I tried to use chi-square but i have only percentage data, so it not suitable for chi-square because frequency(count) data is used for chi-square. any suggestion?
.
it does not make sense : in order to implement a goodness of fit statistical test, you need the counts
the "goodness of fit test" with respect to an unbiased coin will not give the same result for 3 heads in 4 draws (won't reject the null at 95% confidence level) and 3000 heads in 4000 draws (rejects the null at 95% confidence level) !
.
If you have a distribution model for the outcomes (possibly a beta model) then you can use the Kolmogorow-Smirnow-test (KS test) to get the significance of your data under the hypothesized distribution model. Another option is the Anderson-Darling test, but you must be careful here, because many software use it especially to test the normal distribution. The R package goftest provides the function ad.test that can test a vector of values against any distribution model.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Applied Statistics
- Survey Design
- Questionnaire