principal component analysis - which variables should I use?

2017-12-13 17:40:29

I am interested in studying different types of tests (oral exam, multiple choice exam, open-question exam) and different study strategies (re-reading the textbook, study guide, flash cards, study group). For example, if the student had an oral exam, he/she might chose to prepare with a study group (for the sake of this example, only one strategy is possible). My ultimate goal is to see which strategies yield the highest test score for each exam. So with this in mind, (if not PCA) what other approaches would be helpful for visualizing trends?

I thought about running a principal component analysis but I am not sure how to arrange my data. Does it make sense if my variables are the type of test: oral_exam, MC_test and openQ_test with categorical variables for each of the study strategies? Or should each study strategy be a variable with categorical variables for the type of test? I assume the former is better since there is more variation in the study strategies than exam t