Who Goes to College? Social Capital and Other Predictors of College Enrollment for Foster-Care Youth
Abstract
Objective: This study investigated social capital, risk factors, and protective factors associated with the likelihood that youth in foster care will enroll in college. We tested three hypotheses: (a) having a greater number of institutional agents promotes college enrollment, (b) encouragement from school personnel to pursue postsecondary education promotes college enrollment, and (c) the number of institutional agents and amount of school encouragement will interact to promote college enrollment. Method: We used a representative sample of adolescents (N = 712), ages 16.75–17.75, in California foster care in 2012 (response rate = 95%) and obtained college enrollment data from the National Student Clearinghouse. We used logistic regression to predict college enrollment. Results: The number of institutional agents who participants nominated, as well as encouragement from school personnel, significantly increased the likelihood of college enrollment. Findings were mediated by the amount of help youths received preparing for college and their participation in extended foster care. Reading ability, educational aspirations, and high school grades were positively associated with college enrollment. Grade repetition, placement in special education, and early parenthood decreased the likelihood of college enrollment. Conclusions: The likelihood of enrolling in college increases when foster youth have supportive relationships with adults who can leverage their positional power and mobilize college-relevant knowledge and resources.
By age 25, young adults in the general U.S. population are more than five times as likely to have earned a college degree as those who were in foster care as adolescents (Courtney et al., 2011). Although some of this gap can be explained by setbacks occurring after students enroll in college (Day, Dworsky, & Feng, 2013; Day, Dworsky, Fogarty, & Damashek, 2011), the disparity starts at a lower rung in the educational ladder. Foster youth have high hopes of enrolling in college and earning a degree (Courtney, Terao, & Bost, 2004; McMillen & Tucker 1999), yet relatively few make it to college (Courtney et al., 2005; Frerer, Sosenko, & Henke, 2013). College enrollment gaps may be explained in part by deficits in social capital. Biological family members of foster youth typically have not attended college, making adults outside of families (e.g., caseworkers, teachers, mentors, and other professionals) increasingly important for mentorship and guidance with the college choice process. However, many foster youth experience multiple changes in living arrangements and schools, which can both sever existing relationships with adults and make youth reluctant to initiate new relationships (Samuels, 2008). Consequently, social capital that was already thin may be further attenuated. The purpose of this study is to examine the role that social capital—and other risk and protective factors—plays in college enrollment for youth in foster care.
College Enrollment and Youth in Foster Care
In 2014, there were over 66,000 youths ages 16–21 in U.S. foster care, comprising about 16% of the total foster-care population (AFCARS, 2016). A growing body of research paints a consistently bleak picture of postsecondary educational attainment for foster youth (Gillum, Lindsay, Murray, & Wells, 2016). As adolescents, about 80% of foster youth aspire to earn a college degree (Courtney et al., 2004; McMillen & Tucker, 1999), yet few ever see the inside of a college classroom. Existing research suggests that youth in foster care attend college at much lower rates than same-aged peers (Brandford & English, 2004; Frerer et al., 2013; Wolanin, 2005). For example, 24% of foster youths in three Midwestern states were enrolled in college at age 19, compared to 56% of youths from a national sample (Courtney et al., 2005). Moreover, among those enrolled in college, most foster youth attended 2-year colleges (70.4%), but most youth from the general population attended 4-year colleges (64.3%; Courtney et al., 2005).
Many explanations have been proposed for the gap in college enrollment between foster youth and their nonfoster peers (e.g., Geenen et al., 2015; Pecora, 2012). These explanations typically point to the enduring effects of maltreatment, disruptions caused by placement and educational instability, difficulties accompanying psychological problems, and lags in educational performance and advancement. A wide body of literature has shown the negative effect that childhood maltreatment can have on psychosocial functioning, such as emotional regulation and cognitive functioning, including the ability to sustain attention (Cicchetti, 2016; Klein, Damiani-Taraba, Koster, Campbell, & Scholz, 2015; Romano, Babchishin, Marquis, & Fréchette, 2015). Multiple placement changes common among adolescents in foster care (Courtney et al., 2004; Sullivan, Jones, & Mathiesen, 2010) lead to gaps in school attendance, create discontinuities in learning, require acclimation to each new school community and culture, and impede the development of lasting connections to school staff and peers (Allen & Vacca, 2010; Romano et al., 2015). Among foster youth, maltreatment and placement instability have been linked to an increased risk of psychopathology, such as depression, posttraumatic stress disorder, externalizing behavior disorders (Norman et al., 2012; Zlotnick, Tam, & Soman, 2012), and substance-use disorders (Edalati & Krank, 2015). Moreover, foster youth experience high rates of school suspension, grade retention, and placement in special-education classrooms (Courtney et al., 2004), as well as delays in high school completion and academic preparedness (Frerer et al., 2013; Unrau, Font, & Rawls, 2012). In addition, foster youth often attend schools in low-income communities that fall below performance benchmarks and lack needed resources (Frerer et al., 2013). Although seldom tested empirically, the cascading risks of maltreatment, instability, psychopathology, and educational setbacks are thought to underlie the low rates of college enrollment for foster youth.
Another factor that may influence foster youths’ likelihood of enrolling in college is social capital—the resources embedded in relational ties (Coleman, 1988). Several qualitative studies have underscored the importance of having an invested and knowledgeable adult to assist foster youth with accessing college (e.g., Batsche et al., 2014; Salazar, Jones, Emerson, & Mucha, 2016). For example, Hines, Merdinger, and Wyatt (2005) interviewed foster-care alumni attending a 4-year college and found that the presence of a “crucial person” who served as role model, advocate, and source of encouragement was instrumental to the youths going to college. However, these studies include just youths who enrolled in college; no data were collected on those who did not enroll. Thus, it is not possible to know whether having crucial support individuals is a distinguishing feature between foster youth who made it to college and those who did not. One quantitative study found marginal support for a positive association between social capital and college enrollment (Ahrens, DuBois, Richardson, Fan, & Lozano, 2008), but only relationships with mentors were considered. Other quantitative studies investigated the association between social capital and college enrollment among foster youth, but these studies are limited by shortcomings such as low response rates and insufficiently accounting for possible confounders (Merdinger, Hines, Lemon-Osterling, & Wyatt, 2005; Finnie, 2012). And, these studies all relied on retrospective recall of social capital, which may be susceptible to recall bias and reporting errors.
Current Study
In the current investigation, we used a representative sample of foster youth from a state with a large population of foster youth to rigorously examine the relationship between social capital and college enrollment. The study addressed limitations of prior investigations by using a representative sample and including covariates at age 17 to predict college enrollment at age 20. Finally, our statistical analyses controlled for a wide range of factors that could plausibly confound the relationship between social capital and college enrollment.
Theoretical Framework
Social capital theory is used as the conceptual framework for this study. Lin (2001) identified four explanations of how social capital can affect the likelihood that actions will lead to attainment of goals: information, influence, social credentials, and reinforcement. We applied these explanations to the goal of college enrollment. First, network ties can supply youth with important information about college that would otherwise be unavailable, either by providing information directly or linking youth to sources of information (e.g., assisting youth with technical aspects of the application process, referring youth to college guidebooks, and exploring viable options for financing college). Second, social ties can exert influence on youths’ decisions, such as the types of colleges that they choose to apply to (e.g., 2-year vs. 4-year, local vs. distant, public vs. private). Third, social ties can influence others’ investment of the time, assistance, and resources needed to get youth into college. Fourth, social ties can reinforce youths’ identities as future college students.
For many aspiring college students, immediate family members provide the social and material resources integral in applying to and enrolling in college (McNamara-Horvat, Weininger, & Lareau, 2003; Sandefur, Meier, & Cambell, 2006). However, like other low-income students, foster youth often come from families where college attainment is not the norm (Barth, Wildfire, & Green, 2010). Parents and family members without college experience are limited in the substantive knowledge they can offer about planning for and executing the college choice process (Choy, 2001; Holland, 2010). Moreover, the strain, demands, and expectations that sometimes arise from these ties can hinder young people from enrolling and remaining in college (Engle & Tinto, 2008).
Consequently, foster youth must rely on adults outside of their families for types of social capital that are instrumental in navigating the college choice process. Stanton-Salazar (2011) refers to these individuals as institutional agents—nonfamily adults who occupy “one or more hierarchical positions of relatively high status and authority” and who act on behalf of adolescents to “directly transmit or negotiate the transmission of highly valued resources” (p. 1067). Such individuals can serve a variety of roles (e.g., resource agent, knowledge agent, advisor, advocate, networking coach, cultural guide) and may leverage their positional power or mobilize personal resources to assist adolescents in achieving a targeted goal (Stanton-Salazar, 1997, 2011). Agents within a youth’s social network serve a bridging function (Lin, 2001; Portes, 1998), connecting the youth to information and resources that would otherwise be limited or unavailable.
