Hispanic immigrants and their children are not only the fastest growing population in this country, but they are also among the poorest (Portes and Rumbaut 2006). For Hispanic youth, the pathway to higher social mobility is fraught with formidable obstacles, like discrimination and suboptimal inner-city schools, that render them unprepared for the challenges of the labor market, and it appears that, counter to traditional patterns of immigrant incorporation, they are experiencing downward adjustment. This becomes a problem for Hispanics, and for U.S. society as a whole, given the increasing size of the Hispanic population. If Hispanic educational and economic indicators do not improve, as their proportions grow, poverty in the United States will grow correspondingly.

Redfield et al. (1936) defined acculturation as the “phenomena which result when groups of individuals having different cultures come into continuous first-hand contact, with subsequent changes in the original cultural patterns of either or both groups.” Yet, time in the United States is typically associated with second-culture acquisition (Rudmin 2010), part of a complex and stressful process (Ward 1996). For Hispanic youth, acculturation often occurs in a context influenced by, often hostile environments that are not easy to navigate. Arguably, the multifaceted process of acculturation can add to the already tumultuous emotions experienced in adolescence and may impact self-esteem. As a result, immigrant youth acculturate in different ways, influenced in large part by the contexts with which they interact (Piedra and Engstrom 2009). The contextual changes that young immigrants face in the normal course of their lives may vary widely and if these contexts are culturally distant from one another, the cultural norms between them can be difficult to reconcile. For instance, a young person who has little mastery of the English language may not have a positive sense of self when in the company of more acculturated peers or classmates. In those cases, they may feel closer to their families with whom they are likely to be more culturally matched (e.g., how they value the importance of family) and despite the generational differences.

Poor self-esteem has been associated problem behaviors, like dropping out of school and poor social and behavioral adjustment (Weiner 2000). More broadly, a recent trend, referred to as the “immigrant health paradox” is based on, a consistent, and growing body of evidence mostly from the field of epidemiology suggests that acculturated adolescents experience lower levels of well-being than less acculturated counterparts in the United States (Harker 2001; Mendoza et al. 2007; Rumbaut 1997). Although findings of this paradox are too numerous to cite and largely focused on Mexicans (Alegría et al. 2008), consider that in a sample of Hispanic adolescents, acculturation has been associated with higher substance use (Greenman and Xie 2008; Turner and Gil 2002), chronic illness (Wickrama et al. 2007), and lower risky sexual behaviors (Weiss and Tillman 2009). Vega and Alegría (2001) found some support to suggest that Hispanics have better mental health upon arrival than after achieving higher levels of acculturation. Researchers exploring the possible reasons for negative association between acculturation and well-being often include loss of family resources such as conflicts within families stemming from acculturation gaps between parents and their children as factors shown to reduce family cohesion (Smokowski et al. 2007, 2008). As Hispanics, defined as people who trace their ancestry to countries that share a common language, history, and certain cultural traits, which may or may not carry across the generations as the peoples acculturate (Cafferty 1985), continue to increase in size it is important to understand their continuing problematic patterns of acculturation, which run counter to the popular notion that immigrants are better off acculturating.

Today more is known about the important role that acculturation plays as a moderator of the well-being of Hispanic youth, and it is acknowledged that full assimilation into mainstream U.S. life does not take place in just one generation. The relationship between acculturation and self-esteem has been found to vary in myriad ways (Berry et al. 2006; Bornstein and Cote 2007; Lang et al. 1982). Much of the variation in findings may be attributed to methodological differences in how acculturation is conceptualized and measured (Rudmin 2009). As an individual-level proxy for acculturation, language acquisition occurs quickly, especially in young people. Furthermore, it captures a complex process that includes affective, cognitive, and behavioral components (Cuéllar et al. 1995). Language fluency has been shown to account for a large portion of the variance in multiple-item measures (Rogler et al. 1991) and it continues to function as the most broadly used measure of cultural practice. All acculturation scales, whether complex or simple, are imperfect in some ways. Escobar and Vega (2000), after reviewing the literature, recommend using individual measures (e.g., language and length of stay) rather than multidimensional ones with uncertain value, as is currently the case in the field, until a single acculturation measure is fully validated scientifically.

Because little is known about a young Hispanic’s well-being as they interact with different and probably widely differing environments, this study set out to understand the inter-relationships between acculturation, the contexts in which youth interact, and their effect on self-esteem. Data collected using the Experience Sampling Method, a repetitive sampling technique that assessed adolescents’ self-esteem in many different contexts throughout the course of one week, is analyzed in this study to argue that acculturation and context moderate self-esteem. Using advanced statistical design that allowed background differences to be held constant to reduce bias contributes to the literature on better understanding the role of acculturation, context, and self-esteem across groups dissimilar in many socio-economic dimensions (e.g., parental income, personality, age, and gender). This study can assist policy makers and practitioners to better position programs and policies in ways that better reflect on the experiences and needs of this population.

Person-Context Perspective

The effects on well-being from interactions between people and the contexts in which time is spent are important aspects of ecologically valid assessment (Bronfenbrenner 1986). Ecologically valid assessment views a person’s well-being holistically, part of a broader system related to the environment of which he or she is a part (Sameroff 1995). Ecological theorizing is a core social-work perspective. The perspective of person-context interaction congruence is predicated on the idea that people’s well-being will be better when they are living in contexts in which they feel there is a good fit (Emmons et al. 1985). For example, if youth are happy when in the context of family they are likely to benefit from socialization better than if they were in conflict with their parents on a regular basis. The person-context congruence perspective is a helpful theoretical lens through which one can observe the effect of acculturation and context on self-esteem. The importance of studying complex person-in-an-environment effects on well-being is highlighted in the literature on stress and coping (Lazarus and Folkman 1984). For example, adolescents who are better able to focus their attention on challenging and productive activities have been found to be more resilient than those who cannot (Garmezy et al. 1984). Therefore, if there is congruence between spending time with friends and acculturation level, an adolescent’s self-esteem should be positive relative to other contexts in which he or she spends time. For instance, if adolescents are interested in fitting into peer culture, and they feel accepted when in the context of peers, it is expected that they will report positive self-esteem.

