This is a simple latent class model used for illustrating MCMC
calibration in Chapter 9 of Almond et al., <i>Bayesian Networks in
Educational Assessment</i>.  There are two binary items and a single
item with three levels.  This directory contains the input to the
calibration using StatShop.
</p>
<p>
Note that anerror was found in the prior model where the probability
and parameter tables were swapped for the level3 evidence model.  If
you have difficulty reproducting this table try change the prior to
have row sums of one (the file level3orig.gm.xml gives this value).