Correspondence concerning this article should be addressed to Alan Meca, Ph.D., Department of Public Health Sciences, University of Miami Miller School of Medicine, 1120 N.W. 14 th Street, Suite 1013, Miami, FL 33172 (ude.imaim.dem@acem.a)
The publisher's final edited version of this article is available at Cultur Divers Ethnic Minor Psychol
This study examined directionality between personal (i.e., coherence and confusion) and cultural identity (i.e., ethnic and US) as well as their additive effects on psychosocial functioning in a sample of recently immigrated Hispanic adolescents.
The sample consisted of 302 recent (Mage = 14.51 years at baseline; SD = .88 years) from Miami and Los Angeles who participated in a longitudinal study.
Results indicated a bidirectional relationship between personal identity coherence and both ethnic and US identity. Ethnic and US Affirmation/Commitment (A/C) positively and indirectly predicted optimism and negatively predicted rule breaking and aggression through coherence. However, confusion predicted lower self-esteem and optimism and higher depressive symptoms, rule breaking, unprotected sex, and cigarette use. Results further indicated significant site differences. In Los Angeles (but not Miami), ethnic A/C also negatively predicted confusion.
Given the direct effects of coherence and confusion on nearly every outcome, it may be beneficial for interventions to target personal identity. However, in contexts such as Los Angeles, which has at least some ambivalence towards recently immigrated Hispanic adolescents, it may be more beneficial for interventions to also target cultural identity to reduce confusion and thus promote positive development.
Keywords: Personal identity, ethnic identity, United States identity, adolescents, Hispanic immigrants
Identity development is a key developmental task of adolescence and emerging adulthood (Kroger, 2007). Developing a coherent sense of self helps to support a consistent set of choices in career aspirations (Côté, 2002) and romantic relationships (Beyers & Seiffge-Krenke, 2010). Moreover, individuals with an internally consistent sense of identity report higher self-esteem, lower internalizing symptoms, and are less likely to engage in risk-taking behaviors compared to individuals with less coherent identities (Crocetti et al., 2014). For adolescents from immigrant and ethnic/racial minority backgrounds, identity development can be more complex compared to United States (US) born, ethnic majority youth (Azmitia, Syed, & Radmacher, 2008; Syed & Mitchell, 2013). In addition to developing a general sense of personal identity, which is a normative developmental task for all adolescents, immigrant and ethnic/racial minority adolescents are also tasked with navigating multiple cultural reference points. As a result, these adolescents are confronted with the task of developing a cultural identity – an understanding of what their ethnic/racial group and membership in the larger nation mean to them.
Although personal and cultural identity are among the most commonly studied identity components, the interplay and directionality between personal and cultural identity remains understudied (Schwartz et al., 2013; Syed & Juang, 2014). To address this gap in the literature, in the current study we sought to examine directionality between personal and cultural identity, as well as their effects on psychosocial functioning. Additionally, we examined whether these processes differ across two contexts of receptions (Miami and Los Angeles) in a sample of recently immigrated Hispanic adolescents. We focus on Hispanic immigrant adolescents for four reasons. First, Hispanics are among the largest and fastest growing minority groups in the US (Ennis, Ríos-Vargas, & Albert, 2011). Second, as immigrants and ethnic minorities, recently immigrated Hispanic adolescents face the additional task of developing a cultural identity, on top of the normative task of developing a personal identity (Schwartz, Montgomery, & Briones, 2006). Third, recently immigrated Hispanic adolescents are likely to experience cultural stressors which may complicate cultural identity development (Rumbaut, 2008) and the development of a coherent personal identity. Finally, our focus on Hispanic adolescents is driven by a growing body of research indicating significant health disparities. Indeed, research indicates that Hispanic adolescents are more likely than other ethnic groups to use drugs and alcohol (Johnston, O’Malley, Bachman, & Schulenberg, 2014), smoke cigarettes (Kann et al., 2014), and report depressive symptoms (McLaughlin, Hilt, & Nolen-Hoeksema, 2007). As a result, more developmentally informed prevention research is needed that focuses on identifying mechanisms underlying Hispanic adolescents’ psychosocial functioning (Umaña-Taylor, 2011).
Theoretical Models of Personal and Cultural Identity Development
Personal and cultural identity have inspired separate literatures (Syed & Juang, 2014). Personal identity focuses on the set of goals, values, and beliefs that an individual has developed and/or internalized (Waterman, 1999). Personal identity therefore represents the answer to the question “Who am I?” Cultural identity, on the other hand, refers to how individuals define themselves in relation to the cultural groups to which they belong (Schwartz et al., 2006), and therefore more closely represents the answer to the question “who am I as a member of my group?” (Schwartz, Zamboanga, & Weisskirch, 2008). Although personal identity is salient for most young people, who must decide who they want to be and what to do with their lives (Côté & Levine, 2002), cultural identity is most salient for immigrants and ethnic minorities youth (Phinney & Ong, 2007). To provide a more complete conceptualization of personal and cultural identity, we provide a brief review of both these identity domains and the underlying theoretical perspectives.
Much of the literature on personal identity traces its roots to Erikson’s (1950) model of psychosocial development. According to Erikson (1950), identity develops in a dynamic manner involving coherence and confusion. Coherence refers to a sense of who one is and where one is going in life, whereas confusion signifies an inability to enact and maintain lasting commitments and a lack of a clear sense of purpose and direction. Although these two dimensions are negatively related, they are not complete opposites. People can show coherence in some domains and confusion in others (Schwartz, Zamboanga, Wang, & Olthuis, 2009b). We focus on coherence and confusion for two primary reasons. First, coherence and confusion refer to the overall sense of where a person is with respect to her/his personal identity development (Schwartz, Luyckx, & Crocetti, 2014b). Second, coherence and confusion uniquely and significantly predict positive and negative psychosocial functioning (e.g., Syed et al., 2013). Therefore, we rely on personal identity coherence and confusion (hereby referred to as coherence and confusion) as indices of personal identity in the present study.
