Examining the influences of gender, race, ethnicity, and social capital on the subjective health of adolescents

ARTICLE IN PRESS

Journal of Adolescence Journal of Adolescence 32 (2009) 109–133 www.elsevier.com/locate/jado

Examining the influences of gender, race, ethnicity, and social capital on the subjective health of adolescents Gunnar Almgrena,, Maya Magaratib, Liz Mogfordc a School of Social Work, University of Washington, 4101 15th Avenue N.E. Seattle, WA 98105-6299, USA Department of Sociology, University of Washington, 4101 15th Avenue N.E. Seattle, WA 98105-6299, USA c Department of Sociology, Western Washington University, 516 High Street, Bellingham, Washington 98225, USA b

Abstract We investigate the factors that influence adolescent self-assessed health, based upon surveys conducted between 2000 and 2004 of high-school seniors in Washington State (N ¼ 6853). A large proportion of the sample (30%) was first and second generation immigrants from Asia, Latin America, and Eastern Europe. Findings include a robust negative effect of female gender on self-reported health that is largely unmodified by demographic, developmental, social capital, and parental support variables, gender differences in the covariates of self-reported health, and the tendency of male adolescents of Cambodian and Vietnamese origin to report lower levels of self-reported health despite controls for other health-related individual characteristics. Social capital dimensions such as positive school affiliation, social network cohesion, and a safe learning environment were found to covary with the self-reported health of adolescent females. r 2007 The Association for Professionals in Services for Adolescents. Published by Elsevier Ltd. All rights reserved. Keywords: Adolescent competence; Adolescent health; Assimilation; Gender; Immigrant health; Cultural assimilation; Self-reported health; Social capital; Social cohesion; Socio-economic gradient

Corresponding author.

E-mail address: [email protected] (G. Almgren). 0140-1971/$30.00 r 2007 The Association for Professionals in Services for Adolescents. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.adolescence.2007.11.003

ARTICLE IN PRESS 110

G. Almgren et al. / Journal of Adolescence 32 (2009) 109–133

Introduction and theoretical background A common and enduring feature of post-demographic transition societies is the linear relationship between individual socio-economic status (SES) and various measures of morbidity and mortality risk. Although some of this relationship can be explained through individual health behaviors and health-care access associated with SES, there is a significant component of individual health that is mediated by the social environment in complex ways that are independent of individual characteristics (Yen & Syme, 1999). For a variety of reasons, the period of adolescence affords an important glimpse of the complex interplay between individual and contextual influences on health throughout the life course (Goodman, 1999). In addition to being a particularly high-risk period for the initiation of lifelong tobacco use, unprotected sex, substance abuse, and violent death, transition through adolescence brings with it the foundations for future educational attainment. It is also during adolescence that the acquisition of social skills essential to the lifelong task of establishing and sustaining supportive social networks takes place. Such interpersonal networks are argued to mediate between various forms of social stress and health (Wilkinson, Kawachi, & Kennedy, 1998). Finally, unlike adulthood, adolescence offers the chance to observe the SES–health relationship with minimal influence of the ‘‘social drift’’ phenomenon; namely, the circumstance in which poor health is the determinant of social class via reduced earning power and educational attainment (Starfield, Riley, Witt, & Robertson, 2002). For all of these reasons, investigators from a wide array of disciplines and theoretical orientations have turned to adolescent health as a point of leverage for disentangling the SES–health gradient. In this paper, we seek to provide two important contributions to the literature on adolescent health. First, we investigate the magnitude of differences in self-reported health by gender, race, ethnicity, and immigrant background. Of particular interest is the extent to which the ‘‘gender gap’’ in adolescent health is common and of similar magnitude across all racial and ethnic groups represented in our diverse sample. Second, we identify a general model of adolescent health that considers the role of social capital. Our investigation is based upon the self-reported health of a pooled sample of four waves of high-school senior classes from 12 high schools in Washington State surveyed over the 2000–2004 period (N ¼ 6853). Aside from the high response rate (71% of the sampling frame) and a rich array of survey items that reflect both known and plausible correlates of adolescent health, nearly one-third (N ¼ 2086) of participants were from immigrant family households. The interpretation of self-reported health in adolescence The measure of health employed is the self-reported health of participants on a five-point scale. Self-reported health is a widely utilized measure of subjective health in public health and the social sciences. Self-reported health is highly correlated with more objective measures of health (Andresen, Catlin, Wyrwich, & Jackson-Thompson, 2003; Idler & Benyamini, 1997). Moreover, self-reported health has predictive effects on longevity independent of objective health status (Idler & Benyamini, 1997). Among adolescents, self-reported health also has a broader interpretation, extending to a more general sense of social competence, coherence, and wellbeing (Mechanic & Hansell, 1987). These psychological traits are highly correlated with subjective health complaints that are common to adolescents and associated with socio-economic inequalities (Goodman, 1999; Leveque, Humblet, Wilmet-Dramaix, & Lagasse, 2002; Mechanic

ARTICLE IN PRESS G. Almgren et al. / Journal of Adolescence 32 (2009) 109–133

111

& Hansell, 1987; Torsheim, Aaroe, & Wold, 2001). For these reasons, as well as the fact that adolescents generally have much lower levels of morbidity relative to adults, variations in the selfreported health of adolescents might provide important insights concerning the pre-adulthood emergence of the SES–health gradient. Established correlates of adolescent health We summarize the established correlates of adolescent health into demographic characteristics (gender, race, social class, and immigrant generation) and developmental characteristics (selfesteem, self-efficacy, and adolescent competency). Our use of the term ‘‘established’’ does not imply that there is a general consensus over either the theoretical role or the specific effects of a given correlate. Rather, we simply mean that there is well-established consensus that the correlate in question might have important effects upon adolescent health. We begin with gender. Although it is generally contended that women suffer higher rates of morbidity across an array of conditions over the life course despite their mortality advantage (Green & Pope, 1999), other recent investigations have revealed that gender differences in morbidity vary in direction and magnitude depending on the dimension of health being considered and the phase of the life course (Macintyre, Hunt, & Sweeting, 1996). It is also suggested that some structural effects on health (e.g. SES, family structure, and social capital) are stronger for women than men across both subjective and objective measures of health (Denton & Walters, 1999). However, some studies find that gender effects on health that are apparent in adulthood do not translate to adolescence (Goodman, 1999; Goodman et al., 1997; Hetland, Torsheim, & Aaro, 2002). Because there has been less research devoted to gender and class disparities in the general health of adolescents (Rahkonen & Lahelma, 1992), this paper seeks to identify both the magnitude of the gender gap in self-reported health and its prevalence across adolescents of diverse racial and ethnic origins. A second established demographic correlate of adolescent health is social class. In contrast to the adult and early childhood populations, where the SES gradient remains robust across a variety of measures of health and social class, in adolescence the manifestation of the SES gradient is more nuanced. The most recent multiple indicator study of SES in adolescence, conducted in the United Kingdom, finds little evidence of SES differentials where the measure considered is longstanding (chronic) illness. However, there is a pronounced SES differentiation in height and in self-reported health, as well as some evidence of a gender–SES interaction, depending upon the health indicator employed (West & Sweeting, 2004). Although another recent cross-national study finds that relative material deprivation operates at multiple levels as a predictor of the selfreported health of adolescents (Torsheim et al., 2004), there are also arguments and compelling evidence to suggest that behavioral norms associated with the health effects of social class have less influence during adolescence than during later stages of the life course. According to the ‘‘equalization’’ theory of adolescent health, the normative influences on health that arise from family and neighborhood class differences are held in abeyance and to some extent counteracted by the larger influences of the peer group and the prevailing youth culture (West, 1997). This juxtaposition of age and class allows us at a general level to identify the conditions under which a process of equalization in youth might occur. This would be so when one or more

