I have two independent variables (both categorical) and one dependent (categorical) variable. I want to look at the predictive factor or significance of those two independent variables (either separately or combined) on the dependent variable. Is logistic regression the test I need to use? What if I run this same test on three different groups?

  • $\begingroup$ You should try $\chi^2$ for contigency tables, logistic regression or loglinear models. $\endgroup$ – user83346 Sep 11 '15 at 6:45

Logistic regression is meant for continuous independent variables. The textbook approach in your case is to use a chi-squared test of independence. To omit a variable, just add together the cell counts for cells that differ only in the omitted variable.

A reference: http://stattrek.com/chi-square-test/independence.aspx?Tutorial=AP

  • $\begingroup$ Can I use chi-square if I have more than one independent variable? $\endgroup$ – student4444 Sep 11 '15 at 12:40
  • $\begingroup$ Yes. You would form the table by listing all possible combinations of the independent variables you have along one axis, then listing the values of the dependent variable along the other. I am new to CV, so if the down-voter would like to make a suggestion, I will gladly consider it. $\endgroup$ – eric_kernfeld Sep 13 '15 at 13:49

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.