0

I have a data collected on five point likert scale; aims at studying, the level of preference individuals gives to the different factors, (such as Tax Benefits, Maturity period, Rate of Returns ...) before making investment. Altogether there are 6 factors.

Respondents are employees from three different sectors, i.e, service sector, Manufacturing sector and Banking & Finance sector. I want to study if respondents from this three sectors differs in their attitude.

I read in one of the R hand book that likert scaled data can be analyzed by applying permutations test. But one of the important assumption of permutation test is homogeneity of variance (If am not wrong). So before applying PERMANOVA , I ran PERMDISP test on data ,in R, to check homogeneity of dispersions. Result was significant, p- value =0.017.

I want to ask what will be the suitable test in such situation. I am not interested in applying test of association. I am not that well versed with statistics. Pls do suggest.

kjetil b halvorsen
  • 63,378
  • 26
  • 142
  • 467
Imran
  • 11
  • 4
  • Bootstrapping does not "improve homogeneity of variance". Right now, your question is very unclear. – Peter Flom Nov 17 '19 at 22:16
  • There is no way that bootstrapping by itself can "solve the problem of heteroskedasticity." Please ask your real problem, describe your data, and the research questions you want answers to. Maybe you can get answers from [some posts in this list](https://stats.stackexchange.com/search?q=heteros*+anova++answers%3A1). – kjetil b halvorsen Nov 17 '19 at 18:45
  • Thanks alot for clearing my doubt. I have provided necessary informations. – Imran Nov 18 '19 at 22:32
  • You may want to look over [this thread](https://stats.stackexchange.com/questions/129485/what-are-the-assumptions-of-permutation-test) on the assumptions of permutation tests. – Peter Flom Nov 19 '19 at 12:30

0 Answers0