HEALTH RISK BEHAVIOR AMONG ADOLESCENTS IN IMMIGRANT FAMILIES

Kathleen Mullan Harris

Department of Sociology and The Carolina Population Center University of North Carolina at Chapel Hill

November 1999

Paper to be presented at the second biannual meeting of the Urban Seminar Series on Children’s Health and Safety entitled “Successful Youth in High-Risk Environments,” on December 2-3, 1999 at Harvard University. The research contained in this paper is work in progress; all findings and results are preliminary. This work has benefited from useful comments from Don Hernandez, Deborah Phillips, Ruben Rumbaut, Glen Elder, Nancy Landale, and Nancy Denton. I am grateful for the generous support from the W.T. Grant Foundation through their Faculty Scholars Program and from the National Institute of Child Health and Human Development through the Family and Child Well-Being Research Network, grant 1 U01 HD37558-01. This paper is based on data from the Add Health project, a program project designed by J. Richard Udry (PI) and Peter Bearman, and funded by grant P01-HD31921 from the National Institute of Child Health and Human Development to the Carolina Population Center, University of North Carolina at Chapel Hill, with cooperative funding participation from 17 other federal agencies.

Health Risk Behavior Among Adolescents in Immigrant Families Kathleen Mullan Harris

After a lapse of half a century, the United States has again become a country of immigration and is being profoundly transformed in the process (Farley 1996; Portes and Rumbaut 1996). Since 1990 roughly one million immigrants have been added to the population per year, projecting that more immigrants will have arrived in this decade than at any time in U.S. history (Rumbaut 1998b). The “immigrant stock” of the United States today numbers about 55 million people–that is, persons who are either immigrants (26.8 million) or U.S.-born children of immigrants (27.8 million). Most of the foreign-born arrived between 1980 and 1997, constituting the “new immigration” from Latin America and the Caribbean (52%) and from Asia and the Middle East (30%) (Rumbaut 1997). By the year 2050, projections indicate that nearly half of the U.S. population will consist of persons of Hispanic, Asian or African origin (Smith and Edmonston 1997). Immigrant children (first generation) and U.S.-born children of immigrants (second generation) are the fastest growing segment of the U.S. child population, already accounting for 15% of all American children in 1990, including about 60% of all Hispanic children and 90% of all Asian-American children (Zhou 1997). Today, one of five American children are children of immigrants (first and second generation). The long-term social and economic consequences of this new mass immigration to the United States hinges on the trajectories of these youths, who have been coming of age, transforming their adoptive society as they themselves are transformed into the newest Americans. Many view the dramatic increase and changing composition of immigration with alarm. In part, this is due to the Third World origins of most recent immigrants. Because the sending countries are generally poor and politically chaotic, Americans believe that the immigrants themselves are poor and uneducated and their assimilation will be slow because of their non-European past and tenuous legal status (Portes 1995). There is much speculation but little consensus that the changing demographic characteristics and social origins of the new immigration have increased poverty and inequality in America (Tienda and Liang 1994). Public concern centers on the adverse impact of immigration on the labor market and on the racial and ethnic composition of the population, the primacy of the English language, the level of crime, and American culture in general (Borjas and Tienda 1987; Smith and

2

Edmonston 1997). As immigration from impoverished countries continued to rise, so did public worry that immigrants will drain public coffers by exploiting a social welfare system that provides food, cash, and medical care for the poor. Indeed, the 1996 welfare law eliminated most forms of federal assistance for legal immigrants in the United States in an effort to reduce any economic incentive inherent in our welfare system for immigration from poor countries. This is a critical time for immigrant families and children in America. Language barriers, low economic status and poverty, and alien social and cultural practices stigmatize and isolate immigrant youth from mainstream youth cultures and slow the process by which immigrant youth assimilate into American society (Hirschman 1996). As state and federal policies seek to restrict health services and health benefits to the immigrant population, immigrant families face increasing rates of poverty and limited access to health care (Wolfe 1994). As a result, the health status and well-being of immigrant youth and families is thought to be especially precarious (Klerman 1993). This paper examines the health behavior of immigrant adolescents and native-born adolescents with immigrant parents relative to adolescent health behavior in native-born families. Generational differences are assessed within country of origin and ethnic group background and the extent to which family factors operate in the assimilation process of immigrant youth is analyzed.

Background Despite considerable research on the experiences and adaptation of immigrant adults, there has been very little attention to the immigrant experiences of children in studies of immigration or in studies of child well-being (e.g., Hernandez 1993; Jasso and Rosenzweig 1990; Lieberson 1980; Portes and Zhou 1993). This is largely due to a lack of data on immigrant children or missing information on nativity (Hogan and Eggebeen 1997; Jensen and Chitose 1996; Portes 1996). Only within the past several years have studies addressed the adaptation processes and outcomes of children in the new immigration, but this research is mainly based on regional surveys that exploit the geographic concentration of immigrant families or on specific immigrant ethnic groups (e.g., Fuligni 1997; Rumbaut 1994; Perez 1994; Waters 1996; Zhou and Bankston III 1996; review in Zhou 1997). Research using national data focus primarily on educational outcomes and rarely have sufficient sample sizes to identify separate ethnic groups (e.g., Bradby 1992; Kao and Tienda 1995). We therefore lack a national and representative view of the well-being of

3

immigrant youth in America and we have little understanding of the mechanisms involved in the assimilation processes of children in immigrant families. Current research does suggest that the social contexts in which immigrant youth grow and develop are uniquely different than those of youth in native-born families. Immigrant youth grow up in homes with more complex family structures, inferior socioeconomic resources, and in linguistic or geographic isolation from the non-immigrant white majority (Jensen and Chitose 1996; Rumbaut 1998a). Low education requires immigrant households to rely on multiple workers and multiple jobs and on jobs in ethnic enclaves to escape poverty (Hogan and Eggebeen 1997). The non-standard and long work hours of immigrant parents leave more time in which immigrant youth are unsupervised in the context of peers, neighborhoods, and schools. Such neighborhoods and schools are typically in inner-city urban areas where the majority of new immigrants to the U.S. settle and live with the urban problems of poverty, unemployment, crime, and social disorganization (Sampson and Groves 1989; Wilson 1987). Of course there is considerable variation in these social contexts between and within the two major immigrant groups of Hispanic and Asian background, but there is little research that has documented ethnic differences in the social contexts of immigrant youth, much less studied their impact on measures of wellbeing (Zhou 1997). Theories of immigrant assimilation have a long history that parallel the tumultuous record, the disjointed laws, and the controversial public discourse on immigration to the U.S. (See Alba and Nee 1997; Rumbaut 1998b). The classical model of assimilation is conceptualized as a linear progression or process whereby immigrants discard Old World traits and adopt “American” ones (Gordon 1964; Lieberson 1980; Sowell 1930). This traditional model argues that the length of residence, coupled with increasing generations, leads to the progressive narrowing of socioeconomic and educational differentials with the native-born population. The first generation of immigrants, those who were not born in the U.S., are rarely expected to achieve socioeconomic parity with the native population. Learning a new language, adjusting to a different educational system, and experiencing native prejudice and hostility toward those with a foreign accent and culture are major obstacles for immigrants. The second generation, U.S. born children of immigrants, are socialized in American schools and neighborhoods, receive a mainstream education and obtain the skills needed to participate in the American occupational structure. Their progress is evidenced

4

by the narrowing of the gap in various educational and socioeconomic outcomes between the second generation and the native population (Hirschman 1996). The third generation of immigrants, native-born children with native-born parents but immigrant grandparents, are thought to differ little from the fourth or higher generations because any ethnic influence of grandparents is thought to be relatively minor in a home in which parents do not speak a foreign language and were educated and socialized in American schools and neighborhoods. This “straight-line” model of immigrant adaptation or “Americanization” can also be applied to an intragenerational process of assimilation. The classical hypothesis argues that longer residence in the United States leads to socioeconomic progress and the narrowing of differentials with the native-born population. There is evidence of this process in the reduction of income differentials (Jasso and Rosenzweig 1990). More specific to immigrant youth, the age at which children arrive in the U.S. may affect their process of adaptation. Children who arrive in their preschool years can more easily adapt to the American educational system, learn the English language, and be less stigmatized without a noticeable accent than children who arrive in the U.S. during their adolescence. Revisionist theories of immigrant assimilation have evolved from the study of today’s new immigrants, whose adaptation contradicted the classical model, often resulting in deteriorating outcomes both over time and generation in the U.S. (Gans 1992; Hirschman 1996; Reimers 1992). Portes and Zhou (1993) advanced a “segmented assimilation” thesis which argues that different groups of immigrants experience different outcomes based on the sector of American society into which they assimilate. For instance, the classical hypothesis would argue that adolescents who arrived in the U.S. at a younger age and who have spent more time in the U.S. will assimilate into society more readily than immigrant adolescents who arrived more recently. If, however, greater exposure to American society has primarily been in inner-city environments, where many new immigrants settle and where the social environment and economic opportunities have been declining, then immigrant children with longer U.S. residence (and a younger age at arrival) may not be doing better than recent arrivals. Most of today’s immigrants are racial minorities who will continue to be recognized as minorities even if they culturally assimilate to mainstream America. Most immigrant families settle within inner cities and urban ghettos, near American minorities of the same racial background who have historically faced socioeconomic barriers and discrimination. Such neighborhoods have high poverty rates and lack jobs that offer decent wages

5

and mobility opportunities (Jensen and Chitose 1996; Portes and Zhou 1993; Rumbaut 1997). This paper examines the health risk behavior of a population of adolescents who represent this “new immigration.” Using data from a national representative study of adolescents in American schools in grades 7 through 12, this study includes immigrant youth who arrived in the U.S. between 1975-1994 and native-born youth of immigrant parents. Health behavior outcomes among first and second generation youth are contrasted with the outcomes of youth in native-born families of the third+ generation. Differentials in health behavior across generations will reveal the assimilation process by the extent to which immigrant youth adopt the health norms and behaviors of youth in native-born families.

Adolescent Health, Development, and Assimilation As a minority group becomes more highly assimilated into the mainstream of American values and customs, changes in health-related attitudes and behaviors may also occur (Earls 1993). Acceptance of the predominant values may make such a group more amenable to the prevailing social norms of health behaviors. The pattern of diseases characterizing the group may also shift toward that experienced by the majority group (Mendoza et al. 1990). However, behavioral changes may yield unwanted outcomes. For example, low levels of assimilation are associated with lower rates of completed suicide among Mexican-Americans (Earls, Escobar, and Manson 1990). An increasingly cited finding is that foreign-born Mexican-Americans experience lower rates of infant mortality and low birth weight than other groups (Bautista-Hayes 1990). Subsequent generations of Mexican-Americans appear to lose this advantage, which may be a consequence of adopting the lifestyle and habits of the dominant culture. Perhaps at no stage of the life course are assimilative processes more intensely experienced, or assimilative outcomes more sharpy exhibited, than during the formative years of adolescence. In a period where being different or “standing out” takes on crucial social significance, acceptance into peer networks and school culture may be especially important for immigrant adolescents who are already different in either their appearance, dress, or speech. Family influence can have especially poignant effects, either coddling youth within the boundaries of their ethnic traditions and values or turning youth away from their own ethnic culture and family traditions which define their difference within American society. As youth increasingly value peer friendships and peer relationships, concurrent distancing from the family

6

origin constitutes a central task of adolescence (Perry et al. 1993). This normal developmental process of waning family involvement and increasing peer influence may be particularly alien to the cultural practices and models of respect in immigrant families and may create family conflict and divided loyalties for immigrant adolescents (Zhou 1997). For adolescents in immigrant families, normal developmental tensions may be exaggerated as they strive to be accepted by the majority population and at the same time cope with family and neighborhood socializing agents, some who may wish to deter the assimilation process and others, especially peers, who may speed the process. How this tension plays out in the lives of immigrant adolescents will furthermore vary by ethnic background. Family influence on adolescent behavior is primarily viewed as operating through socialization and social control mechanisms. Socialization theory explains how various socializing agents in the family and larger community influence the development of children and adolescents (Maccoby 1992; Maccoby and Martin 1983). The main function of socialization within the family is to expose or teach youth the values, attitudes, and norms of behavior most valued or adopted by parents and other adults in the family and to prepare youth to assume an adult role in society that embodies the values and norms being taught. As socializing agents, parents and other family members serve as role models who shape the perceptions and goals of adolescents regarding future roles and lifestyles they are likely to assume. There is a prevalent assumption that parental influence wanes as children age, but the socialization literature indicates that parents continue to influence their children during adolescence and long after they have left the parental home, and that events that occur in the context of parent-child interactions affect children’s social behavior in other settings and at later times (Maccoby 1992; Majoribanks 1987; Amato and Booth 1997). The influence of parents as socializing agents, however, will vary by the developmental stage of the individual and by the individual’s gender, race, and ethnicity. The socialization of children and youth also involves monitoring and regulation of social behaviors. Social control theory (Hirschi 1969; Liska and Reed 1985) further elaborates on the family mechanisms that operate to supervise and monitor adolescent development and behavior (e.g., Hogan and Kitagawa 1985; Thornton and Camburn 1987; Udry 1988). Parents may inhibit or facilitate health risk behaviors by providing or restricting opportunities through direct supervision and monitoring of adolescents’ behavior. The structural characteristics of family contexts may modify socialization and social control

