yao lu posted on Tuesday, December 13, 2011 - 4:28 pm
I have two group respondents, group A using product A, and group B using product B. Group A has a very high familiarity with product A, group B has a very low familiarity with product B.
the product familiarity is the moderator in the model, and is a continuous variable.
So in order to test the moderating effect, i can (a) conducting one SEM based on the combined sample from group A and group B (similar to a regression test), or (b) conducting a SEM on each group, and comparing the equity of path coefficients associated with the product familiarity construct in each SEM.
Could you please advice me if any of these two methods is significantly better than the other, and if so, why?