My survey data contains 10 different questions all recoded into 'Correct' (1) and 'Incorrect'. I have 2 IVs which are also categorical. I need to find out whether each treatment condition affects the answer to the questions. In order to do this in SPSS I have 2 options:

  1. Run binomial logistic regression with each question as my DV and the two categorical IVs. However, do I run binomial logistic regression for each question separately? Does SPSS allow to run this with multiple dependent variables?

  2. Run multinomial logistic regression. In this case, I created a variable called 'correct_answers' which indicate the number of correct answers given by each participant. So, in order to run multinomial logistic regression, I would put 'Correct_answers' as my DV (reference as last by default) and Factors (since categorical) or IVs would contain each question and the two previous IVs. Will this work or am I doing something wrong?

Thank you.

  • 1
    $\begingroup$ This seems to mostly be about statistics, not programming so I vote to leave it open. $\endgroup$
    – Peter Flom
    Commented Mar 17, 2018 at 14:48

1 Answer 1


Questions about how to code are off topic here, but your question has a statistics component as well.

Multinomial logistic will not be right here, as far as I can tell. If you are interested in the number of correct answers then that would be the DV and you would use a count regression model such as Poisson or negative binomial regression with your two IVs.

Whether to run multiple logistic regressions with each DV or one regression with all of them depends on whether you are concerned about the relationships among the DVs. If you are, then you would need multivariate logistic regression. I don't know if SPSS can do this. Be careful in searching as some people use this term to mean one DV and several IVs (which is not what you want). These models are tricky, it's much simpler to run separate models for each DV.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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