Due to significant differences across impulsivity quartiles (Table 1), alternate models were tested including age, years of education and baseline alcohol use as covariates. None of these variables significantly predicted subjective response and their inclusion in models did not change the study results, thus findings from these alternate models were not reported here.
