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

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


A: success = 0.085 n=50
B: Success = 0.167 n=50
C: Success = 0.380 n=50

If I do post-hoc contrasts using pairwise.prop.test() then I sacrifice power because there is no need to test C against A, especially given C against B is already significant.

The pairwise.prop.test() function provides p-value adjustment such as bonferroni, holm or BH, BY (see R documentation for p.adjust {stats} ).

  1. Which of these adjustments is most appropriate if choosing to do all post-hoc comparisons?
  2. Would it be OK to two conduct 2 Chi-squared tests and test B against C and A against B and double the p-values?


  • $\begingroup$ This is just a variation of your previous question at stats.stackexchange.com/questions/21061/…. Please combine them and indicate which one you would prefer to be closed. $\endgroup$ – whuber Jan 13 '12 at 21:34
  • 1
    $\begingroup$ I deleted the other posts. Also this is different because I am asking about plannned comparisons, not posthoc comparisons. Also, I am asking about right adjustment question 2 is about an alternative... $\endgroup$ – user670186 Jan 14 '12 at 20:22

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.