1

I am running a Fisher's exact test on SPSS. There are 2 variables - 3 groups and their Pass/Fail frequencies:

Group 1: 8 pass, 18 fail

Group 2: 1 pass, 10 fail

Group 3: 11 pass, 6 fail

I am using Fisher's exact test because some cells have a count less than 5. I have a significant result (p=.01) but I don't know how to run post hoc tests for this analysis. Should I rerun the test but using pairs of groups - then adjust my p-value using Bonferroni correction?

Any suggestions would help!

Statsyy
  • 11
  • 3
  • "some cells have a count less than 5" -- the common (and sometimes overly strong) rule is not about the observed counts but rather about the expected counts. – Glen_b Mar 19 '23 at 21:45
  • Fair enough - still my data has counts less than 5 for expected counts. I know that 5 is overly strong... I did read one source that said the minimum count should be between 1 and 2, but most sources say to be conservative. I opted for the conservative approach and got significant results - now, the question is how do I find out where the differences are? – Statsyy Mar 19 '23 at 22:34
  • https://stats.stackexchange.com/search?q=Fisher+exact+post+hoc – Glen_b Mar 19 '23 at 23:00

0 Answers0