R function for Multiple Imputation of Longitudinal Binary Data Based on Propensity Score

Description 

R function time1(x, m) and timex(x, inacol, nacol, m) return the imputed data based on propensity score and beta distribution (see details in reference paper). This multiple imputation is to impute the missing values according to time series and the later imputed values are dependent on the prior imputed data, which is different from R package "MICE" and SAS PROCE MI where the imputation is performed based on regression models.

Usage

time1(x, m)

timex(x, inacol, nacol, m)

Arguments

x              data frame of longtitudinal binary data to impute
m             number of replicates
inacol      a list of the column numbers/names before timex without NAs
nacol       the number of time before time x, for example total times is 5, to impute the time 3, nacol=2.

Examples

Copy the full R code from HERE or run the following scripts in R Gui:

Download the function code on Mac:

download.file("https://raw.github.com/devilszhang/blog/master/timex.r", destfile="./timex.r", method="curl")

Download the function code on Windows:

download.file("https://raw.github.com/devilszhang/blog/master/timex.r", destfile="./timex.r")

Source the code:

source("./timex.r")

Access the demo data from HERE or using:

require(RCurl)
x<-read.table(textConnection(getURL("https://raw.github.com/devilszhang/blog/master/longdata.txt")), header=T)

#Impute day0
x$t1imp<-time1(polio52, 5)#Impute day4

inacol<-c("t1imp")   
nacol<-c("day4")   
x$t2imp<-timex(x, inacol, nacol, 1)    #Impute day11

 inacol<-c("t1imp", "t2imp")
nacol<-c("day11")
x$t3imp<-timex(x, inacol, nacol, 2)    #Impute day18

 inacol<-c("t1imp", "t2imp", "t3imp")
nacol<-c("day18")
x$t4imp<-timex(x, inacol, nacol, 3)    #Impute day25

 inacol<-c("t1imp", "t2imp", "t3imp", "t4imp")
nacol<-c("day25")
x$t5imp<-timex(x, inacol, nacol, 4)   
#complete data
complx<-x[, c("t1imp", "t2imp", "t3imp", "t4imp", "t5imp")]
complx    

References

Li X.M. et al. Statistics in Medicine, 2006, 25: 2107-2124.