r-pedmod

Provides functions to estimate mixed probit models using, for instance, pedigree data like in <doi:10.1002/sim.1603>. The models are also commonly called liability threshold models. The approximation is based on direct log marginal likelihood approximations like the randomized Quasi-Monte Carlo suggested by <doi:10.1198/106186002394> with a similar procedure to approximate the derivatives. The minimax tilting method suggested by <doi:10.1111/rssb.12162> is also supported. Graph-based methods are also provided that can be used to simplify pedigrees.