.._ mfiwnm .fimmmw! {g} ........s .i yummvv. 4J9": ‘ H. S. . 3”!!!“ i .. _ . .1: . 1...}..minhi .... [:31 a: ‘. ”Uta. u.» .. 1 .ng 5.3.. .. 1 u,.....i.fi3$... , . :ezzhtrtr. -‘an . u L 3:1”. 2: .1! $- ‘ a ‘ 1).? )A [11... .1.1.\ a.‘ .‘l :.\,z;9| .k k I 5251.. Lust-:3!" v. x I. til I455. ‘0‘ 'n 1.: .Eiszwisxéi. -.ln..l’nuv|-}"”i, V .1. 11114... a $41.51.: itfl. E: :l. x... :5. .. It}; , ‘ o. .VI-O 1., .In‘ tr}ttt.l. '42.... .IUO. ‘L.il.llb.‘o. 3. tv. \ntl‘x..‘t‘ . 13!! .i ‘3. tAu lv-u al.3AIO.‘ A ‘ ¢| ol‘.‘|‘ ,r, .P ‘1‘ . ~llul .....-..t.l ‘11., P1!» Iii}, 1V. u .‘."..lo| t0.) ~(L.IILEY ‘13.... ....!U.v .. .11.... IREWI .‘cfi $1.3?qu -u. \‘ I'lli‘l‘uo‘l. r...» ll: UVMMFH aw’v ‘- 131‘ II v)~f.t‘». , I51» u» , . ,3? 1. \r -103 It ... $1». ‘10.: -.i (2... u... . t. . . . .. hi: 81h? .0 § .1 f .u,.._.A .f‘ 10.....- 1 LI luv. In. . viii! duvet .sbtfi I u u . .‘D‘ LU rrl u H5849 lllllllllllllllllllllllllllllllllllll ll! lllllllllllllll 3 1293 00885 9922 This is to certify that the thesis entitled Familial and Self-Concept Variables Related to Substance Abuse in a National Study of Disadvantaged Young Adults presented by Lisa C. Jordan has been accepted towards fulfillment of the requirements for M.A. Psychology/ Urban Studies degree in fl/fia ' \ / Major professor Jul 22 1993. Date )7 ’ 0-7639 I MS U is an Affirmative Action/Equal Opportunity Institution LIBRARY Mlchlgan Stale Unmmw PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. H DATE DUE DATE DUE DATE DUE MSU Is An Aflirmative Action/Equal Opportunity Institution cmmS-M .————--——'" ‘ ’ —'k"‘—'* ' ‘ "—"‘—'—_"“ — —“ FAMILIAL AND SELF-CONCEPT VARIABLES RELATED TO SUBSTANCE ABUSE IN A NATIONAL STUDY OF DISADVANTAGED YOUNG ADULTS BY Lisa C. Jordan A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Psychology 1993 John Hurley, Advisor ABSTRACT FAMILIAL AND SELF-CONCEPT VARIABLES RELATED TO SUBSTANCE ABUSE IN A NATIONAL STUDY OF DISADVANTAGED YOUNG ADULTS BY Lisa C. Jordan This study tested a model of self-concept as a "buffer" between exposure to substance abuse in the family system and outcomes of substance abuse in children of alcoholics. Descriptive analyses were conducted with data from the National Longitudinal Survey of Youth (NLSY). The analyses focused on a supplemental sample from the NLSY (g = 4,777) including African American, Hispanic, and low-income Caucasian youth. It was anticipated that high self-concept (as measured by indices of self-esteem, locus of control, and level of aspirations] expectations for academic success) would act as a "buffer,” decreasing rates of substance abuse in high risk youth. This study found minimal support for the "buffer" model. The only confirmatory evidence was obtained from analyses using the self-esteem measure with females aged 24- 27. ‘While the evidence for the other groups was less conclusive, some interesting gender and age effects were identified. The findings suggested that self-concept may exert variable effects on substance use depending on the age and gender of the individual. ACKNOWLEDGEMENTS I acknowledge with gratitude the assistance of my chairperson John Hurley for his assistance with this project, his attention to details, and his encouragement of my persistence and assertiveness in accomplishing the necessary steps to finish this Thesis. Special thanks to the other members of my committee, Robert Zucker, John Schweitzer, and Gloria Smith for their patience, support, and guidance. You have each contributed uniquely to my understanding of what it means to be a researcher and a psychologist. I give thanks to my family and friends for always believing in me, listening to me, and giving me the inspiration that I needed during the most difficult times. To my Aunt Tina who has firmly instilled in me a "never give up" attitude and continues to lovingly mold me into the person that I am. I also offer special thanks to Vernita Marsh who has been a mentor, a wonderful friend and part.of my "Michigan family." iii TABLE OF CONTENTS LISTOF TABLES00000 0000000000000000 00 ..... 00000000000 00000000 v LISTOF FIGURES00000000000000000000000000000000000000000000Vii INTRODUflIONO00O00OO00.00000000000000000.000000000000000000001 Prevalence of Substance Abuse Among Young Adults. . . . . . . .2 Theoretical Background .................................4 RWIWOFTHELITERATURE00000000000000000 00000000000000000 0011 Children of Alcoholics................ .............. ...12 Resilience as a Protective Mechanism. . ........ . ...... . . 16 STATEMENTOFTHEPROBLEM............ ............. . ........ ..19 HYMHESES0000000000000000000000000000000000000000000000000023 METHOD......................................................25 Sample.................................................25 Procedures.............................................26 Measures.............................. ...... ...........31 Family History..... ....... ............. ..... ...........36 Self-Concept.......................... ...... ...........36 Dependent‘Variables................ ............ ........37 RESULTS................................. ......... ...........40 Data Analysis Strategy.................................40 Demographic Characteristics............................42 Descriptive Statistics.................................42 Hypothesis I: Correlation of Self -Concept Measures. . . . . 47 Hypothesis II: Family Alcoholism and Respondent's Self-Concept......................................47 Hypothesis III: Moderator Effects of Self-Concept Measures..........................................49 Hypothesis IV: Age-Related Differences. . . . . . . . . . . . . . . . . 55 DISCUSSION0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 000000000 0 00000 0 0 0 0 0 0 0 0 0 0 0 0 0 57 Limitations of the Study. . . . . ..... . . . ........... . ...... 60 Future Directions................ ................... ...62 REFWCES 0 0 0 0 0 0 0 0 0 0 0 0000000000000000000000000000000000000 0 0 64 APPmDIx. 0 0 0 0 0 0 0 0 000000000 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 74 iv A "1' LIST OF TABLES 1. Selected NLSY Variables, Dates Collected, and Respondents 2. Sociodemographic Characteristics of the Sample by Family A1c°h0118m00000000000000000000000000000000000000000000043 3. Correlations Among Family Alcoholism and Self-Concept Variables 0000000 00000 00000 000000000000 000000 0000000000044 4 . Correlations Among Dependent Measures. . . . . . . ..... . . . . . . . . 44 5. Correlations Between Independent Variables, Moderator Variables, Alcohol-Related Problems, Alcohol Dependency Symptoms, and Drug Experimentation. . . . . . . ...... . . . . . . . . 45 6. Descriptive Statistics for Dependent Variables. . . . . . . . . . . 46 7. Summary Results of Multivariate Regression Analysis Predicting Substance Use Variables: The Main Effects of Family Alcoholism on the Outcome Measures ....... . . . . . . . 50 8. Hierarchical Regression Analyses Predicting Outcomes from Family Alcoholism, Self-Esteem, and their Interactions.51 9. Hierarchical Regression Analyses Predicting Outcomes from Family Alcoholism, Educational Aspiration, and their Interactions000000000000000000000000000000000000000000052 10. Hierarchical Regression Analyses Predicting Outcomes from Family .Alcoholism, Educational .Aspiration-Expectation Discrepancy, and their Interactions. . . . . . . . . . . . . . . . . . . . 53 11. Hierarchical Regression Analyses Predicting Outcomes from Family Alcoholism, Locus of Control, and their Interactions................. ....... ...................54 ~._._‘.. (~41.- ..n—efi‘l 1-5" .-1 -‘ 1A. 2A. 3A. 4A. 5A. 6A. 7A. 8A. 9A. Appendix Correlation Between Independent and Dependent Measures. . 74 Correlations Between Independent Variables, Moderator Variables, Ever Drank, Average Daily Quantity, Frequency of Heavy Drinking, Use of Illicit Drugs, and Marijuana 088000000000000000000000000000000000000000000000000000075 Summary of Hierarchical Regression Analyses Predicting Outcomes from Family Alcoholism, Self-Esteem, and their Interactions000000000000000000000000000000000000000000077 Summary of Hierarchical Regression Analyses Predicting Outcomes from Family Alcoholism, Educational Aspiration, and their Interactions.................................79 Summary of Hierarchical Regression Analyses Predicting Outcomes from Family Alcoholism, Educational Aspiration- Expectation Discrepancy, and their Interactions. . . . . . . . 81 Summary Table for the Hierarchical Regression Analyses Predicting the Outcome Variables from Family Alcoholism, Locus of Control, and their Interactions. . . . . . . . . . . . . . .83 Summary Table for the Hierarchical Regression Analyses Predicting the Outcome Variables from Family Alcoholism, Self-Esteem, and their Interactions for Total Sample. . .85 Summary Table for the Hierarchical Regression Analyses Predicting the Outcome Variables from Family Alcoholism, Educational Aspiration, and their Interactions for Total Sample................................. ..... ...........86 Summary Table for the Hierarchical Regression Analyses Predicting the Outcome Variables from Family Alcoholism, Educational Aspiration-Expectation Discrepancy, and their InteractionsforTotalSample..........................87 10A. Summary Table for the Hierarchical Regression Analyses Predicting the Outcome Variables from Family Alcoholism, Locus of Control, and their Interactions for Total sample00000000000000000000000000000000000000000000000088 vi LI ST OF FIGURES 10Thenauffer.'MOde10000000000000000000000 0000000000 000000021 vii V INTRODUCTION There has been considerable documentation and national attention to the relatively high rates of substance abuse among low-income and other disadvantaged populations. Lex's (1987) review of the literature on demographics and patterns of drug use among ethnic groups indicated that disadvantaged populations do not necessarily engage in.more substance abuse across the board. Moreover, ethnic groups may be quite heterogeneous in ‘their’ patterns of drug 'use and. abuse. Membership in a disadvantaged group does not appear to be a sufficient cause for substance abuse (Lex, 1987; Windle, 1990). Intervening variables may significantly contribute to the decision to abuse substances. These findings seem to indicate a need for broader analyses of the underlying phenomena in substance abuse. Studies that control for between-group differences and/or encompass within-group designs would be helpful in this regard. Because it provided a large subsample of low-income Caucasians, African Americans, and Hispanics, the National Longitudinal Survey of Youth (NLSY, conducted by the Center for Human Resource Research at Ohio State University) was of particular interest. 2 Prevalence of substance Abuse among Young Adults Data from national surveys on adolescent substance use indicate that, by the end of their senior year in high school, 93% of students have tried alcohol, 69% report having used alcohol in the past 30 days, and 37% drank heavily (five or more drinks at once) on at least one occasion during the past two weeks (Johnston, O’Malley, & Bachman, 1986). The same authors' Monitoring the Future Study found the following lifetime prevalence rates for drug use among 18-year-olds: 56% for marijuana, 23% for stimulants, 17% for cocaine, 11% for tranquilizers, 8% for opiates, and 8% for barbituates. Prevalence rates were substantially higher for a comparison group of 27— to 28-year-olds. The National Institute on Drug Abuse (NIDA, 1986) found that rates of alcohol use were highest among young adults aged 18-25, with 72% of those surveyed indicating that they drank alcohol during the past month. Other studies indicate that, while this may be the period associated with the heaviest rates of alcohol abuse among Caucasians, drinking patterns among other ethnic groups may vary (e.g., African Americans and Hispanics, Brown & Tooley, 1989; Office for Substance Abuse Prevention, 1990). Getting and Beauvais (1990) reviewed the data on alcohol and drug use from national epidemiological studies, including the 1986 National Senior Survey and the 1988 American Drug and Alcohol Survey. Their analysis revealed that while reported 3 drug use by Mexican American youth was similar to that of Caucasian American youth, reported rates of use were significantly lower for African Americans. In these studies, 93-94% of white youth, 84-91% of Mexican Americans and 83% of African American youth indicated that they had drank alcohol and gotten drunk at least once in their life. The National Longitudinal Survey of YOuth (NLSY) data show that 75.6% of youth aged 17-24 in a cross-sectional sample reported drinking alcohol during the past month, with the rate being higher for males than females. Approximately 48% were categorized as heavy drinkers (Grant, Harford, & Grigson, 1988). Other investigations revealed that the heaviest drinking was reported by males between the ages of 20-23; for females, the heaviest drinking group were 19-year- olds (Crowley, 1983; 1985a). Regarding alcohol-related problems, NIDA (1990) reported that the most common negative experiences from drinking were feeling aggressive (26%) and getting into heated arguments while drinking (19.8%) . Less commonly reported problems were: fear'of becoming'anlalcoholic (11%), difficulty stopping (9%), and nearly losing a job because of drinking (2%). Rates of problems experienced were significantly related to the number of the times respondents reported getting drunk over the past year. The prevalence rates for negative alcohol-related consequences in the NLSY sample are comparable to those reported by NIDA (see Crowley, 1985a). Negative consequences __ 3., l-‘ 4 were correlated with each other and also with level of alcohol consumption. Many constructs have been posited to be associated with the use of illicit substances during adolescence and young adulthood; these include both environmental and intrapsychic processes. Self-concept and exposure to norms of social deviance are prominent among the reasons posited. The present study attempts to test a longitudinal model that incorporates both environmental exposure (in the family system) and the intrapsychic construct of self-concept to predict young adults' drug usage. Theoretical Background The following section includes a development and rationale for the intrapsychic constructs of self-esteem, locus. of control, and. aspirations/expectations of future success in relation to substance abuse among children of alcohol-abusing parents. W Many theories on the etiology of substance abuse propose at least an indirect link between self-esteem level and the tendency to use mood-altering substances destructively. For example, Kaplan (National Institute on Drug Abuse [NIDA], 1980) hypothesized a central role of self-rejecting attitudes (resulting from previous negative experiences with important norm groups, such as same-age peers) in his Self-derogation theory of substance abuse. According to this theory, substance "”“l‘l 5 use is usually initiated in an attempt by the individual to reduce the experience of negative, self-rejecting attitudes. Steffenhagen (NIDA, 1980) further elaborated on this concept in his Self-esteem theory of substance abuse. Steffenhagen contended that while low self-esteem may serve as an impetus for drug initiation, especially for those seeking immediate gratification (or relief), low self-esteem.alone is insufficient to account for drug initiation. The individual's social milieu (e.g., peer group, family environment) provides the opportunities for, and approval of, drug initiation. In