Although the operationalization of cultural identity has varied in the literature, the current study focused on ethnic and US identity, two components particularly salient for recent immigrants in the US (Phinney & Ong, 2007). Broadly, ethnic and US identity can each be represented as multidimensional psychological constructs that reflect individuals’ beliefs and attitudes about their ethnic and national group membership as well as the process by which these beliefs and attitudes develop over time (Umaña-Taylor et al., 2014). In this study, we focus on the process by which individuals consider the subjective meaning of their ethnic group or membership in the larger US society, and develop emotional attachments to both of these cultural groups (Umaña-Taylor et al., 2014). Process models (Phinney, 1990; Schwartz et al., 2012) have drawn largely on Erikson’s (1950) work and on social identity theory (Tajfel & Turner, 1986), positing exploration (i.e., considering what it means to belong to a particular cultural group), commitment (i.e., making a decision regarding the subjective meaning of one’s cultural group membership), and affirmation (i.e., positive/negative feelings about one’s cultural group membership) as underlying processes.
Most cultural identity research has focused on ethnic identity. Among Hispanics, studies have generally found ethnic identity affirmation and commitment to be positively and negatively associated, respectively, with adaptive and maladaptive psychosocial functioning. However, as reviewed by Rivas-Drake et al. (2014) and Umaña-Taylor (2011), these relationships have not emerged consistently across studies. One potential explanation for this inconsistency is the reliance on ethnic identity as the sole indicator of cultural identity. Because the development of an integrated sense of self and identity incorporates elements from one’s ethnic group and the US (Berry, 1997), it is essential for researchers to evaluate the joint effects of ethnic and US identifications on psychosocial functioning—particularly among recently immigrated adolescents.
In this study, to facilitate appropriate comparisons between personal and cultural identity, drawing on social identity theory (Tajfel & Turner, 1986), we focused on the ethnic and US affirmation and commitment. Specifically, whereas coherence is strongly related to self-concept clarity (Schwartz et al., 2011), commitment and affirmation are conceptually similar to “cultural identity clarity,” or a sense of belonging (commitment) and solidarity (affirmation) with one’s cultural group(s) (Usborne & Taylor, 2010). Thus, cultural commitment and affirmation represent key components of a consolidated sense of ethnic and US identity (Schwartz et al., 2013). Although commitment and affirmation represent distinct yet related cultural identity processes (Umana-Taylor, Yazedjian, & Bamaca-Gomez, 2004), within the current sample, these subscales were highly correlated (r > .90). As a result, in the current study, we employed ethnic affirmation/commitment (A/C) and US A/C as indicators of a consolidated sense of ethnic and US identity, respectively.
Although empirical evidence suggests that personal and cultural identity are positively associated with well-being and inversely associated with internalizing and externalizing symptoms (Crocetti et al., 2014; Rivas-Drake et al., 2014), the ways in which these variables work together are not well understood (Schwartz et al., 2013). It is possible that the additional task of developing a cultural identity might make the development of a personal identity more difficult. However, given the theoretical links between personal and social identity, it is likely that personal and cultural identity (where cultural identity may be considered as a dimension of social identity) may be reciprocally connected (Schwartz et al., 2008). Indeed, some aspects of personal identity are assigned, informed, constrained, or internalized based on group memberships (Bosma & Kunnen, 2001; Hitlin, 2003). Moreover, personal goals, values, and beliefs can be influenced by cultural orientation (Triandis, 1995). For example, highly collectivistic values may lead to internalization of goals from close social networks (e.g., family, close friends, etc.), whereas individualistic values may lead to independent establishment of personal goals. In the other direction, a coherent personal identity provides individuals with a sense of structure within which to understand self-relevant information (Adams & Marshall, 1996) and thus shapes how individuals negotiate with the social environment and the opportunities they choose to pursue (Côté, 2002).
Although several studies have found significant and positive associations between ethnic and personal identity among adolescents (e.g., Branch, Tayal, & Triplett, 2000; Miville, Darlington, Whitlock, & Mulligan, 2005), most have been cross-sectional. Moreover, only a handful of these studies have examined the relationship of personal and cultural identity with psychosocial functioning (e.g., Schwartz, Zamboanga, Weisskirch, & Wang, 2010; Usborne & Taylor, 2010). Both Schwartz et al. (2010) and Usborne and Taylor (2010) found support for the idea that that cultural identity is indirectly associated with psychosocial functioning through personal identity. However, the cross-sectional nature of these and earlier studies has made it difficult to establish directionality (Maxwell & Cole, 2007), which is critical for determining strategic points of intervention. For example, if personal identity predicts psychosocial functioning by bolstering cultural identity, then interventions might focus on promoting cultural identity to prevent negative outcomes. On the other hand, if cultural identity predicts psychosocial functioning by bolstering personal identity, then interventions might focus on promoting personal identity instead. However, if the relationship between personal and cultural identity is bidirectional, then interventions should focus on both personal and cultural identity as a means to prevent negative outcomes.
Identity Processes and the Receiving ContextBecause identity emerges within the opportunities and constraints provided by sociocultural factors (Côté & Levine, 2002; Umana-Taylor et al., 2014), it is important to determine whether identity processes work similarly across cultural contexts. In the current study, we explored the directional relationship between personal and cultural identity across two sites: Miami and Los Angeles. Miami is a thriving metropolis as a result of the influx of Cuban migration (Portes & Stepick, 1994), aided by a law that allows them to become legal US residents immediately upon arrival (Stepick & Stepick, 2002). Miami is also a highly bicultural city where Hispanics are the majority of the population (65%; US Census Bureau, 2011) and hold many positions of political and economic power (Stepick, Grenier, Castro, & Dunn, 2003). In contrast, although Los Angeles has been home to a large Mexican population for several decades, the majority of city officials and leaders have been White (Hayes-Bautista, 2004; Light, 2006). Indeed, studies using the current dataset indicate that recently immigrated Hispanics in Los Angeles encounter higher perceived discrimination and negative context of reception than their Miami counterparts (Schwartz et al., 2015).
