 683 – 705 . doi:10.1111/0022-4537.00191 Google Scholar Crossref Search ADS Hosmer , D. W. , & Lemeshow , S. ( 1992 ). Confidence interval estimation of interaction . Epidemiology, 3 , 452 – 456 . Google Scholar Crossref Search ADS Joint Center for Housing Studies . ( 2017 ). America’s rental housing. Retrieved from http://www.jchs.harvard.edu/research-areas/reports/americas-rental-housing-2017 Kalil , A. , & Ryan , R. M. ( 2010 ). Mothers’ economic conditions and sources of support in fragile families . Future of Children, 20 ( 2 ), 39 – 61 . Google Scholar Crossref Search ADS Leonard , T. , & Di , W. ( 2014 ). Is household wealth sustainable? An examination of asset poverty reentry after an exit . Journal of Family and Economic Issues, 35 , 131 – 144 . Google Scholar Crossref Search ADS Letiecq , B. L. , Anderson , E. A. , & Koblinsky , S. A. ( 1998 ). Social support of homeless and housed mothers: A comparison of temporary and permanent housing arrangements . Family Relations, 47 , 415 – 421 . Google Scholar Crossref Search ADS Lusardi , A. , Schneider , D. J. , & Tufano , P. ( 2011 ). Financially fragile households: Evidence and implications (No. w17072). Cambridge, MA : National Bureau of Economic Research . Retrieved from http://www.nber.org/papers/w17072 McKernan , S. M. , Ratcliffe , C. , & Vinopal , K. ( 2009 ). Do assets help families cope with adverse events? Washington, DC : Urban Institute . Retrieved from http://www.urban.org/uploadedpdf/411994_help_family_cope.pdf McLanahan , S. S. , Wedemeyer , N. V. , & Adelberg , T. ( 1981 ). Network structure, social support, and psychological well-being in the single-parent family . Journal of Marriage and Family, 43 , 601 – 612 . Google Scholar Crossref Search ADS Meadows , S. O. ( 2009 ). Is it there when you need it? Mismatch in perception of future availability and subsequent receipt of instrumental social support . Journal of Family Issues, 30 , 1070 – 1097 . Google Scholar Crossref Search ADS Mills , G. , & Amick , J. ( 2010 ). Can savings help overcome income instability? Washington, DC : Urban Institute . Retrieved from http://www.urban.org/UploadedPDF/412290-savings-overcome-income-instability.pdf Mills , G. , & Zhang , S. ( 2013 ). Social support networks and their effects on hardship avoidance among low-income households. Retrieved from http://www.urban.org/sites/default/files/publication/22201/412991-Social-Support-Networks-and-Their-Effects-on-Hardship-Avoidance.PDF National Low Income Housing Coalition . ( 2013 , February). America’s affordable housing shortage, and how to end it . Housing Spotlight, 3 ( 2 ). Retrieved from http://nlihc.org/sites/default/files/HS_3-1.pdf National Low Income Housing Coalition . ( 2015 ). 2015-2016 public policy agenda. Retrieved from nlihc.org/sites/default/files/Sec1.02_Policy-Agenda_2015.pdf National Women’s Law Center . ( 2016 ). National snapshot: Poverty among women & families, 2015. Retrieved from http://nwlc.org/resources/national-snapshot-poverty-among-women-families-2015/ Proctor , B. D. , Semega , J. L. , & Kollar , M. A. ( 2016 ). Income and poverty in the United States: 2015 . Washington, DC : U.S. Census Bureau . Retrieved from https://www.census.gov/library/publications/2016/demo/p60-256.html Shaefer , L. ( 2013 ). The Survey of Income and Program Participation (SIPP): Critical issues for data analysis using the SIPP. Retrieved from https://www.census.gov/content/dam/Census/programs-surveys/sipp/about/Shaefer-SIPP-Lecture%203.pdf Sherraden , M. S. ( 2013 ). Building blocks of financial capability. In J. Birkenmaier , M. S. Sherraden , & J. Curley (Eds.), Financial education and capability: Research, education, policy, and practice (pp. 3 – 43 ). New York : Oxford University Press . Stack , C. ( 1974 ). All our kin: Strategies for survival in a black community . New York : Harper & Row . Stata SE (Version 14) [Computer software] . ( 2015 ). Retrieved from https://www.stata.com/stata14/ Steffen , B. , Carter , G. , Martin , M. , Pelletiere , D. , Vandenbroucke , D. , & Yao , Y. ( 2015 ). Worst case housing needs: 2015 Report to Congress. Retrieved from https://www.huduser.gov/portal/publications/affhsg/wc_HsgNeeds15.html Stone , S. I. , & Rose , R. A. ( 2011 ). Social work research and endogeneity bias . Journal of the Society for Social Work and Research, 2 ( 2 ), 54 – 75 . Google Scholar Crossref Search ADS U.S. Census Bureau . ( 2008 ). 2008 SIPP user’s guide. Retrieved from http://www.census.gov/programs-surveys/sipp/methodology/users-guide.html VanderWeele , T. J. , & Knol , M. J. ( 2014 ). A tutorial on interaction . Epidemiologic Methods, 3 ( 1 ), 33 – 72 . Wang , K. , & Tran , T. ( 2017 , January). Analyzing and presenting additive and multiplicative interaction effects between categorical variables. Oral presentation at annual meeting of Society for Social Work and Research, New Orleans. West , S. ( 2016 ). Associations between financial inclusion, liquid assets, income shocks, and later housing instability in households headed by a single mother. Retrieved from http://hdl.handle.net/1808/21552 Williams , J. H. ( 2017 ). Political landscape: The times they are a-changing [Editorial] . Social Work Research, 41 , 3 – 5 . Google Scholar Crossref Search ADS © 2018 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)

Hodge, David, R;Hong, Philip Young, P;Choi,, Sangmi

doi: 10.1093/swr/svy034pmid: N/A

Abstract Interest in the construct of grit has increased across disciplines due to its ability to predict success in a wide variety of settings. Grit is a particularly important construct among disadvantaged populations, who typically must overcome a disproportionate number of obstacles to reach their goals. This study develops and tests a model of grit among one such population: underemployed urban African Americans. The study sample consists of 1,045 consecutive attendees at a two-week job readiness training program in a large urban area. The results of the structural equation modeling indicate that spirituality has a direct, positive effect on grit and that this relationship is partially mediated by employment hope. The results suggest that spirituality and employment hope are protective factors that may be leveraged in practice settings to potentially enhance grit. The psychological concept of grit can be understood as a combination of passion and perseverance directed toward achieving an important goal (Duckworth, 2016). As such, grit is an important determinant of success in many settings (Hammond, 2017). Grit enables people to overcome barriers and obstacles that stand in the way of successful outcomes. Grit is often an especially salient factor among disadvantaged populations (Griffin, McDermott, McHugh, Fitzmaurice, & Weiss, 2016; Guerrero, Dudovitz, Chung, Dosanjh, & Wong, 2016). These groups typically face more challenges relative to other, more advantaged populations. In the face of such environmental constraints, the importance of grit in surmounting obstacles becomes more pronounced. Yet, despite its importance, little research on grit has been conducted among such populations (Hammond, 2017; Wolters & Hussain, 2015). To address this gap in the literature, we developed and tested a model of grit among a sample of primarily urban African Americans who are either underemployed or unemployed. Literature Review Grit is a strengths-based, noncognitive construct that has recently emerged from the positive psychology movement. Most of the early research on the topic was conducted in education. Duckworth (2016), a key popularizer of the concept, observed that grit is often a better predictor of academic success than more traditional measures, such as cognitive ability. Compared with their counterparts, gritty students exhibit greater persistence in pursuit of their academic goals, overcoming setbacks and distractions. For instance, grit has been shown to play a significant role in predicting academic success among black college students at predominately white institutions (Strayhorn, 2014). Over time, the stick-to-itiveness results in better scholastic performance. In addition to academic achievement, the merit of grit has been validated in other areas. For instance, men with higher levels of grit are more likely to keep sales jobs, complete military Special Forces training, and stay married (Eskreis-Winkler, Duckworth, Shulman, & Beal, 2014). Grittier novice teachers in underresourced public schools recorded better end-of-year student performance on academic tests relative to their less gritty peers (Duckworth, Quinn, & Seligman, 2009). Grittiness has also been linked to lower levels of attrition among medical residents (Burkhart, Tholey, Guinto, Yeo, & Chojnacki, 2014; Salles, Cohen, & Mueller, 2014), higher engagement in intense exercise (Reed, Pritschet, & Cutton, 2013), and a lower likelihood of engaging in substance use and delinquent behaviors (Guerrero et al., 2016). In short, the extant research suggests grit is a key construct that enables people to achieve success by overcoming adversity. Given the effectiveness of grit, researchers have sought to unpack and understand its correlates (Von Culin, Tsukayama, & Duckworth, 2014). For instance, studies have examined variables that predict grit. Perhaps particularly important in this area are studies that feature populations that face challenges. Examples of such work include studies featuring samples comprised of Latino college students (Vela, Lu, Lenz, & Hinojosa, 2015), people with substance use disorders (Griffin et al., 2016), and African American college students (Yates et al., 2015). Such research has important real-life implications. Specifically, it has the potential to help members of these groups persist in moving toward their goals despite the many obstacles they frequently encounter. One population for whom additional research is warranted is unemployed African Americans. According to the Bureau of Labor Statistics (2017), the unemployment rate among African Americans (8.4%) is significantly higher than the rate among Latinos (5.8%), white Americans (4.3%), and Asians (3.6%). Grit may play an especially salient role among African Americans in urban settings who are under- or unemployed, because finding appropriate employment in such settings can be more challenging (Bolland, Lian, & Formichella, 2005). Although job training programs can assist some participants to obtain employment in urban locations, evidence suggests that they are not universally successful (U.S. Department of Labor, 2014). As alluded to earlier, this is due to the barriers African Americans face when seeking and finding appropriate employment in urban environments (Bonilla-Silva, 2018). Specific barriers that tend to exist in these settings include lack of employment opportunities, comparatively few jobs with good wages and benefits, and employer discrimination (Gibbs & Bankhead, 2000). In turn, these obstacles can engender a sense of hopelessness that further impedes people’s ability to secure appropriate employment (Bolland et al., 2005; Gibbs & Bankhead, 2000). This suggests that understanding how grit functions among urban African Americans is critical to achieving positive outcomes. In impoverished urban settings, grit can play an instrumental role in helping job seekers push through the structural and psychological challenges that exist. Spirituality as a Potential Predictor of Grit Spirituality is commonly understood in terms of an individual’s relationship with God (Wuthnow, 2007), although it is important to note that many other views of spirituality exist (Canda & Furman, 2010). According to Gallup data, the overwhelming majority of African Americans report belief in God (Newport, 2012). Indeed, according to some measures, African Americans are the most spiritually animated population in the nation. A substantial body of research links spirituality with a diverse array of positive health outcomes (Koenig, King, & Carson, 2012). To explain these salutary findings, theorists have noted that spirituality is a relational construct that can shape understandings of reality (Granqvist & Kirkpatrick, 2013). Individuals motivated by their spirituality tend to perceive events through a distinct cognitive lens. In turn, this lens engenders positive health outcomes. This line of theoretical reasoning suggests that spirituality is positively associated with grit (Yates et al., 2015). More specifically, spirituality may engender grit through creating narratives that frame events from a transcendent perspective (Pargament, 2007). Common transcendent scripts that may foster grit include the belief that God has a plan for one’s life, that he cares for people, and that he will not allow more challenges than one can bear (Naeem et al., 2015). These perceptions may directly affect grit by providing spiritually motivated people with a transcendent rationale to persist in the face of difficult circumstances. Indeed, this line of thought is reflected in prior qualitative research featuring African American college students (Yates et al., 2015). Respondents cited the importance of such beliefs in the course of articulating how spirituality helped enhance levels of grit. Hope as a Potential Mediator In broad relief, hope can be understood as a positive psychological state that relates to obtaining expected outcomes (Snyder, Lopez, & Pedrotti, 2011). Put simply, if people hope for something, they expect certain outcomes to be realized in the future. Employment hope is essential to the process of obtaining employment in workforce development. It can be viewed as a component of psychological self-sufficiency that, in turn, enables individuals to design and implement plans to overcome perceived