I have a complex experimental design: 3 groups, each of which come on 4 occassions and I obtain recordings from 250 skin electrodes. I am interested in knowing which of the electrodes provides a useful indicator for predicting the group x occassion interaction. So, if I do 250 ANOVAs (3x4), surely I will need to correct for multiple comparisons? Once I have identified any electrodes of interest I will conduct further experiments.
This thread (Do I need p-value adjustment for multiple ANOVAs on the same dataset?) deals with a similar question with a simpler experimental design and therefore a smaller number of ANOVAs, where arguably, multiple comparisons would not be such an issue. In my case, there are 250 ANOVAs. A 3-way ANOVA seems hard to interpret (perhaps it's just me). Would an FDR adjustment be appropriate for the 250 tests?