ARTICLE IN PRESS 112

G. Almgren et al. / Journal of Adolescence 32 (2009) 109–133

influences associated with age (the school, the peer group or youth culture) cut across those associated with class (the family, home background, and neighbourhood), the net effect of which is to reduce, remove, or even reverse social class differences in a characteristic present in the earlier period of childhood (West, 1997 p. 836). In the United States, where there are significant disparities in access to health care by social class (Collins, Davis, Doty, & Ho, 2004), it may be that the SES gradient is more resilient in adolescence than in the UK and other Western democracies with universal access to health care. The most recent and comprehensive findings on this question are derived from a sample of 3309 adolescents aged 11–17 surveyed in public high schools in Baltimore, Western Maryland, and Maine between 1992 and 1999 (Starfield et al., 2002). Because the investigators employed a composite measure of social class that included both education and occupation, as well as a complex taxonomy of health indicators, their findings were arguably more sensitive to evidence of the SES gradient than other studies done outside the US. The central findings of Starfield et al. point to a strong SES gradient in subjective health and well-being, and in developmental indicators that are associated with successful transition to adulthood. However, it is also shown that there is an absence of a statistically significant SES gradient effect on other measures of health, most notably in acute disorders. The essential conclusion reached by the authors also stands as a fair appraisal of the state of the literature on the role of the SES gradient in adolescent health. They find that the presence of the gradient depends upon not only the dimensions of social class considered, but also the type of health outcomes observed (Starfield et al., p. 359). A third category of established covariates of self-reported health is race. The relationship between race and health in adolescence is less ambiguous in the sense that race is correlated with various measures of adolescent health, health behaviors, and access to health care. For example, the plethora of studies yielded from National Longitudinal Study of Adolescent Health (Add Health) data show that the distribution of somatic complaints (headache, stomachaches, and general malaise) among adolescents varies significantly by race (Udry, 2003), as do risk behaviors (Blum et al., 2000) and access to health care (Ford, Bearman, & Moody, 1999). On the other hand, the confounding of race with socio-economic status and a variety of latent structural effects has led to serious questions about the utility of race in and of itself as a useful construct to explain differences in health outcomes, except for the limited number of known diseases with well established race-based risks (e.g. cystic fibrosis, sickle-cell anemia). Like most other studies, we employ race as a known correlate of adolescent health without explicit assumptions as to the role of race or its effects. A fourth category of established correlates of adolescent health we include in our analysis concerns key dimensions of adolescent development: self-esteem, self-efficacy, and adolescent competency. To the extent that subjective health reflects overall psychological well-being, we anticipate covariance between self-esteem, self-efficacy and self-reported health. Indeed, this is the prevailing pattern in other research on adolescent health (Ge, Elder, Regnerus, & Cox, 2001; Grubbs et al., 1992; Haugland, Wold, Stevenson, Aaroe, & Woynarowska, 2001; Konu, Lintonen, & Rimpela, 2002; Mechanic & Hansell, 1987; Torsheim & Wold, 2001). Another dimension of adolescent development that appears critical to adolescent health is described as adolescent competence, that is, the adolescent’s ability to fulfill the various activities and role expectations of adolescence that are prescribed and mutually reinforced by parents and schools. These

ARTICLE IN PRESS G. Almgren et al. / Journal of Adolescence 32 (2009) 109–133

113

expectations include participation in athletic and non-athletic school activities, attending to homework, achieving acceptable grades, and earning money through part-time work. The conceptual pioneers of adolescent competence, Mechanic and Hansell (1987), found in a large survey sample of high-school students that adolescent competence (measured as participation in sports and other school activities, doing homework, and GPA achievement) makes a substantial contribution to self-reported health, independent of other aspects like psychological well-being and physical health. The fifth and final category of established correlates to adolescent health we consider is parental support. It has been shown that parental support, involvement, and monitoring collectively reduce the likelihood that adolescents will engage in a variety of behaviors that undermine health (Atkins, Oman, Vesely, Aspy, & McLeroy, 2002; Hawkins, Catalano, Kosterman, Abbott, & Hill, 1999; Li, Stanton, & Feigelman, 2000). Not surprisingly, the general state of adolescent health is also influenced by the quality of parental involvement and support, both through direct and indirect paths (Mellin, Neumark-Sztainer, Story, Ireland, & Resnick, 2002; Ransom & Fisher, 1995). Presumably, the availability of parental involvement and support will remain an important factor in adolescent health even where gender, race, SES, and other forms of social capital outside the family are considered. Social capital, social cohesion, and adolescent health Central in much of the thinking in social epidemiology is the role of social cohesion as a determinant attribute of the social environment throughout the life course. Although conceptually elusive and contested, a socially cohesive environment is argued to be one characterized by egalitarian relationships, mutual trust, and reciprocity (Wilkinson, 1999). In social environments marked by a high level of social cohesion, it is argued that individuals have access to social capital in the form of interpersonal networks characterized by reciprocity, trust, mutual aide, and beneficial collective action (Lochner, Kawachi, & Kennedy, 1999). In contrast, social environments that are characterized by extreme status inequalities, weak affiliations, and distrust are argued to have deleterious health effects both through the internalization of low status, weakened defenses against stress, and at the macro-level a weakened commitment to investment in human capital (Kaplan, Pamuk, Lynch, Cohen, & Balfour, 1996; Wilkinson et al., 1998). The literature on the linkages between social cohesion, social capital, and health can be described as intriguing and contentious. To its proponents (Kaplan et al., 1996; Kawachi, Kennedy, Lochner, & Prothrow-Stith, 1997; Wilkinson, 1999), the theoretical and empirical linkages between social capital, social cohesion, and various measures of income inequality are central to the problem of social structure and health. To its critics, this line of theory is a conceptually muddy distraction from the more compelling and empirically justified linkage between class structure and absolute material deprivation (Finch, 2003; Forbes & Wainwright, 2001; Muntaner & Lynch, 1999; Muntaner et al., 2002). Clearly, one way of sorting out this controversy is to take advantage of individual-level studies of health outcomes that have measures of both social class and access to social capital at various points throughout the life course, including studies of adolescent health. Although there are many studies that link the etiologies and management of specific adolescent health risks to various dimensions of the social environment, there are relatively few studies that explicitly examine the role of social cohesion and

ARTICLE IN PRESS 114

G. Almgren et al. / Journal of Adolescence 32 (2009) 109–133

social capital in adolescent health (Gold, Kennedy, Connell, & Kawachi, 2002; Konu et al., 2002; Torsheim & Wold, 2001). Immigration, cultural assimilation, and adolescent health A significant and increasing proportion of the US adolescent population is comprised of immigrants or children born to immigrant families (Malone, Baluja, Costanzo, & Davis, 2003). From an intuitive perspective, it seems that negotiating the dual processes of cultural assimilation and transition from adolescence to adulthood would result in various forms of psychological and physical distress for adolescents from immigrant families. There is a body of research suggesting that children born to refugee families, a substantial minority of the sample in this study, may express the sequela of direct and indirect trauma in somatic terms many years after the migration experience (Rousseau, Drapeau, & Platt, 1999; Sack et al., 1993, 1994). It is also possible that children from some immigrant families may express poorer levels of health as a reflection of cultural beliefs about the meaning of good health (Uehara, 2001). There is another theoretical thread that speaks of the role of assimilation dissonance in adolescent health. Briefly stated, it is the idea that adolescents who perform the duty of language and cultural translator for their non-English speaking parents confront a distressing conflict between their subordinate role as a child and their role as family mediator with the dominant culture (Portes & Hao, 2002). It is speculated that this role conflict, compounded with the immigrant adolescent experience of daily living in two different worlds, will express itself in somatic terms (Portes & Hao, 2002). Another source of somatic distress for many immigrant adolescents is parental emphasis on academic achievement, despite obstacles of language and dominant group culture. Among some Asian immigrant groups, there is a paradox involving anxiety and low self-esteem despite high academic achievement that is attributed to an immigrant ethos that places adolescents under extreme pressure to excel academically (Bankston & Zhou, 2002). Finally, it is also noted that various studies of immigrant populations suggest assimilation into American culture degrades immigrant health, although the evidence is difficult to evaluate given the lack of consistent theoretical specificity and different measures of health (Salanta & Lauderdale, 2003).

