Regression for dependent variable with 4 categories I want to do a regression where my dependent variable has four categories (1,2,3,4) which represents the number of dependents. Can I do this with logistic regression? I read somewhere that link=glogit option is useful in this, can somebody please shed some light? I am new to this.
Writing the syntax here would be very useful for me.
 A: I think this document: logistic holds all the information you need (with pointers to how you can do it in both R and SAS). It explains the concept of the proportional odds model and indicates that glogit is indeed the way to go.
If you need more, just google for "SAS proportional odds logistic regression"...
A: If you need imputation (as your comment suggests), look into PROC MI. It is specifically designed for this purpose. One of the many options it has is imputation of an ordinal outcome.
For example, the following code will use proportional odds regression to impute ndependents based on x1, x2 and their interaction.
proc mi;
   class ndependents;
   var x1 x2;
   monotone logistic(ndependents= x1 x2 x1*x2);
run;

You can do only one imputation, but multiple imputations are preferrable. After the analysis you can use PROC MIANALYZE to combine the results.
A: You'll want to look up the literature on multinomial logistic regression, a.k.a. nominal regression.  It's an expanded version of the usual logistic regression.  Coefficients and odds ratios obtained deal with the likelihood of the outcome being A, B, or C as opposed to D, the reference category.  Thus for 4 levels of a dependent variable you'll have 3 tables with coefficients, odds ratios, etc.  You'll want to make sure you choose that reference category intentionally:  to which result will you and your readers most want to make comparisons?  Sorry I can't provide syntax but SAS documentation should be a help there.
A: I'd suggest to you to consider using decision trees (such as CART) for such problems.
I see that SAS Enterprise Miner has some functions for decision trees:
http://sas-x.com/2011/01/decision-trees-in-sas-enterprise-miner-and-spss-clementine/