7

processes and define the social and economic resources available to youth. Families characterized by high poverty or family disruption increase the likelihood of health risks among youth (National Research Council 1993). One-parent families are less effective in supervising youth and socialize children to be more accepting of alternative family forms (Hogan and Kitagawa 1985; Wu and Martinson 1993; McLanahan 1988; McLanahan and Sandefur 1994). Economic stress reduces parents’ abilities to be supportive and nurturing of children (Elder et al. 1992; McLoyd 1990). The structural characteristics of family contexts also define social resources in the form of “social capital” (Coleman 1988). Social capital is embodied in the relations among persons and accrues through social connections and interactions in various and multiple social contexts. Within families, social capital is primarily developed through parent-child relations. Like other forms of capital, social capital is productive, making possible the achievement of certain ends that in its absence would not be possible. Without close parent-child relationships, parental communication and monitoring would be ineffective in promoting positive health outcomes among adolescents. Without shared activities and time together, opportunities for discussions about safe sex or drinking and driving are precluded. The absence of a father in the home limits adolescents’ access to important social capital resources associated with the father’s information sources, social contacts, and other opportunities through his membership in larger networks (McLanahan and Sandefur 1994). Social capital also exists in relations with others in the community which are facilitated by the family’s activities or involvement in the community. To the extent that the family is involved in religious institutions or has associations with community members of the same race or ethnic group, youth will meet and feel connected to members outside their immediate family who may provide key sources of social and emotional support. These social networks, especially those that involve connections with other adults, can reinforce normative expectations and supervise adolescents’ behavior in a way that is consistent with that of the family of origin. Because the family is the primary arena for the socialization and social control of children (Elder 1980; Maccoby 1992), social capital in families is the most accessible social resource for youth (Coleman 1988), and family context plays a central role in the assimilation processes of immigrant children (Fernandez-Kelly 1995; Fuligni 1997; Rumbaut 1994), this research focuses on the role of family mechanisms in understanding the health behavior outcomes of immigrant youth. The objective of the

8

analysis is to examine how ethnic background and immigrant generation influence health risk behavior, and whether family context explains any of the observed differences in health behavior outcomes. To the extent that an assimilation process is observed such that immigrant youth increasingly adopt the health norms and behaviors of the mainstream youth culture across immigrant generations and across time in the U.S., the analysis further examines whether the process varies by ethnic origin and how family structural, family process, and cultural adaptation factors operate in the assimilation process through socialization, social control, and social capital mechanisms. I explore three sets of mediating factors to understand differentials in the health risk behavior of immigrant youth. Family structural factors include family structure and socio-economic status which determine the economic resources in the household as well as levels of household social capital, supervision, and role models. Family process factors include parenting behaviors and parent-child relationships. Parental monitoring and supervision, and the quantity and the quality of parent-child relations will influence adolescent risk behavior through social control and social capital mechanisms and will vary by generation and ethnic origin. The third set of mechanisms are conceptualized as cultural adaptation factors that are central to the assimilation process of immigrant youth. These factors include measures of the strength of ties to the adolescent’s cultural identity and ethnic group values and practices. Such factors involve religiosity and the importance of religion, the primary language spoken in the home, perceived social support of family members and other adults in the community, and the presence of other relatives in the home. To the extent that such cultural factors are prevalent in immigrant families, assimilation into mainstream youth culture may be delayed or prevented by slowing the adaptation process of Americanization.

Data Data come from the National Longitudinal Study of Adolescent Health (Add Health), a nationally representative study of over 20,000 adolescents in grades 7 through 12 in the U.S. in 1995. Add Health was designed to help explain the causes of adolescent health and health behavior with special emphasis on the effects of multiple contexts of adolescent life. The study used a multistage, stratified, school-based, cluster sampling design. A stratified sample of 80 high schools was selected with probability proportional to size.1 For each high school, a feeder school was also selected with probability

9

proportional to its student contribution to the high school. The school-based sample therefore has a pair of schools in each of 80 communities.2 An in-school questionnaire was administered to every student who attended each selected school on a particular day during the period of September 1994 to April 1995. The in-school questionnaire was completed by more than 90,000 adolescents, and was intended to measure school-context variables; to obtain friendship networks; to measure a variety of health conditions; and to obtain data to make it possible to select special samples of individuals in rare but theoretically crucial categories. In a second level of sampling adolescents and parents were selected for in-home interviews. The school rosters of the sampled schools were used as the sampling frame for the in-home sample. From the school rosters, a random sample of some 200 students (100 of each gender) from each school pair was selected, irrespective of school size, to produce the core in-home sample of about 12,000 adolescents. A number of special over-samples were also selected for in-home interviews using screeners from the inschool questionnaires. With varying probabilities, the following special samples were drawn: physically disabled adolescents; black adolescents from highly educated families; several ethnic samples (Cuban, Puerto Rican, and Chinese adolescents); a genetic sample (identical and fraternal twins, full sibs, half sibs, and unrelated adolescents in the same household); and saturated samples in 14 schools. The in-home interviews were conducted between April and December 1995, yielding the Wave I data. The core sample plus the special samples produced a sample size of 20,745 adolescents in Wave I. A parent, generally the mother, was also interviewed in Wave I. The one- to two-hour adolescent interview was interviewer assisted using a laptop computer and audio-CASI for the more sensitive questions. Minority populations are represented in proportion to their size in the population, yielding sufficient samples for separate analysis of major ethnic groups nationwide. The special over-samples permit separate analysis of smaller ethnic groups in America. The special ethnic samples, large sample size, and national representation make this an ideal dataset with which to study immigrant youth and families. Add Health data provide an unprecedented view of the experiences of immigrant youth and the influence of family contexts on immigrant health status and health behavior. Data are available on nativity of youth and parents, both resident and nonresident biological parents. Ethnic background is reported for youths and parents of all Add Health respondents, providing comparisons of family context

10

and health behavior outcomes by immigrant generation within ethnic group background. This study uses data from the Wave I in-home sample. Descriptive analysis is based on 20,457 cases with valid generation and ethnic origin data. The sample size drops somewhat in multivariate analysis due to missing values. All analyses use sampling weights to adjust for the complex sampling design and standard errors are adjusted for the clustered design by using a robust cluster estimator. See Bearman, Jones, and Udry, 1997 for a more detailed description of the Add Health study.

Measures This study focuses on the health dimension of risk behavior. Health risk behaviors are self-reported by the adolescent and include the following: 1) ever had sexual intercourse; 2) age at first intercourse; 3) delinquency; 4) violence; and 5) use of substances.3 Delinquency, violence, and substance use are measured as continuous indexes in multivariate models only, while dichotomous measures indicating high involvement in these health risk behaviors are used in descriptive analysis. A risk behavior index based on sexual behavior, delinquency, violence, and use of substances is presented as well. For the entire sample of adolescents aged 11-21 at Wave I, 38 percent reported ever having had sexual intercourse (see Table 1, Total column, for sample means). Among those who have had sex, the average age at first intercourse was 14.5, reflecting the selective nature of the sample of sexually active adolescents in high school at a point in time. Delinquency is measured by constructing an index of 9 delinquent or illegal behaviors in which the youth engaged such as paint graffiti, damage property, shoplift, run away from home, steal a car, sell drugs, and burglary. In descriptive analysis, a dichotomous measure indicating whether the youth engaged in 3 or more delinquent acts is used (16%) and in multivariate analysis the count of delinquent acts expressed as a proportion of all possible and non-missing responses is used. The reliability (cronbach’s alpha) of the index is .80. Violence is measured as an index based on 9 items in which the youth reports violent behavior and weapon use including such items as fighting, pulled a knife or gun on someone, shot or stabbed someone, and used a weapon in a fight. The dichotomous measure used in descriptive analysis identifies youth who engaged in 3 or more acts of violence (21%), and the multivariate analysis uses the proportionate measure with a reliability of .80.

11

Use of various substances is examined in descriptive analysis, shown in Table 1. The intensity of smoking and drinking alcohol is examined, along with use of marijuana and hard drugs. Fifty-eight percent of adolescents have ever tried smoking, while 21.5 percent smoke regularly (at least one cigarette every day for 30 days). Thirty percent of youth drink alcohol one or more times a month while 17 percent report getting drunk once or more a month. Nearly 28 percent of youth have ever used marijuana, while 14.4 percent used marijuana in the last month. Use of hard drugs, which includes inhalants, cocaine, other illegal drugs, or injected illegal drugs, is only prevalent among 12.3 percent of youth. To achieve parsimony in the multivariate analysis, I constructed an ordinal scale of substance use that ranges from 0 to 4 and captures the severity of risk involved with use of specific or multiple substances. Values on the scale are assigned in the following manner: (0) never used substances; (1) tried smoking or drink once a month or more only; (2) regular smoker or get drunk once or more a month and no use of marijuana or hard drugs; (3) used marijuana in the last month, smoked or drank alcohol but no use of hard drugs; and (4) use hard drugs in any combination with other substances. The sample mean on this scale is 1.32, with 36 percent using no substances, 28 percent with a value of 1, and 12 percent using hard drugs with a value of 4. A risk behavior index is constructed to create a summary measure of the degree of health risk in which adolescents engage across the range of behavioral domains studied and to serve as a powerful comparison measure across immigrant generation and ethnic origin. The index sums the dichotomous indicators of ever had sex, engaged in 3 or more delinquent acts, 3 or more acts of violence, being a regular smoker or getting drunk once or more a month, and recent use of marijuana or hard drugs. The index ranges from 0, involved in no risk behaviors, to 5, involved in all 5 risk behaviors with a sample average of 1.25 risk behaviors.

Youth Characteristics and Ethnic Background. Gender and age capture numerous aspects of both the biological and social readiness of adolescents to be responsible for their health care and to engage in certain health behaviors, as well as differential exposure to health risks in the social environment. Socialization processes also vary by developmental age and gender, as do social control mechanisms (Liska and Reed 1985; Maccoby and Martin 1983) . Moreover, certain social capital resources, such as close parent-child relations, are probably more relevant to the

12

health of younger adolescents, whose health behavior may be more influenced by parental information sources and other adults, than older adolescents. The gender distribution is even and the average age of the sample is 15.5. Immigrant status and ethnic background are fundamental exogenous variables that structure this analysis. Immigrant status is defined by a three-category variable signifying immigrant generation: foreign-born adolescents to foreign-born parents (1st generation), native-born adolescents to foreign-born parents (2nd generation) and native-born adolescents to native-born parents (3+ generation). Foreignborn adolescents with foreign-born parents are those children who were not born in the United States nor were born U.S. citizens abroad, and thus migrated to the U.S. as children (in most cases with their immigrant parents). Native-born adolescents with foreign-born parents are children born in the United States (and thus are U.S. citizens) but who have at least one parent who is foreign-born. Adolescents in native-born families (3+ generation) may have grandparents or great grandparents who were immigrants, but because the immigration experience is much farther removed from the social context of their childhood and adolescent development, this category is considered the native population and the fundamental comparison group for immigrant children and the children of immigrants. Parallel ethnic origin classifications are used for immigrant children, for the children of immigrants, and for youth in native-born families. Adolescents who indicate that they are of Hispanic or Asian background are then asked to identify their specific Hispanic or Asian ethnic background. For immigrant youth and the children of immigrants, I check this self-reported ethnic identity against country of birth reports of youth and parents for consistency or to achieve a more specific ethnic identity.4 For immigrant youth who are not Hispanic or Asian, I use their country of birth to assign ethnic origin.5 This classification scheme results in a nine-category variable of race and ethnic background that permits comparisons of immigrant generations within the following groups: Mexican, Cuban, Puerto Rican, Central and South American, Chinese, Filipino, Other Asian, African and Afro-Caribbean, and Canadian and European. A four-category race and ethnicity variable is defined for all adolescents, but the measure is used to classify youth in native-born families in aggregate comparisons to youth in immigrant families (e.g., Table 1). The categories are: non-Hispanic white, non-Hispanic black, Asian, and Hispanic.6 A small number (about 200) of American Indian youth were not included in this study.

