<br>I have been flirting with generalized mixed effects models. It is said that if you model, for example, a binomial (logit) link, and use the persons as random effects and the items as fixed effects, this is the equivalent of a Rasch model. Indeed, using the R package and the lme4 program, the results are equivalent to standard rasch software e.g. BigSteps. Leaving aside the fact that it is not yet possible for lme4 to build a kidmap or an item-hierarchy plot and the like, are there fundamental theoretical reasons for experienced and well trained Raschers like me not to migrate to R package? I am seduced by the fact that I can fit multilevel Rasch models with explanatory fixed effects. Please give me your opinion since you are substantially more experienced than me.
<br>P.S. (of course I know that Winsteps and the other Rasch-specific packages offer an amazing array of facilities and they are user friendly, but here, I refer to a well-trained specialist, not to beginners)
<br><br><br><br><font color=# style='font-size: 9pt;'>Dr. Iasonas Lamprianou
<br>CFAS, School of Education
<br>The University of Manchester
<br>Oxford Road, Manchester M13 9PL, UK
<br>Tel. 0044 161 275 3485