As Hispanic immigrant youth acculturate to U.S. life, they face many linguistic and cultural challenges along the way and sometimes these adversely affect their well-being—including self-esteem. Hispanic youth upon arrival struggle to adapt to the culture and language of the United States and during this time they must reconcile among the different contexts in which they spend time. Usually by virtue of being in school, younger individuals learn the language of the host culture relatively easily and adapt rapidly, although acculturation can be moderated by the surrounding cultures of the communities in which they live. Parents are typically slower to learn English. Cultural and linguistic retention are likely to be of more practical value in communities where the immigrant’s native language is widely spoken. Among immigrant adolescents in the United States, unless they are recently arrived the preferred language is typically English. For many, learning a second language means losing their first (Fillmore 1991). If their first language was Spanish, it is possible that their parents are bilingual or have yet to learn English. As a result, language use is an effective proxy measure. For example, youth who speak Spanish are likely to have arrived recently in the United States. Among those whose first language was Spanish, they are likely to be speaking English after a few years in the United States, but their parents are likely to retain Spanish longer and, by proxy, most likely their culture as well. Youth whose first language was English likely have highly acculturated parents with whom they are probably culturally matched.

Context and Self-Esteem

The important role that context—defined as time spent with people such as family, friends, at school, with others, or alone—plays in the lives of immigrant youth cannot be understated. Immigrant youth on any given day can enter into many different types of contexts ranging from situations that require varying levels of acculturation. For instance, with parents who are less acculturated and who do not speak English, they must accommodate to their acculturation level. A reverse situation more than likely occurs at school, where the language of instruction is typically English. As youth enter and exit these widely different contexts they must on their own reconcile their beliefs and ideas and make decisions about which set of norms and values they will adopt. This can be terribly confusing during times where the contexts differ significantly from one another.

Self-esteem is one of many variables that are often used to assess well-being among youth. However, self-esteem has been assessed as a global measure or one that is context-dependent. This study focuses on the latter definition, where less is known. A context-dependent measure of self-esteem means that assessing it can vary greatly depending on the circumstances that individuals are experiencing when they are being assessed. James (1890) first identified the importance of context regarding self-esteem when he described it as the ratio of expectations to achievement. In this definition, self-esteem fluctuates according to the particular conditions of the context, or state. An individual’s perceived achievement of a goal is related to the importance that the individual places on achievement of that particular goal. Hence, under this definition, self-esteem depends on the frame of reference, including the context, where the subjective judgments that individuals use to evaluate their accomplishments take place. In this two-part process, the individual’s perception of success or failure to reach goals influences the outcome. Thus, lowered expectations can increase self-esteem, and expectations that are too high can result in unfulfilled accomplishments and therefore lower self-esteem. Students with higher self-esteem are believed to be better adjusted, be more accomplished in school, and have close, trusting relationships with their parents (Gove et al. 1990).

The level of acculturation and contexts in which youth spend time play an important role in how youth feel about themselves. Psychologists and sociologists have used self-esteem measures to evaluate levels of well-being between immigrants and nonimmigrants and between immigrants of different generations (Berry et al. 2006; Leondari 2001). Some studies have explored the relationship between the level of acculturation and global self-esteem; however, findings have been inconclusive. The effects of contextual factors on self-esteem warrant consideration (Boden et al. 2008). For example, in the context of the Hispanic family, family cohesion has been observed to weaken with acculturation to U.S. life (Sabogal et al. 1987). A consequence of family conflict as a young person adapts to U.S. life contributes to lowered well-being, including self-esteem (Portes and Zady 2002; Smokowski and Bacallao 2007).

In families, self-esteem can be used to detect the presence or absence of conflict, although conflict within a family is a difficult variable to define and measure (Pawlak and Klein 1997). Vega et al. (1995) reported that both low- and high-acculturation groups experienced conflict related to language use. The less acculturated reported experiencing stress both at home and at school. The more acculturated reported experiencing language conflicts, perceiving discrimination, and feeling that they were in a closed society. The authors speculated that the native-born/high-acculturation group might perceive discrimination to be a greater hindrance to their desire to succeed than the less acculturated group. Success is an integral part of the self-esteem equation as defined by James (1890).

The importance of family as a perceived source of support has been found to be robust across different Hispanic subgroups (e.g., Mexicans, Cubans, and Central Americans) (Sabogal et al. 1987) but has been found to manifest differently between Anglo and Mexican cultures (Luna et al. 1996). Familism, which refers to the within-family support systems is typically identified as protective factors for young Hispanics. Family cohesion has been found to be a protective factor against adolescent acting out (e.g., conduct problems, aggressive behavior, and rule-breaking) (Marsiglia et al. 2009). Hispanics who follow familismo and retain aspects of native culture are believed to enjoy closer family bonds in a collective and respectful orientation. Respeto, or respect for elders, is an example of another Hispanic/U.S. cultural difference. Acculturating Hispanic youth may offend their elders if they internalize the U.S. norm, which emphasizes the individual’s rights and autonomy.

