In addition to testing for a significant main effect of the stimulation we tested for significant stimulation by group interaction. We first evaluated the true difference between the effect sizes for the two groups as:30Δtrue=(μsJAWS−μnJAWS)−(μsGFP−μnGFP)The approximate distribution of this difference under the null hypothesis that there was no difference between the groups was evaluated by creating an ensemble of permuted datasets in which we randomly assigned subjects to the JAWS and GFP groups and the interaction P value was calculated as above.