barriers and achieve their goals (Hong, 2013; Hong, Choi, & Key, 2018; Hong, Choi, & Polanin, 2014). For mediation to occur, hope must be related to both spirituality and grit. Regarding the former relationship, at least 40 studies have examined the relationship between spirituality and hope (Koenig et al., 2012). Of these, the vast majority reveal a significant positive relationship between the two constructs. In a sample of underemployed and unemployed urban African American job seekers, hope was found to mediate the relationship between spirituality and economic self-sufficiency (Hong, Hodge, & Choi, 2015). These findings may be explained by the same transcendent scripts discussed in the preceding section. Such scripts—for example, believing that God has a benevolent plan for one’s future—are likely to foster perceptions of hope for the future. Prior research has also documented a significant association between hope, in tandem with other forms of self-efficacy, and grit (Guerrero et al., 2016; Vela et al., 2015). For instance, in a sample of Latino college students, hope was the strongest predictor of grit among the examined variables (Vela et al., 2015). Such findings are theoretically plausible. The ability to persevere through challenging circumstances is likely enhanced by the belief that positive outcomes will be realized in the future. Hypotheses The preceding reasoning can be summarized in the form of four testable hypotheses (H). More specifically, the extant research and theory suggest that spirituality has a positive influence on employment hope (H1); spirituality has a positive influence on grit (H2); employment hope has a positive influence on grit (H3); and spirituality has a positive indirect influence on grit, mediated through employment hope (H4). Figure 1 provides a graphical depiction of these relationships in the form of a testable model. Figure 1: View largeDownload slide Mediation Model with Standardized Parameter Estimates Figure 1: View largeDownload slide Mediation Model with Standardized Parameter Estimates The present study tested these four hypotheses with a sample consisting primarily of urban African Americans seeking suitable employment. To the best of our knowledge, no previous studies on grit have focused on this population. This represents a significant gap in the literature in light of the potential utility of grit in overcoming the challenges that urban African Americans encounter when obtaining employment. Method Data Collection and Sample Characteristics The study sample consisted of 1,045 adults participating in a two-week job readiness training (JRT) program at the Chicago Urban League (CUL). CUL is a community-based agency that works for economic, educational, and social progress for African Americans. CUL promotes strong, sustainable communities through progressive advocacy, effective collaboration, and innovative programming. Its mission is to (1) prepare people to work at all strata in a continually changing economy, (2) develop engaged citizens, and (3) build strong families. In keeping with this ethos, CUL’s JRT program serves individuals who are out of, and looking for, work. The research was conducted with the approval of a university institutional review board. The survey was administered by the research team at CUL’s orientation session and the first day of the JRT program. To help mitigate selection bias, all program attendees from 2012 through 2015 were invited to participate in the study. Essentially all participants completed the survey (the refusal rate was <1%). The survey took approximately 30 to 40 minutes to complete. The sample was primarily African American (93.4%) and male (58.0%) (see Table 1). A majority of the respondents were unemployed (83.6%) and had participated in a job training program within the last 10 years (74.8%). Most (59.5%) reported currently receiving Temporary Assistance for Needy Families or other welfare benefits. Table 1: Demographic Characteristics (N = 1,045) Characteristic N % Gender  Male 574 58.0  Female 416 42.0 Age group (M, SD) (Min.–Max.) (37.87, 12.99) (17–78) Race  African American 987 93.4  Non–African American 65 6.6 Education level  Less than high school 61 6.2  High-school/GED 268 27.2  Some college but no degree 308 31.2  Diploma/certificate from technical, vocational, or trade school 131 13.3  Above associate degree 219 22.2 Employment status  Employed 137 16.4  Unemployed 699 83.6 Job training participation  Yes 650 74.8  No 219 25.2 Receipt of welfare benefits/TANF  Yes 539 59.5  No 367 40.5 Characteristic N % Gender  Male 574 58.0  Female 416 42.0 Age group (M, SD) (Min.–Max.) (37.87, 12.99) (17–78) Race  African American 987 93.4  Non–African American 65 6.6 Education level  Less than high school 61 6.2  High-school/GED 268 27.2  Some college but no degree 308 31.2  Diploma/certificate from technical, vocational, or trade school 131 13.3  Above associate degree 219 22.2 Employment status  Employed 137 16.4  Unemployed 699 83.6 Job training participation  Yes 650 74.8  No 219 25.2 Receipt of welfare benefits/TANF  Yes 539 59.5  No 367 40.5 Note: TANF = Temporary Assistance for Needy Families. Table 1: Demographic Characteristics (N = 1,045) Characteristic N % Gender  Male 574 58.0  Female 416 42.0 Age group (M, SD) (Min.–Max.) (37.87, 12.99) (17–78) Race  African American 987 93.4  Non–African American 65 6.6 Education level  Less than high school 61 6.2  High-school/GED 268 27.2  Some college but no degree 308 31.2  Diploma/certificate from technical, vocational, or trade school 131 13.3  Above associate degree 219 22.2 Employment status  Employed 137 16.4  Unemployed 699 83.6 Job training participation  Yes 650 74.8  No 219 25.2 Receipt of welfare benefits/TANF  Yes 539 59.5  No 367 40.5 Characteristic N % Gender  Male 574 58.0  Female 416 42.0 Age group (M, SD) (Min.–Max.) (37.87, 12.99) (17–78) Race  African American 987 93.4  Non–African American 65 6.6 Education level  Less than high school 61 6.2  High-school/GED 268 27.2  Some college but no degree 308 31.2  Diploma/certificate from technical, vocational, or trade school 131 13.3  Above associate degree 219 22.2 Employment status  Employed 137 16.4  Unemployed 699 83.6 Job training participation  Yes 650 74.8  No 219 25.2 Receipt of welfare benefits/TANF  Yes 539 59.5  No 367 40.5 Note: TANF = Temporary Assistance for Needy Families. Measures Spirituality. The independent latent construct—spirituality—was measured using the Intrinsic Spirituality Scale (ISS) (Hodge, 2003). The six-item ISS is designed to assess the degree to which spirituality functions as a source of intrinsic motivation. The ISS measures both theistic and nontheistic forms of spirituality, irrespective of whether the spiritual impulse is expressed inside or outside of a religious context. The ISS uses the phrase completion methodology in which respondents complete a phrase on an 11-point response key in which zero denotes the absence of the construct in question and 10 denotes a theorized maximum (Hodge & Gillespie, 2003). A sample item is, “In terms of the questions I have about life, my spirituality answers . . .” (with responses ranging from 0 = no questions to 10 = absolutely all my questions). A Cronbach’s alpha of .801 was obtained in the present study for the ISS. Employment Hope. The mediating latent construct, employment hope, was assessed with the Employment Hope Scale (EHS) (Hong et al., 2014). The scale consists of 14 items that assess hope in the context of seeking employment. More specifically, the questions tap the degree of psychological empowerment or confidence in moving toward achieving job-related goals—that is, I am worthy of working in a good job. Respondents rated each statement on an 11-point scale ranging from 0 = not at all to 10 = all the time. In the current study, a Cronbach’s alpha of .936 was obtained for the EHS. Grit. The dependent latent construct, grit, was measured with the Short Grit Scale (Grit-S) (Duckworth & Quinn, 2009). The eight-item Grit-S measures trait-level perseverance and passion for long-term goals. The Grit-S conceptualizes grit as a two-factor construct: (1) consistency of interest (4 items) and (2) perseverance of effort (4 items). Each item uses a five-point Likert response key (ranging from 1 = not like me to 5 = very much like me). A Cronbach’s alpha of .795 for the Grit-S was recorded in the present study. Data Analysis Preliminary analysis was conducted using SPSS 20.0. Less than 2% of cases were missing data across variables, a relatively inconsequential level of missingness (Kline, 2016). To retain the full sample size, missing data were imputed using the expectation–maximization algorithm procedure (Schafer & Graham, 2002). Structural equation modeling (SEM) was performed with AMOS 20.0. Relative to procedures such as ordinary least squares regression, SEM is better suited to testing complex mediation models while simultaneously taking into account measurement error (Schumacker & Lomax, 2016). To test the study hypotheses, a traditional two-step approach was used (Anderson & Gerbing, 1988). In the first step, a measurement model was constructed to ensure that the observed indicators measure the latent construct with some degree of accuracy. After establishing the validity of the measurement model using confirmatory factor analysis, we proceeded to examine the subsequent structural model. Maximum likelihood estimation was used for estimation (Schumacker & Lomax, 2016). Bootstrapping was used to estimate the mediation effect of employment hope (Dearing & Hamilton, 2006; Hayes, 2009). To assess the fit of each model, three different methods were used (that is, incremental, parsimonious, and absolute). More specifically, the following three indices were used: Comparative Fit Index (CFI), root mean square error of approximation (RMSEA) with 90% confidence intervals (CIs), and standardized root mean square residual (SRMR). In addition to providing three different approaches to assess model fit, these indices are also among the most commonly recommended indices (Byrne, 2016; Garson, 2012; Mueller & Hancock, 2008; West, Taylor, & Wu, 2012). Results Treatment of Control Variables Prior research has investigated the relationship between grit and numerous variables including demographic traits (Duckworth & Quinn, 2009). Regarding the latter, research suggests grit may be linked with age and education (Bailey, Jenkins, & Leinbach, 2005; Duckworth, Peterson, Matthews, & Kelly, 2007). Accordingly, the subsequent structural model controlled for the effects of age and education level. Different approaches to handling control variables in SEM have appeared in the literature (Fletcher, Selgrade, & Germano, 2006; Kammeyer-Mueller & Wanberg, 2003). In the present study, the approach recommended by Mueller and Hancock (2010) was used. Specifically, the control variables were allowed to covary as exogenous predictors within the model. To simplify the presentation of the final model, these relationships were not presented in Figure 1. Descriptive Analysis The means and standard deviations for the study variables are presented in Table 2. In addition, bivariate analyses are also reported. As expected, all the study variables—spirituality, employment hope, and grit—were positively correlated with each other. Table 2: Descriptive and Bivariate Statistics for the Study Variables (N = 1,045) Variable M (SD) Range 1 2 3 1. Spirituality 6.93 (2.20) 0.00–10.00 (.801)a 2. Employment hope 8.94 (1.42) 0.00–10.00 .192** (.936)a 3. Grit 3.77 (0.92) 1.00–5.00 .146** .191** (.795)a Variable M (SD) Range 1 2 3 1. Spirituality 6.93 (2.20) 0.00–10.00 (.801)a 2. Employment hope 8.94 (1.42) 0.00–10.00 .192** (.936)a 3. Grit 3.77 (0.92) 1.00–5.00 .146** .191** (.795)a Notes: Spirituality and hope use Likert scales from 0 to 10 and the scale measuring grit from 1 to 5.aCronbach’s alpha.**p < .01. Table 2: Descriptive and Bivariate Statistics for the Study Variables (N = 1,045) Variable M (SD) Range 1 2 3 1. Spirituality 6.93 (2.20) 0.00–10.00 (.801)a 2. Employment hope 8.94 (1.42) 0.00–10.00 .192** (.936)a 3. Grit 3.77 (0.92) 1.00–5.00 .146** .191** (.795)a Variable M (SD) Range 1 2 3 1. Spirituality 6.93 (2.20) 0.00–10.00 (.801)a 2. Employment hope 8.94 (1.42) 0.00–10.00 .192** (.936)a 3. Grit 3.77 (0.92) 1.00–5.00 .146** .191** (.795)a Notes: Spirituality and hope use Likert scales from 0 to 10 and the scale measuring grit from 1 to 5.aCronbach’s alpha.