Composite Likelihood for probit latent variable models

Source: R/complik.R

Estimate parameters in a probit latent variable model via a composite likelihood decomposition.

complik(
  x,
  data,
  k = 2,
  type = c("all", "nearest"),
  pairlist,
  messages = 0,
  estimator = "normal",
  quick = FALSE,
  ...
)

Arguments

x

lvm-object

data

data.frame

k

Size of composite groups

type

Determines number of groups. With type="nearest" (default) only neighboring items will be grouped, e.g. for k=2 (y1,y2),(y2,y3),... With type="all" all combinations of size k are included

pairlist

A list of indices specifying the composite groups. Optional argument which overrides k and type but gives complete flexibility in the specification of the composite likelihood

messages

Control amount of messages printed

estimator

Model (pseudo-likelihood) to use for the pairs/groups

quick

If TRUE the parameter estimates are calculated but all additional information such as standard errors are skipped

...

Additional arguments parsed on to lower-level functions

Value

An object of class estimate.complik inheriting methods from lvm

See also

estimate

Author

Klaus K. Holst

Examples

m <- lvm(c(y1,y2,y3)~b*x+1*u[0],latent=~u)
ordinal(m,K=2) <- ~y1+y2+y3
d <- sim(m,50,seed=1)
if (requireNamespace("mets", quietly=TRUE)) {
   e1 <- complik(m,d,control=list(trace=1),type="all")
}