logistic regression

[Campenni, Estelle]

In running a series of logistic regressions, I have noted that
classification does not change at all from step 0 with the use of the mode
to that found using the regression model. Yet, the chi-square test of the
overall model is significant as are the coefficients as determined by the
Wald stat. How could that be and how should I interpret this? Also, why
would an exp(B) of 1.078 be significant?

Estelle

[SR Millis-3]

Have you assessed the degree of collinearity among your predictors/covariates?

  SR Millis

Estelle Campenni <[hidden email]> wrote:
  In running a series of logistic regressions, I have noted that
classification does not change at all from step 0 with the use of the mode
to that found using the regression model. Yet, the chi-square test of the
overall model is significant as are the coefficients as determined by the
Wald stat. How could that be and how should I interpret this? Also, why
would an exp(B) of 1.078 be significant?

Estelle



Scott R Millis, PhD, MEd, ABPP (CN & RP)
Professor & Director of Research
Department of Physical Medicine & Rehabilitation
Wayne State University School of Medicine
261 Mack Blvd
Detroit, MI 48201
Email: [hidden email]
Tel: [phone]
Fax: [phone]

*********************************************************
This electronic message may contain information that is confidential and/or legally privileged. It is intended only for the use of the individual(s) and entity named as recipients in the message. If you are not an intended recipient of this message, please notify the sender immediately and delete the material from any computer. Do not deliver, distribute or copy this message, and do not disclose its contents or take any action in reliance on the information it contains. Thank you.

[paulandpen]

What is the proportion of 0 and 1 cases in your dependent variable?

> SR Millis <[hidden email]> wrote:
>
> Have you assessed the degree of collinearity among your
> predictors/covariates?
>
>   SR Millis
>
> Estelle Campenni <[hidden email]> wrote:
>   In running a series of logistic regressions, I have noted that
> classification does not change at all from step 0 with the use of the
> mode
> to that found using the regression model. Yet, the chi-square test of
> the
> overall model is significant as are the coefficients as determined by
> the
> Wald stat. How could that be and how should I interpret this? Also, why
> would an exp(B) of 1.078 be significant?
>
> Estelle
>
>
>
> Scott R Millis, PhD, MEd, ABPP (CN & RP)
> Professor & Director of Research
> Department of Physical Medicine & Rehabilitation
> Wayne State University School of Medicine
> 261 Mack Blvd
> Detroit, MI 48201
> Email: [hidden email]
> Tel: [phone]
> Fax: [phone]
>
> *********************************************************
> This electronic message may contain information that is confidential
> and/or legally privileged. It is intended only for the use of the
> individual(s) and entity named as recipients in the message. If you are
> not an intended recipient of this message, please notify the sender
> immediately and delete the material from any computer. Do not deliver,
> distribute or copy this message, and do not disclose its contents or
> take any action in reliance on the information it contains. Thank you.

[Hector Maletta]

The significance test means that you can be 95% confident that the odds
ratio is not really 1.0 in the population. In other words, your factor
actually increases (or decreases) the odds of the outcome. If your sample is
large enough this may happen with pretty low odd ratios such as 1.078.
At the same time, the outcome for each case is predicted by its estimated
probability compared with a criterion for prediction (which is 0.50 by
default). All cases with p>0.50 are predicted to suffer the event, and all
cases below 0.50 are predicted not to suffer it. If the new factor added in
a step increases the odds only a little, the expected probability of the
event for a given case would not change by much. It may change, say, from
0.10 to 0.12, which is always below 0.50 and the case will be predicted not
to suffer the event, just as it was in the previous step, before the new
factor was introduced. As this would happen to most cases, the number of
predicted events may change little or not change at all from one step to the
next.
Therefore there is no contradiction in the fact that the odds ratio of your
new factor is significant but the classification table (predicted vs
observed) does not change from the previous step.

Hector


-----Mensaje original-----
De: SPSSX(r) Discussion [mailto:[hidden email]] En nombre de
Estelle Campenni
Enviado el: Wednesday, June 21, 2006 8:19 PM
Para: [hidden email]
Asunto: logistic regression

In running a series of logistic regressions, I have noted that
classification does not change at all from step 0 with the use of the mode
to that found using the regression model. Yet, the chi-square test of the
overall model is significant as are the coefficients as determined by the
Wald stat. How could that be and how should I interpret this? Also, why
would an exp(B) of 1.078 be significant?

Estelle