Substance‐free and substance‐related reinforcement in relation to substance use: A meta‐analysis
Brian M. Bird, Molly L. Garber, Samuele F. Acuff, Keanan J. Joyner, James G. Murphy, James MacKillopAbstract
Background and aims
Contemporary behavioral economic theory conceptualizes addiction as an operant reinforcer pathology in which substance use is substantively influenced by reinforcement obtained through substance‐free sources. The aim of the current meta‐analysis was to quantitatively synthesize cross‐sectional associations between both substance‐related and substance‐free reinforcement in relation to substance use and substance‐related problems.
Methods
Systematic bibliographic searches were conducted in PubMed and PsychINFO from 1998 to 20 March 2026. Main eligibility criteria included human, peer‐reviewed studies that measured substance use, the reinforcement ratio (proportionate substance‐related reinforcement), substance‐related reinforcement and/or substance‐free reinforcement, which yielded 28 studies and 31 samples. Pearson's r correlations were extracted for the reinforcement ratio, substance‐related reinforcement and substance‐free reinforcement in relation to substance use problem/severity and quantity/frequency outcomes. Six random effects meta‐analyses were conducted based on these key variables, with an average aggregated sample size of n = 5663 (range = 4344–7258).
Results
The reinforcement ratio (RR) and substance‐related reinforcement (SR) were statistically significantly associated with both substance use problem/severity [RR r = 0.33, 95% confidence interval (CI) = 0.27–0.39, number of effect estimates ( k ) = 19, I 2 = 85.06, SR r = 0.34, 95% CI = 0.29–0.60, k = 12, I 2 = 94.28, P s < 0.001] and quantity/frequency (RR r = 0.38, 95% CI = 0.31–0.46, k = 17, I 2 = 86.22, SR r = 0.34, 95% CI = 0.25–0.43, k = 10, I 2 = 87.99, P s < 0.001). Substance‐free reinforcement was statistically significantly negatively associated with quantity/frequency ( r = −0.11, 95% CI = −0.17 to −0.04, k = 13, I 2 = 81.57, P = 0.002) but not problem/severity of substance use ( r = −0.07, 95% CI = −0.14 to 0.00, k = 16, I 2 = 83.38, P = 0.051). Continuous study moderators (% male, % White, study year) and categorical study moderators (reinforcement measurement tool) explained some heterogeneity across analyses. Overall, there was minimal evidence of publication bias.
Summary
Substance‐related reinforcement and the reinforcement ratio appear to be moderately strong correlates of substance problems/severity and quantity/frequency outcomes.