**p < .01. Measurement Model A measurement model was constructed consisting of the latent constructs spirituality, employment hope, and grit. The individual item reliability was tested to assess the proposed dimensionality with each latent construct. Three spirituality indicators and four grit indicators were deleted due to factor loadings less than .70 (Fornell & Larcker, 1981). After deleting these seven items, the fit of the measurement model was tested. Model fit was assessed using the three fit indices mentioned earlier (CFI, RMSEA, and SRMR). For the CFI, values greater than .90 indicate a marginal fit and values greater than .95 indicate a good fit (Byrne, 2016). For the RMSEA, values less than .08 represent a reasonable fit and values less than .05 represent a good fit (Byrne, 2016), with values greater than .10 indicating a poor fit (Garson, 2012). For the SRMR, values less than .09 represent a good fit between the proposed model and the data (Mueller & Hancock, 2008). Although the results of the traditional chi-square test are reported, it should be noted that this test is not widely used to assess model fit because the null hypothesis is too strict and affected by the sample size (Fabrigar, Wegener, MacCallum, & Strahan, 1999). The values for the measurement model indicated the model fit the data well: [χ2(41, N = 1,045) = 107.041, p =.000], CFI = .987, RMSEA = .039 [90% CI: .030, .048], SRMR = .0268. Standardized factor loadings were calculated for each of the observed variables. As presented in Table 3, the factor loadings function as validity coefficients, indicating how accurately the item measures the latent construct. Coefficients greater than .70 indicate relatively high loadings (Kline, 2016). Table 3: Confirmatory Factor Analysis of the Measurement Model Latent and Observed Variable λ Employment hope  3. When working or looking for a job, I am respectful toward who I am. .728  4. I am worthy of working in a good job. .742  5. I am capable of working in a good job. .751  6. I have the strength to overcome any obstacles when it comes to working. .761  11. I am going to be working in a career job. .683  15. I feel energized when I think about future achievement with my job. .785  17. I am aware of what my skills are to be employed in a good job. .778  18. I am aware of what my resources are to be employed in a good job. .700  19. I am able to use my skills to move toward career goals. .801  20. I am able to use my resources to move toward career goals. .743  21. I am on the road toward my career goals. .694  22. I am in the process of moving forward reaching my goals. .729  23. Even if I am not able to achieve my financial goals right away, I will find a way to get there. .734  24. My current path will take me to where I need to be in my career. .698 Spirituality  2. Growing spiritually is more important than anything else in my life. .700  4. Spirituality is the master motive of my life, directing every other aspect of my life. .950  6. My spiritual beliefs affect absolutely every aspect of my life. .814 Grit  1. I often set a goal but later choose to pursue a different one. .682  3. I become interested in new pursuits every few months. .719  5. I have been obsessed with a certain idea or project for a short time but later lost interest. .690  6. I have difficulty maintaining my focus on projects that take more than a few months to complete. .771 χ2 = 107.041 (df = 41), p = .000, CFI = .987, RMSEA = .039 (90% CI: .030, .048), SRMR = .0268 Latent and Observed Variable λ Employment hope  3. When working or looking for a job, I am respectful toward who I am. .728  4. I am worthy of working in a good job. .742  5. I am capable of working in a good job. .751  6. I have the strength to overcome any obstacles when it comes to working. .761  11. I am going to be working in a career job. .683  15. I feel energized when I think about future achievement with my job. .785  17. I am aware of what my skills are to be employed in a good job. .778  18. I am aware of what my resources are to be employed in a good job. .700  19. I am able to use my skills to move toward career goals. .801  20. I am able to use my resources to move toward career goals. .743  21. I am on the road toward my career goals. .694  22. I am in the process of moving forward reaching my goals. .729  23. Even if I am not able to achieve my financial goals right away, I will find a way to get there. .734  24. My current path will take me to where I need to be in my career. .698 Spirituality  2. Growing spiritually is more important than anything else in my life. .700  4. Spirituality is the master motive of my life, directing every other aspect of my life. .950  6. My spiritual beliefs affect absolutely every aspect of my life. .814 Grit  1. I often set a goal but later choose to pursue a different one. .682  3. I become interested in new pursuits every few months. .719  5. I have been obsessed with a certain idea or project for a short time but later lost interest. .690  6. I have difficulty maintaining my focus on projects that take more than a few months to complete. .771 χ2 = 107.041 (df = 41), p = .000, CFI = .987, RMSEA = .039 (90% CI: .030, .048), SRMR = .0268 Notes: λ = standardized factor loading of the observed variable on the latent construct. CFI = Comparative Fit Index; RMSEA = root mean square error of approximation; CI = confidence interval; SRMR = standardized root mean square residual. Table 3: Confirmatory Factor Analysis of the Measurement Model Latent and Observed Variable λ Employment hope  3. When working or looking for a job, I am respectful toward who I am. .728  4. I am worthy of working in a good job. .742  5. I am capable of working in a good job. .751  6. I have the strength to overcome any obstacles when it comes to working. .761  11. I am going to be working in a career job. .683  15. I feel energized when I think about future achievement with my job. .785  17. I am aware of what my skills are to be employed in a good job. .778  18. I am aware of what my resources are to be employed in a good job. .700  19. I am able to use my skills to move toward career goals. .801  20. I am able to use my resources to move toward career goals. .743  21. I am on the road toward my career goals. .694  22. I am in the process of moving forward reaching my goals. .729  23. Even if I am not able to achieve my financial goals right away, I will find a way to get there. .734  24. My current path will take me to where I need to be in my career. .698 Spirituality  2. Growing spiritually is more important than anything else in my life. .700  4. Spirituality is the master motive of my life, directing every other aspect of my life. .950  6. My spiritual beliefs affect absolutely every aspect of my life. .814 Grit  1. I often set a goal but later choose to pursue a different one. .682  3. I become interested in new pursuits every few months. .719  5. I have been obsessed with a certain idea or project for a short time but later lost interest. .690  6. I have difficulty maintaining my focus on projects that take more than a few months to complete. .771 χ2 = 107.041 (df = 41), p = .000, CFI = .987, RMSEA = .039 (90% CI: .030, .048), SRMR = .0268 Latent and Observed Variable λ Employment hope  3. When working or looking for a job, I am respectful toward who I am. .728  4. I am worthy of working in a good job. .742  5. I am capable of working in a good job. .751  6. I have the strength to overcome any obstacles when it comes to working. .761  11. I am going to be working in a career job. .683  15. I feel energized when I think about future achievement with my job. .785  17. I am aware of what my skills are to be employed in a good job. .778  18. I am aware of what my resources are to be employed in a good job. .700  19. I am able to use my skills to move toward career goals. .801  20. I am able to use my resources to move toward career goals. .743  21. I am on the road toward my career goals. .694  22. I am in the process of moving forward reaching my goals. .729  23. Even if I am not able to achieve my financial goals right away, I will find a way to get there. .734  24. My current path will take me to where I need to be in my career. .698 Spirituality  2. Growing spiritually is more important than anything else in my life. .700  4. Spirituality is the master motive of my life, directing every other aspect of my life. .950  6. My spiritual beliefs affect absolutely every aspect of my life. .814 Grit  1. I often set a goal but later choose to pursue a different one. .682  3. I become interested in new pursuits every few months. .719  5. I have been obsessed with a certain idea or project for a short time but later lost interest. .690  6. I have difficulty maintaining my focus on projects that take more than a few months to complete. .771 χ2 = 107.041 (df = 41), p = .000, CFI = .987, RMSEA = .039 (90% CI: .030, .048), SRMR = .0268 Notes: λ = standardized factor loading of the observed variable on the latent construct. CFI = Comparative Fit Index; RMSEA = root mean square error of approximation; CI = confidence interval; SRMR = standardized root mean square residual. Structural Model Given the validated measurement model, it is appropriate to test the hypothesized model in which the relationship between spirituality and grit is mediated by employment hope (see Figure 1). The fit indices for the structural model indicated a satisfactory fit: [χ2(59, N = 1,045) = 176.880, p = .000], CFI = .977, RMSEA = .044 [90% CI: .036, .051], SRMR = .0332. In light of these values, the individual path coefficients can be examined. Coefficients for the various paths that comprise the structural model are presented in Table 4. All hypotheses were supported. The relationship between spirituality and employment hope was significant (H1) as was the relationship between employment hope and grit (H3), which implies that spirituality may enhance grit, mediated by employment hope. Spirituality also exhibited a direct, positive effect on grit (H2). Regarding the control variables, age was not significantly related to grit. Conversely, education level was related to grit, which suggests education may contribute to the development of grit. Table 4: Result of Structural Equation Modeling with the Hypothesized Model (N = 1,045) Path Path Coefficients p SE CR H1: Spirituality → Employment hope .025 (.064) .050* .013 1.878 H2: Spirituality → Grit .033 (.118) .0*** .010 3.307 H3: Employment hope → Grit .147 (.205) .0*** .027 5.528 Age → Grit .004 (.063) .067 .002 1.835 Education level → Grit .146 (.098) .005** .052 2.813 χ2 = 176.880 (df = 59), p =.000, CFI = .977, RMSEA = .044 (90% CI: .036, .051), SRMR = .0332 Path Path Coefficients p SE CR H1: Spirituality → Employment hope .025 (.064) .050* .013 1.878 H2: Spirituality → Grit .033 (.118) .0*** .010 3.307 H3: Employment hope → Grit .147 (.205) .0*** .027 5.528 Age → Grit .004 (.063) .067 .002 1.835 Education level → Grit .146 (.098) .005** .052 2.813 χ2 = 176.880 (df = 59), p =.000, CFI = .977, RMSEA = .044 (90% CI: .036, .051), SRMR = .0332 Notes: Standardized parameter estimates are reported in parentheses. CR = critical ratio; CFI = Comparative Fit Index; RMSEA = root mean square error of approximation; CI = confidence interval; SRMR = standardized root mean square residual.*p < .05. **p < .01. ***p < .001. Table 4: Result of Structural Equation Modeling with the Hypothesized Model (N = 1,045) Path Path Coefficients p SE CR H1: Spirituality → Employment hope .025 (.064) .050* .013 1.878 H2: Spirituality → Grit .033 (.118) .0*** .010 3.307 H3: Employment hope → Grit .147 (.205) .0*** .027 5.528 Age → Grit .004 (.063) .067 .002 1.835 Education level → Grit .146 (.098) .005** .052 2.813 χ2 = 176.880 (df = 59), p =.000, CFI = .977, RMSEA = .044 (90% CI: .036, .051), SRMR = .0332 Path Path Coefficients p SE CR H1: Spirituality → Employment hope .025 (.064) .050* .013 1.878 H2: Spirituality → Grit .033 (.118) .0*** .010 3.307 H3: Employment hope → Grit .147 (.205) .0*** .027 5.528 Age → Grit .004 (.063) .067 .002 1.835 Education level → Grit .146 (.098) .005** .052 2.813 χ2 = 176.880 (df = 59), p =.000, CFI = .977, RMSEA = .044 (90% CI: .036, .051), SRMR = .0332 Notes: Standardized parameter estimates are reported in parentheses. CR = critical ratio; CFI = Comparative Fit Index; RMSEA = root mean square error of approximation; CI = confidence interval; SRMR = standardized root mean square residual.*p < .05. **p < .01. ***p < .001. Indirect Effect As implied in Figure 1, employment hope partially mediated the relationship between spirituality and grit. To estimate the statistical significance of the indirect effect from spirituality to grit—in other words, the mediating effect of employment hope on spirituality and grit—bootstrapping was performed. Bootstrapping is the most plausible method to determine indirect effects (Dearing & Hamilton, 2006; Hayes, 2009). If zero exists in the interval of standardized estimates, no significant indirect effect exists. Accordingly, a bootstrapping test was performed using 800 bootstrap replicated samples and 95% CIs. The results revealed that the indirect effect of spirituality on grit was significant without zero scores between lower and upper bounds: standardized estimate (p) = .131 (.002), range between lower bound and upper bound [95% CI = .069, .195]. This finding demonstrates that employment hope mediates the path from spirituality to grit. Discussion The concept of grit has been the recipient of a substantial degree of interest across disciplines due to its robust ability to predict success in a wide variety of settings (Griffin et al., 2016; Hammond, 2017). Grit may be a particularly salient variable among disenfranchised populations. Because members of these groups typically encounter a disproportionate number of obstacles, the ability to persevere in the face of challenges is often instrumental to achieving success. Reflecting this line of reasoning, the present study developed and tested a model of grit among a sample of underemployed and unemployed urban African Americans seeking to improve their employment prospects. Based on the extant research and theory, the study proposed that spirituality would have a direct, positive impact on grit. In addition, it was posited that the relationship between spirituality and grit would be mediated by employment hope. These hypotheses were confirmed with the study sample. The results revealed a positive relationship between spirituality and grit, which was partially mediated by employment hope. These findings represent an important contribution to the existing literature in at least four ways. First, the study adds to the limited research on grit using African American samples (Strayhorn, 2014). Given the challenges members of this population continue to encounter in obtaining employment, studies that focus on variables that can facilitate successful outcomes represent important contributions