0

I am trying to compare how often members from two groups of turtles (adults vs juveniles), fed in 1 of 3 locations (sea floor, free-floating, or surface).

I don't think I can use a chi-squared or Fisher's as the turtles could have fed in multiple locations in the same 'test'.e.g 10 times on the sea floor, 2 times in free floating, and 0 at the surface.

So far, I have come the the conclusion that I can maybe work out the % frequency for each of the 3 locations for each group (adults n=9 vs juveniles n=5). Is there a test that would allow me to do this?

Alternatively, is it possible to work out the % frequency for each turtle (so each turtle had a total of 100% of their feeding events recorded across all 3 locations), and then perhaps compared the groups in a rank sum test and see if their was any relationship between the groups this way?

Or is this possible with chi-square/ Fishers? Many thanks

RMM
  • 1
  • Fisher's exact test can be applied here. You have a 2x3 contingency table, and Fisher's test tells you whether the adult and juvenile are significantly different. – frank Jul 23 '22 at 09:45
  • Hi and thank you for the feedback. Is it still statistically valid to use the 'raw counts' for each time they were recorded feeding in a particular location when each of the turtles was potential feeding in all three of the locations in the same test? As opposed to a percentage frequency for each group together. I have seen that a chi-squared should only be used on raw counts and not percentages before you compute the statistic. But I have also not been able to find any example where they do not have a mutually exclusive 'choice' from the predictor variable. – RMM Jul 24 '22 at 12:50
  • I.e. they can only choose one of the options of the dependent variable levels...Is the chi-squared or Fishers ok for a percentage frequency analysis of difference between two groups? – RMM Jul 24 '22 at 12:50

0 Answers0