A Rasch model is a generalized linear model, a type of item-response model in which the probability of a 'positive' output is distributed according to a logistic function of the 'skill' and 'difficulty' of the item and recipient, respectively.

This code is an experiment with building a hierarchical generalized linear version of the Rasch model (a la Gelman) and then evaluating it through a sampler that uses slice-sampling as a sub-routine.

<http://bayes.wordpress.com/2009/03/12/still-we-are-nowhere/>
<http://en.wikipedia.org/wiki/Rasch_model>