BRIEF REPORT PREDICTING DRUG USE: APPLICATION OF BEHAVIORAL THEORIES OF CHOICE

Transcription

1 Pergamon Addictive Behaviors, Vol. 23, No. 5, pp , 1998 Copyright 1998 Elsevier Science Ltd Printed in the USA. All rights reserved /98 $ PII S (98) BRIEF REPORT PREDICTING DRUG USE: APPLICATION OF BEHAVIORAL THEORIES OF CHOICE CHRISTOPHER J. CORREIA, JEFFREY SIMONS, KATE B. CAREY, and BRIAN E. BORSARI Syracuse University Abstract The current study sought to test the utility of Herrnstein s (1970) matching law in predicting drug use occurring in the natural environment. Participants were 206 college students. Behavioral allocation was measured across two concurrently available sets of activities: those engaged in while using or under the influence of drugs and/or alcohol (drug related) and those engaged in when drug free. Results from regression analyses indicate that predictions of drug use are improved with the addition of reinforcement received from drug-free activities, which enters the model with a negative coefficient value. The addition of a reinforcement ratio, based on matching law equations, also accounted for unique variance. Results demonstrate the utility of applying behavioral theories of choice to drug use and highlight the importance of viewing behaviors within their broader environmental context Elsevier Science Ltd Behavioral theories of choice recognize that preferences for drugs or alcohol arise within a broader context involving the availability or utilization of other competing reinforcers and associated environmental constraints (Vuchinich & Tucker, 1996). Vuchinich and Tucker (1988) reviewed literature on the connection between alcohol use and the presence of alternative reinforcers. These authors concluded that alcohol consumption is increased when access to alternative reinforcers is constrained and that consumption decreases when access to alcohol itself is constrained. Because direct constraints on alcohol are relatively rare in most natural environments, constraints on nondrug alternative reinforcers may be the more salient determinant of actual consumption. In a more recent review, Carroll (1996) concluded that the availability of nondrug reinforcers can slow or prevent acquisition of drug self-administration and that the presence of nondrug alternatives may suppress withdrawal under some conditions. Herrnstein s (1970) matching law is a mathematical account of choice behavior. The theory and accompanying equations specify that an individual s behavior is distributed across concurrently available options in proportion to the amount of reinforcement received for engaging in each behavior. Within this framework, the relative amount of reinforcement received for a behavior is viewed as more predictive of choice behavior than the absolute amount of reinforcement received (McDowell, 1988). Reviews of the literature suggest that the matching law can adequately describe human choice behavior in both controlled and natural environments (cf. McDowell, 1988; Pierce & Epling, 1983). Christopher J. Correia, Jeffrey Simons, Kate B. Carey, and Brian E. Borsari are at the Department of Psychology, Syracuse University. This work was supported in part by National Institute on Drug Abuse Grant DA Requests for reprints should be sent to Kate B. Carey, Department of Psychology, 430 Huntington Hall, Syracuse University, Syracuse, NY ; kbcarey@psych.syr.edu 705

2 706 C. J. CORREIA et al. The current study sought to test the utility of contextual variables, similar to those specified by the matching law, in predicting naturally occurring substance use. Participants were asked to report the frequency and enjoyability of two concurrently available sets of activities: those engaged in while under the influence of alcohol or other drugs (drug related) and those engaged in when drug free. Retrospective reports of drug use were collected for a parallel time frame. Regression analyses were conducted to test the relative contributions of various measures of behavioral allocation in the prediction of substance use. We hypothesized that the amount of reinforcement received from drug-free activities would add unique variance over and above the variance accounted for by the amount of reinforcement received from drug-related activities. Further, we hypothesized that a proportional variable accounting for reinforcement from drug-related activities relative to total reinforcement would account for additional unique variance beyond that accounted for by reinforcement from drug-related and drug-free activities. METHOD Participants Participants were undergraduates at a large private university who volunteered to participate in this research as partial fulfillment of course requirements. Analysis for the current study concerned only 206 participants who reported alcohol and/or other drug use in the last 30 days. The mean age of the sample was years (SD 1.78); 58% were female, and 11% were minorities. Measures Self-report measures were administered as part of a larger questionnaire packet. To ensure confidentiality, the packets contained no identifying information. Participants were asked to report use in the last 30 days for 10 specific drugs (e.g., How many days in the past 30 days have you used marijuana? ) as well as substance use in general (e.g., How many days in the past 30 days have you used alcohol or any other drugs? ). The Pleasant Events Schedule (PES; MacPhillamy & Lewinsohn, 1982) is a 320- item instrument designed to measure the frequency and subjective pleasure of potentially reinforcing events or activities over a previous 30-day period. Each item yields a frequency score and an enjoyability score (subjective pleasure rating), and the crossproduct of the two ratings is an index of reinforcement potential. Frequency and enjoyability scores range from 0 to 2 for each item, producing a cross-product ranging from 0 to 4. A higher cross-produce score indicates that the activity was engaged in with a high amount of reinforcement potential, which is considered to be a useful approximation of obtained positive reinforcement. Averaging across items produces a summary index for each of the three scores. In previous studies, the PES has demonstrated adequate internal consistency and test-rest reliability; peer ratings, expert ratings, and subsequent choice behavior have provided evidence of validity (MacPhillamy & Lewinsohn, 1982). Three modifications were made to the PES for the current study. First, participants were asked to provide two frequency and enjoyability ratings for each activity; one set of ratings assessed times when the participants were drug free ( time when you were not using or under the influence of any psychoactive drug including alcohol ), and the second set assessed times when participants were using or under the influence of

