Luci M. posted on Saturday, May 07, 2011 - 1:05 pm
I started out with an SEM model and got significant findings. I then wanted to see if age moderated those paths and went on to create product terms. I reran the model with the mean centered predictors and product terms and found that age was not a significant moderator. What worries me is that some of the initial findings went away (that is from the model I ran first with the raw variables -- not mean centered ones).
Will it be accurate to report model 1 findings and then report the non-significant findings for age? Or do I need to report the coefficients from the second model throughout? It is my understanding that when moderation is not found one can go to report the main effect findings and drop the product term. Is it only the product term that ones drops? Or is it using the mean centered version of those variables to? With other words taking me back to the original model?