A problematic aspect of acculturation is that often younger family members outpace older ones by learning the language of the host society more rapidly, something that facilitates their access to societal institutions. Asymmetrical acculturation such as this can introduce an acculturation gap with the potential for family discord (Kwak 2003; Portes and Rumbaut 2006; Szapocznik et al. 1986) and increased intergenerational conflict (Szapocznik et al. 1986). Acculturation gaps in Hispanic families have been found to be inversely associated with familial obligation (Sabogal et al. 1987) and family cohesion (Smokowski et al. 2008). Parent-adolescent conflict, which has been found to increase over subsequent immigrant generations has been shown to moderate the negative association between acculturation and self-esteem (Dennis et al. 2010). Indeed, when children acculturate faster than their parents, as often happens, conflicts in the family occur, creating additional stressors for its members (Szapocznik et al. 1986).

Prior studies found that adolescents in the United States allocate more of their time to their friends than to their families. They also reported that they enjoyed the time they spent with their friends much more than the time they spent with their families. Although this is likely to be normative behavior (and attitude) in adolescence, it is of concern because it may get in the way of receiving needed sanctions from parents, which is a necessary part of the socialization process (Csikszentmihalyi and Larson 1984). For instance, some research has found lack of delinquency related to time spent with family on weekends (Pabon 1998), and, more generally, parental monitoring (Halgunseth et al. 2006). Of course, it should be noted that some research has found that peers can play a positive role in the lives of adolescents (Youniss and Smollar 1985), and this can be an important source of self-esteem. Thus, the existing literature has not focused much on understanding how Hispanic adolescents differ on the amount of time they spend across different contexts and this too will be explored in this study.

Other contexts besides family and friends have also been found to affect the self-esteem of young people (Csikszentmihalyi and Larson 1984; Csikszentmihalyi and Schneider 2000). Some studies found students to be unhappy or disengaged, whereas others found that Hispanic students, more than any other ethnic/racial group, experience higher levels of enjoyment and affect (measured by scores on a scale measuring sociable, proud, happy, and relaxed) in school (Shernoff et al. 2000). Time spent alone is not associated with positive emotions (Hunter and Csikszentmihalyi 2003; Larson and Csikszentmihalyi 1980), but it can act as a stress buffer (Larson and Lee 1996) and arguably is needed to engage in productive tasks (Csikszentmihalyi and Larson 1984). For introverted personality types, spending time alone has been found to be more problematic than for the extroverted (Brandstätter 1994).

Summary and Hypothesis

This study is motivated by the important and growing concern of the downward effects of acculturation (and its concomitant constructs, lowered family cohesion and conflict) on the well-being of Hispanic adolescents. Using the lens of person-context congruence to understand the dynamic changes in self-esteem corresponding with changes in the contexts where adolescents spend time, this research set out to better understand whether acculturation and context together moderate self-esteem. There have been studies that explored the effects of acculturation on self-esteem and there are studies that support the idea that self-esteem has state and trait qualities. However, not much is known about how self-esteem changes in response to changes in context together with the level of acculturation. It is also important to consider the effect of background differences of participants (e.g., personality) on self-esteem because of their potential for biasing results. In nonexperimental research, it is difficult to compare groups, because people vary in many ways. Because self-esteem is thought to have both state (i.e., momentary or fluid qualities varying according to the circumstances in which one finds oneself) and trait (i.e., global personality) qualities, one way to isolate the effect of contextual changes while holding trait constant is to employ person-context congruence models, which help researchers to assess moment-by-moment changes in self-esteem. It is equally helpful to control for background differences by employing statistical techniques (e.g., fixed-effects models) that hold background differences constant (Wooldridge 2006).

This study explores differences in language use, context, and the associated self-esteem changes while comparing cross-cultural groups. Drawing on person-context congruence theories, the focus is on how self-esteem changes when adolescents move from one context to another. This study explores how youth at different levels of linguistic acculturation feel in the various contexts where they spend significant amounts of time and that are important to their development. Particular emphasis is placed on how youth at different levels of linguistic acculturation feel when with family versus nonfamily. If, for example, poorly linguistically acculturated youth have better self-esteem with family than with friends, and the reverse is true of more linguistically acculturated youth, linguistic acculturation can be said to moderate the experience between context (e.g., family) and self-esteem. Controlling for unobserved background differences, and the type of activity youth were engaged in, we would predict that linguistic acculturation would moderate the effect of context on self-esteem. Specifically, given prior research, we hypothesize that with increased linguistic acculturation, self-esteem will be lower with family than with friends and others.

Methodology

Research Design

This study used data from a large, nationally representative sample employing the Experience Sampling Method (ESM). ESM, a technique popular in positive psychology, employs a repeated-measure assessment technique over the course of 1 week to assist with in situ data collection, which provides higher reliability than methods based on one-time questionnaires that rely on recall (Csikszentmihalyi and Schneider 2000). In this study, ESM helps us understand how linguistic acculturation moderates the adolescent experience of emotional well-being in important contexts, especially between home and friends. Use of a fixed-effects model to analyze ESM data is advantageous because it can control for background differences, often a major problem in cross-cultural assessments (Wooldridge 2006). Although many ESM researchers aggregate the data to the person level, this study uses a fixed-effects statistical model, which makes it possible to analyze how self-esteem changes from moment to moment as participants enter different contexts—with family, friends, school, alone, or elsewhere. The model also allows the researcher to control for background differences (i.e., fixed effects).

Sample

Participants for this study were drawn from the first wave of data, gathered in 1992–1993, from the longitudinal Alfred P. Sloan Study of Youth and Social Development, which had three waves of ESM data extending to 1997. Sloan Center researchers sought participants from 33 public middle and high schools, in 12 locations across the country, with diverse levels of urbanization, racial/ethnic composition, and economic diversity, in grades 6, 8, 10, and 12. A protocol for collecting this data in a way that protected human subjects was in place at the University of Chicago, where a team of faculty and students received IRB approval for the study.

