Package: birt 0.2.0

birt: BIvariate Response Time Modeling with 'lavaan'-Style Syntax

Fits item response theory models with one or two latent variables, for example ability and speed in joint models of response accuracy and response time (van der Linden, 2007, <doi:10.1007/s11336-006-1478-z>), from 'lavaan'-style model syntax. The main engine fits the syntax in 'RTMB' by approximate Bayesian inference as an extended latent Gaussian model (Stringer, Brown and Stafford, 2023, <doi:10.1080/10618600.2022.2099403>): the persons are integrated out by adaptive Gauss-Hermite quadrature and the remaining parameters by an outer adaptive quadrature, with marginal likelihoods for model comparison; the same engine gives maximum likelihood, with casewise scores for score-based invariance tests and DIF trees. Ordinal and continuous indicators, moderated nonlinear factor analysis statements, sinh-arcsinh residuals (Jones and Pewsey, 2009, <doi:10.1093/biomet/asp053>) whose quantile effects at every level come from one fit, conditional reliability and the information of each quantile regression of the response times. An optional 'JAGS' engine translates the model into a plain 'JAGS' model file, for spike-and-slab priors and asymmetric Laplace (quantile) distributions, with marginal-likelihood fit indices (DIC, WAIC, PSIS-LOO) and posterior predictive checks.

Authors:Jie-Wen Tsai [aut, cre]