The Present Study
In the present study, we used data from a longitudinal study of acculturation among recently arrived Hispanic immigrant families from Miami and Los Angeles, to achieve three primary aims: (1) examine the directionality between personal (coherence and confusion) and cultural (ethnic and US) identity, (2) evaluate the effects of personal and cultural identity on psychosocial functioning and health outcomes, and (3) examine the moderating role of site. For the first aim, we anticipated a bidirectional relationship between personal and cultural identity. For the second aim, although these processes may operate on psychosocial functioning in complex ways, prior cross-sectional research has suggested that personal identity predicts psychosocial functioning more strongly than cultural identity does and may even mediate the relationship between cultural identity and psychosocial functioning (Schwartz et al., 2010; Usborne & Taylor, 2010). As a result, we hypothesized that: (2a) coherence and confusion would positively predict positive psychosocial functioning and negatively predict negative psychosocial functioning and health risk behaviors more strongly than ethnic and US identity would; and (2b) cultural identity would predict outcomes indirectly through personal identity. Finally, with regard to Aim 3, given the fact that discriminatory events can activate cultural identities and render them more salient (Rumbaut, 2008) and the protective role of ethnic identity vis-à-vis the negative experiences associated with discrimination (Umaña-Taylor, 2011), we hypothesized that cultural identity would predict psychosocial functioning and health risk behaviors more strongly for adolescents in Los Angeles versus Miami.
Data came from a six-wave longitudinal study on acculturation, identity, family functioning, and health among recent Hispanic immigrant families (Schwartz et al., 2014c). The sample included 302 recent-immigrant Hispanic adolescents (53% boys; Mage=14.51 years at baseline; SD=.88 years, range 14–17 years). Eighty-five percent of the sample was retained across the four waves. The Miami sample (n=152) was primarily from Cuba (61%), the Dominican Republic (8%), Nicaragua (7%), Honduras (6%), Colombia (6%), and other Hispanic countries (12%); and the Los Angeles sample (n=150) was primarily from Mexico (70%), El Salvador (9%), Guatemala (6%), and other Hispanic countries (15%).
Baseline data were gathered during the summer and fall of 2010, and subsequent time points occurred every six months thereafter. Participants were recruited from randomly selected public high schools whose student bodies were at least 75% Hispanic. The study was approved by the Institutional Review Boards at both participating universities and by the Research Review Committees for the participating school districts. Each participant completed the assessment in English or Spanish, depending on her/his preference. Although 84% of adolescents completed their baseline assessments in Spanish, this percentage decreased to 66% by Time 4.
Unless otherwise specified, 5-point Likert scales were used for all study measures, with response options ranging from 1 (Strongly Disagree) to 5 (Strongly Agree). Alpha coefficients presented are from the current sample across relevant time points and for both English and Spanish versions. All measures were administered at each time pointed and developed using a two-step translation process (Sireci, Wang, Harter, & Ehrlich, 2006) whereby measures are translated by one translator from English to Spanish, back-translated by a second translator (Spanish to English), and then evaluated by both translators to resolve discrepancies.
The 12-item identity subscale from the Erikson Psychosocial Stage Inventory (EPSI; Rosenthal, Gurney, & Moore, 1981) was used to assess coherence and confusion. Six items are worded in a “positive” direction towards coherence (α=.69–.87; Sample item: “I know what kind of person I am”), and 6 items are worded in a “negative” direction towards confusion (α=.62–.77; Sample item: “I don’t really know who I am”). Prior analyses have not only established the validity of the measure in Spanish (Schwartz et al., 2014a) but also supported the two-factor structure, even after controlling for potential methodological effects of item wording (Schwartz et al., 2009b).
Ethnic A/C was assessed by combining the 4-item commitment and 3-item affirmation subscales from the Multi-Group Ethnic Identity Measure (MEIM; Roberts et al., 1999; α=.84–.93; Sample item: “I feel good about my cultural or ethnic background’). US A/C was assessed using matching items from the American Identity Measure (AIM; Schwartz et al., 2012; α= 87–.90; Sample item: “I am happy that I am an American”), again combining the commitment and affirmation subscales. The AIM was adapted from the MEIM, with “the United States” inserted in place of “my ethnic group.” Prior findings have established the validity of both the AIM and MEIM in Spanish (Schwartz et al., 2014a).
We assessed this construct in terms of self-esteem and optimism. The 10-item Rosenberg Self-Esteem Scale (RSES; Coopersmith, 1981) was used to measure self-esteem (α=.72–.84; Sample item: “On the whole, I am happy with myself”). This measure has been used widely with Spanish-speaking samples (Schmitt & Allik, 2005). The 6-item Children’s Hope Scale (CHS; Edwards, Ong, & Lopez, 2007) was used to measure optimism (α=.87–.95; Sample item: “I think I am doing pretty well”). This measure was designed specifically for use with Hispanics.
We assessed this construct in terms of depressive symptoms, rule breaking, and physical aggression. The 20-item Center for Epidemiologic Studies Depression Scale (CES-D; Radloff, 1977) assessed adolescents’ depressive symptoms (α=.81–.89, sample item: “I felt sad this week”). Items are rated on a 4-point scale ranging from 1 (Seldom) to 4 (Most of the Time) and ask participants how often they experienced various depressive symptoms during the week prior to assessment. The CES-D has been translated into Spanish and used frequently with Hispanic individuals (e.g., Todorova, Falcón, Lincoln, & Price, 2010). The Youth Self-Report (YSR; Achenbach, Newhouse, & Rescorla, 2004) was used to assess adolescents’ aggressive (17 items, α=.87–.93, sample item: “I physically attack people”) and rule breaking (15 items, α=.87–.92, sample item: “I lie or cheat”) behaviors. Response choices ranged from 0 (Not true) to 2 (often or very true). The YSR has been validated with Spanish-speaking adolescents (Lopez et al., 2008).
