Contingency Table Analysis by chi square problem

I am trying to figure out a chi square problem, and I keep getting stuck. The problem reads: A researcher tested single mosquitoes from five different locations for the presence of a particular virus and obtained the following data:

Location A B C D E
w/ virus 3 7 9 11 3
w/o virus 26 32 21 16 30

First I performed a chi square test about found that the data indicate there are significant differences between the locations.
Specifically, I found that my test statistic, T=12.88, X^2=9.488, and my p-value=0.01187.
However, I have no idea what those differences are. I don't know how to describe the patter of differences/determine which locations are significantly different from each other. I looked into doing a Tukey's test but it doesn't really make any sense with this problem (I think). I don't know if it makes sense to just do a bunch of separate chi square tests between all the variables. Any help would be much appreciated!


Less is more. Stay pure. Stay poor.
I used to do bunch of pairwise comparisons (Chi squares) and correct for false discovery - but I know some people use the standardize residuals from the original omnibus Chi square.

Side note Chi square is usually frowned upon if any cell is 5 or less. With Fishers exact test being used instead.


TS Contributor
Indeed, Tukey's test is for comparison of means (interval scaled data), not for categorical data.

I suppose you meant expected cell frequencies, not actual cell frequencies? I suppose that all expected frequencies are > 5 here.

Wth kind regards



Less is more. Stay pure. Stay poor.
Yes, good point Karabiner - I should have said expected frequencies.

Also, with the OP providing data we can run the analyses. One cell had an expectancy of 5.6, so it is up to them to interpret it and make the next step as they see fit.