Despite the influential role that institutional agents could play in helping foster youth attend college, there are often barriers to access and utilization. Foster youth frequently attend schools classified as low performing (Frerer et al., 2013), where guidance counselors are typically responsible for many students, have extraneous tasks competing for their limited time, and tend to place greater emphasis on high school completion than college enrollment (Bryan et al., 2011; Plank & Jordan, 2001). Youth in foster care have access to other professionals who could hypothetically serve as institutional agents, such as child welfare caseworkers, surrogate caregivers (e.g., foster parents, group home staff), independent-living service providers, and legal advocates. However, large caseloads and a high staff turnover common in child welfare (Kim & Kao, 2014), as well as frequent placement changes experienced by many foster youth (Courtney et al., 2004), can attenuate or sever ties to adults (Samuels, 2008).
Research Questions
Drawing on the preceding theoretical concepts, we proposed three hypotheses about the relationship between foster youths’ social capital and the likelihood of enrolling in college:
• H1—a greater number of institutional agents who youth can turn to for tangible support and/or advice will increase youths’ likelihood of enrolling in college;
• H2—youth who receive encouragement from school professionals to pursue higher education will be more likely to enroll in college than youth who receive little or no encouragement; and
• H3—having both school encouragement and institutional agents will amplify youths’ likelihood of enrolling in college.
In addition, we tested six competing hypotheses that contest specific parts of our main hypotheses:
• A1—having a greater number of noninstitutional agents who youth can turn to for tangible support and advice will increase youths’ likelihood of enrolling in college (contests H1);
• A2—institutional agents who provide emotional support (but not tangible support or advice) will increase youths’ likelihood of enrolling in college (contests H1);
• A3—a larger overall support network will increase youths’ likelihood of enrolling in college (contests H1);
• A4—youth who report having enough people to turn to for tangible support and advice are more likely to enroll in college (contests H1);
• A5—youth who receive encouragement from foster-care professionals to pursue higher education will be more likely to enroll in college than youth who receive little or no encouragement to enroll in college (contests H2); and
• A6—youth who receive encouragement from their biological family to pursue higher education will be more likely to enroll in college than youth who receive little or no encouragement (contests H2).
Method
Sample
The sample for this study included young people who participated in the first wave of the California Youth Transition to Adulthood Study (CalYOUTH; Courtney, Charles, Okpych, Napolitano, & Halsted, 2014; Courtney et al., 2016). Youths were eligible for the study if they were between the ages of 16.75 and 17.75 at the time the sample was selected in 2013 and if they had been in California foster care for at least 6 months. Once the population of youths meeting these criteria was identified, a stratified random sample was selected (N = 880). The 58 California counties were divided into six groups based on the number of eligible youths residing in each county, and counties with fewer youth were oversampled to ensure adequate representation in the study (for more information, see Courtney et al., 2014). Youths were excluded if they were incarcerated, physically or mentally unable to complete the study interviews, ran away from their foster-care placement for at least 2 months, returned home for at least 2 months, and/or relocated out of state. After excluding ineligible youths, the final sample consisted of 763 adolescents.
Data Collection
Baseline in-person interviews were completed (a 95% response rate) with 727 of the 763 eligible youths between April and October of 2013. Response rates did not significantly differ across the six sampling strata, ranging from 93.5% to 96.8%. The interviews covered 20 content areas and took approximately 90 minutes to complete. Sections of the survey instrument that addressed sensitive topics (e.g., mental health, delinquency, past maltreatment) were administered using audio-enhanced, computer-assisted self-interviewing. Youths received a $50 cash incentive for participation.
Of the 727 participants, 713 granted permission for CalYOUTH researchers to access administrative data on their college enrollment. One participant died after the baseline interview, and the 712 remaining youths constitute the sample for the present analysis. Data from the baseline interviews were linked to National Student Clearinghouse (NSC) data on college enrollment. These data were acquired in February 2016, when most study participants were 20 years old. The NSC is a 501(c)(6) nonprofit, nongovernmental organization that provides information on enrollment status and degree records for more than 3,600 public and private U.S. postsecondary institutions, which comprise about 98% of the nation’s postsecondary student body (NSC, 2016a).
In total, NSC college enrollment records were found for 379 of the 712 participants. There were two main concerns that could lead to errors in describing participants’ enrollment status. First, participants could be classified as being enrolled when they should not be designated as such. This can occur if participants had not completed a high school credential but were enrolled in basic education courses at 2-year and community colleges (e.g., to help complete a GED). Second, participants could have enrolled in college but were not identified as such in NSC records, either because the student’s record was blocked or because the college they attended did not report data to NSC (undercoverage; Dynarski, Hemelt, & Hyman, 2013).
We used self-report data from the 2013 CalYOUTH baseline interview and the 2015 Wave 2 follow-up interview to address both data issues. To identify college enrollees who had not completed a secondary credential, we cross-walked NSC enrollment dates with CalYOUTH interview dates and self-reported secondary credential status. A total of 17 youths appeared in NSC records but had not earned a secondary credential at the time they were enrolled in college. These 17 participants were recoded as being not enrolled in college.
CalYOUTH data were also used to handle blocked records and undercoverage. There were 27 blocked records, meaning that 27 participants had been enrolled in a postsecondary institution but could not be identified. In terms of NSC coverage, the nationwide coverage rate in fall 2015 was 96.7%, which was the same as the coverage rate for California (NSC, 2016b), where about 95% of college enrollees in our sample attended school. Assuming that 379 participants identified in NSC data comprised about 96.7% of the sample enrolled in college, we estimated that approximately 13 additional participants had enrolled in college but were not included in the NSC records. Combining the 27 blocked records and 13 students missing due to undercoverage, we estimated that 40 participants had enrolled in college but were not identified as such in the NSC data.
Overall, there was a high rate of agreement (92.9%) between the college status appearing in NSC records and the self-reported college status obtained from interviews (Cohen’s kappa = .85, Z = 21.1, p <.0001). We subsequently used self-report data from the baseline and Wave 2 CalYOUTH interviews to identify the students missing from NSC records. Participants were counted as being enrolled in college if (a) at the baseline interview they had completed a high school credential and were currently enrolled in college (n = 6); (b) they completed a high school credential by their baseline interview, and in their Wave 2 interview they reported being currently enrolled in college or enrolled in college since the baseline interview (n = 9); or (c) at the Wave 2 interview they had completed a high school credential and were currently enrolled in college (n = 22). A total of 37 students met one of these criteria, which was close to our estimate of the number of missing students. Drawing on NSC data and CalYOUTH self-reports, a total of 399 participants were designated as being enrolled in college.
Measures
Outcome measure: College enrollment status
The outcome is a binary measure of whether a study participant had ever enrolled in a 2-year or 4-year college by February 2016.
Main predictors: Social capital
Institutional agents (tangible support/advice)
The Social Support Network Questionnaire, a brief instrument used to capture information on a wide range of characteristics of individuals’ social support networks, was used to gather information on the number and quality of support relationships (Gee & Rhodes, 2007). Some modifications were made to the original instrument to meet time constraints and characteristics of the foster-care population. For example, the number of individuals who could be nominated by participants was limited to three for each of the support types, and several new relationship-type categories were added to reflect the types of individuals foster youth may encounter (e.g., foster parent, social worker, independent-living-program staff, attorney).
Participants were asked to nominate up to three individuals who they would most likely turn to for emotional support (talk to about something personal or private), up to three individuals they would turn to for tangible support (give or lend the youth something that is needed, or assist with something the youth needed to do), and three individuals they would turn to for advice (provide guidance or information if the youth did not know where to get something or do something). In total, participants could nominate up to nine distinct support individuals; for each nominee, participants were asked about their relationship to the person. Institutional agents was a count of the number of individuals who the participant turned to for tangible support and/or advice and who belonged to one of the following relationship roles: teacher, school counselor, caseworker, therapist/counselor, mentor, nonrelative foster parent, or other professional. These roles include individuals with firsthand experience with college. The number of distinct institutional agents ranged from zero to six. The Pearson’s correlation for the number of institutional agents providing tangible support and the number of institutional agents providing advice/guidance was .49 (p < .001), and the Chronbach’s alpha for these two items was .65. Chronbach’s alpha for the institutional-agents measure was .70 (Peterson, 1994).
School encouragement
The measure of school encouragement was constructed from a question that asked participants how much encouragement they received from professionals at their school (i.e., teachers, guidance counselors, and administrators) to purse their education beyond high school. A binary variable was created, which distinguished participants who received “a lot” of encouragement from those who received less encouragement (“none,” “some,” and “a little”). Combining the three categories reflected a meaningful cut point, avoided issues of statistical power (the original categories could be statistically underpowered to detect true differences), and was supported by correlates of related covariates.
Noninstitutional agents (tangible support/advice)
Drawing on data from the Social Support Network Questionnaire, a measure for noninstitutional agents included the number of nominated individuals who participants would turn to for tangible support or advice but who did not fit one of the seven institutional agent relationship roles (teacher, school counselor, caseworker, therapist/counselor, mentor, nonrelative foster parent, or other professional). Noninstitutional agents include roles such as relatives (e.g., siblings, biological parents, step-parents, aunts/uncles, and grandparents); peers (e.g., friends, coworkers, classmates, and romantic partners); and other individuals (e.g., relative foster parents, coaches). The variable ranged from zero to six distinct individuals.