Cigarette and alcohol use were assessed using a modified version of the Monitoring the Future survey (Johnston, O’Malley, Bachman, & Schulenberg, 2014). Adolescents were asked about frequency of cigarette and alcohol use 90 days prior to each assessment point. Following Jemmott, Jemmott, and Fong (1998), we asked adolescents how often they have had vaginal or anal sex without using a condom during the past 90 days. The scale ranged from 0 (Never) to 4 (Always). Because of low base rates, we dichotomized the responses to create binary variables for alcohol use (use vs. nonuse), cigarette use (use vs. nonuse), and unprotected sex (abstinence/condom vs. sex without a condom) at Time 4.
Analyses were conducted in Mplus v7.2 (Muthén & Muthén, 1998–2012) using a sandwich covariance estimator (Kauermann & Carroll, 2001) to adjust the standard errors and account for nesting of participants within schools. Missing data were handled using full-information maximum likelihood estimation. Model fit was evaluated using the comparative fit index (CFI), the root mean square error of approximation (RMSEA), and the standardized root mean square residual (SRMR). According to values suggested by Little (2013), good fit is represented as CFI≥.95, RMSEA≤.06, and SRMR≤.06 1 . The analytic process proceeded in five steps. Across all steps, we controlled for age, gender, and years in the US. First, we calculated descriptive statistics and a correlation matrix for all study variables.
Second, to establish directionality of effects, we fit a model to the first three waves of data, such that cross-lagged paths from Times 1 to 2 and Times 2 to 3 were included between personal and cultural identity indicators (see Figure 1a ). Directional paths between personal and cultural identity were estimated from Times 1 to 2 (paths A1 and B1) and Times 2 to 3 (paths A2 and B2) as a way to examine replicability of the findings. Additionally, we allowed directional paths between indicators within each domain (e.g., coherence → confusion, US A/C → ethnic A/C). Moreover, to statistically ensure that any directional effects could be replicated, we explored stationarity, or non-varying cross-lagged paths across time (Allison, 1990). This was done by imposing equality constraints on corresponding cross-lagged relationships from Times 1–2 and Times 2–3, such that path A1 was constrained to be equal to A2, and B1 constrained to be equal to B2 (see Figure 1b ). For example, the path estimate between coherence at Time 1 (t) and ethnic A/C at Time 2 (t+1) was constrained to be equal to the path estimate between coherence at Time 2 (t) and ethnic A/C at Time 3 (t+1), producing one set of lagged path estimates corresponding to Time t and Time t+1 (paths A and B). To evaluate the tenability of these stationarity constraints, we compared the fit of models with and without these constraints using the ΔCFI (>.010) and ΔRMSEA (>.010) criteria to determine whether the stationarity assumption should be statistically rejected (Little, 2013) 2 .
Overview of Stationarity Constraints
To examine how these relationships may influence later positive development and risk behaviors, as part of Step 3, Time 4 outcome variables were added to the stationarity model, predicted by Time 3 personal and cultural identity processes (i.e., Personal/Cultural Identity Time 1–(paths A/B)→ Personal/Cultural Identity Time 2 –(paths A/B)→ Personal/Cultural Identity Time 3 –(path C)→ Outcomes Time 4). For each Time 4 outcome variable, Time 1 levels of that variable were controlled. In Step 4, we examined whether Time 3 cultural identity mediated the effect of invariant Time 1 and 2 personal identity on Time 4 outcome variables, and/or whether Time 3 personal identity mediated the effect of invariant Time 1 and 2 cultural identity on Time 4 outcome variables. Finally, in Step 5, we tested for significant site differences.
Table 1 provides means and standard deviations or frequencies for categorical outcomes and a zero-order correlation matrix for identity processes at T3 and outcomes at T4.
Descriptive Statistics and Bivariate Correlations between Time 3 Identity Processes and Time 4 Outcomes
| Mean (SD) | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1. Coherence (T3) | 2.98 (.70) | −.25* | .57* | .46* | −.27* | .18* | .29* | −.10 | −.15 | .05 | .02 | .04 |
| 2. Confusion (T3) | 1.52 (.79) | −.40* | −.14 | .01 | .40* | −.39* | −.24* | .27* | .25* | .04 | .03 | .09 |
| 3. Ethnic A/C (T3) | 2.84 (.75) | .51* | −.24* | .46* | −.24* | .15 | .10 | −.06 | −.11 | .10 | .02 | −.02 |
| 4. US A/C (T3) | 2.56 (.75) | .07 | .05 | .17 | −.21* | .04 | .15 | −.02 | .01 | .01 | −.01 | .10 |
| 5. Depression (T4) | 1.48 (.74) | −.22* | .28* | −.23* | −.13 | −.48* | −.25* | .52* | .57* | .06 | .19* | .06 |
| 6. Self-Esteem (T4) | 3.01 (.66) | .53* | −.33* | .29* | .34* | −.49* | .58* | −.31* | −.38* | −.14 | −.24* | −.13 |
| 7. Optimism (T4) | 3.93 (.95) | .41* | −.25* | .32* | .34* | −.46* | .60* | −.09 | −.15 | −.22* | −.09 | .01 |
| 8. Rule-Breaking (T4) | 0.27 (.34) | −.41* | .23* | −.23* | −.15 | .23* | −.32* | −.41* | .90* | .18* | .28* | .29* |
| 9. Aggression (T4) | 0.30 (.36) | −.27* | .20* | −.07 | −.13 | .37* | −.33* | −.29* | .76* | .11 | .23* | .23* |
| 10. Unprotected Sex (T4) 1 | 8.6% | −.07 | .20* | −.09 | −.14 | .32* | −.23* | −.11 | .05 | .07 | .30* | .30* |
| 11. Alcohol Use (T4) 1 | 8.9% | −.13 | .21* | −.12 | .01 | .15 | −.21* | −.06 | .45* | .23* | .33* | .47* |
| 12. Cigarette Use (T4) 1 | 5.6% | −.33* | .13 | .01 | −.11 | .01 | −.29 | −.04 | .49* | .27* | .16 | .67* |
Notes. Bivariate correlations for the Miami site are presented above the diagonal while the bivariate correlations for Los Angeles are provided below the diagonal.
