Vietnamese Ethnicity and Adolescent Substance Abuse more

vietnamese ethnicity and adolescent substance abuse: evidence for a community-level approach Carl L. Bankston Ill Department of Sociology, Loyola University, New Orleans, Louisiana, USA This paper examines how membership in a highly concentrated Vietnamese-American community affects the drug and alcohol use of Vietnamese-American secondary school students. It suggests that research on adolescent substance abuse has had a tendency to concentrate on the family environment and the peer group. For this ethnic group, however, the present study finds that involvement in the ethnic community has a strong negative effect on drug and alcohol abuse, both directly and indirectly, through lessening the likelihood that adolescents will have substance-abusing friends. Vietnamese language use rs found to be an especially influential aspect of ethnicity. It is suggested that research on adolescent substance abuse should place more emphasis on community-level explanations, such as the effect of ethnic and other sorts of social groups that surround individuals and families. Theoretical explanations of adolescent substance abuse have generally sought the etiology of drug and alcohol use in two environments: the family and the peer group. It has been recognized that the predictors of substance abuse vary in strength and significance among racial and ethnic groups, but relatively little attention has been devoted to the ethnic group itself as an environment that may affect adolescent behavior. Rece1ved 23 June 1994; accepted 29 June 1994. The author w1shes to thank Dr Stephen ]. Caldas, Dr Mm Zhou, Msgr Dommie Luong, Ms. Susan Egnew, and Mr Tom R1gsby Address correspondence to Carl l. Bankston Ill, 312 Shrewsbury Court, Jefferson, LA 70121. E-ma1l: bankston@mus1c loyno edu. Dev1ant Behav1or: An fnterdiSCiplmary journal, 16 59-80, 1995 Copynght © 1995 Taylor & FranCIS 0763-9625195 $10 00 59 + 00 60 C. L. BANKSTON Ill The present paper examines how membership in a highly concentrated Vietnamese-American community affects the drug and alcohol use of Vietnamese-American secondary school students. Specifically, it considers whether a high degree of involvement in this Vietnamese ethnic community should be seen as a major influence on this type of delinquent behavior. If so, this example suggests that researchers in the area of substance abuse may be giving insufficient attention to the wider webs of social relations of which the ethnic community is one type. REASONS FOR STUDYING VIETNAMESE ADOLESCENTS Vietnamese youth offer an illuminating subject for study for anumber of reasons. First, they are one of the newest large ethnic groups in the United States. Prior to 1975, there were virtually no Vietnamese-Americans. By 1990, as a result of the exodus of refugees from Southeast Asia following the Vietnam war, the number of Vietnamese in the U.S. had increased to over 614,000. By the year 2000, the Vietnamese are expected to be the third largest Asian American group, outnumbered only by Chinese and Filipinos (Lee 1991, p. 55). Thus, in a very short time, Vietnamese ethnicity has become a matter of consequence in American society, and it is worthwhile to see how this ethnicity affects social outcomes, such as propensity toward substance abuse. The second reason for adopting the Vietnamese as the subject of study is that they have shown a propensity toward settling in distinctive ethnic communities. Although the government and private voluntary agencies that were primarily responsible for resettling Indochinese in the U.S. initially attempted to scatter the migrants around the country, secondary migration and family sponsorship of new arrivals from Vietnam has resulted in noticeable ethnic residential concentrations (Montero 1979; Haines et al. 1981). The community in which the students in this paper reside is particularly interesting example of a Vietnamese residential concentration. This community, located in eastern New Orleans, contains over 4,500 Vietnamese in a neighborhood that radiates out from a Vietnamese Catholic church; and Vietnamese shops, restaurants, and other places of business predominate. Studying young people who live in such a dense ethnic cluster offers a unique opportunity to see how membership in a distinctive ethnic group may reduce the tendency towards problematic behavior on the part of young people. VIETNAMESE ETHN/CITY 61 The third reason for looking at Vietnamese young people is that youth is a major characteristic of this group. The median age of Vietnamese in the United States in the 1990 census was only 25.7, compared with 35.6 for whites and 29.4 for blacks; 34.1% of Vietnamese were under 18, compared with 23.9% of whites and 32.0% of blacks. Because substance abuse is a major issue facing young people, it is particularly worthwhile to look at this issue in the context of a group that contains such a large proportion of young people. THEORETICAL CONSIDERATIONS Sources of Adolescent Substance Abuse Bahr et al. (1993), in a paper on family and religious influences on substance abuse, note that studies of substance abuse generally approach the issue using one of two general theoretical frameworks of deviant behavior. Social control theory takes conformity, rather than deviance, as problematic and seeks to find mechanisms that force people to conform and prevent them from engaging in the forms of deviance they might enjoy if all social constraints were removed. "Rather than asking 'Why do people take drugs,' the question is 'Why don't people take drugs?' " (Bahr et al. 1993, p. 447). Social control theory, in other words, emphasizes constraint, the different ways in which social ties and structures keep adolescents and others from indulging in tempting, but