13

Family Context Structural features of the family include family structure at Wave I, parents’ education, family poverty status, number of siblings, and urban residence. Family structure is measured as a five-category variable, operationalized as a set of dummy variables that identify youth who live with 1) two biological parents (reference); 2) step parents (step, adopted, or foster); 3) mother only; 4) father only; and 5) other family forms (with grandparents, aunts, uncles or other relatives, or in group homes). Parents’ education is measured as the higher of either mother’s or father’s education and is represented as four dummy variables: 1) less than high school (reference); high school diploma or GED; some college; and college or post-graduate schooling. Poverty status is indicated if reported 1994 income was less than the poverty threshold for a family of 4 in 1994.7 These family structural effects may operate through multiple mechanisms. For example, single parents serve as role models for adolescents, transmitting values and norms of behavior that run counter to mainstream views on work and family formation (McLanahan 1995). Single parents also exercise less supervision and social control over adolescent behavior, and provide fewer resources to adolescents, possibly restricting opportunities for youth, thereby lowering the costs of health risk behaviors. Finally, social capital resources are notably less in single-parent families and less-educated families (McLanahan and Sandefur 1994; Furstenberg and Hughes 1995). The number of siblings may determine the amount of direct supervision parents can expend on children, as well as the opportunities and potential camaraderie in experimenting in risky behaviors. The process of migration itself often disrupts the traditional family system and truncates family networks which in turn weaken traditional mechanisms of control and social support (Fernandez-Kelly 1995). In addition to urban residence, I control for region of residence (west, south, midwest, and northeast as the reference) because immigrant groups are concentrated in only a handful of states in the U.S. (Farley 1996). Not surprisingly, 88 percent of first generation immigrant youth and 77 percent of second generation youth live and attend school in urban areas (Table 2), and live an urban lifestyle in a declining social and economic environment that lacks safety, public resources and institutions, and opportunities to engage in healthy behaviors (Jensen and Chitose 1996). Family process measures include parental supervision, parental monitoring, parent-child closeness and shared activities. Supervision and monitoring capture key social control mechanisms. Social capital,

14

on the other hand, is manifest in parent-child interactions that involve shared activities and time together, and the emotional bond between parents and adolescents. I argue that family socialization is facilitated by the social capital that is present in adolescent-parent relationships. That is, the more social capital in parent-child relations, the more effective parenting behaviors are in controlling and responding to adolescent health issues and behaviors. Parental supervision is a count variable ranging from 0 to 4 indicating whether a parent is present in the home most or all of the time the adolescent 1) goes to school in the morning, 2) comes home from school in the afternoon, 3) eats the evening meal (5-7 dinners a week), and 4) goes to bed at night. A lack of parental presence reduces supervision of youth and reduces the effectiveness of socialization and social capital resources since such mechanisms can not operate in the absence of time shared by parents and adolescents. Parental monitoring is measured by an index constructed from a series of questions that ask whether the adolescent makes all his or her own decisions about 1) the time to be home on weekends; 2) friends; 3) clothes; 4) how much television to watch; 5) which TV shows to watch; 6) time to go to bed on week nights; and 7) what to eat (cronbach’s alpha=.63). The index is reverse coded such that high measures indicate high monitoring. The quantity of time spent together is measured by the count of shared activities in which the parent and adolescent engaged in the last 4 weeks, such as religious or church events, shopping, sports, attending sports events, going to a movie, or working on a school project. Parent-child closeness is measured by the mean response (ranging from 0=low to 4=high) of adolescent reports on the level of closeness, satisfaction, warmth, and satisfaction with communication in the parentchild relationship (alpha=.89).8 The last set of family context variables measure dimensions of cultural adaptation. These measures will be especially salient in immigrant families because high values will represent stronger ties to the ethnic culture and traditions of the family, which has implications for youth assimilation into American culture. Religiosity is measured by church attendance which ranges from 0 (never attend) to 4 (attend once a week or more). The importance of religion is measured on a 4-point scale where 0 is not important at all and 4 is very important. The frequency of prayer is measured on a 5-point scale where 0 is never and 5 is at least once a day. Those who are missing on religion are given a value of -1 on each of these measures, and are assigned a 1 on a missing religion variable. Three dummy variables are created if the adolescent reports that the primary language spoken at home is English; Spanish, or an other language.

15

Only 31 percent of first generation youth speak English at home and this more than doubles to 69% among second generation youth (Table 2). An index of social support from adults and family members is measured by adolescents’ mean responses to 8 items on a scale from 1 (not at all) to 5 (very much) of how much adolescents feel that adults, teachers, parents, friends, and people in their family care about them, and whether people in their family understand them, pay attention to them, and have fun together (alpha=.78). Finally, a dummy variable indicates whether any other relatives, including grandparents, aunts, uncles, or other relatives, live in the youth’s household. These measures are key indicators of cultural adaptation because the stronger such cultural measures are within the family context, the more likely socialization influence will be toward family ethnic traditions and values, the more effective social control by parents or family members will be toward normative behaviors of the family cultural background, and the more social capital will exist in relations with persons of a similar ethnic identity. For example, adolescents with high religiosity are less likely to deviate from normative behaviors and are more responsive to social control mechanisms than adolescents for whom religion is not important. For highly religious immigrant youth, the normative behaviors and values will be those of the parents or family, tying the adolescent to their cultural origins. The presence of additional family members or other adults in the household, a common experience for immigrant families (Jensen and Chitose 1996), could represent important sources of social capital, as well as additional sources of socialization and social control of youth. The greater the social support from family members and adults in the community, the more connected the immigrant adolescent will be with his or her cultural norms. Such mechanisms, therefore, should operate to isolate and shield immigrant youth from American norms of behavior and slow the adaptation process of assimilation.

Analysis Plan The analysis will proceed in two stages. The first stage involves descriptive analysis of differences in health risk behaviors by immigrant generation and then explores whether differences in risk behavior by immigrant generation persist within ethnic origin. At this stage, I also examine the assimilation process among immigrant children by contrasting behavioral outcomes by age at entry and length of time in the U.S. Stage two of the analysis employs multivariate regression methods to assess the relative effects of

16

immigrant generation by ethnic group categories on the various health behavior outcomes compared to the baseline effect for non-Hispanic white adolescents in native-born families. This baseline model serves to demonstrate degrees of assimilation by the extent to which immigrant generations are similar to or different from the native-born mainstream youth culture of non-Hispanic whites in their health risk behavior. Subsequent regression models then explore the extent to which these effects operate through differences in family context, including structural characteristics, family processes, and cultural adaptation factors. This final stage of the analysis addresses the role that family context plays in the assimilation process of immigrant children. Multivariate analyses use two forms of regression: logit models to estimate the risk of first sexual intercourse, and OLS models to estimate the linear proportionate indexes of risk behavior, as well as the risk behavior index. All analyses use sampling weights and correct standard errors for the clustered design.

Health Behavior Differentials by Immigrant Generation If we accept the view that second generation immigrant youth have adopted the values and culture of American society more so than first generation youth, we can observe health behavior differentials across immigrant generation as representing an assimilation process. This perspective is based on the argument that second generation youth represent the first members of their family to be educated and socialized in American institutions, whereas first generation youth may carry with them the socialization and educational experience that they received in their country of origin, depending on the age at which they entered the U.S. This perspective is addressed in Table 1 which examines various health risk behaviors of first and second generation immigrant youth in comparison to adolescents in native-born families of the third+ generation who are further subgrouped into native populations of non-Hispanic whites, non-Hispanic blacks, Asians, and Hispanics. If we view generation as representing the degree of assimilation, the results indicate a pattern of increasing adoption of the mainstream behaviors of adolescents in native-born families with increasing assimilation into American culture. For all of the outcomes examined, second generation youth have a greater propensity to engage in risk behavior than first generation youth. Nativeborn children of immigrants are more likely than immigrant children to ever have had sex (34.3% vs. 26.9%), to engage in 3 or more delinquent acts (20.6% vs. 10.7%), to be involved in 3 or more violent

17

acts (21.7% vs. 15.6%), to ever tried smoking (55.4% vs. 40.7%), to smoke regularly (16.7 vs. 8.7), to drink alcohol one or more times a month (31.1% vs. 15.9%), to get drunk one or more times a month (15.5% vs. 7.4%), to have ever used marijuana (27.9% vs. 14.8%), to have used marijuana last month (14.0% vs. 5.4%), and to have used hard drugs (11.1% vs. 4.6%). All of these differences are significant at the .05 level, while differences in the age at first intercourse are negligible between the two immigrant generations. Reflecting the heterogeneity of native-born families, health behavior outcomes vary considerably across race and ethnic groups in the native population. Asian youth are the least likely to have ever had sex; Asian and African-American youth are the least likely to use substances; African-American youth are the least likely to engage in delinquent behavior; and non-Hispanic white youth are the least likely to engage in violence. Native-born Hispanic youth seem to be the most involved in risk behavior across all behavioral domains, with the exception of first sexual intercourse. Knowing that the ethnic composition of first and second generation youth largely represents the “new immigration” from Latin America and Asia, the native subpopulations of Asian and Hispanic adolescents are most similar in their ethnic background and may represent the “segment” of the U.S. native population to which immigrant youth assimilate. If we contrast the prevalence of risk behavior for first and second generation youth with the native populations of Asian and Hispanic adolescents, for certain health behaviors we observe a fairly linear pattern. For example, involvement in delinquent behavior increases from 10.7 % in the first generation to 20.6% in the second generation to 22.1% among Asian youth and 24.7% among Hispanic youth in the third+ generation. The same pattern is observed for violent behavior. A linear pattern of increasing involvement in risk behavior across generations also occurs for first sexual intercourse and use of various substances if we contrast immigrant generations with Hispanic youth in the third+ generation. The risk behavior index displays this linear pattern as well. The other relevant contrast for immigrant generations is with native-born non-Hispanic white adolescents, with whom a similar linear pattern is found for substance use only. In other health behavior domains, second generation immigrant youth are equally or more involved in risk behavior as white adolescents in the third+ generation. For example, nearly the same percentage of second generation youth have had first sexual intercourse and a larger percentage are engaged in delinquent and violent behavior than non-Hispanic white youth in native-born families. However, the risk behavior index, which

18

summarizes involvement across all domains, reveals a linear pattern of increasing involvement in risk behavior with increasing generation. In sum, the most consistent finding from Table 1 is a pattern of assimilation displayed by the increasing propensity to engage in health risk behaviors across immigrant generations of youth, especially comparing immigrant children and the children of immigrants to the native populations of non-Hispanic white youth and Hispanic youth on use of substances, and comparing immigrant generations to the native populations of Asian and Hispanic youth on delinquent and violent behavior. The most striking finding, however, is that first generation youth engage in the least risk behavior. Foreign-born youth have less experience with sex, are less likely to engage in delinquent and violent behavior, and are less likely to use controlled substances than native-born youth.

Health Behavior Differentials among First Generation Youth To further test the assimilation model from an intragenerational perspective, health behavior outcomes by time in the U.S. for immigrant youth are examined. The results are dramatically consistent: the longer the time since arrival in the U.S., the greater the likelihood of engaging in health risk behaviors. Age at arrival is correlated with time in the U.S., and the same results are obtained when health behavior differentials are examined by age at arrival to the U.S. Figure 1 summarizes these results by presenting means on the risk behavior index by years in the U.S. for first generation youth. Data on the specific health risk behaviors by time in the U.S. are shown in Appendix I. A remarkably linear and statistically significant pattern is evident indicating that with more years of exposure and assimilation into American culture, health risk behavior increases.

Health Behavior Differentials by Immigrant Generation and Ethnic Origin The aggregate differences by immigrant generation documented thus far mask likely variation in the relationships between generation and health behavior by ethnic background. Differences in the ethnic composition of first and second generation youth and in the relationship between ethnic origin and health behavior may partially explain the aggregate patterns found in Table 1. I therefore introduce controls for ethnic background and examines differences in risk behavior outcomes by immigrant generation within ethnic origin.