Institutional agents (emotional support)
A count variable was created for the number of nominees who participants would turn to for emotional support and who belonged to one of the seven institutional agent relationship roles. The variable ranged from zero to three distinct individuals.
Estimated network size
One survey item asked participants to estimate the number of individuals they could turn to for emotional support, tangible support, or advice. The original response range was 0 to 99, but the measure was capped at 30. Fewer than 5% of participants reported having more than 30 individuals in their support network.
Adequacy of support (tangible support/advice)
Participants rated whether they had enough people to turn to for tangible support and for advice. The response choices included “enough people you can count on,” “some but not enough people you can count on,” and “no one you can count on.” A dummy variable was created for adequacy of tangible support and advice; 1 indicated that they had “enough” support in both areas, and 0 indicated otherwise.
Biological family encouragement
Participants were asked to report the amount of encouragement they received to pursue education beyond high school from their biological family (e.g., birth parents, aunts and uncles, etc.). A dummy variable was created for participants who received “a lot” of encouragement versus “no,” “a little,” or “some” encouragement.
Foster-care encouragement
Participants also reported the amount of encouragement they received to pursue postsecondary education from professionals in the foster-care system (e.g., caseworkers, foster parents, group home staff). A dummy variable was created for participants who received “a lot” of encouragement versus “no,” “a little,” or “some” encouragement.
Covariates
Demographic characteristics
Covariates included participants’ gender, race/ethnicity, age at the time of their baseline interview, age at the time that NSC data were acquired, and county size groups. The 51 counties with study participants were divided into the following groups: rural (n = 18, all municipalities in the county had a population smaller than 50,000); urban (n = 19, at least one municipality in the county with a population between 50,000 and 250,000); large urban (n = 12, at least one municipality in the county with a population greater than 250,000); and Los Angeles County.
Personality traits
Because differences in personality traits may influence youths’ social capital and likelihood of pursuing college, a second group of covariates included measures of five personality traits: extroversion, agreeableness, conscientiousness, neuroticism, and openness to new experiences. These measures were taken from a brief 10-item version of the Big Five personality assessment (Gosling, Rentfrow, & Swann, 2003). Each trait had a response range of 0 to 12, with a greater score indicating a stronger presence of that trait.
Education status and achievement
A dummy variable was created to assess whether participants had or had not earned a secondary credential (i.e., high school diploma, GED, or alternate certificate). A measure of the participants’ reading proficiency was obtained from the Wide-Range Achievement Test: Fourth Edition (Wilkinson & Robertson, 2006), a standardized test that provides a brief assessment of basic academic skills. Raw scores were converted to a standardized scale that is similar to the IQ scale (M = 100, SD = 15). High school grades were measured by self-report reflecting mostly As, mostly Bs, mostly Cs, or mostly Ds or lower.
Educational aspirations and preparedness
Participants were asked about the highest level of education they aspired to attain, which included graduating from high school or less, some college, earning a college degree, and more than a college degree. Another survey item asked participants how prepared they felt to continue their education beyond high school. The four original response options were combined into three categories: (a) “not prepared” or “somewhat prepared,” (b) “prepared,” and (c) “very prepared.” Similarly, the four response options to a survey question that asked participants about the amount of preparation, support services, and training they received to complete their education or job training goals were collapsed into three categories: (a) “none” or “a little,” (b) “some,” and (c) “a lot.”
History of education difficulties
The fifth group of control variables were self-reported measures of school difficulties and involvement in special education. Three dummy variables indicated whether participants had ever repeated a grade, ever been expelled from school, and ever been in a special education classroom.
Foster-care characteristics
Six measures were used to capture different dimensions of the participants’ foster-care history that may influence their educational trajectories. Participants reported the number of days they missed school because of foster care (e.g., court hearings), as well as the number of school changes due to a foster-care placement change or a family move. These two items originally had ranges of 0 to 99 but were both top-coded at 20. Current foster-care placement captured participants’ type of living arrangement at the time of their baseline interview and included a nonrelative foster-care home, a relative foster-care home, a group home or residential treatment facility, and other placements (e.g., independent living facility, adoptive home). Three variables examined experiences of physical abuse by caregivers, neglect by caregivers, and sexual abuse. For each maltreatment type, participants were asked multiple yes or no questions about different types of maltreatment (e.g., “your caregiver beat you up, such as hitting or kicking you repeatedly”). Count variables of affirmative responses were created for physical abuse (range: 0–7), neglect (range: 0–9), and sexual abuse (range: 0–2; see Courtney et al., 2014, for specific items). Chronbach’s alphas for physical abuse, neglect, and sexual abuse were 0.88, 0.83, and 0.80, respectively.
Potential hindrances
Three dummy variables indicated whether participants screened positive for depression (major depressive episode), a substance-use problem (dependence or abuse), or an alcohol-use problem (dependence or abuse) at the time of the baseline interview. These measures were taken from the results of a computerized version of the MINI-KID, a standard mental health screening tool for nonprofessionals used widely to assess behavioral health problems in children and adolescents (Sheehan et al., 2010). Additionally, dummy variables were created to indicate if participants ever spent a night in a correctional facility and if they had a child by the time of their baseline interview.
Employment history
A dummy variable was created to indicate whether a participant had ever worked for pay outside of their home in the 4 weeks prior to the baseline interview.
Mediators of social capital on college enrollment
College help
As described in the following section, the amount of help participants received with college preparation is examined as a possible mediating link between institutional agents and college enrollment. During the Wave 2 interview, when most study participants were 19 years old, participants were asked to report the amount of help they received with college preparation (e.g., deciding about going to college, assistance with applications and financial aid forms, etc.). Participants could select from the following six options: “no help,” “only a little help,” “some help but not enough,” “enough help,” “more than enough help,” and “not interested in going to college.”
Participation in extended foster care
A variable was created to count the number of months (1 month = 30 days) participants remained in care past the age of 18. Youths who exited care before age 18 were coded as 0 months. Remaining in care past age 18 is a second suspected mediator of the relationship between social capital (institutional agents, in particular) and college enrollment.
Analyses
Analyses were conducted using Stata 14. Survey weights were used in all analyses, accounting for features of the sampling design and nonresponse, and expanding results to the population of California foster youths meeting the study criteria. Our original analysis plan entailed modeling college enrollment as three categories (never enrolled, enrolled in a 2-year college, and enrolled in a 4-year college). However, only a small proportion of our sample had enrolled in 4-year colleges, which yielded large standard errors and inadequate statistical power in our multinomial regression analyses. Thus, we decided to model college enrollment as two categories, distinguishing participants who had enrolled in 2-year or 4-year colleges from participants who had never enrolled in college. As a supplement, we present results from our three-category analyses for just our main predictors (social capital), with the caveat that these findings should be interpreted cautiously.
To test our three hypotheses and the competing hypotheses, we ran separate logistic regression models in which the log odds of college enrollment were regressed on each measure of social capital. If a given measure of social capital was significantly related to the likelihood of enrolling in college (p < .05), we then added the covariates described earlier and reexamined the results. To test the interaction between the count of institutional agents and school encouragement, we added an interaction term for these two variables to the full model. Finally, to test whether the amount of help with college preparation mediates the relationship between institutional agents and college enrollment, we added the measure of college help (measured at age 19) to the full model. Running a test of mediation is important because of a limitation with our measure of institutional agents. Although we know that these agents are individuals who participants turn to for tangible support and advice, it is not possible to know if they provided help with getting into college, specifically. However, if having more institutional agents increases the likelihood of college enrollment, and if the strength of the association weakens when the amount of college help participants received is added to the model, this suggests that institutional agents played a role in participants getting into college through help they offered to the participants. A similar test was run to examine the extent to which remaining in care past age 18 mediated the relationship between social capital and college enrollment.
In total, 167 of the 712 participants had missing data on at least one of the covariates included in the full regression model (23.5% of the sample). However, most of these participants were missing data on just one or a few variables. We used multiple imputation by chained equations to address missing data (Royston & White, 2011). For each multiple imputation model, we executed 20 imputations and 50 cycles of regression switching between imputations.
Findings
Sample Characteristics
About 60% of participants were female, and the greatest proportion of participants identified as Hispanic. On average, participants were 17.5 years old at the time of their baseline interview and 20.2 years old at the time NSC data were accessed. The standardized reading score for participants was more than two thirds of a standard deviation below the average reading score for same-aged peers. Over 80% hoped to earn a college degree or more, and about one fifth indicated that they were not prepared to achieve their education goals. One third of participants had repeated a grade, a little over one quarter had been expelled, and one third had been in a special-education classroom.
