topic Re: Mianalyze on GLM contrasts ? in Statistical Procedures

Mianalyze on GLM contrasts ?

Giampaolo — Fri, 24 Sep 2021 16:25:41 GMT

Hi,

I would like to pool the results of contrasts statements after multiple imputation. Is there a way to do it?

Thank you

My code is the following:

proc glm data = MI_dataset outstat= stat;
by _imputation_;
class &catvar;
model iNPI=&catvar;
contrast 'linear' &catvar -3 -1 1 3;
contrast 'quadratic' &catvar 1 -1 -1 1;
contrast 'cubic' &catvar -1 3 -3 1;
run;quit;

Re: Mianalyze on GLM contrasts ?

SAS_Rob — Fri, 24 Sep 2021 17:22:00 GMT

Because MIANALYZE only works and point estimates and their standard errors, you would need to convert these to ESTIMATE statements.

See the example at the bottom of this usage note for how to combine the ESTIMATE statements.

https://support.sas.com/kb/48/700.html

Re: Mianalyze on GLM contrasts ?

Giampaolo — Fri, 24 Sep 2021 17:38:22 GMT

Thank you!

Is it possible then to use the estimate statement, instead of the contrast statement, to test for a trend ? As in my code obtained from https://support.sas.com/kb/22/912.html?

Re: Mianalyze on GLM contrasts ?

Saifulinfs — Fri, 04 Aug 2023 08:08:57 GMT

Hello, I am also facing the same issue; testing for trend with multiply imputed data. How did you solve that?