Not sure exactly what you want to do. Are you trying to get the mean for each obs across the five datasets? Then you can just use aggregate for example or lappy. You can also do it in an array. Here is an example with two datasets: x1 <- data.frame(ind = 1:10, a = 1:10, beta = exp(1:10), logic = rep(c(TRUE,TRUE,FALSE,FALSE,TRUE),2)); x2 <- data.frame(ind = 1:10, a = 11:20, beta = log(1:10), logic = .5*rep(c(TRUE,TRUE,FALSE,FALSE,TRUE),2)); store <- rbind(x1,x2); aggregate(store[,-1],by=list(ind=store$ind),mean); From: gov2001-l-bounces at lists.fas.harvard.edu [mailto:gov2001-l-bounces at lists.fas.harvard.edu] On Behalf Of Brett Logan Carter Sent: Saturday, April 26, 2008 9:14 PM To: gov2001-l at lists.fas.harvard.edu Subject: [gov2001-l] Simulation with multiply imputed data sets Hi everyone, I have a quick question. I'm working with several multiply imputed data sets and would like to simulate quantities of interest at mean/mode values of explanatory variables. Because of the way I subset data iteratively in a for loop, I would like to consolidate the five data sets into a single data set, with each observation simply taking the mean (or, in some cases, mode) value of the five data cells from the imputed data sets. So my question: How can I coerce the five imputed data sets into a single data set with mean/mode values? There must be some straightforward way to do this. Thanks in advance for the help. Best, Brett