I have 23 samples of milk and 14 samples of cheese. For each of the 7 antibiotics there is a given number of samples that were resistant to it. Note there is only a single strain of bacteria.
Which method should I use to answer these questions?
TLDR: Is (each of) the given antibiotic more effective against cheese bacteria, milk bacteria, or is there no statistical difference at p=0.05?
Another way to put it:
For each antibiotic (and at p=0.05):
- Are bacteria in milk sample more resistant to given antibiotic than bacteria in cheese sample?
- Are bacteria in cheese sample more resistant to given antibiotic than bacteria in milk sample?
- Or is there no statisticaly significant difference in the resistance to the given antibiotic between milk and cheese samples?
I think I only need to count p-values somehow. The problem is the different number of samples. Could I use standardized mean difference (Cohen's d) to overcome this?
Seeing that 5 of the 6 listed antibiotics contain too few observations in the contingency table, one cannot use the chi-square test. When 5 or less observations are present, you have to use the Fisher's exact test instead.
https://www.statology.org/fishers-exact-test-excel/ https://youtu.be/p2c4C2D7ZZc