Date:         Wed, 7 Sep 2005 19:29:44 +0100
Reply-To:     starborn@statisticsdoc.com
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         Keith Starborn <starborn@statisticsdoc.com>
Subject:      Re: Collinearity Statistics
Comments: To: Omar Farook <omarfdawood@yahoo.com>
In-Reply-To:  <20050907094555.81777.qmail@web90204.mail.scd.yahoo.com>
Content-Type: text/plain; charset=ISO-8859-1; format="flowed"

Keith Starborn
www.statisticsdoc.com

Dear Omar,

Many investigators will exclude a predictor variable for reasons of
multicollinearity if the VIF is greater than or equal to 4 (likewise if the
tolerance is lesser than or equal to .25).  Your tolerances and VIFs
look good.
Even so, you might want to consider looking at the bivariate correlations
between the predictor that has a VIF of 2.3 and the other predictors.  I might
be a little concerned that this predictor and one other predictor are highly
related, leading to potential supression effects when both are entered into a
multiple regression.  There might be no problem, and if there is, then perhaps
the two predictors could be entered as a composite.

Best,

KS


Quoting Omar Farook <omarfdawood@yahoo.com>:

> Dear Friends.
>
>
>
> I have a multiple regression model consist of 7 independent variables.
>
> the tolerance figures between 0.435 & 0.806 &  the VIF six values
> between 1.241 & 1.956  & one is 2.300.
>
> are there any evidence from the figures above that the model
> suffering from multicollinearity problem?
>
> I will be very appreciate if  you support your answers with some references.
>
> thanks in advance
>
>
>
> Omar.
>
> omarfdawood@yahoo.com
>
>
>
> ---------------------------------
> Click here to donate to the Hurricane Katrina relief effort.
>



--
For personalized and experienced statistical consulting,
visit www.statisticsdoc.com