3 Predicting drug use 707 drugs or alcohol. In the interest of reducing participant burden, enjoyability ratings were obtained only for those events or activities in which participants actually engaged in during the previous 30 days. The second modification eliminates the original enjoyability score while still allowing the calculation of the frequency score and the crossproduct. Third, to keep the number of drug-free and drug-related activities equivalent, the 15 original PES items that explicitly mention drug use (e.g., drinking beer) were not used in the calculations of any scores or considered in any of the analyses presented. Thus, the drug-related reinforcement refers to reinforcement from activities enhanced by intoxication and not the subjective experience of intoxication itself. This final modification will also reduce the possibility that correlations between drug-related scores and actual drug use would be artificially inflated. In addition to the frequency and cross-product scores, the current study made use of a reinforcement ratio score. The reinforcement ratio, based on Herrnstein s (1970) matching law, was calculated for each subject by dividing the average drug-related cross-product score by the sum of the average drug-related cross-product score and the average drug-free cross-product score. The ratio ranges from 0 to 1, with a higher score indicating a greater proportion of reinforcement received from drug-related activities relative to total reinforcement. Thus, our modification of the PES allows for distinctions between drug-related and drug-free experiences and results in the following summary statistics: drug-related activity (frequency), drug-free activity (frequency), drug-related cross-product (reinforcement), drug-free cross-product (reinforcement), and reinforcement ratio. In all cases, the term drugs is used to refer to alcohol as well as other illicit substances. Procedure Participants attended group sessions to complete a set of questionnaires addressing alcohol use and other topics. After providing written informed consent and demographic data, participants completed the previously described measures. RESULTS Gender comparisons on key measures are summarized in Table 1. Significant gender differences were found on the number of drug use days in the previous 30 days, t(204) 2.23, p.05; drug-related activity score, t(204) 2.53, p.05; drug-free Table 1. Mean and standard deviations of key variables by gender Men (n 87) Women (n 119) M SD M SD Drug use days a activity activity a cross-product a cross-product Reinforcement ratio b Notes. Drug(s) refers to alcohol and other illicit substances. Scores on the drug-free and drug-related activity scores range from 0 2; scores on the drug-related and drugfree cross-product range from 0 4. a Significant gender differences indicated by p.05. b Significant gender differences indicated by p.001.

