Developmental heterogeneity of perceived social support among former foster youth

Developmental heterogeneity of perceived social support among former foster youth

Children and Youth Services Review 76 (2017) 51–58 Contents lists available at ScienceDirect Children and Youth Services Review journal homepage: ww...

971KB Sizes 0 Downloads 77 Views

Children and Youth Services Review 76 (2017) 51–58

Contents lists available at ScienceDirect

Children and Youth Services Review journal homepage: www.elsevier.com/locate/childyouth

Developmental heterogeneity of perceived social support among former foster youth Andrew Zinn ⁎, Ashley N. Palmer, Eunji Nam University of Kansas, United States

a r t i c l e

i n f o

Article history: Received 19 October 2016 Received in revised form 24 February 2017 Accepted 25 February 2017 Available online 28 February 2017 Keywords: Social support Social connections Foster youth Emerging adulthood, transition to adulthood Longitudinal

a b s t r a c t Research suggests that social support influences a number of important young adult outcomes. Unlike other youth, however, foster youths' ties to their families, peers, and communities may be attenuated by the experiences of maltreatment, removal from home, and disrupted foster care placements. To address these challenges, public child welfare agencies have implemented a number of policy and practice reforms. In many ways, however, these efforts have not been well-informed by empirical research. A better understanding of how social support develops during this crucial stage of life could help to inform the formulation and targeting of interventions designed to facilitate social connections and support among foster youth. Based on data from a longitudinal panel survey of foster youth in three Midwestern states, the current study attempts to elucidate the nature of social support development among former foster by using a combination of descriptive and person-centered analyses. Results suggest several important features of social support development among this population. For example, differences in social support appear to be amplified and distilled during emerging adulthood. Together, the results suggest that efforts intended to assess and fortify social connections and support should start well before emancipation and continue well into young adulthood. © 2017 Elsevier Ltd. All rights reserved.

1. Introduction Research suggests that social support influences a number of important young-adult outcomes, including physical, emotional, and psychological well-being, college adjustment, and socioeconomic status and employment (Azmitia, Syed, & Radmacher, 2013; Guan & Fuligni, 2015; Holt-Lunstad, Smith, & Layton, 2010; Matthews, Stansfeld, & Power, 1999). Unlike other youth, however, foster youths' ties to their families, peers, and communities may be severely attenuated by the experiences of maltreatment, removal from home, and disrupted foster care placements (Jones, 2014; Perry, 2006). Indeed, several studies have found that victims of maltreatment, and children who have been place in substitute care, report lower levels of social support and fewer social connections in adulthood than other groups (Perry, 2006; Pitzer & Fingerman, 2010; Sperry & Widom, 2013). Not surprisingly, a growing body of work suggests that social support may mediate or moderate the impacts of maltreatment and substitute care on a variety of outcomes, including psychological and emotional functioning, selfesteem, and resilience (e.g., Horan & Widom, 2015; Jones, 2014; Salazar, Keller, & Courtney, 2011; Sperry & Widom, 2013).

⁎ Corresponding author. E-mail addresses: [email protected] (A. Zinn), [email protected] (A.N. Palmer), [email protected] (E. Nam).

http://dx.doi.org/10.1016/j.childyouth.2017.02.032 0190-7409/© 2017 Elsevier Ltd. All rights reserved.

More broadly, several child welfare scholars (Antle, Johnson, Barbee, & Sullivan, 2009; Propp, Ortega, & NewHeart, 2003), building on the earlier work of Stiver (1991), have argued that normative young adulthood is characterized not by autonomy and self-reliance, but by an effective, “blending of self-sufficiency and dependency” (Propp et al., 2003). Recasting independent living as interdependent living, these scholars point out that continued reliance on others throughout young adulthood is both typical and essential and, thus, child welfare services should endeavor to fortify the ability and willingness of foster youth to develop and maintain social connections after emancipating from care. Partly in response to these findings, public child welfare agencies have sought to implement policies and programs designed to help youth to (1) develop individual social-emotional competencies (e.g., emotional regulation, decision making) necessary to form and maintain relationships and (2) establish and strengthen social networks prior to emancipation. Examples range from counseling and relationship education to kinship care, co-placement with siblings, interventions to identify relatives and kin (i.e., family finding), and mentoring programs. In many ways, however, these efforts have not been well-informed by conceptual or empirical research (Courtney & Heuring, 2005; Stein, 2006b). In particular, we have only a rudimentary understanding of the developmental course of social support among foster youth during emerging adulthood (Collins, Spencer, & Ward, 2010). A better understanding of how social support develops during this crucial stage of

52

A. Zinn et al. / Children and Youth Services Review 76 (2017) 51–58

life could help to inform the formulation and targeting of interventions designed to facilitate social connections and social support, and serve as an important prerequisite to examining the relationships among social support and the young-adult outcomes of foster youth. The goal of the current study is to examine the nature of social support development among former foster youth over the course of emerging adulthood. As described in the Methods section, we use a combination of descriptive statistics and person-centered analyses to elucidate the important features of social support development among this population. In a forthcoming study (Zinn, Palmer, & Nam, under review), we build on the findings of the current study by examining the interrelationships among foster youths' individual characteristics, social connections with different types of individuals and formal institutions, and youths' perceived social support during emerging adulthood.

