I completely agree that the development of full-information maximum likelihood (FIML) estimation for use in packages like lm, lme, lmer, etc. would make R much more attractive. FIML is better than other approaches to missing data (e.g., multiple imputation; see Graham, Olchowski, & Gilreath, 2007), and it is much better than what R currently uses for most functions (listwise deletion). Right now, the best implementation of FIML is in Mplus (http://statmodel.com/), which is useful in the structural equation model (SEM) framework. There is an R package that incorporates FIML in SEM called OpenMx (http://openmx.psyc.virginia.edu/openmx-features), in which one could run multiple regression with FIML. It would be nice to see FIML estimation incorporated into other functions, as well though, as FIML seems to be the way to go. I second the call to get the R community talking about FIML.
Thank you Andrew for starting this -- it's long overdue. Hopefully, other people are interested, as well. -- View this message in context: http://r.789695.n4.nabble.com/FIML-in-R-tp4515074p4538436.html Sent from the R help mailing list archive at Nabble.com. ______________________________________________ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.