Wilcoxon rank sum test with binary data

[Friday12]

Hello Everybody,

Can the Wilcoxon rank sum test be used when the variable of interest is binary? For example, say I have a control group and a treatment group, and I am interested in seeing whether the treatment group goes to college more than the control group. The variable "college" takes on a value of 1 if the subject goes to college and 0 if the subject does not. Can I use the Wilcoxon rank sum test to discern whether the two groups are drawn from the same distribution?

It seems that when ranking the observations, most would be ties, and this might be problematic for the Wilcoxon rank sum test.

If not the Wilcoxon rank sum test, what nonparametric method would be appropriate for this situation?

Thanks!

 

[Dason]

What's wrong with a test of proportions?

 

[Friday12]

Test of proportions would be most appropriate in most of these situations, I was just wondering whether there is something inherently wrong with using the Wilcox rank sum test with binary data.

After doing a bit more research I think the Fischer test would be more appropriate as a nonparametric method for binary data