To date, only one study, Greeson, Garcia, Kim, and Courtney (2015), has examined the development of social support during emerging adulthood among former foster youth. Using longitudinal data from an evaluation of a life skills training program for foster youth in Los Angeles (Courtney et al., 2008), Greeson and colleagues examined the predictors of perceived social support among (former) foster youth age 17 to 21. Unlike the findings from studies using non-foster-care or general-population samples, Greeson and colleagues reported that the level of perceived social support among foster youth decreased modestly (0.16 s.d.) from age 17 to 21. However, consistent with the findings from these other studies, Greeson also found that the rates of change in social support vary significantly across sampled youth.

1.1. Nature and development of perceived social support

The goal of the current study is to examine heterogeneity in the developmental course of perceived social support among former foster youth during emerging adulthood. Based on the conceptual and empirical work described above, we have two general hypotheses. First, by virtue of having experienced maltreatment, removal from home, and (in many cases) disrupted foster care placements, foster youths' ties to their families, peers, and communities may be tenuous and strained (Jones, 2014; Perry, 2006). Also, the social ties that foster youth do have may contain a disproportionate number of professional, contextdependent relationships, such as substitute caregivers, social workers, and formal mentors (Collins et al., 2010). Finally, evidence suggests that foster youth report relatively high levels of social and emotional problems (Courtney, Terao, & Bost, 2004; McMillen et al., 2005). Thus, in many cases, foster youth are faced with the challenge of having to navigate the transition to adulthood, including the impending exit from substitute care, with only a tenuous support network and without the requisite social and emotional skills needed to develop and maintain supportive relationships. Consequently, we expect that, consistent with the findings reported by Greeson et al. (2015), the average level of social support will decline during emerging adulthood. Second, a defining feature of emerging adulthood is the profound population-level heterogeneity in employment and school enrollment, residential stability, friendship and intimate partner relationships, and participation in various social institutions (e.g., religious organizations) (Arnett, 2000), all of which serve as potentially important sources of supportive relationships. More specifically, recent work involving foster youth has documented similar types of heterogeneity (Courtney, Hook, & Lee, 2012; Keller, Cusick, & Courtney, 2007; Stein, 2006a). Consequently, we expect to observe substantive heterogeneity in the developmental course of social support among this sample. However, given the lack research to date concerning the nature of social support among foster youth, we make no specific hypothesis about the nature of this heterogeneity.

In general, studies have suggested that level of social support remains relatively stable within specific developmental periods (Levitt, Guacci-Franco, & Levitt, 1993; Sarason, Sarason, & Shearin, 1986). However, conceptual models of social support development, as well as emerging empirical evidence, suggest that the level of social support can change substantially during important developmental transitions, including the transition to adulthood (a.k.a., emerging adulthood). For example, Kahn and Antonucci's (1980) social convoy model argues that the constellation of people from whom an individual derives social support changes as a function of changes in the characteristics of an individual and their social environment. In other words, changes in the level of available support will reflect changes in (a) the availability of potentially supportive relationships within an individual's social ecologies, and (b) an individual's ability to identify, cultivate, and maintain supportive relationships. During emerging adulthood per se, transitions into work and higher education, exit from parents' homes, and development of intimate relationships, can result in profound changes in an individuals' social network (Wrzus, Hänel, Wagner, & Neyer, 2012). Also during this period, individuals continue to undergo significant social and emotional maturation (Arnett, 2006). In light of these changes, substantive change in social support might also be expected. In general, the findings of empirical studies of social support appear to support the hypothesis that youth experience significant changes in social support during emerging adulthood. For example, in a 4-year prospective study of high school seniors in Los Angeles, Guan and Fuligni (2015) found that the reported levels of emotional support received from parents, siblings, and peers either remain stable or increase modestly during the four years following adolescents' senior year in high school. Similarly, Pettit, Roberts, Lewinsohn, Seeley, and Yaroslavsky (2011), in a study of Oregon high school students, found that the average level of perceived social support from friends and family increases modestly between the ages of 21 and 30. Finally, in a meta-analysis of studies examining change in social networks over the life course, Wrzus et al. (2012) found that the size of individuals' global social networks increases during adolescence and young adulthood, plateauing around age 30. The findings of empirical studies also suggest substantial heterogeneity in the developmental course of social support during emerging adulthood. For example, in the studies by Guan and Fuligni (2015) and Pettit et al. (2011), the parameter estimates of growth models suggest that the rates of change in social support vary significantly across sampled youth. Also, in a recent longitudinal study of Australian women ages 18 to 40, Holden, Dobson, Ware, Hockey, and Lee (2015) identified four distinct developmental trajectories, including two that are characterized, respectively, by a long-term net increase or decrease in perceived social support. Broadly speaking, these findings are consistent with the conceptual understanding of emerging adulthood as a time of life that is characterized by a profound heterogeneity in experiences and life choices (Arnett, 2000).