to the literature (Bonilla-Silva, 2018). Second, the results confirm and extend prior research on grit. Previous qualitative research with African Americans indicates that spirituality plays a key role in enhancing grit (Yates et al., 2015). Respondents perceived their relationship with God, and associated beliefs and values, to directly facilitate higher levels of grit. The results of the present study confirm these qualitative findings and extend them by documenting a quantitative relationship between spirituality and grit. Third, the study adds to the wider body of research on spirituality. Prior research has illustrated a positive relationship between spirituality and a wide array of salutary outcomes (Koenig et al., 2012). The present study suggests that spirituality is also associated with grit. This is particularly significant because spirituality, as well as the constructs of employment hope and grit, is a strengths-based variable that resonates with the profession’s ethical calling to recognize and leverage clients’ strengths (Saleebey, 2013). Fourth, the study provides important insights into how grit may be enhanced. The importance of understanding the processes by which grit might be increased has been repeatedly cited as a crucial area of inquiry (Duckworth et al., 2009; Griffin et al., 2016; Guerrero et al., 2016). The present study directly addresses this concern. The findings suggest that both spirituality and employment hope may be key constructs that contribute to grit. Particularly notable is the finding that employment hope—a key component of psychological self-sufficiency—mediates the relationship between spirituality and grit. This suggests that employment hope, in tandem with grit, may function as dynamic noncognitive factors that foster positive employment outcomes (Hong et al., 2018). Given the sense of hopelessness that job seekers can experience (Bolland et al., 2005; Gibbs & Bankhead, 2000), these findings have important implications for practice, particularly with urban African American who are under- or unemployed. Implications for Practice Urban African American job seekers encounter many challenges locating employment (Bonilla-Silva, 2018). These barriers are both structural (for example, a paucity of decent paying jobs) and psychological (for example, hopelessness and despair) in nature (Bolland et al., 2005; Gibbs & Bankhead, 2000). Grit is imperative in dealing with these employment barriers. Accordingly, implementing strategies that facilitate grit is an important concern in practice settings (Duckworth et al., 2009). The results suggest that interventions that focus on spirituality and employment hope may enhance grit. Toward that end, social work practitioners might incorporate the administration of a spiritual assessment into their work with clients (Hodge, 2015). Such assessments can be used to identify relevant spiritual beliefs, including beliefs that foster both employment hope and grit. It should be noted that life problems can overwhelm clients (Saleebey, 2013). When faced with repeated rejections, job seekers can lose sight of the larger picture (Bolland et al., 2005; Gibbs & Bankhead, 2000). In such an emotional state, strengths that foster grit can be overlooked. Present challenges are frequently perceived to require continuous focused attention. As a result, sources of strength—such as spirituality and employment hope—can be neglected (Snyder et al., 2011). Practitioners can assist clients by identifying salutary spiritual beliefs and behaviors. During the assessment, practitioners might look for practices that enhance hope and perseverance (Saleebey, 2013). For instance, qualitative work suggests that meditating on biblical passages, such as “I can do all things through Christ who strengthens me,” can help develop grit (Yates et al., 2015). One social work practice model that could facilitate this process is the Transforming Impossible into Possible (TIP) program. The TIP program targets psychological self-sufficiency—seeking to reorient perceptions from barriers to hope—by drawing from individuals’ sources of strength, including spirituality (Hong, 2016). Assisting clients to refocus on intrinsic assets can help them persevere in the face of obstacles that were previously perceived to be overwhelming. Furthermore, social workers with expertise in cognitive–behavioral therapy (CBT) might consider adapting CBT protocols to include culturally relevant statements that enhance grit (Naeem et al. 2015). For instance, traditional CBT self-statements might be modified with content drawn from clients’ spiritual value system (Hodge, 2015). Statements that promote hope and grit could be substituted for unproductive thoughts that foster discouragement, depression, and other detrimental outcomes. Altering guiding cognitions in this manner may have a direct, positive effect on grit, positioning clients to press on in difficult circumstances (Strayhorn, 2014). Qualitative research on grit with African Africans suggests that such strategies may have some utility in practice settings (Yates et al., 2015). It is important to note, however, that intervention studies are needed before any definite conclusions can be drawn. The present study lays the groundwork for such research with underemployed and unemployed African American job seekers, and perhaps other groups as well. Limitations As is the case with any research, key limitations should be mentioned. The use of a nonprobability sampling procedure precludes generalization of the results to other groups of underemployed urban African Americans. The findings are only valid with the study sample. Conversely, the use of a consecutive admissions methodology to obtain study participants, in tandem with the relatively large sample size, helps mitigate bias. This suggests that the findings may have some applicability to other groups. The cross-sectional study design is also a limitation. As a result of the study design, no definitive conclusions can be made about the directionality of the relationships tested in the study. However, the findings are consistent with existing theory and research. For example, the identified relationship between spirituality and grit reflects the perspective of African Americans who, when asked how they acquired grittiness, reported that their spirituality played a direct, positive role in enhancing grit (Yates et al., 2015). Finally, it is important to mention that grit is not an unmitigated good. For example, among people with suicidal ideation, higher levels of grit may be associated with more frequent suicide attempts (Anestis & Selby, 2015). Grit is not a panacea, especially for disadvantaged populations. Rather, it is one variable that holds potential for helping people succeed in achieving their goals and aspirations. Conclusion Despite growing interest in the concept of grit across disciplines, comparatively little research has been conducted on this topic (Hammond, 2017; Wolters & Hussain, 2015). To the best of our knowledge, this is the first study to develop and test a model of grit among underemployed urban African Americans. Given the potential importance of grit in obtaining employment, this study represents a critical contribution to the literature on African American wellness. The results suggest spirituality has a direct, positive effect on grit and that this relationship is partially mediated by employment hope. The findings have important implications for social work practitioners working with members of this population, namely that spirituality and employment hope are protective factors that may be leveraged to potentially enhance grit. Further outcome studies are needed, however, to build on these foundational findings. Given the potential salience of grit to African Americans and other disadvantaged populations, such research should be a priority. David R. Hodge, PhD, is professor, School of Social Work, Arizona State University, Phoenix, and senior nonresident fellow, Program for Research on Religion and Urban Civil Society, University of Pennsylvania, Philadelphia. Philip Young P. Hong, PhD, is associate dean for research and Lucian and Carol Welch Matusak Endowed Professor and director, Center for Research on Self-Sufficiency, School of Social Work, Loyola University Chicago. Sangmi Choi, PhD, is assistant professor, Department of Social Welfare Counseling, Dongguk University, Seoul, Republic of Korea References Anderson , J. C. , & Gerbing , D. W. ( 1988 ). Structural equation modeling in practice: A review and recommended two-step approach . Psychological Bulletin, 103 , 411 – 423 . Google Scholar Crossref Search ADS Anestis , M. D. , & Selby , E. A. ( 2015 ). Grit and perseverance in suicidal behavior and non-suicidal self-injury . Death Studies, 39 , 211 – 218 . Google Scholar Crossref Search ADS Bailey , T. , Jenkins , D. , & Leinbach , T. ( 2005 ). Graduation rates, student goals, and measuring community college effectiveness (CCRC Brief Number 28 ). New York : Community College Research Center, Columbia University . Bolland , J. M. , Lian , B. E. , & Formichella , C. M. ( 2005 ). The origins of hopelessness among inner-city African American adolescents . American Journal of Community Psychology, 36 , 293 – 305 . Google Scholar Crossref Search ADS Bonilla-Silva , E. ( 2018 ). Racism without racists: Color-blind racism and the persistence of racial inequality in America ( 5th ed. ). Lanham, MD : Rowman & Littlefield . Bureau of Labor Statistics . ( 2017 ). Unemployment rate and employment-population ratio vary by race and ethnicity. Retrieved from https://www.bls.gov/opub/ted/2017/unemployment-rate-and-employment-population-ratio-vary-by-race-and-ethnicity.htm Burkhart , R. A. , Tholey , R. M. , Guinto , D. , Yeo , C. J. , & Chojnacki , K. A. ( 2014 ). Grit: A marker of residents at risk for attrition? Surgery, 155 , 1014 – 1022 . Google Scholar Crossref Search ADS Byrne , B. M. ( 2016 ). Structural equation modeling with AMOS: Basic concepts, applications and programming ( 3rd ed. ). New York : Taylor & Francis . Canda , E. R. , & Furman , L. D. ( 2010 ). Spiritual diversity in social work practice: The heart of helping ( 2nd ed. ). New York : Oxford University Press . Dearing , E. , & Hamilton , L. C. ( 2006 ). Best practices in quantitative methods for developmentalists: V. Contemporary advances and classic advice for analyzing mediating and moderating variables. Monographs of the Society for Research in Child Development. Duckworth , A. L. ( 2016 ). Grit: The passion and power of perseverance . New York : Scribner . Duckworth , A. L. , Peterson , C. , Matthews , M. D. , & Kelly , D. R. ( 2007 ). Grit: Perseverance and passion for long-term goals . Journal of Personality and Social Psychology, 92 , 1087 – 1101 . Google Scholar Crossref Search ADS Duckworth , A. L. , & Quinn , P. D. ( 2009 ). Development and validation of the Short Grit Scale (GRIT–S) . Journal of Personality Assessment, 91 , 166 – 174 . Google Scholar Crossref Search ADS Duckworth , A. L. , Quinn , P. D. , & Seligman , M. E. ( 2009 ). Positive predictors of teacher effectiveness . Journal of Positive Psychology, 4 , 540 – 547 . Google Scholar Crossref Search ADS Eskreis-Winkler , L. , Duckworth , A. L. , Shulman , E. P. , & Beal , S. ( 2014 ). The grit effect: Predicting retention in the military, the workplace, school and marriage . Frontiers in Psychology, 5 ( 36 ), 1 – 12 . Fabrigar , L. R. , Wegener , D. T. , MacCallum , R. C. , & Strahan , E. J. ( 1999 ). Evaluating the use of exploratory factor analysis in psychological research . Psychological Methods, 4 ( 3 ), 272 – 299 . Google Scholar Crossref Search ADS Fletcher , T. D. , Selgrade , K. A. , & Germano , L. M. ( 2006 ). On the use of partial covariances in structural equation modeling. Paper presented at the 21st Annual Conference of the Society for Industrial and Organizational Psychology, Dallas. Fornell , C. , & Larcker , D. F. ( 1981 ). Evaluating structural equation models with unobservable variables and measurement error . Journal of Marketing Research, 18 ( 1 ), 39 – 50 . Google Scholar Crossref Search ADS Garson , G. D. ( 2012 ). Structural equation modeling . Asheboro, NC : Statistical Publishing Associates . Gibbs , J. T. , & Bankhead , T. ( 2000 ). Joblessness and hopelessness: The case of African American youth in South Central Los Angeles . Journal of Ethnic & Cultural Diversity in Social Work, 9 ( 1–2 ), 1 – 20 . Google Scholar Crossref Search ADS Granqvist , P. , & Kirkpatrick , L. A. ( 2013 ). Religion, spirituality, and attachment. In K. I. Pargament (Ed.), APA handbook of psychology, religion, and spirituality: Vol. 1. Context, theory, and research (pp. 139 – 155 ). Washington, DC : American Psychological Association . Griffin , M. L. , McDermott , K. A. , McHugh , R. K. , Fitzmaurice , G. M. , & Weiss , R. D. ( 2016 ). Grit in patients with substance use disorders . American Journal on Addictions, 25 , 652 – 658 . Google Scholar Crossref Search ADS Guerrero , L. R. , Dudovitz , R. , Chung , P. J. , Dosanjh , K. K. , & Wong , M. D. ( 2016 ). Grit: A potential protective factor against substance use and other risk behaviors among Latino adolescents . Academic Pediatrics, 16 , 275 – 281 . Google Scholar Crossref Search ADS Hammond , D. A. ( 2017 ). Grit: An important characteristic in learners . Currents in Pharmacy Teaching and Learning, 9 ( 1 ), 1 – 3 . Google Scholar Crossref Search ADS Hayes , A. F. ( 2009 ). Beyond Baron and Kenny: Statistical mediation analysis in the new millennium . Communication Monographs, 76 , 408 – 420 . Google Scholar Crossref Search ADS Hodge , D. R. ( 2003 ). The Intrinsic Spirituality Scale: A new six-item instrument