19

This descriptive analysis produces a multitude of numbers to shift through in order to grasp general patterns and are presented in Appendix II. The risk behavior index serves as a summary measure of the extent to which youth are engaged in health risk behavior, and facilitates the emergence of the fundamental patterns found in Appendix II. The mean risk behavior index by immigrant generation and ethnic origin is shown graphically in Figure 2.9 The general pattern shown in Figure 2 confirms the aggregate findings: first generation youth engage in less risk behavior than second generation youth, who generally engage in less risk behavior than the native youth population in the third+ generation. Exceptions to this pattern are found for Chinese youth, among whom the second generation is most involved in risk behavior (this pattern holds across all health behavior outcomes except violence shown in Appendix II), and for European and Canadian youth, among whom the second generation is equally involved in risk behavior as the third+ generation. The classical linear pattern of increasing involvement in risk behavior across generations is most prominent for Mexican, other Asian, and African and Afro-Caribbean youth, and is roughly found for Central and South American, Puerto Rican, and Filipino youth. For certain ethnic groups, the first and second generation are more similar in their levels of risk behavior and distinctly different from the third+ generation, while for others it is the second and third+ generations that are more similar and distinctly different from the first generation. For example, the main difference in levels of risk in the health behavior of youth from Central and South America is between children in immigrant families (first and second generations) and children in native-born families (third+ generation). Among Cuban, Puerto Rican, Filipino, and European and Canadian youth, foreign-born children, those in the first generation, display much lower involvement in risk behavior than native-born youth–those in the second and third+ generations. The extent to which certain ethnic groups experience segmented assimilation can be seen clearly in Figure 2 where the risk behavior of second generation youth is greater than that of the mainstream youth population of non-Hispanic whites (the third+ generation of European/Canadian youth–the very last bar on the right). Only mainland-born Puerto Rican youth with island-born parents (i.e., second generation) are more involved in risk behavior than non-Hispanic white youth in the third+ generation, although second generation Mexican youth engage in nearly the same level of risk behavior as nonHispanic whites in native-born families.

20

Note finally that Chinese youth stand out as the ethnic group least engaged in risk behavior. Because of intense familial expectations and cultural pressure to maintain the family reputation and focus only on achievement goals in American society, it is possible that Chinese youth may underreport their engagement in risk behavior in this survey, however, this potential bias can not be explored with the data at hand. Youth of Hispanic origin, on the other hand, display the greatest involvement in risk behavior. In sum, the dominant finding of Figure 2 (and Appendix II) is the prevalence of a protective advantage in health behavior associated with immigrant status among the major ethnic groups in America. As the immigrant experience becomes more distant across generations, youth become more similar to the native population in their health behavior. To the extent that immigrant adjustment and assimilation involves the adoption of health and behavioral norms in the native ethnic subgroup with which immigrants are identifying, a segmented assimilation process is evident for Mexican and Puerto Rican ethnic groups.

Multivariate Analysis The second stage of analysis entails testing for the significant and independent effects of immigrant generation and ethnic origin categories displayed in Figure2 and exploring the family mechanisms through which the protective effects of first generation immigrant status potentially operate. Multivariate regression analysis is used to model health behavior outcomes as a function of immigrant generation by ethnic origin (essentially an interaction effect) relative to the reference effect of native non-Hispanic whites, the majority native ethnic group of youth in America. The modeling strategy involves first estimating a baseline model, and then adding in sets of mediating variables according to their theoretical import in the likelihood of engaging in risk behavior according to potential differences by generation and ethnic group origin. A selection of the main mediating variables explored are shown in Table 3 by generation with the third+ generation broken down by race and ethnicity, and in Table 4 by ethnic origin for the first generation only. To the extent that such mediating factors explain the effects of generation by ethnic origin, we gain some insight into the family processes that lead to differential health behavior outcomes across generations within ethnic origin.

Family Context Differences

21

Table 2 shows that there are important structural differences in family context by immigrant generation that appear to disadvantage immigrant youth. For instance, first generation youth experience one of the highest poverty rates (28%) and first and second generation youth are the most likely to have parents with less than a high school education (32% and 23%, respectively). Immigrant youth also have more siblings than youth in native-born families (not shown). As noted earlier, 88% of first generation youth and over three-quarters of second generation youth live in urban areas. Family process differences are not as marked. Immigrant parents monitor their children to a greater extent than native-born parents with the exception of Asian parents, but they do not foster as close emotional bonds with their children as do third+ generation parents. Cultural adaptation factors are generally more important in immigrant families. While first generation youth do not necessarily score the highest on the individual measures of cultural adaptation, they are consistently higher than the mean score across all measures. In particular, first generation youth place high importance on religion (but less than native non-Hispanic blacks), have higher than average social support (but less than native Asians), are likely to live with other relatives (but not as likely as native non-Hispanic blacks), and are the least likely to speak English at home. Cultural adaptation characteristics among second generation youth are not very different from those of the various race and ethnic groups of the third+ generation, with the exception of speaking English at home. Table 3 examines differences in the mediating factors by ethnic origin for first generation youth. There is substantial variance in the family contexts of immigrant youth by ethnic origin, highlighting the importance of controlling for ethnic background. In general, structural features of families tend to disadvantage immigrant youth of Hispanic descent with lower percentages living in two-parent families, and higher percentages living with a single mother, living in poverty, having parents without a high school education, and living in urban areas. Certain disadvantages are particular to certain Hispanic ethnic groups. For example, Mexican youth have the highest percentage of parents without a high school education; while Cuban youth have the highest family poverty rates and practically all of the Cuban sample lives in an urban area. Puerto Rican youth are relatively better off among the Hispanic groups with lower poverty rates and fewer less-educated parents. In contrast, Asian immigrant youth appear to live in better economic and social conditions. Chinese immigrant adolescents, in particular, are more likely to live in two-parent families, while Filipino youth enjoy very low poverty rates and have more

22

educated parents. African and Afro-Caribbean immigrant youth are mainly disadvantaged in terms of family structure, but do not seem to face particularly high poverty or have uneducated parents. Finally, European and Canadian immigrant youth enjoy the most advantageous social and economic conditions. Parental supervision and monitoring is more common among Hispanic parents, particularly among Cuban, Puerto Rican, and Mexican parents, as is parent-child closeness. Perhaps more striking are the low levels of parental supervision and parent-child closeness in Chinese and Filipino families. Parental monitoring is also rather low among African, Afro-Caribbean, European, and Canadian families. The main differences in the cultural adaptation characteristics seem to lie between the “new immigrants” of Hispanic and Asian background and immigrants from Africa, the Afro-Caribbean, and Europe and Canada. For example, about three-fourths of African, Afro-Caribbean, European, and Canadian immigrants speak English at home compared to less than 20 percent of youth from Mexico, Cuba, and Central and South America. Moreover, it is largely the various Hispanic and Asian immigrant families that place importance on religion much more so than African and European immigrant families.

Multivariate Results Tables 4-8 display the results of regression models estimating the risk behavior index, first sexual intercourse, delinquent behavior, violent behavior, and substance use, respectively. For each outcome, 4 models are estimated. The first model estimates the effects of immigrant generation by ethnic origin relative to the baseline effect for native non-Hispanic whites, controlling for the age and sex of the adolescent. Model 2 then controls for family structural effects; model 3 adds family process measures; and model 4 explores the influences of cultural adaptation differences. Note that the estimated effects of family context on health behavior outcomes are not of central interest in this analysis; rather the interest is the extent to which health behavior differentials by immigrant generation and ethnic origin are due to the structural, family process, and cultural adaptation factors of the different immigrant and ethnic groups, thereby playing an important role in the assimilation processes observed in Figure 2. The analysis therefore focuses on how the effects of generation by ethnic origin change as sets of theoretical variables are added to the model, and so I only show the coefficients on generation by ethnic origin in the tables. In Appendix III the full results of the model estimating the risk behavior index are shown for interested readers.

23

These tables contain a substantial amount of data analysis and can not be reviewed in detail without making this paper exceedingly long (and very boring). Therefore, in an effort to present only the most important and general findings, I will first discuss results from the risk behavior index model in Table 4, and then simply highlight important differences and key results from the models of specific health risk behaviors in Tables 5-8. Model 1 in Table 4 essentially replicates the patterns shown in Figure 2 by comparing the level of risk behavior for each generation and ethnic group combination with that of non-Hispanic white youth in the 3+ generation, controlling for the age and gender of the adolescent. The results show that there are consistently beneficial and significant effects of first generation immigrant status on risk behavior for all ethnic groups, except Puerto Rican youth. First generation Mexican, Cuban, Central and South American, Chinese, Filipino, other Asian, African and Afro-Caribbean, and European and Canadian youth engage in fewer risk behaviors than the third+ generation of non-Hispanic whites. Because the negative coefficients indicate that immigrant children experience less risk behavior than native-born non-Hispanic white youth, foreign birth is viewed as protective. Island-born Puerto Rican youth do not enjoy these beneficial effects, suggesting a commonality with U.S. native youth culture. Nor is this protective quality conferred to second generation youth. Other noteworthy findings in Model 1 are the greater prevalence of risk behavior among nativeborn Puerto Rican youth with island-born parents and among Mexican youth in the 3+ generation relative to third+ generation non-Hispanic white youth. The second generation Puerto Rican effect reflects the segmented assimilation we observed in Figure 2 in which levels of risk behavior among the second generation have exceeded levels of risk in the mainstream white majority youth culture and are approaching the levels of Puerto Rican youth in the third+ generation. The higher levels of risk behavior among Mexican youth in the 3+ generation reflect the disadvantaged status of Mexican-Americans in U.S. society. Finally, the very low levels of engagement in risk behavior among third+ generation Chinese youth shown in Figure 2 is significantly different from levels of risk behavior among the majority white youth in the third+ generation, demonstrating the apparent uniqueness of Chinese youth in nativeborn families in their resistence to adopt mainstream norms of health behavior in American youth society. In model 2, differences in the structural features of family context are controlled and the effect is rather dramatic. Once the structural conditions of families are held constant, the protective effect of first

24

generation status increases for Mexican, Cuban, Central and South American, and African and AfroCaribbean youth. This suggests that differences in family social and economic conditions operate to increase health risk behavior for immigrant youth from these regions, and when these factors are controlled, the beneficial influence of foreign birth increases. As shown in Table 3, poverty, less educated parents, and urban residence disadvantage first generation youth from Mexico, Cuba, and Central and South America, while non-intact family structures and urban residence disadvantage African and AfroCaribbean youth. The effects of first generation status for Chinese, Filipino, Other Asian, and European and Canadian are relatively unchanged in model 2, indicating that their family social and economic circumstances are not important in explaining health behavior differences from the non-Hispanic white population of youths in the third+ generation. Family structural conditions also operate to increase the likelihood of risk behavior among third+ generation Mexican youth, second generation Puerto Rican youth, and third+ generation African American youth such that once these factors are controlled in model 2, levels of risk behavior among third+ generation Mexican youth and second generation Puerto Rican youth are no longer different from those of third+ generation of non-Hispanic whites, and African American youth are now significantly less likely to engage in risk behavior than non-Hispanic white youth. Model 3 examines the importance of family process differences in mediating the effects of first generation immigrant status. For most ethnic groups, parenting behaviors and parent-child relations are unimportant in understanding the beneficial effect of foreign birth. However, family processes which operate to reduce risk behavior do explain part of the beneficial effect of first generation status for Central and South American youth. This is seen by a reduction in the size of the negative coefficient from -.64 in model 2 to -.55 in model 3. Note that the negative coefficients for first generation Chinese and Filipino youth increase slightly when family process differences are controlled, indicating that these youths experience less parental supervision and monitoring and less close emotional bonds with parents (shown in Table 3), providing less protection from risk behavior than youths in other families. Once parenting behaviors are controlled, the protective effect of first generation status increases for Chinese and Filipino youth. Other noteworthy findings in model 3 are the emergence of several negative effects (i.e., protective) for second generation Central and South American, Chinese, and African and Afro-Caribbean

