تفکر سازنده، رفتار ضد اجتماعی، و استفاده از مواد مخدر در نوجوانان پسر با و بدون سابقه خانوادگی یک اختلال مصرف مواد

Abstract The purpose of this study was to determine the manner in which constructive thinking (CT) and antisocial behavior (ASB) are related to drug use in 295, 15–17 year old, adolescent males with a positive (FHP; n=126) and a negative (FHN; n=169) family history of a substance use disorder (SUD). CT is considered to be a “nonintellective” cognitive skill that reflects the ability to effectively deal with problem situations through the use of different thinking styles and behavioral, social, and emotional coping strategies. CT and ASB were measured using self-report inventories in a laboratory setting. Three types of drug use were assessed: (a) number of drugs used, (b) drug use frequency, and (c) drug use problems. Results showed that the FHP group had significantly lower CT and significantly higher ASB and drug use scores compared with the FHN group. CT was significantly related to each type of drug use and these relations were all mediated by ASB. Moreover, the relation between CT and drug use frequency was moderated by ASB and the relations between ASB and each form of drug use were moderated by family history. Although few in number, this and other studies show that deficits in nonintellective forms of cognition such as CT are important liability factors for SUD that require further investigation.

Introduction Males with a positive family history (FHP) of a substance use disorder (SUD) are at high risk for the development of an SUD compared with those with a negative family history (FHN) of the disorder (for reviews see Pihl et al., 1990 and Sher, 1991). It is well known that FHP males differ from FHN controls on a number of variables considered to be risk factors for SUD (Tarter et al., 1999). For example, there is a large literature describing mild cognitive deficits in FHP individuals. Compared with their FHN counterparts, FHP males exhibit poorer performance on tests of visuospatial abilities, perceptual-motor skills, language processing, reading comprehension, vocabulary, educational achievement, and executive functioning which includes planning, problem solving, attentional control, categorization, organization, abstracting ability, and working memory (for reviews see Giancola and Moss, 1998, Pihl et al., 1990 and Tarter et al., 1999). Given their nature, the cognitive abilities listed above fall under the rubric of “intellective” processes. However, in addition to intellective skills, there also exist “nonintellective” forms of cognition which refer to particular biases or styles of thinking, interpreting information, and coping. Epstein and Meier (1989) recently identified a nonintellective cognitive ability which they termed constructive thinking (CT). CT is defined as “a person's ability to think in a manner that solves everyday problems in living at a minimal cost in stress” (Katz & Epstein, 1991, p. 789). More specifically, CT is a form of experiential intelligence (“common sense” information that is acquired through experience) that reflects the ability to effectively deal with problem situations through the use of different thinking styles and behavioral and emotional coping strategies (Epstein & Meier, 1989). FHP children have been well-studied with regard to their intellective cognitive abilities, however, the same cannot be said for their nonintellective abilities. Investigating these processes and the manner in which they are related to drug use in FHP children is important because persons with distorted thinking styles or low CT are probably more likely to use drugs compared with their more adaptive-thinking counterparts. Specifically, adolescents with poor nonintellective cognitive abilities may be at high risk for drug use and SUD because of their inability to effectively deal with stress and their reduced capacity to invoke the necessary behavioral and emotional skills needed to cope with, and solve, everyday problems. Of the few studies investigating the relation between nonintellective cognitive functions and drug use, one found that FHP boys exhibit a distorted cognitive style insofar as they explain negative situational events by making maladaptive internal, stable, and