1
$\begingroup$

I ran a study for my Master's thesis that counted the number of people who attended different courses that were advertised in different ways, and how many passed/failed each. I used a Chi Square analysis to compare the numbers of people in each category across the four courses. However, there is a chance some of the people attended multiple courses but I don't know for sure.

One of my coursemates suggested I use McNemar's Chi Square instead which is for repeated measures tests, but I don't feel this is right as although they could be the same people, they could also not be.

What is the best approach for this? Stick with Pearson's Chi Square, despite the chance they could be the same people or use a repeated measures test?

$\endgroup$
1
$\begingroup$

The only way you could use McNemar's test is if you could pair the observations for each individual. That is, you would have to know if Bob Belcher passed this class and if Bob Belcher passed the other class. In addition, you could only use McNemar for those individuals that had paired data. Finally, I think McNemar would be asking a different question. For example, on the pass/fail question, McNemar asks if one course is more difficult than another course. Whereas you want to ask if there is an association between advertising and passing.

$\endgroup$

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.