MImix (version 1.0)

MImix: Multiple imputation summaries via mixture of normals

Description

Combines results of analyses on multiply imputed data sets using a mixture of normal distributions.

Usage

MImix(results, ...)
"MImix"(results,variances,weights = 1/length(results), percentiles = c(0.025, 0.5, 0.975), ...)

Arguments

results

A list of results from inference on separate imputed datasets

variances

If results is a list of parameter vectors, variances should be the corresponding variance-covariance matrices

weights

A vector of weights for each imputed dataset. The default is to use equal weights.

percentiles

A vector of percentiles to be returned from the mixture summary distribution. The default is to return the 2.5th, 50th, and 97.5th percentiles.

...

Other arguments, not used

Value

An list containing the desired percentiles from the mixture summary distribution.

Details

This function combines results of analyses on multiply imputed data sets using a mixture of normal distributions according to the approach described in Steele, R.J., Wang, N., and Raftery A.E. (2009). This package contains a generic function default method, although other methods may be available in future releases. The results argument in the default method may be either a list of parameter vectors or a list of objects that have coef and vcov methods. In the former case a list of variance-covariance matrices must be supplied as the second argument. This corresponds to the same structure that is used by MIcombine in the mitools package.

References

Steele, R.J., Wang N., and Raftery, A.E. (2009) Inference from Multiple Imputation for Missing Data Using Mixtures of Normals Sociological Methodology Accepted.

Examples

### See help(diaph.data) for example