0

I am comparing proportions of different maturity stages (3 levels) between different treatments (4 levels). As I have some values less than 5 in some cells, I decided to use a Fisher exact test instead of a chi square.

After doing the fisher test, I got a significant results. However, when i did the posthoc test, the results were not significant. I used the function: row_wise_fisher_test because I am only interested to know if the proportions of individuals of a same maturity stage are equivalent or different between two treatments.

How come do I get a non-significant results in the posthoc test after a significant fisher test ?

I am confused....what am i doing wrong?

Thanks for your help.

nathmar
  • 31

1 Answers1

1

I don't have enough reputation, so I will write an answer. The reason for this result is very likely the following: the Fisher test uses all the available data, while a post hoc test only analyzes part of the data. The amount of data used in the post hoc test may not be enough to detect a significant effect, while the full data shows a significant effect.

Quertiopler
  • 118
  • Thank you for your reply. I think I found a way to avoid this problem. I did another fisher exact test by comparing 2 treatments at the time and comparing the proportion of 1 stage over the total between two treatments. – nathmar Jan 20 '23 at 15:14