Full Information Maximum Likelihood (FIML) for nested binary outcome analysis with missing data

Haris — Tue, 24 Sep 2019 18:08:05 GMT

Does SAS currently have PROC that can can use Full Information Maximum Likelihood (FIML) to analyze datasets with missing data and binary nested outcomes. I am working with an outcome of mortality in patients nested in hospitals with missing patient-level data. Currently we use GLIMMIX to run a mixed effects model and PROC MI/ANALYZE to impute the data. I know CALIS and MIXED have METHOD=FIML available but previously, SAS could not handle logistic regression.

Thanks for any help.

Re: Full Information Maximum Likelihood (FIML) for nested binary outcome analysis with missing data

Haris — Wed, 25 Sep 2019 22:51:54 GMT

According to SAS tech support, this is still not a feature in SAS. Vote early and often!

topic Full Information Maximum Likelihood (FIML) for nested binary outcome analysis with missing data in Statistical Procedures

Full Information Maximum Likelihood (FIML) for nested binary outcome analysis with missing data

Re: Full Information Maximum Likelihood (FIML) for nested binary outcome analysis with missing data