Significant two-way interaction, but simple slopes are n.s.

#1
Hi everyone,

I've seen similar questions like mine posted here before, but none of the replies were ever clear enough, so here's my Q. Thanks in advance. :)

I conducted a a variety of simple and multiple regression analyses with two predictors (IV1 and two dummy codes for IV2). So, the complete "steps" of the analyses I have conducted are as follows:

Model 1: IV1 predicts DV
Model 2: Dummy1 and Dummy 2 predict DV
Model 3: IV1, Dummy1 and Dummy 2 are now in same model, plus all interaction terms (IV1 x Dummy1, IV1 x Dummy2)

In Model 3, one of the two-way interaction terms (IV1 x Dummy1) is significant. Visually, it is a cross-over interaction. However, neither of the simple slopes are significant, so I am wondering how to interpret this data. Thanks in advance!
 

hlsmith

Omega Contributor
#2
You keep the basic main effects in the model, but ignore interpreting them (they are conditional on an interaction). So focus on presenting and interpreting the interaction terms.