global cognitive misattributions (Perez-Bouchard, Johnson, & Ahrens, 1993). Another study demonstrated that cognitive distortions in the form of overgeneralization, selective abstraction, personalization, and catastrophizing, are related to increased drug use in adolescent boys with and without a family history of SUD (Giancola, Mezzich, Clark, & Tarter, 1999). These results are supported by clinical data indicating that increased cognitive distortions are related to a diagnosis of SUD in dually-diagnosed male and female adolescents (Kempton, Van Hasselt, Bukstein, & Null, 1994). Although the relation between CT and drug use also remains largely unexplored, there is some evidence to suggest that the two constructs are linked. For example, compared with FHN children, FHP children are more likely to respond to their problems with avoidant, emotionally reactive nonconstructive coping strategies (Clair & Genest, 1987). More specifically, CT has been found to correlate negatively with drug use in college undergraduates (Epstein & Meier, 1989) and pregnant women (Park, Moore, Turner, & Aldler, 1997). Other research has demonstrated that adolescent females with an SUD have lower CT scores compared with non-SUD controls and that CT is negatively related to drug use in these girls (Giancola, Shoal, & Mezzich, 2001). Together, these findings suggest that low CT (e.g. weak behavioral and emotional coping skills) may be a risk factor for drug use problems. Nevertheless, further study is clearly needed in this area inasmuch as FHP and FHN adolescents have not yet been compared with regard to CT nor has the relation between CT and drug use been investigated in adolescents at high-risk for SUD. In addition to CT, another well established risk factor for SUD is antisocial behavior (ASB; Gillen and Hesselbrock, 1992 and Sher and Trull, 1994). Compared with controls, FHP males are more likely to meet criteria for psychiatric diagnoses characterized by ASB such as oppositional defiant disorder, attention deficit hyperactivity disorder, conduct disorder, and antisocial personality disorder (Clark et al., 1997, Hesselbrock and Hesselbrock, 1992, Pihl et al., 1990 and Sher, 1991). FHP individuals also report relatively higher levels of aggression, delinquency, and affiliations with delinquent peers (Blackson and Tarter, 1994, Finn et al., 1992 and Giancola et al., 1996). In fact, a number of recent studies have shown that ASB is one of the strongest, and most reliable, correlates/predictors of drug and alcohol use (Brook et al., 1996, Chassin et al., 1993 and O'Donnell et al., 1995), an early age of onset of drug use (Hesselbrock et al., 1985 and Windle, 1993), problem drug use (Pulkkinen and Pitkanen, 1994 and Stice et al., 1998), a diagnosis of SUD (Biederman et al., 1997 and Halikas et al., 1990), and SUD relapse (Myers, Brown, & Mott, 1995). It can be argued that ASB has a more proximal or direct influence on drug use than CT. In fact, one can logically make the case that low CT will lead to increased ASB. Specifically, persons with an inability to effectively deal with stress as well as those who have difficulty generating and implementing adaptive coping strategies might be more likely to resort to ASB, particularly when faced with precarious or provoking situations. Given this argument and the fact that ASB is a strong and proximal predictor of drug use, it can be hypothesized that the relations between CT and drug use will be mediated by ASB. This hypothesis is supported by a recent study demonstrating this exact finding in adolescent girls with an SUD (Giancola et al., 2001). Nevertheless, this hypothesis is yet to be tested in FHP and FHN adolescents. Based on the this literature review, the aims of this study were to determine whether: (1) FHP adolescent boys have lower CT scores than FHN controls; (2) CT is related to drug use and whether this relation is mediated by ASB; (3) CT and ASB interact to account for unique variance in drug use; and (4) a family history of SUD moderates the relations between CT and ASB with drug use. With regard to the last two aims, the examination of these interactive effects is important because it will help identify subgroups of boys who are especially at risk for the development of SUD.

