hIRT: Hierarchical Item Response Theory Models

An R package implementing hierarchical item response theory (IRT) models where both the mean and the variance of latent preferences (ability parameters) may depend on observed covariates (Zhou 2018). The current implementation includes both the two-parameter latent trait model for binary data and the graded response model for ordinal data. Both are fitted via the Expectation-Maximization (EM) algorithm. Asymptotic standard errors are derived from the observed information matrix. Available at https://cran.r-project.org/web/packages/hIRT/