I am conducting a study in which the Impact of Event Scale–Revised (https://studylib.net/doc/18321853/impact-of-events-scale-revised--ies) was administered to 3 different cohorts. The questionnaire consists of 22 questions that can be answered on a 0-4 scale. The score for each question is them summed to determine a total score. The 3 cohorts consist of 6, 15, and 3 subjects. What would be the correct test(s) to compare mean score for each question and total mean score between the 3 cohorts? ANOVA? Thanks for your help.
If you want to compare each question, you can use chi-square. If you want to compare the total score of the scale or sub-scales, ıt is not suitable to use ANOVA because the last cohort has quite a small sample size. Therefore you can use the Kruskal Wallis test.
I agree with Dr. Sevilay Kilmen as Likert scale is not a continuous finding actually. You can consider it continuous after total counts only. You can check the distribution before finalizing anything.
Given your small sample sizes, I would suggest you to use a Kruskal Wallis test to compare the groups for the individual score and the total score. The Kruskal Wallis test can be used on ordinal and continuous data. If you do have a significant result, you can follow it with a Dunn's test to obtain pairwise comparisons.
- Badges
- Science topic
- Similar topics
- Correction