an object of class
"priorLoglin" containing the type of the prior
and other information, see
. The prior type may be one
of
"ml",
"noninformative", or
"data.dependent".
OPTIONAL ARGUMENTS:
algorithm
character, either
"em" for the EM algorithm, or
"da" for data augmentation.
object
an object of class
"preLoglin" used for generating data dependent priors.
VALUE:
for
algorithm == "da", if the prior type is "ml",
check.Loglin.prior assigns a prior of 0 to each cell in the table,
while if prior type is "noninformative" then a prior of 0.5 is
assigned to each cell.
For
algorithm == "em",
check.Loglin.prior assigns a prior of 1 to
each cell in the table with both the maximum likelihood and the
"noninformative" prior.
When a data dependent prior is used,
check.Loglin.prior computes the
marginal probabilities for the observed data using the input
"preLoglin" object. These marginal probabilities are used to
compute the cell probabilities. The data dependent prior is the
product of the
nobs value found in the
"priorLoglin" object, and
these cell probabilities.