Implications of low standard deviation for a moderator


New Member

I have a model where Y = b0 +b1*X+ b2*Z +b3*XZ. Z is a moderator in my theory. In my data, Z is measured on a 5 point Likert type scale with mean = 3.27 and s.d. =.43. A reviewer wants to me to discuss the implications of what they consider "low" s.d. for the significance and interpretation of the moderation effect. Intuitively, low s.d. should provide a more conservative test because of range restriction. That is, it should make it harder to detect moderation. How do I discuss this in more statistical and substantive terms?