This function computes posterior summary statistics for model parameters stored in an object containing MCMC samples. It supports different parameter types (e.g., loadings, factor correlations, residual covariances) and automatically constructs index labels based on the parameter structure.
For each parameter element, the function computes: - Posterior mean or median - Posterior standard deviation - Highest Posterior Density (HPD) interval - Significance indicator (HPD interval excluding zero) - Potential scale reduction factor (PSRF)
stat(
classname,
varName,
J = NULL,
K = NULL,
P = NULL,
Q = NULL,
Qb = NULL,
LD = F,
med = FALSE,
start = 0,
end = -1,
SL = 0.05,
sig = F
)An object containing MCMC samples and a
getVariable() method to extract parameter chains.
Character string indicating the parameter name
Number of items.
Number of latent factors.
Number of predictors.
Loading pattern matrix for factor loadings.
Loading pattern matrix for structure coefficients.
Logical. Whether local dependence structure is included.
Logical. If TRUE, compute posterior median instead of mean.
Starting iteration for posterior summary.
Ending iteration for posterior summary.
Significance level for HPD interval (default 0.05).
Logical. If TRUE, return only significant parameters.
A matrix containing posterior summaries for each parameter element, including estimates, standard deviations, HPD interval bounds, significance indicator, PSRF, and corresponding index labels.
Let theta denote a model parameter. Posterior summaries are computed from MCMC samples theta^(s) as:
Mean: E(theta | data) SD: sd(theta | data) HPD interval: 100(1 - SL)
A parameter is considered significant if its HPD interval does not contain zero.
Index positions are automatically constructed depending on the parameter type: A: factor loadings C: factor correlation elements V: residual covariance elements