Cross Validated is a question and answer site for people interested in statistics, machine learning, data analysis, data mining, and data visualization. It only takes a minute to sign up.
Sign up to join this community
An experiment about the review process was made. Participants were students visiting a course. One group had to solve some exercises and the other group had to review their work and give feedback.
At the end, an exam for both groups was held. Exam grades for every participant were collected (aggregated to a score between 0 and 100%).
Which statistical method could be used to prove the hypothesis that says that reviewers learn more than people being reviewed?
Because these are test scores (bounded, etc.), it may be best not to assume normality and homogeneity of variance. In addition, your groups are non-independent in an interesting way. With these facts in mind, given that you want to compare two groups, I might try the Wilcoxon signed rank test. Each matched reviewer-reviewee would be a pair. In essence, you are testing to see if one tends to score higher than the other. This should be available in most standard statistical software. For example, in R you could use ?wilcox.test(, paired=TRUE).
