P1: GDB Journal of Child and Family Studies [jcfs]
PH018-296862
February 2, 2001
12:50
Style file version Nov. 19th, 1999
Journal of Child and Family Studies, Vol. 9, No. 4, 2000, pp. 461–479
A Regional Study of Risk Factors for Drug Abuse and Delinquency: Sex and Racial Differences Diana H. Fishbein, Ph.D.1,3 and Deanna M. P´erez, Ph.D.2
Many juveniles are at high risk for delinquency and drug abuse by virtue of the characteristics of their neighborhoods, which are often economically impoverished, socially unstable, physically neglected, and rife with crime, drug use, and drug selling. However, a significant proportion of these youth do not engage in deviant behavior, while others participate to varying degrees. This study examined what factors discriminate between high-risk youth who do and who do not exhibit deviant behavior. Self-reported measures of several attitudes and behaviors known to correlate with drug abuse and delinquency were assessed in relation to reports of property offenses, person offenses, drug use, and drug selling. Additional analyses were conducted to distinguish between sex and racial groups to further qualify the relationships between deviant behaviors and attitudes. The most notable finding was that types of deviance that primarily involve material and monetary gain may be largely influenced by relationships with significant others, while deviance with potentially harmful consequences to both self and others are influenced largely by personal attitudes. Examination of individual variables showed that negative peer influences and positive attitudes toward fighting significantly increased the likelihood of reporting involvement in three of the four deviance measures, and that positive relationship with the father and prosocial values were inversely related to two of the behavioral deviance measures. There were very few differences in significant predictors of deviance for males versus females. Several self-reported attitudes and behaviors distinguished African Americans from Hispanics and Whites, even though Whites were reportedly responsible for a higher incidence of deviant behaviors. KEY WORDS: delinquency; drug abuse; risk; sex; race.
1 Senior Research Scientist, Transdisciplinary Behavioral Science Program, Research Triangle Institute,
Rockville, MD. Associate, Department of Criminology and Criminal Justice, University of Maryland, College Park, MD. 3 Correspondence should be directed to Diana Fishbein, Research Triangle Institute, 6110 Executive Blvd., Suite 420, Rockville, MD 20852. e-mail:
[email protected]. 2 Research
461 C 2000 Human Sciences Press, Inc. 1062-1024/00/1200-0461$18.00/0 °
P1: GDB Journal of Child and Family Studies [jcfs]
462
PH018-296862
February 2, 2001
12:50
Style file version Nov. 19th, 1999
Fishbein and P´erez
There is widespread interest in determining why similar environmental conditions produce divergent behavioral outcomes (see review in Loeber & StouthamerLoeber, 1998). The “hot spots” literature suggests that particular pockets or clusters of addresses within a community are more saturated with criminal activity than other areas or neighborhoods due largely to environmental conditions (Sherman, Gartin, & Buerger, 1989). Given that these areas tend to be infested with drug use, drug selling, and crime (Sherman & Weisburd, 1995), juveniles who live in or frequent these hot spots are considered to be at high risk for the same. However, not all juveniles who live in or are exposed to these conditions engage in deviant behavior. Numerous reports have cited several risk factors that predict drug abuse and delinquency with considerable validity (Brook et al., 1997; Bry, McKeon, & Pandina, 1982; Hawkins, Catalano, & Miller, 1992; Newcomb, Maddahian, & Bentler, 1986) that may increase understanding of why environmental risk factors do not provide a complete explanation of variability in deviance within high risk neighborhoods. Recognition of individual differences in vulnerability to, and resiliency against, social stressors (Anisman, Zaharia, Meaney, & Merali, 1998) would lead to more effective interventions to reduce these high risk behaviors. The notion of multiple causation asserts that propensity for delinquency and drug abuse can be understood as the result of dynamic interactions between many varied and diverse risk factors, rather than any one risk factor model alone. Social conditions which are most commonly associated with these behavioral problems include family functioning, parenting techniques, peer influences, and school environments (Hawkins et al., 1992). Studies also indicate that certain individuallevel characteristics (e.g., attitudes, personality, temperament, learning experiences, and biological factors) increase the risk for delinquency and drug problems (see Fishbein & Pease, 1996). The final behavioral outcome results from a combination of one or more of these social conditions uniquely interacting with individual characteristics. Thus, similar social and environmental conditions could, theoretically, produce variable outcomes in different individuals. In order to assess the origins of these differences, subjects exposed to similar environmental stressors with varying rates of delinquency and drug use can be examined with respect to distinguishing risk factors. The social development model, one of the most widely accepted schemata to identify risk factors, integrates theories of both social learning and social control perspectives by focusing on the socializing effects of social bonds with parents, peers, and societal institutions, and adherence to norms and values of society (Catalano & Hawkins, 1996; Hawkins & Weis, 1985; Kazdin, 1990). This model focuses on the extent to which individuals bond to prosocial family and peers, school, and societal norms which, in turn, act as protective factors against prevailing risks that otherwise contribute to the development of adverse behavioral outcomes, including both delinquency and drug abuse. Individual differences in the type and strength of these attachments are likely to explain variability in deviance among
P1: GDB Journal of Child and Family Studies [jcfs]
PH018-296862
Risk for Drug Abuse and Delinquency
February 2, 2001
12:50
Style file version Nov. 19th, 1999
463
individuals residing in the same neighborhood. Thus, the concepts of multiple causation, individual differences, and behavioral flux are inherently captured by application of the social development perspective to the study of predictors of drug use and delinquency. There is evidence to suggest that dimensions of the social development model (e.g., relationships with parents, commitment to school, and prosocial values) differ between sex and racial groups (Graham, 1997; Liu & Kaplan, 1999; Rienzi et al., 1996; Vaccaro & Wills, 1998; Vega, Zimmerman, Warheit, Apospori, & Gil, 1993), and that these differences may predict varying rates of offending and drug abuse. The gender gap in rates of both crime and drug use have been widely documented (Opland, Winters, & Stinchfield, 1995; Steffensmeier & Allan, 1996; Thomas, 1996). Although sex differences in predispositions to substance abuse have not been sufficiently studied, differences in age of onset, patterns of use, and consequences have been documented (Opland et al., 1995; Thomas, 1996). For example, males show higher rates of onset for most drugs and are less likely to quit using a drug than females (DeWit, Offord, & Wong, 1997), and males report significantly more involvement in delinquency, especially serious delinquency (Liu & Kaplan, 1999), while females are more often arrested for minor crimes and status offenses (Rhodes & Fischer, 1993). It remains unclear, however, as to whether the gender gap in crime and drug use is due to differential exposure to risk factors, differential responses to predisposing conditions, different risk factors altogether, or a combination of any of the above mentioned factors. Numerous racial/ethnic differences also exist in the nature, extent and possibly etiology of drug use and delinquency. Studies consistently report that African American adolescents have the lowest rate of substance use, with Hispanics showing intermediate rates, and Whites having the highest rates (SAMHSA, 1998; Vacarro & Wills, 1998; Vega et al., 1993; Wells et al., 1992). On the other hand, there is evidence that African American youth commit more crimes against persons than Whites (Farrington, Loeber, Southamer-Loeber, Van Kammen, & Schmidt, 1996), and that African American youth have higher levels of self-reported delinquent behavior and school trouble, as well as teacher reported aggressive behavior (Wells et al., 1992). When proper controls are intact for socioeconomic status (SES), however, prevalence rates for violent behavior have been found to be significantly higher for African Americans when compared to Whites only among the low SES group (Paschall, Flewelling, & Ennett, 1998). Racial differences in the incidence of drug abuse and delinquency may be largely due to differential exposure to prevailing risk factors, and the quality or quantity of factors necessary to increase risk may differ by group as well (Vega et al., 1993). In the context of the social development model, differential responses to risk factors by racial groups may be related to differences in socializing experiences, social bonds, attachment to normative systems, learning experiences, and associations. Importantly, differences in opportunities for involvement in particular activities and interactions
