Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

A Latent Class Analysis of Parent Involvement Subpopulations

A Latent Class Analysis of Parent Involvement Subpopulations Abstract Using nationally representative data from the Early Childhood Longitudinal Study, Kindergarten Class of 2010–11, this study relied on latent class analysis (LCA) to advance a subpopulation view of parent involvement (PI) in elementary school. Four PI subpopulation profiles were yielded using the LCA approach. Two of these parent subpopulations were involved in a limited number of school-based PI activities. Two others were involved in multiple activities at the school. It is significant that additional latent class regression analyses indicated that membership in these PI profile groups could be predicted by parents’ sociodemographic characteristics, especially their ethnicity, occupational status, family income, and social capital. Together, these findings highlight needs for school social workers to help schools develop PI programs and policies that are more nuanced. PI initiatives need to be tailored to fit the characteristics of particular parent subpopulations and particular school community contexts. Parent involvement (PI) is widely viewed as a critical ingredient for children’s academic achievement and overall school success (see, for example, Mapp & Kuttner, 2014). Yet, in spite of its known importance, PI remains weak in many school communities (Warren, Hong, Rubin, & Uy, 2009). This challenge of “low” PI is most frequently reported in low-income school communities, especially those urban and rural places challenged by social isolation and exclusion dynamics, racism, and other forms of concentrated disadvantage (Lareau, 2011). Two recent developments hold promise for enhancing PI in these and other challenged places. One is data-driven decision making (see, for example, Moss, 2012). The other is intervention science, which frames PI as a specialized intervention (H. Lawson, Alameda-Lawson, Lawson, Briar-Lawson, & Wilcox, 2014). Together, these twin developments signal opportunities for researchers to develop data-driven PI models that are action oriented and for school social workers, educators, and other school professionals to use these models to develop tailor-made PI strategies and interventions. The current study of school-based PI was designed to capitalize on this important opportunity. It proceeded with a person-centered approach to statistical analysis. Person-centered statistical methods (for example, cluster analysis, latent class analysis [LCA]) are salient for today’s PI research, practice, and policy agenda because they allow researchers to distill parents’ involvement in a broad range of activities into identifiable patterns or subpopulation profiles (Nylund, Bellmore, Nishina, & Graham, 2007). Demonstrable benefits follow. For instance, instead of rough-cut, catch-all categorizations of “the parents” that force a reliance on singular or standardized PI practices, person-centered models enable school social workers and other professionals to proceed with interventions that are customized to identified subpopulations. This study’s three research objectives were structured to illustrate this potential. The purposes were (a) to identify and describe subpopulation profiles of PI, (b) to explore the influence of family income on each PI profile, and (c) to identify and describe the sociodemographic characteristics associated with each PI profile group. Rationale: Needs and Opportunities in the Related Literature Educational leaders, researchers, and policymakers assert that PI is a driver for children’s academic achievement (Epstein, 2011; Lee & Bowen, 2006). Decades of correlational research provide the foundation for this enduring claim. Study after study has revealed consistent, positive associations between PI and children’s school outcomes, even when variables related to social class and ethnic group affiliation have been controlled in study designs (see, for example, Epstein, 2011; Jeynes, 2011). However, some recent research challenges PI’s efficacy for improving child and school outcomes. For example, a line of sociological research suggests that conventional PI activities (for example, volunteering at the school or participating in the Parent–Teacher Association [PTA]) may not attend to the strengths, needs, and challenges of low-income parents (see, for example, Lareau, 2011). Moreover, whereas much educational practice and policy has followed PI’s long-standing association with children’s academic achievement outcomes, recent research suggests that this relationship may not be automatic. In fact, some of the more nascent quantitative PI studies from sociology and education have yielded nonsignificant relationships between PI in conventional activities like volunteering at the school and academic achievement (see, for example, Robinson & Harris, 2014), and others have yielded significant inverse associations between PI activities like helping kids with homework and children’s academic outcomes (Robinson & Harris, 2014; Rogers, Theule, Ryan, Adams, & Keating, 2009). What factors might explain these emergent, contradictory findings? One answer to this question may lie in the often limited ways that PI is operationalized and then analyzed in quantitative research designs. For example, one strand of extant quantitative research operationalizes PI as the frequency of involvement in discrete activities such as volunteering at the school or participating in the PTA (see, for example, Robinson & Harris, 2014). In this activity-centered approach, researchers typically use regression modeling to explore the relationship between PI and academic achievement (see, for example, Park & Holloway, 2017). The goal is to isolate the effect that particular PI activities may exert on children’s academic outcomes by controlling for the influence of other PI activities in the statistical model. The primary strength of this activity-centered approach is that it helps practitioners and policymakers identify those PI activities that, on average, carry the most consequence for enhancing children’s academic outcomes. The primary limitation of this strategy is that it can mask the possibility that, in the real world, improvements in children’s school learning may not follow PI in singular activities or events, but instead may reflect the cumulative result of PI in multiple forms of activity (for example, volunteering at the school and participating in the PTA and helping children with homework). In light of these limitations, some quantitative researchers have operationalized PI more broadly to include parent participation in multiple school-based and home-based activities (see, for example, Fan & Chen, 2001). In these studies, scholars typically create a composite view of PI by summing the values of parents’ survey responses together to form a summative PI score (see, for example, Holloway et al., 2016). The strength of this “summative approach” is that it increases the variance of researchers’ PI variable, and this increased variance allows researchers to more precisely estimate PI’s relationship with children’s academic outcomes. The primary limitation is that it can create instances where parents share the same “PI score” as others who are involved in a qualitatively different set of PI activities. To illustrate this possibility, consider the following simple case example. In this scenario, parent A receives an involvement score of 8 by virtue of reporting that she “strongly agrees” that she participates regularly in the PTA (4) and the school-site council (4), but not other activities like back-to-school nights and open houses. Meanwhile, parent B receives the same PI score of “8” by strongly agreeing that she participates in back-to-school nights (4) and open houses (4), but not other PI activities like school governance. Thus, to the extent that different configurations of PI activity matter for children’s academic outcomes, the summative strategy is largely ill-equipped to capture their influence. Moreover, the lack of measurement precision inherent in this approach may explain why recent studies of the PI–academic achievement relationship have yielded inconsistent results. This study’s subpopulation profile models help address these aforementioned research challenges in two important ways. First, they provide a unique data-driven view of the pathways for improving PI. Second, they advance a more integrative and ecologically valid framework for analyzing PI’s relationship with children’s academic achievement outcomes. Method Participants in this study were recruited into the Early Childhood Longitudinal Study, Kindergarten Class of 2010–11 (ECLS-K:2011), a nationally representative cohort study of children attending both public and private elementary schools in the United States (Mulligan, Hastedt, & McCarroll, 2012). The ECLS-K:2011 provides rich data on children and parents’ early school experiences beginning with kindergarten and following children through fifth grade (see also Holloway et al., 2016). The sample for the present study was delimited to about 15,600 parents who had first-grade children attending 500 public elementary schools in the United States. The PI experiences of first-grade children are featured in this study because the ECLS-K:2011 administered its PI survey to parents when their children were in the first grade. Descriptive statistics of our ECLS-K:2011 parent sample are provided for readers in Table 1. All parent data were drawn using a restricted ECLS-K:2011 data file. Table 1: Descriptive Statistics of Primary Study Variables (N = 15,600) Demographic Variable M SE Min Max African American .123 .003 0 1 Hispanic .251 .004 0 1 Asian .046 .001 0 1 Other .058 .002 0 1 White .464 .001 0 1 English language learner .175 .003 0 1 Single parent .166 .003 0 1 Married .512 .003 0 1 Unmarried with domestic partner .322 .004 0 1 Number of parental contacts at school 3.55 .031 0 40 Parent education of high school or less .275 .004 0 1 Receives Temporary Assistance for Needy Families .06 .002 0 1 Income is $30,000 a year or less .363 .002 0 1 Income is between $30,001 and $74,999 .341 .004 0 1 Income is between $75,000 and $99,999 .123 .003 0 1 Earns $100,000 or more per year .173 .003 0 1 Social geography  Suburb .333 .004 0 1  Town .109 .003 0 1  Rural .234 .004 0 1  Urban .324 .020 0 1 Parent involvement barriers  Not feeling welcome at school .059 .002 0 1  Lack of child care .253 .004 0 1  Lack of transportation .058 .002 0 1  Working >35 hours a week .352 .004 0 1 Latent class manifest variables  Parent has contacted the school .77 .004 0 1  Parent attends school open house .806 .004 0 1  Parent participates in the Parent–Teacher Association or other parent–teacher organization .366 .004 0 1  Parent is involved in school council .126 .004 0 1  Parent attends parent–teacher conferences .951 .002 0 1  Parent attends school events .777 .004 0 1  Parent volunteers at the school .558 .003 0 1 Demographic Variable M SE Min Max African American .123 .003 0 1 Hispanic .251 .004 0 1 Asian .046 .001 0 1 Other .058 .002 0 1 White .464 .001 0 1 English language learner .175 .003 0 1 Single parent .166 .003 0 1 Married .512 .003 0 1 Unmarried with domestic partner .322 .004 0 1 Number of parental contacts at school 3.55 .031 0 40 Parent education of high school or less .275 .004 0 1 Receives Temporary Assistance for Needy Families .06 .002 0 1 Income is $30,000 a year or less .363 .002 0 1 Income is between $30,001 and $74,999 .341 .004 0 1 Income is between $75,000 and $99,999 .123 .003 0 1 Earns $100,000 or more per year .173 .003 0 1 Social geography  Suburb .333 .004 0 1  Town .109 .003 0 1  Rural .234 .004 0 1  Urban .324 .020 0 1 Parent involvement barriers  Not feeling welcome at school .059 .002 0 1  Lack of child care .253 .004 0 1  Lack of transportation .058 .002 0 1  Working >35 hours a week .352 .004 0 1 Latent class manifest variables  Parent has contacted the school .77 .004 0 1  Parent attends school open house .806 .004 0 1  Parent participates in the Parent–Teacher Association or other parent–teacher organization .366 .004 0 1  Parent is involved in school council .126 .004 0 1  Parent attends parent–teacher conferences .951 .002 0 1  Parent attends school events .777 .004 0 1  Parent volunteers at the school .558 .003 0 1 Note: SE = standard error. Table 1: Descriptive Statistics of Primary Study Variables (N = 15,600) Demographic Variable M SE Min Max African American .123 .003 0 1 Hispanic .251 .004 0 1 Asian .046 .001 0 1 Other .058 .002 0 1 White .464 .001 0 1 English language learner .175 .003 0 1 Single parent .166 .003 0 1 Married .512 .003 0 1 Unmarried with domestic partner .322 .004 0 1 Number of parental contacts at school 3.55 .031 0 40 Parent education of high school or less .275 .004 0 1 Receives Temporary Assistance for Needy Families .06 .002 0 1 Income is $30,000 a year or less .363 .002 0 1 Income is between $30,001 and $74,999 .341 .004 0 1 Income is between $75,000 and $99,999 .123 .003 0 1 Earns $100,000 or more per year .173 .003 0 1 Social geography  Suburb .333 .004 0 1  Town .109 .003 0 1  Rural .234 .004 0 1  Urban .324 .020 0 1 Parent involvement barriers  Not feeling welcome at school .059 .002 0 1  Lack of child care .253 .004 0 1  Lack of transportation .058 .002 0 1  Working >35 hours a week .352 .004 0 1 Latent class manifest variables  Parent has contacted the school .77 .004 0 1  Parent attends school open house .806 .004 0 1  Parent participates in the Parent–Teacher Association or other parent–teacher organization .366 .004 0 1  Parent is involved in school council .126 .004 0 1  Parent attends parent–teacher conferences .951 .002 0 1  Parent attends school events .777 .004 0 1  Parent volunteers at the school .558 .003 0 1 Demographic Variable M SE Min Max African American .123 .003 0 1 Hispanic .251 .004 0 1 Asian .046 .001 0 1 Other .058 .002 0 1 White .464 .001 0 1 English language learner .175 .003 0 1 Single parent .166 .003 0 1 Married .512 .003 0 1 Unmarried with domestic partner .322 .004 0 1 Number of parental contacts at school 3.55 .031 0 40 Parent education of high school or less .275 .004 0 1 Receives Temporary Assistance for Needy Families .06 .002 0 1 Income is $30,000 a year or less .363 .002 0 1 Income is between $30,001 and $74,999 .341 .004 0 1 Income is between $75,000 and $99,999 .123 .003 0 1 Earns $100,000 or more per year .173 .003 0 1 Social geography  Suburb .333 .004 0 1  Town .109 .003 0 1  Rural .234 .004 0 1  Urban .324 .020 0 1 Parent involvement barriers  Not feeling welcome at school .059 .002 0 1  Lack of child care .253 .004 0 1  Lack of transportation .058 .002 0 1  Working >35 hours a week .352 .004 0 1 Latent class manifest variables  Parent has contacted the school .77 .004 0 1  Parent attends school open house .806 .004 0 1  Parent participates in the Parent–Teacher Association or other parent–teacher organization .366 .004 0 1  Parent is involved in school council .126 .004 0 1  Parent attends parent–teacher conferences .951 .002 0 1  Parent attends school events .777 .004 0 1  Parent volunteers at the school .558 .003 0 1 Note: SE = standard error. Measures Our subpopulation parent profile models were analyzed using the seven available ECLS-K:2011 survey items that best reflected Epstein’s (2011) typology of parents’ school-based PI practices. These items measured whether each child’s primary parent or caregiver had ever (1) contacted the school, (2) attended an open house, (3) participated in the PTA or another parent–teacher organization (PTO), (4) participated in the school council, (5) attended a parent–teacher conference, (6) attended a school event, and (7) volunteered at the school. Significantly, each PI variable was coded in the ECLS-K:2011 data set as a dichotomous item. As a result of this coding, the frequency of PI in each of these activities cannot be discerned. Our analysis of the sociodemographic correlates of PI were conducted using 10 sociodemographic variables included in the ECLS-K:2011 data set. The first of these measures was parents’ ethnicity, which we coded into four binary variables (African American, Hispanic, Asian, and Other), with white representing the omitted reference category. The second variable was a dummy-coded measure of parents’ status as English language learners (ELLs), with native English speakers representing the omitted reference group. The third sociodemographic predictor variable was a dummy-coded measure of a caregiver’s single-parent status, with married or cohabitating parents representing the omitted reference category. The fourth variable was a dummy-coded measure of parents’ educational status, with 1 = parents who received a high school diploma or less and 0 = parents who had received at least some postsecondary education. Our fifth predictor variable was a count variable that measured the number of children parents had in the home. Our next set of predictor variables measured some of the key research-identified facilitators and barriers to PI (see, for example, Epstein, 2011; Mannan & Blackwell, 1992). Here, we analyzed the number of social contacts parents had with other parents at the school as a proxy measure for parents’ school social capital—a known facilitator for PI (Lareau, 2011). We also analyzed two dummy-coded measures of parents’ occupational status—specifically, whether parents received Temporary Assistance for Needy Families (TANF) and whether they worked for 35 hours or more each week. Last, we analyzed three dummy-coded items that measured key PI barriers—specifically, whether parents felt unwelcome at school, whether they lacked child care, and whether they faced consistent transportation barriers. The final predictor variable included in our analysis captured the social geography of parents’ home communities. This social geography construct was coded into three dichotomous items that captured whether parents lived in a suburb, town, or rural location. Parents who lived in an urban setting represented the omitted reference category for this variable. Analytic Approach Our PI profile models were