Research questions The first major question addressed in this study concerns whether and to what extent there are differences in subjective health by gender, race, ethnicity, and immigrant background. A derivative question pertains to evidence that the magnitude of a gender gap in self-reported health differs racial and ethnic groups. The second major question pertains to the relationship between the adolescent social capital and self-reported health. Specifically, we ask whether there is an explanatory contribution of social capital to the self-reported health of adolescents that adds to the variance explained by demographic and developmental covariates.

ARTICLE IN PRESS G. Almgren et al. / Journal of Adolescence 32 (2009) 109–133

115

Methods Participants The data employed in our sample were drawn from the pooling of four survey waves of senior class cohorts from 12 high schools in Washington State, covering the period between 2000 and 2004 (N ¼ 6853). Nine of the 12 high schools participating in the study were public, two were parochial and one was private. It should be noted that only five public high schools participated in all four waves of the study, with the balance of the sample including students from the seven additional high schools in the 2003 and 2004 cohort years. Because the survey waves were concentrated over a 5-year period (2000, 2002, 2003, and 2004) and school context variables are included in the analysis, we do not anticipate that these aspects of the sampling design will significantly bias the results. Study participants were recruited from the class lists of students identified by each participating high school as completing their final semester of high school. All participants were surveyed within 2 months of commencement, after obtaining either direct consent or (in the case of students under age 18) both direct consent and surrogate consent from a custodial parent/guardian. Participation was voluntary, with the nominal incentive of a movie pass upon survey completion. The overall response rate (surveys completed/all enrolled seniors) was 71%. Of students completing the survey, 55% were females and 56% non-Hispanic white. The principal racial/ ethnic categories other than non-Hispanic white were Asian (17%), African American (14%) and Hispanic (9%). Notably, 30% of the students surveyed were either first or second generation immigrants. Measures The independent variable measures employed in our analysis fall into four broad categories: demographic background characteristics, developmental characteristics, measures of social capital and measures of parental support. Demographic background characteristics The demographic background characteristics utilized in the analysis include gender, race/ ethnicity, parental educational attainment, family intactness, immigration generation and primary language spoken at home. Gender was identified based upon the survey responses and consistency with the gender recorded on school records. There were multiple items in our survey pertaining to racial and ethnic background, and the category assigned to each respondent was based both upon the internal consistency of item responses and the primary racial/ethnic identity preference of the respondent. Eight racial/ethnic groupings are employed in the analysis: non-Hispanic white, Hispanic, African American, AIHPI (American Indian/Hawaiian/Pacific Islander), Filipino/other S.E. Asian, East Asian (Chinese, Japanese or other East Asian), Vietnamese and Cambodian. Of the eight categories mentioned, only the AIHPI category is regarded as a residual category without particular theoretical significance. Because the research agreements with the cooperating school districts precluded the inclusion of items pertaining to income, the demographic measures

ARTICLE IN PRESS 116

G. Almgren et al. / Journal of Adolescence 32 (2009) 109–133

employed for social class background are the educational attainment of participants’ mothers and fathers measured on a three-item ordinal scale (high school or less, some college, completion of a 4-year college degree or higher). The final demographic indicator, ‘‘family intactness,’’ is a dichotomous measure of family structure that it is assigned a value of (1) if both biological and adoptive parents co-reside with the participant and (0) if otherwise. Developmental characteristics The measures pertaining to developmental characteristics include items from two widely used subscales that were included in the survey instrument design (self-efficacy and locus of control) and a third scale that is a composite of survey latent constructs and individual survey items that represent distinct domains of adolescent competency. The self-efficacy scale is comprised of seven items that are scored on a four-point Likert scale, with the means of each item summed to a total score that ranged from 1 to 7. The self-efficacy scale items yielded a Cronbach’s alpha of 0.73. The locus of control scale is similarly constructed from six items that yield a score range from 1 to 6 and a Cronbach’s alpha of 0.63. Based on earlier work by Mechanic and Hansell (1987), we conceptualize adolescent competency as the ability to engage successfully in the various normatively prescribed activities of adolescents: being a successful learner, participating in extracurricular activities like clubs and student athletics, being successful in peer social relations, and earning money through paid employment. Thus, our measure of adolescent competency is constructed from the summed standard scores of cumulative high school GPA, weekly hours spent doing homework, hours of participation in extracurricular activities and a factor score from items loading on a single dimension we term ‘‘social competency.’’ This method produced an adolescent competency scale with normally distributed values that ranged from a lowest competency of 10.62 to a highest competency value of 9.42. The two other measures of developmental characteristics are the body mass index and the number of days absent in school per month, both expressed as standard scores. Social capital We employ four measures of social capital that we believe represent core aspects of the concept. Positive school affiliation and a safe learning environment pertain to the ‘‘local opportunity structure’’ available to the adolescent (Baum & Ziersch, 2003). The two other measures, social network cohesion and parents having knowledge of friends’ plans, concern the extent to which there is connectedness and mutual concern (social cohesion) among the children and adults that comprise the adolescent’s social world (Baum & Ziersch, 2003; Wilkinson, 1999). The particular measures employed are factor scores produced from an exploratory factor analysis of 14 survey items that on face examination appeared strongly relevant to various dimensions of social capital. For example, five Likert scale items that load on the dimension we employ as our measure of positive school affiliation include such positive statements about the school as ‘‘the school provides a caring environment,’’ ‘‘the teaching is good,’’ and ‘‘teachers are interested in students.’’ The factor procedures utilized to derive each of the four social capital measures are described in more detail at the beginning of the Results section.

ARTICLE IN PRESS G. Almgren et al. / Journal of Adolescence 32 (2009) 109–133

117

Parental support measures Two parental support measures that pertain to parents’ participation in the academic and social life of adolescents were derived from an exploratory factor analysis of nine items in the survey that referred to parent/adolescent interactions. The first measure, which we name talks to parents, is the factor score representing the factor loadings of six items broadly pertaining to the frequency and quality of the communication between the parent(s) and the adolescent. The first three items refer to the frequency with which the student reports discussions with parents that pertain to school activities, going to college, and receiving guidance and support. The other three items are Likert scale measures that report the extent to which the student feels she/he receives high levels of love and support from his/her parents, whether the student feels she/he has frequent in-depth conversations with parents, and whether the adolescent feels parents are usually disappointed with what she/he does. The second parent support measure focuses more narrowly on academic support and is derived from three items pertaining to the amount of involvement parents have in homework and whether parents limit their children’s time with friends on school nights. We have named this factor score homework support. The dependent variable (self-reported health) The dependent variable is the value on a five-point (1–5) scale of self-reported health that has been widely utilized in both adolescent and adult populations as a general measure of subjective health. Participants were asked to rate their health as either poor, fair, good, very good, or excellent in response to the question ‘‘In general, how is your health?’’ As previously mentioned, although this measure of selfreported health is highly correlated with more objective measures of health in adult populations (Andresen et al., 2003; Idler & Benyamini, 1997), among adolescents subjective health is argued to extend to a more general sense of social competence, coherence, and well-being (Mechanic & Hansell, 1987).

Procedures All measures utilized in this study were derived from a common survey instrument completed by senior class students from 12 urban Washington State high schools in four survey waves over a 5-year period (2000, 2002, 2003, and 2004). The surveys were offered and completed within 6 weeks of the end of the final semester of high school, on site at the high school during the normal school day. Although every attempt was made to contact and offer survey participation to all students deemed eligible for graduation from high school in the coming weeks, the 71% of students who participated are likely to under-represent students who have high rates of truancy or are more likely to take advanced placement classes outside the high school. Thus, we conceptualize our high-school senior class participants as fairly representative of the typical high school senior rather than the universe of all high-school seniors.

