I have 21 students who have taken two exams. On each exam there are 30 completion questions. 15 randomly selected questions of the 30 have been treated, meaning that certain words or phrases have been bolded and underlined to help student focus on key elements of the question. The remaining 15 questions have not been treated. Question responses are either correct or incorrect, so I am using 1 = correct, and 0 = incorrect.
For each exam, I want to see if there is a statistical difference between student performance on the Treated vs. Untreated questions
Comparing the two exams, I want to know if there is an increase in performance between the first trial and the second on the Treated questions vs. the Untreated questions
How do I organize the data and analyze the data for the two questions above?
Thank you.
Usually for such comparisons you can go for two sample t test
But as your data seems to be binay and might not follow the assumptions of t test so can go for non-parametric alternative of the same which is mann-whiteney u test
http://www.statisticssolutions.com/mann-whitney-u-test/
- Badges
- Science topic
- Similar topics
- Correction