A sample size near to 20 with pre and post assessment after psychotherapy sessions. Can we apply paired sample T test? or other suggested tests that we can apply on it?
A paired t-test is an appropriate statistic in this case, but with a small N, you may well lack the "power" necessary to find significant results.
You can try the nonparametric Wilcoxon signed-rank test as an alternative.
https://en.wikipedia.org/wiki/Wilcoxon_signed-rank_test#:~:text=The%20Wilcoxon%20signed%2Drank%20test,is%20a%20paired%20difference%20test).
Yes paired t-test would be fine as long as you check the assumptions. As Prof. David L Morgan suggested you should check the power of your test. That can be done with any statistical package. Best, David Booth
P.S. Jos Feys is correct and an added placebo is a better design. However, I would not regard that as required in this particular instance. I would keep it in mind for future research. If you intend to publish this work you should add Jos Feys comment as a possible limitation.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size