Effect sizes were estimated as partial ω2 for the main effects of strain and sex, and for the sex × strain interaction (Table 3). Consistent with statistically significant strain effects, 146 measures under consideration had large effect sizes (ω2 > 0.30), 93 had intermediate effect sizes (0.30 > ω2 > 0.10) and 19 had small effect sizes (ω2 < 0.10) (Fig. 1a). Locomotor activity in a variety of apparatus and under different drug exposure conditions typically had large strain effect sizes, consistent with the previous reports (Wahlsten et al. 2006). Other traits with large effect sizes include acoustic startle response, pre-pulse inhibition, morphine withdrawal and alcohol withdrawal. Intermediate effect sizes were observed for nociception-related traits, morphine side-effects, ethanol-induced ataxia, baseline HIC, blood ethanol concentration, cocaine-conditioned place preference and anxiety. Only a small number of measures had effect sizes below 0.1. These were most often measures derived from linear combinations of two measures, for which the expected variance of the derived scores is at least four times the variance of either measure alone (Johnson & Wichern 1998). The 30% effect size is a conventional guide for traits amenable to genetic analysis.