Results The first results reported pertain to the exploratory principal components factor analysis procedures employed to derive the social capital and parental support measures, each expressed as

ARTICLE IN PRESS 118

G. Almgren et al. / Journal of Adolescence 32 (2009) 109–133

a factor score selected by the regression method. The four social capital measures that were derived using PC factor analysis were based on a varimax rotation procedure with Kaiser normalization, applied to 14 items pertaining to the participant’s impression of her/his school and social connections among friends and parents of friends. The criterion applied to the identification of any orthogonal factor was a minimum eigenvalue of 1.0 and item selection was based on a minimum factor loading minimum absolute values of 0.40. Thirteen items were ultimately retained that correlated with four orthogonal factors. The first factor, which we have named positive school affiliation, correlates with five Likert scale items pertaining to positive sentiments about the school context (caring environment, teaching is good, teachers are interested in students, students are graded fairly, discipline is fair). The Cronbach’s alpha for the items comprising positive school affiliation is 0.80. The second social capital factor, network cohesion, correlates with three Likert scale items: whether the participant’s parents know her/his friends parents, whether the participant’s parents know her/his friends, and whether the participant perceives encouragement from friends to do well. The alpha reliability coefficient for the network cohesion items is 0.57. The third social capital measure, safe learning environment, also correlates with three Likert scale items: whether the participant feels safe at school, the extent to which the school environment is disruptive to learning, and whether fights occur between different racial/ ethnic groups (Cronbach’s alpha ¼ 0.86). The fourth social capital factor, parents have knowledge of friends plans, is correlated with two items: the extent that the participant believes she/he has knowledge of friend’s plans (scaled 1–6) and the number of other participants in the sample that have the reference participant listed as a ‘‘best friend.’’ The Cronbach’s alpha for the items comprising this social capital measure is 0.28. The parental support measures were derived from nine survey items using the identical PC factor analysis procedures. As previously mentioned, the first of the parent support measures, talks with parents, is correlated with six items pertaining to how often the student discusses school and personal topics with parents (Cronbach’s alpha ¼ 0.81). The second parent support measure, homework support, is correlated with three items that pertain to parental assistance with homework and parents’ limiting activities with friends in school nights (Cronbach’s alpha ¼ 0.56). A more detailed presentation of the PC factor analysis results is provided in the appendix. Differences in self-reported health by demographic background characteristics The first research question, concerning the extent to which there are large differences in selfreported health among the adolescent sample by gender, race/ethnicity and immigration background, is largely addressed by the descriptive statistics shown in Table 1 and Fig. 1. As shown in Table 1, there is a gender gap in self-reported health that is large and quite consistent with that observed in the national Add Health sample. That is, boys in the 17–18 age range are much more likely to report themselves in excellent health than girls (t-test po0.001).1 As shown in Fig. 1, there also appears to be a significant interaction between race/ethnicity and gender. Male adolescents have an advantage in self-reported health across all ethnic/racial groups (except for 1

Both a t-test for gender differences in means of the self-reported health scale and a t-test for gender differences in the proportion reporting their health as ‘‘excellent’’ were significant at the po0.001 level. Full results available from authors.

ARTICLE IN PRESS G. Almgren et al. / Journal of Adolescence 32 (2009) 109–133

119

Table 1 Comparison of self-reported health between the add health and the beyond high school surveys Poor

Add health (%) BHS (%)

Fair

Good

V. good

Excellent

Total

Ma

Fa

M

F

M

F

M

F

M

F

M

F

2 1

0 1

6 7

8 10

19 21

31 30

36 32

38 35

37 38

23 23

100 100

100 100

Note on data sources: (1) Add health wave I in-school sample and (2) University of Washington Beyond High School (BHS). a M ¼ male, F ¼ female.

4.2 Male Mean subjective health

4.0

Female

3.8

3.6

3.4

3.2 te

hi

W

k

ac

Bl

ic

an

sp Hi

n ica er ian m A Ind

o

n

in

lip

Fi

e

es

sia

Ea

A st

Vi

et

m na

an

di

bo

m

Ca

Race/ethinicity

Fig. 1. Mean health students reported in the survey, by race/ethnicity and gender.

S.E. Asians), but the size of the gender gap clearly fluctuates across racial/ethnic groups. For example, the gender gap appears nearly closed among S.E. Asians (Vietnamese and Cambodians), fairly large for Whites and Filipinos, and even larger for all other racial/ethnic groups. It also appears that the magnitude of the gender gap by racial/ethnic group is principally a function of differences in self-reported health among girls of different racial/ethnic background rather than differences among boys. With the notable exception of the S.E. Asian adolescents in the sample, adolescent boys across all racial/ethnic groups appear similarly disposed to rating their health as very good or better. On the other hand, adolescent girls from different racial and ethnic backgrounds appear more likely to show differences in self-reported health. The only adolescents that fail to display the appearance of a significant gender gap, those who are of S.E. Asian origin, have nearly identical low levels of self-reported health. The American

ARTICLE IN PRESS 120

G. Almgren et al. / Journal of Adolescence 32 (2009) 109–133

Indian, Hawaiian and Pacific Islander (AIPI) adolescent girls appear to be similarly disadvantaged as S.E. Asian with respect to self-reported health. However, unlike the S.E. Asian adolescents, the gender gap among the AIPI adolescents is significantly large. As shown in Table 2, in addition to gender and race/ethnicity, immigrant generation, nonEnglish language use in the home, parental education level, and family intactness also appear to be associated with differences in self-reported health. Bivariate results suggest that adolescents from more recent immigrant households have somewhat lower levels of self-reported health (see means for immigrant generation and non-English language use in the home, Table 2), and that higher levels of parental education and family intactness favor self-reported health. These results may reflect effects of social class and/or effects related to processes of assimilation or acculturation—questions to be taken up in the results from multivariate analysis. We first present the results of bivariate correlations between self-reported health and measures of developmental, social capital and parental support. Developmental and social capital contributions to adolescent health As shown in Table 3, all of the developmental and social capital variables correlate with selfreported health, with the exception of the parental support variable pertaining to assistance with homework. However, we elected to retain this variable in the multivariate analysis because it is correlated with other social capital variables and may contribute to a suppressor effect. Of the developmental variables, adolescent competency, self-esteem, and locus of control are the most highly correlated with self-reported health. Although statistically significant, none of the social capital variables is strongly correlated with self-reported health. However, it is interesting to note that network cohesion is modestly correlated with some of the developmental measures that do appear important to self-reported health. In order to discern whether there remains a distinct contribution of social capital to adolescent health apart from demographic background and developmental factors, we show the results of a series of regression models that add the previously described measures for adolescent development, social capital, and parental support to the demographic background variables. Because the gender differences in self-reported health are so dramatic, we have stratified this segment of the analysis by gender. Results of multivariate analysis In order to assess the relative effects of the demographic background, developmental attributes and social capital on self-reported health and entertain the possibility that ethnic and racial differences are largely mediated by these covariates, we test a series of ordinary least square (OLS) models. We begin with the demographic variables and then advance through models that introduce the developmental, social capital, and parental support variables via the block entry method. The first series of models on Table 4 pertain to the sub-sample of adolescent females (N ¼ 3748) and the second series of models on Table 5 pertain to the sub-sample of adolescent males (N ¼ 3103). The first model on Table 4 shows that while most racial/ethnic groups tend to report lower levels of health than whites, the racial/ethnic effect is itself quite trivial (adjusted R2o0.01). Somewhat contrary to our expectations, the second model shows that immigration

ARTICLE IN PRESS G. Almgren et al. / Journal of Adolescence 32 (2009) 109–133

121

Table 2 Mean self-assessed health by demographic variables, health ranges from low (1) to high (5) N

Mean

Std. deviation

t-Value

F-value

3749 3104 6853

3.69 3.99 3.82

0.98 1.00 1.00

12.525***

–

3838 948 591 300

3.90 3.77 3.76 3.68

0.96 1.02 1.06 1.04

285 498 220 173 6853

3.81 3.78 3.48 3.45 3.82

0.96 1.00 1.06 1.06 1.00

–

12.284***

Immigrant generation First generation Second generation Third generation or higher Total

964 1122 4767 6853

3.68 3.80 3.86 3.82

1.01 0.98 1.00 1.00

–

14.428***

Non-English language spoken at home Yes No Total

1648 5205 6853

3.72 3.86 3.82

1.05 0.97 0.99

4.652***

–

Mother’s education High school or less Some college College or higher Total

2400 2369 1774 6543

3.69 3.88 3.97 3.84

1.01 0.98 0.95 0.99

–

46.557**

Father’s education High school or less Some college College or higher Total

2179 2064 2026 6269

3.69 3.82 4.02 3.84

1.03 0.98 0.93 0.99

–

61.169***

Family intactness Both parents living together Parents not living together Total

2179 2026 6269

3.69 4.02 3.84

1.03 0.93 0.99

7.067***

–

Gender Female Male Total Race White African American Hispanic American Indian, Hawaii, Pacific Islander Filipino other S.E. Asian Chinese, Japanese other East Asian Vietnamese Cambodian Total

*po0.05, **po0.01, ***po0.001.