estimated using a particular person-centered statistical technique, LCA. LCA is a statistical method that allows researchers to distill multiple combinations of PI activity into characteristically discrete patterns or subpopulation profiles (see, for example, Nylund et al., 2007). These profiles, also called latent classes, are identified statistically in LCA through a process known as latent class enumeration. Consistent with best practices in LCA modeling (see Masyn, 2013), we began our class enumeration procedure by estimating a one-class model using the statistical software program Mplus 7.3 (Muthén & Muthén, 2015). We then successively added latent classes to our models until such time that there were no empirical improvements in model fit. This determination of model fit was made by consulting the three primary statistical indices (that is, the Bayesian information criterion, the consistent Akaike information criterion, and the approximate weight of evidence criterion) typically used to analyze LCA models (M. A. Lawson & Masyn, 2015; Masyn, 2013). Lower values on these indices indicate an improved fit from a model with one fewer classes. Ultimately, we performed this latent class enumeration procedure on multiple subsets of ECLS-K:2011 data. We began by conducting an LCA of the full sample of 15,600 ECLS-K:2011 public school parents. The goal here was to capture a nationally representative view of parents’ school-based PI profiles. As a part of these analyses, we closely monitored the percentage of parents who were assigned to each PI profile group (that is, the class probabilities of each profile) and the model estimated chances that parents in each PI profile were involved in particular school-based PI activities (that is, each profile’s corresponding item probabilities). Once parents’ PI profiles were identified using the complete ECLS-K:2011 sample of public school parents, we performed this same LCA procedure on subsets of ECLS-K:2011 parent data that were organized according to family income. As part of these analyses, we used the family income coding available in the ECLS-K:2011 data set to divide our broader ECLS-K:2011 parent sample into four family income subgroups: (1) families with annual incomes of $30,000 or less; (2) families with annual incomes between $30,001 and $74,999; (3) families who earned between $75,000 and $99,999 annually; and (4) families who earned $100,000 or more annually. We then performed separate LCAs on each of these subgroups (subsamples) to evaluate the replicability of our original PI profile models. Second-Stage Analysis Our final research objective was to explore the sociodemographic features associated with each PI profile. To accomplish this objective, we estimated a series of logistic regression models. In these models, we used the “modal class assignment” feature in Mplus 7.3 to classify parents into particular PI profile groups. We then used these modal class assignments to create several binary outcome variables that could be analyzed by way of logistic regression modeling. We selected this logistic regression strategy in lieu of the more rigorous multinomial logistic regression approach for two reasons. First, logistic regression models yield results that are more user friendly to readers, and this accessibility is important in light of the journal’s diverse readership, including practitioners and policymakers. Second, although not detailed here due to page limits, the results yielded from these logistic regression models were similar (with respect to effect size and p value) to those generated using a more rigorous three-step approach to latent class regression modeling (see, for example, Vermunt, 2010). Last, because interest resided in understanding whether relations between our PI predictors and profiles might be moderated by family socioeconomic status (SES), we estimated separate logistic regression models for each category of family income. These models were estimated using 10 sets of imputed predictor variables, with standard errors clustered around the “school ID” variable. Improvements in model fit were evaluated using a simple likelihood ratio test. Results Our initial latent class enumeration procedure on the full sample of ECLS-K:2011 supported a four-class PI profile model (see Table 2). The item and class probabilities yielded for each PI profile are presented in Figure 1. As evident in the figure, the first PI profile was the low involvement class (20.6% of full ECLS-K:2011 parent sample). This profile includes parents whose involvement patterns were characterized by efforts to contact the school and attend parent–teacher conferences. Table 2: Results from a Latent Class Enumeration of School Involvement Profiles (Full ECLS-K:2011 Sample) (N = 15,600) Class LL BIC CAIC AWE LMR Entropy 1 –42,989 86,046 86,053 86,135 NA NA 2 –41,203 82,550 82,565 82,739.8 0.001 0.565 3 –40,888 81,997 82,020 82,288.9 0.001 0.589 4 –40,796 81,890 81,921 82,283.1 0.001 0.687 5 –40,769 81,914 81,953 82,407.9 0.149 0.655 Class LL BIC CAIC AWE LMR Entropy 1 –42,989 86,046 86,053 86,135 NA NA 2 –41,203 82,550 82,565 82,739.8 0.001 0.565 3 –40,888 81,997 82,020 82,288.9 0.001 0.589 4 –40,796 81,890 81,921 82,283.1 0.001 0.687 5 –40,769 81,914 81,953 82,407.9 0.149 0.655 Notes: Optimal class solution is indicated by the lowest yielded value for each fit index. ECLS-K:2011 = Early Childhood Longitudinal Study, Kindergarten Class of 2010–11; LL = log-likelihood value; BIC = Bayesian information criterion; CAIC = consistent Akaike information criterion; AWE = approximate weight of evidence criterion; LMR = Lo–Mendell–Rubin likelihood ratio test. Table 2: Results from a Latent Class Enumeration of School Involvement Profiles (Full ECLS-K:2011 Sample) (N = 15,600) Class LL BIC CAIC AWE LMR Entropy 1 –42,989 86,046 86,053 86,135 NA NA 2 –41,203 82,550 82,565 82,739.8 0.001 0.565 3 –40,888 81,997 82,020 82,288.9 0.001 0.589 4 –40,796 81,890 81,921 82,283.1 0.001 0.687 5 –40,769 81,914 81,953 82,407.9 0.149 0.655 Class LL BIC CAIC AWE LMR Entropy 1 –42,989 86,046 86,053 86,135 NA NA 2 –41,203 82,550 82,565 82,739.8 0.001 0.565 3 –40,888 81,997 82,020 82,288.9 0.001 0.589 4 –40,796 81,890 81,921 82,283.1 0.001 0.687 5 –40,769 81,914 81,953 82,407.9 0.149 0.655 Notes: Optimal class solution is indicated by the lowest yielded value for each fit index. ECLS-K:2011 = Early Childhood Longitudinal Study, Kindergarten Class of 2010–11; LL = log-likelihood value; BIC = Bayesian information criterion; CAIC = consistent Akaike information criterion; AWE = approximate weight of evidence criterion; LMR = Lo–Mendell–Rubin likelihood ratio test. Figure 1: View largeDownload slide Parents’ School Involvement Profiles (Full Sample, N = 15,600) Figure 1: View largeDownload slide Parents’ School Involvement Profiles (Full Sample, N = 15,600) The second PI profile yielded from LCA was the school investment class (8.1% of the full ECLS-K:2011 parent sample). This profile was characterized by parent participation in parent–teacher conferences and the PTO or PTA. The third PI profile was the school involvement class (42.6% of the full ECLS-K:2011 parent sample). Parents who fit this profile were generally involved in all school-based PI activities with the exception of school governance and PTO. The fourth and final LCA-derived PI profile was the school engagement class (28.5% of full parent sample). Members of this class were involved in nearly all of the PI activities analyzed in the statistical model. Significantly, school-engaged parents also had the highest relative chances of all parent groups of participating in their school’s advisory council. Distribution of PI Profiles across Categories of Family Income The second research objective was to examine the extent to which our original profile models might be replicated at discrete levels of family income. To evaluate this possibility, we performed separate LCAs on each of the four data subsets described earlier in the article. Each of these LCA models (not detailed here due to space limits) supported a four-class PI profile solution. Moreover, each of these LCA models yielded class-specific item probabilities that were nearly identical to those presented in Figure 1. Although the item probabilities of each profile were consistent across family income groups, the percentage of parents assigned to each PI profile group (that is, each profile’s class probabilities) varied considerably by family income. For example, as depicted in Table 3, over 50% of parents who earned $30,000 or less annually belonged to either the low-involvement (32.8%) or the school-invested (19.6%) PI profiles. However, fewer than 15% of parents with annual family incomes of between $75,000 and $99,999 belonged to these profile groups. Instead, the vast majority (80%) of families in that income category belonged to either the school-involved or the school-engagement PI profile groups. Consistent with extant research (see, for example, Lareau, 2011), these findings suggest that as family income increases, so do parent tendencies to engage in a broader configuration of school-based PI activities. Table 3: Distribution of School Involvement Profiles, by Family Income Category (N = 15,600) Profile (% of ECLS-K:2011 Population) Family Income Category (%) ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 Low involved (20.6) 32.8 24.1 9.2 9.9 School invested (8.1) 19.6 7.0 6.4 1.7 School involved (42.6) 30.6 42.0 53.0 57.0 School engaged (28.5) 17.0 27.0 31.3 31.0 Profile (% of ECLS-K:2011 Population) Family Income Category (%) ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 Low involved (20.6) 32.8 24.1 9.2 9.9 School invested (8.1) 19.6 7.0 6.4 1.7 School involved (42.6) 30.6 42.0 53.0 57.0 School engaged (28.5) 17.0 27.0 31.3 31.0 Note: ECLS-K:2011 = Early Childhood Longitudinal Study, Kindergarten Class of 2010–11. Table 3: Distribution of School Involvement Profiles, by Family Income Category (N = 15,600) Profile (% of ECLS-K:2011 Population) Family Income Category (%) ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 Low involved (20.6) 32.8 24.1 9.2 9.9 School invested (8.1) 19.6 7.0 6.4 1.7 School involved (42.6) 30.6 42.0 53.0 57.0 School engaged (28.5) 17.0 27.0 31.3 31.0 Profile (% of ECLS-K:2011 Population) Family Income Category (%) ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 Low involved (20.6) 32.8 24.1 9.2 9.9 School invested (8.1) 19.6 7.0 6.4 1.7 School involved (42.6) 30.6 42.0 53.0 57.0 School engaged (28.5) 17.0 27.0 31.3 31.0 Note: ECLS-K:2011 = Early Childhood Longitudinal Study, Kindergarten Class of 2010–11. Predictors of PI Profile Membership Results from our logistic regression analyses of PI profile membership on parents’ sociodemographic characteristics are presented in Tables 4 through 7. The logistic regression coefficients presented in these tables are odds ratios (ORs). ORs that are greater than 1 indicate a positive association between the predictor and the PI profile of interest, whereas ORs less than 1 indicate a negative or inverse association, all else being equal. Each model was initially estimated using the full complement of predictor variables detailed earlier in the article. However, our final models did not include the variables for TANF or the number of children in each parent’s household. These variables were omitted from our final models because they did not predict membership in any PI profile, and because their inclusion did not improve model fit. Table 4: Predictors of Membership in the Low-Involvement Profile, by Family Income Category (N = 15,600) Variable ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 OR SE OR SE OR SE OR SE Race or ethnicity  African American 1.15 (0.14) 1.14 (0.17) 1.47 (0.57) 1.65 (0.58)  Hispanic 0.99 (0.12) 1.22 (0.15) 0.03 (0.44) 1.42 (0.38)  Asian 1.72** (0.33) 1.44 (0.27) 2.99*** (0.97) 1.96* (0.56)  Other 1.12 (0.21) 1.01 (0.19) 0.78 (0.38) 1.59 (0.52) ELL 1.06 (0.12) 1.64*** (0.23) 2.11* (0.71) 2.87** (0.81) Single parent 1.01 (0.10) 1.13 (0.14) 0.88 (0.44) 1.48 (0.53) High school education or less 1.75*** (0.13) 1.63*** (0.15) 1.64* (0.49) 3.29*** (0.81) Number of social contacts 0.81*** (0.02) 0.79*** (0.02) 0.86** (0.05) 0.83*** (0.03) Involvement barriers  Not welcome at school 1.12 (0.17) 1.67** (0.31) 0.93 (0.55) 1.15 (0.73)  Lack of child care 1.34*** (0.11) 1.28* (0.12) 1.3 (0.30) 1.85 (0.28)  Lack of transportation 1.57*** (0.19) 1.81** (0.37) 2.12 (1.35) 1.62 (0.33)  Works >35 hours weekly 0.78** (0.07) 1.27** (0.18) 1 (0.07) 0.44 (0.21) Social geography  Suburban 0.89 (0.07) 1.02 (0.11) 1.75* (0.47) 1.45 (0.31)  Town 1.15 (0.19) 0.97 (0.16) 1.04 (0.51) 1.41 (0.57)  Rural 0.88 (0.10) 0.91 (0.12) 1.13 (0.41) 1.17 (0.29) Variable ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 OR SE OR SE OR SE OR SE Race or ethnicity  African American 1.15 (0.14) 1.14 (0.17) 1.47 (0.57) 1.65 (0.58)  Hispanic 0.99 (0.12) 1.22 (0.15) 0.03 (0.44) 1.42 (0.38)  Asian 1.72** (0.33) 1.44 (0.27) 2.99*** (0.97) 1.96* (0.56)  Other 1.12 (0.21) 1.01 (0.19) 0.78 (0.38) 1.59 (0.52) ELL 1.06 (0.12) 1.64*** (0.23) 2.11* (0.71) 2.87** (0.81) Single parent 1.01 (0.10) 1.13 (0.14) 0.88 (0.44) 1.48 (0.53) High school education or less 1.75*** (0.13) 1.63*** (0.15) 1.64* (0.49) 3.29*** (0.81) Number of social contacts 0.81*** (0.02) 0.79*** (0.02) 0.86** (0.05) 0.83*** (0.03) Involvement barriers  Not welcome at school 1.12 (0.17) 1.67** (0.31) 0.93 (0.55) 1.15 (0.73)  Lack of child care 1.34*** (0.11) 1.28* (0.12) 1.3 (0.30) 1.85 (0.28)  Lack of transportation 1.57*** (0.19) 1.81** (0.37) 2.12 (1.35) 1.62 (0.33)  Works >35 hours weekly 0.78** (0.07) 1.27** (0.18) 1 (0.07) 0.44 (0.21) Social geography  Suburban 0.89 (0.07) 1.02 (0.11) 1.75* (0.47) 1.45 (0.31)  Town 1.15 (0.19) 0.97 (0.16) 1.04 (0.51) 1.41 (0.57)  Rural 0.88 (0.10) 0.91 (0.12) 1.13 (0.41) 1.17 (0.29) Notes: OR = odds ratio; SE = standard error; ELL = English language learner. *p < .05. **p < .01. ***p < .001. Table 4: Predictors of Membership in the Low-Involvement Profile, by Family Income Category (N = 15,600) Variable ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 OR SE OR SE OR SE OR SE Race or ethnicity  African American 1.15 (0.14) 1.14 (0.17) 1.47 (0.57) 1.65 (0.58)  Hispanic 0.99 (0.12) 1.22 (0.15) 0.03 (0.44) 1.42 (0.38)  Asian 1.72** (0.33) 1.44 (0.27) 2.99*** (0.97) 1.96* (0.56)  Other 1.12 (0.21) 1.01 (0.19) 0.78 (0.38) 1.59 (0.52) ELL 1.06 (0.12) 1.64*** (0.23) 2.11* (0.71) 2.87** (0.81) Single parent 1.01 (0.10) 1.13 (0.14) 0.88 (0.44) 1.48 (0.53) High school education or less 1.75*** (0.13) 1.63*** (0.15) 1.64* (0.49) 3.29*** (0.81) Number of social contacts 0.81*** (0.02) 0.79*** (0.02) 0.86** (0.05) 0.83*** (0.03) Involvement barriers  Not welcome at school 1.12 (0.17) 1.67** (0.31) 0.93 (0.55) 1.15 (0.73)  Lack of child care 1.34*** (0.11) 1.28* (0.12) 1.3 (0.30) 1.85 (0.28)  Lack of transportation 1.57*** (0.19) 1.81** (0.37) 2.12 (1.35) 1.62 (0.33)  Works >35 hours weekly 0.78** (0.07) 1.27** (0.18) 1 (0.07) 0.44 (0.21) Social geography  Suburban 0.89 (0.07) 1.02 (0.11) 1.75* (0.47) 1.45 (0.31)  Town 1.15 (0.19) 0.97 (0.16) 1.04 (0.51) 1.41 (0.57)  Rural 0.88 (0.10) 0.91 (0.12) 1.13 (0.41) 1.17 (0.29) Variable ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 OR SE OR SE OR SE OR SE Race or ethnicity  African American 1.15 (0.14) 1.14 (0.17) 1.47 (0.57) 1.65 (0.58)  Hispanic 0.99 (0.12) 1.22 (0.15) 0.03 (0.44) 1.42 (0.38)  Asian 1.72** (0.33) 1.44 (0.27) 2.99*** (0.97) 1.96* (0.56)  Other 1.12 (0.21) 1.01 (0.19) 0.78 (0.38) 1.59 (0.52) ELL 1.06 (0.12) 1.64*** (0.23) 2.11* (0.71) 2.87** (0.81) Single parent 1.01 (0.10) 1.13 (0.14) 0.88 (0.44) 1.48 (0.53) High school education or less 1.75*** (0.13) 1.63*** (0.15) 1.64* (0.49) 3.29*** (0.81) Number of social contacts 0.81*** (0.02) 0.79*** (0.02) 0.86** (0.05) 0.83*** (0.03) Involvement barriers  Not welcome at school 1.12 (0.17) 1.67** (0.31) 0.93 (0.55) 1.15 (0.73)  Lack of child care 1.34*** (0.11) 1.28* (0.12) 1.3 (0.30) 1.85 (0.28)  Lack of transportation 1.57*** (0.19) 1.81** (0.37) 2.12 (1.35) 1.62 (0.33)  Works >35 hours weekly 0.78** (0.07) 1.27** (0.18) 1 (0.07) 0.44 (0.21) Social geography  Suburban 0.89 (0.07) 1.02 (0.11) 1.75* (0.47) 1.45 (0.31)  Town 1.15 (0.19) 0.97 (0.16) 1.04 (0.51) 1.41 (0.57)  Rural 0.88 (0.10) 0.91 (0.12) 1.13 (0.41) 1.17 (0.29) Notes: OR = odds ratio; SE = standard error; ELL = English language learner. *p < .05. **p < .01. ***p < .001. Table 5: Predictors of Membership in the School Investment Profile, by Income Category (N = 15,600) ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 Variable OR SE OR SE OR SE OR SE Race or ethnicity  African American 2.53*** (0.38) 1.72* (0.43) 8.47*** (3.37) 3.32* (1.91)  Hispanic 1.97*** (0.29) 1.62* (0.32) 2.31** (0.81) 0.93 (1.14)  Asian 2.09** (0.21) 3.16* (0.82) 2.18 (1.01) 2.34 (1.03)  Other 0.78 (0.21) 1.53 (0.51) 1.28 (0.71) 0.41 (1.06) ELL 1.93** (0.22) 2.57*** (0.51) 1.01 (0.42) 2.62* (1.21) Single parent 1.04 (0.11) 1.13 (0.26) 0.73 (0.42) 0.81 (0.64) High school education or less 1.05 (0.09) 1.14 (0.19) 1.53 (0.53) 3.31* (1.81) Number of social contacts 0.98 (0.02) 0.89*** (0.02) 0.97 (0.03) 0.85* (0.06) Involvement barriers  Not welcome at school 1.46* (0.22) 0.85 (0.35) 1.67 (1.26) 1.1 (0.22)  Lack of child care 0.89 (0.08) 1.2 (0.19) 1.44 (0.41) 1.31 (0.51)  Lack of transportation 0.89 (0.13) 0.64 (1.2) 0.77 (0.61) 2.56 (2.03)  Works >35 hours weekly 1.01 (0.10) 1.31 (0.18) 1.01 (0.07) 1.29 (0.45) Social geography  Suburban 0.87 (0.08) 1.03 (0.17) 0.68 (0.19) 1.11 (0.42)  Town 0.89 (0.19) 1.13 (0.36) 0.74 (0.39) 1.55 (1.18)  Rural 0.87 (0.12) 0.92 (0.19) 0.81 (0.25) 0.42 (0.31) ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 Variable OR SE OR SE OR SE OR SE Race or ethnicity  African American 2.53*** (0.38) 1.72* (0.43) 8.47*** (3.37) 3.32* (1.91)  Hispanic 1.97*** (0.29) 1.62* (0.32) 2.31** (0.81) 0.93 (1.14)  Asian 2.09** (0.21) 3.16* (0.82) 2.18 (1.01) 2.34 (1.03)  Other 0.78 (0.21) 1.53 (0.51) 1.28 (0.71) 0.41 (1.06) ELL 1.93** (0.22) 2.57*** (0.51) 1.01 (0.42) 2.62* (1.21) Single parent 1.04 (0.11) 1.13 (0.26) 0.73 (0.42) 0.81 (0.64) High school education or less 1.05 (0.09) 1.14 (0.19) 1.53 (0.53) 