for assessing the salience of spirituality as a motivational construct . Journal of Social Service Research, 30 ( 1 ), 41 – 61 . Google Scholar Crossref Search ADS Hodge , D. R. ( 2015 ). Spiritual assessment in social work and mental health practice . New York : Columbia University Press . Hodge , D. R. , & Gillespie , D. ( 2003 ). Phrase completions: An alternative to Likert scales [Note on Research Methodology] . Social Work Research, 27 , 45 – 55 . Google Scholar Crossref Search ADS Hong , P.Y.P. ( 2013 ). Toward a client-centered benchmark for self-sufficiency: Evaluating the ‘process’ of becoming job ready . Journal of Community Practice, 21 , 356 – 378 . Google Scholar Crossref Search ADS Hong , P.Y.P. ( 2016 ). Transforming Impossible into Possible (TIP): A bottom-up practice in workforce development for low-income jobseekers . Environment and Social Psychology, 1 , 93 – 104 . Google Scholar Crossref Search ADS Hong , P.Y.P. , Choi , S. , & Key , W. ( 2018 ). Psychological self-sufficiency: A bottom-up theory of change in workforce development . Social Work Research, 42 , 22 – 32 . Google Scholar Crossref Search ADS Hong , P.Y.P. , Choi , S. , & Polanin , J. R. ( 2014 ). A multisample confirmatory factor analysis of the Short Employment Hope Scale (EHS-14) . Journal of Social Service Research, 40 , 339 – 352 . Google Scholar Crossref Search ADS Hong , P.Y.P. , Hodge , D. R. , & Choi , S. ( 2015 ). Spirituality, hope, and self-sufficiency among low-income job seekers . Social Work, 60 , 155 – 164 . Google Scholar Crossref Search ADS Kammeyer-Mueller , J. D. , & Wanberg , C. R. ( 2003 ). Unwrapping the organizational entry process: Disentangling multiple antecedents and their pathways to adjustment . Journal of Applied Psychology, 88 , 779 – 794 . Google Scholar Crossref Search ADS Kline , R. B. ( 2016 ). Principles and practices of structural equation modeling ( 4th ed. ). New York : Guilford Press . Koenig , H. G. , King , D. , & Carson , V. B. ( 2012 ). Handbook of religion and health ( 2nd ed. ). New York : Oxford University Press . Mueller , R. O. , & Hancock , G. R. ( 2008 ). Best practices in structural equation modeling. In J. Osborne (Ed.), Best practices in quantitative methods (pp. 488 – 508 ). Thousand Oaks, CA : Sage Publications . Mueller , R. O. , & Hancock , G. R. ( 2010 ). The reviewer’s guide to quantitative methods in the social sciences . New York : Routledge . Naeem , F. , Phiri , P. , Munshi , T. , Rathod , S. , Ayub , M. , Gobbi , M. , & Kingdon , D. ( 2015 ). Using cognitive behaviour therapy with South Asian Muslims: Findings from the culturally sensitive CBT project . International Review of Psychiatry, 27 , 233 – 246 . Newport , F. ( 2012 ). God is alive and well: The future of religion in America . New York : Gallup Press . Pargament , K. I. ( 2007 ). Spiritually integrated psychotherapy: Understanding and addressing the sacred . New York : Guilford Press . Reed , J. , Pritschet , B. L. , & Cutton , D. M. ( 2013 ). Grit, conscientiousness, and the transtheoretical model of change for exercise behavior . Journal of Health Psychology, 18 , 612 – 619 . Google Scholar Crossref Search ADS Saleebey , D. ( 2013 ). The strengths perspective in social work practice ( 6th ed. ). Upper Saddle River, NJ : Pearson . Salles , A. , Cohen , G. L. , & Mueller , C. M. ( 2014 ). The relationship between grit and resident well-being . American Journal of Surgery, 207 , 251 – 254 . Google Scholar Crossref Search ADS Schafer , J. L. , & Graham , J. W. ( 2002 ). Missing data: Our view of the state of the art . Psychological Methods, 7 , 147 – 177 . Google Scholar Crossref Search ADS Schumacker , R. E. , & Lomax , R. G. ( 2016 ). A beginner’s guide to structural equation modeling ( 4th ed. ). New York : Taylor & Francis . Snyder , C. R. , Lopez , S. J. , & Pedrotti , J. T. ( 2011 ). Positive psychology: The scientific and practical explorations of human strengths ( 2nd ed. ). Thousand Oaks, CA : Sage Publications . Strayhorn , T. L. ( 2014 ). What role does grit play in the academic success of black male collegians at predominantly white institutions? Journal of African American Studies, 18 , 1 – 10 . Google Scholar Crossref Search ADS U.S. Department of Labor . ( 2014 ). What works in job training: A synthesis of the evidence . Washington, DC : Author . Vela , J. C. , Lu , M.T.P. , Lenz , A. S. , & Hinojosa , K. ( 2015 ). Positive psychology and familial factors as predictors of Latina/o students’ psychological grit . Hispanic Journal of Behavioral Sciences, 37 , 287 – 303 . Google Scholar Crossref Search ADS Von Culin , K. R. , Tsukayama , E. , & Duckworth , A. L. ( 2014 ). Unpacking grit: Motivational correlates of perseverance and passion for long-term goals . Journal of Positive Psychology, 9 , 306 – 312 . Google Scholar Crossref Search ADS West , S. G. , Taylor , A. B. , & Wu , W. ( 2012 ). Model fit and model selection in structural equation modeling. In R. H. Hoyle (Ed.), Handbook of structural equation modeling (pp. 209 – 231 ). New York : Guilford Press . Wolters , C. A. , & Hussain , M. ( 2015 ). Investigating grit and its relations with college students’ self-regulated learning and academic achievement . Metacognition and Learning, 10 , 293 – 311 . Google Scholar Crossref Search ADS Wuthnow , R. ( 2007 ). After the baby boomers: How twenty- and thirty-somethings are shaping the future of American religion . Princeton, NJ : Princeton University Press . Yates , L. III , Moore , J. , Vairez , M. R. Jr. , Barber-Freeman , P. T. , Ross , W. , Parker , W. H. , & Bautista , R. ( 2015 ). The grit of African American male pre-service teachers . Journal of the Texas Alliance of Black School Educators, 1 ( 2 ), 11 – 38 . © 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)

Young, Diane, S;Casey, Erin, A

doi: 10.1093/swr/svy026pmid: N/A

Qualitative researchers often must make decisions about anticipated sample sizes in advance of data collection. Estimates are typically required for human subjects review committees, grant applications, and resource planning purposes. Once a study is underway or completed, researchers must evaluate whether the sample has been robust enough to address the research aims. The challenge is to find a sample that will produce thorough and meaningful findings while minimizing unnecessary burden on participants and expenditure of scarce resources such as time and research dollars. Currently, little guidance is available regarding what minimum sample size is needed to adequately identify the themes and codes in an area of inquiry. In addition, the issue of sample sizes needed to reach theme and code saturation across different qualitative methodologies or data analysis approaches is understudied. Although researchers often cite having achieved saturation as a reason to conclude sampling, details regarding how saturation was determined are not provided for the most part (Bowen, 2008; Francis et al., 2010). Gentles, Charles, Ploeg, and McKibbon (2015) conducted an overview of the literature from influential authors within the traditions of grounded theory, phenomenology, and case study, and they noted the lack of clarity relative to sample size and saturation. Guetterman (2015) looked at the most-cited empirical articles in the fields of education and health sciences from 2008 through 2012 within five qualitative research approaches to assess specific samples sizes and the rationale for sample sizes. Sample size across the 51 studies varied widely, and most articles did not include a discussion of saturation or the adequacy of the sample. In an effort to provide empirically based guidance about appropriate minimum sample sizes for qualitative studies, researchers have recently begun to conduct methodological studies that examine the point at which data saturation occurs. Guest, Bunce, and Johnson (2006) operationalized data saturation “as the point in data collection and analysis when new information produces little or no change to the codebook” (p. 65). They reviewed transcripts from a previous study in sets of six, according to the order in which individual interviews had been conducted at two research sites. They noted theme and code development, asking whether six interviews yielded as much data as 12, 18, 24, and so on interviews. They found that 73% of codes were identified in the first six interviews and 92% within the first 12 interviews. Examining this same question with focus group data, as opposed to individual interviews as in the earlier study, Guest, Namey, and McKenna (2016) found that 60% of their 94 codes were found in the first focus group, 84% in the first three groups, and 90% by six. When the focus groups were randomly ordered to assess for temporal bias, the results remained consistent. Other researchers have examined the question of minimum sample size using different definitions of data saturation, sometimes referred to as code saturation. Extending Guest et al.’s (2006) findings to cross-cultural research, Hagaman and Wutich (2016) considered three repetitions of a theme by different interviewees as identification of that theme and found that 16 interviews were enough to identify themes from homogeneous groups, with 20 to 40 needed to identify metathemes that cut across cultures and study sites. Francis and colleagues (2010) considered data saturation to be achieved when no new ideas emerged with three additional interviews. Using this strict stopping criterion of three, they examined interview data from two different studies and found that saturation was achieved in one study at 17 interviews, with no new data emerging after the 14th interview, and was not yet determined in the 14 interviews available to them from the second study. Even so, the majority of themes (92% and 86% in study one and two, respectively) emerged in the first six interviews. Finally, Hennink, Kaiser, and Marconi (2016) proposed that code saturation is the point at which you have “heard it all,” but that meaning saturation is the point when you “understand it all” (p. 15). In their study on patient retention with 25 individuals, they found that 84% of codes were identified by the sixth interview and 91% by the ninth interview. It took 16 to 24 interviews, however, to understand all the dimensions of the nine central codes, achieving meaning saturation. These findings suggest that under some study conditions, rich qualitative findings can be discovered with relatively small sample sizes. Further determining the parameters under which this applies would be helpful to researchers and research participants alike. Most efforts thus far have been done with studies relying on individual interviews, and many are within the medical field. In addition, examinations of minimal required sample sizes that examine available interviews once, in the order they were collected, raise concerns about possible temporal bias. We sought to examine the minimum sample sizes needed to adequately include the themes and codes in areas of inquiry within the field of social work. Considering three distinct qualitative research studies inclusive of both individual interviewing and focus group data collection approaches, we addressed four research questions: (1) What minimum sample size is needed to adequately identify codes (smaller units of meaning) within the data? (2) What minimum sample size is needed to ensure that all larger themes are partially represented by at least one of the codes that comprise that theme? (3) What minimum sample size is needed to fully realize the complete dimensionality of all themes by including all assigned codes? (4) Are minimum sample sizes needed consistent across different substantive areas of exploration and different modes of data collection, specifically individual interviews and focus groups? To address temporal bias, we addressed these questions by examining multiple random draws of various sample sizes within each included qualitative study. Method For the purpose of addressing the stated questions related to sample size and data redundancy, this article presents analyses done on data we previously collected for three distinct qualitative studies. Each original study is described briefly, outlining each one’s research aims, sample size and participant criteria, mode of data collection, analytic process, and number of resulting themes and codes. These brief synopses are presented to indicate the diversity of substantive areas and approaches used. More detail about each, including original research findings, are referenced. For the present methodological study, data from the original studies were not reanalyzed. Rather, the presence or absence of the themes and codes originally identified and described in the cited, published studies were examined in random subsamples. The Men Against Violence Study The Men Against Violence (MAV) study (Casey, 2010) consisted of individual interviews with 27 U.S. men between the ages of 20 and 72 who identified as allies in the prevention of gender-based violence. The primary aim of the study was to assess the strategies used and challenges faced by the participants as they work to engage other men and boys in violence prevention. Respondents represented all regions of the United States and were recruited via topic-relevant Listservs and referrals from violence prevention organizations. Data were gathered in person or over the phone via a uniform, semistructured interview guide that assessed the nature of men’s antiviolence involvement, their use and perceptions of effective and ineffective strategies for engaging other men, and the barriers they encountered in efforts to reach men. Once all interviews were conducted, transcripts resulting from the