From: D on
I'm running a logistic regression in SAS and I am getting two inflated
point estimate. What might be causing this problem. Any thoughts/
suggestions on how to fix this problem would be appreciated. -Thanks.

Analysis of Maximum Likelihood Estimates

Standard
Wald
Parameter DF Estimate Error Chi-Square Pr > ChiSq

VAR1 1 -10.9605 0.6418 291.6207 <.
0001
VAR2 1 0.000391 0.00143 0.0747
0.7846
VAR3 1 0.00630 0.0109 0.3329
0.5640
VAR4 1 6.4819 0.6571 97.2911 <.
0001



Odds Ratio Estimate

Point 95%
Wald
Effect Estimate Confidence Limits

VAR1 <0.001 <0.001
<0.001
VAR2 1.000 0.998 1.003
VAR3 1.006 0.985 1.028
VAR4 653.187 180.166 >999.999
From: Garnett McMillan on
My guess is that there is a scaling issue with Var1 and Var4.
A unit increase in Var1 is causing a decrease in the probability of
the event from 1 to 0. Var4 has the opposite effect.

My first approach would be to standardize the predictors to mean = 0,
sd = 1, using, for example, PROC STANDARD.
From: shiling99 on
On May 12, 10:40 am, D <dana201...(a)gmail.com> wrote:
> I'm running a logistic regression in SAS and I am getting two inflated
> point estimate. What might be causing this problem. Any thoughts/
> suggestions on how to fix this problem would be appreciated. -Thanks.
>
> Analysis of Maximum Likelihood Estimates
>
> Standard
> Wald
> Parameter DF Estimate Error Chi-Square Pr > ChiSq
>
> VAR1 1 -10.9605 0.6418 291.6207 <.
> 0001
> VAR2 1 0.000391 0.00143 0.0747
> 0.7846
> VAR3 1 0.00630 0.0109 0.3329
> 0.5640
> VAR4 1 6.4819 0.6571 97.2911 <.
> 0001
>
> Odds Ratio Estimate
>
> Point 95%
> Wald
> Effect Estimate Confidence Limits
>
> VAR1 <0.001 <0.001
> <0.001
> VAR2 1.000 0.998 1.003
> VAR3 1.006 0.985 1.028
> VAR4 653.187 180.166 >999.999

Two estimates have very low chi-squire value. If the data is highly
collinear, it may result in the inflated estimates. This is just one
of posiblities.