Compute an Anova Table for a Multiple Imputations Object - Generic function

DESCRIPTION:

Produces anova summaries; these may be objects of class anova, which inherits from data.frame, or miList objects whose components are of class anova.

miAnovaAux is generic (see Methods); method functions can be written to handle specific classes of data. Classes which already have methods include: (none yet).

USAGE:

miAnova(object, ...) 
miAnovaAux(object1, object, ...) 

REQUIRED ARGUMENTS:

object
an mi object.
object1
this must be the first imputation from object, i.e. miSubscript(object, 1).

OPTIONAL ARGUMENTS:

...
optional arguments to be passed to methods for miAnovaAux, and (depending on the method) ultimately to anova.

VALUE:

in the default case, an miList whose components are obtained by calling anova for each imputation.

DETAILS:

miAnova checks that all imputations have the same class, then calls miAnovaAux, which dispatches based on that class. The appropriate method to use in combining multiple imputations usually depends not on the class of object but rather on the class of the imputations of object. For example, suppose that object is an miList object whose imputations are lm objects; this generic function then dispatches to miAnovaAux.lm (if it exists), otherwise to miAnovaAux.default.

Methods for miAnovaAux can be written to provide appropriate summaries which consolidate information across multiple imputations. Use getMethods("miAnovaAux") to see what methods have been defined (currently none, because it is unclear how to properly combine Anova tables across imputations).

If no method is found, then control passes to miAnovaAux.default , which calls anova for each imputation, but does not consolidate information across imputations.

REFERENCES:

Schafer, J. L. (1997), Analysis of Incomplete Multivariate Data , Chapman & Hall, London.

SEE ALSO:

, , .

EXAMPLES:

fit <- miEval(lm(chol14~., data = cholesterolImpExample)) 
miAnova(fit)