Responses were obtained on likert scale.User were just asked to rate the relative importance of factors in their eyes.
Researcher "believes", all factors are interlinked/correlated hence AHP alone won't serve the purpose.
No dependent variable!
How to rank them?
Dear Saad,
You could try a Friedman's ANOVA, although 5 point Likert responses are not ideal. If you are using SPSS and you obtain a significant result, double click on the output table to open the Model Viewer, then change the view to pairwise comparisons. This will perform a post hoc test of each pair of responses with adjusted significance values.
The alternative is a dimension reduction technique, such as exploratory factor analysis, although you don't sound very keen on the idea. There is a requirement of correlations to be less than 0.8 between all pairs of items for this technique to be used.
Dear Saad,
To begin with, study the structure for each of your answers to your questions. Review the resulting structures - they will show where the largest accumulation of cases (at 1, 2 .. or at 5 for example) on individual issues.
Try also with correlation coefficients at weak rocks.
They will show you the dependence between pairs of variables.
You can also use the nonparametric ANOVA method
I would also recommend some of the nonparametric regressions (Logit, Probit, Ordinal) using stepwise patterns. Much depends on the type of your data, the number of observations.
It all depends on what you are trying to prove and what information you have!
Success
M.Kaneva
- Badges
- Science topic
- Similar topics
- Mathematics
- Applied Mathematics