One-tailed p values in repeated ANOVA?


I am examining the results of 3 groups on a series of tasks using repeated ANOVAs (mixed model), which include 2 within subject factors. I hypothesise that my two clinical groups will perform worse than the control group on all the task, but I don't have priori hypotheses regarding the two clinical groups (i.e. I am not expecting one to peform worse/better than the other). My questions are the following:

1) Is it ok to divide my p value by two for the results of the main group effect?
2) can I divide the p values for any interactions involving the group and a within subject factors?