1
$\begingroup$

In my data set I have one dependent variable (dead or alive) and 37 predictor variables. 35 of my predictor variables are dichotomous (Obese: 1 or 0, Female 1 or 0, etc), however 2 of my variables are coded into ordinal categories (ie, age <69: 1, age 70-74: 2, age 75-79: 3, and age >79: 4). In order to run a binary logistic regression with all 37 variables, should I split my ordinal variables such as age into 4 different columns (one for each category) and make them binary or is there an easier, better way to accomplish this?

I tried leaving the ordinal variables as they are and running the logistic regression, however, for some reason they don't show up in the statistical output...

$\endgroup$
1
$\begingroup$

The binary logistic procedure is going to treat the variables as continuous covariates, which would not be appropriate for ordinal variables. The NOMREG procedure accepts factors, so if you use that one, you can just use the variables as they are.

The SPSSINC CREATE DUMMIES extension command makes it very easy to create dummy variables such as what you would need to do here. With a little more tedium you can do this with COMPUTE, of course.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.