1.2. Current study

2. Methods 2.1. Sample and data sources Data for this study came from the Midwest Evaluation of Former Foster Youth (Courtney et al., 2007). The sampling frame included adolescents in substitute care placements supervised by public child welfare agencies in three Midwestern states: Illinois, Wisconsin, and Iowa. The sample included youth who (a) were between the ages of 17 and 17.5 years, (b) had been in care for at least one year, and (c) were deemed able to complete an in-person or phone interview. Youth were excluded because of developmental disability, severe mental illness, current incarceration or psychiatric hospitalization, and inadequate English language skills. In Wisconsin and Iowa, the samples constituted a census of qualifying youth, whereas in Illinois a random sample was drawn from the population of qualifying youth. A total of 767 youth were deemed eligible for inclusion.

A. Zinn et al. / Children and Youth Services Review 76 (2017) 51–58

Data were collected during five waves of computer-assisted interviews. The baseline interview was administered between May 2002 and March 2003. Among the sample of eligible youth, 732 (Illinois: 474, Wisconsin: 195, Iowa: 63) were interviewed at baseline, yielding a participation rate of 95.4%. Between 80.7 and 82.4% of these youth were interviewed at each of the four successive interview waves, which were administered approximately 15 months apart. For a full description of the sampling procedures and survey response rates, see Courtney et al. (2004) and Courtney et al. (2011). 2.2. Measures Social support was measured using the Medical Outcome Study Social Support Survey (Sherbourne & Stewart, 1991), which is a measure of perceived social support. Unlike received social support, which is thought to be confounded by situational factors, including an individual's current level of need for social support, perceived social support is a reflection of an individual's internalized understanding of the degree to which other people are supportive and available (Pierce, Sarason, & Sarason, 1991). This internalized understanding is understood to be a product of an individual's interpersonal history, including their relationships with caregivers. Although more stable than received social support, perceived social support is thought to be continuously updated by an individual's ongoing interactions with other people and the changing characteristics of an individual's social environment (see Uchino (2009) for a comprehensive review of the differences between received and perceived social support). The Medical Outcome Study Social Support Survey scale is comprised of 19 questions about how often a person feels they have someone to provide support for various types of needs (e.g., How often do you feel that there is someone you can count on to listen to you when you need to talk) (Sherbourne & Stewart, 1991). Response options are specified as a 5-level Likert scale, ranging from none of the time [0] to all of the time [4]. These questions are included in all five interview waves. Prior work suggests this scale is a valid and reliable measure of perceived social support (Gjesfjeld, Greeno, & Kim, 2007). Although this scale is intended to measure four different types of perceived social support (i.e., emotional/informational, tangible, affectionate, and positive social interaction), exploratory factor analysis on these data strongly suggests that the scale items capture a univariate construct across all 5 waves of data. Thus, the scale items are averaged to obtain a total perceived social support score (0.95 ≤ α ≤ 0.97). 2.3. Analytic approach The goal of the current study was to examine the development of perceived social support over the course of emerging adulthood, and explore how the course of this development differs across former foster youth. To examine these questions, we used a combination of descriptive and person-centered methods. 2.3.1. Descriptive analyses In addition to standard measures of central tendency (e.g., median), we calculated three measures intended to summarize the change in social support over time. First, we calculated the standardized net change in perceived social support, which measures the net difference between social support at age 17 and 26. This measure was calculated by taking the difference between perceived social support scores reported on youths' first and last interviews, dividing this difference by the number of years between the administrations of the first and last interviews, and then multiplying by 9 (i.e., number of years between the target ages at waves 1 and 5, respectively). The standardized net change in perceived social support has a potential range between −4 (decrease from all the time [4] to none of the time [0]) and 4 (increase from none of the time [0] to all of the time [4]). The second measure we calculated was the absolute value of the standardized percentile rank change, which measures

53

the extent to which youth experience changes relative to other youth. This measure was calculated in a similar fashion as the standardized net change, except that within-wave percentile ranks were used instead of raw perceived social support scale scores. Also, because the dispersion of percentile rank change is, by definition, symmetric, we presented the absolute value, which has a potential range between 0 (indicating no relative change) to 100 (indicating a change from the lowest to the highest, or the highest to the lowest, relative percentile ranks). Finally, we calculated a measure intended to characterize the degree to which youth experienced reversals in the direction of change from one measurement occasion to the next. In brief, this measure, which we labeled maximum valence change, corresponds to the degree to which the slope of a successive change in social support deviates from the slope of the preceding change in social support. For example, if a youth first experienced an increase from a little of the time [1] to all the time [4] followed by a decrease to some of the time [3], then the value of the valence change would be equal 4 ((4 − 1) − (3 − 4)). Alternatively, if a youth experienced changes from a little of the time [1] to some of the time [2] to most of the time [3], the rates of change would be equal and, thus, the valence change would be 0 ((2 − 1) − (3 − 2)). The maximum valence change has a potential range between 0 (no changes in valence across time) to 8 (maximum increase followed by maximum decrease or maximum decrease followed by maximum increase). Note that the standardized net change in perceived social support and the absolute value of the standardized percentile rank change can only be calculated for youth with 2 or more measurement occasions (n = 712, 97.3%), and the maximum valence change can only be calculated for youth with 3 or more measurement occasions (n = 664, 90.7%).