4 708 C. J. CORREIA et al. Table 2. Correlation matrix for drug use days and reinforcement variables Drug use days activity activity cross-product cross-product Reinforcement ratio Drug use days activity.02 activity.51**.46** cross-product.03.91**.32** cross-product.55**.41**.94**.37** Reinforcement ratio.59**.08.79**.16*.79** *p.05. **p.001. cross-product, t(204) 2.5, p.05; and the reinforcement ratio, t(204) 3.4, p.001, with males scoring higher on all of the variables except the drug-free cross-product. Correlational data, presented in Table 2, indicate that drug-free and drug-related cross-product scores are moderately correlated, r(217).37, and both scores are related to the reinforcement ratio in predicted directions, r(216).16, and r(216).79, respectively. Drug use days in the last 30 days were moderately correlated with both the drug-related cross-product, r(206).55, and the reinforcement ratio, r(206).59, but were not significantly correlated with the drug-free cross-product. A series of multiple regressions was performed to determine the relative contributions of drug-related reinforcement, drug-free reinforcement, and a ratio of drug-related reinforcement to total reinforcement in the prediction of drug use days. Results are presented in Table 3. Drug use days in the previous 30 served as the criterion variable. In Step 1, gender and the amount of reinforcement received for drug-related activities were entered as predictors, accounting for 32% of the variance, F(2, 203) 46.94, p For Step 2, reinforcement from drug-free activities was added, and it accounted for a 4% increment, F(1, 202) 14.87, p.001. In this model, the reinforcement re- Table 3. Regression of drug use days on PES variables Predictor B Standard error B R 2 R 2 Model F Step 1.32** 46.94** Gender cross-product ** Step 2.04**.36** 38.39** Gender cross-product ** cross-product ** Step 3.02*.38** 30.33** Gender cross-product * cross-product Reinforcement ratio * Notes. N 206. PES Pleasant Events Schedule. *p.05. **p.001.

5 Predicting drug use 709 ceived from drug-free activities entered into the model with a negative coefficient value; thus, after controlling for the influence that drug-related reinforcement has on substance use, as reinforcement from drug-free activities decreases, the number of substance use days increases. In Step 3, the reinforcement ratio was entered, and it accounted for a 2% increment in variance, F(1, 201) 4.29, p.05. In the full model, the amount of reinforcement received from drug-free activities was no longer a significant predictor. Gender did not contribute to any of the tested models. DISCUSSION The results of this study demonstrate that contextual variables, derived from behavioral theories of choice, have predictive validity with regard to drug use behavior. Reinforcement from drug-free activities contributed to the prediction of drug use, even after controlling for the contribution of drug-related reinforcement. Specifically, a negative relationship was observed, indicating that, as reinforcement from drug-free activities decreases, the frequency of drug use increases. These findings are consistent with previous experimental research and further highlight the connection between drug-free reinforcement and drug-related behaviors. The addition of the reinforcement ratio and its predictive ability further demonstrates the utility of expanding the study of drug use to include nondrug contextual variables. Although the amount of reinforcement received directly from drug use certainly remains an important variable, its value as a predictor of drug use is strengthened when viewed in the context of competing drug-free reinforcers. Thus, the reinforcement ratio highlights the relative importance drugs play in the lives of our participants. In addressing the relevance of matching theory in the natural human environment, McDowell (1988) described choice as selecting a particular alterative over all other alternatives considered as an aggregate (p. 96). Given McDowell s conceptualization of choice, our study adopted a methodology capable of assessing choices regarding a specific behavior in this case drug use relative to a measurable aggregate of alternative sources of reinforcement. Thus, this study provides further evidence regarding the importance of understanding the context in which drug use occurs. The current study also demonstrates that behavioral principles derived in laboratory settings can stimulate theory-driven research and lead to methodological advances for studies conducted in the natural environment. REFERENCES Carroll, M. E. (1996). Reducing drug abuse by enriching the environment with alternative non-drug reinforcers. In L. Green & J. Kagel (Eds.), Advances in behavioral economics (Vol. 3, pp ). Norwood, NJ: Ablex. Herrnstein, R. J. (1970). On the law of effect. Journal of the Experimental Analysis of Behavior, 13, MacPhillamy, D. J., & Lewinsohn, P. M. (1982). The pleasant events schedule: Studies on reliability, validity, and interscale correlations. Journal of Consulting and Clinical Psychology, 50, McDowell, J. J. (1988). Matching theory in natural human environments. The Behavior Analyst, 11, Pierce, W., & Epling W. (1983). Choice, matching, and human behavior: A review of the literature. The Behavior Analyst, 6, Vuchinich, R. E., & Tucker, J. A. (1988). Contributions from behavioral theories of choice to an analysis of alcohol abuse. Journal of Abnormal Psychology, 97, Vuchinich, R. E., & Tucker, J. A. (1996). The molar context of alcohol abuse. In L. Green & J. Kagel (Eds.), Advances in behavioral economics (Vol. 3, pp ). Norwood, NJ: Ablex.