topic Re: Fishers Exact Test Interpretation in Discussions

Fishers Exact Test Interpretation

MedianPuppy4982 — Tue, 09 Apr 2024 13:49:13 GMT

Hello! I am analyzing an experiment that I conducted which was made using JMP DOE. I am interested in using the Fishers Exact Test results (pictured below). I was doing some reading on the JMP help site and it says that values in the tables are not to be considered as P-Values, can these values be taken has regular probability values or is there another way to interpret the data.

Re: Fishers Exact Test Interpretation

Mark_Bailey — Tue, 09 Apr 2024 14:26:00 GMT

They are p-values. Can you send me a link to the page in the documentation from which you interpreted otherwise?

What is your alternative hypothesis in the experiment? Isi it two-sided or one-sided?

Re: Fishers Exact Test Interpretation

MRB3855 — Wed, 10 Apr 2024 14:10:42 GMT

Hi @MedianPuppy4982 and @Mark_Bailey : I see where the confusion is coming from.

Here: Fisher’s Exact Test Report (jmp.com)

Therein, it says:

"The Fisher’s Exact Test report includes the following information:

Table Probability (P)

The probability for the observed table. This is not the p-value for the test.

Two-sided Prob ≤ P

The significance probability (p-value) for the two-sided test."

The above is true for the general r x c table case as in my example screenshot below (I have a binary response, and a three level factor) where my table is 3 rows by 2 columns.

But, in the 2 x 2 case (like @MedianPuppy4982 has) it gives p-values only as @MedianPuppy4982 output shows. .

Re: Fishers Exact Test Interpretation

MedianPuppy4982 — Thu, 11 Apr 2024 16:55:09 GMT

Hi @MRB3855!

Thank you so much for your help! Do we know why that is? or is it just the general rule for 2x2 tables for the Fishers Exact Test?

Re: Fishers Exact Test Interpretation

MRB3855 — Thu, 11 Apr 2024 20:40:12 GMT

Good question @MedianPuppy4982 . My guess is (JMP folks may want to jump in here) that with the 2x2 case there are more direct ways (e.g., based on the binomial distribution) to calculate p-values. In the general rxc case, the p-value is calculated as the sum the probabilities from all tables (with the same row and column totals) having a probability less than or equal the probability of the observed table.

Re: Fishers Exact Test Interpretation

ParkerGordon — Thu, 09 May 2024 12:29:39 GMT

Thank you so much for the solution.