P1: GDB Journal of Child and Family Studies [jcfs]
PH018-296862
February 2, 2001
464
12:50
Style file version Nov. 19th, 1999
Fishbein and P´erez
with others may also play a role (Catalano, Kosterman, Hawkins, Newcomb, & Abbott, 1996). In light of evidence for individual differences in response to similar environmental conditions, the present study was designed to test the relative ability of various dimensions of the social development model (individual personality, attitudinal, social, and contextual factors) to explain deviant behavior in a sample of high risk juveniles. The research questions posed included the following: (1) Which subset of self reported variables (e.g., personality, attitude, immediate social, and community contextual factors) possess the greatest predictive value with respect to reports of property crimes, person crimes, drug use, and drug sales?; (2) Which individual variables best predict deviance measures?; (3) Do individual variables differentially predict deviance measures in males and females?; and, (4) Do individual variables differentially predict deviance measures in various racial/ethnic groups? Differences were expected between males and females in number and type of risk factors associated with their deviance; however, given discrepancies in the literature, no directional hypotheses were posed. Racial groups were also expected to show differences in the predictive value of community contextual factors in relation to self-reported deviance; i.e., low ethnic identification, school commitment and perceived lack of opportunities are expected to be more predictive of deviance in African Americans and Hispanics than in Whites.
METHODS This investigation of self-reported individual and social characteristics of high-risk youths was part of a larger study designed to evaluate the effects of programs and services provided by the Washington/Baltimore High Intensity Drug Trafficking Areas (HIDTA) through the Office of National Drug Control Policy (ONDCP). As such, objectives of the larger evaluation design dictated how sites and subjects were selected. Participants and control subjects were tested at baseline and each year for three years after exposure to HIDTA programs. The present analysis examines pre-exposure data only to identify significant predictors of drug use, drug sales, juvenile crimes against persons and juvenile crimes against property.
Subjects In response to a solicitation requesting proposals for neighborhood prevention programs, six sites in the Washington/Baltimore region were chosen for funding from ONDCP. Specific “high-risk” neighborhoods were targeted for services and programs within each of the following areas: Anacostia, Washington, D.C.; Montgomery County, Maryland; Baltimore City, Maryland; Loudoun County, Virginia; Fairfax-Falls Church, Virginia; and Prince William County, Virginia.
P1: GDB Journal of Child and Family Studies [jcfs]
PH018-296862
February 2, 2001
Risk for Drug Abuse and Delinquency
12:50
Style file version Nov. 19th, 1999
465
Designation of these target locations as high risk is based on official and anecdotal reports, and needs assessments which reflect crime rates, drug trafficking and usage patterns, poverty level, educational level, school attendance and drop-out records, fear of crime, perceptions of community, child abuse and neglect reports, racial tensions and other neighborhood features often associated with drug abuse, delinquency and violence. Subject selection procedures varied by site; however, all subjects resided in one of the six specific geographic regions targeted for prevention services. There was no random selection for program participation. Rather, youth were referred for services, typically by their school (e.g., guidance counselor or teacher), parent, or the juvenile court. Random selection was not employed in the present study because the aim was to provide programming to youth deemed to be at highest risk for drug abuse or delinquency. As a result, there exists the potential for selection bias in the final sample towards youth with an increased incidence or severity of risk factors. Control subjects were selected based on their comparability to programparticipating youth by virtue of neighborhood characteristics (e.g., infestation of drugs and crime), demographics, and prevailing risk factors (e.g., low grade point averages and high suspension rates, a high prevalence of single-parent homes, low parental supervision, parental substance abuse, or minimal parental involvement in their children’s lives). In the initial wave, 567 youth aged 10–17 years old were surveyed. Measures In this baseline cross-sectional survey, demographic, personality, attitudinal, and behavioral measures were obtained from 269 participants and 298 comparisons completing the Gang Resistance, Education and Training (GREAT) instrument, a self-administered questionnaire developed to evaluate the effects of the GREAT program nationwide (Esbensen & Deschenes, 1998). Several of the scales were adapted from the National Youth Survey (Elliott, Huizinga, & Ageton, 1985) or the Denver Youth Survey (Huizinga, Esbensen, & Weiher, 1991). A slightly shortened version of this questionnaire was administered to youths once they were recruited either for program participation (i.e., target youth) or for the comparison group (i.e., control youth). All baseline measures were collected prior to participation in the prevention program. Background information provided by this instrument included sex, race/ethnicity (white, African American, Hispanic, Asian or other), family structure (i.e., with whom they live), and age. Independent Variables Independent variables were examined both individually and in subsets constructed according to the social development model of prevailing risk factors
P1: GDB Journal of Child and Family Studies [jcfs]
PH018-296862
February 2, 2001
466
12:50
Style file version Nov. 19th, 1999
Fishbein and P´erez
(Hawkins & Weis, 1985). According to this scheme, “personality variables” included the scales self esteem, impulsivity, and social isolation (which contains items reflective of both depression and shyness); “personal attitudes” included attitudes toward police, attitudes toward fighting, and prosocial values (e.g., honesty, unfavorable attitudes toward cheating and stealing); “immediate social variables” included negative peer influences, relationship with mother, relationship with father, and parental supervision; and “contextual community variables” included school commitment, perceived limited opportunities, and ethnic identification (measuring the extent to which subjects feel positively about membership in their racial/ethnic group). High scale scores reflect the perception of high levels of that factor (e.g., high impulsivity or high school commitment). Subjects lacking any relationship with their father were instructed to leave blank the corresponding section; thus, items were coded as missing. Controls for sex, race/ethnicity, and age of the respondent were included in the analyses. Analyses within racial/ethnic sub-samples were restricted to African Americans, Whites, and Hispanics due to sample size limitations for the other groups.
