Imputation at Level 2 (in miceadds)

The imputation method mice.impute.2lonly.function is a general imputation function for level 2 imputation which allow any defined imputation function at level 1 in mice.

mice.impute.2lonly.function(y, ry, x, wy=NULL, type, imputationFunction,
     cluster_var, ...)

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. Only numeric variables are permitted for usage of this function.
wy: Logical vector of length(y) indicating at which positions imputations should be conducted.
type: Cluster identifier can be specified by -2 for aggregation. However, we recommend to use the argument cluster_var for specifying the cluster variable at Level 2. Predictors must be specified by 1.
imputationFunction: Imputation function for mice. Any imputation method which is defined at level 1 can be used for level 2 imputation.
cluster_var: Cluster identifier for Level 2 units
...: Other named arguments.

Value

A vector of length nmis with imputations.

Examples

if (FALSE) {
#############################################################################
# EXAMPLE 1: Imputation of level 2 variables
#############################################################################

#**** Simulate some data
# x,y ... level 1 variables
# v,w ... level 2 variables

set.seed(987)
G <- 250            # number of groups
n <- 20             # number of persons
beta <- .3          # regression coefficient
rho <- .30          # residual intraclass correlation
rho.miss <- .10     # correlation with missing response
missrate <- .50     # missing proportion
y1 <- rep( stats::rnorm( G, sd=sqrt(rho)), each=n ) + stats::rnorm(G*n, sd=sqrt(1-rho))
w <- rep( round( stats::rnorm(G ), 2 ), each=n )
v <- rep( round( stats::runif( G, 0, 3 ) ), each=n )
x <-  stats::rnorm( G*n )
y <- y1 + beta  * x + .2 * w + .1 * v
dfr0 <- dfr <- data.frame( "group"=rep(1:G, each=n ), "x"=x, "y"=y,
        "w"=w, "v"=v )
dfr[ rho.miss * x + stats::rnorm( G*n, sd=sqrt( 1 - rho.miss ) ) <
                stats::qnorm(missrate), "y" ] <- NA
dfr[ rep( stats::rnorm(G), each=n ) < stats::qnorm(missrate), "w" ] <- NA
dfr[ rep( stats::rnorm(G), each=n ) < stats::qnorm(missrate), "v" ] <- NA

#* initial predictor matrix and imputation methods
predM <- mice::make.predictorMatrix(data=dfr)
impM <- mice::make.method(data=dfr)

#...
# multilevel imputation
predM1 <- predM
predM1[c("w","v","y"),"group"] <- c(0,0,-2)
predM1["y","x"] <- 1        # fixed x effects imputation
impM1 <- impM
impM1[c("y","w","v")] <- c("2l.continuous", "2lonly.function", "2lonly.function" )
# define imputation functions
imputationFunction <- list( "w"="sample", "v"="pmm5" )
# define cluster variable
cluster_var <- list( "w"="group", "v"="group" )

# impute
imp1 <- mice::mice( as.matrix(dfr), m=1, predictorMatrix=predM1, method=impM1, maxit=5,
            imputationFunction=imputationFunction, cluster_var=cluster_var )
}

Arguments

Value

See also

Examples