R Graphical Manual

Browse All

Last data update: 2014.03.03

R: Imputation of a Variable with a Known Posterior Distribution

mice.impute.2l.eap

R Documentation

Imputation of a Variable with a Known Posterior Distribution

Description

This function imputes values of a variable for which the mean and the standard deviation of the posterior distribution is known.

Usage

mice.impute.2l.eap(y, ry, x, eap, ...)

Arguments

`y`	Incomplete data vector of length `n`
`ry`	Vector of missing data pattern (`FALSE` – missing, `TRUE` – observed)
`x`	Matrix (`n` x `p`) of complete covariates.
`eap`	List with means and standard deviations of the posterior distribution (see Examples). If for multiple variables posterior distributions are known, then it is a list named in which each list entry is named according th variable to be imputed and each list entry contains the variable's EAP and standard deviation of the EAP.
`...`	Further arguments to be passed

Value

A vector of length nmis=sum(!ry) with imputed values.

Author(s)

Alexander Robitzsch

Examples

## Not run: 
#############################################################################
# EXAMPLE 1: Imputation based on known posterior distribution
#############################################################################	
	
data(data.ma03)
dat <- data.ma03

# definiere variable 'math_PV' as the plausible value imputation of math
dat$math_PV <- NA
vars <- colnames(dat)
dat1 <- as.matrix( dat[,vars] )

# define imputation methods
impmethod <- rep( "pmm" , length(vars ))
names(impmethod) <- vars
# define plausible value imputation based on EAP and SEEAP for 'math_PV'
impmethod[ "math_PV" ] <- "2l.eap"
eap <- list( "math_PV" = list( "M" = dat$math_EAP  , "SE" = dat$math_SEEAP  ) )
# define predictor matrix
pM <- 1 - diag(1,length(vars))
rownames(pM) <- colnames(pM) <- vars
pM[,c("idstud","math_EAP" , "math_SEEAP") ] <- 0     
    # remove some variables from imputation model

# imputation using three parallel chains
imp1 <- mice::mice( dat1 , m=3 , maxit=5 , imputationMethod=impmethod ,  
               predictorMatrix = pM , allow.na =TRUE , eap=eap )
summary(imp1)   # summary

# imputation using one long chain
imp2 <- mice.1chain( dat1 , burnin=10 , iter=20 , Nimp =3 , imputationMethod=impmethod ,  
              predictorMatrix = pM ,  allow.na =TRUE , eap=eap )
summary(imp2)    # summary

## End(Not run)