3.31* (1.81) Number of social contacts 0.98 (0.02) 0.89*** (0.02) 0.97 (0.03) 0.85* (0.06) Involvement barriers  Not welcome at school 1.46* (0.22) 0.85 (0.35) 1.67 (1.26) 1.1 (0.22)  Lack of child care 0.89 (0.08) 1.2 (0.19) 1.44 (0.41) 1.31 (0.51)  Lack of transportation 0.89 (0.13) 0.64 (1.2) 0.77 (0.61) 2.56 (2.03)  Works >35 hours weekly 1.01 (0.10) 1.31 (0.18) 1.01 (0.07) 1.29 (0.45) Social geography  Suburban 0.87 (0.08) 1.03 (0.17) 0.68 (0.19) 1.11 (0.42)  Town 0.89 (0.19) 1.13 (0.36) 0.74 (0.39) 1.55 (1.18)  Rural 0.87 (0.12) 0.92 (0.19) 0.81 (0.25) 0.42 (0.31) Notes: OR = odds ratio; SE = standard error; ELL = English language learner. *p < .05. **p < .01. ***p < .001. Table 5: Predictors of Membership in the School Investment Profile, by Income Category (N = 15,600) ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 Variable OR SE OR SE OR SE OR SE Race or ethnicity  African American 2.53*** (0.38) 1.72* (0.43) 8.47*** (3.37) 3.32* (1.91)  Hispanic 1.97*** (0.29) 1.62* (0.32) 2.31** (0.81) 0.93 (1.14)  Asian 2.09** (0.21) 3.16* (0.82) 2.18 (1.01) 2.34 (1.03)  Other 0.78 (0.21) 1.53 (0.51) 1.28 (0.71) 0.41 (1.06) ELL 1.93** (0.22) 2.57*** (0.51) 1.01 (0.42) 2.62* (1.21) Single parent 1.04 (0.11) 1.13 (0.26) 0.73 (0.42) 0.81 (0.64) High school education or less 1.05 (0.09) 1.14 (0.19) 1.53 (0.53) 3.31* (1.81) Number of social contacts 0.98 (0.02) 0.89*** (0.02) 0.97 (0.03) 0.85* (0.06) Involvement barriers  Not welcome at school 1.46* (0.22) 0.85 (0.35) 1.67 (1.26) 1.1 (0.22)  Lack of child care 0.89 (0.08) 1.2 (0.19) 1.44 (0.41) 1.31 (0.51)  Lack of transportation 0.89 (0.13) 0.64 (1.2) 0.77 (0.61) 2.56 (2.03)  Works >35 hours weekly 1.01 (0.10) 1.31 (0.18) 1.01 (0.07) 1.29 (0.45) Social geography  Suburban 0.87 (0.08) 1.03 (0.17) 0.68 (0.19) 1.11 (0.42)  Town 0.89 (0.19) 1.13 (0.36) 0.74 (0.39) 1.55 (1.18)  Rural 0.87 (0.12) 0.92 (0.19) 0.81 (0.25) 0.42 (0.31) ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 Variable OR SE OR SE OR SE OR SE Race or ethnicity  African American 2.53*** (0.38) 1.72* (0.43) 8.47*** (3.37) 3.32* (1.91)  Hispanic 1.97*** (0.29) 1.62* (0.32) 2.31** (0.81) 0.93 (1.14)  Asian 2.09** (0.21) 3.16* (0.82) 2.18 (1.01) 2.34 (1.03)  Other 0.78 (0.21) 1.53 (0.51) 1.28 (0.71) 0.41 (1.06) ELL 1.93** (0.22) 2.57*** (0.51) 1.01 (0.42) 2.62* (1.21) Single parent 1.04 (0.11) 1.13 (0.26) 0.73 (0.42) 0.81 (0.64) High school education or less 1.05 (0.09) 1.14 (0.19) 1.53 (0.53) 3.31* (1.81) Number of social contacts 0.98 (0.02) 0.89*** (0.02) 0.97 (0.03) 0.85* (0.06) Involvement barriers  Not welcome at school 1.46* (0.22) 0.85 (0.35) 1.67 (1.26) 1.1 (0.22)  Lack of child care 0.89 (0.08) 1.2 (0.19) 1.44 (0.41) 1.31 (0.51)  Lack of transportation 0.89 (0.13) 0.64 (1.2) 0.77 (0.61) 2.56 (2.03)  Works >35 hours weekly 1.01 (0.10) 1.31 (0.18) 1.01 (0.07) 1.29 (0.45) Social geography  Suburban 0.87 (0.08) 1.03 (0.17) 0.68 (0.19) 1.11 (0.42)  Town 0.89 (0.19) 1.13 (0.36) 0.74 (0.39) 1.55 (1.18)  Rural 0.87 (0.12) 0.92 (0.19) 0.81 (0.25) 0.42 (0.31) Notes: OR = odds ratio; SE = standard error; ELL = English language learner. *p < .05. **p < .01. ***p < .001. Table 6: Predictors of Membership in the School Involvement Profile, by Family Income Category (N = 15,600) Variable ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 OR SE OR SE OR SE OR SE Race or ethnicity  African American 0.37*** (0.04) 0.54*** (0.07) 0.41*** (0.1) 0.46** (0.001)  Hispanic 0.9*** (0.07) 0.75** (0.08) 0.72 (0.14) 1.03 (0.17)  Asian 0.47*** (0.09) 0.45*** (0.07) 0.88 (0.22) 0.73 (0.12)  Other 1.05 (0.18) 0.82 (0.13) 1.39 (0.36) 0.96 (0.18) ELL 0.53*** (0.05) 0.49*** (0.06) 0.78 (0.19) 0.75 (0.15) Single parent 0.83 (0.07) 0.94 (0.10) 1.23 (0.33) 1.25 (0.27) High school education or less 0.73*** (0.73) 0.86 (0.08) 1.01 (0.21) 0.74 (0.14) Number of social contacts 1.02* (0.01) 1 (0.02) 0.92** (0.02) 0.93** (0.01) Involvement barriers  Not welcome at school 0.87 (0.14) 0.87 (0.15) 1.25 (0.41) 0.75 (0.21)  Lack of child care 0.96 (0.08) 1.18 (0.10) 1.25 (0.19) 0.97 (0.11)  Lack of transportation 0.83 (0.10) 0.67 (0.67) 1.17 (0.62) 2.72 (1.45)  Works >35 hours weekly 1.25* (0.11) 0.86 (0.07) 1.43** (0.18) 1.24* (0.12) Social geography  Suburban 1.29** (0.11) 1.09 (0.17) 0.84 (0.13) 0.99 (0.12)  Town 1.01 (0.16) 1.42 (0.19) 0.95 (0.24) 0.99 (0.22)  Rural 1.31* (0.15) 1.2 (0.13) 1.27 (0.22) 1.01 (0.18) Variable ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 OR SE OR SE OR SE OR SE Race or ethnicity  African American 0.37*** (0.04) 0.54*** (0.07) 0.41*** (0.1) 0.46** (0.001)  Hispanic 0.9*** (0.07) 0.75** (0.08) 0.72 (0.14) 1.03 (0.17)  Asian 0.47*** (0.09) 0.45*** (0.07) 0.88 (0.22) 0.73 (0.12)  Other 1.05 (0.18) 0.82 (0.13) 1.39 (0.36) 0.96 (0.18) ELL 0.53*** (0.05) 0.49*** (0.06) 0.78 (0.19) 0.75 (0.15) Single parent 0.83 (0.07) 0.94 (0.10) 1.23 (0.33) 1.25 (0.27) High school education or less 0.73*** (0.73) 0.86 (0.08) 1.01 (0.21) 0.74 (0.14) Number of social contacts 1.02* (0.01) 1 (0.02) 0.92** (0.02) 0.93** (0.01) Involvement barriers  Not welcome at school 0.87 (0.14) 0.87 (0.15) 1.25 (0.41) 0.75 (0.21)  Lack of child care 0.96 (0.08) 1.18 (0.10) 1.25 (0.19) 0.97 (0.11)  Lack of transportation 0.83 (0.10) 0.67 (0.67) 1.17 (0.62) 2.72 (1.45)  Works >35 hours weekly 1.25* (0.11) 0.86 (0.07) 1.43** (0.18) 1.24* (0.12) Social geography  Suburban 1.29** (0.11) 1.09 (0.17) 0.84 (0.13) 0.99 (0.12)  Town 1.01 (0.16) 1.42 (0.19) 0.95 (0.24) 0.99 (0.22)  Rural 1.31* (0.15) 1.2 (0.13) 1.27 (0.22) 1.01 (0.18) Notes: OR = odds ratio; SE = standard error; ELL = English language learner. *p < .05. **p < .01. ***p < .001. Table 6: Predictors of Membership in the School Involvement Profile, by Family Income Category (N = 15,600) Variable ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 OR SE OR SE OR SE OR SE Race or ethnicity  African American 0.37*** (0.04) 0.54*** (0.07) 0.41*** (0.1) 0.46** (0.001)  Hispanic 0.9*** (0.07) 0.75** (0.08) 0.72 (0.14) 1.03 (0.17)  Asian 0.47*** (0.09) 0.45*** (0.07) 0.88 (0.22) 0.73 (0.12)  Other 1.05 (0.18) 0.82 (0.13) 1.39 (0.36) 0.96 (0.18) ELL 0.53*** (0.05) 0.49*** (0.06) 0.78 (0.19) 0.75 (0.15) Single parent 0.83 (0.07) 0.94 (0.10) 1.23 (0.33) 1.25 (0.27) High school education or less 0.73*** (0.73) 0.86 (0.08) 1.01 (0.21) 0.74 (0.14) Number of social contacts 1.02* (0.01) 1 (0.02) 0.92** (0.02) 0.93** (0.01) Involvement barriers  Not welcome at school 0.87 (0.14) 0.87 (0.15) 1.25 (0.41) 0.75 (0.21)  Lack of child care 0.96 (0.08) 1.18 (0.10) 1.25 (0.19) 0.97 (0.11)  Lack of transportation 0.83 (0.10) 0.67 (0.67) 1.17 (0.62) 2.72 (1.45)  Works >35 hours weekly 1.25* (0.11) 0.86 (0.07) 1.43** (0.18) 1.24* (0.12) Social geography  Suburban 1.29** (0.11) 1.09 (0.17) 0.84 (0.13) 0.99 (0.12)  Town 1.01 (0.16) 1.42 (0.19) 0.95 (0.24) 0.99 (0.22)  Rural 1.31* (0.15) 1.2 (0.13) 1.27 (0.22) 1.01 (0.18) Variable ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 OR SE OR SE OR SE OR SE Race or ethnicity  African American 0.37*** (0.04) 0.54*** (0.07) 0.41*** (0.1) 0.46** (0.001)  Hispanic 0.9*** (0.07) 0.75** (0.08) 0.72 (0.14) 1.03 (0.17)  Asian 0.47*** (0.09) 0.45*** (0.07) 0.88 (0.22) 0.73 (0.12)  Other 1.05 (0.18) 0.82 (0.13) 1.39 (0.36) 0.96 (0.18) ELL 0.53*** (0.05) 0.49*** (0.06) 0.78 (0.19) 0.75 (0.15) Single parent 0.83 (0.07) 0.94 (0.10) 1.23 (0.33) 1.25 (0.27) High school education or less 0.73*** (0.73) 0.86 (0.08) 1.01 (0.21) 0.74 (0.14) Number of social contacts 1.02* (0.01) 1 (0.02) 0.92** (0.02) 0.93** (0.01) Involvement barriers  Not welcome at school 0.87 (0.14) 0.87 (0.15) 1.25 (0.41) 0.75 (0.21)  Lack of child care 0.96 (0.08) 1.18 (0.10) 1.25 (0.19) 0.97 (0.11)  Lack of transportation 0.83 (0.10) 0.67 (0.67) 1.17 (0.62) 2.72 (1.45)  Works >35 hours weekly 1.25* (0.11) 0.86 (0.07) 1.43** (0.18) 1.24* (0.12) Social geography  Suburban 1.29** (0.11) 1.09 (0.17) 0.84 (0.13) 0.99 (0.12)  Town 1.01 (0.16) 1.42 (0.19) 0.95 (0.24) 0.99 (0.22)  Rural 1.31* (0.15) 1.2 (0.13) 1.27 (0.22) 1.01 (0.18) Notes: OR = odds ratio; SE = standard error; ELL = English language learner. *p < .05. **p < .01. ***p < .001. Table 7: Predictors of Membership in the School Engagement Profile, by Family Income Category (N = 15,600) Variable ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 OR SE OR SE OR SE OR SE Race or ethnicity  African American 1.54** (0.22) 1.56** (0.22) 0.85 (0.24) 1.65* (0.39)  Hispanic 1.1 (0.15) 1.07 (0.11) 1.14 (0.24) 0.79 (0.14)  Asian 0.68 (0.17) 1.12 (0.21) 0.48* (0.17) 1.11 (0.21)  Other 0.82 (0.22) 1.16 (0.20) 0.65 (0.20) 0.76 (0.16) ELL 0.98 (0.12) 0.82 (0.12) 0.71 (0.19) 0.44*** (0.11) Single parent 1.08 (0.11) 0.92 (0.11) 0.91 (0.26) 0.57* (0.15) High school education or less 0.61*** (0.06) 0.70*** (0.07) 0.68 (0.16) 0.58* (0.13) Number of social contacts 1.18** (0.02) 1.14* (0.02) 1.16* (0.03) 1.14*** (0.03) Involvement barriers  Not welcome at school 0.51** (0.11) 0.69 (0.15) 0.61 (0.41) 1.13 (0.35)  Lack of child care 0.72** (0.08) 0.58*** (0.06) 0.62** (0.19) 0.81 (0.11)  Lack of transportation 0.57** (0.11) 0.9 (0.21) 0.19* (0.63) 0.25 (0.21)  Works >35 hours weekly 1.03 (0.11) 0.88 (0.07) 0.66** (0.18) 0.72*** (0.07) Social geography  Suburban 0.92 (0.09) 0.87 (0.09) 1.13 (0.13) 0.89 (0.12)  Town 0.87 (0.19) 0.63** (0.11) 1.13 (0.23) 0.89 (0.22)  Rural 0.91 (0.13) 0.9 (0.10) 0.77 (0.22) 0.78 (0.13) Variable ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 OR SE OR SE OR SE OR SE Race or ethnicity  African American 1.54** (0.22) 1.56** (0.22) 0.85 (0.24) 1.65* (0.39)  Hispanic 1.1 (0.15) 1.07 (0.11) 1.14 (0.24) 0.79 (0.14)  Asian 0.68 (0.17) 1.12 (0.21) 0.48* (0.17) 1.11 (0.21)  Other 0.82 (0.22) 1.16 (0.20) 0.65 (0.20) 0.76 (0.16) ELL 0.98 (0.12) 0.82 (0.12) 0.71 (0.19) 0.44*** (0.11) Single parent 1.08 (0.11) 0.92 (0.11) 0.91 (0.26) 0.57* (0.15) High school education or less 0.61*** (0.06) 0.70*** (0.07) 0.68 (0.16) 0.58* (0.13) Number of social contacts 1.18** (0.02) 1.14* (0.02) 1.16* (0.03) 1.14*** (0.03) Involvement barriers  Not welcome at school 0.51** (0.11) 0.69 (0.15) 0.61 (0.41) 1.13 (0.35)  Lack of child care 0.72** (0.08) 0.58*** (0.06) 0.62** (0.19) 0.81 (0.11)  Lack of transportation 0.57** (0.11) 0.9 (0.21) 0.19* (0.63) 0.25 (0.21)  Works >35 hours weekly 1.03 (0.11) 0.88 (0.07) 0.66** (0.18) 0.72*** (0.07) Social geography  Suburban 0.92 (0.09) 0.87 (0.09) 1.13 (0.13) 0.89 (0.12)  Town 0.87 (0.19) 0.63** (0.11) 1.13 (0.23) 0.89 (0.22)  Rural 0.91 (0.13) 0.9 (0.10) 0.77 (0.22) 0.78 (0.13) Notes: OR = odds ratio; SE = standard error; ELL = English language learner. *p < .05. **p < .01. ***p < .001. Table 7: Predictors of Membership in the School Engagement Profile, by Family Income Category (N = 15,600) Variable ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 OR SE OR SE OR SE OR SE Race or ethnicity  African American 1.54** (0.22) 1.56** (0.22) 0.85 (0.24) 1.65* (0.39)  Hispanic 1.1 (0.15) 1.07 (0.11) 1.14 (0.24) 0.79 (0.14)  Asian 0.68 (0.17) 1.12 (0.21) 0.48* (0.17) 1.11 (0.21)  Other 0.82 (0.22) 1.16 (0.20) 0.65 (0.20) 0.76 (0.16) ELL 0.98 (0.12) 0.82 (0.12) 0.71 (0.19) 0.44*** (0.11) Single parent 1.08 (0.11) 0.92 (0.11) 0.91 (0.26) 0.57* (0.15) High school education or less 0.61*** (0.06) 0.70*** (0.07) 0.68 (0.16) 0.58* (0.13) Number of social contacts 1.18** (0.02) 1.14* (0.02) 1.16* (0.03) 1.14*** (0.03) Involvement barriers  Not welcome at school 0.51** (0.11) 0.69 (0.15) 0.61 (0.41) 1.13 (0.35)  Lack of child care 0.72** (0.08) 0.58*** (0.06) 0.62** (0.19) 0.81 (0.11)  Lack of transportation 0.57** (0.11) 0.9 (0.21) 0.19* (0.63) 0.25 (0.21)  Works >35 hours weekly 1.03 (0.11) 0.88 (0.07) 0.66** (0.18) 0.72*** (0.07) Social geography  Suburban 0.92 (0.09) 0.87 (0.09) 1.13 (0.13) 0.89 (0.12)  Town 0.87 (0.19) 0.63** (0.11) 1.13 (0.23) 0.89 (0.22)  Rural 0.91 (0.13) 0.9 (0.10) 0.77 (0.22) 0.78 (0.13) Variable ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 OR SE OR SE OR SE OR SE Race or ethnicity  African American 1.54** (0.22) 1.56** (0.22) 0.85 (0.24) 1.65* (0.39)  Hispanic 1.1 (0.15) 1.07 (0.11) 1.14 (0.24) 0.79 (0.14)  Asian 0.68 (0.17) 1.12 (0.21) 0.48* (0.17) 1.11 (0.21)  Other 0.82 (0.22) 1.16 (0.20) 0.65 (0.20) 0.76 (0.16) ELL 0.98 (0.12) 0.82 (0.12) 0.71 (0.19) 0.44*** (0.11) Single parent 1.08 (0.11) 0.92 (0.11) 0.91 (0.26) 0.57* (0.15) High school education or less 0.61*** (0.06) 0.70*** (0.07) 0.68 (0.16) 0.58* (0.13) Number of social contacts 1.18** (0.02) 1.14* (0.02) 1.16* (0.03) 1.14*** (0.03) Involvement barriers  Not welcome at school 0.51** (0.11) 0.69 (0.15) 0.61 (0.41) 1.13 (0.35)  Lack of child care 0.72** (0.08) 0.58*** (0.06) 0.62** (0.19) 0.81 (0.11)  Lack of transportation 0.57** (0.11) 0.9 (0.21) 0.19* (0.63) 0.25 (0.21)  Works >35 hours weekly 1.03 (0.11) 0.88 (0.07) 0.66** (0.18) 0.72*** (0.07) Social geography  Suburban 0.92 (0.09) 0.87 (0.09) 1.13 (0.13) 0.89 (0.12)  Town 0.87 (0.19) 0.63** (0.11) 1.13 (0.23) 0.89 (0.22)  Rural 0.91 (0.13) 0.9 (0.10) 0.77 (0.22) 0.78 (0.13) Notes: OR = odds ratio; SE = standard error; ELL = English language learner. *p < .05. **p < .01. ***p < .001. Predictors of Low-Involvement Profile Membership Table 4 presents our logistic regression analysis of the sociodemographic features of the low-involvement PI profile group. These analyses indicated that, for most family income groups, the significant predictors of low involvement were having a high school education or less (OR range = 1.63–3.29), knowing fewer parents at the school (OR range = 0.79–0.86), and being Asian (relative to being white) (OR range = 1.42–2.99). Meanwhile, parents with family incomes less than $75,000 annually were more likely belong to the low-involvement profile group when they experienced transportation (OR range = 1.57–1.81) or child care barriers (OR range = 1.28–1.34). Last, these analyses revealed somewhat complex, nuanced relationship between low-involvement profile membership and full-time work. Specifically, parents with family incomes of $30,000 or less were significantly less likely to belong to low-involvement profile when they worked full time (OR = 0.78, p < .01). In contrast, parents residing in the $30,001 to $74,999 income category were significantly more likely to be low involved when they were engaged in full-time work (OR = 1.27, p < .01). Predictors of School Investment Table 5 presents our logistic regression analysis of the sociodemographic features of the school investment PI profile group. These analyses indicated that, across most family income groups, the significant predictors of school investment were being ELL (OR range = 1.93–2.62) and being African American (relative to being white) (OR range = 1.72–8.47). These models also indicated that Hispanic (OR range = 1.62–2.31) and Asian parents (OR range = 2.09–3.16) living in families earning less than $75,000 annually were significantly more likely than white parents to belong to the school investment profile. Last, parents in families earning $30,000 or less annually were significantly more likely to belong to the school investment profile when they felt unwelcome by the school (OR = 1.46, p < .05). Predictors of School Involvement Table 6 presents our logistic regression analysis of the sociodemographic predictors of the school involvement profile membership. As indicated in the table, these models generally supported a significant, positive relationship between full-time work and school involvement (OR range = 0.86–1.43). However, these same models yielded a significant negative relationship between being an ethnic minority parent and school involvement profile membership. Specifically, our models indicated that, across all levels of family income, African American parents were significantly less likely than white parents to belong to the school involvement group (OR range = 0.37–0.54). In the same vein, Hispanic (OR range = 0.75–0.90) and Asian (OR range = 0.45–0.47) parents earning less than $75,000 annually were significantly less likely than white parents to belong to the school involvement profile group, all else being equal. Last, when the analysis was restricted to parents living in the lowest family income stratum, our models indicated that parents who were ELL (OR = 0.53, p < .001) or who had a high school diploma or less (OR = 0.73, p < .001) were significantly less likely to be school involved. Meanwhile, these models indicated that low-income school-involved parents were more likely to live in the suburbs (OR = 1.29, p < .01) or rural communities (OR = 1.31. p < .05) than urban locations. Predictors of School Engagement Table 7 presents our logistic regression analysis of the sociodemographic predictors of the school engagement profile membership. These analyses indicated that, across most family income categories, school-engaged parents were more likely to be African American than white (OR range = 0.85–1.65), and they were also more likely to have a higher number of social contacts at the school (OR range = 1.14–1.18). Within the lowest income stratum, school-engaged parents were less likely to experience common PI barriers such as child care (OR = 0.72, p < .01) and transportation (OR = 0.57, p < .01) constraints, and they were also highly unlikely to have a high school diploma or less (OR = 0.61, p < .05). Last, parents who worked full time and lived in families earning $75,000 or more were significantly less likely to belong to the school engagement profile (OR range = 0.66–0.72). This latter finding signals the amount of time that may be needed for parents to engage in a full range of school-based PI activities. Summary and Implications This study used LCA to analyze subpopulation profiles of PI in elementary school. These analyses were conducted using a nationally representative sample of first-grade parents who had children attending public schools in the United States. Four characteristically distinct profiles of PI were yielded from the LCA approach: (1) A low involvement profile characterized by PI in a limited number of school-based activities; (2) a school investment profile characterized primarily by participation in the PTA or other PTOand parent–teacher conferences; (3) a school involvement profile characterized by PI in multiple activities, but not governance; and (4) a school engagement profile characterized by parent