Can I run a mixed model ANOVA with a within-subject factor of time (baseline vs. after the training) and a between-subject factor of condition (2 experimental groups and 1 control)? In that case, are contrast tests pointless to look at? I will also split the file to compare two groups (each with 3 conditions). Would you suggest a different analysis? In addition to this, is the normality of the outcome variables important for mixed ANOVAs? If I transform the variables that are not normal would that be okay?