Sam,
1. Compute XYMOD = (Y-Mean(Y)) * (X-Mean(X))
2. Enter X, Y to predict Z as model#1 (SPSS uses something like block 1, or
step 1 or such)
3. Enter X, Y, XYMOD to predict Z as model#2 (aka, the full model)
4. Subtract Rsquared for model#1 from Rsquared for model#2 to get Rsquared
change (aka, Rsquared delta), and test for significance. SPSS should have
this test come out automatically, I believe.
Tabachnick & Fidell's Multivariate stats book is a good reference.
JR
Jeff Rasmussen, PhD
The Art of The Experiment:
experimental design and statistics tutorial cd-rom
http://symynet.com/educational_software
|