I have fitted a null mixed model to my binary outcome variable for 96 groups. I have included SAS and R codes and outcomes. The problem for me is that the outcomes from two software do not match. I will appreciate it very much for your insight. Thanks

SAS code: format db db.; run;

ods output fitstatistics=fitNull;

title "Model Null,No covariate"; proc glimmix data=data noclprint; class group ; model db = / solution link=logit dist=binary cl ; random intercept / type=un subject=group; format db db.; weight WEIGHT; run;

Output: Fit Statistics -2 Res Log Pseudo-Likelihood 26057026 Generalized Chi-Square 26088235 Gener. Chi-Square / DF 237.31

Covariance Parameter Estimates Cov Parm Subject Estimate Standard Error UN(1,1) Group 0.08987 0.01299

Solutions for Fixed Effects Effect Estimate StandardError DF t Value Pr >|t| Alpha Lower Upper Intercept -2.5054 0.03046 96 -82.26 <.0001 0.05 -2.5659 -2.4450

R code: MOdelNull <- glmer(db ~ 1 +(1|Group),data=data,family = binomial,weights=WEIGHT)

Outcome: AIC BIC logLik deviance df.resid 25118593 25118612 -12559294 25118589 109930

Random effects: Groups Name Variance Std.Dev. Group (Intercept) 0.06343 0.2519
Number of obs: 109932, groups: Group, 97

Fixed effects: Estimate Std. Error z value Pr(>|z|)
(Intercept) -1.35310 0.02538 -53.31 <2e-16 ***


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.