Statistical analysis of non-numerical data

Question

[aminijavad4]

Hi In one study for Alzheimer's, I made a few non-numerical observations (like below: T,S,M,D). These observations have been seen in both patients and healthy individuals. But the number of times is different. The only thing I could do was the measure percentage of each observation in Alzheimer and healthy. But I do not know how to use statistical methods like t-test for evaluation association of each observation with Alzheimer's disease.

Alzheimer's disease: M,D,S,T,S,T,D,S Heathy: D,S,M,D,S,S,S,D

I appreciate if anybody share his/her comment with me.

Thanks in advance.

[aminijavad4]

Thanks for your help Kevin

Answer 1

[Kevin Blighe]

You could use a Fisher Test to test the hypothesis that the frequency of each observation does not differ between Alzheimer's and Healthy.

In R, that is fisher.test(). The data would look something like:

              M   D   S   T
Alzheimer    67  54   2   5
Healthy       4  15  16  20

Kevin