How do I perform Fisher's exact post-hoc pairwise comparisons?

[patrick5]

I have a 3 x 2 comparison with expected frequencies low enough to make Fisher's exact preferred over chi-square.  I'm trying to figure out how to output the Fisher's exact p-values for the pairwise comparisons to feed into PROC MULTTEST.  I found and modified an example from SAS documentation that works for chi-square, but can't figure out how to make it work for Fisher's (see below).  I assume I need to replace anything that refers to chi-square in the first line, but can't seem to make it work.  The data are structured as one line per subject:

subjectID  species  examtype

12345      CANINE         1

12346      FELINE          2

12347      EQUINE         1

.... 

And the output I'm looking for looks like this (copied from running the below using chi-square in SAS 9.4):

ODS OUTPUT chisq(PERSIST)=chisq(WHERE=(statistic="Chi-Square") rename=(Prob=Raw_P));
      PROC FREQ DATA= six;
         WHERE species in ('CANINE', 'FELINE');
         TABLES examtype * species / FISHER;
		 RUN;
	  PROC FREQ DATA= six;
         WHERE species in ('CANINE', 'EQUINE');
         TABLES examtype * species / FISHER;
		 RUN;
      PROC FREQ DATA= six;
         WHERE species in ('FELINE', 'EQUINE');
         TABLES examtype * species / FISHER;
		 RUN;
ODS OUTPUT CLEAR;

      PROC PRINT NOOBS;
         VAR value raw_p;
         RUN;

      PROC MULTTEST PDATA=chisq BON;
	  RUN;

Thanks in advance for any help!

[FreelanceReinh · Accepted solution]

Hello @patrick5 and welcome to the SAS Support Communities!

The ODS output datasets for chi-square and Fisher's exact test are structured differently. This should work:

ods output fishersexact(persist)=fisher(where=(name1='XP2_FISH') rename=(nvalue1=raw_p));

Dataset fisher will not contain a variable value, so please delete this from your PROC PRINT step.