Dependent Variables A self-report inventory containing 23 delinquency items was administered to the study participants. Respondents were asked to identify whether they had ever committed each of these delinquent acts. Four self-reported indices of deviant behavior were created: drug use, drug sales, juvenile crimes against persons, and juvenile crimes against property. The following items comprise the individual indices. Property offenses: purposely damaged or destroyed property that did not belong to you; illegally spray painted a wall or a building; stole or tried to steal something worth less than $50; stole or tried to steal something worth more than $50; went into or tried to go into a building to steal something; stole or tried to steal a motor vehicle Person offenses: hit someone with the idea of hurting them; attacked someone with a weapon; used a weapon or force to get money or things from people; shot at someone because you were told to by someone else; been involved in gang fights Drug Use: used tobacco products; used alcohol; used marijuana; used paint, glue or other things you inhale to get high; used other illegal drugs Drug Sales: sold marijuana; sold other illegal drugs such as heroin, cocaine, crack, or LSD. Individual items were summed to create a total score for each of these deviance indicators. However, because most of the observations were
P1: GDB Journal of Child and Family Studies [jcfs]
PH018-296862
February 2, 2001
Risk for Drug Abuse and Delinquency
12:50
Style file version Nov. 19th, 1999
467
confined to values of 0 and 1, prevalence measures indicating any involvement in a particular type of delinquency were created. The four prevalence measures were coded 0 for no instance of deviance or 1 for at least one self-reported deviant act. Although a number of the items comprising the property and person offense measures differed in degree of severity, a substantial percentage of the sample was under the age of 15 (47%). Given that more serious forms of offending occur in the later teen years (Elliott, Huizinga, & Menard, 1989; Farrington, 1986), it is likely that many of the delinquent-involved adolescents in our sample have not yet progressed to more serious offending. Furthermore, research has consistently shown that early aggressive behavior and involvement in minor offending significantly predicts later serious delinquent behavior (Farrington, 1995; Moffitt, 1993); therefore, the inclusion of minor delinquency items was warranted in the present analyses.
Analytic Technique In the first phase, logistic regression analyses were conducted to identify which of the four theoretical constructs most robustly predicted our deviance measures. Accordingly, each set of factors comprising a theoretical construct was regressed on each dependent variable. For all of the models, the demographic variables sex, race/ethnicity and age were entered first and then the theoretical variables were added. The sample size was controlled for each dependent variable across the four theoretical blocks in order to compare sample statistics across models. The Model Chi-square (χ 2 ) can be compared across the equations for each dependent variable to determine which of the estimated models provides a better fit to the data; the higher the value, the better the fit. Next, a comparison of the Akaike’s information criterion (AIC) can be made across each model to determine the relative influence of each construct on the dichotomous outcome measures; the lower the value, the better the model is at predicting the dependent variable (Allison, 1999). Next, a full model including all of the theoretical variables and the demographic controls was estimated for each dependent variable to identify the individual variables that were most strongly related to each of the dependent variables. The relative effect and strength of each factor in the overall model was measured by the regression coefficient (B), odds ratio (Odds), and Wald statistic (Wald). Finally, additional logistic regression models were run within groups (i.e., male and female; Black, White, and Hispanic) to identify which of the individual factors were most predictive of deviance for each of the racial/ethnic and sex groups separately.
P1: GDB Journal of Child and Family Studies [jcfs]
PH018-296862
February 2, 2001
12:50
468
Style file version Nov. 19th, 1999
Fishbein and P´erez Table I. Demographic Characteristics of Sample
Sex of Respondent Male Female
%
N
50.4 49.6
286 281
Average Age Race/Ethnicity White Black Hispanic Other
14.1
567
13.6 60.7 14.5 9.2
77 344 82 52
Family Structure Mother only Father only Both parents Other
39.0 5.6 36.7 15.7
221 32 208 89
RESULTS The demographic characteristics of this sample are presented in Table I. The mean age for this sample was 14.1 (SD = 2.23). Males comprised 50.4% and females were 49.6% of the sample. The majority were African American (60.7%), followed by Hispanics (14.5%), Whites (13.6%), and Other (9.2%) (due to their small numbers, Native Americans, Asians and others were combined into one category). Most (39.0%) lived only with their mother, 5.6% lived only with their father, 36.7% lived with both parents and 15.7% had other living arrangements. Given the number of conceptually related variables under consideration in this model, correlation coefficients were computed to check for multicollinearity. Most of the independent variables were significantly related to each other, albeit weakly (<.30). There were a few significantly moderate (between .40 and .60) relationships; relationship with mother was significantly correlated with relationship with father (r = .547) and parental supervision (r = .462). Commitment to school correlated with several variables: parental supervision (r = .471), attitudes towards police (r = .464), negative peer influences (r = −.521), and self esteem (r = .496). Attitudes towards police was also significantly associated with negative peer influences (r = −.463). Despite the number of significant bivariate relationships between our independent variables, the low values of the correlation coefficients suggest weak correlations. Given these findings, we can assume multicollinearity between our independent variables will not affect the stability of our model. The Chi-square test for independence was used to examine sex and racial group differences in lifetime prevalence of deviant behaviors. As seen in Table II, males were significantly more likely to report involvement in property and person
P1: GDB Journal of Child and Family Studies [jcfs]
PH018-296862
February 2, 2001
12:50
Style file version Nov. 19th, 1999
Risk for Drug Abuse and Delinquency
469
Table II. Prevalence of Self-Reported Deviant Behavior
Propertya,b Persona Drug useb Drug salesa,b a Difference b Difference
Male
Female
White
Black
Hispanic
Total Sample
52.7 52.7 52.2 23.4
41.5 42.8 51.3 13.1
61.8 50.7 75.7 39.7
44.2 50.1 51.6 16.4
38.5 39.7 32.5 9.0
44.9 47.1 50.5 18.1
is significant for males and females at the p ≤ .05 level of significance. is significant for three ethnic groups at the p ≤ .05 level of significance.
