Examine collinearity diagnostics: see if there are any condition indexes >30. For those that are >30, next examine the associated variance decomposition proportions. See if there are any that are >.50: this will allow you to identify variables that have high levels of collinearity.
This will work for funny variables and non-funny variables---even the tragic ones. Of course, all of this assumes that you have a sufficiently large sample, ie, 10-20 subjects per variable.
Scott R Millis, PhD, ABPP (CN,CL,RP), CStat, CSci
Professor & Director of Research
Dept of Physical Medicine & Rehabilitation
Associate - Dept of Emergency Medicine
Wayne State University School of Medicine
261 Mack Blvd
Detroit, MI 48201
Email:
[hidden email]
Tel: 313-993-8085
Fax: 313-966-7682
--- On Sun, 3/29/09, Pushpender Nath <
[hidden email]> wrote:
> From: Pushpender Nath <
[hidden email]>
> Subject: help in treating multi-colinearity
> To:
[hidden email]
> Date: Sunday, March 29, 2009, 6:56 AM
> Hi Fellows
> I apologise as my question is more of Statistics and less
> of SPSS.
> Could you please elaborate best practices in dealing with
> Multi-Colinearity in multiple linear regression. I usually
> deal with
> around 30 explanationary variables including some funny
> variables.
> Dummy variables are in form of 1 and 0.
>
> --
> Sent from Gmail for mobile | mobile.google.com
>
> Regards
>
> Pushpender Nath
>
> =====================
> To manage your subscription to SPSSX-L, send a message to
>
[hidden email] (not to SPSSX-L), with no body
> text except the
> command. To leave the list, send the command
> SIGNOFF SPSSX-L
> For a list of commands to manage subscriptions, send the
> command
> INFO REFCARD
=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD
Locked
