Fit linear models, estimating score distributions for groups of people, following Cohen and Jiang (1999) <doi:10.2307/2669917>. In this model, the response is a latent trait (such as student ability) and raw item responses are combined with item difficulties in an item response theory (IRT) framework to form a density for each unit (student). This latent trait is then integrated out. This software is intended to fit the same models as the existing software 'AM' <http://am.air.org/>.