offenses, as well as drug sales, than females. Race and ethnic differences were observed for three of the deviant behavior measures, with Whites reporting greater prevalence of property offenses, drug use, and drug sales than Blacks and Hispanics. Social Development Model Predictors of Deviance Measures Separate logistic regressions were conducted with each subset of factors regressed on each outcome to identify the theoretical construct that best predicts each of the deviance measures. In all models, the demographic characteristics were entered first followed by the block of factors to determine the relative contribution of each set of factors beyond the influence of sex, race/ethnicity, and age. The results are presented in Table III. A comparison of the Model Chi-square statistics revealed that, holding constant the demographic characteristics, the model including the immediate social variables provided the best fit to the data for property offending (χ 2 = 84.32, p < .001) and drug sales (χ 2 = 102.57, p < .001), while the personal attitudes model provided the best fit to the data for person offending (χ 2 = 64.34, p < .001) and drug use (χ 2 = 156.03, p < .001). The Table III. Logistic Regression Model Estimates of Four Types of Self-Reported Deviant Behavior on Social Development Factors Property Offense (N = 444) χ2 Demographic traits Construct social Personality Attitudes Contextual < .05. < .01. c p < .001.
ap bp
Person Offense (N = 442)
AIC
χ2
14.55a 84.32c 49.10c 75.30c 64.72c
Drug Use (N = 441)
AIC
χ2
610.65
11.18a
548.89
53.94c
582.10 555.90 566.49
35.40c 64.34c 46.37c
Drug Sales (N = 441)
AIC
χ2
AIC
612.03
90.20c
533.15
52.34c
368.12
577.27
133.39c
497.96
102.57c
325.90
593.81 564.87 582.84
103.18c 156.03c 114.48c
526.18 473.33 514.87
72.95c 92.99c 89.78c
353.52 333.48 336.68
P1: GDB Journal of Child and Family Studies [jcfs]
PH018-296862
February 2, 2001
12:50
470
Style file version Nov. 19th, 1999
Fishbein and P´erez
immediate social variables were the best predictors of both property offending (AIC = 548.89) and drug sales (AIC = 325.90). In contrast, personal attitudes best predicted person offending (AIC = 564.87) and drug use (AIC = 473.33). Sample-wide Predictors of Deviance Measures Logistic regressions were conducted with all GREAT scales entered as independent variables, and sex, race/ethnicity, and age entered as controls, to identify the variables that significantly predict each of the deviance measures. The results for the four measures of deviance are summarized in Table IV. The scales predictive of property offending, in descending order of strength (as measured by the Wald statistic), are prosocial values, attitudes toward fighting, and relationship with the father. For person offenses, attitudes toward fighting, commitment to school, and negative peer influences were significantly predictive. Drug use was best predicted by all three personal attitude variables, attitudes toward police, prosocial values, Table IV. Logistic Regression of Four Types of Self-Reported Deviant Behavior Property Offense B Variables Mother Father Supervis Negpeer Impulse Esteem Isolate Police Prosocial Fight Schcomit Limopp Ethnic Sex Black Hispanic Other Age Intercept −2 LL Model χ 2 Sample size ap bp cp
< .05. < .01. < .001.
Odds Wald
−.09 −.16a −.30 .26 .07 −.17 .02 −.11 −.49b .27a −.27 .21 .29 .28 −.44 −.07 .56 −.07
.92 .85 .74 1.30 1.07 .84 1.02 .89 .61 1.31 .77 1.23 1.34 1.32 .65 .93 1.75 .93 3.66 493.95 119.26c 444
.63 4.27 3.53 2.97 .15 .95 .02 .30 7.67 5.57 1.87 1.46 3.10 1.43 1.49 .02 1.29 1.60
Person Offense B −.13 −.02 −.05 .35a .24 .24 −.14 −.36 .01 .39c −.54b .04 .20 .17 .11 −.36 −.22 −.16b
Odds Wald
Drug Use B
Odds Wald
Drug Sales B
Odds Wald
.87 1.64 −.07 .93 .35 .39b 1.47 .98 .05 −.12 .88 2.00 −.23a .79 .94 .12 −.24 .78 1.85 −.20 .81 1.41 5.17 .38a 1.46 4.85 .49a 1.63 1.27 1.89 −.14 .87 .51 .31 1.36 1.28 1.97 .28 1.33 2.18 .11 1.12 .86 1.25 .23 1.26 2.61 −.11 .89 .66 .69 3.19 −.73b .48 10.20 −.41 1.01 .00 −.50b .60 6.81 −.28 .75 1.47 12.15 .26a 1.29 4.77 .14 1.15 .58 8.01 −.30 .74 1.90 −.45 .63 1.04 .07 −.26 .77 1.90 .29 1.33 1.22 1.56 .21 1.23 1.26 −.36 .69 1.18 .52 −.18 .83 .48 .53 1.70 1.11 .10 −1.21b .29 8.74 −.88a .41 .39 .70 .65 −1.55b .21 8.77 −.94 .80 .21 −1.05 .35 3.81 −1.91a .14 .85 7.61 .22c 1.25 12.88 .16 1.17 1.82 1.79 −1.99 514.88 433.52 283.58 96.33c 177.83c 124.88c 442 441 441
7.18 5.49 .89 5.95 1.54 .25 .35 2.08 1.46 .76 3.37 1.63 2.42 2.46 4.67 2.27 4.82 3.22
P1: GDB Journal of Child and Family Studies [jcfs]
PH018-296862
February 2, 2001
12:50
Style file version Nov. 19th, 1999
Risk for Drug Abuse and Delinquency
471
and attitudes toward fighting, as well as by negative peer influences. Finally, drug sales was best predicted by three of the immediate social variables, relationship with mother, negative peer influences, and relationship with father. The common theme noted in these findings was the value of attitudes toward fighting and peer influences, and to some extent relationship with father, prosocial values, and attitudes toward police, in predicting self-reported deviance. To interpret the effect of an independent variable on the probability of delinquency, the following formula was used, b(Pi )(1 − Pi ), where b is the logistic regression coefficient and Pi is the probability of an event occurring. Examining the effect of each significant scale on the change in probability of delinquency reveals that a positive relationship with the father decreased the probability of property offending by 4.0% and the probability of drug sales by 3.3%, holding constant all other independent variables. Controlling for all relevant variables, prosocial values decreased the probability of property offending by 12.2% and drug use by 12.6%. Positive attitudes toward fighting increased the probability of property and person offending, and drug use by 6.6%, 9.7%, and 6.4%, respectively. At the mean of the dependent variable, negative peer influences and positive attitudes toward fighting increased the probability of person offenses by 8.6% and 9.7%, respectively, and the probability of drug use by 9.4% and 6.4%, respectively. Further, negative peer influences increased the probability of drug sales by 6.9%. Commitment to school reduced the probability of reporting involvement in person offenses by 13.5% and positive attitudes toward police decreased the probability of drug use by 18.2%. Oddly, a positive relationship with the mother was positively related to drug sales, increasing its probability by 5.5%. Findings also show that sex was not significantly related to any of the four outcome measures, and that African Americans and Hispanics were significantly less likely to report ever having used drugs than Whites. Further, African Americans and those in the Other category were significantly less likely to report having sold illegal drugs than White adolescents. Sex and Racial/Ethnic Group-Wise Predictors of Deviance In order to further test the hypothesis that relationships between the theoretical constructs and deviance measures would vary across sex and racial groups, logistic regressions were conducted separately for males and females, and for Whites, African Americans and Hispanics including each of the GREAT factors. Males and females in this population were distinguishable only by a few indices (see Table V). Relationship with father and attitudes toward fighting were significantly related to both property offending and drug use for males, while social isolation and attitudes toward police were significantly predictive of drug use for females. Poor school commitment was predictive of person offenses for both males and females, and was related to drug sales for males only, while impulsivity predicted
< .00.