. Results 3.1. Group differences data Group differences were calculated using bi-directional t-tests (a chi-square test was used to assess differences in ethnicity). The FHP group had a significantly lower SES compared with the FHN group, however, the groups did not differ with regard to age, years of education, and ethnicity. The first aim of this study was to determine whether the FHP group had lower CT scores than the FHN group. As can be seen in Table 1, this hypothesis was confirmed. The FHP group also had significantly greater ASB and drug use scores. Table 1. Group differences Measure FHP FHN M S.D. M S.D. Age 15.5 0.65 15.4 0.57 Years of education 9.1 0.87 9.2 0.69 Socioeconomic status 37.5 13.00 45.69 14.00* Constructive thinking 100.47 14.42 105.87 13.14* Antisocial behavior 6.23 4.57 4.85 3.27* Number of drugs used 3.51 3.41 2.47 2.94* Drug use frequency 2.76 3.46 1.82 3.45* Drug use problems 7.35 15.41 4.26 10.92* Ethnic distribution n % n % Caucasian 86 68 140 83 African American 36 29 24 14 Other 4 3 5 3 ∗ P<0.05. Table options 3.2. Correlations In order to provide a more complete description of the sample, a correlation matrix was computed for all of the variables. Age was the only demographic variable that was significantly related to the drug use measures. None of the demographic variables were significantly related to CT or ASB. As expected, CT and ASB were significantly related to one another and with all of the drug use variables in the expected directions. These data are displayed in Table 2. Table 2. Pearson product–moment correlations Measure 1 2 3 4 5 6 7 8 (1) Age 0.52* 0.14* −0.05 −0.03 0.18* 0.18* 0.12* (2) Years of education 0.13* 0.09 −0.16* −0.01 −0.03 0.01 (3) Socioeconomic status 0.09 −0.08 0.00 −0.04 −0.10 (4) Constructive thinking −0.33* −0.19* −0.24* −0.23* (5) Antisocial 0.34* 0.40* 0.36* (6) Number of drugs used 0.66* 0.56* (7) Drug use frequency 0.59* (8) Drug use problems ∗ P<0.05. Table options 3.3. Mediation analyses The second aim of this study was to determine whether CT is related to drug use and whether this relation is mediated by ASB. In order for a variable to be considered a mediator, four conditions must be satisfied (Baron and Kenny, 1986 and Holmbeck, 1997). First, the independent variable (e.g., CT) must be significantly correlated with the dependent variable (e.g. drug use). Second, the independent variable must be significantly correlated with the proposed mediator (e.g. ASB). Third, the proposed mediator must be significantly correlated with the dependent variable (e.g. drug use). Fourth, the relation between the independent variable and the dependent variable must diminish considerably with the introduction of the mediator into the equation. As can be seen in Table 2, the first three conditions were all satisfied thus allowing us to proceed with the test for mediation. Mediation was tested using three, two-step, hierarchical multiple regression equations (one for each dependent variable). Inasmuch as age was found to correlate significantly with the drug use variables, it was included as a covariate. Age and CT were entered in the first step. Even when controlling for age; CT was still significantly negatively related to number of drugs used, frequency of drug use, and drug use problems (see Table 3). The ASB variable was then added in the next step (for each model) to test its potential role as a mediator. As can be seen in Table 3, whereas the relations between ASB and drug use are statistically significant (for each of the equations), the relations between CT and drug use are no longer significant, thus suggesting a mediation effect for each equation. In fact, the addition of ASB accounted for 62%, 53%, and 50% of the variance for the relation between CT and number of drugs used, frequency of drug use, and drug use problems, respectively. Table 3. Mediation analyses Step and measure β P-value Number of drugs used Step 1: Age 0.17 0.003 Constructive Thinking −0.17 0.002 Step 2: Age 0.18 0.001 Constructive Thinking −0.07 ns Antisocial Behavior 0.33 0.000 F(3, 294)=17.93, P=0.000; R2=0.16 Drug use frequency Step 1: Age 0.17 0.003 Constructive Thinking −0.24 0.000 Step 2: Age 0.19 0.000 Constructive Thinking −0.11 ns Antisocial Behavior 0.37 0.000 F(3, 294)=25.59, P=0.000; R2=0.21 Drug use problems Step 1: Age 0.11 ns Constructive Thinking −0.22 0.000 Step 2: Age 0.12 0.02 Constructive Thinking −0.10 ns Antisocial Behavior 0.33 0.000 F(3, 294)=18.17, P=0.000; R2=0.16 Table options 3.4. Moderation analyses The third and fourth aims of this study were to determine whether CT and ASB would interact to account for unique variance in drug use and to determine whether a family history of SUD would moderate the relations between CT and ASB with drug use. According to the recommendations put forth by Aiken and West (1991), it was decided that the most parsimonious manner in which to test these aims would be to conduct three “step-down” moderated regression equations, one for each drug use variable. All variables were