 Treatment effect moderation

Variation in the treatment effect with a baseline characteristic. An interaction tests differences in treatment effects; significance in one subgroup and nonsignificance in another is not itself evidence for an interaction.