Hello, everyone. I am working on analyzing data from a pre and post test questionnaire. There are 4 Likert scales made up of 8-10 Likert items each, and each Likert item has 5 options ranging from "Very unlikely" to "Very likely" or something similar. Each item will be given a numeric value of 1 to 5. I have read that I should find a composite score (sum or average) for each scale. I want to then compare the pre and post tests, perhaps using a t-test or Wilcoxon signed rank test.

I have many questions, but to start:

  • I would like to do a pre/post test comparison for each scale--would it be valid to do a comparison for an overall score as well?
  • There will definitely be a pre and post test, but we may give an additional post test (same questionnaire) at 3 months out and 6 months out. My main interest would be in a difference from pre to any of the post tests, but it might be interesting to test for differences across all tests (pre, post1, post2, and post 3). Would I use ANOVA for such an analysis?
  • Later on, once the sample size has grown to a sufficient size, I may also want to compare the pre and post tests between two groups. For instance, say we want to look at the data by gender or by schooling method (private, public, or home). How would I approach that?
  • I appreciate any help and direction you can provide.

    More Daniel Stone's questions See All
    Similar questions and discussions