Minimizing Measurement Error in Treatment Outcome Estimates: A Meta-Analysis Comparing Estimates Between the Gambling Timeline Followback and Other Self-Report Assessments of Gambling Behavior

Objective: The aim of the current meta-analysis was to examine potential differences in posttreatment effect size estimates for gambling frequency (i.e., the number of days gambled) and gambling expenditure (i.e., the amount of money gambled) when using the gambling timeline followback (G-TLFB) versus other self-report assessments. Method: Using an open-access meta-analysis database of studies on cognitive behavioral treatment for gambling disorder, 22 studies representing 2,824 participants were identified for inclusion. Hedges's g effect sizes representing posttreatment differences on gambling frequency and expenditure between cognitive behavioral treatment versus inactive and minimal treatment controls were calculated, and mixed-effect subgroup analyses examined the effect sizes for each outcome between studies using the G-TLFB versus other self-report assessments. Results: Mixed-effect subgroup analyses indicated that the effect size for gambling frequency was significantly lower for studies using the G-TLFB (g = −0.15) than studies using other self-report assessments (g = −0.71). When examining whether the use of the G-TLFB was associated with the posttreatment effect size for gambling frequency in a random-effect metaregression model that controlled for study grant funding status, the use of the G-TLFB was not significantly associated with effect size. The effect size for gambling expenditure was not significantly different between studies using the G-TLFB (g = −0.22) versus studies using other self-report assessments (g = −0.38). Conclusions: The G-TLFB yields more conservative and precise effect size estimates of posttreatment gambling frequency, but not gambling expenditure, than other self-report assessments. Public Health Significance Statement The current meta-analysis indicated that the effect size estimates of gambling frequency are lower when using the gambling timeline followback assessment compared to other self-report assessments. Resource-intensive assessment methods are necessary to accurately understand the magnitude of differences in posttreatment outcomes between treatment and control.