socially undesirable, forms of behavior. Social control may be exercised through official sanctions (Meier et al. 1984), religious organizations and other institutions (Kandel 1980), and parental and family supervision of behavior. However social controls are imposed, this explanation of substance abuse assumes exposure to drugs or alcohol and seeks to determine which controls are most effective to prevent the transformation of exposure into experience. The major alternative to a social control view of adolescent substance abuse is the social learning view. The social learning model, in the words of Walter et al. (1993), "posits that young people initiate drug use because the behavioral norms, values, and beliefs of their primary reference groups encourage such behavior" (Walter et al. 1993, p. 975). Thus those who associate with drug or alcohol users tend also to use drugs or alcohol. 62 C. L. BANKSTON Ill Researchers have generally found more evidence of the influence of social learning factors than of social control factors, but there is some indication of interaction and complementarity between the two sets of factors. Using a survey of urban minority high school students in New York City, Walter et al. (1993) found that the social learning model was a better predictor of adolescent substance abuse than individual-level social psychological factors such as stress/ strain or disaffiliation. Meier et al. (1984) found that peer substance abuse had the greatest effect on use of marijuana by adolescents, but that sanction threats did seem to have short-term effects on those in groups of users, and that sanctions appeared to inhibit those not in groups of users from joining those groups. Bahr et al. (1993), recognizing the potentially complementary nature of the social control and social learning perspectives, examined the effects of indicators of both social learning and social control and found that adolescent substance abuse is significantly related to peer and family substance abuse and to parental monitoring, with peer substance abuse as the major influence. Religion, in the study of Bahr et al. (1993), has relatively little effect. Krohn (1974) has also offered evidence that peer groups are more important, as a direct influence, than family relations; this paper found that ties between parents and children have little effect on substance abuse when ties to other adolescents are taken into account. In short, social control and social learning have been recognized as the two primary means of influencing adolescent drug and alcohol use, with the family seen as a source of both control and learning and the peer group seen primarily as a source of social learning. In an attempt to integrate empirical findings on adolescent substance abuse into a causal model, Simons et al. (1988, p. 245) therefore used as the primary foci of their model"parenting factors, peer group influences, and adolescent substance abuse." Elliott et al. (1985) and Elliott et al. (1989) give recognition to community and other broader environmental influences on substance use and other forms of delinquency, but they do not treat these as direct sources of control and social learning. Elliott et al. (1985) see both social disorganization and inadequate family socialization as exogenous variables with indirect effects on drug and alcohol use and other forms of delinquency. Similarly, Elliott et al. (1989, pp. 138-9) see social disorganization in the secondary environment affecting drug and alcohol use indirectly by affecting "involvement with and exposure to delinquent others." VIETNAMESE ETHNJCITY 63 Ethnicity as a Social Environment In a discussion of the concept of "social capital," James Coleman considers how comparatively closed and close-knit communities, in particular ethnic communities, influence the behavior of their young. Closed systems of social networks in a community, in his view, allow parents " ... to establish norms and reinforce each other's sanctioning of the children" (Coleman 1990, p. 318). In this way, the efficacy of social control may vary from one ethnic group to another, depending on the social ties within each group. Although Coleman's terminology here inclines one to see ethnicity as a form of social control, it is easy to see how the "closed systems of social networks" he describes may also function as means of social learning-as mechanisms for instilling, as well as enforcing, patterns of behavior. The "dense set of associations" provided by a relatively closed ethnic community (Coleman 1990, p. 316) limits the exposure of young people to members of the larger society as primary references and may thereby help young people in minority groups avoid the dangers of the mainstream society. Coleman has referred to these sets of ethnic network associations as "information channels" (Coleman 1989), and it is in the nature of such channels to provide particular types of knowledge and particular interpretations of social realities. Recent studies have tended to support the view that minority ethnicity is negatively related to substance abuse. "Most studies of drug use," observes one group of researchers, "report that alcohol and drug use are more prevalent among white than among black or Asian American adolescents," with Asian Americans the least likely of all groups to engage in substance abuse (Catalano et al. 1992, p. 208). Windle (1990) has found that both gender and ethnicity are significant predictors, with adolescent males reporting higher levels of substance abuse than females, and non-blacks reporting higher levels than blacks. Researchers have found that the effect of ethnicity is not limited to differences among ethnic groups in levels of family and community supervision or exposure to peer substance