2.3.2. Latent class growth models To further elucidate the nature of change in social support over time, we estimated a series of latent class growth models. Latent class growth models are a restricted form of growth mixture models in which the error terms for each trajectory are set equal to zero. In contrast to unrestricted growth mixture models, which capture a portion of the variation in a longitudinal variable using parametric variance parameters, latent class growth models characterize this variation using only the trajectory intercept and slope parameters. Thus, latent class growth models can be viewed as a non-parametric approach for characterizing the distribution of longitudinal variables (Vermunt, 2010). The latent class growth models were specified using youths' perceived social support scale scores as the dependent variable and youths' age at each interview wave as the indicator of time. Because prior studies have found that social support may vary across gender, race/ethnicity, and urbanicity (Airaksinen et al., 2015; Guan & Fuligni, 2015; House, 1987; Mickelson & Kubzansky, 2003; Pettit et al., 2011), variables for gender, race, Hispanic ethnicity, and state were also included in the models as predictors of social support. As a result, the estimated latent class growth model trajectories were conditional on mean-level differences in perceived social support scores across these variables. Finally, to address the modest sample attrition that occurred in waves 2 through 5, records for waves 2 through 5 were weighted to reflect the distribution of the predictor variables among the baseline interview respondents.

3. Findings 3.1. Sample characteristics The sample-level descriptive statistics are presented in Table 1. The mean age of youth at baseline was 17.9 years. A majority of the sample identified as black (57.0%), while just under one-third (30.9%) identified as white, and less than one in ten (8.6%) identified as Hispanic.

54

A. Zinn et al. / Children and Youth Services Review 76 (2017) 51–58

Table 1 Sample characteristics.

State IL WI IA Age at baseline Female Race Black White Other Hispanic

Freq. / Mean

(Percent / S.D.)

474 195 63 17.9 378

(64.8%) (26.6%) (8.6%) (0.37) (51.6%)

417 226 89 63

(57.0%) (30.9%) (12.2%) (8.6%) Fig. 2. Change statistics.

3.2. Perceived social support In Fig. 1 are plotted the mean, median, and 10th, 25th, 75th, and 90th percentiles of the perceived social support scale for the entire sample over all measurement occasions (right panel) and by youth age at each interview (left panel). In Fig. 2 are plotted the change statistics of perceived social support for the sample as a whole. Over all measurement occasions, the mean and median are 2.83 and 3.00, respectively, which roughly correspond to most of the time (3). The standard deviation is 0.96 and the 25th and 75th percentiles are, respectively, 2.16 and 3.68. Across youth age, there is a modest decrease in the mean of the perceived social support scale from 2.98 at age 17 to 2.75 at age 26, corresponding to an effect-size decrease of 0.24 (0.23 ÷ 0.96). The change in the median of perceived social support is also negative,

albeit somewhat smaller (0.14 s.d.). There also appears to be a small increase in the dispersion of the social support scale across youth age, with the interquartile range increasing from 1.32 (3.74 − 2.42) at age 17 to 1.63 (3.68 − 2.05) at age 26. Although the sample mean values of perceived social support do not change dramatically over time, substantial minorities of youth appear to experience a more significant change (Fig. 2). For example, a fifth of the sample experience a standardized net change in perceived social support of N 1.5 or less than −2.0. Also, a quarter of the sample experience percentile rank changes higher than 47%, and a tenth experience percentile rank changes N 76%. Finally, 1 in 10 youth report experiencing a maximum valence change in excess of 1.79. 3.3. Latent class growth models of perceived social support Model specification and selection proceeded in several steps. First, we estimated a set of models in which the trajectory slopes were specified using cubic functions. Using various statistical fit criteria, the best fitting model contained five trajectory classes. Second, based on this model, we tested various restrictions on the slope parameters. The final model selected consisted of five linear trajectories, three of which had slope parameters equal to zero (fit statistics for all evaluated models are available upon request). Based on the final model, none of the included covariates, save for gender (female: B = 0.034, s.e. = 0.012),

Fig. 1. Perceived social support by youth age. Note: Subsample size is parsed by interview wave and age at interview. For example, 731 youth were interviewed at wave 1, of which 387 and 344, respectively, were 17 and 18 years old at the time of the interview.

Fig. 3. Predicted social support by trajectory group. Note: Slope parameters not shown for trajectories for which parameters were constrained to equal 0.

