I am having students take a test in which they have an option of choosing "don't know" to test questions. I would then like to compare the likelihood of students choosing "don't know" for type A test questions compared to type B test questions based on the number of don't know responses. Is an odds ratio or relative risk appropriate here or should I use a different statistic entirely? I would consider administration of a test to be prospective in nature.
Seeking help from the Talk Stats community! Thank you.
Seeking help from the Talk Stats community! Thank you.