On average, participants missed school about three times due to reasons associated with foster care (Mdn = 1) and changed schools six times because of family moves or changes in their foster-care placement (Mdn = 4). About one fifth of participants screened positive for a major depressive episode, and substance-use problems were more common than alcohol-use problems. One quarter of participants had spent a night in a correctional facility, and about 7% were parents at the time of the interview. In terms of mediators, after excluding participants who were not interested in going to college, about half of the participants said they received “enough” or “more than enough” help preparing for college. Most study participants were in care on or after their 18th birthday (93%), and on average they spent about 15 months in care after age 18 (range: 0–33.4 months). See Table 1 for additional sample characteristics.
| Sample Characteristics | |
|---|---|
| Demographic Characteristics | |
| Male | 40.0% |
| Race/ethnicity | |
| White | 18.0% |
| Black | 17.3% |
| Asian/Pacific Islander/American Indian/Alaskan Native | 2.5% |
| Mixed race | 15.4% |
| Hispanic | 46.8% |
| Age at baseline interview (Mean/SD) | 17.5 (0.3) |
| Age at National Student Clearinghouse data draw (Mean/SD) | 20.2 (0.3) |
| County group | |
| Rural/suburban | 4.8% |
| Urban | 20.7% |
| Large urban | 43.1% |
| Los Angeles County | 31.5% |
| Personality traits (Mean/SD) | |
| Extroversion | 6.8 (2.6) |
| Agreeableness | 7.5 (2.0) |
| Conscientiousness | 8.9 (2.4) |
| Neuroticism | 4.7 (2.6) |
| Openness | 8.6 (2.3) |
| Education status and achievement | |
| Highest completed grade | |
| 9th grade or lower | 8.6% |
| 10th grade | 31.6% |
| 11th grade | 48.5% |
| 12th grade | 11.3% |
| High school credential status | |
| No high school credential | 89.2% |
| Diploma, GED, or certificate | 10.8% |
| Reading level, standardized (Mean/SD) | 89.3 (11.3) |
| High school grades | |
| Mostly As | 13.8% |
| Mostly Bs | 32.5% |
| Mostly Cs | 43.7% |
| Mostly Ds | 10.1% |
| Education aspirations and preparedness | |
| Education aspirations | |
| High school credential or less | 13.1% |
| Some college | 5.7% |
| College degree | 48.0% |
| More than college degree | 33.2% |
| Preparedness to achieve education goals | |
| Not/somewhat prepared | 21.3% |
| Prepared | 35.3% |
| Very prepared | 43.4% |
| Amount of education services | |
| None/few | 18.3% |
| Some but not enough | 48.5% |
| Enough/a lot | 33.2% |
| History of education difficulties | |
| Ever repeated a grade | 33.2% |
| Ever expelled | 27.6% |
| Ever in special education | 33.6% |
| Foster-care characteristics | |
| Number of times missed school (Mean/SD) | 2.7 (3.9) |
| Number of school changes (Mean/SD) | 5.9 (5.2) |
| Current living placement | |
| Nonrelative foster home | 44.4% |
| Relative foster home | 18.2% |
| Group care | 24.0% |
| Other | 13.5% |
| Physical abuse (Mean/SD) | 1.5 (2.2) |
| Neglect (Mean/SD) | 1.8 (2.3) |
| Sexual abuse (Mean/SD) | 0.5 (0.8) |
| Potential hindrances | |
| Depression | 20.5% |
| Alcohol-use problem | 12.5% |
| Substance-use problem | 21.3% |
| Ever incarcerated | 25.2% |
| Parented a child | 6.8% |
| Employment history | |
| Ever had a job | 32.5% |
| Mediators | |
| Help with college (measured at Wave 2) | |
| No help | 13.3% |
| Only a little help | 13.9% |
| Some help, but not enough | 19.0% |
| Enough help | 23.3% |
| More than enough help | 20.3% |
| Not interested in going to college | 10.3% |
| Months in care past age 18 (Mean/SD) | 15.2 (7.5) |
Table 2 displays descriptive statistics for the measures of social capital. On average, participants nominated less than one institutional agent who they could turn to for tangible support and/or advice. Indeed, more than half of participants nominated no institutional agents who provided these types of support (53.8%). Participants nominated 510 institutional agents. About half of these individuals were nonrelative foster parents (49.2%). Caseworkers made up about 15% of the nominees (14.8%), and school personnel, mentors, and therapists each comprised less than 10% of the nominees (8.1%, 9.3%, and 8.3%, respectively). Other types of professionals made up 10.2% of the agents. Figure 1 displays the types of support (advice only, tangible support only, and advice and tangible support) that different institutional agents provided. Half or more of caseworkers, teachers, school counselors, therapists, and other professionals were people the participants turned to for advice only. Foster parents, mentors, and other professionals were more likely than the other institutional agents to provide tangible support or a combination of tangible support and advice.
| Main hypotheses | |
| Institutional agents (tangible/advice) (Mean/SD) | 0.7 (0.8) |
| School encouragement | 63.1% |
| Competing hypotheses | |
| Noninstitutional agents (tangible/advice) (Mean/SD) | 2.3 (1.3) |
| Institutional agents (emotional) (Mean/SD) | 0.1 (0.4) |
| Estimated network size (0–30) (Mean/SD) | 7.6 (6.6) |
| Adequate amount of tangible/advice | 54.2% |
| Biological family encouragement | 60.6% |
| Foster-care encouragement | 69.0% |
Types of support(s) provided by institutional agents (n = 510 institutional agents, weighted). “Other professional” includes another type of professional nominated by participants but not included in the available survey response categories (e.g., court-appointed special advocate).
Less than two thirds of participants said they received “a lot” of encouragement from professionals at their school to pursue postsecondary education. On average, participants nominated 2.3 people who were not institutional agents who they turned to for tangible support and/or advice (Mdn = 2), and most participants did not have an institutional agent who they turned to for just emotional support. Overall, participants’ social support network consisted of approximately eight people (Mdn = 5). Most participants said they had “enough” people to turn to for both tangible support and advice. More than 60% of participants reported receiving “a lot” of encouragement to pursue postsecondary education from their biological family and from foster-care professionals. About 55% of participants (n = 399) had enrolled in college by the date the NSC data were accessed. Among those who enrolled, 84.8% were attending a 2-year college, and 15.2% were attending a 4-year college.
Social Capital as a Predictor of College Enrollment
Table 3 summarizes the results of bivariate logistic regression analyses of the likelihood of enrolling in college on each of the social-capital measures. Results are presented as odds ratios (OR). Both the number of institutional agents providing tangible support and/or advice and receipt of school encouragement increased the estimated odds of college enrollment. The third hypothesis purporting an interaction effect between institutional agents and school encouragement was not supported. In terms of the competing hypotheses, none of the measures of social capital predicted college enrollment except support network size (A3); larger networks predicted a greater likelihood of college enrollment.
| Measure | Bivariate Model | p value | 95% Confidence Interval | |||
|---|---|---|---|---|---|---|
| Odds Ratio | Standard Error | t value | Lower Limit | Upper Limit | ||
| H1. Institutional agents (tangible/advice) | 1.36 | 0.14 | 2.96 | .003 | 1.11 | 1.67 |
| H2. School encouragement | 1.43 | 0.25 | 2.02 | .044 | 1.01 | 2.02 |
| H3. Interaction of IA and school encouragement | 1.31 | 0.28 | 1.26 | .209 | 0.86 | 1.98 |
| A1. Noninstitutional agents (tangible/advice) | 0.99 | 0.07 | −0.12 | .903 | 0.87 | 1.13 |
| A2. Institutional agents (emotional) | 1.18 | 0.24 | 0.81 | .416 | 0.79 | 1.77 |
| A3. Estimated network size (0–30) | 1.04 | 0.01 | 2.71 | .007 | 1.01 | 1.06 |
| A4. Adequate amount of tangible/advice | 1.15 | 0.20 | 0.82 | .415 | 0.82 | 1.61 |
| A5. Biological family encouragement | 0.79 | 0.14 | −1.35 | .178 | 0.56 | 1.11 |
| A6. Foster-care encouragement | 0.98 | 0.18 | −0.10 | .917 | 0.68 | 1.41 |
Results from the full model that included school encouragement and institutional agents are presented in Table 4. After controlling for the wide array of youth, educational, and foster-care characteristics, institutional agents and school encouragement both predicted an increase in the likelihood of college enrollment. The estimated odds of enrolling in college were about 39% greater for each additional institutional agent nominated, and the estimated odds of enrolling in college were 70% greater for participants who received “a lot” of school encouragement than for those who received “no” or “some” school encouragement.
