Investigating the Joint Development of Approach Bias and Adolescent Alcohol Use.

ABSTRACT

BACKGROUND: We investigated the joint development between implicit approach bias and early adolescent alcohol use, and examined whether the link between approach bias and alcohol use was moderated by working memory (WM). METHODS: The current study used data from a 2-year, 4-wave online sample of 378 Dutch early adolescents (mean age 14.9 years, 64.8% female). First, using latent growth curve modeling, we examined trajectories of approach bias and alcohol use over time. Second, we examined relations between baseline approach bias and WM and the development of alcohol use. Third, we examined the joint development of approach bias and alcohol use. Fourth, we examined whether the nature of this joint development varied for different levels of WM. RESULTS: Unconditional growth curve model analyses indicated that the functional forms of alcohol use and cognitive bias were best captured by quadratic and linear trajectories, respectively. We found that cognitive bias decreased over time. We found no significant relations between baseline predictors and observed increases in alcohol use. We found relations between the intercepts, but not to growth factors, in the joint development of alcohol use and approach bias. WM was not found to moderate relations between growth in approach bias and alcohol use in this sample. CONCLUSIONS: While we observed evidence of association between approach bias and alcohol use at baseline, there was no evidence of relations between development trajectories of the two. These findings replicate prior research demonstrating a role of implicit approach bias in predicting early adolescent alcohol use but do not demonstrate, in a light drinking early adolescent sample, the importance of interrelations between changes in approach bias and alcohol use over time, or a moderating role of WM. It is important to consider the potential consequences of repeated online approach bias assessment (e.g., changes in stimulus valence) when interpreting these results.