Drinking Intention–Behavior Links Vary by Affect Among Heavy-Drinking Young Adults: An Ecological Momentary Assessment and Transdermal Sensor Study

Objective: Drinking intention is a predictor of heavy-drinking episodes and could serve as a real-time target for preventive interventions. However, the association is inconsistent and relatively weak. Considering the affective context when intentions are formed might improve results by revealing conditions in which intention–behavior links are strongest and the predictive power of intentions is greatest. Method: We investigated the links between drinking intentions reported in the morning and same-day drinking behavior, moderated by positive and negative affect (PA, NA) in a sample of heavy-drinking young adults. Participants wore the SCRAM continuous alcohol monitor transdermal alcohol sensor anklet for 6 consecutive days in their natural environments and responded to daily ecological momentary assessments that included morning intentions to drink and PA/NA items. Drinking events and patterns were measured usingmorning-report counts and features from the sensor. Bayesian gamma-hurdle and Poisson multilevel models with noninformative priors tested day-level associations. We hypothesized that drinking intention–behavior associations would be strongest on days with high levels of PA, but we did not hypothesize directionality for the NA effect given the conflicting results in previous literature. Results: Day-level drinking intention–behavior associations were stronger on days with higher versus lower PA according to sensors features. Associations were also stronger on days with lower versus higher NA. Conclusions: The strength of intention–behavior links may partly depend on the affective contexts in which intentions are formed. Results could fine-tune intervention approaches by elucidating the affective contexts in which intentions may more clearly link to drinking behavior to reduce the intensity of an episode—better anticipating problematic drinking among young adults.