| Odds Ratio | Standard Error | t value | p value | 95% Confidence Interval | ||
|---|---|---|---|---|---|---|
| Lower Limit | Upper Limit | |||||
| Social capital | ||||||
| Number of institutional agents | 1.39 | 0.17 | 2.66 | .008 | 1.09 | 1.78 |
| School encouragement | 1.70 | 0.39 | 2.31 | .022 | 1.08 | 2.66 |
| Demographic characteristics | ||||||
| Male (ref: female) | 1.01 | 0.23 | 0.04 | .964 | 0.64 | 1.59 |
| Race/ethnicity (ref: White) | ||||||
| Black | 1.54 | 0.51 | 1.30 | .195 | 0.80 | 2.97 |
| Asian/PI/AI/AK | 2.97 | 1.67 | 1.94 | .053 | 0.99 | 8.94 |
| Mixed race | 1.37 | 0.47 | 0.89 | .371 | 0.69 | 2.70 |
| Hispanic | 0.85 | 0.22 | −0.63 | .527 | 0.51 | 1.41 |
| Age at Wave 1 interview | 0.18 | 0.17 | −1.78 | .075 | 0.03 | 1.19 |
| Age at NSC data draw | 9.24 | 9.06 | 2.27 | .024 | 1.35 | 63.34 |
| County group (ref: rural/suburban) | ||||||
| Urban | 0.70 | 0.26 | −0.97 | .330 | 0.34 | 1.44 |
| Large urban | 0.91 | 0.33 | −0.24 | .807 | 0.45 | 1.87 |
| Los Angeles County | 0.80 | 0.33 | −0.54 | .591 | 0.36 | 1.78 |
| Personality traits | ||||||
| Extroversion | 1.07 | 0.46 | 1.53 | .126 | 0.98 | 1.16 |
| Agreeableness | 1.03 | 0.06 | 0.52 | .606 | 0.92 | 1.16 |
| Conscientiousness | 0.98 | 0.05 | −0.51 | .613 | 0.89 | 1.07 |
| Neuroticism | 1.01 | 0.04 | 0.23 | .818 | 0.93 | 1.10 |
| Openness | 0.89 | 0.04 | −2.51 | .012 | 0.80 | 0.97 |
| Education status | ||||||
| Highest completed grade (ref: 9th grade or less) | ||||||
| 10th grade | 1.23 | 0.52 | 0.50 | .615 | 0.54 | 2.82 |
| 11th grade | 1.70 | 0.71 | 1.26 | .207 | 0.75 | 3.85 |
| 12th grade | 0.77 | 0.58 | −0.35 | .723 | 0.18 | 3.34 |
| High school credential (ref: none) | 4.23 | 3.21 | 1.90 | .058 | 0.95 | 18.76 |
| Reading level (standardized) | 1.03 | 0.01 | 3.07 | .002 | 1.01 | 1.06 |
| High school grades (ref: mostly As) | ||||||
| Mostly Bs | 0.82 | 0.27 | −0.59 | .557 | 0.43 | 1.58 |
| Mostly Cs | 0.38 | 0.12 | −3.00 | .003 | 0.20 | 0.72 |
| Mostly Ds or lower | 0.55 | 0.23 | −1.45 | .146 | 0.24 | 1.23 |
| Education aspirations and preparedness | ||||||
| Education aspirations (ref: high school) | ||||||
| Some college | 1.13 | 0.62 | 0.22 | .824 | 0.39 | 3.30 |
| College degree | 1.83 | 0.61 | 1.79 | .073 | 0.94 | 3.53 |
| More than college degree | 2.17 | 0.76 | 2.22 | .027 | 1.09 | 4.31 |
| Preparedness to achieve education goals (ref: not/somewhat prepared) | ||||||
| Prepared | 0.74 | 0.22 | −1.04 | .300 | 0.41 | 1.31 |
| Very prepared | 0.80 | 0.24 | −0.76 | .450 | 0.45 | 1.43 |
| Amount of education services (ref: none/few) | ||||||
| Some but not enough | 1.17 | 0.32 | 0.57 | .570 | 0.68 | 2.00 |
| Enough/a lot | 0.89 | 0.27 | −0.39 | .693 | 0.49 | 1.61 |
| History of education difficulties | ||||||
| Ever repeated a grade | 0.52 | 0.13 | −2.71 | .007 | 0.32 | 0.83 |
| Ever expelled | 1.00 | 0.26 | −0.01 | .990 | 0.60 | 1.66 |
| Ever in special education | 0.54 | 0.13 | −2.55 | .011 | 0.34 | 0.87 |
| Foster-care characteristics | ||||||
| Number of times missed school | 1.01 | 0.03 | 0.21 | .830 | 0.96 | 1.06 |
| Number of school changes | 0.98 | 0.02 | −1.12 | .265 | 0.94 | 1.02 |
| Current living placement (ref: nonrelative foster home) | ||||||
| Relative foster home | 1.97 | 0.59 | 2.28 | .023 | 1.10 | 3.54 |
| Group care | 1.13 | 0.31 | 0.45 | .651 | 0.66 | 1.95 |
| Other | 1.33 | 0.41 | 0.93 | .354 | 0.73 | 2.45 |
| Physical abuse | 1.20 | 0.08 | 2.84 | .005 | 1.06 | 1.38 |
| Neglect | 1.01 | 0.07 | 0.13 | .894 | 0.89 | 1.15 |
| Sexual abuse | 1.19 | 0.18 | 1.19 | .233 | 0.89 | 1.59 |
| Potential hindrances | ||||||
| Depression | 0.63 | 0.17 | −1.71 | .088 | 0.36 | 1.07 |
| Alcohol-use problem | 0.75 | 0.27 | −0.78 | .434 | 0.37 | 1.54 |
| Substance-use problem | 0.68 | 0.20 | −1.34 | .180 | 0.38 | 1.20 |
| Ever incarcerated | 0.85 | 0.23 | −0.60 | .552 | 0.50 | 1.45 |
| Parented a child | 0.31 | 0.14 | −2.55 | .011 | 0.12 | 0.76 |
| Employment history | ||||||
| Ever had a job | 1.16 | 0.24 | 0.70 | .485 | 0.77 | 1.74 |
We also examined a regression model that included the interaction between the number of institutional agents and school encouragement. The interaction term approached but was not below the conventional p < .05 cutoff (OR = 1.49, p = .094, 95% CI [0.93, 2.37]).
The next step in our analyses was to investigate whether the amount of college help and participation in extended foster care mediated the relationship between institutional agents and college enrollment. Abbreviated results, which do not display covariate estimates, are displayed in Table 5. For comparison, the first column reproduces the results from Table 4. After college help is added to the model (second column), the estimated odds ratio for the number of institutional agents decreased by 18%. Additionally, the estimated odds ratio for school encouragement decreased by 19% and is no longer statistically significant. The decreases in the strengths of the associations suggest that the relationship between social capital and college enrollment is at least partly explained by the amount of college help participants received. As reported in the third column, months in care after age 18 also accounts for some of the association between the two social-capital measures and college enrollment. Moreover, months in care past age 18 is a significant predictor of college enrollment, with each additional month in care increasing the odds of enrollment by about 7%. The fourth column displays the results when both mediators (college help and months in care past 18) were added to the model. In this model, the association between the two social-capital measures both fall below the .05 significance level.
| Odds Ratio [95% CI] | p value | Odds Ratio [95% CI] | p value | Odds Ratio [95% CI] | p value | Odds Ratio [95% CI] | p value | |
|---|---|---|---|---|---|---|---|---|
| Number of institutional agents | 1.39 | .008 | 1.32 | .035 | 1.35 | .017 | 1.29 | .053 |
| [1.09, 1.78] | [1.02, 1.71] | [1.06, 1.74] | [0.99, 1.68] | |||||
| School encouragement | 1.70 | .022 | 1.57 | .070 | 1.70 | .026 | 1.57 | .074 |
| [1.09, 2.67] | [0.96, 2.54] | [1.07, 2.71] | [0.96, 2.58] | |||||
| Help with college (ref: no help) | ||||||||
| Only a little help | – | – | 2.41 | .075 | – | – | 2.26 | .108 |
| [0.92, 6.33] | [0.84, 6.13] | |||||||
| Some help, but not enough | – | – | 1.80 | .218 | – | – | 1.74 | .262 |
| [0.71, 4.60] | [0.66, 4.61] | |||||||
| Enough help | – | – | 4.63 | .001 | – | – | 4.06 | .003 |
| [1.91, 11.2] | [1.59, 10.3] | |||||||
| More than enough help | – | – | 3.70 | .005 | – | – | 3.06 | .021 |
| [1.50, 9.14] | [1.18, 7.93] | |||||||
| Not interested in going to college | – | – | 0.41 | .093 | – | – | 0.43 | .115 |
| [0.15, 1.16] | [0.15, 1.23] | |||||||
| Months in foster care past age 18 | – | – | – | – | 1.07 | < .001 | 1.06 | < .001 |
| [1.04, 1.11] | [1.03, 1.10] |
We also tested whether estimated network size remained a statistically significant predictor of college enrollment after controlling for the sets of covariates. In the full model that included the covariates and school encouragement, the relationship between estimated network size was close to but below the p < .05 convention (OR = 1.03, p = .060, 95% CI [0.99, 1.06]). Moreover, when the number of institutional agents was also added to the model, institutional agents remained a statistically significant predictor of college enrollment (OR = 1.39, p = .010, 95% CI [1.08, 1.78]), but network size was not a significant predictor (OR = 1.03, p = .066, 95% CI [0.99, 1.06]). School encouragement significantly predicted college enrollment in this model (OR = 1.63, p = .037, 95% CI [1.03, 2.57]).