1 For categorical outcomes, frequency of engagement in health risk behavior is reported.The cross-lagged model provided a poor fit to the data: χ 2 (25)=101.075, ppp=.060; ΔCFI=.010; ΔRMSEA=.007. Given these stationarity constraints, the findings can be simplified to comparing the cross-lagged path coefficients between Times t and t+1. As shown in Table 2 , coherence at Time t significantly predicted both ethnic and US A/C at Time t+1. Additionally, both ethnic and US A/C at Time t predicted coherence at Time t+1. However, confusion did not predict, and was not predicted by, any of the other identity variables. There were no significant predictive relationships between ethnic and US A/C.
Cross-Lagged Paths for Constrained Directionality Model
| Outcome (t+1) | Predictor (t) | Estimate 1 | p-value | 95% CI |
|---|---|---|---|---|
| Coherence | Confusion | −.082 | .273 | −.229 to .065 |
| Ethnic A/C | .069 | .003 | .024 to .114 | |
| U.S. A/C | .099 | .007 | .028 to .171 | |
| Confusion | Coherence | −.081 | .099 | −.177 to .015 |
| Ethnic A/C | .005 | .920 | −.084 to .094 | |
| U.S. A/C | −.032 | .621 | −.159 to .095 | |
| Ethnic A/C | U.S. A/C | −.050 | .987 | −.098 to .100 |
| Coherence | .245 | .004 | .058 to .297 | |
| Confusion | −.007 | .138 | −.021 to .156 | |
| U.S. A/C | Ethnic A/C | .001 | .199 | −.128 to .027 |
| Coherence | .179 | .171 to .319 | ||
| Confusion | .067 | .852 | −.092 to .076 |
Notes. All estimates are standardized regression coefficients.
To compare these path coefficients formally, we conducted four sets of invariance tests (e.g., coherence = ethnic A/C, confusion = US A/C). We compared the stationarity model to a model where the path coefficient for identity process A at Time t predicting identity process B at Time t+1 was constrained equal to the path coefficient for identity process B at Time t predicting identity process A at Time t+1. The results indicated a significant decline in model fit for ethnic A/C and coherence [Δχ 2 (1)=19.99, pp=.005; ΔCFI=.013; ΔRMSEA=.007] suggesting that the predictive effects of coherence on ethnic and US A/C were significantly stronger than vice versa.
Note. With the exception of categorical outcomes (i.e., cigarette use, alcohol use, and unprotected sex), which are Odds Ratio, all estimates are standardized regression coefficients. Parameter values from Time 1 to Time 2 and Time 2 to Time 3 are identical as a result of the stationarity constraints imposed in Step 1. For simplicity we have excluded 1) the autoregressive paths and 2) Time 1 controls.
Effects of Personal and Cultural Identity on Psychosocial Functioning
| Outcome (T4) | Predictor (T3) | Estimate | p-value | 95% CI |
|---|---|---|---|---|
| Depression | Coherence | −0.041 | 0.549 | −0.174 to 0.093 |
| Confusion | 0.264 | 0.173 to 0.354 | ||
| U.S. A/C | −0.095 | 0.045 | −0.190 to −0.002 | |
| Ethnic A/C | −0.110 | 0.085 | −0.236 to 0.015 | |
| Self Esteem | Coherence | 0.176 | 0.149 | −0.063 to 0.415 |
| Confusion | −0.245 | −0.358 to −0.133 | ||
| U.S. A/C | 0.057 | 0.542 | −0.126 to 0.240 | |
| Ethnic A/C | 0.009 | 0.925 | −0.178 to 0.196 | |
| Optimism | Coherence | 0.201 | 0.001 | 0.078 to 0.324 |
| Confusion | −0.147 | −0.202 to −0.091 | ||
| U.S. A/C | 0.114 | 0.092 | −0.019 to 0.247 | |
| Ethnic A/C | −0.014 | 0.874 | −0.183 to 0.156 | |
| Rule Break | Coherence | −0.128 | 0.011 | −0.227 to −0.030 |
| Confusion | 0.146 | 0.050 | 0.001 to 0.291 | |
| U.S. A/C | 0.010 | 0.736 | −0.048 to 0.068 | |
| Ethnic A/C | 0.001 | 0.998 | −0.159 to 0.159 | |
| Aggression | Coherence | −0.157 | −0.242 to −0.072 | |
| Confusion | 0.129 | 0.067 | −0.009 to 0.268 | |
| U.S. A/C | 0.026 | 0.604 | −0.072 to 0.124 | |
| Ethnic A/C | 0.011 | 0.893 | −0.147 to 0.168 | |
| Unprotected Sex 1 | Coherence | 1.096 | 0.654 | 0.732 to 1.642 |
| Confusion | 1.478 | 0.034 | 1.029 to 2.125 | |
| U.S. A/C | 0.880 | 0.601 | 0.544 to 1.422 | |
| Ethnic A/C | 1.278 | 0.405 | 0.718 to 2.273 | |
| Alcohol Use 1 | Coherence | 1.009 | 0.973 | 0.619 to 1.644 |
| Confusion | 1.412 | 0.048 | 1.002 to 1.990 | |
| U.S. A/C | 1.067 | 0.841 | 0.553 to 2.022 | |
| Ethnic A/C | 0.876 | 0.628 | 0.514 to 1.493 | |
| Cigarette Use 1 | Coherence | 0.482 | 0.193 | 0.160 to 1.447 |
| Confusion | 1.065 | 0.860 | 0.530 to 2.140 | |
| U.S. A/C | 1.149 | 0.773 | 0.448 to 2.950 | |
| Ethnic A/C | 1.579 | 0.237 | 0.740 to 3.372 |
Notes. Autoregressive paths not displayed. All estimates are standardized regression coefficients.
