Multiple imputations object

DESCRIPTION:

data with attached multiple imputations.

GENERATION:

The generator function for this class of objects is miVariable. Multiple imputation functions such as impLoglin, impCgm, and impGauss create objects of this class.

METHODS:

The "miVariable" class of objects has methods for the following generic functions:
, [[, concat, dim, dim<-, dimnames, dimnames<-, data.frameAux, format, , length<-, miPrint, miTrim, numRows, numRows<-, , and . Some additional methods exist which just print an error message, in order to prevent miVariable objects from being accidentally used in calculations directly. Calculations should be carried out using or

STRUCTURE:

An object of class "miVariable" has slots:

VALUE:

Data
the original data (this may have its own class) which generally has missing values,
whichNA
integer vector indicating which elements of the data have multiple imputations (usually the elements which are missing),
Imputations
data frame containing multiple imputations. There are as many rows as there are missing values, with one column for each set of multiple imputations. There should be at least two columns.

DETAILS:

This is a very general class. The Data component may have additional classes and attributes. Atomic objects of any class may be converted into "miVariable" objects without loss of other class or attributes information.

miVariable objects may be converted into miList objects. The reverse is may not be true, e.g. if the elements of the miList object have different length, mode, attributes, or are lists.

The whichNA component governs how rows of Imputations are matched with elements of the original data; the basic rule is that x$Data[x$whichNA] <- x$Imputations[[k]] should replace missing values with the kth set of multiple imputations.

The row names of the data frame need not correspond to the numerical positions of elements to be imputed, and whichNA need not be sorted.

SEE ALSO:

, , , , .