25

youth. The emergence of negative effects once family process differences are controlled indicates that parenting behaviors are less effective in protecting youth from risk behavior and depress the beneficial impact of second generation immigrant status for these ethnic groups. Conversely, when a positive effect emerges, as it does for third+ generation Mexican youth, this indicates that parenting behaviors are protective in their levels of supervision, monitoring, shared activities and parent-child closeness by reducing involvement in risk behavior such that once these parenting behaviors are held constant, levels of risk behavior exceed those among non-Hispanic whites in the third+ generation. Note finally that parenting behaviors in native-born Chinese families are also protective and explain part of the persistent beneficial effect for third+ generation Chinese youth in reducing their involvement in risk behavior. Cultural adaptation characteristics of immigrant families play an important role in explaining the protective effect of immigrant status documented throughout this study. Once differences in cultural adaptation are controlled in model 4, the beneficial effects associated with first generation status reduce for all ethnic groups. The negative coefficients are reduced by one-third to one-half for Mexican, Central and South American, Filipino, and African and Afro-Caribbean youth, and by one-quarter for Cuban, Chinese, and Other Asian youth. For Mexican youth, the protective effect of first generation status is fully explained by cultural adaptation factors, such as religiosity and social support, as the coefficient is no longer significant in model 4. For the other ethnic groups, however, the coefficients on first generation status remain significant, indicating that the protective effect is due to factors in addition to cultural adaptation that are not examined in this study. Moreover, the only significant beneficial effects associated with second generation status among Central and South American, Chinese, and African and AfroCaribbean youth, are also fully explained by cultural adaptation factors as these coefficients are rendered not significant in model 4 once such factors are controlled. Lastly, cultural adaptation factors are expected to exert little influence on native ethnic families, and thus the lower levels of health risk behavior among third+ generation Chinese youth and the higher levels of health risk behavior among third+ generation Mexican youth remain significant in model 4 relative to the majority white population of youth in the third+ generation. Results from the analyses of specific health behaviors in Tables 5 through 8 reveal similar patterns of findings. The beneficial effects of foreign birth are prevalent across most ethnic groups for all the risk behaviors of sexual activity, delinquency, violence, and substance use. The most consistent beneficial

26

effects of foreign birth are found for Mexican, Cuban, Central and South American, Chinese, and Other Asian youth. While first generation immigrants are universally protected from risk behavior, two minority groups stand out as particularly at risk. Third+ generation Mexican youth and second generation Puerto Rican youth have higher levels of engagement in sexual activity, violence, and delinquency than the mainstream third+ generation youth culture of non-Hispanic whites. The ways in which the mediating factors operate to explain health risk behavior differentials are largely consistent across the behavioral domains examined. Where there are significant beneficial effects of first generation status for an ethnic group, structural conditions tend to depress the beneficial effects primarily for Hispanic origin groups who are most disadvantaged in their social and economic conditions; family processes differences, while important in determining health risk behavior, do not explain many of the differences in behavioral outcomes across generations for ethnic groups; and cultural adaptation factors are most influential in explaining the beneficial effects of first generation status for health risk behavior. For example, cultural adaptation factors fully explain the beneficial effects of first generation status on first sexual intercourse for Mexican, Central and South American, Chinese, and Filipino youth and partially explain the effects for Cuban and Other Asian youth. Cultural adaptation fully explains the beneficial effects of first generation status on delinquency for Central and South American, Chinese, and Other Asian youth and partially explains the effects for Puerto Rican, African and Afro-Caribbean, and European and Canadian youth. Cultural adaptation fully explains the effects of first generation status on violent behavior for Mexican and Cuban youth. Finally, cultural adaptation factors partially explain the beneficial effects of first generation status on substance use for all ethnic groups except Puerto Rico. A few differences in the model results should be highlighted. First, there is less protection of first generation immigrant status from violent behavior. For only 4 ethnic groups (Mexico, Cuba, China, and Other Asia) are first generation youth significantly less likely than third+ generation non-Hispanic whites to engage in violent behavior, and this effect only emerges after controlling for family structural differences for two of the groups (Mexico and Cuba). Second, there is less family context mediation of generation effects for substance use, with the exception of cultural adaptation. Third, there is a fair amount of protection conferred to second generation status from substance use. Second generation Central and South American, Filipino, Other Asian, and African and Afro-Caribbean youth use fewer or less dangerous substances than non-Hispanic white youth in the third+ generation, controlling for

27

structural, family process, and cultural adaptation factors (Table 8). Finally, for certain health risk behaviors, native minority youth engage in more risky behavior than native non-Hispanic white youth. For example, third+ generation Mexican, Filipino, and AfricanAmerican youth are more likely to have had first sexual intercourse; third+ generation Mexican and Central and South American youth are more likely to engage in delinquent behavior; and third+ generation Mexican, Puerto Rican, Filipino, and African American youth are more likely to engage in violent behavior than third+ generation white youth. The mediating factors that explain such effects operate in the following manner. Structural conditions place native minority youth at a disadvantage, increasing their involvement in risk behavior, while family processes advantage third+ generation minority youth, protecting youth from engaging in risk behavior. Cultural adaptation factors have little influence on the third+ generation. Native minority groups most at risk are Mexican, Filipino, African American, and Puerto Rican (both second and third+ generation) youth.

Conclusion This research has provided evidence to support both classical and revisionist theories about the assimilation process of immigrant youth. With greater time and socialization in U.S. institutions, neighborhoods, and youth culture, immigrant children increasingly adopt mainstream behavioral norms regarding health risk and health behavior. Across a broad array of health risk behaviors, both the intragenerational (Figure 1) and intergenerational (Figure 2) perspectives revealed a classical assimilation model of health behavior. When health behavior differentials were examined within ethnic groups, some support for “segmented” assimilation was evident in that Mexican and Puerto Rican youth over the generations tend to adopt the health behavior and norms of their native ethnic group in the U.S. population more so that any other segment of the U.S. population. What is unusual about this assimilation process is that acculturation results in worse health outcomes for immigrant youth, not the typical improved outcomes we observe in educational and achievement studies. The classical pattern of assimilation is well demonstrated such that the gap in health risk behavior between the native-born population and the immigrant population narrows over time in the U.S. and across successive generations, but because immigrant youth engage in less health risk behavior and native-born youth in more risk behavior, the narrowing of the gap through assimilation signifies

28

declining health status and behavior for immigrant youth. The greatest degree of assimilation was displayed by youth from Mexico, Cuba, Puerto Rico, the Philippines, other Asian countries, Africa and the Afro-Caribbean, and Europe and Canada (Figure 2 and Appendix II). With each successive generation of exposure and socialization in American culture, the risk behavior of youth in immigrant families approached the levels manifest in the respective native ethnic population of youth. Other ethnic groups displayed a different pattern of adjustment. For instance, the largest differences in health behavior outcomes among youth from Central and South America were between youth in immigrant families, both first and second generation, and youth in native-born families of the third+ generation. Chinese youth were an exception throughout the analysis in that second generation youth displayed a higher level of risk behavior relative to the third+ generation, who had extremely low involvement in risk behavior. Although family processes played a relatively minor role in explaining health behavior differentials in this study, parenting practices and parent-child relations appear to be especially important in explaining some of the generational differences for Chinese youth. For example, part of the protective effect for third+ generation Chinese operates through favorable parenting behaviors (Tables 4, 5, 7,and 8). For first and second generation Chinese, however, parenting behaviors operate in the opposite direction, providing less protection from involvement in risk behavior. When family process factors are controlled in model 3 for all behavioral outcomes, the significant beneficial effects of first or second generation increase, indicating that less supervision and less close parent-child relations tend to depress the protective effect of immigrant status (Tables 4, 5, 6, 7, and 8). Protective parenting behaviors are especially high in third+ generation families and much lower in second generation families. Measures of parental supervision, monitoring, and parent-child closeness are 3.39, 2.35, and 3.44, respectively, for the third+ generation, compared to 3.12, 1.59, and 2.89 for the second generation and 2.71, 1.95, and 2.96 for the first generation. For most other ethnic groups, there is a decline in parental supervision and monitoring with increasing generation. Chinese families display the exact opposite pattern and the high levels of monitoring by Chinese-American parents explain, in part, the anomalous beneficial effect of being a third+ generation Chinese youth. A clear and consistent finding in this research is the protective nature of first generation immigrant status. First generation youth experience less involvement in risk behavior than second and third+

29

generation youth of their own ethnic group and than the mainstream youth culture of non-Hispanic whites in the third+ generation. These findings for a broad range of health behavior outcomes can be added to the small set of studies showing that immigrant status confers a health advantage for birth outcomes (Hummer et al. 1999; Landale et al. 1999) and for mental health (Vega et al. 1998). Analysis showed that this protective quality is largely due to cultural adaptation characteristics of immigrant families which operate to tie adolescents to the family values and traditions of their own ethnic origins thereby slowing or retarding the process of assimilation, which this research has resoundingly shown is detrimental to their health behavior. The retention of traditional culture buffers and protects immigrant youth from readily adopting the health behavior norms of the native-born population. This research has documented health behavior differentials by immigrant generation and has explored the family context mechanisms that might explain such differences. While family cultural adaptation factors explained a large part of the protective effect associated with foreign birth, for many ethnic groups and across several health behavior outcomes, a significant beneficial effect remains to be explained. Future research should continue this line of inquiry to uncover the mechanisms that operate in other social contexts of adolescents’ lives that might further explain immigrant protection from risk behavior. School, neighborhood, and peer network mechanisms may all play a role in the assimilation process by which immigrant youth lose their health advantage over time and across generations. There are several implications of this research for policies or programs focused on improving the well-being of youth. First, this research can speak to the prevailing stereotypes about immigrant youth and their behavior. Because the new immigration is largely from poor and sometimes chaotic countries, most perceptions of immigrant youth are negative. Immigrants are thought to be unruly, undisciplined, low achieving, and engaging in illegal and dangerous behavior on the fringes of American society. This research has shown that such perceptions are wrong. Despite living in disadvantaged social and economic conditions in the inner-cities of America, most immigrant youth engage in less risky behaviors than native-born youth. Rather than fear the recent immigration to our nation, we should capitalize on the position of immigrant youth as soon as they enter our culture and foster their goals for achievement in the context of less exposure and less experimentation with normal adolescent risk behaviors. Second, the results might lead one to ask how can we reduce the gap between immigrant youth and native-born youth in their health behavior, not by facilitating the assimilation of immigrant youth, but

30

by lowering the health behavior risks among native youth toward the levels of immigrant youth. Here, the results on the importance of the intervening family mechanisms provide some suggestions. Because variation in family processes across generation was minimal, these factors did not explain much of the beneficial effect of foreign birth. Moreover, parental supervision and monitoring tends to decline across generation while parent-child closeness tends to increase. However, we can learn from the Chinese results that when parental supervision and monitoring and parent-child closeness is high, native-born youth are protected from engaging in risk behavior. As families are increasingly headed by a single parent who must work, and two-parent families increasingly include two workers, the supervision of youth activities and behavior has naturally declined. Thus, policies that can help parents monitor their children’s behavior and structure their work lives so that family time and activities can be fostered will enable parents to better protect their children from engaging in risk behavior. The mechanisms of cultural adaptation represent family context factors that tie youth closer to their family’s interests, activities, and channels of social support. These factors are especially salient to immigrant youth because they facilitate socialization and social capital in the cultural traditions of the family’s ethnic origins. These factors are also more relevant to immigrant families as they break social ties in their country of origin, come to a foreign culture, may not be able to speak English, and therefore seek out institutional and social support within their own culture. Nevertheless, we can import these notions to native families by encouraging greater involvement by families in community institutions such as churches, neighborhood groups, sports clubs, and volunteer organizations. This will increase family time together as well as social networks of adults who can provide social support and emotional support to America’s native youth, which in turn helps to monitor their behavior and foster their well-being. Finally, this research showed that native minority youth are especially at risk of engaging in health risk behavior, particularly Mexican, Puerto Rican, and African American youth. The structural conditions of native minority families are part of the explanation for their greater risk, as they face higher poverty rates and live largely in urban areas. Family processes and cultural adaptation factors are relatively inconsequential for youth in native minority families. Thus, policies concerned about the well-being of minority youth should focus on improving the structural position of youth from native minority families and changing family process and cultural mechanisms so that they can have positive consequences for the well-being of minority youth. Native minority families should be targeted for policies that might

31

facilitate greater parental supervision and monitoring of youth behavior, such as structured work hours so parents can be home for dinner and in the evening. Moreover, community programs that engage families in community organizations and institutions will provide additional sources of social control and social capital for minority youth.