122

Table 3 The Pearson correlation matrix: health (dependent variable) with developmental, social capital, and parent support measures A Health

E F G H I J SelfLocus of School Network School Parents esteem control affiliation cohesion environment know frd’s plan

K Student talk to parents

1 0.05*** 0.02 0.06*** 0.09*** 0.02 0.05*** 0.04**

1 0.04*** 0.03** 0.02 0.01 0.04**

0.09*** 0.07***

0.11*** 0.04** 0.03* 0.01

N ¼ 6.853.***po0.001, **0.001op40.01, *0.01op40.05.

1 0.55*** 0.14 0.31*** 0.18***

1 0.11*** 0.22*** 0.19***

1 0.00 0.00

1 0.00

1

0.12*** 0.01

0.00

0.00

1

0.15*** 0.01

0.07*** 1 0.00 0.00

0.52*** 0.43*** 0.00 0.03**

0.19*** 0.44*** 0.07*** 0.18***

ARTICLE IN PRESS

A In general, how is 1 your health? B Adolescent competency 0.24*** 1 scale 0.04*** 0.02 C Standardized absent in class D Standardized BMI 0.12*** 0.11*** E Self-esteem 0.33*** 0.41*** F Locus of control 0.26*** 0.41*** G Positive school affiliation 0.10*** 0.14*** H Network cohesion 0.14*** 0.23*** I Safe learning school 0.12*** 0.15*** environment 0.06*** 0.26*** J Parents’ knowledge of friend’s plans K Talks with parents 0.23*** 0.34*** L Parents help with 0.02 0.09 homework, etc.

D BMI

G. Almgren et al. / Journal of Adolescence 32 (2009) 109–133

B C Adol. Absent competency in class

ARTICLE IN PRESS G. Almgren et al. / Journal of Adolescence 32 (2009) 109–133

123

background and language use among adolescent females have no independent effects on selfreported health and mediate very little of race/ethnicity background. Clearly however, the family background effects introduced in model three (parental education and family intactness) are relevant and for most racial/ethnic groups fully mediate the race/ethnicity effect. The most significant advance in explained variation occurs when measures pertaining to adolescent development are introduced. That is, the adjusted R2 increases from a trivial 0.037 to a more substantial 0.155. Interestingly, it is at this point that Vietnamese ancestry re-emerges as statistically significant. In fact, it remains the only racial/ethnic category that retains statistical significance in the final models. The final two blocks of variables, those representing social capital effects and parental support effects, introduce only a slight improvement in explained variation. The social capital effects shown, while interesting, appear too weak to sort out the relative effects of different dimensions of social capital. As Table 5 shows, while the relative relationships between demographic and developmental variables and self-reported health are about the same across gender in terms of explained variation, the detrimental effects of Vietnamese and Cambodian ancestry appear more persistent for males than for females. That is, for males of Vietnamese and Cambodian ancestry their disadvantages in self-reported health are only very modestly mediated by family background effects and individual developmental factors (see the coefficients for the first four models). Perhaps more remarkably, the male Vietnamese and Cambodian disadvantages in self-reported health are essentially unaffected by immigrant generation and household language assimilation (see the coefficients for the first and second models). In comparing the graphic results shown previously from Fig. 1 with the OLS coefficients shown in Table 5, a picture emerges of dramatically lower levels of self-reported health for males from S.E. Asia that are substantially unmediated by individual differences in socio-economic status (as measured by parental education) and differences in developmental attributes. There are two other interesting gender differences that emerge from a comparison of the results shown in Tables 4 and 5. First, it appears that the effects of individual developmental characteristics on self-reported health are somewhat more pronounced among the adolescent males in the sample than adolescent females. Second, among males, even the very slight social capital and parental support effects observed for females are absent. Notably, among females, the social capital dimensions that appear at least somewhat relevant to self-reported health pertain to the social context of the school (positive school affiliation, social network cohesion, and a safe school learning environment).

Discussion The analysis reported focused on two principal questions, one pertaining to the magnitude of differences in the self-reported health of adolescents by gender, race/ethnicity and immigration background and the other considering whether there is a distinct contribution of social capital to adolescent health apart from demographic background and developmental factors. Clearly, among the adolescents of this sample there are large gender differences in self-reported health that are similar in magnitude to those found in the national sample reported in the Add Health study (Udry, 2003). For a variety of reasons that go beyond the data and limited purposes of this study,

124

Table 4 OSL regression of subjective health self-reported by female adolescents Model 1

Model 2

Model 3

Model 4

Model 5

Model 6

Unstd.

Unstd.

Unstd.

Unstd.

Unstd.

Unstd.

Constant

s.e.

B

s.e.

B

s.e.

1.78***

0.14

B

s.e. 1.93***

0.15

0.29*** 0.05 0.31*** 0.33** 0.14** 0.14* 0.16**

0.08 0.08 0.10 0.11 0.05 0.06 0.06

0.29*** 0.04 0.26* 0.30** 0.15** 0.14 0.15*

0.08 0.09 0.11 0.12 0.05 0.07 0.06

0.21** 0.05 0.19 0.14 0.06 0.14 0.06

0.08 0.09 0.11 0.12 0.05 0.07 0.06

0.12 0.01 0.24* 0.10 0.06 0.13* 0.03

0.08 0.08 0.10 0.11 0.05 0.07 0.06

0.12 0.02 0.25* 0.10 0.04 0.12 0.03

0.08 0.08 0.10 0.11 0.05 0.07 0.06

0.11 0.01 0.23* 0.08 0.04 0.12 0.03

0.08 0.08 0.10 0.11 0.05 0.07 0.06

0.05

0.07

0.05

0.06

0.02

0.06

0.02

0.06

0.02

0.05

0.05 0.02

0.05 0.05

0.04 0.00

0.05 0.05

0.02 0.05

0.05 0.05

0.02 0.04

0.05 0.05

0.02 0.05

0.05 0.05

Mother’s education (reference ¼ high school or less) Some college College or higher education

0.12** 0.16**

0.04 0.05

0.06 0.07

0.04 0.05

0.07 0.06

0.04 0.05

0.07 0.06

0.04 0.05

Father’s education (reference ¼ high school or less) Some college College or higher education

0.09* 0.23***

0.04 0.05

0.04 0.16***

0.04 0.04

0.04 0.14**

0.04 0.04

0.03 0.13**

0.04 0.04

0.16***

0.03

0.10**

0.03

0.08**

0.03

0.08*

0.03

0.04** 0.03 0.12** 0.38** 0.24**

0.01 0.01 0.02 0.04 0.04

0.03** 0.03* 0.12** 0.34** 0.23**

0.01 0.01 0.02 0.04 0.04

0.03** 0.04** 0.12** 0.31** 0.21**

0.01 0.01 0.02 0.04 0.05

ARTICLE IN PRESS

0.02

Adolescent development Adolescent competency scale Absent in school ZBMI standardized BMI Self-esteem Locus of control

0.14

s.e.

3.77***

Family intactness (reference ¼ not together) Both parents living together

1.58***

B

0.02

Immigration and language use First generation (reference ¼ 3rd generation) Second generation Non-English spoken at home (reference ¼ no)

0.04

s.e.