This study focuses principally on Hispanic adolescents. For this analysis, 796 participants met selection criteria. Each of these participants filled out up to 56 Experience Sampling Forms (ESFs), thereby contributing 22,335 response sets, which were then analyzed. (This approach will be discussed in the section on procedures and data analysis, below.) Of the Hispanic subsample for this study, which totaled 105, about 66% were of Mexican descent, 10% were Puerto Rican, 1% were Cuban, and the remainder were from various countries in Latin America. All Hispanic participants were included in order to attain sufficient power, given the large number of variables needed for analysis. Hispanics were contrasted with three comparison groups: 487 White non-Hispanic, English-dominant participants who met the inclusion criteria for comparison purposes; 158 non-Hispanic others whose first and current language was English (mostly African Americans); and 46 who reported speaking multiple languages as children and/or in the present adolescent period.

Procedures

To ensure that participants understood the instruments they would be completing, participants met in groups with a researcher before and after the study. Student participants were asked to fill out the same ESM instrument multiple times during the course of a week, to a maximum of 56 response sets per participant. Participants were instructed to fill out their questionnaires each time their preprogrammed wristwatches signaled them to do so. These signals were set to go off at eight random times per day for up to 7 days.

Measures and Variables

Instruments

A one-time questionnaire, NELS, based on the National Education Longitudinal Study 1988–1994, provided language data (Hafner et al. 1990). Following the ESM, an ESF was filled out each time a participant was signaled (up to a possible total of 56 ESFs). Although not all participants responded to all questions, multiple responses were recorded for most participants. All instruments were administered in Spanish or English, according to participants’ language preference.

Techniques such as ESM (Prescott et al. 1981) facilitate the exploration of person-context interactions (Csikszentmihalyi and Csikszentmihalyi 1988; Csikszentmihalyi and Larson 1984; Csikszentmihalyi and Schneider 2000), and because they obtain repeated measures for each individual, score high in reliability (Hektner et al. 2002). ESM gathers multiple scores per respondent unobtrusively, because no interviewer is present, and in situ, in the context in which the participant is engaged when surveyed (Csikszentmihalyi and Larson 1987; Hormuth 1986; Moneta and Csikszentmihalyi 1996). There is no reliance on participants’ ability to recall deep emotions, which are often difficult to remember, much less express.

Outcome Variable: Self-Esteem

The self-esteem variable is a composite (average) score based on select items from the Rosenberg Self-Esteem Scale; to reduce participant fatigue, some scales were shortened (Csikszentmihalyi and Larson 1987). Self-esteem items included “living up to the expectations of others,” “living up to your own expectations,” “feeling good about self,” “succeeding at present activity,” and “feeling in control” on a scale from 0 to 9 (0 = “not at all”; 9 = “very much”). One additional variable, “Were you succeeding at what you were doing?,” originally on a 1–9 scale, was transformed to a 0–9 scale.

Explanatory Set of Variables: Context

The mutually exclusive binary variables for context are defined by time when the respondent filled out the ESF: when engaged in schoolwork, alone, with friends, with families, or in other contexts.

  • Schoolwork. The first was time defined by occasions when the respondent specified being at a place categorized by a coder as schoolwork; if these times overlapped with other contexts, only time in this context counted.

  • Alone. Time alone consists of time with no one else.

  • Family. The family variable includes times when the participant reported being with mother, father, both, a sibling, or another relative and in no other context.

  • Friends. Time with friends consists of times when the participant was with a friend but in no other context.

  • Other contexts. Times in other contexts include occasions when the respondent was with relatives, siblings, strangers, or others, as well as in multiple contexts, but not at school or work. When a participant’s time overlapped with other contexts, the responses were placed into the “other context” variable.

The categories were forced to be mutually exclusive and exhaustive and the variable “family” was left out of the regression equations. Main effects for “context” may be included in the fixed-effects model because context varies for each individual (variation by response signal rather than just between individuals).

Explanatory Set of Variables: Acculturation Group

Based on two binary variables (race/ethnicity and language), participants were placed into six acculturation groups representing a continuum between unacculturated and acculturated states.

  • Less-Linguistically Acculturated Hispanics. For the first of these, the less-linguistically acculturated Hispanics group, Spanish was the first language they learned to speak and still spoke most often at the time of the study.

  • Somewhat-Linguistically Acculturated Hispanics. For the second group, the somewhat linguistically acculturated Hispanics, the first language learned was Spanish, but English was currently spoken most often.

  • Linguistically Acculturated Hispanics. The third, the linguistically acculturated Hispanics group, consists of Hispanics whose first and current language was English.

  • White Non-Hispanic English-Dominant. The principal benchmark group, against which all of the Hispanic subgroups were compared, constituted the fourth group, White, non-Hispanic English-dominant, whose first and current language was English. Other benchmark groups were included to better compare Hispanics.

  • Multiple Languages. The fifth group, a multiple-languages group, consisted of participants of any ethnic heritage who spoke multiple languages, either from birth or at the time of the interview.

  • Non-Hispanic Other. The sixth group, non-Hispanic other, included all non-Hispanic participants who were neither Hispanic nor White; these were non-Hispanic and non-White participants whose first and current dominant language was English.

