We polled an online survey to 2 groups. We received 56 responses. Category choices for respondents were: 'Satisfied', 'Dissatisfied', and 'Neither satisfied or dissatisfied'. What type of statistical analysis could be used to determine a) differences between the 2 groups, and/or b) differences between the 3 types of responses to questions?
For (a), you could probably treat the data as ordinal and do ordinal regression to estimate the difference between groups. I'm not sure what you mean for (b).
As you seem to have an ordinal variable (satisfied, neither, dissatisfied) and a nominal variable (group) I'd suggest an association test like chi squared.
when categories are ordinal a chi2 test wastes information by treating them as nominal. a more sensitive test of association is based on G&C gamma, that is you will get more clear evidence against the null of no association, if any. Also, if cell frequencies are small you better use an exact test based on all possible permutations of cell frequencies yielding the observed marginal distributions.
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Analysis