use. Vega et al. (1993), for example, found that risk factors (including family pride, family substance problems, family smoking, psychosocial factors, perceptions of peer use, perceptions of peer approval, and inclinations toward deviant behavior) had different predictive values for different ethnic groups. These differences in predictive values were 64 C. L. BANKSTON Ill ascribed, somewhat vaguely, to "cultural differences" (Vega et al. 1993, p. 189). Similarly, another group of researchers found that "... families and family process may not exert the same influence across ... racial groups. To the extent that cultural differences are rooted in family traditions, the differences produced in this study may reflect cultural differences" (Catalano et al. 1992, p. 216). Although a comparison of different ethnic groups is beyond the scope of the present study, it is possible to examine the effect of Vietnamese ethnicity on substance abuse in a manner that will not involve introducing the rather ill-defined term "culture" as deus ex machina to resolve all theoretical complications. If ethnicity is regarded as the web of social relations in which an individual is involved by virtue of ethnic group membership, then it becomes possible to conceptualize individual group members as, in a sense, having different degrees of ethnicity. Some individuals are more bounded by the closed system of ethnic ties than others. Some utilize the distinctive language of the group in all situations, so that all information must flow through ethnic channels of communication, while others speak the language of the larger society almost exclusively, receiving much the same types of information as non-group members. More subjectively, some members of a minority identify with the minority strongly and have an intense commitment to it; others identify with the majority society. Using this idea that there are different degrees of ethnicity, or involvement in the complex of ethnic relations, enables one to examine the effect of Vietnamese ethnicity on substance abuse by asking whether adolescents who are more wrapped up in Vietnamese ethnic relations, patterns of communication, identification, and commitments report more or less substance use than other Vietnamese adolescents. RESEARCH QUESTIONS These theoretical considerations of sources of substance abuse and the effect of ethnicity on substance abuse lead to several specific research questions. 1. Does Vietnamese ethnicity, in the sense of involvement with the ethnic group, promote or hinder substance abuse? 2. If Vietnamese ethnicity has an effect, to what aspects of ethnicity (subjective identification with the group, channels of com- VIETNAMESE ETHNICITY 65 munication, distinctive family traditions, ethnic network ties) can this effect be ascribed? 3. Tying the issues of sources of substance abuse and ethnicity together, if ethnicity has an influence on adolescent alcohol and drug use, does it appear to exercise this influence directly, indirectly by its effect on family relationships, or indirectly by its effect on peer group associations? DATA AND METHODS The data are drawn from a survey of Vietnamese high school students at two high schools located near the Vietnamese community in eastern New Orleans and at a public high school for honors students in February, 1994. The two schools located near the community are biracial in their makeup (one is approximately 77% black and 20% Vietnamese, the other is approximately 85% black and 12% Vietnamese), and they contain most of the high school students who live in the Vietnamese neighbhorhood. The third school, the high school for honors students, has a majority white student body, with a large black minority, and only 15 Vietnamese students. This third school was included in the survey because it was thought that students attending an elite school might have less exposure to drug and alcohol use than those attending other schools, and that therefore students at this third school might have experiences of the larger society different from those of students at the other schools. The survey encompassed the entire Vietnamese population at the schools who were present on the day of the survey (N = 402). Students were assured of the complete anonymity of the survey, and each student was allowed, upon completion of the questionnaire, to place his or her own questionnaire in the stack, in order to emphasize the fact that no answers could be connected to any respondent. Since these two schools contain most of the students, the 402 students included constituted over 75% of all 9th-, 10th-, 11th-, and 12th-grade students in the Vietnamese neighborhood. The variable indicating the history of the students' own substance abuse was created from questions about inebriation and drug use. With regard to alcohol, students were asked "Have you ever been drunk? If so, how many times?" A "no" answer to the first question was coded as "0." A "yes" answer to the first question, followed 66 C. L. BANKSTON Ill by the answer that the respondent had been drunk 1 or 2 times was coded as "1." If the respondent indicated having been drunk 3 to 5 times, the response was coded as "2." If the respondent indicated having been drunk 6 to 10 times, the response was coded a "3." If the respondent indicated having been drunk more than 10 times, the response was coded as "4." The question about drug use was similar to the one about alcohol. Students were asked "Have you ever used drugs? If so, how many times?" This item was coded in the same manner as the item dealing with alcohol use. For both items, responses such as "not much" or "some" were coded as "2" and responses such as "every day," "all the time," "beaucoup," and "can't count" were coded as "4." Alcohol use to the point