Explanatory Set of Variables: Moderators of Linguistic Acculturation and Context

The self-esteem of Hispanic youth at various levels of linguistic acculturation is expected to vary with the context that respondents are in. These mutually exclusive and exhaustive binary variables consist of designations formed by the interaction between the binary linguistic acculturation group variable (less-linguistically acculturated Hispanics, somewhat linguistically acculturated Hispanics, linguistically acculturated Hispanics, White, non-Hispanic English-dominant, multiple-language, non-Hispanic other) and the binary context variable (with family, with friends, in school or work, alone, or other contexts). The linguistic acculturation group variable does not vary for an individual, and cannot be used in a fixed-effects model; however, their inclusion is allowed by matching them with context variables that fluctuate from observation to observation.

Control Variables: Activities

Explicit control variables for the types of activities engaged in were measured by the ESM survey question, “What was the main thing you were doing?” Responses to this question were classified into five categories of binary variables: leisure, work, maintenance (e.g., brushing teeth), deviant, and distressed. The criteria used to categorize activities are roughly similar to those found by Csikszentmihalyi and Schneider (2000). Cases in which the respondent failed to specify the activity being performed at the time of the response signal were dropped from the model. For instance, activities involving personal care, sitting, relaxing, or thinking were classified as maintenance. Talking to someone, going to a movie, playing sports, and so on, were classified as leisure. A “distressed” activity consisted of a situation in which the person was crying or in emotional pain (e.g., attending a funeral). The reference group in the equation is leisure; because these variables represent all leisure activities and the remaining variables are mutually exclusive and exhaustive, leisure was left out of the regression model.

Data Analysis Using a Fixed-Effects Model

To explore how the effect of linguistic acculturation on self-esteem varies by context, while controlling explicitly for the type of activity participants were involved in, this study used an analytic framework with a fixed-effects, multiple-regression model. Fixed-effects regression models allowed participants’ self-esteem scores in one context (e.g., when with parents) to be observed relative to another context (e.g., friends). Relative within-person contextually based measures have, among other advantages, the ability to avoid inexact comparisons across people and allow person-by-context factors to be assessed firsthand. Exploration of relative changes in how a person feels across different contexts, as opposed to comparisons between groups, helps to control for group heterogeneity. Fixed-effects models are useful in controlling for unobserved background differences across participants and in helping to reduce autocorrelation in the data.

Figure 1 shows a schematic of the proposed general fixed-effects model. The people with whom time is spent define context. Acculturation groups are implicitly controlled because demographic and other person-level characteristics that do not change within the person automatically drop out of a fixed-effects model. The only way in which the person-level variables do not drop out of the equation is by matching them with variables that do not drop out. So, this study created interaction terms between the person-level variables that do not change from person to person in the course of 1 week (e.g., linguistic acculturation) with each of the context variables (e.g., family, friends), which do change as participants went from one context to the next during the course of the study. The interaction terms are change scores showing how participants’ self-esteem was experienced. Observations of linguistic acculturation group by context are measurable and were useful for answering the research questions of interest. In addition, any unobserved person-level characteristics that do not change within the person, as well as specified (i.e., observed) activities in which adolescents engage, are controlled.

Fig.
figure 1

A fixed-effects model where linguistic acculturation group moderates context. * A main effect of linguistic acculturation cannot be explicitly estimated using fixed effects because it does not vary within an individual

The fixed-effects model demonstrates the relationship between self-esteem and other variables using ordinary least squares with fixed effects.

$$ y = \beta_{o} + X_{1} \beta_{1} + X_{2} \beta_{2} + X_{3} \beta_{3} + X_{4} \beta_{4} + a_{i} + \varepsilon $$

where y stands for one of three well-being indicators, β 0 is the constant, X 1 is a vector of activity variables, X 2 is a vector of context variables, X 3 is a vector of interaction terms, X 4 stands for the time varying intercept (response signal dummies 1… 56), a i is the expected unobserved time invariant effects where i indicates person 1… N.

Fixed-effects models correct for methodological problems, such as omitted variable bias and some autocorrelation, while allowing full use of the dataset. Fixed-effects models implicitly control for any unmeasured trait that varies between, but not within, individuals over time, and eliminate the dependence found in longitudinal data.

Results

Descriptive

Descriptive statistics are shown in Tables 1 and 2. Applying chi-square statistics to person-level data, Table 1 shows that participants’ demographic characteristics differ across the language use groups with respect to grade, socioeconomic class of community, and gender. Also, participants appear to spend the bulk of their time on leisure activities, followed by work and maintenance tasks (see Table 2). The context in which they spent the bulk of their time was on schoolwork (ranging from 31% of the less linguistically acculturated Hispanics to 40.5% for the most linguistically acculturated Hispanics). Time with family occupied the next highest time allocation with White non-Hispanics spending the least amount of time (20%) with family, to less acculturated Hispanics spending 33% of their time with family. Across all groups, time spent with friends rather than schoolwork was minor; less acculturated Hispanics allocated 5% of their time with friends, whereas the most linguistically acculturated Hispanics spent 11% of their time with friends. Interestingly, less acculturated Hispanics spent the least amount of time on schoolwork and the most with family. Also, employing Wald chi-square statistics on disaggregated data, Table 2 shows that the groups differ significantly in some but not all of the types of activities in which they engaged and the contexts in which they spent time. Participants in the different language-use groups differed significantly only on time spent at leisure and maintenance. Within each context, there were significant differences between the groups on schoolwork, family, and friends. All in all, these results confirm the presence of intergroup heterogeneity, which was controlled by the fixed-effects model. Fixed-effects models, by holding all person-level differences constant, minimize measurement bias.