essence, low self—esteem may be a moderator variable that makes the individual more prone to social pressures to use drugs 9; increases the probability that drug use will be seen as a viable coping mechanism. This would explain the failure of many studies to findldirect and strong correlations between self-derogation and substance abuse. The reasons for initiation (intrapsychic and interpersonal) may make the person with low self-esteem more vulnerable to move quickly to drug abuse after initiation. Steffenhagen posited that drug use provided immediate gratification, a defense against personal insecurities and feelings of inferiority, and some degree of freedom from social responsibility (NIDA, 1980). In essence, this theory suggests that the abuse of substances is an attempt to deal with painful emotions or conflicts by using the mode of adaption most readily available in the person's ego 6 functioning at that moment (Krystal & Raskin, 1970). Due to the dysfunctional relationship patterns in families of substance abusers, and the modelling of maladaptive coping mechanisms, children of these families may develop defensive structures that make them more vulnerable to self-devaluing experiences. Their family dysfunction may also leave them devoid of means of dealing with negative experiences in ways other than the abuse of substances. Ultimately, children of substance abusers may be expected to have poorer self-concepts. There is a body of literature, including empirical studies, that supports the self-esteem or self-derogation model of substance abuse. Several studies have found that substance-abusing youth score lower on measures of self- concept and self-esteem than their peers (Pandina & Schuele, 1983; Parish & Parish, 1991; Yanish & Battle, 1985). In a study comparing adolescents in treatment for substance abuse with non-treatment controls, treatment group members reported that low self-concept and inability to cope were factors that precipitated their drug use (Svobodny, 1982). Similarly, Newcomb, Bentler’ and Collins' (1986) longitudinal study of 640 adolescents found a correlation between self-derogation and later alcohol use. This study corroborated Kaplan's theory by showing that use of alcohol during adolescence was significantly related to decreases in reported self-derogation during young adulthood. W Ego/Self theory (Khantzian, cited in NIDA Research Monograph 30; 1980) also highlighted the role of disturbances in self-concept as related to substance abuse and other lapses in self-care. According to Khantzian, drug use is one way in which the ”addiction-prone" person may use the external world to fulfill the need for a sense of well-being, security, and pleasure. Poor defenses and low self-esteem were considered as underlying factors in the "addiction-prone" individual's heavy reliance on the external environment for satisfaction of intrapsychic needs and desires. Substance abuse also was posited as a general failure in self-protective functions that deterred persons from dangerous and/or self-destructive behaviors. The self-protective mechanisms were considered to be related to self-esteem in that a person must have "sufficient self-esteem to feel oneself to be worth protecting” (Khantzian & Mack, 1983, p. 210). If this theory is correct, there is reason to believe that drug initiation and progression to abusive usage is affected by predisposing intrapsychic factors, such as low self-esteem and high externality. Given such predisposing factors, one would expect that persons who abused drugs would score lower on measures of self-esteem and be more externalizing than nonabusers or nonusers. This hypothesis has been corroborated by the results of several empirical studies. Jurich and Polson (1984) surveyed a paired sample of ._..—__-. ”—— ”fl 1 O .. Q I ‘. 8 48 drug-using and drug-abusing high school students and found that the drug users were more likely to report recreational reasons for use. Drug abusers were more likely to report using drugs to ameliorate feelings of distress and disillusion, and to cope with low self-concept or external locus of control. Martin and Pritchard (1991) reported that, among 8,661 college students aged 20-21, White males from higher socioeconomic classes, with weak family orientations and external locus of control, tended to drink more frequently and consume more alcohol per drinking episode than other students. WM Closely related to self-esteem are the individual’s hopes and expectations for the future. Much research documents the negative correlation between academic involvement, high aspirations, or hopefulness regarding future success and the tendency to engage in deviant behavior, like substance abuse (Bechtel & Swisher, 1992; Jessor & Jessor, 1977; Labouvie & MCGee, 1986; Newcomb, Bentler, & Collins, 1986; Newcomb 8 Bentler, 1988a; Newcomb, Fahy, & Skager, 1988). Jessor and Jessor (1977) posited this relationship as a.social control factor, and subsequent research has supported this hypothesis. It has been found that attachment to the family and school, and a commitment to traditional values decreases the likelihood of engaging in deviant behavior (Kaplan, Martin, & Robbins, 1984). 9 Regarding educational aspirations and substance abuse, Jessor and Jessor (1977) found that students who had strong attachments to school, valued academic achievement, and had high academic aspirations and expectations of success were less likely to abuse substances. The academic involvement measures were significantly correlated with rates of alcohol consumption over the past year and marijuana use. On repeated measures of substance abuse, those students who shifted from nonuse to use during a one year period.had.previously assigned less value to academic achievement than those who did not become substance users. Similarly, Newcomb and Bentler's (1986) study of 479 high school students revealed that educational plans (i.e., how much school do you expect to complete?) and a measure of Academic Potential (including Grade Point Average) were negatively correlated with self-reported used of drugs over the past six months. The authors concluded that "a negative perception of one’s future opportunities is significantly predictive of increased alcohol use from adolescence to young adulthood” (p.492). Donovan, Jessor and Jessor (1983) found that adolescent problem drinkers had lower expectations of academic recognition and placed less value on academic achievement prior to excessive drinking. In a study linking self- derogation, general deviance and academic involvement, Kaplan, Martin, and. Robbins (1984) found. a direct path. between 10 perceived rejection by people at school, the perceived self- enhancing potential of deviant behavior, and subsequent alcohol or marijuana use. RBVIBI 01' THE LITERATURE Although several researchers have explored the relationship between self-esteem/self-concept and substance abuse, the findings have been mixed. The mechanisms by which perceptions of self relate to deviant responses, such as drug abuse, remain unspecified. Some researchers have found negative correlations between alcohol consumption patterns and self-esteem or feelings of self-worth (Yanish & Battle, 1985) . These studies have typically compared alcoholics with nonalcoholic drinkers and consistently shown that the latter scored higher on 'measures of self-esteem. other investigations, however, have failed to yield significant correlations between self-esteem or self-concept and measures of substance use (Jessor & Jessor, 1977; Labouvie & McGee, 1986; Martin & Pritchard, 1991; Wells & Rankin, 1983). A noted concern with this literature is the fact that many studies have been cross-sectional, precluding determination of the direction of causality between self- derogating experiences and substance abuse (Newcomb, Bentler, & Collins; 1986). Additionally, most studies have examined the direct link between self-concept and substance use without considering other risk factors, such as the family history of 11 12 substance abuse, which also contribute to the respondents' use of substances. Due to the transitional nature of the period between adolescence and young adulthood, the likelihood of naturally occurring changes in self-esteem and feelings of self- derogation must also be considered when using constructs related to self-concept. Kaplan, Robbins, and Martin (1984) cited research which suggested that between the ages of 13 and 14 marking the transition from junior high school to high school, there is a slight decrease in ratings of global self- esteem. Self-ratings were found to increase between the ages of 14 and 15, and stabilize by age 18 (Kaplan, Robbins, 8 Martin, 1984). These factors will be taken into consideration by this study as the relationship between family alcohol abuse and self-esteem of the participants is examined. W The literature on the vulnerability of children of alcoholics is equivocal, much like that on self-concept and substance abuse. In a well-cited metanalytic review of the literature on family incidence of alcoholism, Cotton (1979) concluded that compared to relatives of non-alcoholics, the rates of alcoholism were substantially higher among those with alcoholic relatives. Similarly, in regards to adolescent substance use, Kandel and Jessor and Jessor suggested that "prior association with users of a particular drug is the strongest predictor of an individual's use of that drug" 13 (Kaplan, Martin, 8 Robbins, 1984, p.279). Research on children of substance abusers indicates that use of illicit drugs among adolescents is correlated with parental use of the same substances and attitudes towards drug use in general (Fawzy, Coombs, 8 Gerber, 1983; Johnson, Shontz, 8 Locke, 1984; Prendergast, 1989). Kandel (1978) reported that 82% of drinking families had youth who drank also, while 72% of abstaining families produced abstainers. Of the families with parents who drank hard liquor frequently, 76% had substance using children. Barnes, Farrell, 8 Cairns (1986) reported similar results although lower percentages with a sample of 124 families. Other studies on children of alcoholics 'have corroborated. the hypothesis of a strong relationship between having alcoholic parents and abusing substances later in life (Coombs 8 Paulson, 1988; Drake 8 Vaillant, 1988; Gfoerer, 1987; Gross 8 McCaul, 1991; Hyphantis, Koutras, Liakos, 8 Marselos, 1991 ; Kandel, Kessler, 8 Margulies, 1978; McDermott, 1984; Newcomb 8 Bentler, 1988b; Sher, Walitzer, Wood, 8 Brent, 1991). Contrarily, Alterman, Searles, and Hall (1989) failed to find significant differences in alcohol involvement between male college students in a sample of 83 respondents with and without alcoholic parents or second degree relatives. These authors cited previous studies (e.g., Knop, Teasdale, Schulsinger, 8 Goodwin, 1985; Schuckit 8 Sweeney, 1987) that also failed to find the hypothesized differences in problem 14 substance use for children of alcoholics. In a study of 1,308 youth, Pandina and Johnson (1989) found that some indicators of problem alcohol use were more prevalent among children of alcoholics (e.g., experiencing negative consequences of drinking), while others were not (e.g., early onset of drinking, frequent intoxication, and drinking for escape reasons). Respondents from alcohol abusing families were slightly more likely to use marijuana heavily, but this effect only held for the oldest age group in the sample. These authors suggested that the level of problem use among youth from substance abusing families may be partly determined by the time frame during which substance use is assessed. The most reliable window was the period between late adolescence and early adulthood (e.g., 18 to 21 years of age). In a re-evaluation of this study, Pandina 8 Johnson (1990) concluded that the vast majority (88%) of the respondents with a family'history of substance abuse could.not be classified as having alcohol or drug-related problems. Additionally, they noted that a small percentage (6%) of respondents with no reported family history of substance abuse were found to have problems as severe as those with a family history; This was interpreted as evidence that. having alcoholic parents was not a sufficient predictor of adult vulnerability to substance misuse. Conversely, not having a family" history of substance abuse ‘was not a sufficient protector against later substance abuse. 15 Some researchers have reported gender differences in substance use by children of alcoholics. For instance, Sher, Walitzer, Wood, and Brent (1991) found that in a group of 253 children of alcoholics, male children of alcoholics reported heavier alcohol consumption, more negative consequences of drinking, and more dependency symptoms than females. In terms of psychosocial characteristics it has been postulated that children of alcoholics may have distinctive personality deficits, including lower self-esteem and more external locus of control (Barnard 8 Spoentgen, 1986; Berkowitz 8 Perkins, 1988; Cermak 8 Rosenfeld, 1987; Kern, et al., 1981; McNeill 8 Gilbert, 1991; Rearden 8 Markwell, 1989; Wallace, 1987) . Some researchers have found evidence of specific adjustment problems in children of substance abusers. For instance, Berkowitz and Perkins' (1988) study of a group of 860 late-adolescent and young adult children of alcoholics found that these children, although similar to their peers by most personality measures, were more likely to report feelings of self-depreciation. McNeill and Gilbert (1991) found that being external orientated was significantly related to having a parent who drank heavily. Regarding academic involvement, Hyphantis, Kourtras, Liakos, and Marselos' (1991) investigation of 8,000 high school students found that children of alcoholics had considerably poorer school performance than did their same- aged peers. This finding has been corroborated by other 16 studies (e.g., Sher, Walitzer, Wood, 8 Brent, 1991). As discussed above, the relationship between academic involvement, social deviance, and substance use has been explicated by other researchers. E iii E ! !'v H l . These studies indicate the importance of modelling of deviant behaviors within the family system, as well as the role of family socialization in determining an individual's self-concept and coping abilities. However, it has been noted that while children of substance abusers may be at a greater risk for negative outcomes including poor self-concept and substance abuse, not all children of alcoholics become substance abusers. Certain protective mechanisms seem to insulate or "buffer" some individuals from even the most dysfunctional families. Several theoretical approaches have been developed to address this issue of resilience. Rutter (1985) noted that.it is unusual for more than half of children of dysfunctional families to succumb and repeat the dysfunction. He hypothesized that protective mechanisms could be found in ‘the environment, in the individual's constitution, or some combination of both. Rutter defined protective mechanisms as "influences that ameliorate, or alter a person's response to some environmental hazard that predisposes to a maladaptive outcome" (1985, p. 600). Thus, these protectors may not make the person healthy, or invulnerable, but may instead mediate responses to situational factors 17 adversity. Protection is posited to develop as a result of experiencing appropriate social controls, positive role models, or at least one good interpersonal relationship. The work of Kaplan and Jessor and Jessor highlighted the importance of social control, school involvement, and attachment to traditional values as protective factors. Marston et a1. (1988) also found that resilient children of alcoholics reported higher academic achievement. Perhaps a connection with positive role models outside the family underlies the protective nature of attachment to conventional values and aides the person in developing a positive self- concept, despite family dysfunction. Positive self-concept has been noted by several authors as a protective mechanism against negative outcomes. Advocates of containment theory suggest that a positive self- concept may serve as an "insulator against deviance” (Reckless, et al., 1956; Reckless, 1967; Schwartz 8 Tangri, 1965; Voss, 1969; cited in Kaplan, Robbins, 8 Martin, 1984, p. 78). This mechanism may be particularly effective when the necessary social controls and socialization are inhibited, as with children of substance abusers. Garmezy (1985, cited in Rutter, 1987) reviewed the literature on stress-resilient children and concluded that protection was due to three broad factors: 1) personality features, such as self-esteem, 2) family cohesion and lack of discord, and 3) the availability of external support systems 18 that encouraged or reinforced the child's coping efforts (p. 316). similarly, Werner and Smith (1982) found that in a longitudinal study of Kauai children, protection from family dysfunction was associated with having a supportive kin relationship, being dispositionally good-natured (as assessed during infancy), and having a positive self-concept among other factors. Werner (1986) suggested that self-esteem moderated for the effects of family alcoholism. The resilient children of alcoholics tended to do well in school, had realistic expectations for the future, had internal loci of control and higher self-esteem. They were also predominantly (72%) female. STATEMENT OF THE PROBE.