ABSTRACT
Modeling effects in meta-analysis of trials is of interest and has been under investigation for quite some time. For an overview see Agresti and Hartzel (2000). In this section we would like to follow a strict likelihood approach. In the likelihood theory, when dealing with nuisance parameters, one approach (see Murphy and Van der Vaart (2000) or Pawitan (2001)) uses the profile likelihood method to eliminate the nuisance parameters. We will show in this section that this approach for a MAIPD leads to a very simple and easy to handle profile likelihood which can be used in more elaborate developments later on.