first converted into z-scores. Interaction terms were then calculated by obtaining the cross-product of pertinent first-order variables. According to Friedrich (1982), it is important to create interaction terms using z-scores rather than raw scores inasmuch as standardizing cross-products after they have already been created does not yield the same regression coefficients as multiplying standardized values. When using this procedure, it is important to interpret the unstandardized, and not the standardized, regression solution. Traditional standardized solutions should not be interpreted because they are not scale invariant for multiplicative terms and will thus yield incorrect regression coefficients for these effects ( Friedrich, 1982). Friedrich's procedure takes into account the problem of scale invariance and yields an approximately “standardized” solution. In addition, another benefit of standardizing variables is the resultant centering (deviation scores with a mean of zero) of values which reduces multicollinearity between interaction terms and their constituent lower-order terms. All possible 2- and 3-way interaction terms were then created for the following variables: CT, ASB, and family history (coded: FHN=−1 and FHP=1). Each linear (i.e. main) effect and all of their interactions were entered into each regression equation simultaneously. Because age was found to correlate significantly with the drug use variables, it was included as a covariate. This resulted in a full model comprising eight variables. Using the step-down procedure, if a significant 3-way interaction is detected, it is interpreted. However, if the 3-way effect is not significant, it is removed from the equation and the model is re-estimated. This procedure is repeated if no 2-way interactions are significant either. This step-down method was chosen in order to increase the efficiency of the regression coefficient estimates (Aiken & West, 1991). It has been argued that removing nonsignificant higher-order terms and re-estimating models is appropriate in analyses where there are no specific hypotheses that a particular higher-order interaction will necessarily be significant (Aiken & West, 1991). This is the case in the present study. Significant interaction terms were interpreted by plotting the effect and determining whether the slopes of the simple regression lines (one SD above and below the overall mean) were significantly different from zero. The 3-way interaction term was not significant for any of the equations. The equations were thus re-estimated without this effect. Each trimmed model contained the same significant 2-way interaction: FH×ASB. ASB was significantly positively related to drug use, for each drug use variable, for both groups, however, in each case, the relations were stronger for the FHN group. Given that the plots of the simple slopes were highly similar for each drug use variable, Fig. 1 only depicts the slopes for the drug use problems variable, (FHP: β=0.30, t=4.20, P<0.05 and FHN: β=0.42, t=4.84, P<0.05). Slopes for the number of drugs used variable and for the drug use frequency variable are, respectively, as follows: (FHP: β=0.24, t=3.38, P<0.05 and FHN: β=0.44, t=5.10, P<0.05) and (FHP: β=0.30, t=4.24, P<0.05 and FHN: β=0.52, t=6.18 P<0.05). In addition to these interaction effects, the CT×ASB term was also significant for the drug use frequency variable. CT was significantly negatively related to drug use frequency for the high ASB group (one SD above the overall mean; β=−0.25, t=−2.79, P<0.05) but not for the low ASB group (one SD below the overall mean; β=−0.04, t=−0.52, P=n.s.). Slopes for this effect are presented in Fig. 2. The results also indicated that ASB was positively related to all drug use variables and that CT was negatively related to drug use frequency and drug use problems. Data for all the regression equations are presented in Table 4. Simple regression slopes for the relation between antisocial behavior and drug ... Fig. 1. Simple regression slopes for the relation between antisocial behavior and drug use problems for the FHN and FHP groups. Figure options Simple regression slopes for the relation between constructive thinking and drug ... Fig. 2. Simple regression slopes for the relation between constructive thinking and drug use frequency for boys with “low” and “high” levels of antisocial behavior. Figure options Table 4. “Standardized” Regression Coefficients [calculated using Friedrich's (1982) procedure] Variable Number of drugs used Frequency of drug use Drug use problems Age 0.18* 0.19* 0.12* FH 0.09 0.02 0.02 CT −0.07 −0.14* −0.14* ASB 0.32* 0.35* 0.31* FH×CT −0.01 0.06 −0.09 FH×ASB −0.11* −0.14* −0.11* CT×ASB −0.03 −0.17* −0.08 FH×CT×ASB – – – FH=Family History; CT=Constructive Thinking; ASB=Antisocial Behavior. All first-order variables were transformed to z-scores. Cross-products were calculated after z-transforming their first-order constituents. Absent values indicate that appropriate restricted models are presented. ∗ P<0.05.