Unstandardized regression coefficient in Multiple Regression Interaction

#1
Hi all,

I am trying to interpret a significant cross product interaction in a Multiple Regression and I am getting more and more confused. There seem to be only a few decent resources for this procedure (e.g. Aiken & West, 1991) and what I have isn't overly user friendly and informative!

Okay - Prior to doing the multiple regression in SPSS, I have centred the single predictors and the cross product. Each of the 3 models is significant in the ANOVA table. In the coefficients table, one of the IV predictors is significant, the other is not significant, and the interaction is significant. Now the Unstandardized regression coefficients for the individual IVs are very small (.007 & .008) but for the interaction the URC is 0.000 (ie. zero). Does this mean that there is actually an interaction or no interaction? When I graph the interaction as per Aiken & West instructions, the lines are totally parallell. I can interpret the graph okay, but I'm calling it an interaction when I'm not convinced it is one.

If anyone can make sense of this and point in the direction of an appropriate, user-friendly resource, I would be grateful!!!
 

vinux

Dark Knight
#2
I am not familiar with 'Aiken & West instructions'. I think the values of IVs are in higher scale than response values.

I would suggest to take the transformation for IVs. Create new IVs by IV1 = IV/1000. Perform the analysis with IV1 s( new IVs).