3.77***

Race/ethnicity (reference ¼ white) AIPI Filipino/S.E. Asians Vietnamese Cambodian African American Chinese including East Asians Hispanic

3.47***

B

G. Almgren et al. / Journal of Adolescence 32 (2009) 109–133

B

Social capital Positive school affiliation Social network cohesion Safe school learning environment Parents have knowledge of friends’ plans

0.07** 0.06** 0.04* 0.00

0.02 0.02 0.02 0.02

Parental support Talks to parents Parents help with homework Model 1 ¼ 0.009 po0.001

Model 2 ¼ 0.009 NS

Model 3 ¼ 0.037 po0.001

***Significant at or less than 0.001, **significant at 0.01, *significant at 0.05. NS ¼ not significant.

Model 4 ¼ 0.155 po0.001

Model 5 ¼ 0.163 po0.001

0.02 0.02 0.02 0.02

0.06** 0.00

0.02 0.02

Model 6 ¼ 0.164 po0.05

ARTICLE IN PRESS

G. Almgren et al. / Journal of Adolescence 32 (2009) 109–133

Adjusted R2 Change in R2 F-test N ¼ 3748

0.07** 0.04* 0.04* 0.00

125

126

Table 5 OSL regression of subjective health self-reported by male adolescents Model 2

Model 3

Model 4

Model 5

Model 6

Unstd.

Unstd.

Unstd.

Unstd.

Unstd.

Unstd.

B Constant

Immigration and language use First generation (reference ¼ 3rd generation) Second generation Non-English spoken at home (reference ¼ no)

B

s.e.

B

s.e.

B

s.e.

B

s.e.

B

s.e.

4.06**

0.02

4.06**

0.02

3.86**

0.04

2.00**

0.15

2.12**

0.15

2.13**

0.16

0.14 0.13 0.56** 0.59** 0.08 0.09 0.10

0.09 0.09 0.10 0.11 0.06 0.07 0.07

0.14 0.12 0.54** -0.58** 0.08 0.08 0.10

0.09 0.10 0.11 0.12 0.06 0.08 0.07

0.11 0.09 0.48** 0.47** 0.02 0.07 0.04

0.09 0.10 0.11 0.12 0.06 0.08 0.07

0.05 0.08 0.40** 0.35** 0.05 0.01 0.03

0.08 0.09 0.10 0.11 0.05 0.07 0.07

0.05 0.09 0.40** 0.35** 0.05 0.01 0.04

0.08 0.09 0.10 0.11 0.05 0.07 0.07

0.05 0.09 0.40** 0.35** 0.05 0.01 0.04

0.08 0.09 0.10 0.11 0.05 0.07 0.07

0.06 0.02 0.03

0.07 0.06 0.06

0.04 0.03 0.03

0.07 0.07 0.06

0.05 0.03 0.05

0.07 0.05 0.05

0.05 0.03 0.05

0.07 0.05 0.05

0.05 0.03 0.06

0.07 0.05 0.05

Mother’s education (reference ¼ high school or less) Some college College or higher education

0.14** 0.06

0.04 0.05

0.08 0.01

0.04 0.05

0.07 0.01

0.04 0.05

0.07 0.02

0.04 0.05

Father’s education (reference ¼ high school or less) Some college College or higher education

0.03 0.18**

0.05 0.05

0.02 0.10*

0.04 0.05

0.02 0.09*

0.04 0.05

0.02 0.09*

0.04 0.05

Family intactness (reference ¼ not together) Both parents living together

0.10**

0.04

0.03

0.04

0.03

0.04

0.03

0.04

0.05** 0.03* 0.12** 0.45** 0.19**

0.01 0.02 0.02 0.04 0.05

0.05** 0.03* 0.12** 0.43** 0.18**

0.01 0.02 0.02 0.04 0.05

0.05** 0.03* 0.12** 0.42** 0.18**

0.01 0.02 0.02 0.04 0.05

Adolescent development Adolescent competency scale Absent in school ZBMI standardized BMI Self-esteem Locus of control

ARTICLE IN PRESS

Race/ethnicity (reference ¼ white) AIHPI Filipino/S.E. Asians Vietnamese Cambodian African American Chinese including East Asians Hispanic

s.e.

G. Almgren et al. / Journal of Adolescence 32 (2009) 109–133

Model 1

Social capital Positive school affiliation Social network cohesion Safe school learning environment Parents have knowledge of friends’ plans

0.03 0.04* 0.03 0.01

Parental support Talks to parents Parents help with homework

0.03 0.04 0.03 0.02

0.02 0.02 0.02 0.02

0.01 0.01

0.02 0.02

Model 1 ¼ 0.017 Model 2 ¼ 0.017 Model 3 ¼ 0.029 Model 4 ¼ 0.174 Model 5 ¼ 0.175 Model 6 ¼ 0.175 po0.001 NS po0.001 po0.001 NS NS

**Significant at 0.01, *significant at 0.05. NS ¼ not significant.

ARTICLE IN PRESS

G. Almgren et al. / Journal of Adolescence 32 (2009) 109–133

Adjusted R2 Change in R2 F-test N ¼ 3103

0.02 0.02 0.02 0.02

127

ARTICLE IN PRESS 128

G. Almgren et al. / Journal of Adolescence 32 (2009) 109–133

it seems that adolescent females report lower levels of adolescent health than males. Indeed, we found in more detailed analysis that the gender effect on self-reported health persists and even grows in magnitude as other covariates for developmental factors were entered into the model.2 Although the lower levels of self-reported health among adolescent females is also consistent with the morbidity disadvantage of women throughout much of the life course, the reasons as to why remain elusive (Denton & Walters, 1999). The significance of these particular findings is that the gender gap in health (at least in the subjective domain) is remarkably salient in adolescence and is prevalent across most racial/ethnic groups sampled. The sole exception, S.E. Asians, presents another puzzle pertaining to the role of race in self-reported health. For most racial and ethnic groups, it was shown that both among males and females the race/ ethnicity effect disappeared when family background variables were introduced, suggesting that the differences in self-reported health among adolescents of different ethnic/racial groups are largely mediated by factors pertaining to social class and family stability. However, it was shown that there is no gender gap in health among adolescents of S.E. Asian origin, with both males and females reporting nearly identical levels of sub-optimal health (see Fig. 1). While the gap between the selfreported health of S.E. Asian female adolescents and females from other groups is significantly mediated by family background variables, for S.E. Asian males their self-reported health disadvantage remains robust. The question of course is why. Although the most intuitive explanation is cultural, it should be kept in mind that Cambodians and Vietnamese have distinct languages and cultures, with different appraisals of the role of individual agency in well-being and the meaning of health (Uehara, 2001). Another cultural explanation, referred to in earlier theoretical discussion, derives from the argument of Bankston and Zhou (2002) that attributes the anxiety and low self-esteem particular to some S.E. Asian groups to a strong expectation to excel academically, placing adolescents under extreme pressure. However, even if this is a partial explanation of why S.E. Asian adolescents appraise their health as lower than other groups, it seems curious that males should exhibit more sensitivity to this phenomenon than females. It also is unclear why this effect should occur for Cambodians and Vietnamese alike—given each group’s very different trends in educational attainment and economic assimilation (Portes & Rumbaut, 2001; Rumbaut, 1996). It is certainly possible that Vietnamese and Cambodian males show their distinct disadvantages in self-reported health for different reasons. For example, in the Vietnamese case the pressure to excel academically per the paradox identified by Bankston and Zhou (2002) may be at play, whereas in the case of Cambodian males their self-reported health may be a reflection of their more disadvantaged prospects for economic assimilation. We also note that assimilation and acculturation effects on self-reported health, at least as measured by generational status and non-English language use in the home, are not apparent in our sample. While this negative finding may be due to sample size and composition limitations, we believe our sample size of first generation immigrants (N ¼ 954) is of sufficient size to have reflected at least modest generational effects on self-reported health were they present. Interestingly, this negative finding as pertains to the effect of nativity status is consistent with a very recent and as yet unpublished study that employs an adult sample of Asian Americans and different methods to control for other individual background effects (Erosheva, Walton, & Tageuchi, 2006). Results pertaining to the second research question, the possibility of any unique contribution of social capital to adolescent self-reported health, is ambiguous. That is, the positive effects of 2

Results available from authors.