“ Although it has been documented that children of alcoholics are at risk for a number of psychological problems, including poor self-concept and a propensity towards substance abuse, it is also acknowledged that many children of alcoholic parents escape these negative outcomes. Studies on resilient children have been limited and the nature of protective or moderator factors has not been explored sufficiently to understand their operation. The available literature has tended to focus on the effects of either self-concept 9;: family history of substance abuse in isolation as opposed to examining their interactive nature. As discussed above, one hypothesis is that children of dysfunctional families who have the benefit of self-enhancing experiences outside the home, are able to develop positive self-concepts and are thus protected from negative outcomes. There is reason to believe that social class may be confounded with availability of self-enhancing experiences. For the purposes of this study, parental educational level will be used as a controlling factor in analyses comparing children of alcoholics and children of non-alcoholics. Previous literature has established this measure as a marker of social 19 20 class (e.g., Richman, Clark, 8 Brown, 1985). Additionally, parental educational level has been found to be correlated with self-esteem.(Richman, Clarkq 8 Brown, 1985) and substance use (Fawzy, et al., 1987; Schinke, et al., 1992; Zucker 8 Harford, 1983). The literature on substance abuse among youth shows fairly consistent gender differences in both patterns of use and experience of negative consequences of use (e.g. , Bachman, Johnston, 8 O'Malley, 1981; Cervantes, Gilbert, de Snyder, 8 Padilla, 1991; Donovan, Jessor, 8 Jessor, 1983; Humphrey, Stephens, 8 Allen, 1983; Lex, 1987). These studies indicate that males generally engage in heavier substance use and experience more problems as a result of drug and alcohol use. Previous studies with the NLSY data have corroborated these gender differences within this sample (Crowley, 1983, 1985a, 1985b; Windle, 1990a). Gender effects will be addressed in this study by separate analyses of drug and alcohol use measures by age and by gender. The present study will attempt to draw connections between the literature on self-concept, family history, and outcomes of substance abuse. Self-concept will be tested as a ”buffer" or moderator of the relationship between exposure to substance abuse in the family system and outcomes of substance abuse in the children of these families (see Figure 1). 21 Figure 1. The ”Buffer” Model Self-Concept _ s ‘ ~ Parental Substance Abuse >Substance Abuse *Note dotted lines indicate negative correlations Since some of the literature reviewed suggested that parental substance abuse leads to deficits in personal adjustment, the issue of whether there is a significant relationship between parental substance abuse and respondents' self-concept will be addressed first. Secondly, the extent to which self-concept moderates the relationship between family history and respondents' substance use will be tested. The relationship between parental substance use and substance use by their offspring' has been fairly' well established. by previous literature and will not be the major focus this study. The focus will be on exploring the moderating effects of self-concept. If children of alcoholics are indeed found to have poorer self-concepts and self-concept is found to moderate the relationship between family substance abuse and respondents' substance use, this will provide evidence for the 22 "buffer” model and add.to the understanding of factors related to risk and resilience among children of alcoholics. Findings from ‘this study 'will be. generalizable to U. S. ethnic minorities (particularly African and Hispanic Americans) and low-income populations, due to the nature of the NLSY sample. It also will not be biased by school attendance, as the data were collected from a stratified national sample. HYPOTHBBBS Previous findings have suggested that different.measures of self-concept (particularly self-esteem and locus of control) are likely to Ibe correlated although they' may represent conceptually distinct realms of personality (Churchill, Broida, 8 Nicholson, 1990; McNeill 8 Gilbert, 1991; Werner 8 Broida, 1991). In light of this, I propose to test the following hypotheses related to self-esteem/self-concept and substance abuse: (1) measures of self-esteem, aspirations of future academic success, and locus of control will be correlated positively; that is, respondents with high scores on the self-esteem scale will be more likely to report higher aspirations of future success and endorse more items reflecting internalization on the locus of control measure. The opposite will be expected for respondents who score low on the self-esteem scale. Conversely, each of these three facets of the self-concept is predicted to be negatively correlated with the reported discrepancy between aspirations and expectations of academic success. (2) children of substance abusers will score lower on measures 23 24 of self-esteem, aspirations for academic success, and have higher externality scores and larger aspiration-expectation discrepancies than children of nonabusers, (3) the ”buffer" hypothesis will be supported, as evidenced.by significant interaction effects of the self-concept variables with family history of alcohol abuse in the prediction of substance use by the respondents, and (4) the relationship between family history and self-concept will be stronger in older adolescents and young adults than in younger adolescents due to the stabilization of the self- concept measures in late adolescence. ._._.... x».A_‘—~.-g—mg..h u . , - METHOD This study was based on data available from the National Longitudinal Surveys of Labor Market Experience (Center for Human Resource Research, 1990) which collected information on substance abuse and self-concept as it:related.to labor market participation in youth. The survey involved three different age cohort groups (3 = 12,686). This study focused on the youth sample (National Longitudinal Survey of Youth, NLSY). Sample The analyses for this study utilized data from a supplemental sample of the NLSY youth cohort (original n = 5,295; number interviewed at the last survey year Q = 4,777), which consisted of Hispanic (3 = 1,480), African American (9 = 2,172), and economically disadvantaged Caucasian youth (n_= 1,643). The NLSY sampling design was based on a stratified national probability sample of people born between the years of 1957 and 1964, with moderate oversampling for African American, Hispanic and low-income Caucasian youth. The selected youths aged 14-22 were interviewed through personal household contacts or telephone interviews beginning in 1979 and surveyed annually thereafter with a 90% retention rate. From this sample, all respondents who reported having 25 ---10—u... u.“ .- L I 26 one or more problem-drinking or alcoholic parents were selected as a high risk group. In the analyses, this sample of youth with one or two alcohol-abusing parents was compared on the self-concept and substance use measures to the youth who did not report having alcohol-abusing parents. Procedures Table 1 lists the collection. dates of the INLSY variables used in this study and ages of respondents at the time of collection. The files that provided these selected data were: (1) the Alcohol and Substance Use file in which data was collected during 1982-5 and 1988 and focused on measures of alcohol use (last 7 days and last 30 days), developmental drinking patterns, and impact of alcohol use on school and/or job behavior for the period of 1982-5. The 1988 survey also collected information regarding relatives of respondents who were alcoholic or problem drinkers, including the relationship of the respondent to the relative and the amount of time the respondent resided with these relatives. Many of the alcohol questions were adapted from the National Health Interview Surveys conducted by the U.S. Bureau of Census. Grant, Harford, and Grigson (1988) conducted a reliability study using the NLSY alcohol measures. Although no reliability data existed previously for the self-report alcohol measures, these authors found that answers to lifetime prevalence questions (collected in 1982 and 1983) Table 1. Years Collected 1979 1980 1981 27 1982 1983 1984 1985 1986 1987 Selected NLSY Variables, Dates Collected, and Respondents Age 1988 Respondents' Age m Common Demographic Information Ethnic self Identification Highest grades completed by parents 14-22 15-23 16-24 17-25 18-26 19-27 20-28 21-29 22-30 23-31 W Educational Aspirations and Expectations Rotter I-B scale Rosenberg B-E scale Alcoholgfise Ever had a drink Q/F" - last month A/D+ symptoms Family history of Alcohol abuse ><><>< Drug Use Lifetime use of marijuana Lifetime use of other drugs Age of first use @6969 ®®® Note: 'Q/F = Quantity/Frequency *A/D = Abuse/Dependency circled values indicate the data used for this study. 28 were fairly'consistentm Only 2% of the youth.who had reported having ever drank in 1982 denied this in 1983. This mild degree of error could be attributed to normal memory processes and suggested little intentional bias in. the reporting. Respondents’ reports of having "ever used" alcohol showed acceptable consistency across years. Responses to problem drinking items were less stable over time. Mensch and Kandel’s (1988) review of the NLSY alcohol and drug use data found that of respondents who reported having drinking-related problems in 1982, only half reported having such problems during their lifetime in 1983. (2) The Drug use file included data from the 1984 and 1988 surveys and provided information on age at first use and extent of use of cigarettes, marijuana, and all other illicit and non-prescription drugs used during those years, including a retrospective account of respondents' use of marijuana] hashish. during 1979-1984. Frequencies of lifetime ‘use, recency of use over the last year, and frequency of use in the last month were asked for each drug classification. For the purposes of this study, lifetime frequencies were used. The most comprehensive drug data were collected during the 1984 survey year, so all analyses were based on those data. For consistency, the alcohol use data from 1984 were also used. It should be noted that respondents' confidentiality was maintained by coding all responses from the alcohol and drug files with identification numbers only. The refusal rates to _. 04—1.;4-nuo. r -_ -.a. ‘ . 29 these questions was less than 1%. This did not insure that respondents were completely comfortable answering the questions or that they were always honest. A comparison of responses by NLSY participants with those of other national drug surveys, such as the Monitoring the Future study and the General Household Survey, revealed evidence of underreporting in the NLSY sample (Mensch 8 Handel, 1988). For example, the reported frequencies of lifetime and current use of marijuana were lower than expected based on. population norms and developmental progression of drug use. NLSY respondents were generally less likely to admit to heavy use of marijuana. Lifetime prevalence for illicit drug use was also lower than expected, with underreporting of inhalants being the most substantial. However, reported lifetime frequencies of illicit drug use were higher in the NLSY sample. Mensch and Kandel indicated that nonreporting of drug use was twice as likely for African Americans and Hispanics as for Caucasians. While the issue of underreporting of drug use among minorities has been identified by previous epidemiological studies, few researchers have dealt with the matter directly. The reasons for nonreporting by minorities remains somewhat of a mystery; however, some authors (e.g., Johnson, Bachman, 8 O'Malley, cited in Mensch 8 Kandel, 1988) have suggested that minority group members are less likely to trust the research process. Feeling threatened by sensitive survey items, minorities may have a tendency to underreport socially 30 undesirable behaviors. There is also considerable debate in the psychological literature over the validity of self-report and questionnaire data for sensitive topics, such as substance abuse. However, self-report and questionnaire data are commonly the most readily available sources of data. The NLSY study relied primarily upon participant's verbal responses to items regarding their own and their relatives' substance use. While the NLSY data revealed patterns of underreporting and overreporting of certain drug categories, overall it was found that respondents reported their substance use with an acceptable level of reliability (Crowley, 1985b; Grant, Harford, 8 Grigson, 1988). Previous researchers (e.g., Gfoerer, 1985; Mensch 8 Kandel, 1988; Polich, 1982) have also concluded that self-report information regarding respondents' own substance use is generally valid and reliable. In a review of studies using self-report measures of substance use, Gfoerer (1985) found that most studies of young adults concluded that reliable and valid self-report measure could be obtained, given the appropriate conditions (e.g. , maintenance of privacy). The substance abuse literature has identified certain biases associated with use of participants’ ratings of parental drinking patterns. This method usually does not address problem severity or allow the researcher to differentiate among different subtypes of alcoholism. 