Interaction terms with non-signficant main effects

#1
I am using a logit model to do the analysis on my dissertation data. 2 of the variables interact, so I have an interaction term. My question is 2 parts:
1. When I run the model without the interaction term, one of the main effects is not significant. Can I still include the interaction term?
2. If so, when I run the model with the interaction term, the interaction is significant, as is one of the main effects, but the other is not. Can I still interpret the interaction term?

thanks,
SK
 

hlsmith

Omega Contributor
#2
Absolutely, if the interaction term is significant you forget about its main effects (though keep them in the model) - since they are conditional on the other variable. How are the two independent variables formatted (e.g., both binary)?
 
#3
thanks, hlsmith! yes, both independent variables are binary.

Absolutely, if the interaction term is significant you forget about its main effects (though keep them in the model) - since they are conditional on the other variable. How are the two independent variables formatted (e.g., both binary)?