Hadfield (2010) talks about "parameter expanded models" (page 11). I never heard of these models before and wonder what he means by this. In the Appendix E he writes more about "Parameter expansion" and cites Liu, Rubin and Wu (1998). As far as I grasped this paper they add parameters to their EM-algorithm models to speed up convergence.
My question is: what are "parameter expanded models" (in the context of generalized linear mixed models)?