1 Estimates provided are Odds Ratio.On the fourth step, we examined two sets of mediated pathways: (1) invariant Time 1 and 2 Personal Identity → Time 3 Cultural Identity → Time 4 psychosocial functioning, and (2) invariant Time 1 and 2 Cultural Identity → Time 3 Personal Identity → Time 4 psychosocial functioning. Using the RMediation package (Tofighi & MacKinnon, 2011), we evaluated indirect effects of personal and cultural identity on outcome variables. All indirect effects were included in a single model to avoid Type I error inflation. Doing so provides more statistical power, and greater rigor, compared to the original Baron and Kenny (1986) approach to testing mediation (Fritz & MacKinnon, 2007). The RMediation package uses the asymmetric distribution of products test, which computes a 95% confidence interval around the product of the two path coefficients that comprise each potential mediating pathway. If this confidence interval does not include zero, then mediation is assumed at p
As displayed in Table 4 , seven indirect effects were found to be significantly different from zero. Specifically, invariant Time 1 and 2 ethnic A/C positively predicted Time 4 optimism, and negatively predicted Time 4 rule breaking and aggression, through Time 3 coherence. Similarly, invariant Time 1 and 2 US A/C positively predicted Time 4 optimism, and negatively predicted Time 4 rule breaking and aggression, through Time 3 coherence. Lastly, invariant Time 1 and 2 coherence negatively predicted Time 4 depressive symptoms through Time 3 US A/C.
Indirect Effects of Personal and Cultural Identity on Psychosocial Functioning
| Outcome (T4) | Predictor (T2) | Mediator (T3) | Estimate | 95% CI |
|---|---|---|---|---|
| Optimism | Ethnic A/C | Coherence | .020 | .004 to .043 |
| U.S. A/C | Coherence | .014 | .004 to .028 | |
| Depression | Coherence | U.S. A/C | −.018 | −.044 to −.001 |
| Rule Breaking | Ethnic A/C | Coherence | −.013 | −.029 to −.002 |
| U.S. A/C | Coherence | −.009 | −.019 to −.002 | |
| Aggression | Ethnic A/C | Coherence | −.016 | −.032 to −.003 |
| U.S. A/C | Coherence | −.011 | −.021 to −.003 |
Notes. All estimates reported are standardized regression coefficients and all indirect paths are significant at p
Finally, we sought to ascertain whether any of the effects differed across sites. To do this, we compared an unconstrained model (with all paths free to vary across site) to a constrained model (with each path constrained to be equal across site) using the likelihood ratio test to evaluate the null hypothesis of equivalent findings across sites. This test provides only a chi-square difference and does not provide any other SEM fit indices. We found a significant difference in fit between these two models, Δχ 2 (51)=115.01, p
Note. With the exception of categorical outcomes (i.e., cigarette use, alcohol use, and unprotected sex), which are Odds Ratio, all estimates are standardized regression coefficients. Parameter values from Time 1 to Time 2 and Time 2 to Time 3 are identical as a result of the stationarity constraints imposed in Step 1. For simplicity we have excluded 1) the autoregressive paths and 2) Time 1 controls. Paths in bold represent paths unique to this particular context while dashed paths are consistent across both sites.
Note. With the exception of categorical outcomes (i.e., cigarette use, alcohol use, and unprotected sex), which are Odds Ratio, all estimates are standardized regression coefficients. Parameter values from Time 1 to Time 2 and Time 2 to Time 3 are identical as a result of the stationarity constraints imposed in Step 1. For simplicity we have excluded 1) the autoregressive paths and 2) Time 1 controls. Paths in bold represent paths unique to this particular context while dashed paths are consistent across both sites.
Significant Site Differences
| Miami | Los Angeles | ||||||
|---|---|---|---|---|---|---|---|
| Outcome | Predictor | Estimate | p-value | 95% CI | Estimate | p-value | 95% CI |
| Confusion (t+1) | Ethnic A/C (t) | 0.070 | .147 | −0.025 to 0.166 | −0.173 | .013 | −0.312 to −0.034 |
| U.S. A/C (t+1) | Coherence (t) | 0.234 | 0.146 to 0.321 | 0.105 | .140 | −0.035 to 0.245 | |
| Self-Esteem (T4) | Cohesion (t) | 0.091 | .301 | −0.081 to 0.263 | 0.366 | .002 | 0.136 to 0.245 |
| U.S. A/C (T3) | −0.030 | .836 | −0.314 to 0.254 | 0.171 | .001 | 0.074 to 0.267 | |
| Optimism (T4) | Ethnic A/C (T3) | −0.142 | .034 | −0.274 to −0.011 | 0.120 | .248 | −0.083 to 0.323 |
| Unprotected Sex 1 (T4) | Confusion (T3) | 1.297 | .116 | 0.937 to 1.796 | 9.757 | .007 | 1.866 to 51.021 |
| Ethnic A/C (T3) | 1.095 | .179 | 0.959 to 1.251 | 0.773 | .011 | 0.634 to 0.942 | |
| Alcohol Use 1 (T4) | Confusion (T3) | 1.103 | .647 | 0.725 to 1.678 | 4.204 | .049 | 1.005 to 17.581 |
| Cigarette Use 1 (T4) | Coherence (T3) | 1.129 | .841 | 0.347 to 3.673 | 0.085 | .009 | 0.013 to 0.539 |
Notes. All estimates are standardized regression coefficients.
1 Estimates provided are Odds Ratio.With regard to the cross-lagged relationships between personal and cultural identity, while maintaining the stationarity constraints, ethnic A/C at Time t negatively predicted confusion at Time t+1 only for participants in Los Angeles. Additionally, coherence more strongly predicted US A/C in Miami than in Los Angeles. For participants in Miami, ethnic A/C at Time 3 predicted optimism at Time 4. For participants in Los Angeles, coherence positively predicted self-esteem and negatively predicted cigarette use, confusion positively predicted unprotected sex and alcohol use, ethnic A/C negatively predicted unprotected sex, and US identity positively predicted Time 4 self-esteem.
