Below is an example of frequency data between observed and expected proportions. Can I simply run a 2-Proportion z test here? Or, would I be violating assumptions? If I were to run dozens of z tests (i.e. O-E cell comparisons), would I have to correct for multiple comparisons? Is there a better way to do this?

Technically I have 5 different sets of expected proportions -- I am just trying to determine which is most predictive of the observed data.


enter image description here

| cite | improve this question | | | | |
  • $\begingroup$ Would a better solution be to perform a chi-square test on the whole contingency table? $\endgroup$ – R Greg Stacey Mar 16 '16 at 20:10
  • $\begingroup$ I suppose I could do a goodness of fit for the whole table, but the residuals (adjusted) would be the equivalent of a z-test, no? Also, the proportions are so small for some of the cells, that I would be well below the minimum acceptable frequency. $\endgroup$ – D. Bruce Mar 16 '16 at 20:19
  • $\begingroup$ I'm not the best person to answer this, but you could try using referring to this question and this question. $\endgroup$ – R Greg Stacey Mar 16 '16 at 20:46

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.