interviews were analyzed using techniques drawn from grounded theory and described by Charmaz (2006). Analysis included inductive, line-by-line coding of transcript content, in conjunction with extensive author memo making to uncover concepts within the data. Axial coding then used a constant comparative method both within and between cases to identify larger themes from a finalized list of more specific codes. This process identified four themes comprising 20 codes, or more specific units of meaning that collectively defined the full dimensionality of each theme. The Social Workers in Criminal Justice Study The Social Workers in Criminal Justice (SWCJ) study (Young, 2014) consisted of individual interviews with 15 experienced social workers working within diverse criminal justice settings in the northwestern United States. Participants shared their perspectives about the definitions of success and attributes needed for effective social work practice in their roles within adult prison, juvenile rehabilitation, treatment court, and offices of prosecution and public defense. Snowball sampling was used to locate individuals with an undergraduate or graduate degree in social work and currently practicing social work in a criminal justice setting. Interviews were conducted in person or over the phone with the use of a semistructured, uniform interview guide. Description rather than theory building shaped the analysis approach. Coding categories were gleaned from the text in relation to the general open-ended research questions: “How do you define success in your work?” and “What personal attributes are needed to be successful in your line of work?” The transcripts in their entirety were reviewed after all interviews were conducted. Once coding categories were identified and all transcripts were coded, the list of initial codes was reviewed and placed into conceptual groupings of major themes and subthemes. Then another thorough review of the transcripts was done, applying the revised set of coding categories to the transcripts and double-checking that the final set of themes and subthemes captured the ideas of the participants. This process identified eight themes comprising 30 specific units of meaning (codes) that collectively defined the full dimensionality of the themes. The Adolescent Bystander Behavior Study The Adolescent Bystander Behavior (ABB) study (Casey, Lindhorst, & Storer, 2017) aimed to identify influences on adolescent bystander decision making in the context of dating violence and bullying. More specifically, the project examined the relevance of two specific behavioral theories (the situational model of bystander behavior and the theory of planned behavior) to explaining bystander behavior. Data were gathered through 12 focus groups with a total of 113 youths ages 14 to 18; eight of these were face-to-face focus groups in local high schools and youth-serving agencies, and four groups were conducted in a real-time online format via text-based chat. Focus groups were facilitated by two researchers and data were gathered using a semistructured, uniform interview guide. Youths were asked to identify common dating violence and bullying scenarios, and then to talk in depth about the range of factors that would influence their decision making regarding how they might respond to these scenarios as bystanders. Data analysis proceeded in two phases once interviews were finished. First, deductive coding (Miles, Huberman, & Saldaña, 2014) was used to identify content in the transcripts relevant to the five constructs that collectively comprise the two guiding theoretical frameworks. Once all the transcripts were analyzed for content relevant to larger theory constructs, inductive thematic content analysis was used to identify codes reflecting the beliefs and ideas that collectively defined each larger theory construct. In addition, content regarding influences on bystander decision making that was not contained within the guiding theories was also inductively coded. These processes resulted in seven larger themes (the five theory constructs and two additional themes), which were defined by a total of 37 codes. The Present Methodological Study In our study, we retrospectively used the data and findings from the three previously described projects because they have important similarities and differences critical to addressing our research aims. Across the projects, interview or focus group data were transcribed and the transcripts thoroughly analyzed, resulting in a specific number of relevant themes and codes. However, in each project researchers addressed different topics and collectively gathered data through two methods: individual interviews and focus groups. Table 1 provides a listing of the number of cases, themes, and codes present in the original studies. Each individual interview or focus group transcript represents a case. Table 1: Sample Size and Number of Themes and Codes in the Reviewed Studies Study Name (Data Collection Method) Cases Themes Codes (n) (n) (n) MAV (individual interviews) 27 4 20 SWCJ (individual interviews) 15 8 30 ABB (focus groups) 12 7 37 Study Name (Data Collection Method) Cases Themes Codes (n) (n) (n) MAV (individual interviews) 27 4 20 SWCJ (individual interviews) 15 8 30 ABB (focus groups) 12 7 37 Notes: MAV = Men Against Violence; SWCJ = Social Workers in Criminal Justice; ABB = Adolescent Bystander Behavior. Table 1: Sample Size and Number of Themes and Codes in the Reviewed Studies Study Name (Data Collection Method) Cases Themes Codes (n) (n) (n) MAV (individual interviews) 27 4 20 SWCJ (individual interviews) 15 8 30 ABB (focus groups) 12 7 37 Study Name (Data Collection Method) Cases Themes Codes (n) (n) (n) MAV (individual interviews) 27 4 20 SWCJ (individual interviews) 15 8 30 ABB (focus groups) 12 7 37 Notes: MAV = Men Against Violence; SWCJ = Social Workers in Criminal Justice; ABB = Adolescent Bystander Behavior. A data set for each original study was created that identified for each transcript the presence or absence of the previously determined themes and codes. Then, using a random number generator, 10 random samples of each size from n = 5 through n = 10 for individual interviews and n = 2 through n = 7 for focus groups were drawn from each project. Because one focus group potentially yields more information than one individual interview and the research aims sought to determine minimum sample sizes, the number of focus groups comprising the subsamples was adjusted downward. To address the research aims, each randomly drawn subsample was then examined to see what proportion of the codes and larger themes from each original study’s full sample were present within each subsample. Finally, results from all 10 subsamples for a given sample size were averaged together to determine the mean presence (expressed as a percentage) of codes and themes. Results The first research aim was to examine at what sample size all final codes within the data were, on average, represented in the randomly selected transcripts. For interview-based projects, near-complete representation of codes was achieved at n = 8 in the MAV project (with an average of 97% of codes represented across random draws), and n = 9 in the SWCJ project (96% of codes represented). Adding one additional transcript to these sample sizes increased representation only to 98% in the MAV project, and did not add new coverage in the SWCJ project. The ABB focus group project achieved near-perfect code coverage at a sample size of six focus groups, with an average of 97% of codes represented across the random draws. Increasing the sample size to seven only increased coverage to an average of 98% of all possible codes. Near total inclusion of codes thus varied between n = 6 and n = 9 across the three qualitative projects. No project evidenced 100% average coverage across all draws at any sample size. Some individual draws reached 100% coverage starting at n = 5 for the MAV and ABB focus group projects, and n = 8 for the SWCJ project. Average code coverage findings are graphed in Figure 1. Figure 1: View largeDownload slide Average Proportion of Codes Present in Each Set of Random Samples of n Transcripts Figure 1: View largeDownload slide Average Proportion of Codes Present in Each Set of Random Samples of n Transcripts The second aim sought to identify the sample size at which all larger themes were at least partially represented by one or more codes within each theme. Findings show that at least some aspect of all larger themes are present at sample sizes ranging from 4 to 6. More specifically, the MAV and SWCJ projects reached 100% average partial representation of themes at n = 5 and n = 6, respectively. The ABB focus group project reached consistent partial theme representation at n = 4. Our third aim was to assess the sample sizes at which themes are fully realized within the data, that is, the point at which themes are defined by the full complement of codes that comprise them. These findings are presented in Figure 2. None of the three projects reached 100% theme completion at any of the examined sample sizes, although the percentage of fully defined themes was relatively high even with small samples. Specifically, the MAV project demonstrated 90% and 95% average theme realization at sample sizes of n = 9 and n = 10, respectively. The SWCJ project showed slightly lower theme completion with 86% average coverage at n = 9 and 85% average coverage at n = 10. For the ABB focus group data, 84% of themes were fully realized, on average, at n = 6, and 92% were completed at a sample size of seven. On some individual draws, however, 100% theme realization was found at n = 5 on the MAV project, n = 6 on the SWCJ project, and n = 5 on the ABB focus group project. Figure 2: View largeDownload slide Average Proportion of Fully Realized Themes in Each Set of Random Samples of n Transcripts Figure 2: View largeDownload slide Average Proportion of Fully Realized Themes in Each Set of Random Samples of n Transcripts Relative to our research aim 4, code and theme representation occurred at similar sample sizes within the three projects examined here across all metrics. As previously summarized, significant coverage of codes ranged from a minimum sample size of six to nine, partial theme representation required minimum sample sizes of four to six, and substantial theme completion necessitated sample sizes of seven to 10 cases across the projects. The ABB focus group project was consistently at the lower end of these ranges, and the more code-heavy of the individual interview projects (SWCJ) typically occupied the higher end. Discussion In three substantive areas, using two methodologies frequently used in qualitative research, findings from small subsamples adequately identified themes and codes in each area of inquiry. These findings agree with previous research (Guest et al., 2006, 2016; Hennink et al., 2016) and provide an important replication and extension of others’ work. The question about what sample size is sufficient is a critical methodological one, affecting almost all qualitative researchers. These findings give strong evidence and reassurance that researchers, under certain conditions, can achieve robust results with small sample sizes. Doing so will minimize participant burden and maximize limited resources. Clarifying the conditions under which small sample sizes yield meaningful findings will further benefit fields that rely heavily on qualitative research approaches. This is an important focus for future research. Aspects of the studies we drew on incorporated factors that are thought to contribute to the ability to achieve thorough findings with small sample sizes: participants met predetermined criteria and described similar experiences, and interviews were relatively structured (Guest et al., 2006; Malterud, Siersma, & Guassora, 2016). Extending the methods previously included in similar examinations of sample size and data redundancy, the studies we drew on incorporated in-person and telephone individual interviews and in-person and real-time online focus groups. That findings were consistent regardless of data collection method strengthens the conclusion that small qualitative samples are adequate for producing robust findings. In guarding against temporal bias by randomly drawing subsamples, we also found that the order in which the transcripts were examined was important. As few as five transcripts included all codes (100%) in some of the individual random sample draws for two out of the three research projects. Using randomization of multiple sample draws helped mitigate against conclusions based on early outliers. This may be a useful approach to continue in future studies. Our findings contribute to the growing body of evidence that robust identification of themes and codes may be achieved relatively quickly in interview and focus group data. Additional cases rounded out or added slight nuance to identified themes, but the vast majority of codes and themes were present in small samples. These findings echo conclusions reached by Hennink et al. (2016), who found near code saturation (“hearing it all”) at six to nine interviews, and additional nuance (“understanding it all”) as additional transcripts were included. The accumulating evidence across studies therefore suggests that rigorously collected qualitative data from small samples can substantially represent the full dimensionality of people’s experiences, with larger sample sizes adding important but perhaps increasingly minute pieces of meaning. Small sample size should not be seen as a limitation, in and of itself, when evaluating the rigor and findings of qualitative