Table 1 Summary statistics for demographic variables
Table 2 Summary statistics for activity variables

Fixed Effects

Using STATA to analyze data, controlling for fixed effects, or rather background differences across participants, and also the type of activity engaged in (e.g., work, maintenance, deviant, or distressed), data support the hypothesis that linguistic acculturation moderates the effect of context on self-esteem. For most of the participant groups studied, results showed that linguistic acculturation level did indeed moderate the effect of context. In other words, the combination of participants’ level of linguistic acculturation and whom they spend time with, tempered participants’ ability to experience positive or negative self-esteem relative to being with family. The hypothesis was that with increased linguistic acculturation, self-esteem was predicted to be lower with family than with friends; prior research showed that familism declines with acculturation and that gaps grow as children outpace their parents in the linguistic acculturation process. However, results are more nuanced than anticipated. Instead of seeing a progression toward higher self-esteem with friends than with family as linguistic acculturation increased, it was the group in the middle, the somewhat linguistically acculturated, rather than those at the high end of the linguistic acculturation process, who had higher self-esteem with friends than with family.

As can be seen in Table 3, coefficients are change scores as participants’ self-esteem changed from one context to the next. These are based on changes to each person’s base response; the first signal responded to acts as the control to which all subsequent responses are compared. Because all the acculturation groups are mutually exclusive and exhaustive, this table shows the main effect, which is essentially the change score for the referent group, the White non-Hispanic English-dominant group when with family. Because the fixed-effects model drops all scores that do not change within the person, all acculturation groups were by design, combined with the contexts, which did change from response signal to response signal to form interaction terms, such as low linguistically acculturated Hispanics with friends for example. The coefficients shown in Table 3 are the resulting change scores from these interactions.

Table 3 Fixed-effects results (n = 796 participants contributing 22,335 response signals)

In Table 3, the first coefficient within each context (friends, schoolwork, alone, and other), which is listed in bold letters at the beginning of each context section, corresponds to the referent group (i.e., White, non-Hispanic, English-dominant group, because the categories are mutually exclusive and exhaustive). This is followed by the interaction terms between each of the other acculturation groups and context. For example, a self-esteem coefficient of .25 (p < .001) for the “friends’” effect means that the White, non-Hispanic, English-dominant group experiences significantly more positive self-esteem when with friends than when with family. The interaction terms that follow within each context represent the moderation effect on self-esteem for each group within each context. So, for example, the coefficient within the friends context corresponding to Spanish-dominant Hispanics showed that self-esteem was moderated by −.27 (not significant [n.s.]). This coefficient, when added to the friend’s effect of .25, implies that being Spanish-dominant translates to −.03 effect on self-esteem, meaning that the less linguistically acculturated Hispanics experience self-esteem no differently with friends than with family. Likewise, going down the column to the schoolwork context main effect (also in bold), when engaged in schoolwork, the White, non-Hispanic, English-dominant group, relative to when with family (the other referent group), experience lower self-esteem when engaged in schoolwork, −.08 (p < .05). The more interesting coefficients are the interaction terms within each context and these will be discussed next.

Assuming that the referent group, by virtue of its much larger numerical representation in this data, is a good proxy for host culture (to which immigrants are assumed to aspire to acculturate), the main-effect score in this analysis can stand as a benchmark to which the different groups may be compared. It was expected that within the friends’ context, the two more linguistically acculturated groups would show a higher self-esteem with friends than with family by virtue of associated loss of familism with acculturation described in the literature. However, the most linguistically acculturated Hispanics in this study, like the less linguistically acculturated, show no significant differences between being with friends and being with family; self-esteem, −.10 (n.s.). They therefore still appear to have more positive self-esteem with friends (as do most adolescents), but this difference is not significant.

For the less linguistically acculturated, the Spanish-dominant group for all the contexts except schoolwork, the effect of self-esteem was moderated downward, −.27 (n.s.); in other words, their negative interaction effect scores suggest that the group is more inclined to higher self-esteem when with family than in any other context but schoolwork. These results imply that the self-esteem of the less linguistically acculturated Hispanics is not significantly different between times spent with family versus friends. This outcome is noteworthy, considering that the effect of familism is likely to be stronger for the less linguistically acculturated and, conceivably, that speaking Spanish most often may distance them from their friends and classmates if English is the dominant language in these contexts, as would be expected in most U.S. cities.

By contrast, the opposite pattern is observed for the somewhat linguistically acculturated—those who learned Spanish first, but as adolescents speak English most often. It is interesting that the somewhat linguistically acculturated have higher self-esteem when with friends than when with family; it is equally interesting that the gap in self-esteem between each of the other contexts (schoolwork, alone, or with others) and the family context is also large for this group, and in the same upward (positive) direction. Data showed an upwardly moderated effect on self-esteem (toward the positive) for the somewhat linguistically acculturated when with friends, creating a self-esteem gap between friends and family that is wider than that of any other linguistic acculturation group. In three of the four contexts, the somewhat linguistically acculturated Hispanics exhibit the widest gaps in self-esteem relative to when with family than any other group. In almost all cases, these effects are statistically significant at p < .05 or better.

It was expected that multiple language participants, having both ability to speak English and retention of native language, and by proxy culture, would afford multiple language speakers more versatility, enhancing their self-esteem across different contexts. However, the data reveal a surprising pattern. These data showed speakers of multiple languages to have lower self-esteem with the friends than with family context, −.41 (p < .10), in a pattern that is similar but more pronounced than that of the least linguistically acculturated Hispanic group.

