Dear Anderson,
Again, many many thanks!!
A final comment, you state that the interpretation of a significant positive or negative Q1*Q2*G interaction is difficult, as it may indicate that the interaction Q1*Q2 varies according to G, that Q1*G varies according to Q2, or Q2*G varies according to Q1.
Is there any way to know what exactly is driving the significant interaction term? Obviously, we would like to indicate that the interaction Q1*Q2 varies according to G. However, is there a method for conforming such claim (perhaps with scatter plots or other graphic visualisation)? If not, how should one actually report and discuss such interaction effect?
Any assistance would be greatly appreciated!!
Regards,
Mauro
|