Dear Amelia Users,

This may be a problem with my understanding of how Amelia is supposed to work, but I’ve encountered the following problem. I give Amelia the following command:

amelia.out <- amelia(smalldat, cs=1, ts=2, polytime=3, lags=1, leads=1, intercs=F, incheck=F, mins=mins, maxs=maxs, conf=conf, means=NULL, sds=NULL,

logs=c(5,6,7,8),

sqrts=c(9,10,11,12,13,14,15,16),

noms=c(17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32))

and I get the following error:

Error in mu.priors - (ones %*% meanx) : non-conformable arrays

using traceback() I tracked it down to the scalecenter subcommand. Since I’ve not provided any means and standard deviations for the priors, rather I’ve provided

ranges, I assume that setting means and sds both to NULL will suggest that mu.priors and sd.priors will also be NULL. If I comment out the troublesome piece of

the scalecenter command as below, source this command in and issue the same command as above to Amelia, it works fine.

scalecenter <- function (x, mu.priors = NULL, sd.priors = NULL)

{

AMn <- nrow(x)

ones <- matrix(1, AMn, 1)

meanx <- colMeans(x, na.rm = TRUE)

stdvx <- apply(x, 2, sd, na.rm = TRUE)

x.ztrans <- (x - (ones %*% meanx))/(ones %*% stdvx)

# if (!identical(mu.priors, NULL)) {

# mu.priors <- (mu.priors - (ones %*% meanx))/(ones %*%

# stdvx)

# sd.priors <- sd.priors/(ones %*% stdvx)

# }

return(list(x = x.ztrans, mu = meanx, sd = stdvx, mu.priors = NULL,

sd.priors = NULL))

}

Can you give me some advice as to what I’m doing wrong in the first step?

Thanks,

Dave.

Dave Armstrong

Department of Politics and International Relations

Manor Road

Oxford

OX1 3UQ

Phone: 08165 285956

E-mail: david.armstrong@politics.ox.ac.uk

Web: http://users.ox.ac.uk/~polf0104