What Laboratory Alcohol Experiments Capture About Real-World Drinking

Psychological science often relies on laboratory paradigms to identify mechanisms assumed to operate in daily life, yet the degree of laboratory-to-life generalizability is rarely tested directly. Alcohol research provides an especially strong test case because acute subjective responses to alcohol are clinically informative candidate mechanisms of alcohol use disorder risk and can be assessed in both experimentally controlled and naturalistic drinking settings. We studied 282 heavy-drinking young adults who completed an unscripted triadic alcohol-administration session and event-contingent ecological momentary assessment (EMA) of real-world drinking episodes. Identical self-report items assessed intoxication, stimulation, sedation, and craving at closely aligned intervals in both contexts. Using multilevel structural equation models, we compared both average response levels (“volume”) and within-episode change (“tempo”). Some average trajectories aligned across contexts, whereas others diverged, underscoring the importance of interpreting person-level correspondence in relation to the trajectory being compared. By contrast, correspondence in within-episode change was weaker and more selective, with sedation showing the clearest positive slope correspondence. Exploratory recentering analyses suggested that slope correspondence may be stronger early in drinking episodes than at the laboratory midpoint. These findings indicate stronger cross-context generalizability in overall response level than in within-episode change, while also showing that dynamic correspondence depends on response domain and timing. More broadly, the results support the value of laboratory and naturalistic methods as complementary tools for characterizing alcohol-response processes.