∗∗ The low a p < .05. b p < .01.
cp
−2.68a Fight Schcomit
Negpeer Fight Schcomit
37.42c 60
58.88c 274
25.60a 64
63.78c 224
1.07a −2.23a
.64b .49b −.50a
−1.13a
.47a −.44a .66c −.70a
n.s
Negpeer Police Prosoc Limopp
**
Father Fight
Isolate Police
.53a
−.59a
Factor
B
18.63 59
83.89c 272
** 66
104.98c 224
84.94c 217
Drug Use
.42a −.88b −.56a −.47a
−.31a .38a
−1.17c
.40a
B
**
Mother Father
Supervis Impulse Isolate
Negpeer Schcomit Limopp
Father
Factor
** 60
42.61c 272
63.28c 65
90.90c 224
52.39c 217
Drug Sales
.38a −.31b
−3.33a 4.73a −2.37a
.53a −.85a 1.03b
−.45a
B
472
contain only those scales that are significantly related to the dependent variable at the p < .05 level for each group. prevalence of Whites reporting non-use and of Hispanics reporting drug sales invalidates estimation of that model.
43.37c 60
62.41c 274
−.21a .46a −.44a
Isolate
Negpeer Isolate Fight Schcomit
45.74c 218
Person
12:50
∗ Cells
Model χ 2 N
Prosoc
Father Negpeer Prosoc
30.16b 66
−.29a .50b
Impulse Schcomit
Factor
February 2, 2001
Hispanic
Model χ 2 N
Black
n.s.
82.01c 226
45.76c 218
B
PH018-296862
Model χ 2 N
White
Father Fight
n.s.
Property
Journal of Child and Family Studies [jcfs]
Model χ 2 N
Male
Model χ 2 N
Female
Factor
Table V. Significant Risk Factors for Four Types of Self-Reported Deviant Behavior, by Sex and Racial/Ethnic Group∗
P1: GDB Style file version Nov. 19th, 1999
Fishbein and P´erez
P1: GDB Journal of Child and Family Studies [jcfs]
PH018-296862
February 2, 2001
Risk for Drug Abuse and Delinquency
12:50
Style file version Nov. 19th, 1999
473
person offenses for females only. Negative peer influences predicted both person offenses and drug sales, social isolation and attitudes toward fighting predicted person offenses, and perceived limited opportunities predicted drug sales for males only. The only factor that predicted drug sales for females was poor relationship with the father. Negative peer influences played a significant role in African Americans for property offenses, person offenses, and drug use, but did not enter the equation for Whites or Hispanics (Table V). Poor relationship with the father predicted property offenses and drug sales, and lack of prosocial values predicted property offenses and drug use only for African Americans. Also for African American subjects, positive attitudes toward fighting and poor school commitment predicted person offenses, poor attitudes toward police and perceived limited opportunities predicted drug use, and interestingly, a positive relationship with mother predicted drug sales. High social isolation was a significant predictor of person offenses and drug sales only for White subjects. In addition, lack of parental supervision and high impulsivity predicted drug sales in Whites only. A lack of prosocial values predicted property offenses and positive attitudes toward fighting and poor school commitment predicted person offenses for Hispanics. Unfortunately, due to low prevalence of nondrug use in Whites and drug sales in Hispanics, the full model for these two dependent variables could not be adequately tested. DISCUSSION These findings suggest that types of deviance that primarily involve material and monetary gain (property offenses and drug selling) may be largely influenced by relationships with significant others, while deviance with potentially harmful consequences to both self and others are influenced largely by personal attitudes. Attitudes are thought to be derived from the immediate social environment (Toby, 1974; Witt, 1997); thus, further examination of these constructs are necessary to determine the source of their differential effects on deviance in this population. It is possible that some of the variables that comprise the “personal attitude” construct, such as self-reported attitudes reflective of positive feelings toward fighting and antisocial values, may be more a function of individual predisposition, e.g., conduct disorder or oppositional defiance disorder, rather than simply socially constructed attitudes. Interestingly, the personality variables (social esteem, impulsivity and social isolation) did not predict any deviance measure. Given the ethnic diversity, transience, and racial tension documented within this population, it was anticipated that poor ethnic identification and perceived limited opportunities would be related to self-reports of deviant behaviors and attitudes (Davis, Moss, Kirisci, & Tarter, 1996; Oetting, Donnermeyer, & Deffenbacher, 1998), and that these variables would be particularly predictive of deviance in
P1: GDB Journal of Child and Family Studies [jcfs]
474
PH018-296862
February 2, 2001
12:50
Style file version Nov. 19th, 1999
Fishbein and P´erez
minority groups. The lack of any association between ethnic identification and deviance in each analysis conducted may be related to the geographic distribution of subjects in this region; the majority of this sample resides in pockets within communities that are dominated by their own ethnic or racial group. Perhaps these homogeneous pockets provide for a higher self estimation of cultural identity, and lesser perceptions of being in the “minority”. In the separate regressions for racial groups, however, perception of limited opportunities significantly predicted drug use exclusively among African American subjects. Thus, ethnic identification may be less important than perception of limited opportunities for these “minority” subjects. The finding that negative peer influences significantly predicted three of the four measures of deviance was consistent with previous research, suggesting that commitment to deviant peers plays a central role in the development of drug abuse (Kandel, Kessler, & Margulies, 1978; Kaplan, Martin, & Robbins, 1984) and delinquent behaviors and attitudes (Fergusson & Horwood, 1999; Paetsch & Bertrand, 1997). In the present study, negative peer influences were particularly related to deviance in males and in African Americans, in spite of the finding that Whites reported a greater involvement in deviant behaviors. These findings are discrepant with reports from investigators who have concluded that, while peer influences are important for all racial groups, Black youths may be somewhat less susceptible to peer influences (Gillmore et al., 1990; Newcomb & Bentler, 1986). Due to the small sample sizes for both Whites and Hispanics, however, it is likely that the regression analyses performed lacked sufficient statistical power to detect significant relationships between peer influences (as well as other risk factors) and behavioral outcomes (Cohen, 1988). Poor relationship with the father played an important role in self-reported deviance measures for the entire sample, and for both males and African Americans in particular. This finding is interesting in light of the fact that the analyses included only subjects with either a father or father-figure in the home or a relationship with the father outside of the home. Most studies on parental influences examine family structural factors, parental relationships, and paternal absence without isolating the effects of a paternal relationship. There is evidence, however, that paternal drinking, drug abuse and psychopathology are associated with negative affect, hostility, and externalizing behaviors in the offspring which is, in turn, related to their eventual drug use and delinquency (Frick et al., 1992; Nigg & Hinshaw, 1998; Sher, Walitzer, Wood, & Brent, 1991). Brook, Whiteman, Gordon, & Brook (1983) reported that, even with adjustments for mother-son relationship, the father’s personality attributes and socialization techniques are associated with the son’s use of marijuana. The present study did not examine parental substance use or psychopathology, although further research should include these paternal conditions as possible mediators of the relationship between the nature of interactions with the father and their children’s drug use and delinquency.