A. Zinn et al. / Children and Youth Services Review 76 (2017) 51–58

are found to be statistically significantly related to perceived social support. Also, the estimated intercept and slope parameters and predicted posterior probabilities from the final model are almost identical to those obtained from a model in which no covariates were included. The predicted trajectories based on the final model are plotted in Fig. 3, and the average initial and final social support scores and change statistics for youth assigned to each trajectory group (based on the modelbased posterior probabilities) are presented in Fig. 4. The intercepts of the five predicted trajectory groups range from 2.22 to 3.94. All of the intercepts statistically significantly differ from one another, with pairwise differences ranging from 0.25 (0.26 s.d.) to 1.72 (1.85 s.d.). Three of the five trajectories ― high stable (4.0%), moderate-high stable (23.9%), and moderate stable (40.6%) ― have slopes equal to zero. One trajectory, high increasing (7.1%), has a statistically significant positive slope of 0.019, which translates into an increase in perceived social support of 0.171 (0.178 s.d.) between age 17 and 26.

55

The final trajectory, low decreasing (24.5%), has a statistically significant negative slope of − 0.077, which translates into a decrease of −0.693 (−0.722 s.d.) between age 17 and 26. 3.4. Change statistics by latent class growth model trajectory group 3.4.1. Initial and final perceived social support Consistent with the latent class growth model intercept estimates, the perceived social support initial (i.e., baseline) scores are statistically significantly lower for youth assigned to trajectories with lower average social support than among youth assigned to trajectories with higher average social support. This is true of all pairwise comparisons. In general, the variation in initial social support scores is greater for youth assigned to trajectories with lower baseline scores. For example, the interquartile range of social support baseline scores among youth assigned to the low decreasing trajectory (1.42) is substantially larger

Fig. 4. Change statistics by trajectory group. †Based on Wilcoxon rank-sum tests.

56

A. Zinn et al. / Children and Youth Services Review 76 (2017) 51–58

than the interquartile range among youth assigned to the high increasing (0.32) and high stable trajectories (0.11). The final perceived social support scores are statistically significantly different for all pairwise comparisons. In general, the dispersion of final social support scores among youth assigned to each trajectory group is narrower than the dispersion in scores at baseline. For example, the interquartile range of scores for youth assigned to the low decreasing, moderate stable, and high increasing trajectories are, respectively, 22, 13, and 17% lower at the final interview than they were at baseline. 3.4.2. Standardized net change Consistent with the latent class growth model slope estimates, the standardized net change in perceived social support among youth assigned to the low decreasing and high increasing trajectories are, respectively, statistically significantly lower and higher than youth assigned to other trajectories. Also, the variation in the rate of change in perceived social support appears to be larger (i.e., broader) for youth assigned to trajectories with lower intercepts. For example, the interquartile range of the standardized net change in social support among youth assigned to the high stable, moderate-high stable, and low decreasing trajectories are, respectively, 0.2, 1.5, and 2.3. Moreover, the distribution of the standardized net change in social support among youth assigned to trajectories with lower intercepts tends to be more negatively skewed than those of youth assigned to trajectories with higher intercepts. For example, the difference between the median and 25th percentiles among youth assigned to the high stable, moderate-high stable, and low decreasing groups are, respectively, 0.1, 0.7, and 1.3. 3.4.3. Absolute value of the standardized percentile rank change The magnitudes of rank order change in social support among youth assigned to the moderate stable and moderate-high stable trajectories are statistically significantly higher than among youth experiencing trajectories with lower or higher mean perceived social support. For example, the median standardized absolute percentile change in social support among youth experiencing these trajectories is 30%, whereas the median values for youth experiencing other trajectories range from 5 to 20%. 3.4.4. Maximum valence change The maximum social support valence change is statistically significantly higher for youth assigned to trajectories with lower average social support than among youth assigned to trajectories with higher average social support. This is true of all pairwise comparisons, save for that between youth assigned to the low decreasing and moderate stable trajectories. 4. Limitations Before discussing the findings described above, it is important to explicitly acknowledge several significant limitations of the current study. First, all of the data were collected from youth self-reports, which may be subject to recall errors and intentional misrepresentations. Second, because these data were collected from a single source (i.e., youth) using a common method (i.e., interview), there is a possibility that the relationships described above are reflections of spurious associations due to common method or single source bias. 5. Discussion 5.1. Nature of social support development during emerging adulthood A majority of sampled foster youth report relatively high levels of social support (‘most of the time’ or ‘all of the time,’) that gradually, but modestly, decrease between the ages of 17 and 26. The finding of a modest decrease over time is consistent with the findings of Greeson et al.

