Pooled degrees of freedom calculation

Edited from mice::barnard.rubin. This function computes the residual degrees of freedom for hypothesis testing, as proposed by Barnard & Rubin (1999). It is used by the vw_pool function.

Usage

barnard.rubin(lambda, m, dfcom = Inf)

Arguments

lambda

: Proportion of total variance due to missingness

m

: Number of imputed datasets

dfcom

: Residual degrees of freedom, estimated using stats::df.residual()

Value

Value for the residual degrees of freedom

References

Barnard, J. and Rubin, D.B. (1999). Small sample degrees of freedom with multiple imputation. Biometrika, 86, 948-955.