Cannabis Aversive Learning Paper Dataset
Description
Paper Abstract: Objective: Learning theories suggest an aversive learning process whereby negative cannabis consequences are associated with reduced cannabis use. Contrary to this expectation, extent research has found a positive prospective association which may reflect continued patterns of heavy use. This study aims to clarify the predictive association of negative cannabis consequences and future cannabis use by considering that both a positive and negative association with future use are simultaneously present and can be isolated with mediation analysis. We tested the hypothesis that negative cannabis expectancies would mediate a negative indirect effect between negative cannabis consequences and future use. We also hypothesized that a positive direct association between consequences and use would remain. Finally, we considered sensitivity to punishment as a moderator of this mediational path. Method: Data from a longitudinal community sample (N=387) assessed annually for three years in young adulthood (ages = 19-21) were analyzed using multilevel models. Monte Carlo simulation was used to test mediation. Results: Negative cannabis consequences were positively directly associated with cannabis use and indirectly negatively associated with cannabis use through negative expectancies. There was no evidence of moderation by sensitivity to punishment. Conclusions: The association between negative cannabis consequences and future use is complex. Contrary to framing the prospective association between consequences and use as either positive or negative, the current study offers evidence of both. Consistent with behaviorist learning theories, negative experiences with cannabis appear to motivate reductions in use through learned associations between use and negative outcomes.
Files
Steps to reproduce
Data Analysis Strategy: Multilevel modeling (MLM) with repeated measures (level 1) nested within participants (level 2) was used to test hypotheses. Analyses were confined to participants who reported any cannabis use during at least one of the three waves (W7-9, N = 214). For the first model, data were organized to examine the association between consequences and negative expectancies within each wave (i.e., W8 consequences -> W8 negative expectancies, W9 consequences ->W9 negative expectancies). This is because the MACQ assesses negative consequences in the past year and negative expectancies are assessed currently. Thus, this model captures the influence of past-year negative consequences on current negative expectancies. We controlled for prior negative expectancies, cannabis use, and parental cannabis use at level 1, SP and SR at level 2, and demographic variables (age and education status at level 1, sex at level 2). For the second model, data were organized to test the prospective association between negative expectancies and frequency of cannabis use (i.e., W7 expectancies -> W8 use, W8 expectancies ->W9 use), controlling for prior use, consequences, parental use, and positive expectancies at level 1, SP and SR at level 2, and demographic variables (age and education status at level 1, sex at level 2). Both models included a random intercept and were run using the PROC MIXED procedure in SAS 9.4 (SAS Institute, Inc., Cary, NC, Copyright © 2016) with Maximum Likelihood estimation. Interaction terms were introduced into each model. SP was crossed with negative consequences to predict negative expectancies and with negative expectancies to predict cannabis use. First-order effects were centered at the sample level to aid in the interpretation of interaction effects and to reduce issues related to non-essential multicollinearity (Aiken et al., 1991; Hox, 2002). The negative consequences by SP interaction was included as a covariate in the use model. Monte Carlo simulation (Preacher and Selig, 2012) was used to test mediation. Selig and Preacher’s (2008) interaction tool was used to generate the 95% CIs for indirect effects.
Institutions
- University at Buffalo - North Campus