topic Re: How do I interpret chi square for dichotomous variables? in Statistical Procedures

How do I interpret chi square for dichotomous variables?

LucasL — Fri, 17 Jul 2020 04:34:49 GMT

Hello,

I am struggling with running tests on two dichotomous variables. I don't think I'm doing it correctly and I don't know how to interpret my output. My advisor said to do Chi-Square. When I do Chi square, I end up with info like this:

chi square test for equal proportions

chi square: 703.7583

DF: 1

PR>ChiSq: <.0001

This is my code:

Proc Freq data=filea;

tables Variable1 Variable2

/chisq;

run;

Thank you for any help you can provide!

Re: How do I interpret chi square for dichotomous variables?

SAS_Rob — Fri, 17 Jul 2020 12:19:00 GMT

The way you have it set up now, you will get two separate Chi-Square tests. Each test will test whether the proportions for the two groups are the same. Assuming that the output you posted was for Variable1, it is testing whether P(Variable1=Level 1)=P(Variable1=Level 2).

It might be that your intention is to test the equality of independent proportions, in which case you need to cross the two variables on the TABLES statement.

tables Variable1*Variable2/chisq;

Re: How do I interpret chi square for dichotomous variables?

Ksharp — Fri, 17 Jul 2020 12:38:26 GMT

What kind of test do you want to do ?
If it is two way contingency table ,you need symbol '*' .

Proc Freq data=filea;
tables Variable1*Variable2
/chisq fisher agree ;
run;