In sum, this study showed that being with family was associated with higher self-esteem for participants at low or at high linguistic acculturation levels, but not for those in the middle of the linguistic acculturation process. Therefore, the idea that the strength of family weakens with linguistic acculturation (Sabogal et al. 1987) is not fully explained or reflected here. The findings of this study demonstrate that acculturation and context did moderate self-esteem, but this did not occur in the expected direction. Self-esteem, defined by changes in momentary affective states as participants move from one context to another, is moderated by the effect of linguistic acculturation and context. However, where it had been expected that self-esteem would decline with increased linguistic acculturation, it was found that Hispanic adolescents at low or high linguistic acculturation levels experience juxtaposed patterns of self-esteem with those in the middle of the acculturation process. For the least and the most linguistically acculturated, there is little difference in self-esteem when with family versus when with others. Being Hispanic Spanish-language monolingual and speaking more than one language moderated the effect of context on self-esteem in a direction favoring positive self-esteem with family. In other words, these groups are more likely to be happy with both family and friends. By contrast, those in the middle of the linguistic acculturation path, termed the “somewhat linguistically acculturated,” appear to have higher self-esteem with anyone other than family.

At the low and high ends of the linguistic acculturation spectrum—the high and the low acculturated more than the bilinguals, Hispanic adolescents are more likely to experience more person-context cultural congruence when with their parents/family. At the low end, those whose first language and current language preference was Spanish, are more than likely youth in monolingual families where (at least from a linguistic perspective) there is more congruence. Participants at the high end of the linguistic acculturation process (i.e., individuals who reported learning English first and but currently speak English most of the time) probably come from English monolingual families; again, there is language congruence, because for the participants to have learned English first, the parents had to have known how to speak English. In these instances, the low and high linguistic acculturation groups’ results showed a clear pattern of linguistic acculturation moderating the experience in favor of family rather than friends. For the middle group, the somewhat linguistically acculturated, who first learned Spanish but reported that currently English is spoken most, results show a marked preference for time with friends rather than with family, supporting the idea that cultural incongruence may be a factor in the family context.

Discussion

These findings contribute to the literature on the effect of linguistic acculturation and context on self-esteem. This study, by showing that linguistic acculturation and context moderated self-esteem, suggests that the complex relationships between the different contexts in which youth spend time, combined with their level of linguistic acculturation, affect their emotional well-being. However, although it had been hypothesized that the emotional distance between parents and children would have been most evident as acculturation increased, this was not the case. The patterns observed in this research were not as predicted by the loss of family cohesion, suspected to be at root of many findings in the immigrant health paradox research—where declines in well-being have been observed as acculturation increases. In a study such as this, where adolescent’s self-esteem was evaluated in different contexts relative to when with parents, it had been expected that youth at higher levels of linguistic acculturation would have lower self-esteem with their families than with their friends by virtue of expected declines in familism over time, as orientation to dominant culture and away from the culture of parents coincides with acculturation. However, the evidence supports a different idea—that it is the middle of the process that more clearly showed a pattern of low self-esteem with family than in any other context.

Specific concern for the middle of the process was evident in these findings. As compared to low or high linguistic acculturation, participants in the middle of the acculturation process experienced significantly higher self-esteem in any context but with their families. More research is needed to unpack this interesting finding. Time with family is important for socialization. Also, familism, especially for Hispanic families (Gonzales et al. 2004) has been identified as a protective factor for immigrant youth against depression (Gil-Rivas et al. 2003). The results of this study indicate that the parent-youth dynamics of Hispanics in the middle of the acculturation process needs to be better understood. The results support various explanations. For instance, as posited long ago by Park and Stonequist, it is possible that home life is very difficult for Hispanic youth mid-process. Alternatively, acculturating Hispanics in the middle of the process may be adapting to the aspect of U.S. culture in which familism gives way to individualism. Or, there may be tension at home due to the clash of cultures, assuming the parents are less acculturated. Unfortunately, this study could not address every aspect of the question. Clearly, a major concern is the finding that self-esteem was consistently higher in any context but with family. This may be associated with a parental inability to socialize children to the culture the parents are most familiar with and, lacking that ability, their children may be adopting peer-culture.

Of course, there are many possible alternative explanations to these findings and these are related to the study’s limitations. First, it may be that loss of family cohesion is but one of many factors that acculturating youth face. At school, the less acculturated who prefer to speak Spanish most often may not be much happier with family, but rather may feel lower self-esteem with classmates and friends by virtue of their lack of mastery of the English language. Likewise, their more linguistically acculturated Hispanic peers who, by virtue of speaking English, attempt to gain entry into social institutions of the dominant culture, yet may feel the ill-effects associated discrimination, alienation, or marginalization that affects many minority groups in the United States (Smokowski et al. 2007; Vega and Aguilar 1995).

Although this study employed rigorous analytical methods and drew on data with a large random sample containing repetitive data collection, considered to have high reliability, many assumptions were stated regarding the challenges involved in assessing the relationship between the variables. In practice, many other complexities are involved in the study of acculturation and language is but one variable. Furthermore, the quality of the interactions as participants go from one context to another is not known in its entirety. It may be that the presence of linguistic acculturation gaps are exacerbated by mixed immigration status within the family and gender/power issues among members, and minimized by such things as collegial communication and empathy among members. Furthermore, the language of the parents is not known, casting some doubt on the true effect of person-context congruence. Finally, before generalizing, one should remember that the context in which people live is always changing and the impact of macro-level factors can have an impact on the micro-level ones (Perez 2011).