of drunkenness and drug use were significantly correlated (r= .681 ), so the two items were added together to create a scale of substance abuse, in which "0" indicated no reported substance abuse and "8" indicated heavy use of both drugs and alcohol. Self-reported data on incidence of deviant behavior should always be used carefully, since there is an inherent tendency toward underreporting. It is believed that the care taken to assure respondents of the confidentiality of the survey has minimized this tendency, but even if the estimates of substance abuse provided by this survey are conservative, this should not present a problem as long as underreporting is not systematically linked to other variables in question. To look at substance abuse among the friends of respondents, respondents were asked "Have any of your friends ever been drunk? If so, how many friends?" Those with no friends who had ever been drunk were coded as "0," those with 1 or 2 friends who had been drunk were coded as "1," those with 3 to 5 friends who had been drunk were coded as "2," those with 6 to 10 friends who had been drunk were coded as "3," and those with more than 10 friends who had been drunk were coded as "4." Responses such as "lots" and "all of them" were coded as "4." Several authors have used concepts such as "family coherence" and "family structure" to indicate the amount of control families may exercise over adolescent behavior (e.g., Krohn 1974, Simons et al. 1988, Bahr et al. 1993). As an indicator of family structure, respondents who I ived alone or with non-parents were coded as "0," those living with only one parent were coded as "1," and those living with both parents were coded as "2." Since parental control may also vary according to how much time parents spend VIETNAMESE ETHN/CITY 67 in the home, respondents were asked to indicate how many hours parents spent working. Those living with two parents who both worked more than 40 hours per week and those living with only one parent or guardian who worked more than 40 hours per week were coded as "1," and all others were coded as "0." In this way, the present study can test "latchkey children" explanations of substance abuse (i.e., that children using drugs are inadequately controlled by parents working excessively long hours). The involvement of parents in their children's educations and in the community was also believed to be important, although this variable might be viewed both from the perspective of social control and from the perspective of social learning. Parents who are involved with schools and other aspects of their children's lives can keep closer watch over their children's activities and have more opportunity to direct those activities. On the other hand, children in families that are highly involved in community organizations and associations tend to be themselves drawn into the community networks. As an indicator of the involvement of parents, respondents were asked if their parents were members of the local Vietnamese Parent-Teacher Association and if their parents were members of any other local Vietnamese organizations. Ethnicity may be seen as having a number of different dimensions. One important aspect of ethnicity is the extent to which social ties are bounded by the ethnic group. Therefore, respondents were asked what proportions of their friends were Vietnamese and what proportions of their friends belonged to other ethnic groups. This question was used to create an "ethnic social circles" variable, which ranged from having social circles composed entirely of nonVietnamese to having social circles composed entirely of Vietnamese. Language is another notable aspect of ethnicity, one that is especially interesting because language determines the flow of information. Respondents were asked about language use in a number of situations, about how well they spoke Vietnamese, and how often they used Vietnamese with parents, siblings, and friends. All of these items were combined to create an index of Vietnamese language use. Still another aspect of ethnicity is self-identification. Respondents were asked, "How do you identify yourself?" Possible responses were "American," which indicated no distinctive ethnic self-identification; "Vietnamese American," which indicated identification 68 C. L. BANKSTON Ill with both the minority and the dominant society; and "Vietnamese," which indicated a primary identification with the minority. These were coded from "0" (no minority self-identification) to "2" (a primary minority self-identification). The one respondent who wrote in "Asian American" was coded as "1." Most sociology texts classify "endogamy" as a chief characteristic of an ethnic group. In order to measure respondents' commitment to endogamy, respondents were asked, "Do you want to marry someone who is Vietnamese or someone who is not Vietnamese?" Possible responses were "I definitely want to marry someone who is not Vietnamese," "I prefer to marry someone who is not Vietnamese," "It isn't important whether I marry a Vietnamese or not," "I prefer to marry someone who is Vietnamese," and "I definitely want to marry someone who is Vietnamese." Participation in religious institutions may be expected to affect levels of substance abuse through social control (religious sanctions and rules) or as a result of social learning (exposure to other churchgoers). To determine frequency of attendance, respondents were asked, "How often do you attend your church or temple?" There were six possible responses, ranging from "never" to "more than one time every week." To judge how the effect of church attendance should be interpreted, it is necessary to look at the relationship between church attendance and other variables of interest. Gender is introduced as a