participation in a full range of school-based PI activities. Following our initial LCA on the complete ECLS-K:2011 sample of public school parents, the next step in the study was to evaluate the relative consistency of our PI profile findings across different categories of family income. Here, our models indicated that the characteristic features (that is, the item probabilities) of each PI profile remained remarkably consistent across family income groups. However, the proportion of parents assigned to each profile group varied according to parents’ SES. Specifically, parents living in families earning $30,000 or less annually were disproportionately assigned to those profiles characterized by PI in a limited number of school-based activities (that is, the low involvement and school investment profiles), whereas the vast majority of parents from families earning $75,000 or more belonged to either the school involvement or the school engagement profile groups. Several important implications for school social work practice accompany these diverse PI profile findings. The first of these implications relates to extant research regarding the PI practices of low-income families. As described earlier in the article, prior PI research has generally characterized low-income parents as less involved than their more affluent counterparts (Lareau, 2011). In practice-embedded research contexts, this finding has been linked to deficit-oriented views regarding low-income parents’ core values and mores, including educator doubts about whether low-income parents truly value their children’s education (M. A. Lawson, 2003). In contrast to these deficit-oriented findings, our LCA models indicated that nearly half the parents earning $30,000 or less annually belonged to either the school involved or the school engagement PI profile groups. Moreover, although these models assigned a disproportionate number of low-income parents to the low involvement and school investment PI profile groups, it is important to note that these two parent subpopulations were characterized, in part, by PI in at least one school-based activity. An immediate implication is that all low-income parents have important PI-related strengths that schools can and should build from to support PI and promote students’ school success. School social workers, with their unique training and experience in strengths-based and empowerment-based practice modalities, can help schools learn to better capitalize on parents’ existing strengths, knowledge, and interests (see, for example, Alameda-Lawson & Lawson, 2016). Recommendations for Improving PI in Low-Income Schools One specific way that school social workers and other school community professionals might go about improving PI in low-income schools is to enhance opportunities for parent participation in school governance activities, especially PTOs. PTOs may represent an important starting point for school social work interventions because the data indicate that low-income African American, Latino, and Asian parents, as well as low-income parents classified as ELL, are initially drawn to them. At the same time, it is important to note that the data indicated that an important subpopulation of PTO parents (that is, low-income, minority parents who belong to the school investment profile group) represented the most likely candidates to report that they felt unwelcome at the school. This latter finding, in particular, underscores the need for social work–assisted programs and supports that help ethnically and linguistically diverse families negotiate the complex (and often difficult) practices, boundaries, and cultures of schools (see, for example, M. A. Lawson & Alameda-Lawson, 2012). It also highlights opportunities for social workers to help educators develop programs and policies that can accommodate the strengths, needs, and challenges of their constituent families and communities (Alameda-Lawson, Lawson, & Lawson, 2010). PI as an Inherently Social Activity One of the more important findings yielded from this study concerns the relationship between PI and parents’ social networks. The importance of parents’ social networks was particularly evident in findings that linked PI profile membership to the number of social contacts that parents enjoyed at the school. Specifically, the data indicated that parents who knew more parents at the school were significantly more likely to belong to the school involvement and school engagement profiles, whereas membership in the low involvement profile was counterindicated by this proxy indicator for parents’ school social capital. The ready implication is that relational or “collective” PI practices and models may hold special promise for enhancing PI in low-income school communities (see, for example, Alameda-Lawson, Lawson, & Lawson, 2013; Warren et al., 2009). School social workers can facilitate the development of these collectivist PI approaches by forging school–family–community partnerships with faith-based groups, youth sports leagues, and other neighborhood-based efforts that can connect heretofore isolated parents to others who have children at the school (Alameda-Lawson & Lawson, 2016; Lareau, 2011; Warren et al., 2009). PI as a Specialized Intervention A final important finding from this study concerns the complex relationship that emerged between PI profile membership and full-time work. This finding was both curious and novel because prior research has identified full-time work as a barrier to parents’ school-based PI (see, for example, M. A. Lawson, 2003). However, in this study, full-time work appeared to predict PI for some parent subpopulations (that is, the school involvement profile), whereas for others, it appeared to have the opposite effect (that is, the school engagement group). In addition to these findings, this study illuminated the conditions in which family income might moderate the association between PI and full-time work. For example, our models indicated that parents living in the lowest income stratum were significantly less likely to belong to the low involvement profile group when they worked full time. In contrast, full-time work was a significant predictor of low involvement among parents residing in the $30,001 to $74,999 family income category. Unfortunately, due to the nature of the ECLS-K:2011 data set, we can only speculate as to why full-time work might act as a PI barrier for select parent subpopulations. For example, it is possible that low-involved parents in the $30,001 to $74,999 income group are more likely than parents in lower income subgroups to face stringent work hours. Relatedly, it is also possible that low-involved parents in this income category represent a parent subpopulation that is more likely to work more than one job. The subpopulation PI profile differences revealed in this study highlight the limitations of standardized school–family interventions and showcase needs for more tailored PI interventions and related supports. For instance, findings from this study indicate that large-scale, standardized PI efforts such as the National Network of Partnership Schools (see, for example, Epstein, 2011)—which encourages family involvement in the full spectrum of PI activities explored in this study—may be unlikely to realize their full potential when parents’ lives are consumed by daily hassles, such as child care, food, housing, and transportation constraints (Posey-Maddox & Haley-Lock, 2016). In contrast, more discrete efforts like Lynn McDonald’s Families and Schools Together (FAST) program, with its special emphasis on enhancing parent–school interactions and increasing parent and family social capital, may be especially useful in helping previously isolated parents forge needed social connections both at school and in the community (see, for example, Kratochwill, McDonald, Levin, Scalia, & Coover, 2009). However, because the service capacity of programs like FAST is typically limited to a handful of school families (see also Webster-Stratton & Reid, 2010), social workers can and should take leadership roles in helping these programs expand into more comprehensive, schoolwide, community-building efforts. Limitations and Future Research Directions Although this study offered an enhanced, person-centered view of parents’ PI practices and profiles, there are several limitations to this study’s design that serve to qualify our results. It is important to note that these limitations double as future research needs. To begin, one of the primary limitations of this study stems from the quality and nature of the ECLS-K:2011 survey questions, especially the coding of school-based PI activities into binary items and dichotomous variables. As a result of the ECLS-K:2011 coding strategy, our parent profile models may have masked important distinctions between parents who had little or occasional involvement in certain PI activities and those who engaged in those activities on a consistent basis. For this reason, future PI research would benefit from person-centered studies that attend to both the breadth and the depth of parental involvement in elementary school. The second limitation concerned our exclusive reliance on the cross-sectional PI experiences of parents of first-grade children. This narrow focus was driven by the particularities of the ECLS-K:2011 data set, but it was not without consequence. For example, because our analysis focused exclusively on the cross-sectional experiences of first-grade parents, we are not able to generalize our findings to parents of elementary school-age children in grades 2 through 6, nor are we able to understand how parents’ PI profiles might change over time. In addition, because of limitations in the ECLS-K:2011 data set, we were not able to discern how other important sociogeographic features and factors—such as the distance between parents’ homes and their children’s school—might influence parents’ school involvement patterns and practices. For this reason, longitudinal PI studies are needed, especially those that detail how parents’ sociogeographies might influence the quality and nature of their PI practices and profiles. The third limitation concerns the primary characteristics of this study’s parent sample. Specifically, because this study analyzed the PI practices of each child’s “primary parent,” the PI practices and profiles of other adults and caregivers in the family (for example, spouses, aunts and uncles, friends, and grandparents) were not included in our LCA models. Given the importance of family assets for children’s educational success and well-being (Lareau, 2011), future research should target a more comprehensive and family-centered view of PI practices and outcomes. The final limitation of this study concerns the absence of home-based PI activities in our statistical models. These activities were not included in the present study because of page limits. Ultimately, these recommendations, and the design limitations from which they are derived, help to highlight the exploratory nature of the present research. All of this study’s results should therefore be considered preliminary until replicated in both local and national parent samples, using a more nuanced array of PI indicators and a more comprehensive set of controls. Tania Alameda-Lawson, PhD, is associate professor, School of Social Work, University of Alabama, Box 870314, Tuscaloosa, AL 35487-0314; e-mail: [email protected]. Michael A. Lawson, PhD, is assistant professor, College of Education, University of Alabama, Tuscaloosa References Alameda-Lawson , T. , & Lawson , M. A. ( 2016 ). Ecologies of collective parent engagement in urban education . Urban Education . Advance online publication. doi:10.1177/0042085916636654 Alameda-Lawson , T. , Lawson , M. A. , & Lawson , H. A. ( 2010 ). Social workers’ roles in facilitating the collective involvement of low-income, culturally diverse parents in an elementary school . Children & Schools, 32 , 172 – 182 . Google Scholar CrossRef Search ADS Alameda-Lawson , T. , Lawson , M. A. , & Lawson , H. A. ( 2013 ). An innovative collective parent engagement model for families and neighborhoods in arrival cities . Journal of Family Strengths, 13 ( 1 ), 1 – 24 . Retrieved from http://digitalcommons.library.tmc.edu/jfs/vol13/iss1/1 Epstein , J. L. ( 2011 ). School, family, and community partnerships: Preparing educators and improving schools . New York : Westview Press . Fan , X. , & Chen , M. ( 2001 ). Parental involvement and students’ academic achievement: A meta-analysis . Educational Psychology Review, 13 , 1 – 22 . Google Scholar CrossRef Search ADS Holloway , S. D. , Campbell , E. J. , Nagase , E. , Kim , S. , Suzuki , S. , Wang , Q. , et al. ( 2016 ). Parenting self-efficacy and parental involvement: Mediators or moderators between socioeconomic status and children’s academic competence in Japan and Korea? Research in Human Development, 13 , 258 – 272 . doi:10.1080/15427609.2016.1194710 Google Scholar CrossRef Search ADS Jeynes , W. ( 2011 ). Parental involvement and academic success . New York : Taylor & Francis/Routledge . Kratochwill , T. , McDonald , L. , Levin , J. , Scalia , P. , & Coover , G. ( 2009 ). Families and schools together: An experimental study of multi-family support groups for children at risk . Journal of School Psychology, 47 , 245 – 265 . Google Scholar CrossRef Search ADS Lareau , A. ( 2011 ). Unequal childhoods: Class, race, and family life, with an update a decade later ( 2nd ed. ). Berkeley : University of California Press . Lawson , H. , Alameda-Lawson , T. , Lawson , M. , Briar-Lawson , K. , & Wilcox , K. ( 2014 ). Three parent and family interventions for rural schools and communities . Journal of Education and Human Development, 3 ( 3 ), 59 – 78 . Retrieved from http://jehdnet.com/journals/jehd/Vol_3_No_3_September_2014/5.pdf Google Scholar CrossRef Search ADS Lawson , M. A. ( 2003 ). School-family relations in context: Parent and teacher perceptions of parent involvement . Urban Education, 38 ( 1 ), 77 – 133 . Google Scholar CrossRef Search ADS Lawson , M. A. , & Alameda-Lawson , T. ( 2012 ). A case study of school-linked, collective parent engagement . American Educational Research Journal, 49 , 651 – 684 . Google Scholar CrossRef Search ADS Lawson , M. A. , & Masyn , K. ( 2015 ). Analyzing profiles and predictors of students’ social-ecological engagement . AERA Open, 1 ( 4 ), 1 – 37 . doi:10.1177/2332858415615856 Google Scholar CrossRef Search ADS Lee , J. S. , & Bowen , N. K. ( 2006 ). Parent involvement, cultural capital, and the achievement gap among elementary school children . American Educational Research Journal, 43 , 193 – 215 . Google Scholar CrossRef Search ADS Mannan , G. , & Blackwell , J. ( 1992 ). Parent involvement: Barriers and opportunities . Urban Review, 24 , 219 – 228 . Google Scholar CrossRef Search ADS Mapp , K. L. , & Kuttner , P. J. ( 2014 ). Partners in education: A dual capacity-building framework for family-school partnerships . Austin, TX, & Washington, DC : SEDL & U.S. Department of Education . Retrieved from http://www.sedl.org/pubs/framework/ Masyn , K. ( 2013 ). Latent class analysis and finite mixture modeling. In T. Little (Ed.), Oxford handbook of quantitative methods (Vol. 2, pp. 551 – 611 ). Oxford, UK : Oxford University Press . Moss , P. A. ( 2012 ). Exploring the macro-micro dynamic in data use practice . American Journal of Education, 118 , 223 – 232 . Google Scholar CrossRef Search ADS Mulligan , G. M. , Hastedt , S. , & McCarroll , J. C. ( 2012 ). First-time kindergartners in 2010-11: First findings from the kindergarten rounds of the Early Childhood Longitudinal Study, Kindergarten Class of 2010-11 (ECLS-K:2011) . Washington, DC : U.S. Department of Education, National Center for Education Statistics . Retrieved from http://nces.ed.gov/pubsearch Muthén , L. , & Muthén , B. ( 2015 ). Mplus user’s guide ( 7th ed. ). Los Angeles : Author . Nylund , K. , Bellmore , A. , Nishina , A. , & Graham , S. ( 2007 ). Subtypes, severity, and structural stability of peer victimization: What does latent class analysis say? Child Development, 78 , 1706 – 1722 . Google Scholar CrossRef Search ADS Park , S. , & Holloway , D. ( 2017 ). The effects of school-based parental involvement on academic achievement at the child and elementary school level: A longitudinal study . Journal of Educational Research, 110 ( 1 ), 1 – 16 . doi:10.1080/00220671.2015.1016600 Google Scholar CrossRef Search ADS Posey-Maddox , L. , & Haley-Lock , A. ( 2016 ). One size does not fit all: Understanding parent engagement in the contexts of work, family, and public schooling . Urban Education . Advance online publication. doi:10.1177/0042085916660348 Robinson , K. , & Harris , A. L. ( 2014 ). The broken compass: Parental involvement with children’s education . Cambridge, MA : Harvard University Press . Google Scholar CrossRef Search ADS Rogers , M. A. , Theule , J. , Ryan , B. A. , Adams , G. R. , & Keating , L. ( 2009 ). Parental involvement and children’s school achievement: Evidence for mediating processes . Canadian Journal of School Psychology, 24 , 34 – 57 . Google Scholar CrossRef Search ADS Vermunt , J. K. ( 2010 ). Latent class modeling with covariates: Two improved “three step” approaches . Political Analysis, 18 , 450 – 469 . Google Scholar CrossRef Search ADS Warren , M. , Hong , S. , Rubin , C. , & Uy , P. ( 2009 ). Beyond the bake sale: A community-based relational approach to parent engagement in schools . Teachers College Record, 111 , 2209 – 2254 . Webster-Stratton , C. , & Reid , M. J. ( 2010 ). A school-family-community partnership: Addressing multiple risk factors to improve school readiness. In S. Christenson & A. Reschly (Eds.), Handbook of school-family partnerships (pp. 204 – 227 ). New York : Routledge . © 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/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Social Work Research Oxford University Press

A Latent Class Analysis of Parent Involvement Subpopulations

Alameda-Lawson, Tania; Lawson, Michael A
Social Work Research , Volume Advance Article (2) – Apr 9, 2018
Download PDF
13 pages

Loading next page...
