chi-squared statistic for incomplete categorical data

DESCRIPTION:

Calculates the observed-data chi-squared statistic (versus the saturated model) for incomplete categorical data under a specified value of the underlying cell probabilities, e.g. one row of the paramIter component of a missmodel object.

USAGE:

chi.Loglin(object, theta) 

REQUIRED ARGUMENTS:

object
an object of class "missmodel" whose paramIter component is of class "Loglin". I.e., an object resulting from the use of the mdLoglin, daLoglin, or emLoglin functions. Alternatively, a class "preLoglin" object.

OPTIONAL ARGUMENTS:

theta
the model cell probabilities. If not specified, these are obtained as the final estimates (the last row) in the "missmodel" object.

VALUE:

the value of the observed-data chi-squared statistic at theta.

DETAILS:

This is the chi-squared statistic that ignores the missing-data mechanism.

SEE ALSO:

, , , , .

EXAMPLES:

aa <- mdLoglin(object = crime, margins = ~Visit.1+Visit.2,
               na.proc = "em") 
chi.Loglin(aa)