Step 1 Determine whether a significant interaction is present between the hypothesized moderator variable (z) and the predictor variable by the moderated regression analysis (MRA) procedure. If a significant interaction is found, proceed to Step 2. Otherwise, go to Step 3.
Step 2 Determine whether z is related to the criterion variable. If it is, z is a quasi moderator variable. If not, z is a pure moderator variable. In either case, the moderator influences the form of the relationship in the classic validation model.
。。。(Sharma et al., 1981, p.296)