How to interpret Odds Ratio in Proc Glimmix with interaction

LiHZ — Mon, 12 Apr 2021 09:25:14 GMT

Hi everyone,

Please bear with me. I'm confused to understand Odds Ratio in Proc Glimmix with interaction.

In Proc Logistic, if there is an interaction between factor A(two levels) and B(levels ≥2), Odds ratio of A will be presented in different level of B. In a word, there are different OR of A under different level of B.

However, in Proc Glimmix, only one OR is presented even if there is an interaction.

Here is my code are result.

proc glimmix data=mas ;
where farm ne 'hy';
class season(ref='summer') treatment parity(ref='1') result farm year ;
model result(event='pregnant')= season | parity | treatment /
dist=binary oddsratio ;
random farm year;
run;

As you can see, there is an interaction between treatment and parity or season.

However, OR of treatment is 1.509.

Why there is a difference between proc logistic and proc glimmix in the matter of OR?

Thanks in advance for the help!

Re: How to interpret Odds Ratio in Proc Glimmix with interaction

ChrisNZ — Mon, 19 Apr 2021 05:31:31 GMT

Moved to stats community so maybe you get an answer.

Re: How to interpret Odds Ratio in Proc Glimmix with interaction

LiHZ — Wed, 21 Apr 2021 08:04:07 GMT

Hope this may help whom have the same problem.

https://support.sas.com/kb/24/455.html

topic Re: How to interpret Odds Ratio in Proc Glimmix with interaction in Statistical Procedures

How to interpret Odds Ratio in Proc Glimmix with interaction

Re: How to interpret Odds Ratio in Proc Glimmix with interaction

Re: How to interpret Odds Ratio in Proc Glimmix with interaction