two differemt groups were assessed by around 20 questions.
1.what test should be used for comparing among each yes or no question?
2.what test should be used for comparing total number of "yes" among the two groups?
1. Student's t-test for two samples using the proportion of yes in each sample.
2. I would create a new variable summing up the total number of yes in each group. Then I'd perform again a student's t-test for two samples on the average of this variable.
1.Run t-test (Basically Paired sampled) because you have a sample that has been paired into YES or NO group.
2. Run a Wilcoxon Signed Ranked Tests to know the number of YESs and NOs and the level the YESs are at a particular point of NOs
OR
You compute variables to get a parametric values for YESs and NOs
1. BINARY DATA may be tested under the DeMoivre-Laplace theorem where the Z score is obtain by:
Z = (YES - nP) / sqrt(nPQ)
where YES = total number of YES and n = sample size, P = (success + 1)/(n + 2) and Q = 1 - P. For probability of the Z score, take the Z score and look up the corresponding % probability in the Z table.
2. TWO GROUPS COMPARISON may be analyzed for the significant difference of group 1 and group 1 where Z1 and Z2 are used. Z = (Z1 - Z2). The % difference of (FZ1) - F(Z2) may be looked up at the Z table.
If each of the questions have to be compared between two groups the proportions test for two groups.
If total scores are taken on the questionnaire in each of the groups and to compare;
If Standard deviation is small the Independent t test
If S.ds are wide than perfom Non parametric test of comparison like Mann Whitney U test.
What are your main research question and research hypothesis?
- Badges
- Science method
- Similar topics
- Teaching