I have pre and post test data evaluating the effectiveness of a lesson about how to help friends who are being bullied. I am trying to see if more students know how to help a friend being bullied after the lesson on bullying compared to before the lesson. For example, one of the questions I am running the analysis on is: "Do you know how to help a friend who is being bullied?" The data are yes/no, i.e., dichotomous.

I am doing McNemar's test using matched pairs in R. I want to see how students answered before the test compared to after the test. There are 302 students in this analysis.

I know I can get an odds ratio by using an exact McNemar's test but how do I do an odds ratio if my sample is relatively big and I don't need to use exact McNemar's test? I don't want to use an exact McNemar's test as it is more conservative than just a McNemar's test.

I know the odds ratio is looking at the discordant pairs (those that moved from No to Yes and Yes to No from the pre to the post test). How do I do an odds ratio with confidence intervals from this data using McNemar's test in R? I know I can do an odds ratio by hand but I need the confidence intervals too.

This is the 2x2 table from the question above. These are the data I want to do a McNemar's test on to get an odds ratio:

             No   Yes
 BEFORE No   197  62
        Yes   6   37
  • $\begingroup$ Since the procedure is exactly the same however you calculate McNemar can you edit to clarify just what you are having problems with? $\endgroup$ – mdewey Jan 1 '19 at 14:41

Your Answer

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

Browse other questions tagged or ask your own question.