31 Additionally, ratings of parental alcoholism by their offspring tend to result in more false negatives than false positives (Sher, Walitzer, Wood, 8 Brent, 1991). MOst research indicates that relatives are much better at determining who was 39; an alcoholic than who was. Thus, reports of family alcoholism are considered to be fairly reliable (Rogosch, Chassin, 8 Sher, 1990; Russell, 1990). 3) The Attitude file provided information on locus of control, self-esteem, and aspirations/expectations (see discussion below). These were collected in the 1979 and 1980 surveys. (4) The Common Demographic Information file provided information on the respondents’ race, gender, and family background which were primarily collected during the initial survey’ year, except for the respondents’ annual income, marital status, and whether they had completed high school or received a GED. The latter information was collected during the 1984 survey year. Measures u Cont Sc Locus of control was assessed using four selected items from the Rotter Internal-External Locus of Control scale (LOC; Rotter, 1966) administered in 1979. Items on the Locus of Control scale were forced choices between two statements reflecting either an internally or externally oriented personality style. Respondents are asked to choose the 32 statement that best applied to them, then to indicate whether the chosen statement corresponded closely or slightly to their true opinion. At this time, no information is available on the reliability or validity of the abbreviated Rotter LOC scale used in this survey. Studies using the full 23—item scale indicate 'that. the IRotter scale Ihas adequate ‘test-retest stability, however; the validity of the unidimensional construct of locus of control that this scale attempts to measure is questionable. Several researchers have noted problems interpreting the internal-external factor based on item endorsements, due to the fact that item alternatives are not symmetrical and have not been proven to measure opposite constructs (e.g., Little, 1977; Roberts 8 Reid, 1978; Tyler, Gatz, 8 Keenan, 1979). Rotter (1966) reported test-retest stability coefficients ranging from .49 to .83 for the 23-item Rotter LOC scale with various samples and with intervals of one week to two months. Internal consistency coefficients ranged from .65 to .79. Subsequent studies have generally supported the full scale's reliability. In a study of 247 recent graduates of a liberal arts college, Little (1979) reported stability coefficients of .64 for the Rotter LOC scale over a thirty-month interval. Factor analytic studies using the Rotter LOC scale have established its multi-dimensionality, with different studies identifying between two and five independent factors on the 33 measure. Tyler, Gatz, and Keenan's (1979) item analysis revealed a structural imbalance of the test in that certain domains were more highly represented than others. Marsh and Richards (1986) tested the scale on 71 participants (mean age = 22) in.a 26-day Outward Bound program designed to alter Locus of Control, among other personality factors. Criterion validity' of Rotter’s LOC scale ‘was modestly supported with a correlation of .34 between self- and observer-responses. Construct validity was supported somewhat in that participants scored significantly higher on the Rotter LOC scale after completion of the program. A study comparing the Rotter LOC scores of 541 high school students with their scores on the MacDonald-Tseng test (which was based on a factor analysis of the ‘Rotter LOC items) supported its concurrent validity by a .42 intermeasure correlation. A discussion by Omizo, Omizo, and Michael (1987), noted that the Rotter LOC scale may not be appropriate for younger individuals, because they are subject to the controlling influence of various institutions of authority (e.g., parental, educational, legal). Thus, scores on the‘Rotter LOC scale may have different connotations for adolescents than for adults. However, these authors also noted that, despite the problems with the Rotter scale, no similar questionnaires have been proven valid for adolescents. Some researchers have found different factor loadings of the Rotter LOC scale between ethnic groups and suggest that cultural values and e-‘u'fl J) C'- 34 experiences may significantly influence responses (Roberts 8 Reid, 1978; Garza 8 Widlak, 1977). These issues are particularly relevant to the present study. W Self-esteem was measured using the Rosenberg Self-Esteem (S-E; Rosenberg, 1965) scale administered in 1980. The measure consisted of ten items rated on a four-point Likert scale of strongly agree to strongly disagree. Seven of the items are worded in a positive direction, indicating high self-esteem; three of the items are worded in a negative direction. The validity and reliability of the Rosenberg S-E scale are consistently supported in the literature. Rosenberg (1965) originally tested the measure on a sample of 5,024 randomly sampled high school students in the U.S. and reported a scale reproducibility coefficient of .92. Subsequent studies have examined the internal consistency and test-retest reliability of the measure and found it to be reasonably stable. Goldsmith (1986) found internal consistency values of .96 with a sample of 97 college students and .90 with a sample of 87 adults from the general population. Byrne (1983) administered the Rosenberg S-E scale to 929 high school students and found a stability coefficient of .62 over seven months. Convergent validity was supported by a .58 correlation between the Rosenberg S-E scale and the General .14e-n'-~'m.zll u,.:' 35 Self subscale of the Self-Esteem Inventory (Eagly, 1967; cited in O'Brien, 1985). O’Brien found support for the unidimensionality of the scale in a sample of 206 female college students. However, Goldsmith (1986) cited studies that extracted two to three independent factors from the Rosenberg S-E items. I i !' 1 E ! !' Expectations of future educational and occupational attainment were measured in 1979 and 1982. The 1979 data were used in this study. For educational obtainment, respondents were asked "what is the highest grade or year of regular school, ..., college, or graduate school that you would like to complete?” and "what is the highest grade or year you think you will actually complete?" (rated on a scale of 1-13, 1 = first grade; 13 = more than five years of college). The difference between these two measures provided an index of contrast between aspirations and expectations or a sense of hope for one’s future. For the purposes of this study, the independent variables fell into categories related to family history of problem drinking and the proposed moderating measures of self-concept. The specific measures used were as follows: I. Family History AL_E§mil¥_Al§Qthifim The extent of exposure to familial substance abuse was measured by selecting respondents who indicated in the 1988 36 survey that they had a biological parent who was a problem drinker. Analyses were conducted based on three groups: those who denied having a parent with a drinking problem, those who reported having one parent with a drinking problem, and those who reported having two parents with drinking problems. LEW The level of exposure to an alcoholic parent was measured by the amount of time the respondent reported living with a problem-drinking parent. Responses to the 1988 survey item "For how many years did you live with your (relative) while (he/she) was an alcoholic or problem drinker?" were selected for respondents with problem drinking parents. A mean number of years was computed for those who reported having lived.with two alcoholic parents. II. Self-Concept AL-BQ£;§I_LQ§_§QQl§ Coded.responses to the four items of the‘Rotter LOC scale from the 1980 survey were summed to form a total score for LOC of each respondent, with high scores indicating high internality. EL_BQ§§nh§IQ_§:E_§QQl§ Endorsements on the ten items from the Rosenberg S-E scale administered in 1980 were summed to form a total Self- Esteem score for each respondent, with high scores indicating high self-esteem. 37 Educational aspirations were assessed using the response to one questionnaire item from the 1979 survey, as discussed above. Q, Aspiration-Expectation Discrepancy The difference between the respondents educational aspirations and expectations was computed by subtracting the reported grade the respondent expected to complete from what they wanted to complete. These measures were both collected in 1979. III. Dependent Variables The outcome measures consisted of eight variables derived from the 1984 survey data. These variables measured the level of substance use and indices of negative consequences of drug involvement. They were: AL_E¥§I_H§Q_Q_D£iDK This measure consisted of responses to one item which was "Have you ever had drink of an alcoholic beverage?" This was a dichotomous measure. Fbr respondents who indicated that they had not had never drank, the other alcohol use items were not administered. B, Aggragg Daily Quantity (ADO) The ADQ of alcohol consumption during the last month was computed by summing across six alcohol frequency items (e.g., number of days had one drink,... number of days had six or more drinks) and dividing by 30 days. This measure was used 38 previously by Windle et. a1 (1989, 1990a, 1991) and Miller- Tutzauer, Leonard, and Windle (1991) for the NLSY data set. Respondents who denied having drank in the last month were not asked the following questions about frequency of heavy drinking, alcohol-related problems, or dependency symptoms. 0 E E H 0 . 1. [EH01 The FHD during the last month was computed by summing across two items assessing the number of times the respondent had five, six, or more drinks per day during the last month. This measure was used by Miller-Tutzauer, Leonard, and Windle (1991). — e ed 0 A measure of alcohol-related problems was derived by summing across ten dichotomous items assessing alcohol-related aggression or interference with work and/or school (e.g., "During the past year, have you gotten into a fight while drinking?"). This scale was constructed by Miller-Tutzauer, Leonard, and Windle (1991). These authors reported internal consistency (Cronbach's alpha) of .65 for the scale using the 1984 data of 10,594 NLSY respondents. MW A measure of self-reported alcohol dependency symptoms was derived by summing across eight dichotomous items assessing the extent of dependence on alcohol (e.g., "During the past year, have you sometimes kept on drinking after promising yourself not to?"). This scale was used previously 39 by Windle et. al (1989, 1990a, 8 1991) and by Miller-Tutzauer, Leonard, and Windle (1991). Windle (1989) reported internal reliability (Kuder-Richardson) of .68 for the 1985 NLSY data. Miller-Tutzauer, Leonard, and Windle reported a Cronbach alpha coefficient of .66 using the 1984 data for 10,594 NLSY respondents. This measure was assessed by the following item: "In your lifetime, on how many occasions have you ever used marijuana or hashish?” 9W Lifetime use of other drugs was computed by summing across nine variables assessing the number of times the respondent had ever used the specified drugs. This summation of lifetime frequencies was divided by the total number of years the respondent had been using drugs. The resulting index was a measure of the average number of times the respondent used illicit drugs during each year since the earliest reported use. f u e ' e t t'o This index was constructed by summing across the dichotomous "ever used" items for each of the ten specified drug categories. The resulting index was a measure of the number of different drugs a respondent had used during his or her life. RESULTS mm In testing the relationship between parental alcohol abuse, self-concept, and substance use by study participants, age and gender effects were controlled for by analyzing the data separately for four different groups: males aged 19-23, females aged 19-23, males aged 24-27, and females aged 24-27 in 1984. Although identical analyses were conducted for the full sample (see Appendix, Tables 7A-10A), this section will focus on the results for the four groups. Hypotheses I and II were tested with Pearson product- moment and partial correlation coefficients. Hypothesis III, the test of the "buffer" model, was tested by two methods of analysis. The first consisted of a Multivariate Multiple Regression test for the main effects of Family Alcoholism on the eight dependent variables. The second step involved a series of hierarchical regression analyses testing the main effects of Family Alcoholism, Family Alcoholism with each moderator variable, and the interactions between them. The hierarchical regression analyses were run for each possible combination of Family Alcoholism, moderator variable, and dependent variable. For each test, the main effects of Family 40 41 Alcoholism were entered on step 1; the main effects for the moderator variable were entered on step 2; and the interaction (the product of the independent variable and the moderator variable) was entered on step 3. This method of testing for moderator effects was based on the work of previous authors (e.g. , Barron 8 Kenny, 1986; Rogosch, Chassin, and Sher, 1990; Jaccard, Turrisi, 8 Wan, 1990). According to Barron and Kenny (1986), a moderator is a "qualitative or quantitative variable that affects the direction and/or strength of the relationship between an independent or predictor variable and a dependent or criterion variable" (p. 1174). Conceptuallyy moderators serve to buffer the impact of a known vulnerability (in this case, family history of alcohol abuse). Statistically, this relationship can be represented as an interaction between the independent variable and the proposed moderator variable. The moderator hypothesis would be supported if the independent and moderator variables each showed significant main effects and the interaction term, or the product of the two variables, was significant. Moderator variables must account for an M percent of the variance which is statistically significant. Jaccard, Turrisi, and Wan (1990) suggested that hierarchical tests are most appropriate for partialling out the additional effects of moderator variables. 42 WW Table 2 depicts the sociodemographic characteristics of the NLSY respondents. The respondents' age and parental education levels were collected in 1979; the annual income, marital status, and completion of high school or GED were collected in 1984. The information is presented by the scores on the Family Alcoholism variables. For subsequent analyses, the three groups were treated independently, except when the two groups with alcoholic parents were combined. As Table 2 indicates, notable discrepancies between the groups were identified on annual income, marital status, and high school completion. Children of alcoholics tended to have lower annual incomes, were more likely to be married and less likely to have completed high school or received a GED than children of non-alcoholics. 