P1: GDB Journal of Child and Family Studies [jcfs]
PH018-296862
Risk for Drug Abuse and Delinquency
February 2, 2001
12:50
Style file version Nov. 19th, 1999
475
Counterintuitively, in the present study, a good relationship with the mother predicted higher levels of drug sales in these subjects. Group analyses showed that this association was primarily a function of the positive relationship between drug sales and relationship with the mother in African American subjects. In accordance with the social development model, deviance in the mother would be highly predictive of deviance in the child. Additional studies are needed to determine whether maternal deviance is responsible for this positive relationship. The variable “prosocial values” showed significant relations with drug use and property crimes for the entire sample, and was a significant predictor of property crimes and drug use for African Americans, and of property crimes for Hispanics. Prosocial values are a component of the social development model in its assessment of the extent to which respondents agree with conventional norms and values of society; specifically, unfavorable attitudes towards lying and stealing (Akers, Krohn, Lanza-Kaduce, & Radosevich, 1979; Krohn & Massey, 1980). Beliefs in the values of social units in which youth develop can be either protective or risk factors, depending upon the particular values espoused within that immediate environment (Catalano & Hawkins, 1996). In this sample, values inconsistent with those of the larger society appear to play a significant role in the actual deviant behavior of subjects. These results are consistent with Esbensen and Deschenes (1998) who concluded that processes of social development have more explanatory value than attachment to social institutions. Overall, these findings provide support for the social development model which highlights the important roles of peer influences, paternal relationships, and adherence to norms and values of society (Hawkins & Weis, 1985; Hawkins et al., 1992). Youths considered to be at high risk by virtue of their neighborhoods are more likely to harbor antisocial values and engage in deviant behavior when they lack these attachments, relative to youngsters in the same neighborhoods who report more bonding to prosocial units. Furthermore, the perception that opportunities for social mobility, employment and advancement are limited may be exacerbated by, or possibly a result of, a lack of bonding to micro and/or macro level social units. In our sample, this combination of social development variables and perception of limited opportunities were specifically predictive of antisocial behaviors among males and African Americans. Several limitations of this study preclude definitive conclusions. First is the absence of measures indicative of underlying mechanisms in development of deviant behavior and attitudes. There is ample literature to suggest that propensity to these behaviors is at least partially a function of temperament, biologic and physiologic processes, and the heritability or environmental influence of parental psychopathology in offspring outcomes (Fishbein, 1998). Nevertheless, as a study of risk factors (associated conditions as opposed to etiological factors), these findings provide compelling evidence for the influence of self-reported peer influences, attitudes toward fighting, and paternal relationships, among others, in deviant
P1: GDB Journal of Child and Family Studies [jcfs]
PH018-296862
February 2, 2001
476
12:50
Style file version Nov. 19th, 1999
Fishbein and P´erez
behavior in general. There also exists the problem of accuracy of self-report measures, particularly in high risk youth (Farrington et al., 1996; Wills & Cleary, 1997); however, perceptions of attitudes, behaviors and relationships may be just as informative as actualities in identifying risk factors (Bouhuys, Geerts, Gordijn, 1999; Wierson, Forehand, & McCombs, 1988). Perhaps the most obvious shortcoming is the small sample sizes for Whites and Hispanics relative to African Americans. In regression analyses especially, the greatest number of significant findings were for the African American sub-sample where numbers were much larger. These findings have implications for the treatment and prevention of delinquency and substance abuse among children considered “high-risk.” Programs that operate to strengthen social bonding may have potential to reduce drug abuse and delinquency (Hawkins et al., 1992). For example, improving parental relationships, particularly with the father, may have direct and positive effects on the bond between father and child and can potentially enhance some parenting traits (e.g., affection, involvement, nonconflictual relations, and child-centeredness) that may be “protective” against the development of negative behavioral outcomes (Brook, Whiteman, Balka, & Cohen, 1995; Knight, Broome, Cross, & Simpson, 1998). Another potential direction for approaches to treatment and prevention includes interventions that focus on peer influences. Negative peer influences have consistently been found to be among the strongest predictors of delinquency and drug use among youth. Programs that enhance self-efficacy, decision-making, judgment, and social skills potentially reduce the impact of these influences. Further, program components that serve to strengthen bonds to conventional norms and teach prosocial values should be particularly effective. In effect, a successful prevention strategy should be comprehensive in its focus, contain a multi-component approach, and target underlying individual and social mechanisms that contribute to negative behavioral outcomes, rather than focusing only on the symptoms of drug abuse and delinquency. ACKNOWLEDGMENTS The authors would like to express their appreciation to Tom Carr, Director of the Washington/Baltimore HIDTA for affording the opportunity to conduct this study. Also, our gratitude is extended to the Office of National Drug Control Policy, which had the foresight to provide support for prevention programs and services to high risk children in disadvantaged neighborhoods. REFERENCES Akers, R. L., Krohn, M. D., Lanza-Kaduce, L., & Radosevich, M. (1979). Social learning and deviant behavior: A specific test of a general theory. American Sociological Review, 44, 636–655. Allison, P. D. (1999). Logistic regression using the SAS system: Theory and application. Cary, NC: SAS Institute Inc.
