>> dave crimkey <d_crimkey@YAHOO.COM> 1/31/2007 11:55 am >>> wrote
I'm running a forward selection logistic regression with one dependent that is dichtomous and some demographics that are categorical as the predictors so they're in a class statement. I'm running 6 different models and two of my models are coming out with nothing entered into the model. So I only get the intercept. How do I interpret this?
First, don't do forward stepwise, or any stepwise selection procedure. All the results are wrong. The parameter estimates are biased, the p-values are too low, the standard errors are too small, the resulting model may be nonsensical and, finally, it's a way of avoiding thought.
(if you want more detail, search the archives for stepwise and Cassell).
Second, please write back to SAS-L telling us more about your problem: Where did the data come from? What is the sample size? What are your hypotheses? etc
Third, to interpret the result of a logistic model with just an intercept is similar to interpreting the results of an OLS model with just an intercept. The intercept will be the overall mean, but, unlike OLS, it will here be the overall mean of the logit.
Peter L. Flom, PhD
Assistant Director, Statistics and Data Analysis Core
Center for Drug Use and HIV Research
National Development and Research Institutes
71 W. 23rd St
New York, NY 10010
(212) 845-4485 (voice)
(917) 438-0894 (fax)