Significant Site Differences in Indirect Effects on Psychosocial Functioning
| Outcome (T4) | Predictor (T2) | Mediator (T3) | Indirect | 95% CI |
|---|---|---|---|---|
| Miami | ||||
| Optimism | Coherence | U.S. A/C | 0.033 | 0.001 to 0.066 |
| Ethnic A/C | Coherence | 0.020 | 0.003 to 0.037 | |
| Coherence | Optimism | 0.015 | 0.004 to 0.026 | |
| Depression | Coherence | U.S. A/C | −0.024 | −0.046 to −.002 |
| Rule Breaking | U.S. A/C | Coherence | −0.009 | −0.018 to −0.001 |
| Aggression | Ethnic Identity | Coherence | −0.015 | −0.028 to −0.002 |
| U.S. A/C | Coherence | −0.011 | −0.019 to −0.003 | |
| Los Angeles | ||||
| Self-Esteem | Ethnic A/C | Confusion | 0.041 | 0.002 to 0.080 |
| U.S. A/C | Coherence | 0.028 | 0.003 to 0.052 | |
| Optimism | Ethnic A/C | Coherence | 0.018 | 0.003 to 0.033 |
| Ethnic A/C | Confusion | 0.028 | 0.003 to 0.053 | |
| U.S. A/C | Coherence | 0.014 | 0.003 to 0.024 | |
| Aggression | Ethnic A/C | Coherence | −0.015 | −0.027 to −0.002 |
| U.S. A/C | Coherence | −0.011 | −0.019 to −0.003 | |
| Depression | Ethnic A/C | Confusion | −0.045 | −0.084 to −0.005 |
| Unprotected Sex 1 | Coherence | Ethnic A/C | 0.920 | 0.876 to 0.967 |
| Ethnic A/C | Confusion | 0.919 | 0.852 to 0.992 | |
| Alcohol Use 1 | Ethnic A/C | Confusion | 0.918 | 0.844 to 0.998 |
| Cigarette Use 1 | Ethnic A/C | Coherence | −0.069 | −0.126 to −0.012 |
| U.S. A/C | Coherence | 0.949 | 0.913 to 0.986 | |
Note: All estimates reported are standardized regression coefficients and all indirect paths are significant at p 1 Estimates provided are Odds Ratio.
This study examined the directionality between personal and cultural identity among recently immigrated Hispanic adolescents. We extend previous cross-sectional studies by evaluating the direct and indirect effects of personal and cultural identity on psychosocial functioning, and we evaluated site differences in these relationships.
Although cross-sectional studies have documented associations between cultural and personal identity, our findings clarify the direction of these effects. More specifically, not only did coherence predict greater ethnic and US A/C longitudinally, but ethnic and US A/C also predicted coherence. Our findings are consistent with arguments for bidirectional links between personal and cultural identity (e.g., Hitlin, 2003; Schwartz et al., 2008). However, the predictive effects of coherence on ethnic and US A/C were greater than the predictive effects of ethnic and US A/C on coherence. Thus, to some degree, understanding who one is and where one is going in life may serve to better orient youth towards understanding where they fit in relation to the cultural groups to which they belong. These findings support Schwartz, Montgomery, and Briones (2006), who argued that a coherent sense of personal identity would anchor the person during the post-immigration cultural identity transition. Indeed, without the development of a coherent personal identity providing a sense of direction and stability, adolescents are likely to experience a sense of aimlessness, anomie, and distress (Côté & Levine, 2002).
Additionally, three other important findings emerged from the cross-lagged analysis. First, results indicated non-significant longitudinal paths between coherence and confusion and between ethnic and US A/C. This lack of within-domain associations suggests that, although coherence and confusion, and US and ethnic A/C, may be conceptualized under personal and cultural identity respectively (Schwartz et al., 2013), the processes within each identity domain appear to overlap considerably less than would be expected. That being said, results indicated significant concurrent correlations between coherence and confusion and ethnic and US A/C. Thus, the processes with each identity domain appear to be interrelated within time but do not predict one another across time. Second, confusion seemed to operate largely independent of the other three identity processes. Confusion appears to represent a maladaptive sense of disorganization that is not necessarily the opposite of coherence but that predicts negative psychosocial and health outcomes. Indeed, there is evidence that a fragmented and confused sense of identity portends low self-esteem, internalizing symptoms, externalizing problems, and health risk behaviors among members of various ethnic groups (Oshri et al., 2014). Finally, while future studies are necessary to explore the directionality between personal and cultural identity across different development periods, as indicated by the stationarity results, these findings held across the first three waves of data, providing evidence for the generalizability of these findings over time.
As part of our second aim, we evaluated the direct effects of personal and cultural identity on psychosocial functioning. Consistent with other empirical studies (e.g., Syed et al., 2013), confusion emerged as a negative predictor of self-esteem and optimism, and as a positive predictor of depressive symptoms, rule breaking, unprotected sex, and cigarette use. Coherence, on the other hand, positively predicted optimism and negatively predicted rule breaking and aggression. Regarding components of cultural identity, after controlling for personal identity, US A/C was associated with lower levels of depressive symptoms, whereas ethnic A/C did not significantly predict any indicators of psychosocial functioning. As a whole, the scarcity of direct effects of cultural identity is consistent with cross-sectional studies where personal identity relates to well-being and internalizing symptoms more strongly than cultural identity does (Schwartz et al., 2009c; Usborne & Taylor, 2010). Nonetheless, our longitudinal findings provide important theoretical and practical insights. Specifically, the small number of direct effects of cultural identity on outcomes, after controlling for personal identity, may indicate that an understanding of who one is and where one is going is more critical than understanding where one fits within one’s ethnic group and the context where one resides.