control variable, because it has been noted that males are more likely to engage in substance abuse than females. Because this study concerns an immigrant population, recency of arrival, indicated by year of arrival in the U.S. or U.S. birth, is an essential control variable; recently arrived young people may have experiences quite different from those who have been in this country through most of their lives. For a preliminary examination of patterns of relations among all variables, this study looks first at all zero-order correlations. Next, reported level of substance abuse is regressed on all other variables to find significant predictors of substance abuse. Then, alcohol use by friends is regressed on other variables to see what factors lead young people to associate with substance users. Factor analysis is used to find patterns of relations among variables, in particular to see how the church attendance and parental membership variables should be interpreted. Finally, after major patterns are identified, the factor scores derived from the factor analysis are regressed on sub- VIETNAMESE ETHNICITY 69 stance abuse to deal with problems of multicollinearity and to provide a relatively straightforward comparison of the major factors. RESULTS Table 1 gives zero-order correlations among all variables included in this analysis. Living in a two-parent situation shows significant positive relations with parental membership in community organizations, with all of the measures of ethnic involvement, and with church attendance. It shows a negative zero-order correlation with substance abuse. The "latchkey" variable, having parents or a parent work over 40 hours per week, has no significant relation to any of the other variables. Parental membership in ethnic community organizations appears to be a key variable; it shows significant positive relationships with all aspects of ethnicity and with church attendance. Parental membership has significant negative relations with both respondents' own substance abuse and with alcohol use by friends, as well as with recency of arrival, which is probably explained by the fact that recent arrivals have had less time to become involved in community organizations. Having a social circle consisting of predominantly Vietnamese friends is significantly related to all of the other aspects of ethnicity and to church attendance. Female gender has a significant positive relationship with church attendance and significant negative relationships with substance abuse and with alcohol use by friends. Self-identification as Vietnamese is significantly and positively related to all of the other aspects of ethnicity and to church attendance, and negatively related to substance abuse and alcohol use by friends. Use of the Vietnamese language shows a positive association with all of the other aspects of ethnicity and with church attendance. Language use is the only aspect of ethnicity that is significantly related to recency of arrival. It is negatively related to both of the substance abuse variables. Church attendance, in addition to being positively related to all aspects of ethnicity and to other variables already mentioned, has expected negative associations with both substance abuse variables. Recency of arrival has no significant zero-order associations, other than the aforementioned negative relation with parental mem- " 0 TABLE 1 Zero-Order Correlations among Variables under Consideration Variable 2 3 4 5 6 7 8 9 10 11 12 .003 .184.. .186** .072 .167.. .136** .185** .009 .153** -.028 -.126** 2 3 4 5 6 7 8 9 10 11 .004 -.076 -.012 -.054 -.013 -.078 .067 .028 -.054 -.007 .188** .032 .265** .246** .251** -.123** .224** - .193** -.212** -.036 .484** .399** .335** .035 .332** -.100* -.218** -.063 -.012 .099* .019 .006 -.173** -.210** .483** .238** .171** .350** -.140** -.227** .388** .001 .435** -.317** -.378** .294** -.294** - .376** -.005 -.049 -.019 - .104** - .235** .607** Note. 1-Living in a two-parent, related family; 2-Parents work over 40 hours per week; 3-Parental membership in Vietnamese PTA and other community organizations; 4-Extent to which friendship circles are Vietnamese; 5-Female gender; 6-Ethnic self-description; 7-Use of Vietnamese language; 8-Church attendance; 9-Recency of arrival in U.S.; 10-degree of commitment to marrying a Vietnamese; 11-Number of friends who abuse alcohol; 12-Respondent's own use of alcohol and drugs. *p < .OS (two-tailed) ••p < .001 (two-tailed) VIETNAMESE ETHNICJTY 71 bership and positive relation with use of the Vietnamese language. Commitment to in-group marriage, in addition to associations mentioned,is negatively related to both of the substance abuse variables. As expected, alcohol use by friends and substance abuse are strongly and positively related. To summarize, this is a highly interrelated set of variables, with both two-parent family structure and parental membership strongly correlated with each other and with the indicators of respondents' ethnic involvement. This complex of zero-order correlations suggests that a multivariate regression equation with these variables may be troubled by collinearity. Nevertheless, although this method may make estimates of effects on the dependent variable overly conservative, using regression is helpful because it can help us judge which are the strongest determinants of substance abuse when a!! variables are included. Table 2 gives unstandardized and standardized regression coefficients for determinants of substance abuse, employing three models. Model 1 presents coefficients only for the family variables and for alcohol abuse to compare the effects of family and peer group. Model 2 presents coefficients for a!! variables, including the demographic variables of recency of arrival and gender and the