 
/lp/ou_press/a-latent-class-analysis-of-parent-involvement-subpopulations-XxB801Pa92

References (28)

Publisher
Oxford University Press
Copyright
© 2018 National Association of Social Workers
ISSN
1070-5309
eISSN
1545-6838
DOI
10.1093/swr/svy008
Publisher site
See Article on Publisher Site

Abstract

Abstract Using nationally representative data from the Early Childhood Longitudinal Study, Kindergarten Class of 2010–11, this study relied on latent class analysis (LCA) to advance a subpopulation view of parent involvement (PI) in elementary school. Four PI subpopulation profiles were yielded using the LCA approach. Two of these parent subpopulations were involved in a limited number of school-based PI activities. Two others were involved in multiple activities at the school. It is significant that additional latent class regression analyses indicated that membership in these PI profile groups could be predicted by parents’ sociodemographic characteristics, especially their ethnicity, occupational status, family income, and social capital. Together, these findings highlight needs for school social workers to help schools develop PI programs and policies that are more nuanced. PI initiatives need to be tailored to fit the characteristics of particular parent subpopulations and particular school community contexts. Parent involvement (PI) is widely viewed as a critical ingredient for children’s academic achievement and overall school success (see, for example, Mapp & Kuttner, 2014). Yet, in spite of its known importance, PI remains weak in many school communities (Warren, Hong, Rubin, & Uy, 2009). This challenge of “low” PI is most frequently reported in low-income school communities, especially those urban and rural places challenged by social isolation and exclusion dynamics, racism, and other forms of concentrated disadvantage (Lareau, 2011). Two recent developments hold promise for enhancing PI in these and other challenged places. One is data-driven decision making (see, for example, Moss, 2012). The other is intervention science, which frames PI as a specialized intervention (H. Lawson, Alameda-Lawson, Lawson, Briar-Lawson, & Wilcox, 2014). Together, these twin developments signal opportunities for researchers to develop data-driven PI models that are action oriented and for school social workers, educators, and other school professionals to use these models to develop tailor-made PI strategies and interventions. The current study of school-based PI was designed to capitalize on this important opportunity. It proceeded with a person-centered approach to statistical analysis. Person-centered statistical methods (for example, cluster analysis, latent class analysis [LCA]) are salient for today’s PI research, practice, and policy agenda because they allow researchers to distill parents’ involvement in a broad range of activities into identifiable patterns or subpopulation profiles (Nylund, Bellmore, Nishina, & Graham, 2007). Demonstrable benefits follow. For instance, instead of rough-cut, catch-all categorizations of “the parents” that force a reliance on singular or standardized PI practices, person-centered models enable school social workers and other professionals to proceed with interventions that are customized to identified subpopulations. This study’s three research objectives were structured to illustrate this potential. The purposes were (a) to identify and describe subpopulation profiles of PI, (b) to explore the influence of family income on each PI profile, and (c) to identify and describe the sociodemographic characteristics associated with each PI profile group. Rationale: Needs and Opportunities in the Related Literature Educational leaders, researchers, and policymakers assert that PI is a driver for children’s academic achievement (Epstein, 2011; Lee & Bowen, 2006). Decades of correlational research provide the foundation for this enduring claim. Study after study has revealed consistent, positive associations between PI and children’s school outcomes, even when variables related to social class and ethnic group affiliation have been controlled in study designs (see, for example, Epstein, 2011; Jeynes, 2011). However, some recent research challenges PI’s efficacy for improving child and school outcomes. For example, a line of sociological research suggests that conventional PI activities (for example, volunteering at the school or participating in the Parent–Teacher Association [PTA]) may not attend to the strengths, needs, and challenges of low-income parents (see, for example, Lareau, 2011). Moreover, whereas much educational practice and policy has followed PI’s long-standing association with children’s academic achievement outcomes, recent research suggests that this relationship may not be automatic. In fact, some of the more nascent quantitative PI studies from sociology and education have yielded nonsignificant relationships between PI in conventional activities like volunteering at the school and academic achievement (see, for example, Robinson & Harris, 2014), and others have yielded significant inverse associations between PI activities like helping kids with homework and children’s academic outcomes (Robinson & Harris, 2014; Rogers, Theule, Ryan, Adams, & Keating, 2009). What factors might explain these emergent, contradictory findings? One answer to this question may lie in the often limited ways that PI is operationalized and then analyzed in quantitative research designs. For example, one strand of extant quantitative research operationalizes PI as the frequency of involvement in discrete activities such as volunteering at the school or participating in the PTA (see, for example, Robinson & Harris, 2014). In this activity-centered approach, researchers typically use regression modeling to explore the relationship between PI and academic achievement (see, for example, Park & Holloway, 2017). The goal is to isolate the effect that particular PI activities may exert on children’s academic outcomes by controlling for the influence of other PI activities in the statistical model. The primary strength of this activity-centered approach is that it helps practitioners and policymakers identify those PI activities that, on average, carry the most consequence for enhancing children’s academic outcomes. The primary limitation of this strategy is that it can mask the possibility that, in the real world, improvements in children’s school learning may not follow PI in singular activities or events, but instead may reflect the cumulative result of PI in multiple forms of activity (for example, volunteering at the school and participating in the PTA and helping children with homework). In light of these limitations, some quantitative researchers have operationalized PI more broadly to include parent participation in multiple school-based and home-based activities (see, for example, Fan & Chen, 2001). In these studies, scholars typically create a composite view of PI by summing the values of parents’ survey responses together to form a summative PI score (see, for example, Holloway et al., 2016). The strength of this “summative approach” is that it increases the variance of researchers’ PI variable, and this increased variance allows researchers to more precisely estimate PI’s relationship with children’s academic outcomes. The primary limitation is that it can create instances where parents share the same “PI score” as others who are involved in a qualitatively different set of PI activities. To illustrate this possibility, consider the following simple case example. In this scenario, parent A receives an involvement score of 8 by virtue of reporting that she “strongly agrees” that she participates regularly in the PTA (4) and the school-site council (4), but not other activities like back-to-school nights and open houses. Meanwhile, parent B receives the same PI score of “8” by strongly agreeing that she participates in back-to-school nights (4) and open houses (4), but not other PI activities like school governance. Thus, to the extent that different configurations of PI activity matter for children’s academic outcomes, the summative strategy is largely ill-equipped to capture their influence. Moreover, the lack of measurement precision inherent in this approach may explain why recent studies of the PI–academic achievement relationship have yielded inconsistent results. This study’s subpopulation profile models help address these aforementioned research challenges in two important ways. First, they provide a unique data-driven view of the pathways for improving PI. Second, they advance a more integrative and ecologically valid framework for analyzing PI’s relationship with children’s academic achievement outcomes. Method Participants in this study were recruited into the Early Childhood Longitudinal Study, Kindergarten Class of 2010–11 (ECLS-K:2011), a nationally representative cohort study of children attending both public and private elementary schools in the United States (Mulligan, Hastedt, & McCarroll, 2012). The ECLS-K:2011 provides rich data on children and parents’ early school experiences beginning with kindergarten and following children through fifth grade (see also Holloway et al., 2016). The sample for the present study was delimited to about 15,600 parents who had first-grade children attending 500 public elementary schools in the United States. The PI experiences of first-grade children are featured in this study because the ECLS-K:2011 administered its PI survey to parents when their children were in the first grade. Descriptive statistics of our ECLS-K:2011 parent sample are provided for readers in Table 1. All parent data were drawn using a restricted ECLS-K:2011 data file. Table 1: Descriptive Statistics of Primary Study Variables (N = 15,600) Demographic Variable M SE Min Max African American .123 .003 0 1 Hispanic .251 .004 0 1 Asian .046 .001 0 1 Other .058 .002 0 1 White .464 .001 0 1 English language learner .175 .003 0 1 Single parent .166 .003 0 1 Married .512 .003 0 1 Unmarried with domestic partner .322 .004 0 1 Number of parental contacts at school 3.55 .031 0 40 Parent education of high school or less .275 .004 0 1 Receives Temporary Assistance for Needy Families .06 .002 0 1 Income is $30,000 a year or less .363 .002 0 1 Income is between $30,001 and $74,999 .341 .004 0 1 Income is between $75,000 and $99,999 .123 .003 0 1 Earns $100,000 or more per year .173 .003 0 1 Social geography  Suburb .333 .004 0 1  Town .109 .003 0 1  Rural .234 .004 0 1  Urban .324 .020 0 1 Parent involvement barriers  Not feeling welcome at school .059 .002 0 1  Lack of child care .253 .004 0 1  Lack of transportation .058 .002 0 1  Working >35 hours a week .352 .004 0 1 Latent class manifest variables  Parent has contacted the school .77 .004 0 1  Parent attends school open house .806 .004 0 1  Parent participates in the Parent–Teacher Association or other parent–teacher organization .366 .004 0 1  Parent is involved in school council .126 .004 0 1  Parent attends parent–teacher conferences .951 .002 0 1  Parent attends school events .777 .004 0 1  Parent volunteers at the school .558 .003 0 1 Demographic Variable M SE Min Max African American .123 .003 0 1 Hispanic .251 .004 0 1 Asian .046 .001 0 1 Other .058 .002 0 1 White .464 .001 0 1 English language learner .175 .003 0 1 Single parent .166 .003 0 1 Married .512 .003 0 1 Unmarried with domestic partner .322 .004 0 1 Number of parental contacts at school 3.55 .031 0 40 Parent education of high school or less .275 .004 0 1 Receives Temporary Assistance for Needy Families .06 .002 0 1 Income is $30,000 a year or less .363 .002 0 1 Income is between $30,001 and $74,999 .341 .004 0 1 Income is between $75,000 and $99,999 .123 .003 0 1 Earns $100,000 or more per year .173 .003 0 1 Social geography  Suburb .333 .004 0 1  Town .109 .003 0 1  Rural .234 .004 0 1  Urban .324 .020 0 1 Parent involvement barriers  Not feeling welcome at school .059 .002 0 1  Lack of child care .253 .004 0 1  Lack of transportation .058 .002 0 1  Working >35 hours a week .352 .004 0 1 Latent class manifest variables  Parent has contacted the school .77 .004 0 1  Parent attends school open house .806 .004 0 1  Parent participates in the Parent–Teacher Association or other parent–teacher organization .366 .004 0 1  Parent is involved in school council .126 .004 0 1  Parent attends parent–teacher conferences .951 .002 0 1  Parent attends school events .777 .004 0 1  Parent volunteers at the school .558 .003 0 1 Note: SE = standard error. Table 1: Descriptive Statistics of Primary Study Variables (N = 15,600) Demographic Variable M SE Min Max African American .123 .003 0 1 Hispanic .251 .004 0 1 Asian .046 .001 0 1 Other .058 .002 0 1 White .464 .001 0 1 English language learner .175 .003 0 1 Single parent .166 .003 0 1 Married .512 .003 0 1 Unmarried with domestic partner .322 .004 0 1 Number of parental contacts at school 3.55 .031 0 40 Parent education of high school or less .275 .004 0 1 Receives Temporary Assistance for Needy Families .06 .002 0 1 Income is $30,000 a year or less .363 .002 0 1 Income is between $30,001 and $74,999 .341 .004 0 1 Income is between $75,000 and $99,999 .123 .003 0 1 Earns $100,000 or more per year .173 .003 0 1 Social geography  Suburb .333 .004 0 1  Town .109 .003 0 1  Rural .234 .004 0 1  Urban .324 .020 0 1 Parent involvement barriers  Not feeling welcome at school .059 .002 0 1  Lack of child care .253 .004 0 1  Lack of transportation .058 .002 0 1  Working >35 hours a week .352 .004 0 1 Latent class manifest variables  Parent has contacted the school .77 .004 0 1  Parent attends school open house .806 .004 0 1  Parent participates in the Parent–Teacher Association or other parent–teacher organization .366 .004 0 1  Parent is involved in school council .126 .004 0 1  Parent attends parent–teacher conferences .951 .002 0 1  Parent attends school events .777 .004 0 1  Parent volunteers at the school .558 .003 0 1 Demographic Variable M SE Min Max African American .123 .003 0 1 Hispanic .251 .004 0 1 Asian .046 .001 0 1 Other .058 .002 0 1 White .464 .001 0 1 English language learner .175 .003 0 1 Single parent .166 .003 0 1 Married .512 .003 0 1 Unmarried with domestic partner .322 .004 0 1 Number of parental contacts at school 3.55 .031 0 40 Parent education of high school or less .275 .004 0 1 Receives Temporary Assistance for Needy Families .06 .002 0 1 Income is $30,000 a year or less .363 .002 0 1 Income is between $30,001 and $74,999 .341 .004 0 1 Income is between $75,000 and $99,999 .123 .003 0 1 Earns $100,000 or more per year .173 .003 0 1 Social geography  Suburb .333 .004 0 1  Town .109 .003 0 1  Rural .234 .004 0 1  Urban .324 .020 0 1 Parent involvement barriers  Not feeling welcome at school .059 .002 0 1  Lack of child care .253 .004 0 1  Lack of transportation .058 .002 0 1  Working >35 hours a week .352 .004 0 1 Latent class manifest variables  Parent has contacted the school .77 .004 0 1  Parent attends school open house .806 .004 0 1  Parent participates in the Parent–Teacher Association or other parent–teacher organization .366 .004 0 1  Parent is involved in school council .126 .004 0 1  Parent attends parent–teacher conferences .951 .002 0 1  Parent attends school events .777 .004 0 1  Parent volunteers at the school .558 .003 0 1 Note: SE = standard error. Measures Our subpopulation parent profile models were analyzed using the seven available ECLS-K:2011 survey items that best reflected Epstein’s (2011) typology of parents’ school-based PI practices. These items measured whether each child’s primary parent or caregiver had ever (1) contacted the school, (2) attended an open house, (3) participated in the PTA or another parent–teacher organization (PTO), (4) participated in the school council, (5) attended a parent–teacher conference, (6) attended a school event, and (7) volunteered at the school. Significantly, each PI variable was coded in the ECLS-K:2011 data set as a dichotomous item. As a result of this coding, the frequency of PI in each of these activities cannot be discerned. Our analysis of the sociodemographic correlates of PI were conducted using 10 sociodemographic variables included in the ECLS-K:2011 data set. The first of these measures was parents’ ethnicity, which we coded into four binary variables (African American, Hispanic, Asian, and Other), with white representing the omitted reference category. The second variable was a dummy-coded measure of parents’ status as English language learners (ELLs), with native English speakers representing the omitted reference group. The third sociodemographic predictor variable was a dummy-coded measure of a caregiver’s single-parent status, with married or cohabitating parents representing the omitted reference category. The fourth variable was a dummy-coded measure of parents’ educational status, with 1 = parents who received a high school diploma or less and 0 = parents who had received at least some postsecondary education. Our fifth predictor variable was a count variable that measured the number of children parents had in the home. Our next set of predictor variables measured some of the key research-identified facilitators and barriers to PI (see, for example, Epstein, 2011; Mannan & Blackwell, 1992). Here, we analyzed the number of social contacts parents had with other parents at the school as a proxy measure for parents’ school social capital—a known facilitator for PI (Lareau, 2011). We also analyzed two dummy-coded measures of parents’ occupational status—specifically, whether parents received Temporary Assistance for Needy Families (TANF) and whether they worked for 35 hours or more each week. Last, we analyzed three dummy-coded items that measured key PI barriers—specifically, whether parents felt unwelcome at school, whether they lacked child care, and whether they faced consistent transportation barriers. The final predictor variable included in our analysis captured the social geography of parents’ home communities. This social geography construct was coded into three dichotomous items that captured whether parents lived in a suburb, town, or rural location. Parents