In addition to examining these direct effects, we evaluated the indirect effects of personal and cultural identity on psychosocial functioning. Both ethnic and US A/C positively indirectly predicted optimism and negatively predicted rule breaking and aggression through coherence. Thus, the effects of cultural identity on psychosocial functioning, which have been documented in numerous studies (for reviews, see Rivas-Drake et al., 2014; Umaña-Taylor, 2011), may operate indirectly via coherence. Although interventions may be wise to target both personal and cultural identity, it is worth noting that coherence predicted cultural identity more strongly than vice versa. Given the direct effects of coherence and confusion on nearly every outcome and the directional effects between coherence and cultural identity, despite the fact that cultural identity was a significant mediator, targeting personal identity directly may serve as the most optimal mechanism for promoting both cultural identity and positive psychosocial functioning. With that said, it is important to note that identity development is likely a complex process. Although the current findings point to the importance of personal identity over cultural identity, it is possible that cultural identity influences personal identity in implicit or indirect ways.
Our results also indicated significant differences across sites. As a whole, cultural identity seemed to have played a larger role in Los Angeles than in Miami. Specifically, ethnic A/C, through confusion, predicted positive and negative psychosocial functioning and engagement in health risk behaviors. Through coherence, both ethnic and US A/C negatively predicted aggression and cigarette use. This finding highlights the role of cultural identity as a resource for personal identity development and points to key differences across contexts of reception. Given the protective role of ethnic A/C (Umaña-Taylor, 2011), adolescents in Los Angeles, where Hispanics may be received more ambivalently (Hayes-Bautista, 2004), might have less of an understanding of where they fit within their ethnic group and thus may experience a sense of confusion that inhibits personal identity development.
In Miami, results did not differ greatly from the findings from the sample as a whole. As such, although ethnic identity may inform adolescents’ sense of self and direction in life, in highly multicultural contexts it may not necessarily provide adolescents with the same amount of purpose and direction that it might in less welcoming contexts of reception. Given that research has indicated that the ethnic density of a community can impact awareness of one’s ethnicity (Feinauer & Whiting, 2012), in the highly bicultural and welcoming context of Miami (Stepick et al., 2003), where the majority of residents are first- or second-generation Hispanic immigrants (US Census Bureau, 2011), Hispanic adolescents may be less likely to encounter discrimination and social exclusion. As a result, ethnic A/C may function more independently and may even be somewhat optional. It should be noted however, that for adolescents in Miami, ethnic A/C was found to negatively predict optimism, despite the non-significant yet positive bivariate correlation. Given that Hispanics represent the majority of Miami residents, one potential explanation for this counter-intuitive finding may be that more highly identified individuals are more likely to attribute negative social experiences to discrimination (Hall & Carter, 2006; Sellers & Shelton, 2003).
In sum, given that ethnic and US identity predicted both coherence, and ethnic identity predicted confusion, which in turn predicted nearly every outcome, identity-based interventions might benefit by targeting both personal and cultural identity in contexts marked by ambivalence towards Hispanic adolescents. On the other hand, in ethnically dense, highly bicultural, and welcoming contexts, interventions might be better served by targeting personal identity directly. That being said, it is important to note that the interplay of identity processes may differ across country of origin or receiving community (which were confounded in our sample). Indeed, within the current sample, adolescents in Miami were largely of Cuban descent whereas those in Los Angeles were primarily of Mexican descent. Future studies should further explore the interplay of identity processes across context of reception and nativity.
The present results should be interpreted in light of several limitations. All of our participants were recent immigrants residing in major metropolitan areas. It is possible that adolescents had already begun developing their personal identity in their country of origin, where cultural identity may have been less salient. While the results from the stationarity results provide evidence for the tenability of these findings over time, cultural identity may have only become salient following immigration. Thus, whether these findings can be generalized to immigrants who migrated before or after adolescence remains an empirical question. Given that some of the adolescents we sampled may still be in the process of exploring what it means to have both US and ethnic identities, future studies should also explore the interplay between personal and cultural identity domains. In addition, results may have been different in other cities or in “new receiving communities” where many Hispanic immigrants are settling (e.g., Kiang, Perreira, & Fuligni, 2011). Given that some of our findings differed between sites, and all of our participants were recruited from heavily Hispanic areas; future studies should examine the interplay between identity processes in less dense Hispanic communities. The current sample was also limited to individuals who could be reached and were not planning to move. Thus, findings may not generalize to adolescents whose families are more transient, such as seasonal workers. Finally, the high correlations between the affirmation and commitment subscales of the MEIM and AIM made it impossible to differentiate these two subscales, which represent two theoretically distinct processes (Umaña-Taylor, Yazedjian, & Bámaca-Gómez, 2004). The lack of differentiation between affirmation and commitment may be due to recency of immigration, the developmental period under study, or problems with the measures. For recent immigrants and/or early adolescents, the process of understanding what ones’ ethnic and US membership means to them may be inextricably tied to their positive regard towards their cultural group membership. As such, it is necessary for future studies to uncouple these potential effects by using different instruments and by studying longer-term or older adolescents.
Our results suggest that personal and cultural identity influence one another in complex ways, and that at least some of this interplay differs according to the social context in which it is occurring. The ways in which personal and cultural identity impact psychosocial and risk-taking outcomes also appear to differ by context. Because we studied only two sites, we do not know which is the “rule” and which is the “exception.” It is therefore essential to conduct more work on personal identity, cultural identity, and their links with psychosocial and health outcomes across multiple contexts. Nonetheless, our findings hold important implications for the development and delivery of identity-focused interventions among recently immigrated Hispanic adolescents. By helping adolescents establish a coherent self-view and reconcile contradictory self-beliefs (cf. Harter, 2012), we may be able to improve their psychosocial functioning. We hope that this study will inspire more work in this direction.
1 Although we report the χ 2 value, we did not use it to gauge model fit because it tests a null hypothesis of perfect fit, which is rarely plausible with large samples or complex models (Davey & Savla, 2010).
2 Although we report the Δχ 2 difference test, because it tests the null hypothesis that two paths or models are exactly equivalent (Meade, Johnson, & Braddy, 2008), we did not rely on the Δχ 2 difference test in our interpretations.
