topic Re: Error with mixed effect model in Statistical Procedures

Error with mixed effect model

rajrao — Tue, 17 Dec 2019 20:53:26 GMT

Hi all,

I am trying to use a mixed effects model with a random intercept for a project, but I am not sure my output is correct. My code is as follows:

proc mixed data=pf;

class timepoint(ref="0") tx_arm(ref="Tx");

model TOTAL_SCORE=timepoint tx_arm timepoint*tx_arm/ solution;

repeated timepoint/ sub=username(tx_arm) type=un r rcorr;

run;

I keep getting an error message in the output as follows:

Convergence criteria met but final hessian is not positive definite.

Can someone tell me where I am going wrong? Thank you.

szm — Wed, 18 Dec 2019 17:52:52 GMT

This can be an indication of too complicated model. Try to change a covariance structure and use a more simpler structure than UN.

rajrao — Thu, 19 Dec 2019 14:25:49 GMT

Okay, thank you for your response!