who lived in an urban setting represented the omitted reference category for this variable. Analytic Approach Our PI profile models were estimated using a particular person-centered statistical technique, LCA. LCA is a statistical method that allows researchers to distill multiple combinations of PI activity into characteristically discrete patterns or subpopulation profiles (see, for example, Nylund et al., 2007). These profiles, also called latent classes, are identified statistically in LCA through a process known as latent class enumeration. Consistent with best practices in LCA modeling (see Masyn, 2013), we began our class enumeration procedure by estimating a one-class model using the statistical software program Mplus 7.3 (Muthén & Muthén, 2015). We then successively added latent classes to our models until such time that there were no empirical improvements in model fit. This determination of model fit was made by consulting the three primary statistical indices (that is, the Bayesian information criterion, the consistent Akaike information criterion, and the approximate weight of evidence criterion) typically used to analyze LCA models (M. A. Lawson & Masyn, 2015; Masyn, 2013). Lower values on these indices indicate an improved fit from a model with one fewer classes. Ultimately, we performed this latent class enumeration procedure on multiple subsets of ECLS-K:2011 data. We began by conducting an LCA of the full sample of 15,600 ECLS-K:2011 public school parents. The goal here was to capture a nationally representative view of parents’ school-based PI profiles. As a part of these analyses, we closely monitored the percentage of parents who were assigned to each PI profile group (that is, the class probabilities of each profile) and the model estimated chances that parents in each PI profile were involved in particular school-based PI activities (that is, each profile’s corresponding item probabilities). Once parents’ PI profiles were identified using the complete ECLS-K:2011 sample of public school parents, we performed this same LCA procedure on subsets of ECLS-K:2011 parent data that were organized according to family income. As part of these analyses, we used the family income coding available in the ECLS-K:2011 data set to divide our broader ECLS-K:2011 parent sample into four family income subgroups: (1) families with annual incomes of $30,000 or less; (2) families with annual incomes between $30,001 and $74,999; (3) families who earned between $75,000 and $99,999 annually; and (4) families who earned $100,000 or more annually. We then performed separate LCAs on each of these subgroups (subsamples) to evaluate the replicability of our original PI profile models. Second-Stage Analysis Our final research objective was to explore the sociodemographic features associated with each PI profile. To accomplish this objective, we estimated a series of logistic regression models. In these models, we used the “modal class assignment” feature in Mplus 7.3 to classify parents into particular PI profile groups. We then used these modal class assignments to create several binary outcome variables that could be analyzed by way of logistic regression modeling. We selected this logistic regression strategy in lieu of the more rigorous multinomial logistic regression approach for two reasons. First, logistic regression models yield results that are more user friendly to readers, and this accessibility is important in light of the journal’s diverse readership, including practitioners and policymakers. Second, although not detailed here due to page limits, the results yielded from these logistic regression models were similar (with respect to effect size and p value) to those generated using a more rigorous three-step approach to latent class regression modeling (see, for example, Vermunt, 2010). Last, because interest resided in understanding whether relations between our PI predictors and profiles might be moderated by family socioeconomic status (SES), we estimated separate logistic regression models for each category of family income. These models were estimated using 10 sets of imputed predictor variables, with standard errors clustered around the “school ID” variable. Improvements in model fit were evaluated using a simple likelihood ratio test. Results Our initial latent class enumeration procedure on the full sample of ECLS-K:2011 supported a four-class PI profile model (see Table 2). The item and class probabilities yielded for each PI profile are presented in Figure 1. As evident in the figure, the first PI profile was the low involvement class (20.6% of full ECLS-K:2011 parent sample). This profile includes parents whose involvement patterns were characterized by efforts to contact the school and attend parent–teacher conferences. Table 2: Results from a Latent Class Enumeration of School Involvement Profiles (Full ECLS-K:2011 Sample) (N = 15,600) Class LL BIC CAIC AWE LMR Entropy 1 –42,989 86,046 86,053 86,135 NA NA 2 –41,203 82,550 82,565 82,739.8 0.001 0.565 3 –40,888 81,997 82,020 82,288.9 0.001 0.589 4 –40,796 81,890 81,921 82,283.1 0.001 0.687 5 –40,769 81,914 81,953 82,407.9 0.149 0.655 Class LL BIC CAIC AWE LMR Entropy 1 –42,989 86,046 86,053 86,135 NA NA 2 –41,203 82,550 82,565 82,739.8 0.001 0.565 3 –40,888 81,997 82,020 82,288.9 0.001 0.589 4 –40,796 81,890 81,921 82,283.1 0.001 0.687 5 –40,769 81,914 81,953 82,407.9 0.149 0.655 Notes: Optimal class solution is indicated by the lowest yielded value for each fit index. ECLS-K:2011 = Early Childhood Longitudinal Study, Kindergarten Class of 2010–11; LL = log-likelihood value; BIC = Bayesian information criterion; CAIC = consistent Akaike information criterion; AWE = approximate weight of evidence criterion; LMR = Lo–Mendell–Rubin likelihood ratio test. Table 2: Results from a Latent Class Enumeration of School Involvement Profiles (Full ECLS-K:2011 Sample) (N = 15,600) Class LL BIC CAIC AWE LMR Entropy 1 –42,989 86,046 86,053 86,135 NA NA 2 –41,203 82,550 82,565 82,739.8 0.001 0.565 3 –40,888 81,997 82,020 82,288.9 0.001 0.589 4 –40,796 81,890 81,921 82,283.1 0.001 0.687 5 –40,769 81,914 81,953 82,407.9 0.149 0.655 Class LL BIC CAIC AWE LMR Entropy 1 –42,989 86,046 86,053 86,135 NA NA 2 –41,203 82,550 82,565 82,739.8 0.001 0.565 3 –40,888 81,997 82,020 82,288.9 0.001 0.589 4 –40,796 81,890 81,921 82,283.1 0.001 0.687 5 –40,769 81,914 81,953 82,407.9 0.149 0.655 Notes: Optimal class solution is indicated by the lowest yielded value for each fit index. ECLS-K:2011 = Early Childhood Longitudinal Study, Kindergarten Class of 2010–11; LL = log-likelihood value; BIC = Bayesian information criterion; CAIC = consistent Akaike information criterion; AWE = approximate weight of evidence criterion; LMR = Lo–Mendell–Rubin likelihood ratio test. Figure 1: View largeDownload slide Parents’ School Involvement Profiles (Full Sample, N = 15,600) Figure 1: View largeDownload slide Parents’ School Involvement Profiles (Full Sample, N = 15,600) The second PI profile yielded from LCA was the school investment class (8.1% of the full ECLS-K:2011 parent sample). This profile was characterized by parent participation in parent–teacher conferences and the PTO or PTA. The third PI profile was the school involvement class (42.6% of the full ECLS-K:2011 parent sample). Parents who fit this profile were generally involved in all school-based PI activities with the exception of school governance and PTO. The fourth and final LCA-derived PI profile was the school engagement class (28.5% of full parent sample). Members of this class were involved in nearly all of the PI activities analyzed in the statistical model. Significantly, school-engaged parents also had the highest relative chances of all parent groups of participating in their school’s advisory council. Distribution of PI Profiles across Categories of Family Income The second research objective was to examine the extent to which our original profile models might be replicated at discrete levels of family income. To evaluate this possibility, we performed separate LCAs on each of the four data subsets described earlier in the article. Each of these LCA models (not detailed here due to space limits) supported a four-class PI profile solution. Moreover, each of these LCA models yielded class-specific item probabilities that were nearly identical to those presented in Figure 1. Although the item probabilities of each profile were consistent across family income groups, the percentage of parents assigned to each PI profile group (that is, each profile’s class probabilities) varied considerably by family income. For example, as depicted in Table 3, over 50% of parents who earned $30,000 or less annually belonged to either the low-involvement (32.8%) or the school-invested (19.6%) PI profiles. However, fewer than 15% of parents with annual family incomes of between $75,000 and $99,999 belonged to these profile groups. Instead, the vast majority (80%) of families in that income category belonged to either the school-involved or the school-engagement PI profile groups. Consistent with extant research (see, for example, Lareau, 2011), these findings suggest that as family income increases, so do parent tendencies to engage in a broader configuration of school-based PI activities. Table 3: Distribution of School Involvement Profiles, by Family Income Category (N = 15,600) Profile (% of ECLS-K:2011 Population) Family Income Category (%) ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 Low involved (20.6) 32.8 24.1 9.2 9.9 School invested (8.1) 19.6 7.0 6.4 1.7 School involved (42.6) 30.6 42.0 53.0 57.0 School engaged (28.5) 17.0 27.0 31.3 31.0 Profile (% of ECLS-K:2011 Population) Family Income Category (%) ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 Low involved (20.6) 32.8 24.1 9.2 9.9 School invested (8.1) 19.6 7.0 6.4 1.7 School involved (42.6) 30.6 42.0 53.0 57.0 School engaged (28.5) 17.0 27.0 31.3 31.0 Note: ECLS-K:2011 = Early Childhood Longitudinal Study, Kindergarten Class of 2010–11. Table 3: Distribution of School Involvement Profiles, by Family Income Category (N = 15,600) Profile (% of ECLS-K:2011 Population) Family Income Category (%) ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 Low involved (20.6) 32.8 24.1 9.2 9.9 School invested (8.1) 19.6 7.0 6.4 1.7 School involved (42.6) 30.6 42.0 53.0 57.0 School engaged (28.5) 17.0 27.0 31.3 31.0 Profile (% of ECLS-K:2011 Population) Family Income Category (%) ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 Low involved (20.6) 32.8 24.1 9.2 9.9 School invested (8.1) 19.6 7.0 6.4 1.7 School involved (42.6) 30.6 42.0 53.0 57.0 School engaged (28.5) 17.0 27.0 31.3 31.0 Note: ECLS-K:2011 = Early Childhood Longitudinal Study, Kindergarten Class of 2010–11. Predictors of PI Profile Membership Results from our logistic regression analyses of PI profile membership on parents’ sociodemographic characteristics are presented in Tables 4 through 7. The logistic regression coefficients presented in these tables are odds ratios (ORs). ORs that are greater than 1 indicate a positive association between the predictor and the PI profile of interest, whereas ORs less than 1 indicate a negative or inverse association, all else being equal. Each model was initially estimated using the full complement of predictor variables detailed earlier in the article. However, our final models did not include the variables for TANF or the number of children in each parent’s household. These variables were omitted from our final models because they did not predict membership in any PI profile, and because their inclusion did not improve model fit. Table 4: Predictors of Membership in the Low-Involvement Profile, by Family Income Category (N = 15,600) Variable ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 OR SE OR SE OR SE OR SE Race or ethnicity  African American 1.15 (0.14) 1.14 (0.17) 1.47 (0.57) 1.65 (0.58)  Hispanic 0.99 (0.12) 1.22 (0.15) 0.03 (0.44) 1.42 (0.38)  Asian 1.72** (0.33) 1.44 (0.27) 2.99*** (0.97) 1.96* (0.56)  Other 1.12 (0.21) 1.01 (0.19) 0.78 (0.38) 1.59 (0.52) ELL 1.06 (0.12) 1.64*** (0.23) 2.11* (0.71) 2.87** (0.81) Single parent 1.01 (0.10) 1.13 (0.14) 0.88 (0.44) 1.48 (0.53) High school education or less 1.75*** (0.13) 1.63*** (0.15) 1.64* (0.49) 3.29*** (0.81) Number of social contacts 0.81*** (0.02) 0.79*** (0.02) 0.86** (0.05) 0.83*** (0.03) Involvement barriers  Not welcome at school 1.12 (0.17) 1.67** (0.31) 0.93 (0.55) 1.15 (0.73)  Lack of child care 1.34*** (0.11) 1.28* (0.12) 1.3 (0.30) 1.85 (0.28)  Lack of transportation 1.57*** (0.19) 1.81** (0.37) 2.12 (1.35) 1.62 (0.33)  Works >35 hours weekly 0.78** (0.07) 1.27** (0.18) 1 (0.07) 0.44 (0.21) Social geography  Suburban 0.89 (0.07) 1.02 (0.11) 1.75* (0.47) 1.45 (0.31)  Town 1.15 (0.19) 0.97 (0.16) 1.04 (0.51) 1.41 (0.57)  Rural 0.88 (0.10) 0.91 (0.12) 1.13 (0.41) 1.17 (0.29) Variable ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 OR SE OR SE OR SE OR SE Race or ethnicity  African American 1.15 (0.14) 1.14 (0.17) 1.47 (0.57) 1.65 (0.58)  Hispanic 0.99 (0.12) 1.22 (0.15) 0.03 (0.44) 1.42 (0.38)  Asian 1.72** (0.33) 1.44 (0.27) 2.99*** (0.97) 1.96* (0.56)  Other 1.12 (0.21) 1.01 (0.19) 0.78 (0.38) 1.59 (0.52) ELL 1.06 (0.12) 1.64*** (0.23) 2.11* (0.71) 2.87** (0.81) Single parent 1.01 (0.10) 1.13 (0.14) 0.88 (0.44) 1.48 (0.53) High school education or less 1.75*** (0.13) 1.63*** (0.15) 1.64* (0.49) 3.29*** (0.81) Number of social contacts 0.81*** (0.02) 0.79*** (0.02) 0.86** (0.05) 0.83*** (0.03) Involvement barriers  Not welcome at school 1.12 (0.17) 1.67** (0.31) 0.93 (0.55) 1.15 (0.73)  Lack of child care 1.34*** (0.11) 1.28* (0.12) 1.3 (0.30) 1.85 (0.28)  Lack of transportation 1.57*** (0.19) 1.81** (0.37) 2.12 (1.35) 1.62 (0.33)  Works >35 hours weekly 0.78** (0.07) 1.27** (0.18) 1 (0.07) 0.44 (0.21) Social geography  Suburban 0.89 (0.07) 1.02 (0.11) 1.75* (0.47) 1.45 (0.31)  Town 1.15 (0.19) 0.97 (0.16) 1.04 (0.51) 1.41 (0.57)  Rural 0.88 (0.10) 0.91 (0.12) 1.13 (0.41) 1.17 (0.29) Notes: OR = odds ratio; SE = standard error; ELL = English language learner. *p < .05. **p < .01. ***p < .001. Table 4: Predictors of Membership in the Low-Involvement Profile, by Family Income Category (N = 15,600) Variable ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 OR SE OR SE OR SE OR SE Race or ethnicity  African American 1.15 (0.14) 1.14 (0.17) 1.47 (0.57) 1.65 (0.58)  Hispanic 0.99 (0.12) 1.22 (0.15) 0.03 (0.44) 1.42 (0.38)  Asian 1.72** (0.33) 1.44 (0.27) 2.99*** (0.97) 1.96* (0.56)  Other 1.12 (0.21) 1.01 (0.19) 0.78 (0.38) 1.59 (0.52) ELL 1.06 (0.12) 1.64*** (0.23) 2.11* (0.71) 2.87** (0.81) Single parent 1.01 (0.10) 1.13 (0.14) 0.88 (0.44) 1.48 (0.53) High school education or less 1.75*** (0.13) 1.63*** (0.15) 1.64* (0.49) 3.29*** (0.81) Number of social contacts 0.81*** (0.02) 0.79*** (0.02) 0.86** (0.05) 0.83*** (0.03) Involvement barriers  Not welcome at school 1.12 (0.17) 1.67** (0.31) 0.93 (0.55) 1.15 (0.73)  Lack of child care 1.34*** (0.11) 1.28* (0.12) 1.3 (0.30) 1.85 (0.28)  Lack of transportation 1.57*** (0.19) 1.81** (0.37) 2.12 (1.35) 1.62 (0.33)  Works >35 hours weekly 0.78** (0.07) 1.27** (0.18) 1 (0.07) 0.44 (0.21) Social geography  Suburban 0.89 (0.07) 1.02 (0.11) 1.75* (0.47) 1.45 (0.31)  Town 1.15 (0.19) 0.97 (0.16) 1.04 (0.51) 1.41 (0.57)  Rural 0.88 (0.10) 0.91 (0.12) 1.13 (0.41) 1.17 (0.29) Variable ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 OR SE OR SE OR SE OR SE Race or ethnicity  African American 1.15 (0.14) 1.14 (0.17) 1.47 (0.57) 1.65 (0.58)  Hispanic 0.99 (0.12) 1.22 (0.15) 0.03 (0.44) 1.42 (0.38)  Asian 1.72** (0.33) 1.44 (0.27) 2.99*** (0.97) 1.96* (0.56)  Other 1.12 (0.21) 1.01 (0.19) 0.78 (0.38) 1.59 (0.52) ELL 1.06 (0.12) 1.64*** (0.23) 2.11* (0.71) 2.87** (0.81) Single parent 1.01 (0.10) 1.13 (0.14) 0.88 (0.44) 1.48 (0.53) High school education or less 1.75*** (0.13) 1.63*** (0.15) 1.64* (0.49) 3.29*** (0.81) Number of social contacts 0.81*** (0.02) 0.79*** (0.02) 0.86** (0.05) 0.83*** (0.03) Involvement barriers  Not welcome at school 1.12 (0.17) 1.67** (0.31) 0.93 (0.55) 1.15 (0.73)  Lack of child care 1.34*** (0.11) 1.28* (0.12) 1.3 (0.30) 1.85 (0.28)  Lack of transportation 1.57*** (0.19) 1.81** (0.37) 2.12 (1.35) 1.62 (0.33)  Works >35 hours weekly 0.78** (0.07) 1.27** (0.18) 1 (0.07) 0.44 (0.21) Social geography  Suburban 0.89 (0.07) 1.02 (0.11) 1.75* (0.47) 1.45 (0.31)  Town 1.15 (0.19) 0.97 (0.16) 1.04 (0.51) 1.41 (0.57)  Rural 0.88 (0.10) 0.91 (0.12) 1.13 (0.41) 1.17 (0.29) Notes: OR = odds ratio; SE = standard error; ELL = English language learner. *p < .05. **p < .01. ***p < .001. Table 5: Predictors of Membership in the School Investment Profile, by Income Category (N = 15,600) ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 Variable OR SE OR SE OR SE OR SE Race or ethnicity  African American 2.53*** (0.38) 1.72* (0.43) 8.47*** (3.37) 3.32* (1.91)  Hispanic 1.97*** (0.29) 1.62* (0.32) 2.31** (0.81) 0.93 (1.14)  Asian 2.09** (0.21) 3.16* (0.82) 2.18 (1.01) 2.34 (1.03)  Other 0.78 (0.21) 1.53 (0.51) 1.28 (0.71) 0.41 (1.06) ELL 1.93** (0.22) 2.57*** (0.51) 1.01 (0.42) 2.62* (1.21) Single parent 1.04 (0.11) 1.13 (0.26) 0.73 (0.42) 0.81 (0.64) High school education or less 1.05 (0.09) 1.14 (0.19) 1.53 (0.53) 3.31* (1.81) Number of social contacts 0.98 (0.02) 0.89*** (0.02) 0.97 (0.03) 0.85* (0.06) Involvement barriers  Not welcome at school 1.46* (0.22) 0.85 (0.35) 1.67 (1.26) 1.1 (0.22)  Lack of child care 0.89 (0.08) 1.2 (0.19) 1.44 (0.41) 1.31 (0.51)  Lack of transportation 0.89 (0.13) 0.64 (1.2) 0.77 (0.61) 2.56 (2.03)  Works >35 hours weekly 1.01 (0.10) 1.31 (0.18) 1.01 (0.07) 1.29 (0.45) Social geography  Suburban 0.87 (0.08) 1.03 (0.17) 0.68 (0.19) 1.11 (0.42)  Town 0.89 (0.19) 1.13 (0.36) 0.74 (0.39) 1.55 (1.18)  Rural 0.87 (0.12) 0.92 (0.19) 0.81 (0.25) 0.42 (0.31) ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 Variable OR SE OR SE OR SE OR SE Race or ethnicity  African American 2.53*** (0.38) 1.72* (0.43) 8.47*** (3.37) 3.32* (1.91)  Hispanic 1.97*** (0.29) 1.62* (0.32) 2.31** (0.81) 0.93 (1.14)  Asian 2.09** (0.21) 3.16* (0.82) 2.18 (1.01) 2.34 (1.03)  Other 0.78 (0.21) 1.53 (0.51) 1.28 (0.71) 0.41 (1.06) ELL 1.93** (0.22) 2.57*** (0.51) 1.01 (0.42) 2.62* (1.21) Single parent 1.04 (0.11) 1.13 (0.26) 0.73 (0.42) 0.81 (0.64) High school education or less 1.05 (0.09) 1.14 (0.19) 1.53 (0.53) 3.31* (1.81) Number of social contacts 0.98 (0.02) 0.89*** (0.02) 0.97 (0.03) 0.85* (0.06) Involvement barriers  Not welcome at school 1.46* (0.22) 0.85 (0.35) 1.67 (1.26) 1.1 (0.22)  Lack of child care 0.89 (0.08) 1.2 (0.19) 1.44 (0.41) 1.31 (0.51)  Lack of transportation 0.89 (0.13) 0.64 (1.2) 0.77 (0.61) 2.56 (2.03)  Works >35 hours weekly 1.01 (0.10) 1.31 (0.18) 1.01 (0.07) 1.29 (0.45) Social geography  Suburban 0.87 (0.08) 1.03 (0.17) 0.68 (0.19) 1.11 (0.42)  Town 0.89 (0.19) 1.13 (0.36) 0.74 (0.39) 1.55 (1.18)  Rural 0.87 (0.12) 0.92 (0.19) 0.81 (0.25) 0.42 (0.31) Notes: OR = odds ratio; SE = standard error; ELL = English language learner. *p < .05. **p < .01. ***p < .001. Table 5: Predictors of Membership in the School Investment Profile, by Income Category (N = 15,600) ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 Variable OR SE OR SE OR SE OR SE Race or ethnicity  African American 2.53*** (0.38) 1.72* (0.43) 8.47*** (3.37) 3.32* (1.91)  Hispanic 1.97*** (0.29) 1.62* (0.32) 2.31** (0.81) 0.93 (1.14)  Asian 2.09** (0.21) 3.16* (0.82) 2.18 (1.01) 2.34 (1.03)  Other 0.78 (0.21) 1.53 (0.51) 1.28 (0.71) 0.41 (1.06) ELL 1.93** (0.22) 2.57*** (0.51) 1.01 (0.42) 2.62* (1.21) Single parent 1.04 (0.11) 1.13 (0.26) 0.73 (0.42) 0.81 (0.64) High school education or less 1.05 (0.09) 1.14 (0.19) 1.53 (0.53) 3.31* (1.81) Number of social contacts 0.98 (0.02) 0.89*** (0.02) 0.97 (0.03) 0.85* (0.06) Involvement barriers  Not welcome at school 1.46* (0.22) 0.85 (0.35) 1.67 (1.26) 1.1 (0.22)  Lack of child care 0.89 (0.08) 1.2 (0.19) 1.44 (0.41) 1.31 (0.51)  Lack of transportation 0.89 (0.13) 0.64 (1.2) 0.77 (0.61) 2.56 (2.03)  Works >35 hours weekly 1.01 (0.10) 1.31 (0.18) 1.01 (0.07) 1.29 (0.45) Social geography  Suburban 0.87 (0.08) 1.03 (0.17) 0.68 (0.19) 1.11 (0.42)  Town 0.89 (0.19) 1.13 (0.36) 0.74 (0.39) 1.55 (1.18)  Rural 0.87 (0.12) 0.92 (0.19) 0.81 (0.25) 0.42 (0.31) ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 Variable OR SE OR SE OR SE OR SE Race or ethnicity  African American 2.53*** (0.38) 1.72* (0.43) 8.47*** (3.37) 3.32* (1.91)  Hispanic 1.97*** (0.29) 1.62* (0.32) 2.31** (0.81) 0.93 (1.14)  Asian 2.09** (0.21) 3.16* (0.82) 2.18 (1.01) 2.34 (1.03)  Other 0.78 (0.21) 1.53 (0.51) 1.28 (0.71) 0.41 (1.06) ELL 1.93** (0.22) 2.57*** (0.51) 1.01 (0.42) 2.62* (1.21) Single parent 1.04 (0.11) 1.13 (0.26) 0.73 (0.42) 0.81 (0.64) High school education or less 1.05 (0.09) 1.14 (0.19) 1.53 (0.53) 3.31* (1.81) Number of social contacts 0.98 (0.02) 0.89*** (0.02) 0.97 (0.03) 0.85* (0.06) Involvement barriers  Not welcome at school 1.46* (0.22) 0.85 (0.35) 1.67 (1.26) 1.1 (0.22)  Lack of child care 0.89 (0.08) 1.2 (0.19) 1.44 (0.41) 1.31 (0.51)  Lack of transportation 0.89 (0.13) 0.64 (1.2) 0.77 (0.61) 2.56 (2.03)  Works >35 hours weekly 1.01 (0.10) 1.31 (0.18) 1.01 (0.07) 1.29 (0.45) Social geography  Suburban 0.87 (0.08) 1.03 (0.17) 0.68 (0.19) 1.11 (0.42)  Town 0.89 (0.19) 1.13 (0.36) 0.74 (0.39) 1.55 (1.18)  Rural 0.87 (0.12) 0.92 (0.19) 0.81 (0.25) 0.42 (0.31) Notes: OR = odds ratio; SE = standard error; ELL = English language learner. *p < .05. **p < .01. ***p < .001. Table 6: Predictors of Membership in the School Involvement Profile, by Family Income Category (N = 15,600) Variable ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 OR SE OR SE OR SE OR SE Race or ethnicity  African American 0.37*** (0.04) 0.54*** (0.07) 0.41*** (0.1) 0.46** (0.001)  Hispanic 0.9*** (0.07) 0.75** (0.08) 0.72 (0.14) 1.03 (0.17)  Asian 0.47*** (0.09) 0.45*** (0.07) 0.88 (0.22) 0.73 (0.12)  Other 1.05 (0.18) 0.82 (0.13) 1.39 (0.36) 0.96 (0.18) ELL 0.53*** (0.05) 0.49*** (0.06) 0.78 (0.19) 0.75 (0.15) Single parent 0.83 (0.07) 0.94 (0.10) 1.23 (0.33) 1.25 (0.27) High school education or less 0.73*** (0.73) 0.86 (0.08) 1.01 (0.21) 0.74 (0.14) Number of social contacts 1.02* (0.01) 1 (0.02) 0.92** (0.02) 0.93** (0.01) Involvement barriers  Not welcome at school 0.87 (0.14) 0.87 (0.15) 1.25 (0.41) 0.75 (0.21)  Lack of child care 0.96 (0.08) 1.18 (0.10) 1.25 (0.19) 0.97 (0.11)  Lack of transportation 0.83 (0.10) 0.67 (0.67) 1.17 (0.62) 2.72 (1.45)  Works >35 hours weekly 1.25* (0.11) 0.86 (0.07) 1.43** (0.18) 1.24* (0.12) Social geography  Suburban 1.29** (0.11) 1.09 (0.17) 0.84 (0.13) 0.99 (0.12)  Town 1.01 (0.16) 1.42 (0.19) 0.95 (0.24) 0.99 (0.22)  Rural 1.31* (0.15) 1.2 (0.13) 1.27 (0.22) 1.01 (0.18) Variable ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 OR SE OR SE OR SE OR SE Race or ethnicity  African American 0.37*** (0.04) 0.54*** (0.07) 0.41*** (0.1) 0.46** (0.001)  Hispanic 0.9*** (0.07) 0.75** (0.08) 0.72 (0.14) 1.03 (0.17)  Asian 0.47*** (0.09) 0.45*** (0.07) 0.88 (0.22) 0.73 (0.12)  Other 1.05 (0.18) 0.82 (0.13) 1.39 (0.36) 0.96 (0.18) ELL 0.53*** (0.05) 0.49*** (0.06) 0.78 (0.19) 0.75 (0.15) Single parent 0.83 (0.07) 0.94 (0.10) 1.23 (0.33) 1.25 (0.27) High school education or less 0.73*** (0.73) 0.86 (0.08) 1.01 (0.21) 0.74 (0.14) Number of social contacts 1.02* (0.01) 1 (0.02) 0.92** (0.02) 0.93** (0.01) Involvement barriers  Not welcome at school 0.87 (0.14) 0.87 (0.15) 1.25 (0.41) 0.75 (0.21)  Lack of child care 0.96 (0.08) 1.18 (0.10) 1.25 (0.19) 0.97 (0.11)  Lack of transportation 0.83 (0.10) 0.67 (0.67) 1.17 (0.62) 2.72 (1.45)  Works >35 hours weekly 1.25* (0.11) 0.86 (0.07) 1.43** (0.18) 1.24* (0.12) Social geography  Suburban 1.29** (0.11) 1.09 (0.17) 0.84 (0.13) 0.99 (0.12)  Town 1.01 (0.16) 1.42 (0.19) 0.95 (0.24) 0.99 (0.22)  Rural 1.31* (0.15) 1.2 (0.13) 1.27 (0.22) 1.01 (0.18) Notes: OR = odds ratio; SE = standard error; ELL = English language learner. *p < .05. **p < .01. ***p < .001. Table 6: Predictors of Membership in the School Involvement Profile, by Family Income Category (N = 15,600) Variable ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 OR SE OR SE OR SE OR SE Race or ethnicity  African American 0.37*** (0.04) 0.54*** (0.07) 0.41*** (0.1) 0.46** (0.001)  Hispanic 0.9*** (0.07) 0.75** (0.08) 0.72 (0.14) 1.03 (0.17)  Asian 0.47*** (0.09) 0.45*** (0.07) 0.88 (0.22) 0.73 (0.12)  Other 1.05 (0.18) 0.82 (0.13) 1.39 (0.36) 0.96 (0.18) ELL 0.53*** (0.05) 0.49*** (0.06) 0.78 (0.19) 0.75 (0.15) Single parent 0.83 (0.07) 0.94 (0.10) 1.23 (0.33) 1.25 (0.27) High school education or less 0.73*** (0.73) 0.86 (0.08) 1.01 (0.21) 0.74 (0.14) Number of social contacts 1.02* (0.01) 1 (0.02) 0.92** (0.02) 0.93** (0.01) Involvement barriers  Not welcome at school 0.87 (0.14) 0.87 (0.15) 1.25 (0.41) 0.75 (0.21)  Lack of child care 0.96 (0.08) 1.18 (0.10) 1.25 (0.19) 0.97 (0.11)  Lack of transportation 0.83 (0.10) 0.67 (0.67) 1.17 (0.62) 2.72 (1.45)  Works >35 hours weekly 1.25* (0.11) 0.86 (0.07) 1.43** (0.18) 1.24* (0.12) Social geography  Suburban 1.29** (0.11) 1.09 (0.17) 0.84 (0.13) 0.99 (0.12)  Town 1.01 (0.16) 1.42 (0.19) 0.95 (0.24) 0.99 (0.22)  Rural 1.31* (0.15) 1.2 (0.13) 1.27 (0.22) 1.01 (0.18) Variable ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 OR SE OR SE OR SE OR SE Race or ethnicity  African American 0.37*** (0.04) 0.54*** (0.07) 0.41*** (0.1) 0.46** (0.001)  Hispanic 0.9*** (0.07) 0.75** (0.08) 0.72 (0.14) 1.03 (0.17)  Asian 0.47*** (0.09) 0.45*** (0.07) 0.88 (0.22) 0.73 (0.12)  Other 1.05 (0.18) 0.82 (0.13) 1.39 (0.36) 0.96 (0.18) ELL 0.53*** (0.05) 0.49*** (0.06) 0.78 (0.19) 0.75 (0.15) Single parent 0.83 (0.07) 0.94 (0.10) 1.23 (0.33) 1.25 (0.27) High school education or less 0.73*** (0.73) 0.86 (0.08) 1.01 (0.21) 0.74 (0.14) Number of social contacts 1.02* (0.01) 1 (0.02) 0.92** (0.02) 0.93** (0.01) Involvement barriers  Not welcome at school 0.87 (0.14) 0.87 (0.15) 1.25 (0.41) 0.75 (0.21)  Lack of child care 0.96 (0.08) 1.18 (0.10) 1.25 (0.19) 0.97 (0.11)  Lack of transportation 0.83 (0.10) 0.67 (0.67) 1.17 (0.62) 2.72 (1.45)  Works >35 hours weekly 1.25* (0.11) 0.86 (0.07) 1.43** (0.18) 1.24* (0.12) Social geography  Suburban 1.29** (0.11) 1.09 (0.17) 0.84 (0.13) 0.99 (0.12)  Town 1.01 (0.16) 1.42 (0.19) 0.95 (0.24) 0.99 (0.22)  Rural 1.31* (0.15) 1.2 (0.13) 1.27 (0.22) 1.01 (0.18) Notes: OR = odds ratio; SE = standard error; ELL = English language learner. *p < .05. **p < .01. ***p < .001. Table 7: Predictors of Membership in the School Engagement Profile, by Family Income Category (N = 15,600) Variable ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 OR SE OR SE OR SE OR SE Race or ethnicity  African American 1.54** (0.22) 1.56** (0.22) 0.85 (0.24) 1.65* (0.39)  Hispanic 1.1 (0.15) 1.07 (0.11) 1.14 (0.24) 0.79 (0.14)  Asian 0.68 (0.17) 1.12 (0.21) 0.48* (0.17) 1.11 (0.21)  Other 0.82 (0.22) 1.16 (0.20) 0.65 (0.20) 0.76 (0.16) ELL 0.98 (0.12) 0.82 (0.12) 0.71 (0.19) 0.44*** (0.11) Single parent 1.08 (0.11) 0.92 (0.11) 0.91 (0.26) 0.57* (0.15) High school education or less 0.61*** (0.06) 0.70*** (0.07) 0.68 (0.16) 0.58* (0.13) Number of social contacts 1.18** (0.02) 1.14* (0.02) 1.16* (0.03) 1.14*** (0.03) Involvement barriers  Not welcome at school 0.51** (0.11) 0.69 (0.15) 0.61 (0.41) 1.13 (0.35)  Lack of child care 0.72** (0.08) 0.58*** (0.06) 0.62** (0.19) 0.81 (0.11)  Lack of transportation 0.57** (0.11) 0.9 (0.21) 0.19* (0.63) 0.25 (0.21)  Works >35 hours weekly 1.03 (0.11) 0.88 (0.07) 0.66** (0.18) 0.72*** (0.07) Social geography  Suburban 0.92 (0.09) 0.87 (0.09) 1.13 (0.13) 0.89 (0.12)  Town 0.87 (0.19) 0.63** (0.11) 1.13 (0.23) 0.89 (0.22)  Rural 0.91 (0.13) 0.9 (0.10) 0.77 (0.22) 0.78 (0.13) Variable ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 OR SE OR SE OR SE OR SE Race or ethnicity  African American 1.54** (0.22) 1.56** (0.22) 0.85 (0.24) 1.65* (0.39)  Hispanic 1.1 (0.15) 1.07 (0.11) 1.14 (0.24) 0.79 (0.14)  Asian 0.68 (0.17) 1.12 (0.21) 0.48* (0.17) 1.11 (0.21)  Other 0.82 (0.22) 1.16 (0.20) 0.65 (0.20) 0.76 (0.16) ELL 0.98 (0.12) 0.82 (0.12) 0.71 (0.19) 0.44*** (0.11) Single parent 1.08 (0.11) 0.92 (0.11) 0.91 (0.26) 0.57* (0.15) High school education or less 0.61*** (0.06) 0.70*** (0.07) 0.68 (0.16) 0.58* (0.13) Number of social contacts 1.18** (0.02) 1.14* (0.02) 1.16* (0.03) 1.14*** (0.03) Involvement barriers  Not welcome at school 0.51** (0.11) 0.69 (0.15) 0.61 (0.41) 1.13 (0.35)  Lack of child care 0.72** (0.08) 0.58*** (0.06) 0.62** (0.19) 0.81 (0.11)  Lack of transportation 0.57** (0.11) 0.9 (0.21) 0.19* (0.63) 0.25 (0.21)  Works >35 hours weekly 1.03 (0.11) 0.88 (0.07) 0.66** (0.18) 0.72*** (0.07) Social geography  Suburban 0.92 (0.09) 0.87 (0.09) 1.13 (0.13) 0.89 (0.12)  Town 0.87 (0.19) 0.63** (0.11) 1.13 (0.23) 0.89 (0.22)  Rural 0.91 (0.13) 0.9 (0.10) 0.77 (0.22) 0.78 (0.13) Notes: OR = odds ratio; SE = standard error; ELL = English language learner. *p < .05. **p < .01. ***p < .001. Table 7: Predictors of Membership in the School Engagement Profile, by Family Income Category (N = 15,600) Variable ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 OR SE OR SE OR SE OR SE Race or ethnicity  African American 1.54** (0.22) 1.56** (0.22) 0.85 (0.24) 1.65* (0.39)  Hispanic 1.1 (0.15) 1.07 (0.11) 1.14 (0.24) 0.79 (0.14)  Asian 0.68 (0.17) 1.12 (0.21) 0.48* (0.17) 1.11 (0.21)  Other 0.82 (0.22) 1.16 (0.20) 0.65 (0.20) 0.76 (0.16) ELL 0.98 (0.12) 0.82 (0.12) 0.71 (0.19) 0.44*** (0.11) Single parent 1.08 (0.11) 0.92 (0.11) 0.91 (0.26) 0.57* (0.15) High school education or less 0.61*** (0.06) 0.70*** (0.07) 0.68 (0.16) 0.58* (0.13) Number of social contacts 1.18** (0.02) 1.14* (0.02) 1.16* (0.03) 1.14*** (0.03) Involvement barriers  Not welcome at school 0.51** (0.11) 0.69 (0.15) 0.61 (0.41) 1.13 (0.35)  Lack of child care 0.72** (0.08) 0.58*** (0.06) 0.62** (0.19) 0.81 (0.11)  Lack of transportation 0.57** (0.11) 0.9 (0.21) 0.19* (0.63) 0.25 (0.21)  Works >35 hours weekly 1.03 (0.11) 0.88 (0.07) 0.66** (0.18) 0.72*** (0.07) Social geography  Suburban 0.92 (0.09) 0.87 (0.09) 1.13 (0.13) 0.89 (0.12)  Town 0.87 (0.19) 0.63** (0.11) 1.13 (0.23) 0.89 (0.22)  Rural 0.91 (0.13) 0.9 (0.10) 0.77 (0.22) 0.78 (0.13) Variable ≤$30,000 $30,001–$74,999 $75,000–$99,999 ≥$100,000 OR SE OR SE OR SE OR SE Race or ethnicity  African American 1.54** (0.22) 1.56** (0.22) 0.85 (0.24) 1.65* (0.39)  Hispanic 1.1 (0.15) 1.07 (0.11) 1.14 (0.24) 0.79 (0.14)  Asian 0.68 (0.17) 1.12 (0.21) 0.48* (0.17) 1.11 (0.21)  Other 0.82 (0.22) 1.16 (0.20) 0.65 (0.20) 0.76 (0.16) ELL 0.98 (0.12) 0.82 (0.12) 0.71 (0.19) 0.44*** (0.11) Single parent 1.08 (0.11) 0.92 (0.11) 0.91 (0.26) 0.57* (0.15) High school education or less 0.61*** (0.06) 0.70*** (0.07) 0.68 (0.16) 0.58* (0.13) Number of social contacts 1.18** (0.02) 1.14* (0.02) 1.16* (0.03) 1.14*** (0.03) Involvement barriers  Not welcome at school 0.51** (0.11) 0.69 (0.15) 0.61 (0.41) 1.13 (0.35)  Lack of child care 0.72** (0.08) 0.58*** (0.06) 0.62** (0.19) 0.81 (0.11)  Lack of transportation 0.57** (0.11) 0.9 (0.21) 0.19* (0.63) 0.25 (0.21)  Works >35 hours weekly 1.03 (0.11) 0.88 (0.07) 0.66** (0.18) 0.72*** (0.07) Social geography  Suburban 0.92 (0.09) 0.87 (0.09) 1.13 (0.13) 0.89 (0.12)  Town 0.87 (0.19) 0.63** (0.11) 1.13 (0.23) 0.89 (0.22)  Rural 0.91 (0.13) 0.9 (0.10) 0.77 (0.22) 0.78 (0.13) Notes: OR = odds ratio; SE = standard error; ELL = English language learner. *p < .05. **p < .01. ***p < .001. Predictors of Low-Involvement Profile Membership Table 4 presents our logistic regression analysis of the sociodemographic features of the low-involvement PI profile group. These analyses indicated that, for most family income groups, the significant predictors of low involvement were having a high school education or less (OR range = 1.63–3.29), knowing fewer parents at the school (OR range = 0.79–0.86), and being Asian (relative to being white) (OR range = 1.42–2.99). Meanwhile, parents with family incomes less than $75,000 annually were more likely belong to the low-involvement profile group when they experienced transportation (OR range = 1.57–1.81) or child care barriers (OR range = 1.28–1.34). Last, these analyses revealed somewhat complex, nuanced relationship between low-involvement profile membership and full-time work. Specifically, parents with family incomes of $30,000 or less were significantly less likely to belong to low-involvement profile when they worked full time (OR = 0.78, p < .01). In contrast, parents residing in the $30,001 to $74,999 income category were significantly more likely to be low involved when they were engaged in full-time work (OR = 1.27, p < .01). Predictors of School Investment Table 5 presents our logistic regression analysis of the sociodemographic features of the school investment PI profile group. These analyses indicated that, across most family income groups, the significant predictors of school investment were being ELL (OR range = 1.93–2.62) and being African American (relative to being white) (OR range = 1.72–8.47). These models also indicated that Hispanic (OR range = 1.62–2.31) and Asian parents (OR range = 2.09–3.16) living in families earning less than $75,000 annually were significantly more likely than white parents to belong to the school investment profile. Last, parents in families earning $30,000 or less annually were significantly more likely to belong to the school investment profile when they felt unwelcome by the school (OR = 1.46, p < .05). Predictors of School Involvement Table 6 presents our logistic regression analysis of the sociodemographic predictors of the school involvement profile membership. As indicated in the table, these models generally supported a significant, positive relationship between full-time work and school involvement (OR range = 0.86–1.43). However, these same models yielded a significant negative relationship between being an ethnic minority parent and school involvement profile membership. Specifically, our models indicated that, across all levels of family income, African American parents were significantly less likely than white parents to belong to the school involvement group (OR range = 0.37–0.54). In the same vein, Hispanic (OR range = 0.75–0.90) and Asian (OR range = 0.45–0.47) parents earning less than $75,000 annually were significantly less likely than white parents to belong to the school involvement profile group, all else being equal. Last, when the analysis was restricted to parents living in the lowest family income stratum, our models indicated that parents who were ELL (OR = 0.53, p < .001) or who had a high school diploma or less (OR = 0.73, p < .001) were significantly less likely to be school involved. Meanwhile, these models indicated that low-income school-involved parents were more likely to live in the suburbs (OR = 1.29, p < .01) or rural communities (OR = 1.31. p < .05) than urban locations. Predictors of School Engagement Table 7 presents our logistic regression analysis of the sociodemographic predictors of the school engagement profile membership. These analyses indicated that, across most family income categories, school-engaged parents were more likely to be African American than white (OR range = 0.85–1.65), and they were also more likely to have a higher number of social contacts at the school (OR range = 1.14–1.18). Within the lowest income stratum, school-engaged parents were less likely to experience common PI barriers such as child care (OR = 0.72, p < .01) and transportation (OR = 0.57, p < .01) constraints, and they were also highly unlikely to have a high school diploma or less (OR = 0.61, p < .05). Last, parents who worked full time and lived in families earning $75,000 or more were significantly less likely to belong to the school engagement profile (OR range = 0.66–0.72). This latter finding signals the amount of time that may be needed for parents to engage in a full range of school-based PI activities. Summary and Implications This study used LCA to analyze subpopulation profiles of PI in elementary school. These analyses were conducted using a nationally representative sample of first-grade parents who had children attending public schools in the United States. Four characteristically distinct profiles of PI were yielded from the LCA approach: (1) A low involvement profile characterized by PI in a limited number of school-based activities; (2) a school investment profile characterized primarily by participation in the PTA or other PTOand parent–teacher conferences; (3) a