(2015), which suggests that, unlike youth in general, former foster youth experience an average decrease in social support during emerging adulthood. Despite the fact that the level of reported social support appears to be relatively high and stable among sampled youth, the findings from latent class growth model and confirmatory analyses also suggest substantive heterogeneity in youths' social support development, including differences in the level of social support reported at baseline, magnitude and valence of net change between ages 17 and 26, and level of instability (i.e., valence change) between measurement occasions. Collectively, these results suggest several important features of social support development among this population. First, the findings from the latent class growth model suggest that social support developmental trajectories are distinguished from one another, in large part, by the level of social support reported at baseline. Although there is some variation in the level of support over time among youth assigned to each trajectory class, this variation appears to exist within proscribed bandwidths. Thus, social support development appears to exhibit, to some degree, statedependence: foster youth who report relatively higher levels of social support at baseline are much more likely to also report higher levels of social support in their mid-twenties than are other youth and visa-versa. There are at least two possible explanations for this state-dependence. First, it may reflect the relative stability of the primary drivers of social support. For example, an individual's ability to identify, cultivate, and maintain supportive relationships may be products of personality traits (e.g., temperament) that are relatively stable over time (Roberts, Walton, & Viechtbauer, 2006). Similarly, the state dependence in social support may reflect an underlying stability in the availability (or lack thereof) of potentially supportive relationships within an individual's social ecologies. Another possible explanation for the state dependence observed for social support may be related to the nature of perceived social support. As described in the Methods section, perceived social support is conceptualized as an internalized understanding of the degree to which other people are supportive and trustworthy (Uchino, 2009). These internalized beliefs can endure over time and be inelastic to changes in the level of received social support. Thus, the relative stability of social support across time (i.e., state-dependence) may be a reflection of the relative inelasticity of perception vis-à-vis changes in received support or social integration. A second potentially important feature of social support development among the foster youth included in this study is that differences in social support appear to be amplified and distilled during emerging adulthood. The amplification of differences is suggested by (a) an increase in the sample-level dispersion of social support scores over time and (b) the negative and positive slopes, respectively, of the lowest and second highest latent class growth model classes. A distillation of differences is suggested by a narrowing of within-trajectory variation (i.e., regression to the mean) of social support scores. As a result of these changes, differences in social support that exist at baseline appear to become increasingly pronounced over time. One potential explanation for this finding is that the consequences of exiting substitute care, and the challenges of emerging adulthood, have a differential effect on youth, depending on their baseline level of social support. In brief, youth who report lower levels of social support at baseline may possess poorer social–emotional skills than other youth. Similarly, the social networks of these youth may be sparser, and contain proportionately greater numbers of professional, context-dependent relationships than the networks of other youth (Collins et al., 2010). As a result, in addition to being less well-equipped to develop new sources of social support than other youth, foster youth with lower levels of baseline social support would also be more likely to experience a proportionately greater loss of social support during the transition to adulthood.

A. Zinn et al. / Children and Youth Services Review 76 (2017) 51–58

Another possible explanation for the distillation and amplification of differences in social support during emerging adulthood is that youths' perceptions of available social support change because of changes in context. For example, the relatively protected environment of substitute care, combined with youths' inexperience, may serve to distort youths' perceptions of the availability of social support (Geenen & Powers, 2007). However, upon emancipation from care, the demands of independent living may yield a more realistic (and differentiated) assessment of the adequacy of that support. The last potentially important feature of social support development among the foster youth is that the magnitude of net changes and reversals of direction (i.e., valence) in social support appear to be greater for youth with lower baseline levels of social support. In other words, youth who begin emerging adulthood with lower levels of perceived social support tend to experience larger changes, and greater instability, in social support than youth who begin with relatively higher baseline levels of social support. The negative correlation between baseline social support and subsequent instability may be a reflection of sparser, less robust social networks among youth with lower levels of social support. For example, for a youth with a smaller social network, the loss or gain of a single source of support may constitute a much larger relative change than it would for a youth with a larger social network. Also, as described above, the social networks of foster youth with lower levels of baseline support may contain proportionately greater numbers of professional, context-dependent relationships than the networks of youth with higher levels of baseline social support. If these types of relationships are more fleeting, and less emotionally close, than other relationships (e.g., family, friends), then we would expect larger changes in perceived social support than those experienced by other youth. 6. Implications The findings suggesting that initial differences in social support are both distilled and differentiated over time serves as a reminder that the developmental course of social support among foster youth during emerging adulthood is a continuation of developmental processes that have spanned a lifetime. These processes influence the development of youths' social–emotional competencies and supportive social networks, which, in turn, serve as the building blocks of social connections and positive social support in young adulthood. Conversely, however, the findings that many youth experience significant net change and reversals of valence underscore that social support development is far from a deterministic process. Together, these findings suggest that the current set of practices intended to fortify social connections and support among foster youth may be inadequate or ill-informed for a significant percentage of youth emancipating from care. However, these findings also suggest that, with a different set of practices and policies, there may be considerable scope for improvement. Collectively, the findings also suggest two specific changes in practice and policy. First, efforts to assess and support youths' social support development should start as early as possible and continue well-after youth emancipate from care. As described above, social support development during emerging adulthood builds upon youths' preexisting social–emotional competencies and social networks. Thus, efforts to assess and support youth need to begin as soon as youth enter care. However, based on the findings described here, it also appears that foster youth experience substantial change in social support during emerging adulthood. Thus, we would argue that efforts to assess and support youth need to be maintained well after emancipation from care. Second, the findings described here suggest that there is substantive heterogeneity in social support development, with a significant minority of youth experiencing lower baseline levels of social support that are subsequently followed by a continuing deterioration in support. This differentiation in experiences suggests that identifying and targeting specific sets of youth for services and supports is both feasible and

