Dear all,
Thanks to everyone who replied to my previous post (‘Comparing measures?’).
I had tried to simplify a more complex problem but don’t think I succeeded, so
here’s the problem in its entirety.
I am trying to compare two measures. Both measures are designed to tap a
construct that is expected to moderate the relationship between Variable X
and Outcome Y (as demonstrated by an interaction between Measure *
Variable X). Scores on Measure 1 represent the mean of 12 items which are
designed to measure this construct. Scores on Measure 2 represent the mean
of 4 of the 12 items used in Measure 1. The data I have for Measures 1 and 2
are taken from one and the same sample. For various reasons, I have
hypothesised that Measure 2 will more strongly moderate the X * Y
relationship than will Measure 1.
So, I have run two regression models. One includes Measure 1, Variable X, and
a Measure 1 * Variable X interaction as predictors of Outcome Y. The second
includes Measure 2, Variable X, and a Measure 2 * Variable X interaction as
predictors of Outcome Y.
How can I examine whether there is a significant difference between the
magnitude of the predictive effect of the Measure 1 * Var X coefficient and
that of the Measure 2 * Var X coefficient?
Many thanks,
Ben
|