As a supplemental analysis, we ran multinomial logistic regression models that compared the estimated odds of no college enrollment to those of 2-year college enrollment and 4-year college enrollment. In a model with only institutional agents and school encouragement, with no college enrollment as the reference group, institutional agents significantly predicted a greater estimated odds of enrolling in 2-year colleges (OR = 1.33, p = .007, 95% CI [1.08, 1.64]) and 4-year colleges (OR = 1.56, p = .008, 95% CI [1.12, 2.16]), and school encouragement was a marginally significant predictor of enrollment in 2-year colleges (OR = 1.43, p = .052, 95% CI [0.99, 2.05]) but did not significantly predict enrollment in 4-year colleges (OR = 1.48, p = .262, 95% CI [0.75, 2.91]). In the full model with all of the covariates, institutional agents predicted a significantly greater estimated odds of enrolling in 2-year colleges (OR = 1.39, p = .008, 95% CI [1.09, 1.78]) but not 4-year colleges (OR = 1.37, p = .224, 95% CI [0.82, 2.28]), and school encouragement was a significant predictor of enrollment in 2-year colleges (OR = 1.68, p = .027, 95% CI [1.06, 2.64]) and enrollment in 4-year colleges (OR = 2.53, p = .034, 95% CI [1.07, 5.99]). As mentioned previously, the results involving 4-year college enrollment should be interpreted with caution due to possibly insufficient statistical power.
Additional Predictors of College Enrollment
Several covariates significantly predicted college enrollment (see Table 4). Demographic characteristics, most personality traits, most foster-care characteristics, behavioral health problems, and employment status were not associated with college enrollment. Not surprisingly, being older at the time NSC data was accessed increased the estimated odds of enrolling in college. Reading ability, educational aspirations, and self-reported high school grades were positively associated with a greater likelihood of enrolling in college. Conversely, having ever repeated a grade, placement in special education, and having a child decreased participants’ likelihood of enrolling in college. There also were some findings in which the direction of the relationship was counterintuitive. Specifically, participants’ openness to new experiences was negatively associated with college enrollment, but residing in relative foster homes (vs. nonrelative foster homes) was positively associated with college enrollment. These findings arose from associations with other covariates and the outcome. For example, since the measure of institutional agents includes nonrelative foster parents, the relationship between nonrelative foster homes and other living settings is altered. Indeed, when the count of institutional agents is removed from the model, the difference in estimated odds of college enrollment between participants in relative and nonrelative foster homes is no longer significant (OR = 1.53, p = .145, 95% CI [0.86, 2.69]). Similarly, openness to new experiences is not associated with college enrollment in a model with no controls, but it becomes negatively associated with enrollment after other covariates are added (e.g., reading proficiency). Another unexpected finding is that physical abuse was positively associated with college enrollment. Since this was also the case in a model with no controls (OR = 1.14, p = .001, 95% CI [1.05, 1.23]), the association in the full model is not simply a result of interrelationships with other variables.
Discussion
This study examined the role of social capital in promoting college enrollment for youth in foster care. Our first two hypotheses (H1 and H2) contended that the number of institutional agents providing tangible support and/or guidance, and encouragement from school personnel to go to college, would both increase the likelihood that foster youth enroll in college. Our findings supported these hypotheses and were consistent even after accounting for a wide range of factors that could explain these associations. Moreover, the role that institutional agents and school encouragement played in the likelihood of enrolling in college was not trivial. Consider a group of youths who, based on their combination of covariates included in the model, had an expected likelihood of enrolling in college that was right at the sample’s average (a 55% probability of enrolling). If a second group of youths was identical to the first in all the covariates included in the regression model (including school encouragement), except that youths in the second group nominated one additional institutional agent, their expected probability of enrolling in college would be about 8 percentage points higher (63%). In a similar hypothetical case of two groups of youths who were identical in all measured ways (including number of institutional agents) but differed only in the amount of school encouragement they received, the expected difference in the probability of enrollment would be about 12 percentage points (67% for youths who received “a lot” of encouragement vs. 55% for youths who received “no” or “some” encouragement). To put these findings into perspective, a one-standard-deviation increase in reading ability is expected to increase the probability of enrolling in college by 12 percentage points.
Although the estimate was in the expected direction and approached statistical significance (p = .09), our third hypothesis (H3) about the interaction between institutional agents and school encouragement was not supported. Our findings also did not support the six competing hypotheses that stood in contrast to our hypotheses about institutional agents and school encouragement. Neither tangible support/advice from individuals without college experience nor emotional support from individuals with college experience increased youths’ likelihood of going to college. Likewise, general features of youths’ support networks, such as the size and perceived adequacy of support, did not predict college enrollment. Ruling out these competing explanations supports the contention that specific types of people providing specific types of support promotes postsecondary enrollment. Specifically, results indicate that the likelihood of going to college increases when youth have supportive relationships with adults who can leverage their positional power and mobilize college-relevant knowledge and resources. Finally, results from the last two competing hypotheses (A5 and A6) suggest that simply receiving encouragement to go to college, even if it is genuine and well-intended, does not necessarily promote college enrollment. Encouragement from biological family members or foster-care professionals did not predict college enrollment. In contrast, school personnel, who stand as representatives of the education institution and who have intimate knowledge of students’ capacities, exert particular influence on youths’ college future. In short, the source and amount of encouragement matter.
Implications
When thinking about potential implications for practice, it is important to remember that our findings about institutional agents are based on individuals who participants nominated. Participants were told that the number of individuals they could name for each type of support was limited, and thus the individuals who came to mind likely play important roles in the lives of the participants. Simply introducing institutional agents into the lives foster youth may not be sufficient to increase their chances of going to college. Some degree of connection with, trust in, and credibility of the individual likely must first develop—enough so that the youth would name this person as someone they rely on for support. Moreover, it is likely that some youths are, for a variety of reasons, better than others at eliciting the support of institutional agents. Efforts to improve educational outcomes for foster youth must find ways to meaningfully connect institutional agents to youths without such support.
Breaks in three distinguishable places can limit foster youths’ connections to institutional agents. First, foster youth may not have regular contact with institutional agents (accessibility). Second, even if foster youth are in regular contact with institutional agents, these agents may not be available to provide needed support (e.g., overworked high school counselors). Third, foster youth may not be ready or willing to utilize available agents for help with college due to factors such as constraints on their time, experiences of fractured trust and cycling relationships with professionals and caregivers, and policies that embody a tension between preparing for self-sufficiency and forming lasting connections with adults (Curry & Abrams, 2015; Samuels & Pryce, 2008). Each potential break warrants distinct programmatic responses, such as connecting youth to college-educated adults with more regularity (accessibility); increasing the amount of face time youth and institutional agents spend together, particularly on the tasks of college planning (availability); and reframing youths’ conceptions about help-seeking, cultivating enduring connections so that trust can be built, and looking beyond professionals as potential institutional agents (e.g., former foster youth; utilization).
The positive and statistically significant association between months in care past age 18 and the likelihood of enrolling in college (see Table 5, column 3) provides support for the recent change in federal policy providing states with funding to allow youth to remain in foster care until their 21st birthday. Prior research has shown a positive relationship between extended care and college enrollment and persistence, but the mechanisms underlying that relationship remain unclear (Courtney & Hook, 2017). One possible way that extended foster care could improve college attainment is by helping youth maintain contact with institutional agents associated with the foster-care system. Youths who were still in care at age 19 in the CalYOUTH study reported having a greater number of individuals to turn to for both tangible support (p < .01) and advice (p < .001) than did youths who had left care (Courtney et al., 2016). In addition, youths who were still in care at age 19 were more likely than those who had left care to identify caseworkers (i.e., institutional agents) as supportive adults; youths who had left care were more likely than those still in care to nominate grandparents. Further research is needed to assess whether the connections to institutional agents afforded by extended foster care contribute to the apparent influence of extended care on college enrollment and persistence.
In addition to social capital, our findings point to other factors that are important for stakeholders interested in promoting college enrollment for foster youth. Low academic performance (e.g., grades and reading skills), school expulsion, and need for special education services each decreased the likelihood of college enrollment. School mobility was not related to college enrollment, either in a bivariate model or the full model. Additional analyses (not reported) also found no associations between school mobility and reading scores, high school grades, grade repetition, or the highest grade that participants had completed. Although school mobility does not help educational progress, it may be that other academic risk factors play a more prominent role in college enrollment for foster youth. These risk factors (e.g., low performance, expulsion, and learning difficulties) likely reflect an accumulation of educational difficulties and needed intervention and will require interventions that start early and persist over time. The estimated odds of enrolling in college were nearly 70% lower for young parents than for participants without children. Additional preventive initiatives are needed to reduce the incidence of unwanted pregnancies and to encourage youth to delay parenthood until later in life, coupled with interventions that support current parents in completing their education. The positive association between physical abuse and college enrollment is a counterintuitive finding. There may be a substantive explanation for the finding (e.g., some physically abused children develop a toughness or tenacity that helps with goal attainment), or it may be an aberration in the sample.