ARTICLE IN PRESS G. Almgren et al. / Journal of Adolescence 32 (2009) 109–133

129

variables pertaining to the four dimensions of social capital considered in the analysis are limited to females, and their contribution to the explained variation is fairly small (R2 increase of 0.08). Moreover, the particular dimensions of social capital, which are statistically significant for females, derive largely from the social context of the school. Although these results fit with arguments of the relative importance of relationships and social environment to a sense of well-being for females (Stafford, Cummins, Macintyre, Ellaway, & Marmot, 2005), it seems a larger gender difference than that observed here is implied by these arguments. The two other possibilities are that there are other dimensions of social capital that are more relevant to adolescent health than the constructs available from the data, and the limits of measures themselves. Conclusion Although the study sample employed in this analysis of the demographic, developmental and social capital covariates of adolescent health is not a nationally representative probability sample of high-school students, the high proportion of youth from immigrant households and the racial/ ethnic diversity of the sample yielded some interesting findings pertaining to the influences of gender and race/ethnicity on adolescent perceptions of their health. Consistent with findings from the Add Health study sample and from international samples (World Health Organization, 2004), it appears that the gender gap in self-perceived health is well entrenched in adolescence and generalized across most (but not all) racial/ethnic groups. We also note important gender differences in the relevance and magnitude of the covariates of adolescent perceptions of health. The developmental covariates appear to have a stronger influence among males, and social capital and parental support appear more relevant to females. Racial and ethnic background is clearly relevant to perceptions of health, with males of most minority groups represented in this sample fairing somewhat worse than whites. However, with the sole exception of adolescent males of S.E. Asian origin, the racial/ethnic disadvantage in health disappears when variables capturing the influence of social class (parental education) are introduced. To this point, it also seems that at least the parent education dimension of social class plays a role in the self-reported health of adolescent that is in the expected direction. As to the remarkable persistence of the S.E. Asian male self-perceived health disadvantage, it can only be said from these findings that these disadvantages are not explained by either social class background or differences in developmental characteristics that are relevant to self-reported health. Pertaining to the role of social capital to the perceptions of health among adolescents, these findings suggest important gender differences in the effects of at least some dimensions of social capital on health. However, gaining traction on this question will require more precise measures of the construct that represent the broad array of social capital’s multiple dimensions. Acknowledgments The authors gratefully acknowledge the support of the Andrew W. Mellon and Bill and Melinda Gates foundations for this research. This manuscript is a revised version of a paper by the same title presented at the Population Association of America 2004 Annual Meeting in Boston, and the authors also gratefully acknowledge the very helpful remarks and suggestions offered by our PAA colleagues.

ARTICLE IN PRESS 130

G. Almgren et al. / Journal of Adolescence 32 (2009) 109–133

Appendix A detailed presentation of the PC factor analysis results is shown in Table A1.

Table A1 Factor loadings, eigenvalues, and Cronbach’s alpha coefficients for variables included in factor analysis Cronbach’s alpha Parental support indicators (exploratory factor analysis results) Talks to parents How often have you discussed school activities or events with parent(s)? How often have you and your parent(s) discussed going to college? I receive high levels of love and support from my family (recoded) I have frequent, in-depth conversations with my parent(s) (recoded) I can go to my parent(s) for advice and support (recoded) Agree? My parent(s) are usually unhappy or disappointed with what I do Homework support How often do your parent(s) help with or check your homework? How often does another adult (tutor) help with or check homework? How often do your parent(s) limit time with friends on school nights? Social capital factor (exploratory factor analysis results) Positive school affiliation School provides a caring, encouraging environment Teaching is good Teachers are interested in students Students are graded fairly The discipline is fair Network cohesion Parents know friend’s parents Parents know friends Best friends encourage me to do well Safe learning environment Do not feel safe in school Disruptions get in the way of learning Fights occur between different racial/ethnic groups Friendship bonds Has knowledge of friend’s plans Number of people listing person as a best friend

Factor loading

0.81

Eigenvalues

3.01 0.58 0.54 0.78 0.72 0.81 0.70

0.56

1.88 0.74 0.66 0.68

0.80

2.82 0.64 0.76 0.77 0.74 0.69

0.57

1.73 0.84 0.84 0.42

0.86

1.64 0.67 0.74 0.73

0.28

1.68 0.79 0.67

Note: Factor loadings and eigenvalues are derived from varimax rotation. Items with factor loadings of lower magnitude than 70.40 are excluded from reliability analysis and the computation of factor scores employed as measures.

ARTICLE IN PRESS G. Almgren et al. / Journal of Adolescence 32 (2009) 109–133

131

References Andresen, E. M., Catlin, T. K., Wyrwich, K. W., & Jackson-Thompson, J. (2003). Retest reliability of surveillance questions on health related quality of life. Journal of Epidemiology and Community Health, 57(5), 339–343. Atkins, L. A., Oman, R. F., Vesely, S. K., Aspy, C. B., & McLeroy, K. (2002). Adolescent tobacco use: The protective effects of developmental assets. American Journal of Health Promotion, 16(4), 198–205. Bankston, C., & Zhou, M. (2002). Being well vs. doing well: Self-esteem and school performance among immigrant and nonimmigrant racial and ethnic groups. International Migration Review, 36(2), 389–415. Baum, F. E., & Ziersch, A. M. (2003). Social capital (glossary). Journal of Epidemiology and Community Health, 57(5), 320–324. Blum, R. W., Beuhring, T., Shew, M. L., Bearinger, L. H., Sieving, R. E., & Resnick, M. D. (2000). The effects of race/ ethnicity, income, and family structure on adolescent risk behaviors. American Journal of Public Health, 90(12), 1879–1884. Collins, S. R., Davis, K., Doty, M. M., & Ho, A. (2004). Wages, health benefits, and worker’s health. New York: The Commonwealth Fund. Denton, M., & Walters, V. (1999). Gender differences in structural and behavioral determinants of health: An analysis of the social production of health. Social Science and Medicine, 48(9), 1221–1222. Erosheva, E., Walton, E., Tageuchi, D. (2006). Self-rated health among foreign and native-born individuals: A test of comparability. Department of Statistics, University of Washington Seattle, unpublished manuscript. Finch, B. K. (2003). Early origins of the gradient: The relationship between socioeconomic status and infant mortality in the United States. Demography, 40(4), 675–699. Forbes, A., & Wainwright, S. P. (2001). On the methodological, theoretical and philosophical context of health inequalities research: A critique. Social Science and Medicine, 53(6), 801–816. Ford, C. A., Bearman, P. S., & Moody, J. (1999). Foregone health care among adolescents. Journal of the American Medical Association, 282(23), 2227–2234. Ge, X., Elder, G. H., Regnerus, M., & Cox, C. (2001). Pubertal transitions, perceptions of being overweight, and adolescents’ psychological maladjustment: Gender and ethnic differences. Social Psychology Quarterly, 64(4), 363–375. Gold, R., Kennedy, B., Connell, F., & Kawachi, I. (2002). Teen births, income inequality, and social capital: Developing an understanding of the causal pathway. Health and Place, 8(2), 77–83. Goodman, E. (1999). The role of socioeconomic status gradients in explaining differences in US adolescents’ health. American Journal of Public Health, 89(10), 1522–1528. Goodman, E., Amick, B. C., Rezendes, M. O., Tarlov, A. R., Rogers, W. H., & Kagan, J. (1997). Influences of gender and social class on adolescents’ perceptions of health. Archives of Pediatric Adolescent Medicine, 151(9), 99–104. Green, C. A., & Pope, C. R. (1999). Gender, psychosocial factors and the use of medical services: A longitudinal analysis. Social Science and Medicine, 48(10), 1363–1372. Grubbs, S., Hardin, S. B., Weinrich, S., Weinrich, M., Garrison, C., Pesut, D., et al. (1992). Self-efficacy in normal adolescents. Issues in Mental Health Nursing, 13(2), 121–128. Haugland, S., Wold, B., Stevenson, J., Aaroe, L. E., & Woynarowska, B. (2001). Subjective health complaints in adolescence. A cross-national comparison of prevalence and dimensionality. European Journal of Public Health, 11(1), 4–10. Hawkins, J. D., Catalano, R. F., Kosterman, R., Abbott, R., & Hill, K. (1999). Preventing adolescent health-risk behaviors by strengthening protection during childhood. Archives of Pediatric Adolescent Medicine, 153(3), 226–234. Hetland, J., Torsheim, T., & Aaro, L. E. (2002). Subjective health complaints in adolescence: Dimensional structure and variation across gender and age. Scandinavian Journal of Public Health, 30(3), 223–230. Idler, E. L., & Benyamini, Y. (1997). Self-rated health and mortality: A review of twenty-seven community studies. Journal of Health and Social Behavior, 38(1), 21–37. Kaplan, G. A., Pamuk, E. R., Lynch, J. W., Cohen, R. D., & Balfour, J. L. (1996). Inequality in income and mortality in the United States: Analysis of mortality and potential pathways. British Medical Journal, 312(7037), 999–1003. Kawachi, I., Kennedy, B. P., Lochner, K., & Prothrow-Stith, D. (1997). Social capital, income inequality, and mortality. American Journal of Public Health, 87(9), 1491–1498.

