Abstract
The present study evaluated the mediating role of implicit cognitive processes in the prediction of alcohol and marijuana use and examined the relationships between dissociative experiences, implicit processes, and sensation seeking in models of drug use and problem experiences. Participants were 467 diverse at-risk adolescents in California. Results from latent variable models revealed that implicit cognition independently predicted alcohol and marijuana use and mediated the predictive effects of sensation seeking on drug use. Dissociative experiences did not predict implicit cognition or drug use in this sample, though this factor was a significant predictor of problem experiences and was positively correlated with sensation seeking. This research provides further evidence suggesting that implicit, associative memory processes are influential in drug-use motivation.
Footnotes
For comparison purposes with the alcohol-use models, the marijuana use measurement model was respecified with the cue-behavior association,outcome-behavior association, and phrase completion tasks as indicators of implicit cognition. The factor loadings and intercorrelations in this measurement model were almost identical to the marijuana measurement model using the cue and outcome-behavior association tasks and the DAPT-S picture cues as indicators of the Implicit Cognition factor.
Additionally, supplementary models for both alcohol use and marijuana use were estimated using measurement models with only previously used implicit cognition indicators, the cue and outcome-behavior association tasks. The factor loadings and intercorrelations of these measurement models were similar to all other measurement models evaluated in this research. The final supplementary structural models using only the cue and outcome-behavior association tasks were also identical in pattern of significance of paths compared with the final structural models noted in the text. The direct predictive effect of implicit cognition on alcohol use and on marijuana use was significant in all structural models evaluated. This degree of replication despite divergent indicators attests to the generality of the primary results across assessment assumptions.
An alternative supplemental structural model for marijuana use was estimated. In this model, the Implicit Cognition factor consisted of the same indicators specified for this factor in the final alternative structural model for alcohol use (i.e., cue and outcome behavior association, and the phrase completion association tasks). This model was identical in pattern of significance of paths compared to the final supplemental model for marijuana use noted earlier. The fit of this model was very good but did not reach statistical nonsignificance, χ2 (85, N = 227) = 124.777, p < .01, NNFI = .952, CFI = .961, RMSEA = .046, CI = .027 to. 062.