32

Table 1. Health Risk Behavior by Immigrant Generation and Race/Ethnicity of Adolescents in Third+ Generation (means) __________________________________________________________________________________________________________________ Third+ Generation _______________________________________ First Second Non-Hisp Non-Hisp Gen Gen White Black Asian Hispanic __________________________________________________________________________________________________________________ Health Risk Behavior

Total

Ever had sex

26.9

34.3

35.2

58.2

25.1

41.9

38.3

Age/1st intercoursea

14.9

14.8

14.7

13.8

15.0

14.0

14.5

3 or more delinquent acts

10.7

20.6

16.2

12.5

22.1

24.7

16.2

3 or more acts of violence

15.6

21.7

19.0

28.4

23.1

30.0

21.0

Ever tried smoking

40.7

55.4

62.3

49.3

45.0

59.8

58.2

8.7

16.7

26.6

9.1

13.9

20.7

21.5

15.9

31.1

33.4

22.0

21.0

33.4

30.4

7.4

15.5

19.7

10.3

10.0

19.7

17.1

14.8

27.9

28.5

27.1

27.3

39.6

27.9

Used marijuana last month

5.4

14.0

14.7

14.7

14.3

20.7

14.4

Hard drugs

4.6

11.1

14.9

4.1

10.8

16.5

12.3

Smoke regularly Drink alcohol 1+/month Drunk 1+/month Ever used marijuana

Risk behavior index

.74

1.20

1.27

1.32

1.13

1.56

N 1,716 2,998 10,167 4,267 187 1,122 __________________________________________________________________________________________________________________ a Based on the sample of adolescents who ever had sex (N = 7,504). Note: Distributions are based on weighted data.

33

1.25 20,457

Appendix II. Health Risk Behavior by Immigrant Generation and Ethnic Origin (means) ____________________________________________________________________________________________________________ Central-South Mexico Cuba America Puerto Rico 1 2 3+ 1 2 3+ 1 2 3+ 1 2 3+ ____________________________________________________________________________________________________________ Health Risk Behavior Ever had sex

31.9

32.5

40.9

28.1

29.8

39.6

36.6

33.9

39.9

23.1

42.8

48.0

Age/1st intercoursea

15.0

14.9

14.0

15.7

14.8

15.2

14.6

14.5

13.7

14.5

13.7

14.2

3+ delinquent acts

14.8

23.8

22.8

20.8

19.5

9.6

7.6

18.8

33.8

.4

23.7

22.7

3+ violent acts

17.5

28.4

30.4

15.2

26.7

1.7

18.7

19.2

26.1

40.0

35.9

32.6

Ever tried smoking

40.8

55.2

56.9

27.9

51.9

76.8

38.2

51.1

61.0

54.8

59.9

66.3

9.2

10.2

16.4

5.3

13.8

10.7

8.3

15.9

25.1

23.3

22.1

30.7

18.3

31.0

37.3

14.7

32.3

38.0

13.4

24.2

25.8

10.5

24.9

25.3

9.8

14.8

21.6

5.7

11.6

36.7

7.4

10.9

15.3

.8

14.2

15.3

18.1

26.6

40.1

8.8

24.4

34.5

9.6

23.5

40.0

24.7

29.7

37.3

6.6

12.2

20.8

.4

12.9

32.3

3.7

9.1

18.0

12.8

15.9

21.4

6.9 .91

13.2 1.26

17.5 1.52

2.9 .78

16.3 1.18

25.1 1.28

3.1 .82

9.1 1.09

17.7 1.56

Smoke regularly Drink alcohol 1+/month Drunk 1+/month Ever used marijuana Used marijuana last month Hard drugs Risk behavior index

7.0 1.01

N 321 805 630 228 257 24 285 249 139 40 ____________________________________________________________________________________________________________

34

9.1 1.48 240

9.3 1.67 300

Appendix II (cont.). Health Risk Behavior by Immigrant Generation and Ethnic Origin ____________________________________________________________________________________________________________ China Philippines Other Asia Africa/Afro-Caribbean 1 2 3+ 1 2 3+ 1 2 3+ 1 2 3+ ____________________________________________________________________________________________________________ Health Risk Behavior Ever had sex

15.7

23.1

12.3

19.6

30.4

41.2

10.8

33.2

27.2

44.8

41.1

58.2

Age/1st intercoursea

14.9

15.7

15.3

15.2

15.0

15.2

15.0

14.0

14.7

14.2

14.5

13.8

3+ delinquent acts

5.7

22.0

17.0

19.7

23.8

14.2

7.6

20.9

32.7

3.2

11.4

12.5

3+ violent acts

6.3

6.9

9.8

10.2

15.0

30.0

11.9

22.5

37.0

19.5

18.4

28.4

31.8

35.2

30.1

49.1

50.1

61.9

36.7

58.5

57.4

38.2

42.7

49.2

9.5

19.6

5.4

7.5

13.1

26.9

7.1

18.8

13.4

5.3

8.8

9.1

11.8

27.0

17.9

16.4

17.6

20.0

11.2

29.9

34.1

20.3

25.7

22.0

Drunk 1+/month

1.8

9.0

8.6

4.8

10.0

4.0

4.8

17.5

26.3

10.8

8.1

10.3

Ever used marijuana

7.9

24.5

7.7

21.4

29.9

29.1

6.4

28.9

40.7

21.1

23.1

27.0

Used marijuana last month

4.6

20.0

5.6

8.5

16.3

13.9

2.7

12.5

24.9

5.8

14.4

14.7

Hard drugs

4.8

20.6

7.6

5.7

8.9

4.3

3.3

10.8

23.8

3.6

2.8

4.1

1.10

1.29

1.04

1.32

Ever tried smoking Smoke regularly Drink alcohol 1+/month

Risk behavior index

.48

.95

.60

.72

.44

1.12

1.69

.80

N 126 179 54 299 286 67 224 300 114 109 ____________________________________________________________________________________________________________

35

225

4,261

Appendix II (cont.). Health Risk Behavior by Immigrant Generation and Ethnic Origin ____________________________________________________________________________________________________________ Europe/Canada 1 2 3+ ____________________________________________________________________________________________________________ Health Risk Behavior Ever had sex

22.0

35.9

35.2

Age/1st intercoursea

14.3

15.4

14.7

7.9

19.0

16.2

3+ violent acts

12.9

16.2

19.0

Ever tried smoking

56.8

61.8

62.3

Smoke regularly

13.6

24.7

26.6

Drink alcohol 1+/month

27.2

40.6

33.4

Drunk 1+/month

12.1

22.0

19.7

Ever used marijuana

31.9

31.9

28.5

Used marijuana last month

10.8

16.8

14.7

4.3

11.2

14.9

3+ delinquent acts

Hard drugs Risk behavior index

.75

1.24

1.27

N 79 440 10,148 ____________________________________________________________________________________________________________ a These outcomes are based on the sample of adolescents who ever had sex (N = 8,260). Note: Distributions are based on weighted data.

36

Appendix I. Health Risk Behavior of First Generation Immigrants by Time in the U.S. (means) ______________________________________________________________________________________ Years in the U.S. 0-5 6-10 11+ Total ______________________________________________________________________________________ Health Risk Behavior Ever had sex

22.8

26.0

35.8

27.1

Age/1st intercourse

15.1

15.3

14.2

14.9

7.3

10.1

17.4

10.8

3+ violent acts

13.4

14.5

17.3

14.7

Ever tried smoking

33.7

42.0

49.2

40.2

7.9

6.5

12.9

8.8

13.3

14.4

22.9

16.1

Drunk 1+/month

4.5

6.1

13.6

7.3

Ever used marijuana

8.6

11.7

23.7

13.5

Used marijuana last month

1.8

3.9

11.4

4.9

Hard drugs

2.2

4.3

8.8

4.6

1.10

.74

3+ delinquent acts

Smoke regularly Drink alcohol 1+/month

Risk behavior index

.55

.70

N 645 430 435 1,510 ______________________________________________________________________________________ Note: Distributions are based on weighted data.

1

Table 2.

Mean Structural, Family Process, and Cultural Adaptation Characteristics by Generation and Race/Ethnicity of Third +Generation (N=20,457) Third + Generation Non-Hispanic White

First Generation

Second Generation

2 Biological Parents

56.0

61.0

58.8

Single Mother

17.4

18.4

Income < Poverty

27.8

Parents Education < High School

Non-Hispanic Black Asian

Hispanic

Total

27.9

71.3

44.1

53.3

15.4

43.1

15.1

24.3

20.4

14.8

10.8

31.4

1.7

16.0

15.2

31.8

22.7

7.6

12.7

2.1

16.4

11.8

Urban Residence

88.0

76.8

42.7

56.4

34.6

74.4

52.5

Parental Supervision

3.01

2.98

3.01

2.74

2.99

2.95

2.96

Parental Monitoring

2.16

2.04

1.81

1.96

2.33

1.84

1.88

Parent-Child Closeness

3.14

3.20

3.24

3.28

3.29

3.24

3.24

Church Attendance

2.04

1.94

1.88

2.19

1.91

1.80

1.94

Importance of Religion

2.39

2.28

2.24

2.64

2.29

2.27

2.31

Speak English at Home

31.4

68.7

99.9

99.8

94.0

92.4

92.3

Social Support

4.03

3.99

4.01

3.99

4.11

4.02

4.01

Live with Relatives

20.8

16.8

7.6

22.9

17.0

14.9

12.0

Note: Means are based on weighted data

2

Table 5. Parameter Estimates of Generation and Ethnic Origin Effects on First Sexual Intercourse 1

2

3

4

(baseline)

(+structural)

(+family process)

(+cultural adapt)

Mexico 1

-.70**

-.91**

-.88**

-.61

Mexico 2

-.24

-.28

-.28

-.11

.51**

.39*

.47**

.54**

Cuba 1

-.75**

-1.13**

-1.14**

-1.04**

Cuba 2

-.34

-.72**

-.73**

-.64**

Cuba 3+

-.61

-.26

-.24

-.22

Puerto Rico 1

-.62

-.69*

-.58

-.33

Puerto Rico 2

.76**

.59*

.68**

.83**

.55

.44

.41

.45

Central-South America 1

-.39

-.57**

-.44**

-.19

Central-South America 2

-.01

-.004

-.09

.08

.29

.21

.17

.23

China 1

-1.38*

-1.24*

-1.35*

-.88

China 2

-.73

-.48

-.76**

-.44

China 3+

-1.01**

-.82*

-.57

-.67

Philippines 1

-1.18**

-1.01**

-1.03**

-.59

Philippines 2

-.24

-.04

-.12

.05

.27

.33

.37

.59*

Other Asia 1

-2.27**

-2.28**

-2.21**

-1.67**

Other Asia 2

-.20

.02

-.001

.11

Other Asia 3+

-.11

.10

.03

.04

Africa/Afro-Caribbean 1

-.15

-.21

-.28

.14

Africa/Afro-Caribbean 2

.06

.07

.12

.29

1.06**

.79**

-.84**

.95**

Europe/Canada 1

-.77

-.72

-.70

-.47

Europe/Canada 2

.01

.11

.09

.14

_

_

_

_

-9315.99

-9023.72

-8725.48

-8573.56

0.00

0.00

0.00

0.00

.16

.19

.22

.23

Mexico 3+

Puerto Rico 3+

Central-South America 3+

Philippines 3+

Africa/Afro-Caribbean 3+

Europe/Canada 3+

Log L prob Pseudo R2

Notes: Estimates are based on weighted data with robust standard errors. ** p < .01 * p < .05

5

Table 6. Parameter Estimates of Generation and Ethnic Origin Effects on Delinquent Behavior (N= 17,065) 1

2

3

4

(baseline)

(+structural)

(+family process)

(+cultural adapt)

Mexico 1

-.05

-.15

-.18

.16

Mexico 2

.43**

.36**

.32**

.52**

Mexico 3+

.40**

.26*

.29*

.35**

Cuba 1

.04

.03

.08

.33*

Cuba 2

.15

.15

.15

.41**

-.20

-.23

-.13

-.10

Puerto Rico 1

-.69**

-.83**

-.75**

-.53**

Puerto Rico 2

.77**

.53**

.56**

.64**

.35

.10

.03

.08

Central-South America 1

-.34**

-.42**

-.35**

-.04

Central-South America 2

.27

.12

.04

.22

.64**

.54*

.52*

.54**

China 1

-.26

-.38

-.52*

-.39

China 2

.26

.16

-.04

.11

China 3+

.14

-.01

.12

.08

Philippines 1

.09

-.13

-.21

.05

Philippines 2

.30

.14

.10

.20

-.20

-.34

-.28

-.11

Other Asia 1

-.22*

-.34**

-.33**

-.14

Other Asia 2

.18

.11

.09

.14

Other Asia 3+

.83

.70

.61

.65

Africa/Afro-Caribbean 1

-.62**

-.86**

-.90**

-.64**

Africa/Afro-Caribbean 2

-.29*

-.45**

-.40**

-.32**

-.09

-.16**

-.15**

-.10*

Europe/Canada 1

-.46**

-.54**

-.51**

-.38**

Europe/Canada 2

.11

.04

.05

.07

-

-

-

-

F stat

14.31

14.13

32.06

36.19

prob

0.00

0.00

0.00

0.00

.03

.05

.10

.14

Cuba 3+

Puerto Rico 3+

Central-South America 3+

Philippines 3+

Africa/Afro-Caribbean 3+

Europe/Canada 3+

Pseudo R2 Notes:

Estimates based on weighted data with robust standard errors. 6

** p < .01

* p < .05

7

Table 7.

