effect size in hierarchical multiple regression (categorical predictor)

#1
Hello all

So glad to have found this useful forum and have already gotten some clarity on other issues related to the hierarchical regression described below.

My question at this point regards effect size for dummy-coded categorical predictors.

I am using hierarchical multiple regression for a DV (e.g., anxiety), with one categorical (Condition - 3 levels [Experimental, Placebo, Control], dummy coded) and one continuous predictor (Secure Attachment).

I have a main effect of Condition (yay!) and hypothesize moderation by an interaction of Condition x Secure.

The model is run (SPSS v. 14.0) as follows:

Step 1: post-manipulation DV= Pre-manipulation DV; Placebo; Control
Step2: Step1 + Secure
Step3: Step2 + PlaceboxSecure; ControlxSecure

The statistician on my dissertation committee wants me to calculate effect size for the dummy variables by dividing B (NOT beta) by the standard deviation of the DV, which yields an effect size in standard deviation units. I cannot find a reference to this anywhere, and she is away until shortly before my defense.

Can anyone help me and tell me:
1. Are these values (B/sd) are comparable to betas for the continuous variable?
2. Should the same procedure (dividing B/sd) be used for the interaction terms?

Thanks so much for any input!
~sybil