Finally, our study focuses on college enrollment. Although enrollment is an important first step to attaining a college degree, it is just that—a first step. Policies and programs that support college completion are a necessary compliment to understanding factors that help foster youth enroll in college (Okpych, 2012; Okpych & Courtney, 2014).
Limitations
Several study limitations should be noted. A lack of information on types of college classes that participants are enrolled in precludes us from distinguishing students taking only remedial courses from those taking credit-bearing courses. This issue is particularly salient in 2-year colleges, where foster youth are disproportionately enrolled. Also, although self-report data from the CalYOUTH study was used to fill gaps from blocked records and undercoverage in the NSC data, a small amount of misclassification of participants’ college enrollment status may still be present.
There are also limitations in our social-capital measures. Institutional agents nominated by the participants may or may not have helped with college enrollment. Although exogenous factors may explain the association between institutional agents and college enrollment, our mediation analysis based on information collected downstream suggests that the amount of help youth received going to college partly explains the association. Second, the survey item used to create the measure of encouragement by school personnel to pursue education beyond high school was general in nature. This item may be measuring encouragement provided by one individual (e.g., a counselor) or multiple individuals (e.g., teachers, a counselor, etc.), or it may be capturing a broader school-level characteristic.
As reported earlier, we originally planned to model college enrollment as three categories. There are distinct differences in the application processes and qualifications needed to gain enrollment to 4-year versus 2-year colleges, and it is likely that some covariates play different roles predicting the likelihood of enrollment. Results from our analyses suggest that school encouragement is positively associated with enrollment in both 2-year and 4-year colleges, and the number of institutional agents predicts enrollment in 2-year colleges. It may be that foster youth who eventually enroll in 4-year colleges are particularly self-resourceful even in the absence of institutional agents, or that other factors in the model (e.g., indicators of academic performance) play a larger role than do institutional agents in gaining admission to 4-year colleges than to 2-year colleges.
A final limitation pertains to the generalizability of study findings. At the time of this writing, California is one of about 25 states that have passed legislation extending the foster-care age limit to 21, and the state has a particularly robust interest in promoting college success for foster youth. Although our sample represents the population of foster youth in California, the demographic makeup of the foster-care population, structure of child-welfare services, and concentration of colleges and college supports may also limit the generalizability of the findings.
Conclusion
Our study found that social capital plays an important role in helping foster youth enroll in college. Only about half of the participants nominated an institutional agent as someone they turn to for advice and help with practical problems. Programmatic efforts are needed to ensure that institutional agents are accessible and available and to encourage foster youth to connect with these individuals.
Notes
Nathanael J. Okpych, PhD, is an assistant professor at the University of Connecticut School of Social Work.
Mark E. Courtney, PhD, is a professor at the University of Chicago School of Social Service Administration.
Correspondence regarding this article should be directed to Nathanael Okpych, 38 Prospect St., Hartford, CT 06103, or via e-mail to [email protected]
References
AFCARS. (2016). The AFCARS report: Preliminary FY 2015 estimates. Washington, DC: U.S. Department of Health and Human Services. Retrieved from http://www.acf.hhs.gov/programs/cb/resource/afcars-report-23 Ahrens, K. R., DuBois, D. L., Richardson, L. P., Fan, M. Y., & Lozano, P. (2008). Youth in foster care with adult mentors during adolescence have improved adult outcomes. Pediatrics, 121(2), e246–e252. doi: 10.1542/peds.2007-0508 Allen, B., & Vacca, J. S. (2010). Frequent moving has a negative effect on the school achievement of foster children makes the case for reform. Children and Youth Services Review, 32(6), 829–832. doi: 10.1016/j.childyouth.2010.02.001 Barth, R. P., Wildfire, J., & Green, R. L. (2010). Placement into foster care and the interplay of urbanicity, child behavior problems, and poverty. American Journal of Orthopsychiatry, 76(3), 358–366. doi: 10.1037/0002-9432.76.3.358 Batsche, C., Hart, S., Ort, R., Armstrong, M., Strozier, A., & Hummer, V. (2014). Post-secondary transitions of youth emancipated from foster care. Child & Family Social Work, 19(2), 174–184. doi: 10.1111/j.1365-2206.2012.00891.x Brandford, C., & English, D. (2004). Foster youth transition to independence study. Seattle, WA: Office of Children’s Administration Research, Washington State Department of Social and Health Services. Retrieved from https://www.dshs.wa.gov/ca/publications/research Bryan, J., Moore-Thomas, C., Day-Vines, N. L., & Holcomb-McCoy, C. (2011). School counselors as social capital: The effects of high school college counseling on college application rates. Journal of Counseling & Development, 89(2), 190–199. doi: 10.1002/j.1556-6678.2011.tb00077.x Cicchetti, D. (2016). Socioemotional, personality, and biological development: Illustrations from a multilevel developmental psychopathology perspective on child maltreatment. Annual Review of Psychology, 67, 187–211. doi: 10.1146/annurev-psych-122414-033259 Choy, S. (2001). Students whose parents did not go to college: Postsecondary access, persistence, and attainment. Findings from the Condition of Education, 2001. Washington, DC: National Center for Education Statistics. Retrieved from http://eric.ed.gov/?id=ED460660 Coleman, J. S. (1988). Social capital in the creation of human capital. American Journal of Sociology, 94, S95-S120. doi: 10.1086/228943 Courtney, M. E., Charles, P., Okpych, N. J., Napolitano, L., & Halsted, K. (2014). Findings from the California Youth Transitions to Adulthood Study (CalYOUTH): Conditions of foster youth at age 17. Chicago, IL: Chapin Hall Center for Children at the University of Chicago. Retrieved from http://www.chapinhall.org/sites/default/files/CY_YT_RE1214.pdf Courtney, M. E., Dworsky, A., Brown, A., Carey, C., Love, C., & Vorhies, V. (2011). Midwest evaluation of adult functioning of former foster youth: Outcomes at age 26. Chicago, IL: Chapin Hall Center for Children at the University of Chicago. Retrieved from http://www.chapinhall.org/research/report/midwest-evaluation-adult-functioning-former-foster-youth Courtney, M. E., Dworsky, A. L., Ruth, G., Keller, T., Havlicek, J., & Bost, N. (2005). Midwest evaluation of the adult functioning of former foster youth: Outcomes at age 19. Chicago, IL: Chapin Hall Center for Children at the University of Chicago. Retrieved from http://www.chapinhall.org/sites/default/files/ChapinHallDocument_4.pdf Courtney, M. E., & Hook, J. L. (2017). The potential educational benefits of extending foster care to young adults: Findings from a natural experiment. Children and Youth Services Review, 72, 124–132. doi: 10.1016/j.childyouth.2016.09.030 Courtney, M. E., Okpych, N. J., Charles, P., Mikell, D., Stevenson, B., Park, K., … Feng. H. (2016). Findings from the California Youth Transitions to Adulthood Study (CalYOUTH): Conditions of youth at age 19. Chicago, IL: Chapin Hall Center for Children at the University of Chicago. Retrieved from http://www.chapinhall.org/sites/default/files/CY_YT_RE0516_1.pdf Courtney, M. E., Terao, S., & Bost, N. (2004). Midwest evaluation of the adult functioning of former foster youth: Conditions of youth preparing to leave state care. Chicago, IL: Chapin Hall Center for Children at the University of Chicago. Retrieved from https://www.chapinhall.org/sites/default/files/CS_97.pdf Curry, S. R., & Abrams, L. S. (2015). Housing and social support for youth aging out of foster care: State of the research literature and directions for future inquiry. Child and Adolescent Social Work Journal, 32(2), 143–153. doi: 10.1007/s10560-014-0346-4 Day, A., Dworsky, A., & Feng, W. (2013). An analysis of foster care placement history and post-secondary graduation rates. Research in Higher Education Journal, 19, 1–17. Retrieved from http://eric.ed.gov/?id=EJ1064665 Day, A., Dworsky, A., Fogarty, K., & Damashek, A. (2011). An examination of post-secondary retention and graduation among foster care youth enrolled in a four-year university. Children and Youth Services Review, 33, 