Parameter Estimates of Generation and Ethnic Origin Effects on Violent Behavior (N= 17,065) 1

2

3

4

(baseline)

(+structural)

(+family process)

(+cultural adapt)

Mexico 1

.05

-.32**

-.33**

-.11

Mexico 2

.42**

.16

.14

.24

Mexico 3+

.72**

.53**

.55**

.59**

Cuba 1

.09

-.17*

-.15*

-.02

Cuba 2

.58**

.35*

.35*

.50**

-.64**

-.46**

-.40*

-.39

Puerto Rico 1

.65

.56

.60

.73

Puerto Rico 2

1.02**

.69*

.72**

.75**

.85**

.60*

.55*

.56*

Central-South America 1

-.03

-.27

-.21

-.04

Central-South America 2

.11

-.02

-.07

.03

Central-South America 3+

.26

.16

.15

.16

China 1

-.51**

-.51**

-.60**

-.62**

China 2

-.29

-.26

-.40

-.35

-.51**

-.48*

-.40*

-.44*

Philippines 1

-.15

-.16

-.20

-.07

Philippines 2

-.13

-.14

-.16

-.11

.11

.09

.14

.28*

Other Asia 1

-.29*

-.37**

-.36**

-.34*

Other Asia 2

.22

.25

.24

.24

Other Asia 3+

.68

.71

.66

.68

Africa/Afro-Caribbean 1

.07

-.11

-.13

.02

Africa/Afro-Caribbean 2

.05

-.04

-.01

.02

.60**

.42**

.44**

.47**

Europe/Canada 1

-.29

-.28

-.27

-.20

Europe/Canada 2

-.11

-.13

-.13

-.13

-

-

-

-

F stat

63.82

59.34

67.13

72.59

prob

0.00

0.00

0.00

0.00

Cuba 3+

Puerto Rico 3+

China 3+

Philippines 3+

Africa/Afro-Caribbean 3+

Europe/Canada 3+

8

R2 Notes:

.09

.11

Estimates based on weighted data with robust standard errors. ** p < .01 * p < .05

9

.13

.16

Table 8.

Parameter Estimates of Generation and Ethnic Origin Effects on Substance Use (N=16,921) 1

2

3

4

(baseline)

(+structural)

(+family process)

(+cultural adapt)

Mexico 1

-.70**

-.73**

-.71**

-.38*

Mexico 2

-.17

-.14

-.16

.04

.10

.03

.06

.11

Cuba 1

-.95**

-.99**

-.97**

-.75**

Cuba 2

-.14

-.16

-.16

.05

.27

.37

.40

.43

Puerto Rico 1

-.38

-.37

-.30

-.06

Puerto Rico 2

-.05

-.15

-.11

-.005

.12

.03

-.03

.01

Central-South America 1

-.92**

-.94**

-.86**

-.56**

Central-South America 2

-.34**

-.35**

-.41**

-.25*

-.01

-.06

-.09

-.07

China 1

-.82**

-.80**

-.90**

-.80**

China 2

-.39

-.33

-.49*

-.37

China 3+

-.53**

-.57**

-.45*

-.47*

Philippines 1

-.66**

-.68**

-.73**

-.50**

Philippines 2

-.26*

-.26*

-.31**

-.22*

-.26

-.31

-.30*

-.17

Other Asia 1

-.91**

-.86**

-.85**

-.69**

Other Asia 2

-.30**

-.24**

-.26**

-.22**

.48

.48

.41

.41

Africa/Afro-Caribbean 1

-.89**

.93**

-.97**

-.76**

Africa/Afro-Caribbean 2

-.53**

-.55**

-.51**

-.47**

Africa/Afro-Caribbean 3+

-.41**

-.51**

-.50**

-.44**

Europe/Canada 1

-.42**

-.41**

-.39**

-.33*

Europe/Canada 2

-.10

-.09

-.09

-.08

-

-

-

-

33.54

42.61

64.93

52.53

Mexico 3+

Cuba 3+

Puerto Rico 3+

Central-South America 3+

Philippines 3+

Other Asia 3+

Europe/Canada 3+

F stat

10

prob R2 Notes:

0.00

0.00

0.00

0.00

.07

.10

.16

.19

Estimates based on weighted data with robust standard errors. ** p < .01 * p < .05

11

Table 4.

Parameter Estimates of Generation and Ethnic Origin Effects on Risk Behavior Index (N=16,782) 1

2

3

4

(baseline)

(+structural)

(+family process)

(+cultural adapt)

Mexico 1

-.48**

-.60**

-.59**

-.32

Mexico 2

-.06

-.11

-.13

.03

.33**

.20

.23*

.27**

Cuba 1

-.56**

-.68**

-.66**

-.51**

Cuba 2

-.08

-.18

-.17

.01

Cuba 3+

-.24

-.11

-.07

-.04

Puerto Rico 1

-.19

-.22

-.16

.05

Puerto Rico 2

.31*

.11

.15

.24

.39

.20

.14

.17

Central-South America 1

-.54**

-.64**

-.55**

-.32**

Central-South America 2

-.13

-.20

-.26*

-.13

.27

.19

.16

.18

China 1

-.78**

-.76**

-.83**

-.63**

China 2

-.30

-.26

-.44*

-.29

China 3+

-.44**

-.47**

-.34*

-.35**

Philippines 1

-.62**

-.66**

-.71**

-.42*

Philippines 2

-.13

-.13

-.18

-.07

.03

-.02

.0002

.16

Other Asia 1

-.85**

-.86**

-.84**

-.62**

Other Asia 2

-.16

-.11

-.13

-.08

.55

.56

.48

.50

Africa/Afro-Caribbean 1

-.62**

-.75**

-.80**

-.54**

Africa/Afro-Caribbean 2

-.26

-.32

-.28*

-.23

.03

-.12*

-.11*

-.04

Europe/Canada 1

-.57**

-.56**

-.54**

-.44**

Europe/Canada 2

-.05

-.04

-.05

-.03

-

-

-

-

F stat

37.35

39.19

54.83

66.05

prob

0.00

0.00

0.00

0.00

.09

.12

.19

.23

Mexico 3+

Puerto Rico 3+

Central-South America 3+

Philippines 3+

Other Asia 3+

Africa/Afro-Caribbean 3+

Europe/Canada 3+

R2

12

Notes:

Estimates based on weighted data with robust standard errors. ** p < .01 * p < .05

13

Appendix III.

Full Model Estimates of Generation and Ethnic Origin Effects on Risk Behavior Index (N=16,782)

Intercept

1

2

3

4

(baseline)

(+structural)

(+family process)

(+cultural adapt)

-1.92**

-2.09**

.89**

2.52**

Age

.19**

.19**

.11**

.11**

Male

.30**

.31**

.37**

.30**

Mexico 1

-.48**

-.60**

-.59**

-.32

Mexico 2

-.06

-.11

-.13

.03

.33**

.20

.23*

.27**

Cuba 1

-.56**

-.68**

-.66**

-.51**

Cuba 2

-.08

-.18

-.17

.01

Cuba 3+

-.24

-.11

-.07

-.04

Puerto Rico 1

-.19

-.22

-.16

.05

Puerto Rico 2

.31*

.11

.15

.24

.39

.20

.14

.17

Central-South America 1

-.54**

-.64**

-.55**

-.32**

Central-South America 2

-.13

-.20

-.26*

-.13

.27

.19

.16

.18

China 1

-.78**

-.76**

-.83**

-.63**

China 2

-.30

-.26

-.44*

-.29

China 3+

-.44**

-.47**

-.34*

-.35**

Philippines 1

-.62**

-.66**

-.71**

-.42*

Philippines 2

-.13

-.13

-.18

-.07

.03

-.02

.0002

.16

Other Asia 1

-.85**

-.86**

-.84**

-.62**

Other Asia 2

-.16

-.11

-.13

-.08

.55

.56

.48

.50

Africa/Afro-Caribbean 1

-.62**

-.75**

-.80**

-.54**

Africa/Afro-Caribbean 2

-.26

-.32

-.28*

-.23

.03

-.12*

-.11*

-.04

-.57**

-.56**

-.54**

-.44**

Mexico 3+

Puerto Rico 3+

Central-South America 3+

Philippines 3+

Other Asia 3+

Africa/Afro-Caribbean 3+ Europe/Canada 1

1

Europe/Canada 2

-.05

-.04

-.05

-.03

--

--

--

--

Step family

.32**

.21**

.19**

Single Mother

.39**

.36**

.26**

Single Father

.57**

.37**

.28**

Other family form

.48**

.45**

.37**

--

--

--

Parents’ educ < H.S.

.26**

.22**

.21**

Parents’ educ = H.S. grad

.18**

.14**

.12**

Parents’ educ = some college

.15**

.13**

.11**

--

--

--

.04

-.03

-.03

-.04**

-.03*

-.03*

.13**

.10*

.09*

.06

.03

.02

--

--

--

West

.02

-.01

-.02

South

-.06

-.05

.0001

Midwest

-.01

-.04

-.04

--

--

--

Parental supervision

-.10**

-.08**

Parental monitoring

-.06**

-.05**

Shared activities

-.12**

-.04**

Parent-child closeness

-.38**

-.11**

Europe/Canada 3+

2 biological parents

Parents’ educ = college or more Parents’ educ missing Number of siblings Urban residence Suburban residence Rural residence

Northeast

Church attendance

-.08**

Importance of religion

-.10**

Frequency of prayer

-.01

Religion missing

-.50**

Speak Spanish at home

-.20

Speak other language at home

-.26**

Speak English at home

--

2

Social support

-.54**

Live with other relatives

-.02

F stat

37.35

39.19

54.83

66.05

prob

0.00

0.00

0.00

0.00

.09

.12

.19

.23

R2 Notes:

Estimates based on weighted data with robust standard errors. ** p < .01 * p < .05

3

References Alba, R. and V. Nee. 1997. “Rethinking Assimilation Theory for a New Era of Immigration.” International Migration Review 31:826-873. Amato, P.R. and A. Booth. 1997. A Generation at Risk: Growing Up in an Era of Family Upheaval. Cambridge, MA: Harvard University Press. Bautista-Hayes, D.E. 1990. “Latino Health Indicators and the Underclass Model: From Paradox to New Policy Models.” Chicano Studies Research Center, UCLA, unpublished manuscript. Bearman, Peter S., Jo Jones, and J. Richard Udry. 1997. The National Longitudinal Study of Adolescent Health: Research Design [WWW document]. URL: http://www.cpc.unc.edu/projects/addhealth/design.html. Borjas, George J., and Marta Tienda. 1987. “The Economic Consequences of Immigration.” Science 235:645-651 Bradby, D. 1992. Language characteristics and academic achievement: A look at Asian and Hispanic eighth graders in NELS:88. Washington, DC: U.S. Government Printing Office. Coleman, J. S. 1988. "Social Capital in the Creation of Human Capital." American Journal of Sociology 94:S95-S120. Earls, F. 1993. “Health Promotion for Minority Adolescents: Cultural considerations.” Pp. 5872 in Promoting the Health of Adolescents: New Directions for the Twenty-first Century, edited by S.G. Millstein, Ph.D., A.C. Petersen, Ph.D., and E.O. Nightingale, M.D. New York, Oxford: Oxford University Press. Earls, F., J.I. Escobar, S.M. and Manson. 1990. “Suicide in Minority Groups: Epidemiological and Cultural Perspectives.” Pp. 571-598 in S.J. Blumenthal and D.J. Kupfer, eds., Suicide over the Life Cycle. Washington, D.C.: American Psychiatric Association Press. Elder, Glen H., Jr. 1980. Family Structure and Socialization. New York: Arno Press. Elder, G. H., Conger, R. D., Foster, E.M., and Ardelt, M. 1992. “Families Under Economic Pressure.” Journal of Family Issues 13:5-37. Farley, R. 1996. The New American Reality: Who We Are, How We got Here, Where We Are Going. New York: Russell Sage Foundation. Fernandez-Kelly MP. 1995. “Social and Cultural Capital in the Urban Ghetto: Implications for the Economic Sociology and Immigration.” In The Economic Sociology of Immigration: Essays on Networks, Ethnicity, and Entrepreneurship, ed. A. Portes, pp. 213-47. New York: Russell Sage Foundation.

