I have three groups taking pre-post tests and I scored the tests items as binary 1 for correct, 0 for incorrect. The issue is that I used the total (Sum Scores) of the items (e.g. 12 correct out of 24 (total)) for exploratory & descriptive data analysis and plotting. However, I used the original binary data for modelling (Generalized Mixed-effect Model)
Also, I ran ANOVA for assessing the baseline knowledge (pre-test) to see if the groups started at the similar level before the intervention.
I am very beginner in data analysis and I would like to ask if I am doing this right.
Thank you very much!
I don't do things the same way, but I understand the intent.
Pretesting, testing, comparing the level of growth and accuracy in the testing, as well as any regression. Store it as Decimal and then convert to Binary. As a Binary Scientist with a few scores under my belt now, I can tell you that your system is broken unless you use a longer form of binary, and Binary alone is a bad system to try to use. Using Decimal as part of the process is natural and works.
Regression, Progression, Retention, and such are good grading metrics, but from a statistical point of view, 24 questions is pretty light. A hundred-question quiz, or a series of quizzes that equal or pass 100, would be better.
Also, the size of the class can be a statistical issue as well. if there are only 4 students, then you will have a lot of deviation in the study. The general rule of thumb is 10% of your population, or at least 40. Then you have a sizeable enough pool to get meaningful results.
- Badges
- Science topic
- Similar topics
- Mathematics