Acculturation research has existed for nearly 100 years, yet there is still no solid cross-cultural measure for how to assess this process (Rudmin 2009). At the same time, there is a relatively recent proliferation of acculturation research (Rudmin 2009), and some recommend its inclusion in all well-being research (Lara et al. 2005). In truth, much of the prior acculturation research on well-being has been based on language, and despite a growing trend toward bicultural and multidimensional measures, other researchers continue to advocate for simple individual-level proxy measures for acculturation (Escobar and Vega 2000). Language acquisition captures a complex process that includes affective, cognitive, and behavioral components (Cuéllar et al. 1995), and has been shown to be influential in more complex acculturation measures (Rogler et al. 1991). Accurately capturing a complex and dynamic process such as acculturation with a single measure is likely to remain difficult considering the large number of multidirectional factors affecting the process. Although the findings of this research help to point out the important effect of context together with acculturation, it is clear that more research is needed to disentangle these important effects among the large and growing sector of immigrant children the United States.

Implications

These findings underscore the need to better understand the effects of linguistic acculturation on self-esteem and the practical utility of a fixed-effects model. Evidence from this study showed that context and language-use moderated self-esteem. Essentially, the participant in the middle had better self-esteem in any context but with family. Although the times have changed from the days when Park and Stonequist wrote about the marginal man, it is interesting that it was adolescents in the middle of the process—those whose first language was Spanish and whose second language was English—who were different from the other two acculturation levels with respect to their families. Clearly, more research is needed before we can determine the benefits—or drawbacks—of the relationship between their families of Hispanic youth and their self-esteem relative to other contexts in which they spend time.

Research suggests that Hispanics will continue to present varying levels of language use for years to come—probably retaining their native language and culture longer than other immigrant group. Thus, the implications of this study for policy, practice, and research are critical given the size and growth rate of this population, in addition to their continual population replenishment from Latin America (Jiménez 2008), and their transnational (Levitt 2003), circular (Massey 1987), and undocumented patterns of migration.

Research

This study also has implications for future research. The combination of a fixed-effects model with a unique data-collection method, ESM, afforded an unusual and beneficial research design that allowed for robust cross-cultural comparisons. Cross-cultural research is often challenged by the difficulty of comparing dissimilar people. Fixed-effects models have the advantage of being able to hold background effects constant so that cross-cultural comparisons are less prone to selection bias. A second implication for research, regarding the effects of linguistic acculturation on development, is that more understanding of the dynamic between immigrant youth at the different levels of linguistic acculturation and their families, is warranted. Another possibly fruitful path for future research is better understanding the effects of language on the brain or, more generally, on healthy development. The multiple-languages group was expected to be relatively indifferent about where they spent time. It was predicted that participants who spoke multiple languages would experience few or no self-esteem differences across the contexts in which they spend time, by virtue of having learned enough and having enough resources to cope in a variety of contexts. The literature on multicompetence, in particular, shows that the structures of the brain develop differently when individuals learn more than one language at the same time (Kim et al. 1997). These differences are hypothesized to lead to higher levels of cognition, in that multilingual people have access to additional linguistic knowledge not otherwise possible (Han 2004; Pavlenko 2003).

Finally, it is hoped that cross-cultural researchers who work with panel data containing multiple measures per person see the advantage of the fixed-effects model. The fixed-effects model as applied to the panel data (i.e., repetitive sampling technique) used in this study allowed us to control for background differences. This was advantageous to reduce bias. This model controls selection bias by way of the demeaning process, which assesses each person’s changes from his or her first recorded response so that each person serves as his or her own control, and only variables that change from response to response are included. Because each person serves as his or her own base of comparison, the variables that do not change over time are simply held constant and thus do not bias the results. In this way, observable variables that do not vary for a person (e.g., gender, grade, socioeconomic status) fall out of the model because nothing changes from response signal to response signal; in this way they are implicitly controlled. Unobservable variables are also held constant. This is useful for controlling such things as aspects of personality, which can predispose people to prefer some contexts or activities and to experience particular affect. For example, an extroverted person may have a low self-esteem when alone, whereas an introverted person may be much happier alone.

Practice

For practitioners working with young people learning a second language, it is critical to understand that important associated cultural differences to their linguistic differences in how they relate with family and friends and also how they affect self-esteem. Human services staff working with Hispanic youth may, first and foremost, want to assess the degree to which the family is experiencing linguistic acculturation gaps, and also to what extent these are merely a nuisance rather than causes of problematic behaviors. Linguistic acculturation in this study was treated as a moderator of self-esteem; in other words, there are more than likely additional familial and cultural issues that underlie problem behaviors. Language use, and more specifically, linguistic retention has been found to be protective factors bridging gaps between parents and youth and contributing to global self-worth (Birman 1998). For among more acculturated adolescents more understanding is needed to regarding the more problematic areas negatively affecting well-being—whether self-esteem, family conflict, or something else that is not fully understood as of yet.

Social workers may use the information found in this study and others like it to help bridge linguistic acculturation gaps between immigrant parents and acculturating offspring as has been found helpful by some practitioners working with troubled Hispanic adolescents (Szapocznik et al. 1986). Exploring bridges between youth and parental culture during adolescence can help facilitate parental socialization. Knowing that most immigrant youth progress from low to high linguistic acculturation, it may be worthwhile to explore how benefits of cultural retention (e.g., familismo, respeto) could be better preserved, so that those in the middle of the process do not feel estranged from their parental culture. Alternatively, children who do not feel positive self-esteem with parents may not be able to receive important enculturation and if they seek their friends out they may not benefit from the quality understanding that their parents are better able to transmit.

Policy

Regarding policy, there is currently no concerted federal effort to incorporate newcomers, linguistically or otherwise, into U.S. society (Jiménez 2007). Furthermore, the problematic adaptation of more acculturated Hispanic youth, as posited by evidence from the immigrant health paradox, suggest that macro-level interventions led by the U.S. government would be justified (Perez 2011). Addressing the needs of Hispanic youth at the age most crucial to the transition to adulthood can be seen as an investment.