sea.bclca.modelselect {ldsa}R Documentation

Select an Entailment Model Using Bayesian Confirmatory Latent Class Analysis


Fits BCLCA models for the entailment structures associated with a series of breakpoints, and returns various fit statistics in each case.


sea.bclca.modelselect(x, draws = 1000, burnin = 500, 
    measure = c("rate", "z-score", "binom.p"), 
    min.thresh = NULL, max.thresh = NULL, 
    sea.table.precomp = NULL, 
    em.prior = sapply(c(1, 9), rep, length(x)), 
    ep.prior = sapply(c(1, 9), rep, length(x)), 
    phi = 1, err.model = c("nonperverse", "beta"), ...)


x a data.frame with observations on rows; missing data is permitted.
draws number of MCMC draws to take from the posterior of each model.
burnin number of burn-in draws to take (and discard) for each model.
measure error measure to use; rate for absolute error rates, z-score for z-scores, or binom.p for p-values under an exact Binomial test.
min.thresh minimum error threshold at which to begin fitting.
max.thresh maximum error threshold at which models may be fit.
sea.table.precomp an object of class sea.table, computed on x (optional).
em.prior priors for the false negative rates (if err.model=="beta").
ep.prior priors for the false positive rates (if err.model=="beta").
phi priors for the latent class probabilities.
err.model error model to employ. Set to "nonperverse" for uniform non-perverse priors, or "beta" for independent beta priors.
... additional arguments to bclca.


sea.bclca.modelselect is intended to facilitate model selection by taking posterior draws for a BCLCA associated with the entailment structure generated by each of a series of breakpoints (see sea.classbybreak). Various diagnostics and fit statistics are returned, which may prove useful in selecting an optimal model.


A matrix of fit statistics.


Carter T. Butts


~put references to the literature/web site here ~

See Also

bclca, sea.classbybreak, sea.entailment, sea.table

[Package ldsa version 0.1-2 Index]