57

warranted. This ability to target could be especially useful, given the apparent need for ongoing support well into emerging adulthood. Finally, the findings of the current study lead to several questions about the nature and correlates of social support among foster youth. For example, do the trajectories observed here represent a continuation of experiences with origins that predated the age of 17, or do they represent substantively different experiences that began (or significantly changed) at the age of 17? Also, do the differences in social support observed here constitute substantive differences vis-à-vis young adult outcomes? If so, are these differences similarly important throughout the range of social support scores, or are there thresholds that demark important differences? Lastly, what are the important determinants of social support among foster youth? Specifically, which combinations of individual- and context-level characteristics are associated with higher levels of sustained social support? For example, in light of the prevalence of policies intended to facilitate contact with family members, and encourage the identification of natural mentors, what are the associations among these types of relationships and youths' perceptions of social support? Addressing these questions will be critical for designing effective strategies for helping foster youth develop and maintain social connections and support after leaving foster care.

References Airaksinen, J., Hakulinen, C., Elovainio, M., Lehtimäki, T., Raitakari, O. T., KeltikangasJärvinen, L., & Jokela, M. (2015). Neighborhood effects in depressive symptoms, social support, and mistrust: Longitudinal analysis with repeated measurements. Social Science & Medicine, 136, 10–16. Antle, B. F., Johnson, L., Barbee, A., & Sullivan, D. (2009). Fostering interdependent versus independent living in youth aging out of care through healthy relationships. Families in Society, 90(3), 309–315. Arnett, J. J. (2000). Emerging adulthood: Theory of development from the late teens through the twenties. The American Psychologist, 55(5), 469–480. Arnett, J. J. (2006). The psychology of emerging adulthood: What is known, and what remains to be known. In J. J. Arnett, & J. L. Tanner (Eds.), Emerging adults in America: Coming of age in the 21st century. Washington, DC: American Psychological Association. Azmitia, M., Syed, M., & Radmacher, K. (2013). Finding your niche: Identity and emotional support in emerging adults' adjustment to the transition to college. Journal of Research on Adolescence, 23(4), 744–761. Collins, M. E., Spencer, R., & Ward, R. (2010). Supporting youth in the transition from foster care: Formal and informal connections. Child Welfare, 89(1), 125–143. Courtney, M. E., & Heuring, D. H. (2005). The transition to adulthood for youth “aging out” of the foster care system. In D. W. Osgood (Ed.), On your own without a net: The transition to adulthood for vulnerable populations. Chicago: University of Chicago Press. 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: Chapin Hall Center for Children at the University of Chicago. Courtney, M. E., Dworsky, A., Cusick, G. R., Havlicek, J., Perez, A., & Keller, T. E. (2007). Midwest evaluation of the adult functioning of former Foster youth: Outcomes at age 21. Chicago: Chapin Hall. Courtney, M. E., Zinn, A., Zielewski, E. H., Bess, R. J., Malm, K. E., Stagner, M., & Pergamit, M. (2008). Evaluation of the life skills training program: Los Angeles county. Washington, D. C.: Administration for Children & Families. Courtney, M. E., Dworsky, A., Brown, A., Cary, C., Love, K., & Vorhies, V. (2011). Midwest evaluation of the adult functioning of former foster youth: Outcomes at age 26. Chicago: Chapin Hall at the University of Chicago. Courtney, M. E., Hook, J. L., & Lee, J. S. (2012). Distinct subgroups of former foster youth during young adulthood: Implications for policy and practice. Child Care in Practice, 18(4), 409–418. Geenen, S., & Powers, L. E. (2007). “Tomorrow is another problem”: The experiences of youth in foster care during their transition into adulthood. Children and Youth Services Review, 29(8), 1085–1101. http://dx.doi.org/10.1016/j.childyouth.2007.04. 008. Gjesfjeld, C. D., Greeno, C. G., & Kim, K. H. (2007). A confirmatory factor analysis of an abbreviated social support instrument—the MOSS-SSS. Research on Social Work Practice. http://dx.doi.org/10.1177/1049731507309830. Greeson, J. K. P., Garcia, A. R., Kim, M., & Courtney, M. E. (2015). Development & maintenance of social support among aged out foster youth who received independent living services: Results from the multi-site evaluation of foster youth programs. Children and Youth Services Review, 53(1–9). Guan, S. S. A., & Fuligni, A. J. (2015). Changes in parent, sibling, and peer support during the transition to young adulthood. Journal of Research on Adolescence, 26(2), 286–299. Holden, L., Dobson, A. J., Ware, R. S., Hockey, R., & Lee, C. (2015). Longitudinal trajectory patterns of social support: Correlates and associated mental health in an Australian national cohort of young women. Quality of Life Research, 24(9), 2075–2086. Holt-Lunstad, J., Smith, T. B., & Layton, J. B. (2010). Social relationships and mortality risk: A meta-analytic review. PLoS Medicine, 7(7), e1000316.

