N=6 Comparing pre and post test likert scale responses. Participants are mix of practicing & preservice teachers.
Even the most robust statistical tests such as the Wilcoxon Signed-Rank Test still has assumptions that do not appear to be met. The most important of these is random selection. With what you have, I suggest you compare the pre- and post-results using descriptive methods such as frequency distribution profiles or even medians. You could judge their meaningfulness using an appropriate effect size. The bottom line is based on your brief description, I do not believe any inferential statistical test would be valid.
Hello Rhonda D. Miller. You said that you have "likert scale responses". I've noticed that people sometimes fail to distinguish between Likert-type items and Likert scales, as explained here:
- https://www.john-uebersax.com/stat/likert.htm
Q. Is your DV a multi-item scale, or is it a single ordinal item?
If you have one ordinal item, then some of the resources I recommended recently in another thread might be of interest.
Agresti, A. (1983). Testing marginal homogeneity for ordinal categorical variables. Biometrics, 505-510.
https://users.stat.ufl.edu/~aa/articles/agresti_1983.pdf
See Section 2 (Ordinal Strategies).
You may find other helpful resources on Dr. Agresti's web-page:
- https://users.stat.ufl.edu/~aa/
With N=6 and comparing pre- and post-test Likert scale responses, a non-parametric test such as the Wilcoxon signed-rank test is appropriate. This test compares paired samples and is suitable for small sample sizes and ordinal data.
- Badges
- Science topic