research. Diane S. Young, PhD, MSW, is director and associate professor and Erin A. Casey, PhD, MSW, is professor, Social Work and Criminal Justice Program, University of Washington–Tacoma References Bowen , G. A. ( 2008 ). Naturalistic inquiry and the saturation concept: A research note . Qualitative Research , 8 ( 1 ), 137 – 152 . Google Scholar Crossref Search ADS Casey , E. A. ( 2010 ). Strategies for engaging men as anti-violence allies: Implications for ally movements . Advances in Social Work , 11 , 267 – 282 . Casey , E. A. , Lindhorst , T. P. , & Storer , H. L. ( 2017 ). The situational-cognitive model of adolescent bystander behavior: Modelling bystander decision making in the context of bullying and teen dating violence . Psychology of Violence , 7 , 33 – 44 . Google Scholar Crossref Search ADS Charmaz , K. ( 2006 ) Constructing grounded theory: A practical guide through qualitative analysis . Thousand Oaks, CA : Sage Publications . Francis , J. J. , Johnston , M. , Robertson , C. , Glidewell , L. , Entwistle , V. , Eccles , M. P. , & Grimshaw , J. M. ( 2010 ). What is an adequate sample size? Operationalising data saturation for theory-based interview studies . Psychology and Health , 25 , 1229 – 1245 . Google Scholar Crossref Search ADS PubMed Gentles , S. J. , Charles , C. , Ploeg , J. , & McKibbon , K. A. ( 2015 ). Sampling in qualitative research: Insights from an overview of the methods literature . Qualitative Report , 20 , 1772 – 1789 . Guest , G. , Bunce , A. , & Johnson , L. ( 2006 ). How many interviews are enough? An experiment with data saturation and variability . Field Methods , 18 ( 1 ), 59 – 82 . Google Scholar Crossref Search ADS Guest , G. , Namey , E. , & McKenna , K. ( 2016 ). How many focus groups are enough? Building an evidence base for nonprobability sample sizes . Field Methods , 29 ( 1 ), 3 – 22 . Google Scholar Crossref Search ADS Guetterman , T. C. ( 2015 ). Descriptions of sampling practices within five approaches to qualitative research in education and the health sciences . Forum: Qualitative Social Research , 16 ( 2 ), Article 25 . Hagaman , A. K. , & Wutich , A. ( 2016 ). How many interviews are enough to identify metathemes in multisited and cross-cultural research? Another perspective on Guest, Bunce, and Johnson’s (2006) landmark study . Field Methods , 29 ( 1 ), 23 – 41 . Google Scholar Crossref Search ADS Hennink , M. M. , Kaiser , B. N. , & Marconi , V. C. ( 2016 ). Code saturation versus meaning saturation: How many interviews are enough? Qualitative Health Research , 27 , 1 – 18 . Malterud , K. , Siersma , V. D. , & Guassora , A. D. ( 2016 ). Sample size in qualitative interview studies: Guided by information power . Qualitative Health Research , 26 , 1753 – 1760 . Google Scholar Crossref Search ADS Miles , M. B. , Huberman , A. M. , & Saldaña , J. ( 2014 ). Qualitative data analysis ( 3rd ed. ). Thousand Oaks, CA : Sage Publications . Young , D. S. ( 2014 ). Social workers’ perspectives on effective practice in criminal justice settings . Journal of Forensic Social Work , 4 ( 2 ), 104 – 122 . Google Scholar Crossref Search ADS © 2018 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)

Wojtalik, Jessica, A;Eack, Shaun, M

doi: 10.1093/swr/svy032pmid: N/A

Schizophrenia is a severe psychiatric condition involving the experience of positive symptoms (hallucinations, delusions, or both), negative symptoms (affective flattening, alogia, anhedonia, asociality, and avolition), disorganized speech or behavior, and significant social and occupational challenges. As such, this diagnosis places substantial limits on achieving personal life goals such as living independently. Substance misuse is also highly prevalent among this population. The lifetime prevalence rates of cannabis (17% to 80.3%) and alcohol (21% to 86%) use are alarmingly higher among people with schizophrenia compared with the general population (cannabis: 5.8% to 16.4%; alcohol: 2.9% to 17.9%) (Volkow, 2009). Consequently, it is not surprising that people with this condition experience a poor quality of life (QoL) (Eack & Newhill, 2007). And, vital to this investigation, results from the large-scale National Institute of Mental Health Clinical Antipsychotic Trials of Intervention Effectiveness (known as CATIE) study demonstrated that people with schizophrenia actively misusing substances experience an even poorer QoL compared with those not misusing substances (Kerfoot et al., 2011). As such, Kerfoot et al. (2011) recommended comprehensive assessment of and attention to substance misuse in this population related to the adverse impact on QoL. QoL is a multidimensional construct that describes how a person perceives his or her life situation relative to the values and norms of society (WHOQOL Group, 1998) and is an important treatment outcome in social work practice. Whereas antipsychotic medications have little impact on QoL (Swartz et al., 2007), cognitive remediation (CR) interventions for schizophrenia are potentially effective for improving QoL (Garrido et al., 2013). Individuals with schizophrenia actively misusing substances, however, are characteristically excluded from CR clinical trials (Keshavan, Vinogradov, Rumsey, Sherrill, & Wagner, 2014), so the impact on their QoL is unknown. Cognitive enhancement therapy (CET) (Hogarty & Greenwald, 2006) is a comprehensive approach to CR that recently demonstrated beneficial effects for people with schizophrenia actively misusing alcohol or cannabis (Eack, Hogarty, Bangalore, Keshavan, & Cornelius, 2016; Eack et al., 2015). Compared with treatment as usual (TAU) in a randomized feasibility trial, participants in CET reported a high degree of treatment satisfaction, with significant improvements in neurocognition, social cognition, social adjustment, and alcohol misuse (Eack et al., 2016; Eack et al., 2015). In addition, these same CET participants had increased emotion regulation–related brain functioning associated with improved emotion processing (Wojtalik et al., 2015). Previous trials of CET in clients with early course and long-term schizophrenia have also shown considerable improvements in functional outcome (Eack et al., 2009; Hogarty et al., 2004), but effects on QoL have not yet been examined. This follow-up analysis of our previous feasibility trial of CET for clients with substance use problems (Eack et al., 2015) sought to examine the impact of CET on QoL. Method Participants and Procedures A total of 31 individuals diagnosed with schizophrenia (n = 17) or schizoaffective disorder (n = 14) with alcohol or cannabis misuse problems were included in an 18-month feasibility trial of CET (Eack et al., 2015). The randomization ratio was two to one, with 22 participants randomly assigned to CET and nine to TAU. QoL and other cognitive and social adjustment measures were completed by participants every six months for a total of four timepoints (baseline, six months, 12 months, and 18 months) over the course of the trial. All participants provided written informed consent before participation. The study protocol was approved by the University of Pittsburgh Institutional Review Board, reviewed annually, and registered with the national clinical trials database (NCT01292577). A full description of the sample, inclusion and exclusion criteria, and the CONSORT diagram of enrollment can be found in the main outcome article (Eack et al., 2015). Briefly, the participants were ethnically diverse (white: n = 16, 52%; nonwhite: n = 15, 48%), ill for an average of 14.19 years (SD = 11.28), just under the age of 40 (M = 38.23, SD = 13.44), and mostly male (n = 22, 71%). Over half of the participants had some college education (n = 21, 68%), although most were unemployed at baseline (n = 25, 81%). Participants in CET and TAU did not significantly differ in common baseline demographics, including age, sex, race, illness length, IQ, education, and antipsychotic medication dose (all p > .252), daily substance use (all p > .355), and addiction severity (all p < .676). Finally, as discussed in the main outcome article (Eack et al., 2015), attrition was high in the CET group (47%), but not significantly different (p = .148) from attrition in the TAU condition (11%). The overall attrition rate was 36%. Measures QoL QoL,the primary outcome variable, was assessed with the World Health Organization Quality of Life Measure-Abbreviated Version (WHOQOL-BREF), a generic, culturally sensitive measure of subjective QoL (WHOQOL Group, 1998). The WHOQOL-BREF includes 26 self-report items that evaluate four general domains (physical health, psychological, social relationships, and environmental) of QoL on a five-point Likert scale, with a higher score indicating a better QoL. The domain scores have demonstrated good psychometric properties (WHOQOL Group, 1998). Cognition and Social Adjustment A comprehensive battery of commonly used and validated cognitive and behavioral assessments for schizophrenia was completed at each timepoint. See Eack et al. (2015) for a description of the measures and the calculation of composite indexes for social cognition, neurocognition, and social adjustment derived from this battery. The social adjustment composite included measures of role functioning, employment, independent living, and interpersonal functioning (see Eack et al., 2015). The Timeline Followback method (Sobell & Sobell, 1992) evaluated substance use in the last 30 days. Confounding Variables All analyses adjusted for the same a priori confounding baseline variables as the parent study (Eack et al., 2015), which included age, sex, IQ, illness length, and drug use severity. IQ was measured with the Ammons Quick IQ Test (Ammons & Ammons, 1962) and illness length was calculated from the Structured Clinical Interview for DSM-IV (First, Spitzer, Gibbon, & Williams, 2002). Drug use severity was derived from the Addiction Severity Index (McLellan, Luborsky, Woody, & O’Brien, 1980; Sobell & Sobell, 1992). Treatments CET CET is an 18-month performance-based comprehensive and developmental approach to the remediation of cognitive challenges experienced by people diagnosed with schizophrenia (Hogarty & Greenwald, 2006). This recovery phase intervention consists of 60 hours of weekly computer-based neurocognitive training in attention, memory, and problem solving and 45 structured small-group sessions to address social–cognitive challenges. A full description of CET is provided in the training manual (Hogarty & Greenwald, 2006). CET is a scientifically based psychosocial intervention listed in the Substance Abuse and Mental Health Services Administration’s National Registry of Evidence-based Programs and Practices. TAU TAU served as the comparison treatment condition in this randomized feasibility trial of CET, consisting of commonly provided community-based social services, mental health treatments, and substance use programs. All participants were maintained on antipsychotic medications approved for the treatment of schizophrenia or schizoaffective disorder by their treating psychiatrist. Data Analysis Differential changes in QoL between CET and TAU were analyzed with four separate (for example, one for each QoL domain) linear growth curve models in R (version 3.1.2.) adjusting for the baseline confounding variables described earlier. The group (CET versus TAU) by time (baseline, six, 12, 18 months) interaction was the parameter of interest. Next, subsequent linear growth curve models were conducted to examine whether changes in the composite indexes or alcohol use were predictive of any observed significant QoL domain changes. CET previously demonstrated significant treatment effects in these areas in this sample (Eack et al., 2016; Eack et al., 2015). All models used an intent-to-treat (ITT) approach, handling missing data at the time of parameter estimation using expectation–maximization (Dempster, Laird, & Rubin, 1977), and used an autoregressive covariance structure (Raudenbush & Bryk, 2002). Results As can be seen in Table 1, CET demonstrated a significant and medium-sized effect on social relationships QoL over the 18 months of treatment (F = 7.15, df = 45, p = .010). In contrast to the participants in TAU, who had a deterioration in their social relationships QoL, the participants treated with CET demonstrated steady improvement. No other significant differential effects emerged favoring CET on the other QoL domains (all p > .109). Table 1: The Impact of Cognitive Enhancement Therapy on Quality of Life in Individuals with Schizophrenia Misusing Substances Timepoints Results Treatment Baseline Six Months 12 Months 18 Months p d Domain 1: Physical health  CET 12.35 (0.32) 12.61 (0.27) 12.88 (0.34) 13.14 (0.48) .838 .09  TAU 13.20 (0.50) 13.39 (0.38) 13.59 (0.41) 13.79 (0.55) Domain 2: Psychological  CET 12.99 (0.60) 13.23 (0.58) 13.47 (0.64) 13.71 (0.76) .678 .15  TAU 13.42 (0.93) 13.51 (0.88) 13.61 (0.90) 13.70 (0.99) Domain 3: Social relationships  CET 13.48 (0.84) 13.91 (0.83) 14.33 (0.86) 14.76 (0.94) .010 .70  TAU 14.17 (1.30) 13.77 (1.27) 13.37 (1.29) 12.97 (1.34) Domain 4: Environment  CET 13.73 (0.46) 14.25 (0.43) 14.76 (0.50) 15.28 (0.63) .109 .70  TAU 14.98 (0.72) 14.94 (0.64) 14.89 (0.66) 14.85 (0.77) Timepoints Results Treatment Baseline Six Months 12 Months 18 Months p d Domain 1: Physical health  CET 12.35 (0.32) 12.61 (0.27) 12.88 (0.34) 13.14 (0.48) .838 .09  TAU 13.20 (0.50) 13.39 (0.38) 13.59 (0.41) 13.79 (0.55) Domain 2: Psychological  CET 12.99 (0.60) 13.23 (0.58) 13.47 (0.64) 13.71 (0.76) .678 .15  TAU 13.42 (0.93) 13.51 (0.88) 13.61 (0.90) 13.70 (0.99) Domain 3: Social relationships  CET 13.48 (0.84) 13.91 (0.83) 14.33 (0.86) 14.76 (0.94) .010 .70  TAU 14.17 (1.30) 13.77 (1.27) 