I have the a dataset like the following:
| Experiment | A | B | C | Group |
|---|---|---|---|---|
| 1 | 10 | 1 | 3 | M |
| 2 | 30 | 10 | 10 | F |
| 3 | 20 | 30 | 10 | M |
| 4 | 1 | 15 | 4 | F |
And I am looking for differences in the proportions of A, B and C between the groups M and F. I am not sure what test should I use.
Should I sum across the columns A, B and C to obtain a contingency table such as:
| A | B | C | |
|---|---|---|---|
| M | 30 | 31 | 13 |
| F | 31 | 25 | 14 |
and perform a chi squared test?
You should do in R chisq.test(data) where the data is second table where you have 2 random variable one being gender and the other is having 3 classes A, B and C, what ever it is : say year. The degrees of freedom will be 2 for your case.
Of course, I assume you need a Test for independence, and Chi Squared distribution is used for Test for independence, because De Moivre theorem.
