@mcc The important thing is to try a number of different "corrections" until you find it creates a statistically significant result you can publish