Next, alcohol response in the new cohort was determined by a repeated measures ANOVA with dose (two levels) and time (three to five levels) entered as within subject factors. Given the large number of comparisons, in order to reduce Type I error a family-wise error correction (Dar et al., 1994) was used to determine significance with α = 0.004 (0.05/14 main dependent measures = 0.004). To examine if alcohol responses differed between cohorts, a final set of repeated measures ANOVAs was also employed for all measures with cohort (first, second) entered as a between-subject factor and dose and time as within-subject factors. For the between-cohort comparison statistical comparison (e.g., dose × time × cohort), α = 0.05 was used due to our concern that a corrected α may be too conservative to detect differences between groups. The results of both analyses are presented in Table 2. In short, replication was supported if the independent analysis of cohort 2 produced a significant dose × time interaction (p ≤ 0.004) in the first model and the cohorts did not significantly differ in the second model 2 (i.e., dose × time × cohort, p > 0.05).
