Help with Contingency Table Setup

Eco22

New Member
#1
I need help figuring out if I am setting up a contingency table for a ChiSq test correctly.

I did surveys for a frog species at 124 sites in 2013 and 2018. I detected the species at 49 sites in 2013 and 60 sites in 2018. I need to determine if the increase in sites is statistically significant - when I run the following table through a ChiSq calculator -- it comes up with p<.00001 -- but it seems a little off that an increase of just 11 sites would be such a significant change. Thanks for any help...

Contingency Table, with number of sites frog was present in 2013 only, 2018 only, and both years:

1614804364392.png
 

Attachments

katxt

Active Member
#2
Your table doesn't quite match your description (as far as I can see) but I agree that the p value is unrealistically small. Chi square isn't suitable here. You want to compare the totals between the years and things are complicated by the fact that these totals both include a common group and so are not independent. It's sort of like a paired test. McNemar's test comes to mind but I'm not sure that is quite right here. Somebody will confirm or deny. I personally would do a randomization test which eliminates all the uncertainty. My feeling is that the difference may be significant but not by much.