4

Fuligni, A. J. 1997. “The Academic Achievement of Adolescents from Immigrant Families: The Roles of Family Background, Attitudes, and Behavior.” Child Development 68: 261-273. Furstenberg, F. F., Jr. and M. E. Hughes. 1995. “Social Capital and Successful Development Among At-Risk Youth. Journal of Marriage and the Family 57:580-592. Gans, H. J. 1992. “Second Generation Decline: Scenarios for the Economic and Ethic Futures of Post-1965 American Immigrants.” Ethnic and Racial Studies 15: 173-192. Gordon, M.M. 1964. Assimilation in American Life: The Role of Race, Religion, and National Origins. New York: Oxford University Press. Hernandez, D.J. 1993. America’s Children: Resources from Family, Government, and the Economy. New York: Russell Sage. Hirschi, T. 1969. Causes of Delinquency. Berkeley: University of California Press. Hirschman, C. 1996. “Studying Immigrant Adaptation from the 1990 Population Census: From Generational Comparisons to the Process of ‘Becoming American’.” Pp. 54-81 in The New Second Generation. A. Portes, ed. New York: Russell Sage Foundation. Hogan, D. P. and E. M. Kitagawa. 1985. “The Impact of Social Status, Family Structure, and Neighborhood on the Fertility of Black Adolescents.” American Journal of Sociology 90(4):825-55. Hogan, D. P. and D. J. Eggebeen. 1997. “Demographic Change and the Population of Children: Race/Ethnicity, Immigration, and Family Size.” Pp. 311-327 in Indicators of Children’s Well-Being, edited by R. M. Hauser, B.V. Brown and W. R. Prosser. New York: Russell Sage Foundation. Hummer, R.A., M. Biegler, P.B. De Turk, D. Forbes, W. P. Frisbie, Y. Hong, and S.G. Pullum. 1999. “Race/Ethnicity, Nativity, and Infant Mortality in the United States.” Social Forces 77:1083-1118. Jasso, G. and M. R. Rosenzweig. 1990. The New Chosen People: Immigrants to the United States. New York: Russell Sage Foundation. Jensen, L. and Y. Chitose. 1996. “Today’s Second Generation: Evidence from The 1990 Census.: In The New Second Generation, Pp. 82-107, edited by Alejandro Portes. New York: Russell Sage Foundation. Kao, G. and M. Tienda. 1995. “Optimism and Achievement: The Educational Performance of Immigrant Youth.” Social Science Quarterly, 76: 1-19. Klerman, L. V. 1993. “The Influence of Economic Factors on Health-Related Behaviors in Adolescents.” Pp.38-57 in S.G. Millstein, A. C. Petersen, and E. O. Nightingale, eds., Promoting the Health of Adolescents: New Directions of the Twenty-first Century. New York: Oxford University Press. 5

Lieberson, S. 1980. A Piece of the Pie: Blacks and White Immigrants since 1880. Berkeley: University of California Press. Liska, A. E. and M. D. Reed. 1985. “Ties to Conventional Institutions and Delinquency: Estimating Reciprocal Effects.” American Sociological Review 50:547-60. Maccoby, E. E., and S. Martin. 1983. “Socialization in the Context of the Family: Parent-Child Interaction.” Pp. 1-102 in Handbook of Child Psychology, edited by E.M. Heatherington. New York: Wiley. Maccoby, E. E. 1992. “The Role of Parents in the Socialization of Children: An Historical Overview.” Developmental Psychology. 28:1006-17. Majoribanks, K. 1987. “Ability and Attitude Correlates of Academic Achievement: FamilyGroup Differences.” Journal of Education Psychology 79:171-78 McLanahan, S. S. 1988. "Family Structure and Dependency: Early Transitions to Female Household Headship." Demography 25:1-16. McLanahan, S. and G. Sandefur. 1994. Growing Up with a Single Parent: What Hurts, What Helps. Cambridge, MA: Harvard University Press. McLoyd, V. 1990. “The Impact of Economic Hardship on Black Families and Children: Psychological Distress, Parenting, and Socioemotional Development.” Child Development 61:311-346. Mendoza, F.S., S.J. Ventura, B. Valdez, R. Castillo, L.E. Saldivar, K. Baisden, and R. Martorelli. 1990. “Selected Measures of Health Status for Mexican-American, Mainland Puerto Rican, and Cuban-American Children.” Journal of the American Medical Association 265:227-232. National Research Council. 1993. Losing Generations: Adolescents in High Risk Settings. Washington, DC: National Academy Press. Perez, L. 1994. “The Households of Children of Immigrants in South Florida: An Exploratory Study of Extended Family Arrangements.” International Migration Review 28, 4: 736747. Perry, C. L., S. H. Kelder, and K. A. Komro. 1993. “The Social World of Adolescents: Family, Peers, Schools, and the Community.” Pp. 73-96 in S. G. Millstein, A. C. Petersen, and E. O. Nightingale, eds., Promoting the Health of Adolescents: New Directions of the Twentyfirst Century. New York: Oxford University Press. Portes, A. and M. Zhou. 1993. “The New Second Generation: Segmented Assimilation and its Variants.” Annals of the American Academy of Political and Social Sciences. M. Zhou. 1993. “The New Second Generation: Segmented Assimilation and its Variants.” Annals of the American Academy of Political and Social Sciences 530: 74-96. Portes, A. 1995. “Children of Immigrants: Segmented Assimilation and Its Determinants. Pp. 6

248-280 in The Economic Sociology of Immigration: Essays on Networks, Ethnicity, and Entrepreneurship, edited by A. Portes. New York: Russell Sage Foundation. Portes, A., ed. 1996. The New Second Generation. New York: Russell Sage Foundation. Portes, A. and R. G. Rumbaut. 1996. Immigrant America: A Portrait. Berkeley, CA: University of California Press. Reimers, D. M. 1992. Still the Golden Door: The Third World Comes to America, 2nd edition. New York: Columbia University Press. Rumbaut, R. G. 1994. “The Crucible Within: Ethnic Identity, Self-esteem, and Segmented Assimilation among Children of Immigrants.” International Migration Review 28:748794. Rumbaut, R. G. 1997. “Assimilation and Its Discontents: Between Rhetoric and Reality.” International Migration Review, 31, 4, Pp. 923-960. Rumbaut, R. G. 1998a. “Transformations: The Post-Immigrant Generation in an Age of Diversity.” Paper presented at the annual meetings of the Eastern Sociological Society, Philadelphia, PA , March 21. Rumbaut, R. G. 1998b. “Assimilation and Its Discontents: Ironies and Paradoxes.” Chapter in Becoming American, America Becoming , edited by J. DeWind, C. Hirschman, and P. Kasinitz. New York: Russell Sage Foundation, forthcoming. Sampson, R. and W. Groves. 1989. “Community Structure and Crime: Testing Social Disorganization Theory.” American Journal of Sociology, 94: 774-802. Smith, J. P. and B. Edmonston (Eds.). 1997. The New Americans: Economic, Demographic, and Fiscal Effects of Immigration. Washington, DC: National Academy Press. Sowell, T. 1930. Ethnic America: A History. New York: Basic Books, Inc. Thornton, A. and D. Camburn. 1987. “The Influences of the Family on Premarital Sexual Attitudes and Behavior.” Demography 24:323-40. Tienda, M. and Z. Liang. 1994. “Poverty and Immigration in Policy Perspective.” Pp. 330-364 in Confronting Poverty Prescriptions for Change, edited by S. H. Danziger, G. D. Sandefur and D. H. Weinberg. New York: Russell Sage Foundation. Udry, J. R. 1988. “Biological Predispositions and Social Control in Adolescent Sexual Behavior.” American Sociological Review 53:709-22. Waters, M. C. 1996. “Ethnic and Racial Identities of Second-Generation Black Immigrants in New York City.” In The New Second Generation, Edited by A. Portes. New York: Russell Sage Foundation, Pp. 171-196. Wilson, W. J. 1987. The Truly Disadvantaged: The Inner-City, the Underclass, and Public 7

Policy. Chicago: The University of Chicago Press. Wolfe, B.L. 1994. “Reform of Health Care for the Nonelderly Poor.” Pp.253-288 in S. H. Danziger, G. D. Sandefur, and D. H. Weinberg, eds., Confronting Poverty: Prescriptions for Change. New York: Russell Sage Foundation. Wu, L. L. and B. B. Martinson. 1993. “Family Structure and the Risk of a Premarital Birth.” American Sociological Review 58:210-32. Zhou, M. and C. L. Bankston III. 1996. “Social Capital and the Adaptation of the Second Generation: The Case of Vietnamese Youth in New Orleans” in The New Second Generation, edited by A. Portes. New York: Russell Sage Foundation. Zhou, M. 1997. “Growing Up American: The Challenge Confronting Immigrant Children and Children of Immigrants.” Annual Review of Sociology, 23: 63-95.

8

Endnotes 1

Schools were stratified by region, urban/rural status, school sector type (public/private/parochial), ethnic mix, and size. School dropouts are not included in the sample. 2

Some high schools spanned grades 7 through 12 and therefore served as their own “feeder” school, so the “pair” was in fact a single school. There are 134 discrete schools in the study. 3

Although self-reports of delinquent and risky behavior are subject to under-reporting bias, Add Health data attempted to minimize such bias by collecting these data in the audio-CASI, selfadministered section of their questionnaire. 4

Some youth checked multiple ethnic categories or checked “other.”

5

The data quality of measures of immigrant generation and ethnic origin is exceptional. A remarkable number of sources in Add Health report on the nativity of youth and parents, permitting reliability and validity checks. Nativity of adolescents is reported by the adolescent in the in-school interview and in the Wave I in-home interview and by parents in the parental interview. In addition, a question asking whether the youth was born abroad as a U.S. citizen prevents an erroneous classification of foreign-born. The nativity of parents is reported by the adolescent in the in-school interview and in the in-home interview for both resident parents and biological nonresident parents if the adolescent is not living with both biological parents, and by the parent responding to the parent questionnaire. Adolescents report on their ethnic backgrounds in both the in-school and in-home interviews and are permitted to check multiple backgrounds. Parents also report on their ethnic background with a similar question asking them to check all backgrounds that apply. Finally, in the Wave I in-home interview, adolescents who report that they were not born in the U.S. and were not born U.S. citizens abroad report on their country of birth, and similarly, report on their parents’ country of birth if they were not born in the U.S. Thus, comparisons between country of birth and ethnic background are possible. Research on immigration and immigrant families has been handicapped by the lack of precision or detailed data on such measures (Zhou 1997; Hogan and Eggebeen 1997). Add Health makes considerable advances in this regard. 6

The non-Hispanic white and non-Hispanic black categories on this race and ethnic variable correspond to the categories European/Canadian and African/Afro-Caribbean categories on the ethnic origin variable for the third+ generation. 7

Poverty status is only shown in descriptive tables because it did not explain any additional variance or add to the fit of multivariate models of risk behavior over and above the contribution of other structural measures. 8

Shared activities and parent-child closeness is measured for both resident mothers and resident fathers. Adolescents who live with only one parent use that parent-specific measure while youth who live with two parents are assigned the average of the mother and father scores. 9

African and Afro-Caribbean background can not be specifically determined for youth in the third+ generation, however, the native population of non-Hispanic blacks probably represents prior immigration from these regions. Therefore, first and second generation youth from Africa and the Afro-Caribbean who also indicated they were non-Hispanic and of the black race are compared with the native population of non-Hispanic blacks in the third+ generation. In a similar 9

manner, first and second generation youth from Europe and Canada who are non-Hispanic and of the white race are compared with non-Hispanic white adolescents in the third+ generation. Because the native populations of non-Hispanic blacks and non-Hispanic whites represent a more heterogeneous ethnic composition than the immigrant populations of youth with African and Afro-Caribbean ancestry and European and Canadian ancestry, respectively, these comparisons are looser than the other ethnic group comparisons.

10