ARTICLE IN PRESS 132

G. Almgren et al. / Journal of Adolescence 32 (2009) 109–133

Konu, A. L., Lintonen, T. P., & Rimpela, A. K. (2002). Factors associated with schoolchildren’s general subjective wellbeing. Health Education Research, 17(2), 155–165. Leveque, A., Humblet, P. C., Wilmet-Dramaix, M., & Lagasse, R. (2002). Do social class differentials in health and health behaviors exist in young people (15-to-24-year-olds) in Belgium? Revue d’Epide´miologie et de Sante´ Publique, 50(4), 371–382. Li, X., Stanton, B., & Feigelman, S. (2000). Impact of perceived parental monitoring on adolescent risk behavior over 4 years. Journal of Adolescent Health, 27(1), 49–56. Lochner, K., Kawachi, I., & Kennedy, B. (1999). Social capital: A guide to its measurement. Health and Place, 5(4), 259–270. Macintyre, S., Hunt, K., & Sweeting, H. (1996). Gender differences in health: Are things really as simple as they seem? Social Science and Medicine, 42(4), 617–624. Malone, N., Baluja, K. F., Costanzo, J. M., & Davis, C. J. (2003). The foreign born population: 2000 (No. C2KBR-34). Washington, DC: US Census Bureau. Mechanic, D., & Hansell, S. (1987). Adolescent competence, psychological well-being, and self-assessed physical health. Journal of Health and Social Behavior, 28(4), 364–374. Mellin, A. E., Neumark-Sztainer, D., Story, M., Ireland, M., & Resnick, M. D. (2002). Unhealthy behaviors and psychosocial difficulties among overweight adolescents: The potential impact of familial factors. Journal of Adolescence Health, 31(2), 145–153. Muntaner, C., & Lynch, J. (1999). Income inequality, social cohesion, and class relations: A critique of Wilkinson’s neo-Durkheimian research program. International Journal of Health Services, 29(1), 525–543. Muntaner, C., Lynch, J. W., Hillemeie, M., Lee, J. H., David, R., Benach, J., et al. (2002). Economic inequality, working-class power, social capital, and cause-specific mortality in wealthy countries. International Journal of Health Services, 32(3), 423–432. Portes, A., & Hao, L. (2002). The price of uniformity: Language, family and personality adjustment in the immigrant second generation. Ethnic and Racial Studies, 25(6), 889–913. Portes, A., & Rumbaut, R. G. (2001). Legacies: The story of the immigrant second generation. New York: Russell Sage Foundation. Rahkonen, O., & Lahelma, E. (1992). Gender, social class and illness among young people. Social Science and Medicine, 34(6), 649–656. Ransom, D. C., & Fisher, L. (1995). An empirically derived typology of families: II. Relationships with adolescent health. Family Process, 34(2), 183–197. Rousseau, C., Drapeau, A., & Platt, R. (1999). Family trauma and its association with emotional and behavioral problems and social adjustment in adolescent Cambodian refugees. Child Abuse and Neglect, 23(12), 1263–1273. Rumbaut, R. G. (1996). The New Californians: Assessing the educational progress of children of immigrants. Washington, DC: US Census Bureau. Sack, W. H., Clarke, G., Him, C., Dickason, D., Goff, B., Lanham, K., et al. (1993). A 6-year follow-up study of Cambodian refugee adolescents traumatized as children. Journal of the American Academy of Child and Adolescent Psychiatry, 32(2), 431–437. Sack, W. H., McSharry, S., Clarke, G. N., Kinney, R., Seeley, J., & Lewinsohn, P. (1994). The Khmer adolescent project: I. Epidemiologic findings in two generations of Cambodian refugees. Journal of Nervous and Mental Disease, 182(7), 387–395. Salanta, T., & Lauderdale, D. S. (2003). Measuring culture: A critical review of acculturation and health in Asian immigrant populations. Social Science and Medicine, 57(1), 71–90. Stafford, M., Cummins, S., Macintyre, S., Ellaway, A., & Marmot, M. (2005). Gender differences in the associations between health and neighbourhood environment. Social Science and Medicine, 60(8), 1681–1692. Starfield, B., Riley, A. W., Witt, W. P., & Robertson, J. (2002). Social class gradients in health during adolescence. Journal of Epidemiology and Community Health, 56(5), 354–361. Torsheim, T., Aaroe, E., & Wold, B. (2001). Sense of coherence and school-related stress as predictors of subjective health complaints in early adolescence: Interactive, indirect or direct relationships? Social Science and Medicine, 53(5), 603–614.

ARTICLE IN PRESS G. Almgren et al. / Journal of Adolescence 32 (2009) 109–133

133

Torsheim, T., Currie, C., Boyce, W., Kalnins, I., Overpeck, M., & Haugland, S. (2004). Material deprivation and selfrated health: A multilevel study of adolescents from 22 European and North American countries. Social Science and Medicine, 59(1), 1–12. Torsheim, T., & Wold, B. (2001). School-related stress, support, and subjective health complaints among early adolescents: A multilevel approach. Journal of Adolescence, 24(6), 701–713. Udry, J. R. (2003). The National Longitudinal Study of Adolescent Health (Add Health), Wave I and II, 1994–1996 [machine-readable data file and documentation]. Chapel Hill, NC: Carolina Population Center, University of North Carolina at Chapel Hill. Uehara, E. S. (2001). Understanding the dynamics of illness and help-seeking: Event-structure analysis and a Cambodian–American narrative of ‘‘Spirit Invasion’’. Social Science and Medicine, 52(4), 519–536. West, P. (1997). Health inequalities in the early years: Is there equalisation in youth? Social Science and Medicine, 44(6), 833–858. West, P., & Sweeting, H. (2004). Evidence on equalisation in health in youth from the West of Scotland. Social Science and Medicine, 59(1), 13–27. Wilkinson, R. G. (1999). Health, hierarchy, and social anxiety. Annals of the New York Academy of Sciences, 896(1), 48–63. Wilkinson, R. G., Kawachi, I., & Kennedy, B. (1998). Mortality, the social environment, crime and violence. In M. Bartley, D. Blane, & G. Davey-Smith (Eds.), The Sociology of Health Inequalities (pp. 19–38). Oxford, England: Oxford Sociology of Health and Illness Monograph. World Health Organization. (2004). Young peoples health in context: Selected key findings from the health behavior in school aged children survey. Copenhagen, Denmark: World Health Organization Regional Office for Europe. Yen, I. H., & Syme, S. L. (1999). The social environment and health: A discussion of the epidemiologic literature. Annual Review of Public Health, 20, 297–308.