topic Re: How to get p value for each cell in freq table in SAS Programming

How to get p value for each cell in freq table

Strs1309 — Mon, 30 Sep 2024 10:50:16 GMT

Hello guys! I have got a problem which is really widespread I guess.
How can I output p-value for each cell in total column in proc freq procedure? Is it possible to output p-value for each row in cell in 2x2 frequency table?

Re: How to get p value for each cell in freq table

ballardw — Mon, 30 Sep 2024 10:54:13 GMT

P-values are the result of a statistical test. What test are you running on "each cell"?

Basically a "cell" isn't going to have a p-value at all because you aren't running a test on the cell. Some tests, such as Chi-square allow you to get the contribution to the overall test statistic.

Re: How to get p value for each cell in freq table

Strs1309 — Mon, 30 Sep 2024 11:05:08 GMT

I have only this kind of table (amount of people with different types of blood in 2003 and 2022)
2003 2022 p-value
O(I) 2345 1028 ?
A(II) 3402 4093 ?
B(III) 9807 3988 ?
AB(IV) 1008 1982 ?

Is it possible to get p-value in the last column ?

Re: How to get p value for each cell in freq table

PaigeMiller — Mon, 30 Sep 2024 13:03:42 GMT

Repeating essentially what @ballardw said

If you are talking about a chi-squared test, you would not get a p-value for each row of the table. So I guess you are not talking about a chi-squared test. So please be specific: what test are your performing that will produce a p-value?

Re: How to get p value for each cell in freq table

ballardw — Mon, 30 Sep 2024 14:18:24 GMT

@Strs1309 wrote:
I have only this kind of table (amount of people with different types of blood in 2003 and 2022)
2003 2022 p-value
O(I) 2345 1028 ?
A(II) 3402 4093 ?
B(III) 9807 3988 ?
AB(IV) 1008 1982 ?

Is it possible to get p-value in the last column ?

In your first post you had:

Is it possible to output p-value for each row in cell in 2x2 frequency table?

And then you show a 4x2 table. So is it a 2x2 or 4x2 or yet something else?

Still have not mentioned what the test is supposed to be either.

Re: How to get p value for each cell in freq table

mkeintz — Mon, 30 Sep 2024 23:39:07 GMT

Does it make sense to get a probability for every cell in a crosstabulation. I think you could not treat all those probabilities as independent.

For example, in the case of a 2*2 table, I believe that getting a "p-value" for a single cell will generate the same "p-value" for each of the other cells. After all, given you start with row totals and columns totals, knowing the frequency of any single cell means you can calculate the frequency of any other cell. After all, chi-square for a 2*2 table has degrees of freedom=1.