ethnicity variables. Model 3 leaves out alcohol abuse by peers to clarify the issue of how the family or the ethnic community may affect substance abuse by control exposure to deviant others. Model 1 tells us that whereas family structure and parental involvement in community organizations have significant negative effects on substance abuse, when we do not control for the wider social context, substance abuse is most strongly correlated with substance abuse in the peer group (13 = .594). However, looking at Model 2, we see that when we introduce the ethnicity variables and church attendance, which were significantly associated with both two-parent family structure and parental community involvement, the family variables not only become insignificant, they also show standardized coefficients that are lower than any of the other independent variables except recency of arrival. Again, the strongest effect of Vietnamese adolescent substance abuse is alcohol abuse on the part of peers. However, gender and frequency of church attendance also show significant negative influences, and Vietnamese language use has a negative effect that is significant if we employ a one-tailed test. When alcohol abuse by friends is left out of the equation, in Model 3, the negative effect of Vietnamese language TABLE 2 Unstandardized and Standardized Regression Coefficients of Effects on Vietnamese Adolescent Substance Abuse. Model1 B Recency of arrival Female gender Two-parent family Overwork by parent Parental membership Alcohol abuse by friends Commitment to in-group Marriage Ethnic friendship circles Ethnic self-description language use Church attendance Constant Model2 B .001 -.410** .001 -.114* -.030 B .003 - .699** -.049 -.076 -.182 Model3 .031 - .195** - .454* .063 - .237** .708** - .090* .023 -.085** .594** -.149 .026 -.091 .583** -.109 -.049 -.188 -.037• -.208* -.010 -.028 -.066 .010 -.033 .489** -.051 -.046 -.055 -.094• -.144* -.024 -.024 -.234 -.094** -.350** 5.580** .277 -.012 -.023 -.069 -.233** -.242** .539* .392 3.269** .471 R2 *p < .05 (two-tailed) **p < .001 (two-tailed) •p < .05 (one-tailed) VIETNAMESE ETHNICITY 73 TABlE 3 Unstandardized and Standardized Regression Coefficients of Effects on Alcohol Abuse among Friends. Modell B B Model2 Regency of arrival Female gender Two-parent family Overwork by parents Parental membership Commitment to in-group marriage Ethnic friendship circles Ethnic self-description language use Church attendance Constant .003 -.496** -.014 -.118 -.422** -.003 -.051 - .180** -.173 -.177 -.157 .144 .41 -.078 - .097** - .245** 1.461** 3.966** .042 - .104** .042 -.079 -.068 .081 .047 -.027 -.285** -.202** R' **p < .OS (two-tailed) **p < .001 (two-tailed) .035 .188 use is almost as great as that of church attendance, and greater than that of female gender. Table 3 presents regression coefficients for determinants of alcohol abuse on the part of friends, to examine how family and ethnicity variables may decrease the probability of substance abuse by controlling friendship groups. Model 1 gives the effects of family variables alone. Only parental membership in ethnic community organizations significantly affects alcohol use in the adolescent peer group, suggesting that although the structure of individual families has little impact on this aspect of friendship groups, family involvement in the wider community does help to control the kinds of contacts made by young Vietnamese. Model 2 includes all of the variables under consideration. Comparing this table to Table 2, we see that the determinants of having friends who abuse alcohol are almost identical to the determinants 74 C. L. BANKSTON Ill of being a substance abuser oneself. Gender, church attendance, and use of the Vietnamese language all have significant negative effects on the number of friends who abuse alcohol an adolescent Vietnamese has. Of the three significant variables in this table, use of the Vietnamese language has the strongest effect on associating with alcohol abusers (13 = - .285). Looking again at Table 2, we see that the effect of use of the Vietnamese language increased greatly when alcohol use on the part of friends was left out of the equation. Language use, as a means of channeling communication, appears to exercise the greatest influence of all these variables on the composition of adolescents' peer groups with respect to substance abuse, and it appears to affect substance abuse on the part of Vietnamese adolescents chiefly by determining with whom they will associate. Correlation analysis has allowed us to do some preliminary examination of patterns of relations among variables. The complex of correlations suggests that collinearity may be resulting in excessively conservative estimates of the effects of individual variables. Also, the high R-squares that we see in Table 1 indicate that the variables combined account for a great deal of variance, although only a few of these variables are significant. This is evidence that, especially for the highly correlated ethnicity variables, individual effects on substance abuse are being diminished by the complex of interrelations among them. Moreover, the meaning of some of the variables requires greater clarification. Pi1rental involvement in ethnic organizations is dearly a family characteristic, but it is also a characteristic that acts as a bridge between individual families and the community. Should we account whatever effect it may have to parental activism in children's lives or to the social integration of the family in the ethnic group? Similarly, church attendance may hinder drug and alcohol use through normative teachings, through the association of church attendance with strong families, or by acting as an aspect of involvement in the ethnic group, because almost