58

A. Zinn et al. / Children and Youth Services Review 76 (2017) 51–58

Horan, J. M., & Widom, C. S. (2015). From childhood maltreatment to allostatic load in adulthood the role of social support. Child Maltreatment. http://dx.doi.org/10.1177/ 1077559515597063. House, J. S. (1987). Social support and social structure. Sociological Forum, 2(1), 135–146. Jones, L. P. (2014). The role of social support in the transition from foster care to emerging adulthood. Journal of Family Social Work, 17(1), 81–96. Kahn, R. L., & Antonucci, T. C. (1980). Convoys over the life course: Attachment, roles, and social support. Life-span development and behavior. 3. (pp. 253–286). Keller, T. E., Cusick, G. R., & Courtney, M. E. (2007). Approaching the transition to adulthood: Distinctive profiles of adolescents aging out of the child welfare system. The Social Service Review, 81(3), 453–484. Levitt, M. J., Guacci-Franco, N., & Levitt, J. L. (1993). Convoys of social support in childhood and early adolescence: Structure and function. Developmental Psychology, 29(5), 811–818. Matthews, S., Stansfeld, S., & Power, C. (1999). Social support at age 33: The influence of gender, employment status and social class. Social Science & Medicine, 49(1), 133–142. McMillen, J. C., Zima, B. T., Scott, L. D., Auslander, W. F., Munson, M. R., Ollie, M. T., & Spitznagel, E. L. (2005). Prevalence of psychiatric disorders among older youths in the foster care system. Journal of the American Academy of Child and Adolescent Psychiatry, 44(1), 88–95. Mickelson, K. D., & Kubzansky, L. D. (2003). Social distribution of social support: The mediating role of life events. American Journal of Community Psychology, 32(3/4), 265–281. Perry, B. L. (2006). Understanding social network disruption: The case of youth in foster care. Social Problems, 53(3), 371–391. Pettit, J. W., Roberts, R. E., Lewinsohn, P. M., Seeley, J. R., & Yaroslavsky, I. (2011). Developmental relations between perceived social support and depressive symptoms through emerging adulthood: Blood is thicker than water. Journal of Family Psychology, 25(1), 127. Pierce, G. R., Sarason, I. G., & Sarason, B. R. (1991). General and relationship-based perceptions of social support: Are two constructs better than one? Journal of Personality and Social Psychology, 61(6), 1028. Pitzer, L. M., & Fingerman, K. L. (2010). Psychosocial resources and associations between childhood physical abuse and adult well-being. The Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 65(4), 425–433.

Propp, J., Ortega, D., & NewHeart, F. (2003). Independence or interdependence: Rethinking the transition from ward of the court to adulthood. Families in Society, 84(2), 259–266. Roberts, B. W., Walton, K. E., & Viechtbauer, W. (2006). Patterns of mean-level change in personality traits across the life course: A meta-analysis of longitudinal studies. Psychological Bulletin, 132(1–25). Salazar, A. M., Keller, T. E., & Courtney, M. E. (2011). Understanding social support's role in the relationship between maltreatment and depression in youth with foster care experience. Child Maltreatment, 16(2), 102–113. Sarason, I. G., Sarason, B. R., & Shearin, E. N. (1986). Social support as an individual difference variable: Its stability, origins, and relational aspects. Journal of Personality and Social Psychology, 50(4), 845. Sherbourne, C. D., & Stewart, A. L. (1991). The MOS social support survey. Social Science & Medicine, 35(6), 705–714. Sperry, D. M., & Widom, C. S. (2013). Child abuse and neglect, social support, and psychopathology in adulthood: A prospective investigation. Child Abuse & Neglect, 37(6), 415–425. Stein, M. (2006a). Research review: Young people leaving care. Child & Family Social Work, 11(3), 273–279. Stein, M. (2006b). Young people aging out of care: The poverty of theory. Children and Youth Services Review, 28(4), 422–434. Stiver, I. P. (1991). The meanings of “dependency” in female–male relationships. In J. V. Jordan, A. G. Kaplan, J. B. Miller, I. P. Stiver, & J. L. Surrey (Eds.), Women's growth in connection: Writings from the stone center. New York: Guilford Press. Uchino, B. N. (2009). Understanding the links between social support and physical health: A life-span perspective with emphasis on the separability of perceived and received support. Perspectives on Psychological Science, 4(3), 236–255. Vermunt, J. K. (2010). Longitudinal research using mixture models. In K. van Montfort, J. Oud, & A. Storra (Eds.), Longitudinal research with latent variables (pp. 119–152). New York: Springer-Verlag Berlin Heidelberg. Wrzus, C., Hänel, M., Wagner, J., & Neyer, F. J. (2012). Social network changes and life events across the life span: A meta-analysis. Psychological Bulletin, 139(1), 53–80. Zinn, A., Palmer, A.N., & Nam, E. (under review). The Predictors of Perceived Social Support Among Former FosterYouth