2235–2341. doi: 10.1016/j.childyouth.2011.08.004 Dynarski, S. M., Hemelt, S. W., & Hyman, J. M. (2013). The missing manual: Using National Student Clearinghouse data to track postsecondary outcomes (No. w19552). Chicago, IL: National Bureau of Economic Research. Retrieved from http://www.nber.org/papers/w19552 Edalati, H., & Krank, M. D. (2015). Childhood maltreatment and development of substance use disorders: A review and a model of cognitive pathways. Trauma, Violence, & Abuse, 17, 454–467. doi: 10.1177/1524838015584370 Engle, J., & Tinto, V. (2008). Moving beyond access: College success for low-income, first-generation students. Washington, DC: Pell Institute for the Study of Opportunity in Higher Education. Retrieved from http://eric.ed.gov/?id=ED504448 Frerer, K., Sosenko, L. D., & Henke, R. R. (2013). At greater risk: California foster youth and the path from high school to college. San Franciso, CA: Stuart Foundation. Retrieved from http://stuartfoundation.org/greater-risk-california-foster-youth-path-high-school-college/ Finnie, R. (2012). Access to post-secondary education: The importance of culture. Children and Youth Services Review, 34(6), 1161–1170. doi: 10.1016/j.childyouth.2012.01.035 Gee, C. B., & Rhodes, J. E. (2007). A social support and social strain measure for minority adolescent mothers: A confirmatory factor analytic study. Child: Care, Health, and Development, 34(1): 87–97. doi: 10.1111/j.1365-2214.2007.00754.x Geenen, S., Powers, L. E., Phillips, L. A., Nelson, M., McKenna, J., Winges-Yanez, N., … Swank, P. (2015). Better futures: A randomized field test of a model for supporting young people in foster care with mental health challenges to participate in higher education. Journal of Behavioral Health Services & Research, 42(2), 150–171. doi: 10.1007/s11414-014-9451-6 Gillum, N. L., Lindsay, T., Murray, F. L., & Wells, P. (2016). A review of research on college educational outcomes of students who experienced foster care. Journal of Public Child Welfare, 10(3), 1–28. doi: 10.1080/15548732.2016.1172056 Gosling, S. D., Rentfrow, P. J., & Swann, W. B. (2003). A very brief measure of the Big-Five personality domains. Journal of Research in Personality, 37(6), 504–528. doi: 10.1016/S0092-6566(03)00046-1 Hines, A. M., Merdinger, J., & Wyatt, P. (2005). Former foster youth attending college: Resilience and the transition to young adulthood. American Journal of Orthopsychiatry, 75(3), 381–394. doi: 10.1037/0002-9432.75.3.381 Holland, N. E. (2010). Postsecondary education preparation of traditionally underrepresented college students: A social capital perspective. Journal of Diversity in Higher Education, 3(2): 111–125. doi: 10.1037/a0019249 Kim, H., & Kao, D. (2014). A meta-analysis of turnover intention predictors among U.S. child welfare workers. Children and Youth Services Review, 47(3), 214–223. doi: 10.1016/j.childyouth.2014.09.015 Klein, B., Damiani-Taraba, G., Koster, A., Campbell, J., & Scholz, C. (2015). Diagnosing attention-deficit hyperactivity disorder (ADHD) in children involved with child protection services: Are current diagnostic guidelines acceptable for vulnerable populations? Child: Care, Health and Development, 41(2), 178–185. doi: 10.1111/cch.12168 Lin, N. (2001). Social capital: A theory of social structure and action. Port Chester, NY: Cambridge University Press. McMillen, J. C., & Tucker, J. (1999). The status of older adolescents at exit from out-of-home care. Child Welfare, 78(3), 339–360. McNamara-Horvat, E., Weininger, E. B., & Lareau, A. (2003). From social ties to social capital: Class differences in the relations between schools and parent networks. American Educational Research Journal, 40(2), 319–351. doi: 10.3102/00028312040002319 Merdinger, J. M., Hines, A. M., Lemon-Osterling, K., & Wyatt, P. (2005). Pathways to college for former foster youth: Understanding factors that contribute to educational success. Child Welfare, 74, 867–896. National Student Clearinghouse (NSC). (2016a). Clearinghouse facts [Web page]. Retrieved from http://www.studentclearinghouse.org/about/clearinghouse_facts.php National Student Clearinghouse (NSC). (2016b). Enrollment coverage workbook [Data file]. Retrieved from https://nscresearchcenter.org/workingwithourdata/ Norman, R. E., Byambaa, M., De, R., Butchart, A., Scott, J., & Vos, T. (2012). The long-term health consequences of child physical abuse, emotional abuse, and neglect: A systematic review and meta-analysis. PLoS Medicine, 9(11), e1001349. doi: 10.1371/journal.pmed.1001349 Okpych, N. (2012). Policy framework supporting youth aging-out of foster care through college: Review and recommendations. Children and Youth Services Review, 34(7), 1390–1396. doi: 10.1016/j.childyouth.2012.02.013 Okpych, N. J., & Courtney, M. E. (2014). Does education pay for youth formerly in foster care? Comparison of employment outcomes with a national sample. Children and Youth Services Review, 43, 18–28. doi: 10.1016/j.childyouth.2014.04.013 Pecora, P. J. (2012). Maximizing educational achievement of youth in foster care and alumni: Factors associated with success. Children and Youth Services Review, 34(6), 1121–1129. doi: 10.1016/j.childyouth.2012.01.044 Peterson, R. A. (1994). A meta-analysis of Cronbach’s coefficient alpha. Journal of Consumer Research, 21(2), 381–391. doi: 10.1086/209405 Plank, S. B., & Jordan, W. J. (2001). Effects of information, guidance, and actions on postsecondary destinations: A study of talent loss. American Educational Research Journal, 38(4), 947–979. doi: 10.3102/00028312038004947 Portes, A. (1998). Social capital: Its origins and applications in modern sociology. In E. L. Lesser (Ed.), Knowledge and social capital: Foundations and applications (pp. 43–67). Boston, MA: Butterworth-Heinemann. Romano, E., Babchishin, L., Marquis, R., & Fréchette, S. (2015). Childhood maltreatment and educational outcomes. Trauma, Violence, & Abuse, 16(4), 418–437. doi: 10.1177/1524838014537908 Royston, P., & White, I. R. (2011). Multiple imputation by chained equations (MICE): Implementation in Stata. Journal of Statistical Software, 45(4), 1–20. doi: 10.18637/jss.v045.i04 Salazar, A. M., Jones, K. R., Emerson, J. C., & Mucha, L. (2016). Postsecondary strengths, challenges, and supports experienced by foster care alumni college graduates. Journal of College Student Development, 57(3), 263–279. doi: 10.1353/csd.2016.0029 Samuels, G. M. (2008). A reason, a season, or a lifetime: Relational permanence among young adults with foster care backgrounds. Chicago, IL: Chapin Hall Center for Children at the University of Chicago. Retrieved from http://www.chapinhall.org/research/report/reason-season-or-lifetime Samuels, G. M., & Pryce, J. M. (2008). What doesn’t kill you makes you stronger: Survivalist self-reliance as a resilience and risk among young adults aging out of foster care. Children and Youth Services Review, 30(10), 1198–1210. doi: 10.1016/j.childyouth.2008.03.005 Sandefur, G. D., Meier, A. M., & Campbell, M. E. (2006). Family resources, social capital, and college attendance. Social Science Research, 35, 525–553. doi: 10.1016/j.ssresearch.2004.11.003 Sheehan, D. V., Sheehan, K. H., Shytle, R. D., Janavs, J., Bannon, Y., Rogers, J. E., … Wilkinson, B. (2010). Reliability and validity of the Mini International Neuropsychiatric Interview for Children and Adolescents (MINI-KID). Journal of Clinical Psychiatry, 71(3), 313–326. doi: 10.4088/JCP.09m05305whi Stanton-Salazar, R. D. (1997). A social capital framework for understanding the socialization of racial minority children and youths. Harvard Educational Review, 67(1), 1–41. doi: 10.17763/haer.67.1.140676g74018u73k Stanton-Salazar, R. D. (2011). A social capital framework for the study of institutional agents and their role in the empowerment of low-status students and youth. Youth and Society, 43(3), 1066–1109. doi: 10.1177/0044118X10382877 Sullivan, M. J., Jones, L., & Mathiesen, S. (2010). School change, academic progress, and behavior problems in a sample of foster youth. Children and Youth Services Review, 32(2), 164–170. doi: 10.1016/j.childyouth.2009.08.009 Unrau, Y. A., Font, S. A., & Rawls, G. (2012). Readiness for college engagement among students who have aged out of foster care. Children and Youth Services Review, 34(1), 76–83. doi: 10.1016/j.childyouth.2011.09.002 Wilkinson, G. S., & Robertson, G. J. (2006). Wide Range Achievement Test (WRAT4) professional manual. Lutz, FL: Psychological Assessment Resources. Wolanin, T. R. (2005). Higher education opportunities for foster youth: A primer for policymakers. Washington, DC: The Institute for Higher Education Policy. Retrieved from http://www.ihep.org/research/publications/higher-education-opportunities-foster-youth Zlotnick, C., Tam, T. W., & Soman, L. A. (2012). Life course outcomes on mental and physical health: The impact of foster care on adulthood. American Journal of Public Health, 102(3), 534–540. doi: 10.2105/AJPH.2011.300285