'v ' 'cs Tables 3, 4, and 5 display the correlations among the independent and dependent variables (also see Appendix, Tables 1A 8 2A) . Table 6 shows the means and standard deviations for each of the dependent variables. These analyses revealed that Family Alcoholism and Length of Exposure to alcoholism were highly correlated (r = .74) . The Length of Exposure variable did not contribute significantly to the prediction of the outcome variables beyond what was accounted for by Family Alcoholism. Thus, in order to avoid redundancy and the problems of multicollinearity, the Length of Exposure variable 43 Table 2. Sociodemographic Characteristics of the Sample by Family Alcoholism FAMALCH=0‘ FAMALCH=1 FAMALCH=2 (N=3497) (N=830) (N=72) Respondent's M = 17.71 M = 17.68 M = 18.08 Age in 1979 £2 = 2.27 ég = 2.34 g9 = 2.18 Mother's M = 9.04 M = 9.14 M = 9.36 I Education Level” g9 = 4.47 so = 4.28 g9 = 4.3 k Father's M = 7.52 M = 7.71 M = 8.43 % Education Level §Q = 5.74 §Q = 5.28 SQ = 5.22 E Respondents Aged 19-23 in 1984 Annual Incomec bedian = 2,000 [Median = 1,500 iMedian = 2,000 Marital Status Eever Married=79% Never Married=79% Never Married=79% arried = 14% arried = 16% [Married = 21% Separated = 2% Separated = 2% Separated = 0% Divorced = 1% Divorced = 1% Divorced = 0% Completion of Yes = 62% Yes = 56% = 41% High School or No = 33% No = 43% = 59% GED Respondents Aged 24-27 in 1984 Annual Income‘ [Median = 6,000 [Median = 4,400 = 3,250 Marital Status ever Married=53%£l:ever Married=49% Never Married=35% arried = 33% arried = 37% arried = 43% Separated = 4% Separated = 6% eparated = 9% Divorced = 3% Divorced = 7% Divorced = 9% Completion of Yes = 71% Yes = 68% 61% High School or No = 23% No = 30% 35% GED 'Number of Alcoholic Biological Parents bHighest Grade Completed by Respondent’s Parent cRespondent's Total Income in Wages and Salary for 1984 44 Table 3. Correlations Among Family Alcoholism and Self-Concept Variables in 1979/1980 (gs range 4931 - 5287) ALCHEXPb ASPc DIFF‘ Loc° s-E‘ FAMALCH‘ .74** -.04** .02 -.02 .02 ALCHEXP -.04* .01 -.01 .02 ASP .32** .19** .08** DIFF -.07** -.01 LOC .09** *p < .05, ** p < .01, two-tailed 'Family Alcoholism (1988 survey) bLength of Exposure (1988 survey) cEducational Aspiration dAspiration-Expectation Discrepancy °Rotter Locus of Control Scale fRosenberg Self-Esteem Scale Table 4. Correlations Among Dependent Measures at Ages 19-28 (as range 4385 - 5034) ALCPROBS" ADQ° FI-ID‘l ALCHDEP‘ AVEDRUGSf VARDRUGS‘ MJ" DRANK' .11 .16 .10 .11 .08 .18 .19 ALCPROBS .39 .34 .56 .21 .30 .15 ADQ .89 .48 .18 .30 .15 FHD .44 .15 .22 .12 ALCHDEP .19 .26 .15 AVEDRUGS .66 .14 VARDRUGS .40 Mgtg. p < .01 for all entries 'Ever Had a Drink bAlcohol-Related Problems °Average Daily Quantity dFrequency of Heavy Drinking eAlcohol Dependency Symptoms (Average Yearly Use of Illicit Drugs “Level of Drug Experimentation l'Lifetime use of Marijuana 45 Table 5. Correlations Between Independent Variables, Moderator Variables, Alcohol-Related Problems, Alcohol Dependency Symptoms, and Drug Experimentation — F CH‘ A.C EXP” ASP“ s- Locc DIFF’ Females Aged 19-23 (gs range 985 - 1166) ALCPROBS‘ .13** .07** -.03 .00 .00 .05 ALCHDEP‘ .14** .07* —.09** -.02 .01 .01 VARDRUGS‘ .23** .13** -.11** -.04 -.00 .00 Males Aged 19-23 (gs range 1036 - 1247) ALCPROBS .11** .11** -.07* -.04 .01 .01 ALCHDEP .14** .11** -.09** -.12** -.06* .03 VARDRUGS .19** .18** -.03 .03 .01 .01 Females Aged 24-27 (ns range 1206 - 1430) ALCPROBS .13** .07** .02 -.09** -.01 .03 ALCHDEP .12** .08** -.03 -.10** -.O4 .04 VARDRUGS .15** .08** .11** -.06* .06* .03 Males Aged 24-27 (gs range 999 - 1189) ALCPROBS .05 .02 .01 .00 -.04 .03 ALCHDEP .09** .09** -.11** -.09** -.13** -.01 VARDRUGS .08** .05 .07* .02 .01 .09** *9 < .05, ** p < .01, two-tailed 'Family Alcoholism bLength of Exposure cEducational Aspiration dRosenberg Self-Esteem Scale ‘Rotter Locus of Control Scale fAspiration-Expectation Discrepancy 'Alcohol-Related Problems hAlcohol Dependency Symptoms iLevel of Drug Experimentation 46 TabLe 6. Descriptive Statistics for Dependent Variables at Ages 19-28 (n = 4576) M £2 Ever Had a Drink‘ .92 .26 Alcohol-related .32 1.05 Problemsb Alcohol dependency .41 .89 Symptomsc Average daily .44 .80 Quantity‘ Frequency of heavy 1.16 3.33 Drinkingc Lifetime use of 50.70 164.04 Illicit drugsf Level of drug 1.12 1.54 Experimentation“ Lifetime use of 381.72 457.59 Marijuanah “Ever Had a Drink (yes/no%) “Alcohol-Related Problems (sum of ten dichotomous items) cAverage Daily Quantity (drinks per day) “Frequency of Heavy Drinking (5+ drinks per day) eAlcohol Dependency Symptoms (sum of eight dichotomous items) fAverage Yearly Use of Illicit Drugs (mean of nine items, range = 0 - 1,000 or more Occasions; 0 = Never Used, 1 = 1-9 Occasions, 2 = 10-39 Occasions, 3 = 40-99 Occasions, 4 = 100-999 Occasions, 5 = 1,000 or more Occasions) 'Level of Drug Experimentation. (sum of ten dichotomous items) "Lifetime use of Marijuana (one item, coding same as for Average Yearly Use of Illicit Drugs) 47 was dropped from the remaining analyses. Another important finding from these exploratory analyses was that Family Alcoholism was most consistently and highly correlated with three of the outcome measures: Alcohol- related problems, Alcohol Dependency Symptoms, and Level of Drug Experimentation. This section will focus on the results with these outcome measures. Correlations and regression tests with the remaining outcome variables are reported in the Appendix. Other preliminary analyses with the independent.measures revealed that the reported mother and father educational levels were correlated positively (r = .64, p < .01). The parents’ educational level and the educational aspirations and expected level of attainment reported by the respondents were also correlated positively (rs ranging from .28 to .33, p < .01). Thus, the parents' educational level was pooled and used as a controlling factor in the correlation of between- group differences on self-concept. Parental education level is considered to be an index of socio-economic status and was used to control for the possible effects of familial social class on the respondents’ reported self-concept. ' o a o se - n e easures It was predicted that the four measures of self-concept would be correlated positively: high self-esteem, high internality, high educational aspirations, and low difference scores would be correlated as would low scores on the same 48 indices. Table 3's positive correlations between the Self- Esteem, Locus of Control, and Educational Aspiration indices and negative correlations between the Aspiration-Expectation Discrepancy score and the other measures supported this hypothesis. All correlations were significant at the .05 level of confidence, except between the Aspiration-Expectation Discrepancy score and Self-Esteem. Except for the .32 correlation between Educational Aspirations and Discrepancies, these associations were relatively small, ranging from -.07 to .19. Educational aspirations and expectations were highly correlated (r = .84, p < .01). 1 no 1“_S_° .au . , o o'sm and eso- dent's - 01‘! The second hypothesis predicted that children of alcoholics would score lower on the measures of self-concept, that is, they would report lower self-esteem, higher externality, lower’ educational aspirations, and. a larger discrepancy between their educational aspirations and expected level of attainment. This hypothesis was tested using partial correlations that controlled for the reported educational level of the respondents' parents. The results partially support this formulation, as the number of alcoholic parents was significantly negatively correlated with the reported Educational Aspiration level (3 [df = 4663] = -.06, p < .01) and the Discrepancy score (; [df = 4663] = -.03, p < .05). No significant correlations were found between Family Alcoholism and Self-Esteem or Locus of 49 Control. ,,.. ,-; ; 7 - 452- . ._ — - ; . ;- -c- -. u-.:u -- In Hypothesis III it was predicted that a moderator effect would be found between familial problem drinking or alcoholism and indices of substance use by the respondents. This hypothesis was tested by two methods, as noted earlier. Table 7 shows the results for the first part of this analysis, the Multivariate Multiple Regression. The overall multivariate model using Family Alcoholism as a predictor of the eight substance use measures with the entire sample was significant. While the results for the overall model were significant, Family Alcoholism only accounted for a small percent of the variance in the outcome measures. The largest percent (2.1) accounted for was in Level of Drug Experimentation. Level of Drug Experimentation increased .14 units for each one-unit increase in Family Alcoholism. Tables 8, 9, 10, and 11 present the independent hierarchical regression analyses by age and gender. These analyses tested for the main effects of Family Alcoholism and the four moderator variables in predicting Alcohol-Related Problems, Alcohol Dependency Symptoms, and Level of Drug Experimentation. (See Appendix, Tables 3A-6A, for the hierarchical regression results for the remaining outcome measures). 50 Table 7. Summary Results of Multivariate Multiple Regression Analysis Predicting Substance Use Variables (M = 4068): The Main Effects of Family Alcoholism on the Outcome Measures Predictor Outcome Variable Beta R? F p Level of drug .144 .021 85.72 <.01 experimentation Lifetime use of .102 .010 42.64 <.01 marijuana Alcohol-related .097 .009 38.74 <.01 problems Family Alcohol dependency .095 .009 36.99 <.01 Alcoholism symptoms Lifetime use of .077 .006 24.43 <.01 illicit drugs Ever had a drink .065 .004 17.24 <.01 Average daily .037 .001 5.58 <.05 quantity Frequency of heavy .030 .001 3.60 .05 drinking 51 Table 8. Hierarchical Regression Analyses Predicting Outcomes from Family Alcoholism, Self-Esteem, and their Interactions Family Alcoholism W/ Self-Esteem Interaction Outcome Variable R2 Beta Rz/Ch Beta Rz’Ch Beta Females aged 19-23 (g = 677) Alcohol-related .016** .128** .000 .129** .000 -.087 Problems F=15.2 §=3.90 F=.28 t=3.92 F=.38 §=-.25 Alcohol dependency .021** .144** .000 .145** .000 .183 Symptoms F=19.29 t=4.39 F=.12 é=4.40 F=.Ol t=-52 Level of drug .060** .245** .000 .243** .003 -.290 Experimentation F=58.04 §=7.62 F=.48 t=7.55 F=2.43 t=-.85 Males aged 19-23 (g = 661) Alcohol-related .017** .130** .000 .130** .006* .902** Problems F=15.48 t=3.94 F=.44 t=3.93 F=5.63 §=2.76 Alcohol dependency .017** .131** .014** .130** .000 .330 Symptoms F=15.73 t=3.97 F=12.77 t=3.96 F=.38 t=1-01 Level Of drug .029** .l7l** .004 .172** .001 -.l96 Experimentation F=27.29 t=5.22 F=3.42 t=5.25 F=1.29 t=-.60 Females aged 24-27 (g = 1343) Alcohol-related .Ol3** .ll4** .009** .lll** .000 .21 Problems F=14.36 t=3.79 F=9.66 t=3.68 F=.11 t=.68 Alcohol dependency .022** .148** .007** .145** .000 .323 Symptoms F=24.38 §=4.94 F=7.28 t=4.84 F=.34 t=l.06 Level of drug .026** .160** .004* .158** .000 .378 Experimentation F=28.66 t=5.35 F=4.77 t=5.28 =.52 t=1.24 Males aged 24-27 (g = 1133) Alcohol-related .002 .049 .000 .049 .000 .166 Problems F=2.12 §=l.45 F=.15 §=1.45 F=.08 §=.39 Alcohol dependency .001 .031 .005* .032 .001 -.434 Symptoms F=.85 §=.92 F=4.75 §=.95 F=l.24 §=-l.0 Level of drug .003 .055 .001 .055 .002 -.458 Experimentation F=2.65 t=1.63 F=.79 t=1.62 F=1.50 t=-1.1 m V g v i Note: These Beta weights are for Family Alconolism only. *9 < .05, **p < .01, two-tailed. 52 Table 9. Hierarchical Regression Analyses Predicting Outcomes from Family Alcoholism, Educational Aspiration, and their Interactions Family Alcoholism W/ Educational Interaction Aspiration Outcome Variable R2 Beta RZ/cn Beta Rz'Ch Beta Females aged 19-23 (g = 677) Alcohol-related .016** .128** .000 .128** .003 .018 Problems F=15.2 t=3.90 F=.09 J§=3.88 F=2.91 _=.25 Alcohol dependency .021** .144** .000 .143** .001 .073 Symptoms F=19.29 t=4.39 F=.l7 t=4.37 F=1.l9 t=1.02 Level of drug .060** .245** .005* .241** .006* .086 Experimentation F=58.04 t=7.62 F=5.08 t=7.52 F=6.14 t=1.23 Males aged 19-23 (M = 661) Alcohol-related .017** .130** .006* .128** .003 .218** Problems F=15.48 t=3.94 F=5.51 §=3.88 F=2.51 t=3.32 Alcohol dependency .017** .131** .013** .128** .000 .160* Symptoms F=15.73 t=3.97 F=12.06 t=3.90 F=.32 §=2.45 Level of drug .029** .l7l** .002 .171** .003 .272** Experimentation F=27.29 t=5.22 F=l.50 §=5.l9 F=3.21 t=4.16 Females aged 24-27 (g = 1343) Alcohol-related .013** .114** .001 .116** .003 .008 Problems F=l4.36 t=3.79 F=.9l t=3.84 F=3.22 t=.12 Alcohol dependency .022** .148** .000 .149** .000 .125 Symptoms F=24.38 t=4.94 F=.29 t=4.96 F=.16 t=1.86 Level of drug .026** .160** .018** .168** .000 .154* Experimentation F=28.66 t=5.35 F=19.90 t=5.64 F=.05 t=2.3l Males aged 24-27 (g = 1133) Alcohol-related .002 .049 .000 .049 .001 .096 Problems F=2.12 t=1.45 F=.162 t=1.46 F=.50 t=1.29 Alcohol dependency .001 .031 .013** .032 .000 .015 Symptoms F=.85 t=.92 F=ll.07 t=.95 F=.07 _=.20 Level of drug .003 .055 .006* .054 .001 -.017 Experimentation F=2.65 t=1.63 F=5.27 t=1.61 F=1.14 t=-.23 Note: These Beta weights are for Family Alcoholism only. *9 < .05, **p < .01, two-tailed 53 Table 10. Hierarchical Regression Analyses Predicting Outcomes from Family Alcoholism, Educational Aspiration-Expectation Discrepancy, and their Interactions Family Alcoholism W/ Discrepancy Interaction Score Outcome Variable R2 Beta Rz/Ch Beta RZ’Ch Beta Females aged 19-23 (g = 677) Alcohol-related .016** .128** .002 .127** .007** .085 Problems F=15.2 i;=3.90 F=1.96J§=3.87 F=6.91 t=2.33 Alcohol dependency .021** .144** .000 .144** .001 126** Symptoms F=19.29 t=4.39 F=.03 2+4.38 F=l.l9 t=3.46 Level of drug .060** .245** .000 .245** .004 .215** Experimentation F=58.04 §=7.62 F=.01 t=7.61 F=3.47 t=6.02 Males aged 19-23 (g = 661) Alcohol-related .017** .130** .000 .131** .000 .142** Problems F=15.48 t=3.94 F=.35 t=3.95 F=.44 t=3.81 Alcohol dependency .017** .131** .001 .132** .000 140** Symptoms F=15.73 t=3.97 F=.52 t=3.99 F=.25 t=3.77 Level of drug .029** .l7l** .000 .l72** .003 146** Experimentation F=27.29 t=5.22 F=.00 t=5.22 F=2.30 t=3.95 Females aged 24-27 (g = 1343) Alcohol-related .Ol3** .ll4** .001 .ll3** .005* .076* Problems F=14.36 t=3.79 F=.99 t=3.73 F=5.43 t=2.24 Alcohol dependency .022** .148** .000 .147** .001 .127** Symptoms F=24.38 t=4.94 F=.33 t=4.90 F=1.59 t=3.76 Level of drug .026** .160** .001 .159** .001 .140** Experimentation F=28.66 t=5.35 F=.86 t=5.30 F=1.49 t=4.14 Males aged 24-27 (g = 1133) Alcohol-related .002 .049 .001 .048 .001 .060 Problems F=2.12 t=1.45 F=.76 t=1.42 F=.08 t=1.57 Alcohol Dependency .001 .031 .000 .031 .000 .026 Symptoms F=.85 §=.92 F=.00 t=.93 F=.08 t=.69 Level of Drug .003 .055 .008** .052 .000 .052 Experimentation F=2.65 t=1.63 F=6.89 t=1.54 F=.00 t=1.37 Note: These Beta weights are for Family Alcoholism only. *p < .05, **p < .01, two-tailed 54 Table 11. Hierarchical Regression Analyses Predicting Outcomes from Family Alcoholism, Locus of Control, and their Interactions Family AlcoholismIW/ Locus of Control Interaction Outcome Variable R2 Beta R2/Ch Beta Rz’Ch Beta Females aged 19-23 (§,= 677) Alcohol-related .016** .128** .000 .128** .005* -.191 Problems F=15.2 t=3.90 F=.01 =3.90 F=4.50};=-1.24 Alcohol dependency .021** .144** .001 .144** .001 .017 Symptoms F=19.29 t=4.39 F=1.00 £=4.40 =.71 t=.11 Level of drug .060** .245** .000 .245** .002 .036 Experimentation F258.04 §=7.62 F=.01 ¢=7.61 F=2.014t=.24 Males aged 19-23 (g = 661) Alcohol-related .Ol7** .l30** .000 .l30** .000 .198 Problems F=15.48 g=3.94 =.00 J£=3.93 F=.22 J£=1.34 Alcohol dependency .017** .131** .001 .130** .000 .204 Symptoms F=15.73J§=3.97 F=.97 §=3.94 F=.26 t=1.38 Level of drug .029** .l7l** .001 .172** .002 .360** Experimentation F=27.29 t=5.22 F=.69 t=5.24 F=1.71 t=2.45 Females aged 24-27 (2.: 1343) Alcohol-related .Ol3** .ll4** .000 .114** .003 -.123 Problems F=14.36j§=3.79 F=.36 t=3.77 F=3.14 _=-.90 Alcohol dependency .022** .l48** .000 .149** .000 .120 Symptoms F=24.384§=4.94 F=.37 t=4.95 F=.05 t=.88 Level of drug .026** .l60** .003* .162** .000 .213 Experimentation F=28.66 t=5.35 F=3.81 t=5.40 F=.15 t=1.57 Males