4

I have Group A, B, C. My hypotheses are"

1) proportions of C are different from proportions of A (=independance)
2) proportions of C are different from proportions of B (=independance)
3) proportions of A are NOT different from proportions of B (=NO independance)

Is there something like post-hoc contrasts for chi-squared, so I can test all my hypotheses in ONE modell and not have alpha inflation?

Intuitively I would test C against a composite of A and B but that still doesnt test if A is different/independent

There is a post on Stackoverflow regarding this topic, but there wasn't an answer.

Community
  • 1
user670186
  • 450
  • 2
  • 16
  • You might be interested in these questions and related pointers: [Multiple Chi-Squared Tests](http://stats.stackexchange.com/q/1133/930), [How to carry out multiple post-hoc chi-square tests on a 2 X 3 table?](http://stats.stackexchange.com/q/9685/930), [Has anyone used the Marascuillo procedure for comparing multiple proportions?](http://stats.stackexchange.com/q/3526/930). – chl Jan 12 '12 at 07:22

1 Answers1

2

The R Package 'pairwise.prop.test()' does the post-hoc job. One can also define several methods of adjusting p-value. However, this is only a post-hoc test and no planned comparisons.

user670186
  • 450
  • 2
  • 16