My study originally set out to measure one thing as an individual differences variable (so basically correlation/regression) and a few other measures on a task paradigm. However, I've since found out that my distribution (for variable 1 - the individual differences one) is bimodal and that the task is better analysed when treated as repeated measures (of which there are 8).
My supervisor is aware that initially I was going to do correlations/regressions but I'm not sure that this is the correct way to proceed due to the bimodal distributions and repeated measures. Does anyone know if its a heinous crime to use the individual differences measure to split into 2 groups (using bimodal distribution as the justification) and just run an ANOVA? This is basically what I've done. I'm an undergrad student and I really don't understand repeated measures regressions...
If I do this, should I do it as though it is post-hoc splitting the groups up or can I say I decided it in advance as part of the design as such?
ANY advice on this matter would be hugely appreciated.
Advertise on Talk Stats