aged 24-27 (g = 1133) Alcohol-related .002 .049 .001 .050 .000 .016 Problems F=2.12 t=1.45 F=.98 t=1.49 F=.05 t=.10 Alcohol dependency .001 .031 .013** .035 .000 -.062 Symptoms F=.85 t_=.92 F=ll.88 _t_=1.06 F=.39 Lynn Level of drug .003 .055 .000 .054 .001 -.109 Experimentation F=2.65 t=1.63 F=.29 t=1.61 F=1.10 t=-.68 Note: These Beta weights are for Family Alcoholism only. *3 < .05, **p < .01, two-tailed 55 The hierarchical regression analyses revealed that while the main effects of Family Alcoholism were almost always significant, the self-concept variables did not consistently add to the variance accounted for by Family Alcoholism. Self- Esteem and Educational Aspiration accounted for significant additional variance in about half of the subsequent analyses (5/ 12 and 6 / 12, respectively) . The Aspiration-Expectation Discrepancy measure and Locus of Control rarely accounted for any significant additional portion of the variance (1/12 and 2/12, respectively). As in the Multivariate Multiple Regression for the overall model, the percent of additional variance accounted for was very small. No moderator variable accounted for more than 2% of the variance in the outcome measures. Each of the interaction effects that was significant accounted for less than 1% of the variance. Mypgthgsig Ty: Age-related differences It was predicted that significant moderator effects would be more likely for older respondents. This hypothesis was alstpartly supported, Some age differences in the effects of the moderators were identified in the regression results. For example, Self-Esteem contributed significantly to the variance accounted for in each of the outcome measures in the group of older females (see Table 9), but not for the other groups. Contrarily to what was hypothesized, Family Alcoholism had the least significant effects in the group of older males and no significant moderator effects were identified. Additionally, 56 the data from the younger females revealed more significant interaction effects overall than any of the other groups. Four of the six statistically significant interaction effects were for this group. The analyses also revealed one other gender effect: Educational Aspirations accounted for significant additional variance more often for males than for females (4/6 and 2/6, respectively, see Table 10). DISCUSSION This study sought to identify linkages in the substance abuse literature on family history of alcoholism and self- concept as independent predictors of drug and alcohol use in a sample of disadvantaged young adults. Due to the fact that analyses were not run with the general sample, the results from this study may be specific to ethnic minority group members and low-income Caucasians, but not to the general population. Analyses revealed that while some of the specific hypotheses of this study were supported, generally the relationships were statistically significant, but not large enough to support interpretations of practical significance. The "buffer" model of self-concept as a moderator between family alcoholism and patterns of substance use by the respondents received only minimal support. While some studies have identified personality variables that moderate the relationship between family risk factors and substance abuse (e.g., Rogosch, Chassin, 8 Sher, 1990) , others have failed to find such moderator effects (e.g., Brook, Lukoff, 8 Whiteman, 1977) . Brook, Lukoff, and Whiteman (1977) tested three models of relationships between personality variables and drug use among a sample of African Americans and West Indians. These authors found support for an independent 57 58 model, but not for a interdependent (i.e. , moderator) or mediational model. They found that peer, family, and personality factors each contributed significantly to adolescent drug use, despite statistical control for variables in the other domains. The results of Brook, Lukoff, and Whiteman's (1977) study and the present investigation suggest that among African Americans and other minority groups, as well as low-income Caucasians, personality factors such as self-concept may be weak moderators to substance abuse. The factors that serve as protectors for children from disadvantaged families need further exploration. Regarding the multidimensionality of self-concept, this study partly corroborated previously reported positive correlations among different measures of self-concept. Statistically significant but quite weak (.19 to -.01) correlations were found among four measures hypothesized to reflect self-concept: self-esteem, locus of control, educational aspirations, and hopefulness about academic success. These limited values likely reflect psychometric weaknesses in at least some of the measures. Consistent with major trends in the substance-abuse literature, only inconsistent findings were identified fOr between-group differences in self-concept based on family history of alcoholism. The prediction of lower self-concept for children of alcoholics was supported for Educational .3 ~ahm‘ 0". a". u. 59 .Aspirations and.the.Aspiration-Expectation Discrepancy score, but not for Self-Esteem or Locus of Control. In the research literature, the negative relationship between academic involvement and substance use has been more consistently supported than that between self-esteem or locus of control and substance use. Findings from studies on locus of control and substance use are particularly inconsistent. Windle’s review (1990b) of the substance abuse literature concluded that research findings generally only provided meager support for premorbid differences in personality characteristics. .A family history of alcoholism was consistently correlated positively with respondents' substance use, corroborating previous research. Interestingly, this relationship was not found for older males. Family alcoholism did not significantly predict substance use for this group. This could be due to social pressures for males to use alcohol and drugs recreationally or to the lesser social stigma associated with alcoholism in males. It might also reflect a tendency for older males to purposely avoid repeating patterns of familial substance abuse. Previous studies using the NLSY data and other samples have consistently reported higher rates of drug and alcohol consumption and more negative alcohol- related consequences for males. The finding of differences by age suggests need for further exploration. Overall, the ”buffer" model was only meagerly supported. The self-concept variables did not consistently add to the 60 variance accounted for by Family Alcoholism, although Self- Esteem and Educational Aspirations were somewhat better moderators than the measures of Locus of Control and hopefulness about academic success. Few significant interactions effects were found between Family Alcoholism and the self-concept measures. There were some interesting age and gender differences in the moderator effects. Self-Esteem consistently added to the variance accounted for by Family Alcoholism only among the older females. This finding was in the predicted direction. For males, Educational Aspirations contributed the most additional variance to Family Alcoholism. This gender difference might be a reflection of societal gender roles, which place more emphasis on academic and professional success for males than females. Contrary to what was expected, significant interaction effects were more often identified among the group of younger females, suggesting that the self- concept measures were exhibiting stronger moderation effects for this group than the others. However, this effect is difficult to interpret since self-concept is thought to be a less reliable index for younger respondents. Limitations of the study Several limitations of this study warrant discussion. Although the NLSY data were based on a nationally representative sample, there were major difficulties involved with using the data for a study that focused on the specific 61 factors related to substance abuse and self-concept. The NLSY surveys were very comprehensive; however, their main focus concerned factors related to labor market participation. Data on psychological well-being and substance use were less specific and were collected with less consistency (i.e. , only during certain years). This contributed to what was perhaps the most critical problem, which was an inability to measure self-concept concurrently with substance abuse or to assess the reliability of the self-concept measures during the period from late adolescence to adulthood. This issue was partly addressed by conducting separate analyses for older and younger respondents. However, this method could not control for individual variation on the self-concept measures during the f ive-year interval between baseline and the measurement of the outcome variables. Previous studies have suggested that self-concept may be unstable during adolescence and young adulthood (Block, 1971; cited in Bates 8 Pandina, 1991; Kaplan, Robbins, 8 Martin, 1984) . While there may be naturally occurring changes in self-concept for a number of reasons, including role changes (e.g., Bachman, O'Malley, 8 Johnston, 1984; Hammer 8 Vaglum, 1990) and maturing, relatively few studies have examined how these naturally occurring changes relate to substance use. One longitudinal study (Bates 8 Pandina, 1991) found that changes in personality were related to increased substance use among males, especially those who were initially considered to 62 be at risk. The inability to assess substance use and self-concept concurrently or to use repeated measures of self-concept in this study precluded determining the direction of the relationship between self-concept and substance use. Additionally, it is impossible to iknow if the lack of significant findings was due to low-order correlations between the selected variables or to the instabilities of the self- concept measures over the four-year delay between assessment of self-concept and assessment of substance use. Other weaknesses of this study have been alluded to previously. These included: the lack of specificity of the self-concept and substance abuse measures and the reliance on self-report data obtained from interviews with the participants. Also, the Locus of Control scale employed in this study was a markedly abbreviated version of the original scale. This diminished the reliability of findings from this measure and likely contributed to the nonsignificant results. Future Directions The present findings suggest that precipitators of drug use in young adulthood may be.different for males and females. The mechanisms that serve to protect children of alcoholics from abusing substances may also vary by gender. Studies on the etiology of substance use suggest that women tend to use alcohol for more.escapist reasons (e.g., Beckman, 1980; Windle 8 Blane, 1989) than do men and to report more psychological 63 problems associated with their alcohol and drug use (Dawson 8 Grant, 1993) . For instance, Hurley (1991) indicated that alcoholic women were more likely than alcoholic men to report having poor self-concepts, distorted self-images, and low self-esteem. Poor self-concept in these women was hypothesized to result from traumatic childhood experiences or stressful life events in adulthood, and to exacerbate the risk for alcoholism. While there have been. a few' studies examining’ the experiences of female alcoholics, the specific variables that precipitate substance use or serve as protective mechanisms for female children of alcoholics seems to be fertile ground for future research. Previous studies of resilient children of alcoholics indicate that they are more likely to be female (Werner 8 Smith, 1982) . Determining why and at what ages females are more likely than males to resist substance abuse seems a promising avenue for further research. REFERENCES REEERENCES Alterman, A.I. and Searles, J.S. (1989). Failure to find differences in drinking behavior as a function of familial risk for alcoholism: A replication. Journal of Abnormal Eaxonoloox. 28. 50-53. Bachman, J.G.; O'Malley, P.M.; Johnston, L.D. (1984). Drug use among young adults: The impacts of role status and social environment. Journal gr Personality and Social onshologY, 41. 629-645- Bachman, J.G.; Johnston, L.D.; O’Malley, P.M. (1981). Smoking, drinking, and drug use among American high school students: Correlated and trends, 1975-1979. Amarigan Jgurnal a; Public Health, ll, 59-69. Barnard, C.P. and Spoentgen, P.A. (1986). Children of alcoholics: Characteristics and treatment. Alcgholism Treatment Quarterly, 1(4), 47-65. Barnes, G.M.; Farrell, M.P.; and Cairns, A.. (1986). Parental socialization factors and adolescent.drinking behaviors. Journal of Marriage and tho Family. 48. 27-36. Baron, R.M. and Kenny, D.A. (1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Earaonalltx and aooial Baronoloox. 51. 1173-1182. Bates, M.E. and Pandina, R.J. (1991). Personality stability and adolescent substanceause behaviors. Algangliam: olinioal and Experimental anaaron. 1a. 471-477- Bechtel, L.J. and Swisher, J.D. (1992). An analysis of the relationships among selected attitudinal, demographic, and behavioral variables and the self- reported alcohol use behaviors of Pennsylvania adolescents. Lournal of Alcohol and Drug Eouoauion. 11, 83-93. 64 65 Beckman, L.J. (1980). Perceived antecedents and effects of alcohol consumption in women. Journal of flies on AlQQDQl. $1. 513-530- Berkowitz, A. and. Perkins, H.W. (1988). Personality characteristics of children of alcoholics. Jggrnal gr tonsultine and olinioal.£sxonoloex. so. 206-209. Brook J.S., Lukoff I.F., and Whiteman, M. (1977). Peer, family, and personality domains as related to adolescent drug behavior. Mayra}, Bapgrra, 41, 1095-1102. Brown, B. and Tooley, J. (1989). Alcoholism in the Black community. In G. W. Lawson and A. W. Lawson (eds. ) Alcoholism .and .suostanoe. Abuse. in .fioeoial Papalariaaa. Aspen Publisher: Rockville. Byrne, B.M. (1983). Investigating measures of self-concept. Maasuramenr and Evaluarioa i3 Guidance, la, 115-126. Center for Human Resource Research (1990). NLS Man_baak raga: The Matignal Lo agitudiaal Survaya 9_ Labor Markar W. (Available from The Ohio State University; Columbus, Ohio). Cermak, T.L. and Rosenfeld, A.A. (1987). Therapeutic considerations with adult children of alcoholics. gouanoes in Alcohol and substance Abuse. 5(4). 17- Cervantes, R.C.; Gilbert, M.J.; de Snyder, S.N.; and Padilla, A.M. (1991). Psychosocial and cognitive correlates of alcohol use in younger adult immigrant and U.S. born Hispanics. Special Issue: Use and misuse of alcohol and drugs: Ethnicity issues. International Journal of the Addietions. 25, 687-708. Churchill, J.G.; Broida, J.P.; and Nicholson, N.L. (1990). Locus of control and self-esteem of adult children of alcoholics. Journal of Studies on Alcohol. 51. 373-375. Coombs, R.H. and Paulson, M.J. (1988). Contrasting family patterns of adolescent drug users and nonusers. Journalofsneniealpepeneeneulreatnemz. 1(2). 59- 7 Cotton, N.S. (1979). The familial incidence of alcoholism: A review. Journal of Studies on Alcohol. so. 89-116. 