Hi Kate,
I'm thinking that the reason why you see this behavior is because you
use exact matching (method = "exact"). This means that an observation
which cannot find an exact match will be discarded even if you set discard
= "none". Let us know if the problem persists even with non-exact
matching.
Many thanks,
Kosuke
-----------------------------------------------------
Kosuke Imai Office: Corwin Hall 041
Assistant Professor Phone: 609-258-6601
Department of Politics eFax: 973-556-1929
Princeton University Email: kimai(a)Princeton.Edu
Princeton, NJ 08544-1012 http://imai.princeton.edu
-----------------------------------------------------
On Mon, 12 Jun 2006, Kate Crespi wrote:
> Dear Prof. Imai,
>
>
>
> I was very happy to discover the MatchIt package in R and would like to us
> it in my research. However, I've come across what might be a bug. I have
> found that even when the discard="none" option is specified, the matchit
> function still discards outlying observations. Below is an example. I'm
> not sure whether the program is intended to work this way, so I thought to
> bring it to your attention. Users of the package would most likely find it
> more useful if they could control whether or not observations are discarded.
>
>
>
> testdata <- matrix(runif(150,0,0.1), nrow=10, ncol=15)
>
> testdata <- rbind(runif(15,0.9,1), testdata, testdata) #add an outlie
>
> tx <- c(rep(0,11),rep(1,10))
>
> testdata <- cbind(tx, testdata)
>
> colnames(testdata) <- c("tx", 1:15)
>
> testdata <- data.frame(testdata)
>
> mtest <- matchit(tx~X1+X2+X3+X4+X5+X6+X7+X8+X9+X10+X11+X12+X13+X14+X15,
> testdata, method="exact",
>
> discard="none")
>
> summary(mtest)
>
>
>
>
>
> Sincerely,
>
> Kate Crespi
>
>
>
>
>
>
>
> Catherine M. Crespi, Ph.D.
>
> Postdoctoral Fellow
>
> Department of Biostatistics
>
> University of California Los Angeles
>
> ccrespi(a)ucla.edu
>
>
>
>
>
>
-
MatchIt mailing list served by Harvard-MIT Data Center
List Address: matchit(a)latte.harvard.edu
Subscribe/Unsubscribe: http://lists.hmdc.harvard.edu/?info=matchit
MatchIt Software and Documentation: http://gking.harvard.edu/matchit/