chi square test to test difference for 5x2

#1
My data has two groups people, VV and NV, and I want to test the difference in their occupation. The occupation were 5 groups: farm/fh, farmer/fh, professional, skw, wcollar. I used chi-square test in SAS and the results show there is no difference between the VV and NV groups. The code is as follows:

proc freq data=vvnv;
tables Occupation * status / fisher;
run;





The P-value is >0.05, so there is no difference between the two groups. But when I looked at the table, it seems there are some difference, for example, VV group has only two types of occupation, 50% (10 people) were farmer/fh, and 50% (10 people) were skilled workers. But NV has 5 types of occupation, 55% skilled workers, and 34% farmer/fh, 9% white collar etc. VV has 0 in the three other types.

My question is: the overall 5x2 chi-square test showed no difference, but I break it into 5 of 2x2 chi-square to test each occupation type?
 

hlsmith

Omega Contributor
#2
You ran a fishers exact test which is more applicable given your sample counts. Splitting them up may be an idea, but you would be conducting 5 tests instead of 1, which increases your false discovery rate. Your overall problem here is your small sample size. You are likely under powered to find a difference that could be of interest.