Now doesn't propose supplement swaps, but shows list of all available items sorted by best to worst mismatch.

Using BRMS to run IRT models on Wordbank data

Fitting model to developmental cross-situational data from Smith & Yu (2008), Yu & Smith (2011), and testing generalization to Vlach & DeBrock (2017 / 2019).

Examining the false positive rate with multiple binomial tests.

We now first prune ill-fitting items before running the CAT simulations.

Determine Bayesian regression power to detect a given effect size with a given sample size.