all of the respondents are members of a Vietnamese Catholic church or Vietnamese Buddhist temples. In order to establish meaningful patterns among the variables, and to deal with the issues of multicollinearity, all variables except for the demographic variables sex and recency of arrival have been VIETNAMESE ETHNICITY 75 TABLE 4 Rotated and Unrotated Factor Loadings of Relevant Variables. Factor 1 Rotated Two-parent familiy Overwork by parents Parental membership Commitment to in-group marriage Ethnic friendship circles Ethnic self-description Language use Church attendance .412 -.053 .503 .670 .707 .728 .752 .640 Unrotated .381 -.146 .479 .663 .737 .727 .759 .647 Factor 2 Rotated .346 .900 .289 .136 -.109 -.003 -.009 -.008 Unrotated .307 .915 .240 .069 -.184 -.078 -.068 -.074 entered into an exploratory factor analysis, the results of which are given in Table 4. Table 4 reports both rotated and unrotated factor loadings, using oblique rotation. Table 4 shows that our variables divide into two factors. The highest loadings on factor 1 are the different aspects of ethnicity and church attendance. The fact that church attendance loads so highly on the ethnicity factor helps us to interpret the meaning of church attendance. In Durkheimian fashion, religion here appears as an expression of community, specifically the ethnic community, and frequency of church attendance is a measure of involvement in the ethnic group. It will be noted that the variable with the highest factor loading on ethnicity is use of the Vietnamese language, which we have noted is best understood as a means of channeling communication into ethnic networks. Factor 2 concerns the nature of respondents' families. All of the family variables show positive loadings, on this factor and the ethnicity variables show negative loadings. In a manner consistent with our interpretation of church attendance as a form of involvement with the ethnic community rather than as an expression of individual family beliefs, church attendance, like the indicators of ethnic involvement, has a low negative loading on the family factor. Two of the measures of family characteristics, parental membership in ethnic community organizations and living in a two-parent family, load on both factors. For parental membership, it is fairly clear why this should be the case: Parental involvement serves as 76 C. L. BANKSTON Ill TABLE 5 Effects of Ethnicity and Family Factors on Substance Abuse Modell Factor Ethnicity factor Family factor Alcohol abuse by friends Constant Model2 8 - .444** -.058 -.554** -.029 .631** .174 .452 -.530** -.015 .530** 8 - .847** -.110 .941** .195 Rz **p < .001 an important link between the individual family and the community and thus may be seen as an indicator of both factors. It may help to understand the dual character of the two-parent family variable if we point out that an overwhelming predominance of two-parent families is actually a distinctive characteristic of the Vietnamese community. According to the 1990 U.S. Census, 81.2% of Vietnamese families in the census tract in which the Vietanmese neighborhood is located (Orleans Parish Census Tract 17.29) are two-parent families, compared with 75.5% of white families and 53.3% of black families; only 5.8% of Vietnamese families in the tract are headed by females, compared with 17.1% of white families and 41.8% of black families. The interpretation of the high positive loading of overwork by parents on the family factor is somewhat less straightforward. However, it appears that working more than 40 hours per week should be understood as an indicator of parental dedication to the material well-being of the family rather than as an indicator of parental neglect of children; long working hours, when we control for the ethnicity variables, are statistically associated with parental involvement and with family structure. It has been mentioned that the regression equations discussed above are giving overly conservative estimates of effects on substance abuse, as a result of collinearity. Because the variables in question fit into two factors, the factor of the individual family and that of the ethnic community, we can combine these variables using factor scores generated by the factor analysis procedure. Table 5 gives regression coefficients of the individual family factor and the VIETNAMESE ETHNICITY 77 TABLE 6 Effects of Ethnicity and Family Factors on Alcohol Abuse by Friends. factor Ethnicity factor Family factor Constant B R' **p < .001 -.463** -.129 1.21s•• .086 -.290** -.081 ethnic community factor, with substance abuse as the dependent variable. Model 1 gives regression coefficients without alcohol abuse on the part of friends included in the equation. Involvement in the ethnic group has a strong negative effect on substance abuse. The individual family factor shows the expected negative sign, but it is not statistically significant. Model 2 includes alcohol abuse by friends, enabling us to compare the effects of peer group, family, and ethnic community. Substance abuse in the peer group and the composite measure of ethnic involvement have identical standardized coefficients, but in opposite directions. Table 6 presents regression coefficients of both factors on the number of friends who abuse alcohol. Involvement in the ethnic community, the ethnicity factor, has a strong and significant effect on having friends who abuse alcohol. The factor of individual family type, again, has the expected negative sign but lacks statistical significance. Not only does involvement in the Vietnamese community directly lessen the likelihood of substance abuse, it also has a significant indirect effect of lowering the probability that Vietnamese high school students will associate with those who engage in excessive