1
$\begingroup$

As the title reveals, does it make sense to estimate an interaction when you already see that a main effect is insignificantly different from 0 in regression?

In my study I am actually interested in this interaction, as its part of my research question. Can I conclude the interaction will be insignificant too, or can it happen that the interaction will be significant? I usually only estimate interactions when the main effects are significant, hence I am asking/unsure.

$\endgroup$
2
$\begingroup$

Yes, the only requirement for an interaction term (between two variables) in a regression model, is that both main effect variables must be present in the model. The main effects do not have to be significant, but rather just be in the model.

$\endgroup$

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.