66 Crowley, J.E. (1985a). The demographics of alcohol use among young americans: Results from the 1983 National Longitudinal Survey of Labor Market Experience of Youth. Center for Human Resource Research: The Ohio State University. Crowley, J.E. (1985b). Status variation in alcohol use among young adults: Results from the 1984 National Longitudinal Surveys of Youth. Center for Human Resource Research: The Ohio State University. Crowley, J.E. (1983). Demographics of alcohol use. among respondents of the 1982 National Longitudinal Survey of Labor Market Experience of Youth Panel. Center for Human Resource Research: The Ohio State University. Dawson, D.A. and Grant, B.F. (1993). Gender effects in diagnosing alcohol abuse and dependence. gagrnal gr EfilgthQQXI i2. 293'307- Donovan, J.E.; Jessor, R.; Jessor, L. (1983) . Problem drinking in adolescence and young adulthood. 193ml 91 Studies on Aloobol. 44. 109-137- Drake, R. E. and Vaillant, G. E. (1988). Predicting alcoholism and personality disorder in a 33-year longitudinal study of children of alcoholics. Bririsn Qaarnal 21 Addiction. 5.. P799'307- Fawzy, F.I.; Coombs, R.H.; and Gerber, B. (1983) . Generational continuity in the use of substances: The impact of parental substance use on adolescent substance use. Addietiue bebauiors. 8. 109- 114 Fawzy, F.I.; Coombs, R.H.; Simon, J.M.; Bowman-Terrell, M. (1987). Family composition, socioeconomic status, and adolescent substance abuse. agaiariya Sebauiors. 12. 79-83- Garza, R. and Widlak, F.W. (1977). The validity of locus of control dimensions for Chicano populations. Jaarnal gr Eersonalitx Assessment. 41. 635-643- Gfoerer, J. (1987). Correlation between drug use by teenagers and drug use by older family members. Amarigag Mrnal Qt Drug and Aloobol Abuse. 13. 95-108- Gfoerer, J. (1985) . Influence of privacy on self-reported drug use by youth. In B. A. Rouse, N. J. Kozel, 8 L. G. Richards (EdS) Selfzreuort netboas of Estimating Drug use: Meeting ourrent onallenoes to _alioitx NIDA Research monograph 57. DHHS Publication 67 No.(ADM) 85-1402. Rockville, MD: U.S. Government Printing Office. Goldsmith, R. E. (1986). Dimensionality of the Rosenberg Self- Esteem Scale. Journal of Social bebaxior and Personalitu. 1. 253- 264. Grant, B.F.; Harford, T. C.; Grigson, M.E. (1988). Stability of alcohol consumption among youth: A national longitudinal survey. Mal a; srndiea an Alanhgl, 52, 253-260. Gross, J. and McCaul, M.E. (1991). A comparison of drug use and adjustment in urban adolescent children of substance abusers. M12921 Mild; 2f the andictions, z§(4-A), 73-88. Hammer, T. and Vaglum, P. (1990). Use of alcohol and drugs in the transitional phase from adolescence to young adulthood. Jaurnal a; Adalascanca, 1;. 129-142. Humphrey, J.A.; Stephens, V.; Allen, D.F. (1983). Race, sex, marihuana use and alcohol intoxication in college students. Journal of Studies on Alcohol. 41. 733- 738. Hurley, D.L. (1991). Women, alcohol and incest: An analytic review. Journal of Studies on Ml. S2. 253-268- Hyphantis, T.; Koutras, V.; Liakos, A.; Marselos, M. (1991). Alcohol and drug use, family situation and school performance in adolescent children of alcoholics. Ina lnternationalJouLIEofScah i l Esucliiatrr. ll. 35-42- Jaccard, J.; Turrisi, R.; and Wan, C.K. (1990). lnraragrign m in Mnlrinla gegrassi on. Newbury Park: Sage Publications. Jessor, R. and Jessor, S.L. (1977). Problem Benavior and Esucnosocial Emlcnnentz A Lon—on 'tudinal _xstud of 1.11.110 t . New York: Academic Press. Johnson, G.M.; Shontz, F.C.; and Locke, T.P. (1984). Relationship between adolescent drug use and parental drug behaviorsw :agolescence, l2, 295-299. Johnston, L.D.; O’Malley, P.M.; and Bachman, J.G. (1987). National Trends in Drug Use and _Lt_dRe a e Factors Angng_ American Mign School Srudents and_ Manna Adnlra, l975-l98g. DHHS Publication No. (ADM) 87- 1535. Rockville, MD: National Institute on Drug Abuse. 68 Johnston, L.D.; O'Malley, P.M.; and Bachman, J.G. (1986). ang useAnoncAnerjtanhiuhSclmlSJtuden. college StudurLs. anddtherYouncAdults: Nationallrends Ihrangn lggfi. Rockville, MD. National Institute on Drug Abuse. Jurich, A.P. and Polson, C.J. (1984). Reasons for drug use: Comparison of drug users and abusers. angnglagigal am: 5.5: 371-378 0 Kandel. 0-8- (Ed- 1978). LonciMinal Research on Jruc use. New York: Halstead Press. Kandel, D.B.; Kessler, RJC.; Margulies, R.z. (1978). Antecedents of adolescent initiation into stages of drug use: A developmental analysis. Mrnal 91 Xanrn ang Adolescenge, Zé. 270-289. Kaplan, H.B.; Martin, S.S.; Robbins, C. (1984). Pathways of adolescent drug use: Self derogation, peer influence, weakening social controls, and early substance use. Journal of Health and Mel W. 22. 270-289. Kaplan, H.B.; Robbins, C.; Martin, 8.8. (1984). Subcultural variation in multivariate models of self-attitudes and delinquent behavior- Research in connuru'tx and Mental Health. 5. 73-108- Kern, J.C.; Hassett, C.A.; Collipp, P.J.; Bridges, C.; Solomon, M.; and Condren, R.J. (1981). Children of Alcoholics: Locus of control, mental age, and zinc level. Journal of Exchi_t_i.s ar'c W and Exaluation. l. 169-173- Khantzian, E.J. and Mack, J.E. (1983). Self-preservation and the care of the self. Ego instincts reconsidered. Esychoanalutic Smdx of _h_t e _h;__C '1d. 38. 209-232- Krystal, H. and Raskin, H. A. (1970). gm 9 Qenendenca A§n_gr§ 2f, Egg mm. Wayne State University Press: Detroit. Labouvie, E.W. and McGee, C.R. (1986) . Relation of personality to alcohol and drug use in adolescence. Jonrnal gr consulting and clinical Esxcholocx. SA. 289-293. Lex, B.W. (1987). Review' of alcohol problems in ethnic minority 9rOIIps- Journal of consulting and clinical Earsholocx. as. 293-300. 69 Little, C.B. (1979). Stability of internal-external control as measured by the Rotter scale. gnarnal at gagial Boreholodx. 198. 127-128. Little, C.B. (1977). Dimensionality of the Internal-External Control Scale: Some further evidence. ganrnal Q: Senetic Psychology. 121. 329-330. Marsh, H.W. and Richards, G.E. (1986). The Rotter locus of control scale: The comparison of alternative response formats and implications for reliability, validity, and dimensionality. Jgnrnal Qf Baaaargn in Eersonalitx. 2d. 509-528- Marsh, H.W. (1986). Global self-esteem: Its relation 'to specific facets of self-concept and their importance- Journal of Bersonalitu and Social EEXQDQLQQY. 51. 1224-1235- Marston, A.R.; Jacobs, D.F.; Singer, R.D.; Widaman, K.F.; and Little, T.D. (1988). Adolescents who apparently are invulnerable to drug, alcohol, and nicotine use. Adolescence. 22. 594-598. Martin, M.J. and Pritchard M.E. (1991). Factors associated with alcohol use in later adolescence. ganrnal at Studies on 8122921. 22. 5'9- McDermott, D. (1984). The relationship of parental drug use and parents' attitude concerning adolescent drug use to adolescent drug use. Adolescence, l_9_(73), 89-97. McNeill, V.N. and Gilbert, B.O. (1991). External locus of control in children of alcoholics. 2W BERQISS. dd. 528-530- Mensch, B.S. and. Kandel, D.B. (1988). Underreporting' of substance use in a national longitudinal youth cohort. Eublic opinion Quarterlx. s2. 100-124. Miller-Tutzauer, C.; Leonard, K.E.; and Windle, M. (1991). Marriage and alcohol use: A longitudinal study of "maturing out-" Journal.of.Studies.on.Alcohol..52. 434-440. National Institute on Drug Abuse (1990). Marianal Manganala Surxex.1mn luau: Abuse: turn: Findings genus. DHHS Publication No. (ADM) 90-1682. Rockville, MD: National Institute on Drug Abuse Press Office. 70 National Institute on Drug Abuse (1986). Highlight: Qf tn; 1282Rationalhousehold§uuexonbmc£>usaNIDA Capsule, Rockville, MD: National Institute on Drug Abuse Press Office. National Institute on Drug Abuse (1980). W Qn Drug Abuse: Selected contemoorarx oersoectiues- Research Monograph 30. DHHS Publication No. (ADM) 80-967. Rockville, MD: U.S. Government Printing Office. Newcomb, M.D. and Bentler, P.M. (1988a). W a; AQQlescant Drag D_a. Newbury Park: Sage Publications. Newcomb, M.D. and Bentler, P.M. (1988b). The impact of family context, deviant attitudes, and emotional distress on adolescent drug use: Longitudinal latent-variable analyses of mothers and their children. anrnal Q: Research in Rersonalitx. 22. 154-176- Newcomb, M.D.; Bentler, P.M.; Collins, C. (1986). Alcohol use and dissatisfaction with self and life: A longitudinal analysis of young adults. The Jonrnal .91 2139 WI 1.9. 479-494- Newcomb, M.D. and Bentler, P.M. (1986). Drug use, educational aspirations, and work force involvement: The transition from adolescence to young adulthood. Anarigan Journal of communitx Radiology. 14. 303-321. O’Brien, E.J. (1985). Global self-esteem scales: Unidimensionality or multidimensionality? Rsxcholocical Reports. S1. 383-389- Oetting, E. R. and Beauvais, F. (1990). Adolescent drug use: Findings of national and local surveys. Q_nrnal gr consultincandclinioaleucholm. SS. 385- 394. Office for Substance Abuse Prevention (1990). AlQQan and otherdrucuseisaMmcernforAfrican Americanfamiliesandcommunities- 0.8. Department of Health and Human Services: Rockville, MD. Omizo, M.M.; Omizo, S.A.; and Michael, W.B. (1987). Relationship of Locus of Control Inventory for Three Achievement Domains to two other locus of control measures: A construct validity study. Educational and Rsvcholocical Reasuement. 41. 737- 742. 71 Pandina, R.J. andeohnson,‘V. (1990). Serious alcohol and.drug problems among adolescents with a family history of alcoholism. Journal of Studies on Alcohol. S1. 278- 282. Pandina, R.J. and .Johnson, V. (1989). Familial. drinking history as a predictor of alcohol and drug consumption among adolescent children. JQnrnal Qf Studies on Alcohol. 5.0. 245-253. Pandina, R.J. and Schuele, J.A. (1983). Psychosocial correlates of alcohol and drug use of adolescent students and adolescents in treatment. anrnal 91 Studies on Athol. 44. 950-973. Parish, J.G. and Parish, T.S. (1991). Support systems functionality, self-concepts and alcohol use. Qallaga Student Journal. 25. 470-472- Polich, J.M. (1982). The validity of self-reports in alcoholism research. Agaigriya Danavigrs, 1, 123- 132. Prendergast, M.L.; et a1. (1989). Substance abuse among Black youth. Drevenrion Rasearcn Update. A, 1-32. Rearden, J.J. and Markwell, B.S. (1989). Self-concept and drinking problems of college students raised in alcohol- abused homes. Agaigriya Banaviars, l1, 225-227. Richman, C.L; Clark, M.L; and Brown, K.P. (1985). General and specific self-esteem in late adolescent students: Race x gender x SES effects. Adglescence, 39, 555-566. Roberts, A.E. and Reid, P.N. (1978). Dimensionality of the Rotter I-E scale. ggnrnal 9; Social W, lQQ, 129-130. Rogosch, F.; Chassin, L.; and Sher, K.J. (1990). Personality variables as mediators and moderators of family history risk for alcoholism: Conceptual and methodological issues. JQnrnal Qr Stuaies Qn Alcohol, 21. 310- 318. Rosenberg, M. (1965). Society and the Adolescent Self-lnaga. Princeton, N.J.: Princeton University Press. Rotter, J.B. (1966). Generalized expectations for internal versus external control of reinforcement. Egycnolggical MQanrapna, AD (1, whole no. 609). 72 Russell, M. (1990) . Prevalence of alcoholism among children of alcoholics. In M. Windle 8 J. S. Searles (Eds.), thldrcn Qt W. The Guilford Press: New York. Rutter, M. (1987). Psychological resilience and protective . mechanisms. Merican m1 of Qrthocsxchiatrx. 5.1. 315-331. Rutter, M. (1985). Resilience in the face of adversity. British Journal of Esxchiatrx. 141. 598-611. Schinke, S.; Orlandi, M.; Vaccaro, D.; Espinoza, R.; McAlister, A.; and Botvin, G. (1992) . Substance use among Hispanic and non-Hispanic adolescents . Addictiue Rehauiors. 11. 117-124. Sher, K.J.; Walitzer, K.S.; Wood, P.K.; and Brent, E.E. (1991). Characteristics of children of alcoholics: Putative risk factors, substance use and abuse, and psychooatholoqy. Journal of Abnormal Esxcholocx. inn, 427-448. Svobodny, L.A. (1982). Biographical, self concept, and educational factors among chemically dependent adolescents. Aaalaaaanga, ll, 847-853. Tyler, F.B.; Gatz, M.; and Keenan, K. (1979) . A constructivist analysis of the Rotter I-E scale. Journal if Rersonalitx. 41. 11-35. Wallace, J. (1987). Children of alcoholics: A population at risk. W13 Treatnenr Quartarly, 913), 13-30. Wells, L.E. and Rankin, J.H. (1983). Self-concept as a mediating factor in delinquency. Sgcial anganng Quarterly. 45. 11-22. Werner, L.J. and Broida, J.P. (1991). Adult self-esteem and locus of control as a function of familial alcoholism and dysfunction. Journal 9; Studies Qn Alcohol. 52. 249-252- Werner, E.E. (1986). Resilient offspring of alcoholism: A longitudinal study from birth to age 18. Mal Qr Studies on Alcohol. 41. 34-40. Werner, E.E. and Smith, R.S. (1982) . Vulnerable but lnyiainla: A Studx of Resilient childr_n. New York: McGraw- Hill. 73 Windle, M. and Miller-Tutzauer, C. (1991). Antecedents and correlates of alcohol, cocaine, and alcohol-cocaine abuse in early adulthood. Journal of Drug Education. 21. 133-148. Windle, M. (1990a). A longitudinal study of antisocial behaviors in early adolescence as predictors of late adolescent substance use: Gender and ethnic group differences. Journal of Abnormal Esxoholodx. an, 86-91. Windle, M. (1990b) . Temperament and personality attributes of children of alcoholics. In M. Windle 8 J.S. Searles (Eds.) , Qnildran at W. The Guilford Press: New York. Windle, M. and Blane, H.T. (1989). Cognitive ability and drinking behavior in a national sample of young adults. Alconolia : Clinical and Erparinanral Research. 12. 43-48. Windle, M. and Barnes, G.M. (1988). Similarities and differences in correlates of alcohol consumption and problem behaviors among male and female adolescents. The International Journal of the Addictions. 22. 707-728. Yanish, D.L. and Battle, J. (1985) . Relationship between self- esteem, depression and alcohol consumption among ' adolescents. Baycnological Reports, 51. 331-334. Zucker, R.A. and Harford, T.C. (1983). National study of the demography of adolescents drinking practices in 1980. Journal of Studies on Alcohol. 11. 974-985. APPENDIX .7 .3.“ ——.I_.'__.._f _ . . . _ Table 1A. 74 Correlations Between Independent and Dependent Measures (ns range 4249 - 5038) FAMALCH' ALCHEXP" ASP“ 0m" LOC“ s-E' DRANK‘ .07" .05" .06** -.00 .03* .O6** ALCPROBS" .05“: .04** .04“: .00 -.00 .18** 11130i .04** .04": -.04** -.00 .01 .01 moi .07“ .07" -.05** .01 -.02 .10" ALCHDEP" .08** .07“ .01 .01 -.04** .19" avnnnuos' .09" .05** -.02 .04" .02 .00 VARDRUGS" .14** .09** .02 .03* .03* .01 MJ‘ .12** .07** -.02 .04" .04“: -.01 *p < .05, ** p < .01, two-tailed llFamily Alcoholism bLength of Exposure °Educational Aspiration “Aspiration-Expectation Discrepancy °Rotter Locus of Control Scale 'Rosenberg Self-Esteem Scale ‘Ever Had a Drink fAlcohol-Related Problems fAverage Daily Quantity ’Freguency of Heavy Drinking kAlcohol Dependency Symptoms lAverage Yearly Use of Illicit Drugs IIILevel of Drug Experimentation nLifetime use of Marijuana 75 Table 2A. Correlations Between Independent Variables, Moderator Variables, Ever Drank, Average Daily Quantity, Frequency of’ Heavy' Drinking, Use of Illicit Drugs, and Marijuana Use FAMALCH‘ ALCHEXP" ASP‘ DIFF“ LOC° s-E‘ Females Aged 19-22 (ns range 985 - 1166) DRANK‘ .09" .08** .04 -.04 .00 -.O6 ADQ' .13" .12** -.09** -.02 -.02 -.06* FHD“ .12** .09** -.11** -.01 -.02 -.04 AVEDRUGSJ .14** .09“: .08* -.02 -.00 -.07* MJ" .13“: .07* -.05 -.02 -.05 -.03 Males Aged 19-22 (ns range 1036 - 1247) DRANK .08** .05 .04 .02 -.01 -.03 ADQ .07* .06* -.04 .01 .04 -.01 FHD .07* .07* -.07* .02 .04 -.00 AVEDRUGS .08* .09** .00 .01 -.Ol .05 NJ .08** .09** -.O7* .03 .04 -.03 Females Aged 23-27 (ns range 1206 - 1430) DRANK .08** .08** .12** .02 .10** .02 ADQ .09** .03 .05 -.02 .02 -.05 FHD .08** .02 -.02 -.02 -.O3 -.05* AVEDRUGS .05 .01 .06* .06* -.04 -.06* NJ .11** .03 .05* .00 .04 -.01 Table Continues. 