convergence.dalmatian: Convergence

Description

Compute convergence diagnostics for a dalmatian object.

Usage

# S3 method for dalmatian
convergence(
  object,
  pars = NULL,
  nstart = start(object$coda),
  nend = end(object$coda),
  nthin = coda::thin(object$coda),
  raftery = NULL,
  ...
)

Arguments

object

Object of class dalmatian created by dalmatian().

pars

List of parameters to assess. If NULL (default) then diagnostics are computed for the fixed effects and random effects standard deviations in the mean, dispersion, and joint components.

nstart

Start point for computing summary statistics (relative to true start of chain).

nend

End point for computing summary statistics (relative to true start of chain).

nthin

Thinning factor for computing summary statistics (relative to full chain and not previously thinned output).

raftery

List of arguments to be passed to raftery.diag(). Any values not provided will be set to their defaults (see help(raftery.diag()) for details).

...

Ignored

Value

List containing Gelman-Rubin and Raftery convergence diagnostics and effective sample sizes for the selected parameters. This information is used to diagnose convergence of the MCMC sampling algorithms.

References

Bonner, S., Kim, H., Westneat, D., Mutzel, A., Wright, J., and Schofield, M.. (2021). dalmatian: A Package for Fitting Double Hierarchical Linear Models in R via JAGS and nimble. Journal of Statistical Software, 100, 10, 1--25. 10.18637/jss.v100.i10.