P1: GDB Journal of Child and Family Studies [jcfs]
PH018-296862
Risk for Drug Abuse and Delinquency
February 2, 2001
12:50
Style file version Nov. 19th, 1999
477
Anisman, H., Zaharia, M. D., Meaney, M. J., & Merali, Z. (1998). Do early-life events permanently alter behavioral and hormonal responses to stressors? International Journal of Developmental Neuroscience, 16, 149–164. Bouhuys, A. L., Geerts, E., & Gordijn, M. C. (1999). Gender-specific mechanisms associated with outcome of depression: Perception of emotions, coping and interpersonal functioning. Psychiatry Research, 85, 247–261. Brook, J. S., Balka, W. B., Gursen, M. D., Brook, D. W., Shapiro, J., & Cohen, P. (1997). Young adults’ drug use: A 17-year longitudinal inquiry of antecedents. Psychological Reports, 80, 1235–1251. Brook, J. S., Whiteman, M., Gordon, A. S., & Brook, D. W. (1983). Paternal correlates of adolescent marijuana use in the context of the mother-son and parental dyads. Genetic Psychology Monograph, 108, 197–213. Brook, J. S., Whiteman, M., Balka, E. B., & Cohen, P. (1995). Parent drug use, parent personality, and parenting. Journal of Genetic Psychology, 156, 137–151. Bry, B. H., McKeon P., & Pandina, R. J. (1982). Extent of drug use as a function of number of risk factors. Journal of Abnormal Psychology, 91, 273–279. Catalano, R. F., & Hawkins, J. D. (1996). The social development model: A theory of antisocial behavior. In Delinquency and Crime: Current Theories. J. D. Hawkins (Ed.) New York: Cambridge University Press. Catalano, R. F., Kosterman, R., Hawkins, J. D., Newcomb, M. D., & Abbott, R. D. (1996). Modeling the etiology of adolescent substance use: A test of the social development model. Journal of Drug Issues, 26, 429–455. Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum Associates. Davis, N., Moss, H., Kirisci, L., & Tarter, R. (1996). Neighborhood crime rates among drug abusing and nondrug abusing families. Journal of Child and Adolescent Substance Abuse, 5, 1–14. DeWit, D. J., Offord, D. R., & Wong, M. (1997). Patterns of onset and cessation of drug use over the early part of the life course. Health Education and Behavior, 24, 746–758. Elliott, D. S., Huizinga, D., & Ageton, S. S. (1985). Explaining delinquency and drug use. Beverly Hills, CA: Sage. Elliott, D. S., Huizinga, D., & Menard, S. (1989). Multiple problem youth. Delinquency, substance use and mental health problems. New York: Springer-Verlag. Esbensen, F. A., & Deschenes, E. P. (1998). A multisite examination of youth gang membership: Does gender matter? Criminology, 36, 799–828. Farrington, D. P. (1986). Age and crime. In M. Tonry & N. Morris (Eds.), Crime and justice: An annual review of research. Chicago: University of Chicago Press. Farrington, D. P. (1995). The Twelfth Jack Tizard Memorial Lecture. The development of offending and antisocial behaviour from childhood: Key findings from the Cambridge Study in Delinquent Development. Journal of Child Psychology and Psychiatry, 36, 929–964. Farrington, D. P., Loeber, R., Southamer-Loeber, M., Van Kammen, W. B., & Schmidt, L. (1996). Selfreported delinquency and a combined delinquency seriousness scale based on boys, mothers, and teachers: Concurrent and predictive validity for African Americans and Caucasians. Criminology, 34, 493–517. Fergusson, D. M., & Horwood, L. J. (1999). Prospective childhood predictors of deviant peer affiliations in adolescence. Journal of Child Psychology Psychiatry, 40, 581–592. Fishbein, D. H. (1998). Differential susceptibility to comorbid drug abuse and violence. Journal of Drug Issues, 28, 859–890. Fishbein, D. H., & Pease, S. (1996). The dynamics of drug abuse. Needham Heights, MA: Allyn and Bacon. Frick P. J., Lahey, B. B., Loeber, R., Stouthamer-Loeber, M., Christ, M. A., & Hanson, K. (1992). Familial risk factors to oppositional defiant disorder and conduct disorder: Parental psychopathology and maternal parenting. Journal of Consulting and Clinical Psychology, 60, 49–55. Gillmore, M. R., Catalano, R. F., Morrison, D. M., Wells, E. A., Iritani, B., & Hawkins, J. D. (1990). Racial differences in acceptability and availability of drugs and early initiation of substance use. American Journal of Drug and Alcohol Abuse, 16, 185–206. Graham, N. (1997). A test of magnitude: Does the strength of predictors explain differences in drug use among adolescents? Journal of Drug Education, 27, 83–104.