13.37 (1.29) 12.97 (1.34) Domain 4: Environment  CET 13.73 (0.46) 14.25 (0.43) 14.76 (0.50) 15.28 (0.63) .109 .70  TAU 14.98 (0.72) 14.94 (0.64) 14.89 (0.66) 14.85 (0.77) Notes: World Health Organization Quality of Life Measure-Abbreviated Version domain scores presented as M (SE) at each timepoint adjusting for baseline age, sex, IQ, illness length, and drug use severity. CET = cognitive enhancement therapy; TAU = treatment as usual. Bold values indicate statistical significance. Table 1: The Impact of Cognitive Enhancement Therapy on Quality of Life in Individuals with Schizophrenia Misusing Substances Timepoints Results Treatment Baseline Six Months 12 Months 18 Months p d Domain 1: Physical health  CET 12.35 (0.32) 12.61 (0.27) 12.88 (0.34) 13.14 (0.48) .838 .09  TAU 13.20 (0.50) 13.39 (0.38) 13.59 (0.41) 13.79 (0.55) Domain 2: Psychological  CET 12.99 (0.60) 13.23 (0.58) 13.47 (0.64) 13.71 (0.76) .678 .15  TAU 13.42 (0.93) 13.51 (0.88) 13.61 (0.90) 13.70 (0.99) Domain 3: Social relationships  CET 13.48 (0.84) 13.91 (0.83) 14.33 (0.86) 14.76 (0.94) .010 .70  TAU 14.17 (1.30) 13.77 (1.27) 13.37 (1.29) 12.97 (1.34) Domain 4: Environment  CET 13.73 (0.46) 14.25 (0.43) 14.76 (0.50) 15.28 (0.63) .109 .70  TAU 14.98 (0.72) 14.94 (0.64) 14.89 (0.66) 14.85 (0.77) Timepoints Results Treatment Baseline Six Months 12 Months 18 Months p d Domain 1: Physical health  CET 12.35 (0.32) 12.61 (0.27) 12.88 (0.34) 13.14 (0.48) .838 .09  TAU 13.20 (0.50) 13.39 (0.38) 13.59 (0.41) 13.79 (0.55) Domain 2: Psychological  CET 12.99 (0.60) 13.23 (0.58) 13.47 (0.64) 13.71 (0.76) .678 .15  TAU 13.42 (0.93) 13.51 (0.88) 13.61 (0.90) 13.70 (0.99) Domain 3: Social relationships  CET 13.48 (0.84) 13.91 (0.83) 14.33 (0.86) 14.76 (0.94) .010 .70  TAU 14.17 (1.30) 13.77 (1.27) 13.37 (1.29) 12.97 (1.34) Domain 4: Environment  CET 13.73 (0.46) 14.25 (0.43) 14.76 (0.50) 15.28 (0.63) .109 .70  TAU 14.98 (0.72) 14.94 (0.64) 14.89 (0.66) 14.85 (0.77) Notes: World Health Organization Quality of Life Measure-Abbreviated Version domain scores presented as M (SE) at each timepoint adjusting for baseline age, sex, IQ, illness length, and drug use severity. CET = cognitive enhancement therapy; TAU = treatment as usual. Bold values indicate statistical significance. After observing that CET had a significant impact on social relationships QoL, we examined whether changes in the composite indexes and alcohol use were predictive of improved social relationships QoL. Findings indicated that social adjustment was significantly related to social relationships QoL across the sample (B = 0.09, df = 44, p = .008), such that improvements in social adjustment over time predicted greater QoL in this domain. Cognition and alcohol use were unrelated to social relationships QoL (all p > .322). Discussion Individuals with schizophrenia often experience a poor QoL as a consequence of the severe and functionally disabling nature of the condition. The misuse of cannabis and alcohol is exceedingly common among people with schizophrenia (Volkow, 2009), and the toll taken on QoL is even greater for this group compared with those not misusing substances (Kerfoot et al., 2011). QoL is an important treatment outcome in social work practice, but few treatments for schizophrenia, such as antipsychotic medications (Swartz et al., 2007), improve QoL. CR interventions, such as CET (Eack et al., 2009; Hogarty et al., 2004), are promising treatments for ameliorating poor QoL in the condition. For example, Garrido et al. (2013) observed a significant QoL improvement in clients with schizophrenia after completing 48 hours of computer-assisted CR. This study, however, excluded individuals who reported active substance use within the past year. Thus, it is unknown if people with schizophrenia who also misuse substances would benefit from CR treatment with regard to QoL because active substance misuse is often a study exclusion criterion. Eack et al. (2015) demonstrated in a feasibility trial that the benefits of CET could be extended to those actively using substances, with significant and large effects on substance use and functional outcomes. To follow up on these beneficial effects of CET for the population of clients with schizophrenia misusing substances, the present study investigated the effects of CET on QoL in the same participants (Eack et al., 2015). The results indicated that CET had a significant effect on social relationships QoL compared with TAU. Over time, CET participants had a progressive increase in satisfaction with their social relationships, which contrasted with a deterioration among the TAU participants, as is common in the disorder (Aki et al., 2008). Furthermore, a link between social adjustment and social relationships QoL emerged, indicating that a greater degree of social adjustment to the condition was important for improving satisfaction with social relationships. Functional challenges, including poor adjustment to interpersonal relationships, are determinants of QoL in people with schizophrenia (Aki et al., 2008). CET places strong emphasis on adjustment to the condition to promote functional recovery (Hogarty & Greenwald, 2006). For example, participants treated with CET are significantly more likely to obtain, maintain, and feel satisfied with employment outcomes (Eack, Hogarty, Greenwald, Hogarty, & Keshavan, 2011). It may be that CET increases satisfaction with social relationships through treatment parameters targeted at functional recovery, especially for clients with schizophrenia struggling with substance misuse. Ultimately, such findings indicate that CET can meaningfully improve QoL in people living with schizophrenia and substance use comorbidity. The results from this study should be interpreted with consideration of a few limitations. First, given that this was a feasibility trial, the sample size was small and likely reduced the power of seeing smaller to moderate treatment effects on QoL, especially for the environmental domain, which had a similar effect size to social relationships (see Table 1). Furthermore, similar to other studies of significant addiction challenges (Mueser et al., 2013), attrition in this trial was considerable in the CET group. Although ITT analyses were used to include all randomized participants, it will be important for future studies to use strategies that can enhance participant retention. Next, using TAU as a comparison treatment may have influenced a greater CET treatment effect. CET is a manualized and intensive 18-month treatment, so future studies may benefit from comparing CET to a treatment of similar intensity that is not often found in community settings. Last, individuals with schizophrenia misusing only cannabis or alcohol were included in this trial. It is unknown how the effects of CET on QoL would generalize to other types of substance (such as cocaine or opioids) misuse. Overall, the findings from this study provide further evidence that CET is a feasible and possibly effective treatment for improving QoL in clients with schizophrenia with alcohol or cannabis misuse problems. Jessica A. Wojtalik, MSW, is a doctoral candidate and Shaun M. Eack, PhD, is James and Noel Browne Endowed Chair and professor of social work and psychiatry, School of Social Work, University of Pittsburgh. Address correspondence to Jessica A. Wojtalik, School of Social Work, University of Pittsburgh, 2117 Cathedral of Learning, 4200 Fifth Avenue, Pittsburgh, PA 15260; e-mail: [email protected]. Funding for this research was provided by National Institutes of Health grants DA-30763 (SME), MH-95783 (SME), RR-24154 (SME), and MH-113277 (JAW). The authors would also like to acknowledge the clinicians for this study: Susan S. Hogarty, Deborah P. Greenwald, and Maralee Y. Litschge. References Aki , H. , Tomotake , M. , Kaneda , Y. , Iga , J.-I. , Kinouchi , S. , Shibuya-Tayoshi , S. , et al. ( 2008 ). Subjective and objective quality of life, levels of life skills, and their clinical determinants in outpatients with schizophrenia . Psychiatry Research , 158 , 19 – 25 . Google Scholar Crossref Search ADS PubMed Ammons , R. B. , & Ammons , C. H. ( 1962 ). The Quick Test (QT): Provisional manual . Psychological Reports , 11 , 111 – 161 . Dempster , A. P. , Laird , N. M. , & Rubin , D. B. ( 1977 ). Maximum likelihood from incomplete data via the EM algorithm. Journal of the Royal Statistical Society : Series B (Methodological) , 39 ( 1 ), 1 – 38 . Google Scholar Crossref Search ADS Eack , S. M. , Greenwald , D. P. , Hogarty , S. S. , Cooley , S. J. , DiBarry , A. L. , Montrose , D. M. , & Keshavan , M. S. ( 2009 ). Cognitive enhancement therapy for early-course schizophrenia: Effects of a two-year randomized controlled trial . Psychiatric Services , 60 , 1468 – 1476 . Google Scholar Crossref Search ADS PubMed Eack , S. M. , Hogarty , S. S. , Bangalore , S. S. , Keshavan , M. S. , & Cornelius , J. R. ( 2016 ). Patterns of substance use during cognitive enhancement therapy: An 18-month randomized feasibility study . Journal of Dual Diagnosis , 12 ( 1 ), 74 – 82 . Google Scholar Crossref Search ADS PubMed Eack , S. M. , Hogarty , G. E. , Greenwald , D. P. , Hogarty , S. S. , & Keshavan , M. S. ( 2011 ). Effects of cognitive enhancement therapy on employment outcomes in early schizophrenia: Results from a 2-year randomized trial . Research on Social Work Practice , 21 , 32 – 42 . Google Scholar Crossref Search ADS PubMed Eack , S. M. , Hogarty , S. S. , Greenwald , D. P. , Litschge , M. Y. , McKnight , S.A.F. , Bangalore , S. S. , et al. ( 2015 ). Cognitive enhancement therapy in substance misusing schizophrenia: Results of an 18-month feasibility trial . Schizophrenia Research , 161 , 478 – 483 . doi:10.1016/j.schres.2014.11.017 Google Scholar Crossref Search ADS PubMed Eack , S. M. , & Newhill , C. E. ( 2007 ). Psychiatric symptoms and quality of life in schizophrenia: A meta-analysis . Schizophrenia Bulletin , 33 , 1225 – 1237 . Google Scholar Crossref Search ADS PubMed First , M. B. , Spitzer , R. L. , Gibbon , M. , & Williams , J.B.W. ( 2002 ). Structured Clinical Interview for DSM-IV-TR Axis I Disorders, Research Version ( Patient ed.). New York : Biometrics Research, New York State Psychiatric Institute . Garrido , G. , Barrios , M. , Penadés , R. , Enríquez , M. , Garolera , M. , Aragay , N. , et al. ( 2013 ). Computer-assisted cognitive remediation therapy: Cognition, self-esteem and quality of life in schizophrenia . Schizophrenia Research , 150 , 563 – 569 . Google Scholar Crossref Search ADS PubMed Hogarty , G. E. , Flesher , S. , Ulrich , R. , Carter , M. , Greenwald , D. , Pogue-Geile , M. , et al. ( 2004 ). Cognitive enhancement therapy for schizophrenia: Effects of a 2-year randomized trial on cognition and behavior . Archives of General Psychiatry , 61 , 866 – 876 . Google Scholar Crossref Search ADS PubMed Hogarty , G. E. , & Greenwald , D. P. ( 2006 ). Cognitive enhancement therapy: The training manual . Pittsburgh : University of Pittsburgh Medical Center . Available at http://www.CognitiveEnhancementTherapy.com Kerfoot , K. E. , Rosenheck , R. A. , Petrakis , I. L. , Swartz , M. S. , Keefe , R. S. , McEvoy , J. P. , et al. ( 2011 ). Substance use and schizophrenia: Adverse correlates in the CATIE study sample . Schizophrenia Research , 132 , 177 – 182 . Google Scholar Crossref Search ADS PubMed Keshavan , M. S. , Vinogradov , S. , Rumsey , J. , Sherrill , J. , & Wagner , A. ( 2014 ). Cognitive training in mental disorders: Update and future directions . American Journal of Psychiatry , 171 , 510 – 522 . Google Scholar Crossref Search ADS PubMed McLellan , A. T. , Luborsky , L. , Woody , G. E. , & O’Brien , C. P. ( 1980 ). An improved diagnostic evaluation instrument for substance abuse patients: The Addiction Severity Index . Journal of Nervous and Mental Disease , 168 , 26 – 33 . Google Scholar Crossref Search ADS PubMed Mueser , K. T. , Glynn , S. M. , Cather , C. , Xie , H. , Zarate , R. , Smith , L. F. , et al. ( 2013 ). A randomized controlled trial of family intervention for co-occurring substance use and severe psychiatric disorders . Schizophrenia Bulletin , 39 , 658 – 672 . Google Scholar Crossref Search ADS PubMed Raudenbush , D.S.W. , & Bryk , D.A.S. ( 2002 ). Hierarchical linear models: Applications and data analysis methods. Thousand Oaks, CA : Sage Publications . Sobell , L. C. , & Sobell , M. B. ( 1992 ). Timeline follow-back. In R. Z. Litten & J. Allen (Eds.), Measuring alcohol consumption: Psychosocial and biological methods (pp. 41 – 72 ). Totowa, NJ : Humana Press . Swartz , M. S. , Perkins , D. O. , Stroup , T. S. , Davis , S. M. , Capuano , G. , Rosenheck , R. A. , et al. ( 2007 ). Effects of antipsychotic medications on psychosocial functioning in patients with chronic schizophrenia: Findings from the NIMH CATIE study . American Journal of Psychiatry , 164 , 428 – 436 . Google Scholar Crossref Search ADS PubMed Volkow , N. D. ( 2009 ). Substance use disorders in schizophrenia—Clinical implications of comorbidity . Schizophrenia Bulletin , 35 , 469 – 472 . doi:10.1093/schbul/sbp016 Google Scholar Crossref Search ADS PubMed WHOQOL Group . ( 1998 ). Development of the World Health Organization WHOQOL-BREF quality of life assessment . Psychological Medicine , 28 , 551 – 558 . Crossref Search ADS PubMed Wojtalik , J. A. , Hogarty , S. S. , Cornelius , J. R. , Phillips , M. L. , Keshavan , M. S. , Newhill , C. E. , & Eack , S. M. ( 2015 ). Cognitive enhancement therapy improves frontolimbic regulation of emotion in alcohol and/or cannabis misusing schizophrenia: A preliminary study . Frontiers in Psychiatry , 6 , 186 . doi:10.3389/fpsyt.2015.00186 Google Scholar PubMed © 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)