Fig. 2: Considering the size of the effect when a treatment effect interacts with sex is critical.

Just because the p value is significant for the interaction doesn’t mean it is biologically relevant. From the statistical modelling, an average treatment effect across males and females can be estimated (orange arrow) or the treatment effect for each sex can be estimated individually (blue arrows). These effects along with their confidence intervals should be considered. A The treatment effect was approximately double the size in the females compared to the males. For this variable this might be considered equivalent and the average effect used. B In contrast the treatment effect here is approximately four-fold smaller in the females and in this case the researchers, depending on the biology, might make the call that the individual estimates should be reported, and attention should focus on why the treatment effect was different.