76 Table 2A (cont’d.) — FAMALCH' ALCHEXP" ASP“ DIFF“ LOC“ S-E‘ Males Aged 23-27 (ns range 999 - 1189) DRANK .01 .00 .03 .01 -.03 -.02 ADQ .01 .03 -.05 .03 -.04 -.09** PRO .01 .05 -.09** .03 -.06* -.08** AVEDRUGS .05 .01 .09** .05 .03 .03 NJ .10** .07* -.04 .04 -.04 -.06 *p < .05, ** p < .01, two-tailed 'Family Alcoholism I’Length of Exposure °Educational Aspiration “Aspiration-Expectation Discrepancy cRotter Locus of Control Scale fRosenberg Self-Esteem Scale 'Ever Had a Drink fAverage Daily Quantity {Frequency of Heavy Drinking JAverage Yearly Use of Illicit Drugs kLifetime use of Marijuana Table 3A. 77 Summary of Hierarchical Regression Analyses Predicting Outcomes from Family Alcoholism, Self-Esteem, and their Interactions Outcome Variable Family Alcoholism W/ Self-Esteem Interaction Females aged 19-23 (n = 677) R“ Beta RZ/cn Beta Rz’Ch Beta Ever had a drink .005* .073* .003 .069* .000 .267 F=4.82 §=2.20 F=2.37 =2.08 F=.32 t=.76 Average daily .027** .165** .002 .162** .001 .472 Quantity F=25.55 r=5.06 F=l.57 =4.96 F=.79 t=l.35 Frequency of heavy .030** .172** .002 .169** .001 .540 Drinking F=27.81 t=5.27 F=l.42 t=5.18 F=l.l4 t=l.55 Lifetime use of .027** .163** .005* .158** .001 -.116 Illicit drugs F=24.83 é=4.98 F=4.64 =4.83 F=.62 =-.33 Lifetime use of .016** .127** .000 .127** .001 -.145 Marijuana F=14.87 t=3.86 F=.07 =3.86 =.61 t=-.41 Males aged 19-23 (n = 661) Ever had a drink .009** .094** .001 .094** .000 -.027 F=8.09 r=2.84 F=.64 t=2.84 F=.14 r=-.08 Average daily .003 .054 .000 .054 .000 .054 Quantity F=2.59 t=l.6l F=.l7 t=l.6l F=.00 §=.l6 Frequency of heavy .003 .057 .000 .057 .001 .317 Drinking F=2.9l t=1.7l F=.10 t=l.7l F=.63 r=.96 Lifetime use of .005* .067* .005* .068* .000 .066 Illicit drugs F=4.10 t=2.02 F=4.73 t=2.05 =.00 r=.20 Lifetime use of .008** .091** .000 .091** .000 -.076 Marijuana F=7.60 t=2.76 =.15 t=2.75 F=.26 r=-.23 Table Continues. Table 3A (cont'd.) 78 Outcome Variable Family Alcoholism W/ Self-Esteem Interaction Females aged 24-27 (n = 1343) 122 Beta Rz/Ch Beta RZ’Ch Beta Ever had a drink .008** .089** .000 .090** .000 -.016 F=8.68 §=2.95 F=.43 r=2.97 F=.12 t=-.05 Average daily .003 .050 .004* .048 .001 .396 Quantity F=2.76 §=l.66 F=4.56 t=1.58 F=1.28 t=1.28 Frequency of heavy .001 .030 .005* .028 .002 .513 Drinking F=l.00 §=1.00 F=5.ll i£-'92 F=2.48 t=1.66 Lifetime use of .003 .053 .005* .050 .000 .248 Illicit drugs F=3.07 r=1.75 F=4.97 t=1.67 F=.41 t=.80 Lifetime use of .014** .119** .000 .119** .000 .008 Marijuana F=15.68 r=3.96 F=.02 t=3.95 F=.13 r=.03 Males aged 24-27 (n = 1133) Ever had a drink .000 .004 .001 .001 .001 .325 F=.00 r=.03 F=.89 r=.04 F=.60 r=.77 Average daily .000 .010 .002 .011 .005* -.835* Quantity F=.09 r=.3l F=2.18 §=.33 F=4.10 §=-2.0 Frequency of heavy .000 -.010 .002 -.009 .003 -.632 Drinking F=.08 _=-.29 F=2.09 t=-.27 F=2.22 r=-1.5 Lifetime use of .001 .038 .003 .037 .000 .068 Illicit drugs F=1.24 §=1.12 F=2.78 t=1.10 F=.Ol _=.l6 Lifetime use of .009** .094** .004 .095** .001 .565 Marijuana F=7.86 §=2.81 F=3.45 t=2.83 F=1.27 §=1.35 Note: These Beta weights are for Family Alcoholism only. *2 < .05, **p < .01, two-tailed. Table 4A. 79 Summary of Hierarchical Regression Analyses Predicting Outcomes from Family Alcoholism, Educational Aspiration, and their Interactions Outcome Variable Family Alcoholism W/ Educational Interaction Aspiration Females aged 19-23 (n = 677) R2 Beta Rz/Ch Beta 87th Beta Ever had a drink .005* .073* .001 .074* .000 .116 F=4.82 §=2.20 F=l.28 §=2.25 F=.4l t=1.60 Average daily .027** .165** .000 .164** .000 .142* Quantity F=25.55 t=5.06 F=.20 t=5.03 F=.12 t=1.98 Frequency of heavy .030** .172** .001 .171** .000 .204** Drinking F=27.81 t=5.27 F=.88 t=5.22 F=.26 t=2.84 Lifetime use of .027** .l63** .003 .161** .000 .165* Illicit drugs F=24.83 t=4.98 F=2.39 t=4.91 =.01 t=2.30 Lifetime use of .016** .127** .001 .125** .001 .069 Marijuana F=14.87 r=3.86 F=1.33 t=3.80 =.76 =.96 Males aged 19-23 (n = 661) Ever had a drink .009** .094** .001 .095** .000 .123 F=8.09 r=2.85 F=.67 £=2.86 F=.23 t=l.86 Average daily .003 .054 .001 .053 .001 .110 Quantity F=2.59 r=l.61 F=.62 §=l.59 F=1.0l t=1.67 Frequency of heavy .003 .057 .001 .056 .002 .132* Drinking F=2.91 r=l.7l F=.73 t=1.68 t=1.99 F=1.75 Lifetime use of .005* .067* .000 .068* .000 .057 Illicit drugs F=4.10 §=2.02 F=.13 t=2.03 F=.04 §=.86 Lifetime use of .008** .091** .000 .091** .000 .117 Marijuana F=7.60 §=2.76 F=.42 t=2.74 =.22 §=l.78 Table Continues. Table 4A (cont'd.) 80 Outcome Variable Family Alcoholism W/ Educational Interaction Aspiration Females aged 24-27 (a = 1343) R2 Beta RZ/Ch Beta Rz’Ch Beta Ever had a drink .008** .089** .025** .098** .004* .227** F=8.68 §=2.95 F=27.S7 r=3.27 F=4.7l r=3.4l Average daily .003 .050 .008** .055 .000 .021 Quantity F=2.76 r=1.66 F=8.33 4§=l.83 F=.32 §=.31 Frequency of heavy .001 .030 .000 .031 .001 -.037 Drinking F=1.00 §=1.00 F=.13 §=1.02 F=l.26 g=-.55 Lifetime use of .003 .053 .007** .058* .002 .136* Illicit drugs F=3.07 r=1.75 F=7.43 r=1.91 F=1.69 r=2.02 Lifetime use of .014** .119** .005* .123** .002 .221** Marijuana F=15.68 r=3.96 F=5.38 §=4.09 F=2.64 3:3.28 Males aged 24-27 (n = 1133) Ever had a drink .000 .004 .000 .004 .000 .004 F=.00 r=.03 F=.32 _=.02 F=.OO r=.01 Average daily .000 .010 .004* .011 .001 .084 Quantity F=.09 r=.31 F=3.82 £=-32 §=l.12 F=1.18 Frequency of heavy .000 -.010 .010** -.009 .001 .061 Drinking F=.08 r=-.29 F=8.88 _=-.26 r=.82 F=l.1l Lifetime use of .001 .038 .010** .037 .002 -.049 Illicit drugs F=l.24 r=l.12 F=8.92 r=1.10 F=1.66 _=-.66 Lifetime use of .009** .094** .001 .095** .000 .106 Marijuana F=7.87 r=2.81 F=.47 r=2.81 F=.03 r=1.42 Note: These Beta weights are for Family Alcoholism only. *2 < .05, **2 < .01, two-tailed Table 5A. 81 Summary of Hierarchical Regression Analyses Predicting Outcomes from Family Alcoholism, Educational Aspiration-Expectation Discrepancy, and their Interactions Outcome Variable Family Alcoholism W/ Discrepancy Interaction Score Females aged 19-23 (n = 677) 122 Beta RZ/Ch Beta RZ’Ch Beta Ever had a drink .005* .073* .004 .074* .000 073* F=4.82 §=2.20 F=3.37 t=2.24 F=.OO t=1.98 Average daily .027** .165** .000 .165** .003 .137** Quantity F=25.55 r=5.06 F=.03 r=5.06 F=3.21 t=3.75 Frequency of heavy .030** .172** .000 .172** .003 .145** Drinking F=27.81 r=5.27 F=.20 t=5.26 F=2.73 r=4.00 H; J.— Lifetime use of .027** 163** .001 .163** .002 .143** Illicit drugs F=24.83 t=4.98 F=.58 t=5.00 F=1.61 t=3.93 Lifetime use of .016** 127** .000 .127** .000 .123** Marijuana F=14.87 §=3.86 F=.05 t=3.86 F=.05 t=3.37 Males aged 19-23 (n = 661) Ever had a drink .009** 094** .000 .094** .000 .104** F=8.09 t=2.85 =.27 t=2.83 F=.33 §=2.78 Average daily .003 .054 .000 .053 .004 .084* Quantity F=2.59 t=1.6l =.38 t=1.59 F=3.32 §=2.25 Frequency of heavy .003 .057 .000 .056 .003 .085* Drinking F=2.9l t=l.7l F =.ll t=1.69 F=2.7O §=2.26 Lifetime use of .005* .067* .000 .066* .003 .041 Illicit drugs F=4.10 t=2.02 F=.27 t=2.01 F=2.29 r-l.09 Lifetime use of .008** .091** .001 .090** .000 .080* Marijuana F=7.60 t=2.76 F=l.l9 t=2.72 F=.36 §=2.15 Table Continues . Table 5A (cont’d.) 82 Outcome Variable Family Alcoholism W/ Discrepancy Interaction Score Females aged 24-27 (n = 1343) R2 Beta RZ/Ch Beta Rz’Ch Beta Ever had a drink .008** .089** .001 .088** .003 .ll4** F=8.68 t=2.95 F=.84 t=2.89 F=2.87 t=3.35 Average daily .003 .050 .000 .051 .001 .063 Quantity F=2.76 gél.66 F=.50 t=1.70 F=.57 t=l.85 Frequency of heavy .001 .030 .000 .031 .000 .040 Drinking F=1.00 t=l.00 F=-47,§=1°04 F=.28 t=l.16 Lifetime use of .003 .053 .006** .049 .000 .059 Illicit drugs F=3.07 r=l.75 F=6.92 t=1.61 F=.44 t=1.74 Lifetime use of .014** .119** .000 .120** .002 .140** Marijuana F=15.68 r=3.96 F=.05 t=3.96 F=1.76 t=4.13 Males aged 24-27 (n = 1133) Ever had a drink .000 .004 .000 004 .002 -.023 F=.00 3:.03 F=.07 t=.02 F=1.74 §=-.6O Average daily .000 .010 .002 .009 .001 .029 Quantity =.O9 §=.31 F=1.49 =.26 F=1.28 _=.76 Frequency of heavy .000 -.010 .002 - 011 .001 .006 Drinking F=.08 r=-.29 F=1.56 t=-.33 F=’91i§='15 Lifetime use of .001 .038 .002 .036 .000 .030 Illicit drugs F=l.24 r=1.12 F=2.14 t=1.06 F=.12 E=-79 Lifetime use of .009** .094** .002 .093** .001 .075* Marijuana F=7.87 r=2.81 F=1.54 t=2.76 F=l.03 §=1.98 Note: These Beta weights are for Family Alcoholism only. *9 < .05, **p < .01, two-tailed 83 Table 6A. Summary of Hierarchical Regression Analyses Predicting Outcomes from Family Alcoholism, Locus of Control, and their Interactions Outcome Variable Family Alcoholism W/ Locus of Interaction Control Females aged 19-23 (n = 677) R2 Beta RZ/Ch Beta RZ’Ch Beta Ever had a drink .005* .073* .000 .073* .000 .040 F=4.82 r=2.20 F=.10 r=2.20 F=.05 r=.26 Average daily .027** .165** .000 .l65** .003 -.087 Quantity F=25.55 §=5.06 F=.12 £=5.05 F=2.82 r=-.56 Frequency of heavy .030** .l72** .000 .172** .001 .011 Drinking F=27.81 r=5.27 F=.13 r=5.27 F=l.l6 r=.07 Lifetime use of .027** .163** .000 .163** .001 .012 Illicit drugs F=24.83 §=4.98 F=.03 r=5.00 F=l.01 §=.08 Lifetime use of .016** .127** .002 .127** .001 -.046 Marijuana F=14.87 §=3.86 F=1.76 E=3.85 F=1.3l §=-.30 Males aged 19-23 (n = 661) Ever had a drink .009** .094** .000 .095** .000 .097 F=8.09 §=2.85 =.15 t=2.85 F=.00 r=.66 Average daily .003 .054 .006* .055 .000 .132 Quantity F=2.59 §=1.61 F=5.15 t=1.67 F=.28 r=.89 Frequency of heavy .003 .057 .009** .059 .000 .081 Drinking F=2.91 r=1.7l F=8.48 t=1.79 =.02 r=.54 Lifetime use of .005* .067* .000 .068* .000 .093 Illicit drugs F=4.10 r=2.02 F=.06 t=2.03 F=.03 r=.63 Lifetime use of .008** .091** .003 .093** .000 .004 Marijuana F=7.60 r=2.76 F=2.73 t=2.80 F=.38 r=.03 Table Continues . 0‘ 's"" ‘M'L‘ahmi‘. Table 6A (cont’d.) 84 Outcome Variable Family Alcoholism W/ Locus of Interaction Control Females aged 24-27 (n =1343) R2 Beta Rz/Ch Beta Rz’Ch Beta Ever had a drink .008** .089** .013** .092** .004* .355** F=8.68 r=2.95 F=14.39 £=3-05 F=3.92 §=2.60 Average daily .003 .050 .003 .052 .000 -.038 Quantity F=2.76 §=l.66 F=3.44 t=1.7l F=.45 r=-.28 Frequency of heavy .001 .030 .000 .030 .001 -.131 Drinking F=l.00 §=l.00 F=.00 t=l.00 F=1.45 §=-.95 Lifetime use of .003 .053 .001 .052 .000 .081 Illicit drugs F=3.07 r=l.75 F=l.23 t=1.73 F=.05 £3.59 Lifetime use of .014** .119** .004* .121** .000 .043 Marijuana F=15.68 r=3.96 F=4.09 t=4.01 F=.33 r=.32 Males aged 24-27 (n = 1133) Ever had a drink .000 .004 .001 .002 .001 .106 F=.00 r=.03 F=.59 r=.05 F=.45 r=.66 Average daily .000 .010 .000 .011 .000 .023 Quantity =.09 r=.31 =.10 _=.32 F=.01 r=.14 Frequency of heavy .000 -.010 .001 -.009 .001 .134 Drinking F=.08 £=’-29 F=.46 _=-.26 F=.84 §=.84 Lifetime use of .001 .038 .002 .036 .003 -.201 Illicit drugs F=l.24 t=1.12 F=1.75 r=l.07 F=2.32 r=-1.26 Lifetime use of .009** .094** .000 .095** .000 .080 Marijuana F=7.87 t=2.81 =.31 §=2.82 =.Ol _=.51 Note: These Beta weights are for Family Alcoholism only. *p < .05, **p < .01, two-tailed Table 7A. 85 Summary of Hierarchical Regression Analyses Predicting Outcomes from Family Alcoholism, Self-Esteem, and.their Interactions for Total Sample (M = 3778) Outcome Variable Family Alcoholism W/ Self-Esteem Interaction R2 Beta RZ/cn Beta Rz’Ch Beta Ever had a drink .004** .063** .001 .063** .000 .052 F=15.27 r=3.91 F=2.06 m=3.87 F=.OO r=.30 Alcohol-related .008** .089** .001 .088** .001 .34* Problems F=29.97 §;5°47 F=3.54 §=5.43 F=2.15 §=l.98 Average daily .001 .028 .002** .027 .000 -.098 Quantity F=3.07 r=1.75 F=6.69 t=1.69 F=.54 §=-.57 Frequency of heavy .001 .024 .002* .023 .000 .067 Drinking F=2.09 §=l.45 F=5.87 §=1.39 F=.07 §=.39 Alcohol dependency .007** .083** .006** .081** .000 .040 Symptoms F=26.09 §=5.ll F=24.32 §=5.00 F=.06 §=.24 Lifetime use of .005** .073** .000 .073** .000 .131 Illicit drugs F=20.43 r=4.52 F=.02 r=4.51 F=.12 _=.77 Level of drug .019** .138** .000 .l38** .000 -.068 Experimentation F=73.72 r=8.59 F=.09 r=8.57 F=l.49 _=-.40 Lifetime use of .010** .099** .000 .098** .000 -.045 Marijuana F=37.35 r=6.11 F=1.83 r=6.08 =.72 r=-.27 Note: These Beta weights are for Family Alcoholism only. *9 < .05, **p < .01, two-tailed. 86 Table 8A. Summary of Hierarchical Regression Analyses Predicting Outcomes from Family Alcoholism, Educational Aspiration, and their Interactions for Total Sample (M = 3778) Outcome Variable Family Alcoholism W/ Educational Interaction Aspiration 122 Beta Rz/Ch Beta R2’Ch Beta Ever had a drink .004** .063** .004** .065** .001 115** F=15.27 §=3.9l F=15.85 E=4.03 F=2.62 t=3.3l Alcohol-related .008** 089** .001 .088** .000 .078* Problems F=29.97 t=5.47 F=3.16Jr=5.42 F=.10 t=2.24 Average daily .001 .028 .001 .028 .000 .039 Quantity F=3.07 t=1.75 F=2.10J§=1.7l F=.12 t=l.ll Frequency of heavy .001 .024 .003** .022 .000 046 Drinking F=2.09 t=1.45 F=10.554§=1.36 F=.60 t=1.31 Alcohol dependency .007** 083** .006** .081** .000 065 Symptoms F=26.09 t=5.1l F=23.58 r=4.98 F=.25 t=l.87 Lifetime use of .005** 073** .001 .074** .000 081* Illicit drugs F=20.43 t=4.52 F=2.65 r=4.57 F=.05 t=2.31 Level of drug .019** 138** .001 .l39** .000 028 Experimentation F=73.72 t=8.59 F=4.57 r=8.65 F=.71 t=1.57 Lifetime use of .010** 099** .000 .099** .000 .ll4** Marijuana F=37.35 t=6.11 F=.02 §=6.10 F=.23 t=3.26 Note: These Beta weights are for Family Alcoholism only. *9 < .05, **p < .01, two-tailed Table 9A. 87 Summary of Hierarchical Regression Analyses Predicting Outcomes from Family Alcoholism, Educational Aspiration-Expectation Discrepancy and their Interactions for Total Sample (M = 3778) Outcome Variable Family Alcoholism W/ Discrepancy Interaction Score R2 Beta R2 / Ch Beta R2” Ch Beta Ever had a drink .004** .063** .000 .064** .000 .070** F=15.2? §=3.9l F=.ll t=3.92 F=.55 r=3.83 Alcohol-related .008** .089** .000 .088** .001 .076** Problems F=29.97 r=5.47 F=.89 g:5.44 F=2.33 §=4.15 Average daily .001 .028 .000 .028 .001 .040* Quantity F=3.07 t=1.75 F=.30 §=l.73 F=1.98 §=2.18 Frequency of heavy .001 .024 .000 .023 .000 .034 Drinking F=2.09 t=1.45 F=.49 r=1.42 F=1.69 r=1.86 Alcohol dependency .007** 083** .000 .083** .000 .077** Symptoms F=26.09 t=5.1l F=.20 i£=5.12 F=.604;=4.21 Lifetime use of .005** .073** .001 .072** .001 .060** Illicit drugs F=20.43 t=4.52 F=2.7IJ§=4.46 F=2.06 £33.32 Level of drug .019** .138** .001* .l37** .001* .121** Experimentation F=73.72 t=8.59 F=3.20 r=8.52 F=4.15 r=6.67 Lifetime use of .010** .O99** .000 .098** .000 .099** Marijuana F=37.35 t=6.11 F=.67 r=6.08 F=.00 r=5.43 Note: These Beta weights are for Family Alcoholism only. *p < .05, **p < .01, two-tailed. 88 Table 10A. Summary of Hierarchical Regression Analyses Predicting Outcomes from Family Alcoholism, Interactions for Total Sample (M = 3778) Locus of Control and their Outcome Variable Family Alcoholism W/ Locus of Interaction Control R2 Beta R2 / Ch Beta Rz’Ch Beta Ever had a drink .004** .063** .002* .064** .000 .055* F=15.27 g=3.91 F=7.42 i=3.“ F=1.76 t=3.03 Alcohol-related .008** .089** .000 .089** .000 -.005 Problems F=29.97 r=5.47 F=.22 g=5.47 F=1.7O _:-.06 j— Average daily .001 .028 .002** .029** .000 .014 Quantity F=3.07 J;=1.75 F=7.99 t=1.78 F=.04 r=.20 Frequency of heavy .001 .024 .001 .024 .000 .044 Drinking F=2.09 §=l.45 F=3.35 t=1.46 F=.08 r=.60 Alcohol dependency .007** .083** .001 .083** .000 .053 Symptoms F=26 ogfggs 11 F=2°34.§=5'1° F=.174r=.73 Lifetime use of .005** .073** .000 .073** .000 .057 Illicit drugs F=20.43|§=4.52 F=.02 4£54.52 F=.05j£=.78 Level of drug .019** .138** .002** .l39** .000 .147* Experimentatlon F=73.721;=8.59 F=7.28 t=8.62 F=.01 r=2.01 Lifetime use of .010** .099** .001 .099** .000 .017 Marijuana F=37.35 r=6.ll F=2.99 t=6.13 F=1.66 §=.96 Mara: These Beta weights are for Family Alcoholism *p < .05, **p < .01, two-tailed. only. are. "Il'llllllllll'llllr