use of intoxicants. DISCUSSION AND CONCLUSION We have seen strong evidence that Vietnamese ethnicity, in the sense of both subjective and participatory involvement in the ethnic community, inhibits substance abuse on the part of Vietnamese high school students. Although all aspects of ethnicity are interconnected, the role of language use is notable. This finding offers sup- 78 C. L. BANKSTON Ill port for the argument that ethnicity chiefly affects this type of behavior by communication channels that direct the exposure of individuals to different forms of behavior. Although alcohol abuse by friends, in a manner consistent with prior research, is a key influence on drug and alcohol abuse, the surrounding social environment appears to be even more important. Not only can a tightly knit community, such as the Vietnamese community under examination here, counterbalance the effect of exposure to substance-abusing peers, it can also lessen the likelihood of exposure to those peers. It appears, then, that the tendency of research on adolescent substance abuse to concentrate on peer groups and family relations has led to an underemphasis on a crucial part of the explanation: the community environment in which peer groups and families exist. Indeed,the evidence given above suggests that those aspects of family life that are most influential in combating substance abuse in this particular ethnic concentration are the aspects that connect individual families to the community. Vietnamese ethnicity appears to affect adolescent drug and alcohol use less through the characteristics of individual families than through providing an ethnic communication network that helps insulate group members from the dangers of the larger society. The role of church attendance is particularly enlightening on this issue, since church attendance is one of the greatest single effects on substance abuse. In the correlation analysis, it has been seen that church attendance is positively related to all aspects of ethnicity, and the factor analysis has provided evidence that church attendance itself should be seen as one of the qualities that links an individual to a set of ethnic ties. These observations can provide a useful basis for further investigation of the sources of adolescent substance abuse. Many of the differences between ethnic and other social groups that have been vaguely ascribed to "cultural traits" may, in fact, be due to variations in degrees of social integration among ethnic and other social groups that provide differing degrees of constraint and support to young people. To understand how these results may apply to other ethnic groups, it is important that researchers consider ethnicities as concrete, definable communities to understand how membership in sets of social relations, as opposed to mere ascription to social categories, can produce desirable or undesirable behavioral outcomes. VIETNAMESE ETHNICITY 79 REFERENCES Bahr, Stephen J. Ricky D. Hawks, and Gabe Wang. 1993. "Family and Religious Influences on Adolescent Substance Abuse" Youth & Society 24:443-465. Catalano, Richard F., Diane Morrison, Elizabeth A. Wells, Mary R. Gilmore, Bonita lritani, and J. David Hawkins. 1992. "Ethnic Differences in Family Factors Related to Drug Initiation." journal of Studies on Alcohol 53:208-217. Coleman, James. 1989. "Social Capital in the Creation of Human Capital." American Sociological Review 94:98-9 (supplement). Coleman, James. 1990. Foundation of Social Theory. Cambridge: The Belknap Press of Harvard University Press. Elliott, Delbert S., David Huizinga, and Suzanne S. Ageton. 1985. Explaining Delinquency and Drug Use. Boulder: Sage. Elliott, Delbert S., David Huizinga, and Scott Menard. 1989. Multiple Problem Youth: Delinquency, Substance Use, and Mental Health Problems. new York: Springer-Verlag. Haines, David, Dorothy Rutherford, and Patrick Thomas. 1981. "Family and Community among Vietnamese Refugees." International Migration Review 15:311-8. Kandel, D. B. 1980. "Drug and Drinking Behavior among Youth." Annual Review of Sociology 6:235-285. Krohn, M. D. 1974. "An Investigation of the Effect of Parental and Peer Association on Marijuana Use: An Empirical Test of Differential Association Theory." Pp. 75-89 in Crime and Delinquency: Dimensions of Deviance, edited by M. Reidel and T. P. Thornberry. New York: Praeger. Lee, Joann Faung Jean. 1991. Asian Americans: Oral Histories of First to Fourth Generation Americans (rom China, the Philippines, Japan, India, the Pacific Islands, Vietnam, and Cambodia. New York: The New Press. Meier, Robert F., Steven R. Burkett, and Carol A Hickman. 1984. "Sanctions, Peers, and Deviance: Preliminary Models of a Social Control Process." The Sociology Quarterly 25:67-82. Montero, Darrel, 1979. Vietnamese Americans: Patterns of Resettlement and Socioeconomic Adaptation in the United States. Boulder: Westview Press. Simons, Ronald L., Rand D. Conger, and Leslie B. Whitbeck. 1988. "A Multi-Stage Social Learning Model of the Influences of Family and Peers Upon Adolescent Substance Abuse." The journal of Drug Issues 18:293-315. Vega, William A., RickS. Zimmerman, George J. Warheit, Eleni Apospori, and Andres S. Gil. 1993. "Risk Factors for Early Adolescent Drug Use in four Ethnic and Racial Groups." American journal of Public Health 83:185-189. 80 C. L. BANKSTON Ill Walter, Heather j., Robert D. Vaughn, and Alwyn T. Cahall. 1993. "Comparison of Three Theoretical Models of Substance Use among Urban Minority High School Students." journal of the American Academy of Child and Adolescent Psychiatry 32, 5:975-981. Windle, Michael. 1990. "A Longitudinal Study of Antisocial Behaviors as Predictors of Late Adolescent Substance Use: Gender and Ethnic Differences." journal of Abnormal Psychology 99:86-91.