I ran a 2-block hierarchical regression to test for moderation. In the first block I entered the centered IV and centered moderator variable (MV) and in the second block I entered my centered interaction term (product of the centered IV and centered MV multiplied by each other). My DV was the dependent variable for both models. My first model was not significant but the second (full) model was. In both models, both predictor variables were not significant predictors of my DV but the interaction term was.

My question is regarding my interpretation of this: Since my interaction term was significant, I know that the relationship between the IV and DV was moderated by the MV. Does it matter that the IV was not significant in the final model? Also, how do I determine the direction of the moderation?

  • $\begingroup$ No, it doesn't matter. $\endgroup$ – Jeremy Miles Apr 20 '15 at 23:58
  • $\begingroup$ On the basis of theory/expert opinion, you found 2 variables (one predictor?IDV) and one moderator variable (I hope you understand the concept of moderating variable). Then you can run multiple regression (no need of HRA) models with and without interaction. Also, understand that if variables are not at the same scale, then centering/standarizing/transformation can be done. But I can't understand, why are you doing centering only in interaction term. $\endgroup$ – Dr Nisha Arora Jun 15 '16 at 7:43

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.