school involvement profile characterized by PI in multiple activities, but not governance; and (4) a school engagement profile characterized by parent participation in a full range of school-based PI activities. Following our initial LCA on the complete ECLS-K:2011 sample of public school parents, the next step in the study was to evaluate the relative consistency of our PI profile findings across different categories of family income. Here, our models indicated that the characteristic features (that is, the item probabilities) of each PI profile remained remarkably consistent across family income groups. However, the proportion of parents assigned to each profile group varied according to parents’ SES. Specifically, parents living in families earning $30,000 or less annually were disproportionately assigned to those profiles characterized by PI in a limited number of school-based activities (that is, the low involvement and school investment profiles), whereas the vast majority of parents from families earning $75,000 or more belonged to either the school involvement or the school engagement profile groups. Several important implications for school social work practice accompany these diverse PI profile findings. The first of these implications relates to extant research regarding the PI practices of low-income families. As described earlier in the article, prior PI research has generally characterized low-income parents as less involved than their more affluent counterparts (Lareau, 2011). In practice-embedded research contexts, this finding has been linked to deficit-oriented views regarding low-income parents’ core values and mores, including educator doubts about whether low-income parents truly value their children’s education (M. A. Lawson, 2003). In contrast to these deficit-oriented findings, our LCA models indicated that nearly half the parents earning $30,000 or less annually belonged to either the school involved or the school engagement PI profile groups. Moreover, although these models assigned a disproportionate number of low-income parents to the low involvement and school investment PI profile groups, it is important to note that these two parent subpopulations were characterized, in part, by PI in at least one school-based activity. An immediate implication is that all low-income parents have important PI-related strengths that schools can and should build from to support PI and promote students’ school success. School social workers, with their unique training and experience in strengths-based and empowerment-based practice modalities, can help schools learn to better capitalize on parents’ existing strengths, knowledge, and interests (see, for example, Alameda-Lawson & Lawson, 2016). Recommendations for Improving PI in Low-Income Schools One specific way that school social workers and other school community professionals might go about improving PI in low-income schools is to enhance opportunities for parent participation in school governance activities, especially PTOs. PTOs may represent an important starting point for school social work interventions because the data indicate that low-income African American, Latino, and Asian parents, as well as low-income parents classified as ELL, are initially drawn to them. At the same time, it is important to note that the data indicated that an important subpopulation of PTO parents (that is, low-income, minority parents who belong to the school investment profile group) represented the most likely candidates to report that they felt unwelcome at the school. This latter finding, in particular, underscores the need for social work–assisted programs and supports that help ethnically and linguistically diverse families negotiate the complex (and often difficult) practices, boundaries, and cultures of schools (see, for example, M. A. Lawson & Alameda-Lawson, 2012). It also highlights opportunities for social workers to help educators develop programs and policies that can accommodate the strengths, needs, and challenges of their constituent families and communities (Alameda-Lawson, Lawson, & Lawson, 2010). PI as an Inherently Social Activity One of the more important findings yielded from this study concerns the relationship between PI and parents’ social networks. The importance of parents’ social networks was particularly evident in findings that linked PI profile membership to the number of social contacts that parents enjoyed at the school. Specifically, the data indicated that parents who knew more parents at the school were significantly more likely to belong to the school involvement and school engagement profiles, whereas membership in the low involvement profile was counterindicated by this proxy indicator for parents’ school social capital. The ready implication is that relational or “collective” PI practices and models may hold special promise for enhancing PI in low-income school communities (see, for example, Alameda-Lawson, Lawson, & Lawson, 2013; Warren et al., 2009). School social workers can facilitate the development of these collectivist PI approaches by forging school–family–community partnerships with faith-based groups, youth sports leagues, and other neighborhood-based efforts that can connect heretofore isolated parents to others who have children at the school (Alameda-Lawson & Lawson, 2016; Lareau, 2011; Warren et al., 2009). PI as a Specialized Intervention A final important finding from this study concerns the complex relationship that emerged between PI profile membership and full-time work. This finding was both curious and novel because prior research has identified full-time work as a barrier to parents’ school-based PI (see, for example, M. A. Lawson, 2003). However, in this study, full-time work appeared to predict PI for some parent subpopulations (that is, the school involvement profile), whereas for others, it appeared to have the opposite effect (that is, the school engagement group). In addition to these findings, this study illuminated the conditions in which family income might moderate the association between PI and full-time work. For example, our models indicated that parents living in the lowest income stratum were significantly less likely to belong to the low involvement profile group when they worked full time. In contrast, full-time work was a significant predictor of low involvement among parents residing in the $30,001 to $74,999 family income category. Unfortunately, due to the nature of the ECLS-K:2011 data set, we can only speculate as to why full-time work might act as a PI barrier for select parent subpopulations. For example, it is possible that low-involved parents in the $30,001 to $74,999 income group are more likely than parents in lower income subgroups to face stringent work hours. Relatedly, it is also possible that low-involved parents in this income category represent a parent subpopulation that is more likely to work more than one job. The subpopulation PI profile differences revealed in this study highlight the limitations of standardized school–family interventions and showcase needs for more tailored PI interventions and related supports. For instance, findings from this study indicate that large-scale, standardized PI efforts such as the National Network of Partnership Schools (see, for example, Epstein, 2011)—which encourages family involvement in the full spectrum of PI activities explored in this study—may be unlikely to realize their full potential when parents’ lives are consumed by daily hassles, such as child care, food, housing, and transportation constraints (Posey-Maddox & Haley-Lock, 2016). In contrast, more discrete efforts like Lynn McDonald’s Families and Schools Together (FAST) program, with its special emphasis on enhancing parent–school interactions and increasing parent and family social capital, may be especially useful in helping previously isolated parents forge needed social connections both at school and in the community (see, for example, Kratochwill, McDonald, Levin, Scalia, & Coover, 2009). However, because the service capacity of programs like FAST is typically limited to a handful of school families (see also Webster-Stratton & Reid, 2010), social workers can and should take leadership roles in helping these programs expand into more comprehensive, schoolwide, community-building efforts. Limitations and Future Research Directions Although this study offered an enhanced, person-centered view of parents’ PI practices and profiles, there are several limitations to this study’s design that serve to qualify our results. It is important to note that these limitations double as future research needs. To begin, one of the primary limitations of this study stems from the quality and nature of the ECLS-K:2011 survey questions, especially the coding of school-based PI activities into binary items and dichotomous variables. As a result of the ECLS-K:2011 coding strategy, our parent profile models may have masked important distinctions between parents who had little or occasional involvement in certain PI activities and those who engaged in those activities on a consistent basis. For this reason, future PI research would benefit from person-centered studies that attend to both the breadth and the depth of parental involvement in elementary school. The second limitation concerned our exclusive reliance on the cross-sectional PI experiences of parents of first-grade children. This narrow focus was driven by the particularities of the ECLS-K:2011 data set, but it was not without consequence. For example, because our analysis focused exclusively on the cross-sectional experiences of first-grade parents, we are not able to generalize our findings to parents of elementary school-age children in grades 2 through 6, nor are we able to understand how parents’ PI profiles might change over time. In addition, because of limitations in the ECLS-K:2011 data set, we were not able to discern how other important sociogeographic features and factors—such as the distance between parents’ homes and their children’s school—might influence parents’ school involvement patterns and practices. For this reason, longitudinal PI studies are needed, especially those that detail how parents’ sociogeographies might influence the quality and nature of their PI practices and profiles. The third limitation concerns the primary characteristics of this study’s parent sample. Specifically, because this study analyzed the PI practices of each child’s “primary parent,” the PI practices and profiles of other adults and caregivers in the family (for example, spouses, aunts and uncles, friends, and grandparents) were not included in our LCA models. Given the importance of family assets for children’s educational success and well-being (Lareau, 2011), future research should target a more comprehensive and family-centered view of PI practices and outcomes. The final limitation of this study concerns the absence of home-based PI activities in our statistical models. These activities were not included in the present study because of page limits. Ultimately, these recommendations, and the design limitations from which they are derived, help to highlight the exploratory nature of the present research. All of this study’s results should therefore be considered preliminary until replicated in both local and national parent samples, using a more nuanced array of PI indicators and a more comprehensive set of controls. Tania Alameda-Lawson, PhD, is associate professor, School of Social Work, University of Alabama, Box 870314, Tuscaloosa, AL 35487-0314; e-mail: [email protected]. Michael A. Lawson, PhD, is assistant professor, College of Education, University of Alabama, Tuscaloosa References Alameda-Lawson , T. , & Lawson , M. A. ( 2016 ). Ecologies of collective parent engagement in urban education . Urban Education . Advance online publication. doi:10.1177/0042085916636654 Alameda-Lawson , T. , Lawson , M. A. , & Lawson , H. A. ( 2010 ). Social workers’ roles in facilitating the collective involvement of low-income, culturally diverse parents in an elementary school . Children & Schools, 32 , 172 – 182 . Google Scholar CrossRef Search ADS Alameda-Lawson , T. , Lawson , M. A. , & Lawson , H. A. ( 2013 ). An innovative collective parent engagement model for families and neighborhoods in arrival cities . Journal of Family Strengths, 13 ( 1 ), 1 – 24 . Retrieved from http://digitalcommons.library.tmc.edu/jfs/vol13/iss1/1 Epstein , J. L. ( 2011 ). School, family, and community partnerships: Preparing educators and improving schools . New York : Westview Press . Fan , X. , & Chen , M. ( 2001 ). Parental involvement and students’ academic achievement: A meta-analysis . Educational Psychology Review, 13 , 1 – 22 . Google Scholar CrossRef Search ADS Holloway , S. D. , Campbell , E. J. , Nagase , E. , Kim , S. , Suzuki , S. , Wang , Q. , et al. ( 2016 ). Parenting self-efficacy and parental involvement: Mediators or moderators between socioeconomic status and children’s academic competence in Japan and Korea? Research in Human Development, 13 , 258 – 272 . doi:10.1080/15427609.2016.1194710 Google Scholar CrossRef Search ADS Jeynes , W. ( 2011 ). Parental involvement and academic success . New York : Taylor & Francis/Routledge . Kratochwill , T. , McDonald , L. , Levin , J. , Scalia , P. , & Coover , G. ( 2009 ). Families and schools together: An experimental study of multi-family support groups for children at risk . Journal of School Psychology, 47 , 245 – 265 . Google Scholar CrossRef Search ADS Lareau , A. ( 2011 ). Unequal childhoods: Class, race, and family life, with an update a decade later ( 2nd ed. ). Berkeley : University of California Press . Lawson , H. , Alameda-Lawson , T. , Lawson , M. , Briar-Lawson , K. , & Wilcox , K. ( 2014 ). Three parent and family interventions for rural schools and communities . Journal of Education and Human Development, 3 ( 3 ), 59 – 78 . Retrieved from http://jehdnet.com/journals/jehd/Vol_3_No_3_September_2014/5.pdf Google Scholar CrossRef Search ADS Lawson , M. A. ( 2003 ). School-family relations in context: Parent and teacher perceptions of parent involvement . Urban Education, 38 ( 1 ), 77 – 133 . Google Scholar CrossRef Search ADS Lawson , M. A. , & Alameda-Lawson , T. ( 2012 ). A case study of school-linked, collective parent engagement . American Educational Research Journal, 49 , 651 – 684 . Google Scholar CrossRef Search ADS Lawson , M. A. , & Masyn , K. ( 2015 ). Analyzing profiles and predictors of students’ social-ecological engagement . AERA Open, 1 ( 4 ), 1 – 37 . doi:10.1177/2332858415615856 Google Scholar CrossRef Search ADS Lee , J. S. , & Bowen , N. K. ( 2006 ). Parent involvement, cultural capital, and the achievement gap among elementary school children . American Educational Research Journal, 43 , 193 – 215 . Google Scholar CrossRef Search ADS Mannan , G. , & Blackwell , J. ( 1992 ). Parent involvement: Barriers and opportunities . Urban Review, 24 , 219 – 228 . Google Scholar CrossRef Search ADS Mapp , K. L. , & Kuttner , P. J. ( 2014 ). Partners in education: A dual capacity-building framework for family-school partnerships . Austin, TX, & Washington, DC : SEDL & U.S. Department of Education . Retrieved from http://www.sedl.org/pubs/framework/ Masyn , K. ( 2013 ). Latent class analysis and finite mixture modeling. In T. Little (Ed.), Oxford handbook of quantitative methods (Vol. 2, pp. 551 – 611 ). Oxford, UK : Oxford University Press . Moss , P. A. ( 2012 ). Exploring the macro-micro dynamic in data use practice . American Journal of Education, 118 , 223 – 232 . Google Scholar CrossRef Search ADS Mulligan , G. M. , Hastedt , S. , & McCarroll , J. C. ( 2012 ). First-time kindergartners in 2010-11: First findings from the kindergarten rounds of the Early Childhood Longitudinal Study, Kindergarten Class of 2010-11 (ECLS-K:2011) . Washington, DC : U.S. Department of Education, National Center for Education Statistics . Retrieved from http://nces.ed.gov/pubsearch Muthén , L. , & Muthén , B. ( 2015 ). Mplus user’s guide ( 7th ed. ). Los Angeles : Author . Nylund , K. , Bellmore , A. , Nishina , A. , & Graham , S. ( 2007 ). Subtypes, severity, and structural stability of peer victimization: What does latent class analysis say? Child Development, 78 , 1706 – 1722 . Google Scholar CrossRef Search ADS Park , S. , & Holloway , D. ( 2017 ). The effects of school-based parental involvement on academic achievement at the child and elementary school level: A longitudinal study . Journal of Educational Research, 110 ( 1 ), 1 – 16 . doi:10.1080/00220671.2015.1016600 Google Scholar CrossRef Search ADS Posey-Maddox , L. , & Haley-Lock , A. ( 2016 ). One size does not fit all: Understanding parent engagement in the contexts of work, family, and public schooling . Urban Education . Advance online publication. doi:10.1177/0042085916660348 Robinson , K. , & Harris , A. L. ( 2014 ). The broken compass: Parental involvement with children’s education . Cambridge, MA : Harvard University Press . Google Scholar CrossRef Search ADS Rogers , M. A. , Theule , J. , Ryan , B. A. , Adams , G. R. , & Keating , L. ( 2009 ). Parental involvement and children’s school achievement: Evidence for mediating processes . Canadian Journal of School Psychology, 24 , 34 – 57 . Google Scholar CrossRef Search ADS Vermunt , J. K. ( 2010 ). Latent class modeling with covariates: Two improved “three step” approaches . Political Analysis, 18 , 450 – 469 . Google Scholar CrossRef Search ADS Warren , M. , Hong , S. , Rubin , C. , & Uy , P. ( 2009 ). Beyond the bake sale: A community-based relational approach to parent engagement in schools . Teachers College Record, 111 , 2209 – 2254 . Webster-Stratton , C. , & Reid , M. J. ( 2010 ). A school-family-community partnership: Addressing multiple risk factors to improve school readiness. In S. Christenson & A. Reschly (Eds.), Handbook of school-family partnerships (pp. 204 – 227 ). New York : Routledge . © 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/about_us/legal/notices)

Journal

Social Work Research – Oxford University Press

Published: Apr 9, 2018

There are no references for this article.