P1: GDB Journal of Child and Family Studies [jcfs]
478
PH018-296862
February 2, 2001
12:50
Style file version Nov. 19th, 1999
Fishbein and P´erez
Hawkins, J. D., Catalano, R. F., & Miller, J. Y. (1992). Risk and protective factors for alcohol and other drug problems in adolescence and early adulthood: Implications for substance abuse prevention. Psychological Bulletin, 112, 64–105. Hawkins, J. D. & Weis, J. G. (1985). The social development model: An integrated approach to delinquency prevention. Journal of Primary Prevention, 6, 73–97. Huizinga, D., Esbensen, F. A., & Weiher, A. W. (1991). Are there multiple paths to delinquency? Journal of Criminal Law and Criminology, 82, 83–118. Kandel, D. B., Kessler, R. C., & Margulies, R. Z. (1978). Antecedents of adolescent initiation into stages of drug use: A developmental analysis. Journal of Youth and Adolescence, 7, 13–40. Kaplan, H. B., Martin, S. S., & Robbins, C. (1984). Pathways to adolescent drug use: Self-derogation, peer influence, weakening of social controls, and early substance use. Journal of Health and Social Behavior, 25, 270–289. Kazdin, A. E. (1990). Prevention of conduct disorder. Rockville, MD: National Institute of Mental Health. Knight, D. K., Broome, K. M., Cross, D. R., & Simpson, D. D. (1998). Antisocial tendency among drug-addicted adults: Potential long-term effects of parental absence, support, and conflict during childhood. American Journal of Drug and Alcohol Abuse, 24, 361–375. Krohn, M. D., & Massey, J. L. (1980). Social control and delinquent behavior: An examination of social learning and social bonding theories. Sociological Quarterly, 25, 353–371. Liu, X., & Kaplan, H. B. (1999). Explaining the gender difference in adolescent delinquent behavior: A longitudinal test of mediating mechanisms. Criminology, 37, 195–215. Loeber, R., & Stouthamer-Loeber, M. (1998). Development of juvenile aggression and violence. Some common misconceptions and controversies. American Psychologist, 53, 242–259. Moffitt, T. E. (1993). Adolescence-limited and life-course-persistent antisocial behavior: A developmental taxonomy. Psychological Review, 100, 674–701. Newcomb, M. D., & Bentler, P. M. (1986). Substance use and ethnicity: Differential impact of peer and adult models. Journal of Psychology, 120, 83–95. Newcomb, M. D., Maddahian, E., & Bentler, P. M. (1986). Risk factors for drug use among adolescents: Concurrent and longitudinal analyses. American Journal of Public Health, 76, 525–531. Nigg, J. T., & Hinshaw, S. P. (1998). Parent personality traits and psychopathology associated with antisocial behaviors in childhood attention-deficit hyperactivity disorder. Journal of Child Psychology and Psychiatry, 39, 145–159. Oetting, E. R., Donnermeyer, J. F., & Deffenbacher, J. L. (1998). Primary socialization theory. The influence of the community on drug use and deviance. III. Substance Use and Misuse, 33, 1629– 1665. Opland, E. A., Winters, K. C., & Stinchfield, R. D. (1995). Examining gender differences in drugabusing adolescents. Psychology of Addictive Behaviors, 9, 167–175. Paetsch, J. J., & Bertrand, L. D. (1997). The relationship between peer, social, and school factors, and delinquency among youth. Journal of School Health, 67, 27–32. Paschall, M. J., Flewelling, R. L., & Ennett, S. T. (1998). Racial differences in violent behavior among young adults: Moderating and confounding effects. Journal of Research in Crime and Delinquency, 35, 148–165. Rhodes, J. E., & Fischer, K. (1993). Spanning the gender gap: Gender differences in delinquency among inner-city adolescents. Adolescence, 28, 879–889. Rienzi, B. M., McMillin, J. D., Dickson, C. L., Crauthers, D., McNeill, K. F., Pesina, M. D., & Mann, E. (1996). Gender differences regarding peer influence and attitude toward substance abuse. Journal of Drug Education, 26, 339–347. Sher, K. J., Walitzer, K. S., Wood, P. K., & Brent, E. E. (1991). Characteristics of children of alcoholics: Putative risk factors, substance use and abuse, and psychopathology. Abnormal Psychology, 100, 427–448. Sherman, L. W., Gartin, P., & Buerger, M. (1989). Hot spots of predatory crime. Criminology, 27, 27–56. Sherman, L. W., & Weisburd, D. (1995). General deterrent effects of police patrol in crime “hotspots”: A randomized, controlled trial. Justice Quarterly, 12, 625–648. Steffensmeier, D., & Allan, E. (1996). Gender and crime: Toward a gendered theory of female offending. Annual Review of Sociology, 22, 459–487.
P1: GDB Journal of Child and Family Studies [jcfs]
PH018-296862
Risk for Drug Abuse and Delinquency
February 2, 2001
12:50
Style file version Nov. 19th, 1999
479
Substance Abuse and Mental Health Survey Administration. (1998). Prevalence of substance use among racial and ethnic subgroups in the United States 1991–1993 (DHHS Publication No. SMA 98-3202). Rockville, MD: National Clearinghouse for Alcohol and Drug Information. Thomas, B. S. (1996). A path analysis of gender differences in adolescent onset of alcohol, tobacco and other drug use (ATOD), reported ATOD use and adverse consequences of ATOD use. Journal of Addictive Diseases, 15, 33–52. Toby, J. (1974). The socialization and control of deviant motivation. In D. Glaser (Ed.), Handbook of Criminology (pp. 85–100). Chicago: Rand McNally. Vaccaro, D., & Wills, T. A. (1998). Stress-coping factors in adolescent substance use: Test of ethnic and gender differences in samples of urban adolescents. Journal of Drug Education, 28, 257–282. Vega, W. A., Zimmerman, R. S., Warheit, G. J., Apospori, E., & Gil, A. G. (1993). Risk factors for early adolescent drug use in four ethnic and racial groups. American Journal of Public Health, 83, 185–189. Wells, E. A., Morrison, D. M., Gillmore, M. R., Catalano, R. F., Iritani, B., & Hawkins, J. D. (1992). Race differences in antisocial behaviors and attitudes and early initiation of substance use. Journal of Drug Education, 22, 115–130. Wierson, M., Forehand, R., & McCombs, A. (1988). The relationship of early adolescent functioning to parent-reported and adolescent-perceived interparental conflict. Journal of Abnormal Child Psychology, 16, 707–718. Wills, T. A., & Cleary, S. D. (1997). The validity of self-reports of smoking: Analyses by race/ethnicity in a school sample of urban adolescents. American Journal of Public Health, 87, 56–61. Witt, S. D. (1997). Parental influence on children’s socialization to gender roles. Adolescence, 32, 253–259.