Dear allstat listers,
I feel a bit shameful to ask such a basic question but:
Looking to measure the relative importance of my independent variables
within a STEPWISE regression equation, I have come across various readings
offering guidance to achieve a basic task.
All my dependent variables have been measured on the same scale - which I
believe allow me to use the coefficients as an indication of relative
importance. However, some people advise the use of the t-test.
Looking at the outcome of my analyses, I observed discrepancies in rank
order between t-tests and coefficients.
I know this is a rather basic question, but it has raised some doubts.
Is it correct to use the coefficient when all variables have been measured
on the same scale ?
What is the richest information ? (As the t-test is based on standardized
data, I got the feeling that I am loosing some valuable information going
that route in the present case).
Looking forward to your insights
Olivier Van Parys
Senior Research Manager @ Amárach Consulting
28 Upper Fitzwilliam Street
Dublin 2
Tel: 01 6619147 Fax: 01 6610312
E-mail: [log in to unmask]
Web: www.amarach.com
|