Dear Marianne,
Thanks for this input. We've added an example sytanx about how to get
unmatched data sets in our software manual; see "Examples" under
"match.data()" http://gking.harvard.edu/matchit/docs/Examples3.html
If you have any other comments and suggestions, please let us know at
our mailing list serve.
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 Thu, 5 Jul 2007, Gary King wrote:
>
> many thanks for your note and the code! That's terrific. I'm CCing my
> coauthors who will know lots about this. We'll take a look, give it a try,
> and be back in touch.
> Many thanks!
> Gary
>
> On Thu, 5 Jul 2007, Hubregtse, M. wrote:
>
>> Dear mr. King,
>>
>> In using your package for scientific research, I needed to find out which
>> specific people were not matched. I therefore rewrote a tiny piece of the
>> function match.data so that the new function became as below. This
>> functions exactly the same as match.data but selects all respondents that
>> were not matched for further analysis. I think this function can be used by
>> many more people and that it is very important to show in some researches
>> that the people you could not match differ significantly on certain
>> variables.
>>
>> Could you or would you want to incorporate this function in your package?
>> If this is not the right address to pose this question, could you maybe
>> relocate the e-mail?
>>
>> Kindest regards,
>>
>> Marianne Hubregtse, Utrecht University
>>
>> P.S. The function I used is: (where in the selection part return(data[wt <
>> 1, ]) is different from the original function).
>>
>> unmatch.data <-
>> function (object, group = "all", distance = "distance", weights =
>> "weights",
>> subclass = "subclass")
>> {
>> data <- eval(object$call$data)
>> treat <- object$treat
>> wt <- object$weights
>> vars <- names(data)
>> if (distance %in% vars)
>> stop("invalid input for distance. choose a different name.")
>> else if (!is.null(object$distance)) {
>> dta <- data.frame(cbind(data, object$distance))
>> names(dta) <- c(names(data), distance)
>> data <- dta
>> }
>> if (weights %in% vars)
>> stop("invalid input for weights. choose a different name.")
>> else if (!is.null(object$weights)) {
>> dta <- data.frame(cbind(data, object$weights))
>> names(dta) <- c(names(data), weights)
>> data <- dta
>> }
>> if (subclass %in% vars)
>> stop("invalid input for subclass. choose a different name.")
>> else if (!is.null(object$subclass)) {
>> dta <- data.frame(cbind(data, object$subclass))
>> names(dta) <- c(names(data), subclass)
>> data <- dta
>> }
>> if (group == "all")
>> return(data[wt < 1, ])
>> else if (group == "treat")
>> return(data[wt < 1 & treat == 1, ])
>> else if (group == "control")
>> return(data[wt < 1 & treat == 0, ])
>> else stop("error: invalid input for group.")
>> }
>>
>
--
MatchIt mailing list served by Harvard-MIT Data Center
List Address: matchit(a)lists.gking.harvard.edu
Subscribe/Unsubscribe: http://lists.gking.harvard.edu/?info=matchit
MatchIt Software and Documentation: http://gking.harvard.edu/matchit/
Hi, Bill. In addition to Kosuke's "experiment", my forthcoming
package, "blockTools", implements several ways to create blocks in
experimental designs. In addition to multivariate blocking, the
package includes assignment routines, some nice output
options, and functions for diagnosing potential interference problems
among units.
Version 0-1.1 will be released within a month -- I'd greatly
appreciate any suggestions you have for useful functionality.
Thanks, and hope this helps.
RTM
--
Ryan T. Moore
Ph.D. Candidate
Government & Social Policy
Harvard University
www.people.fas.harvard.edu/~rtmoore
***************************************************************************
---------- Forwarded message ----------
Date: Thu, 5 Jul 2007 18:02:21 -0400 (EDT)
From: William Mabe <billmabe(a)rci.rutgers.edu>
To: matchit(a)lists.gking.harvard.edu
Subject: [matchit] MatchIt and experiments
For an experiment, can MatchIt be used to group together similar units into
blocks prior to random assignment to treatment? Any suggestions on how to
implement this? I didn't see anything on this issue in either the MatchIt
documentation or the archives.
Thanks,
Bill
************************************************************************
--
MatchIt mailing list served by Harvard-MIT Data Center
List Address: matchit(a)lists.gking.harvard.edu
Subscribe/Unsubscribe: http://lists.gking.harvard.edu/?info=matchit
MatchIt Software and Documentation: http://gking.harvard.edu/matchit/
For an experiment, can MatchIt be used to group together similar units
into blocks prior to random assignment to treatment? Any suggestions on
how to implement this? I didn't see anything on this issue in either the
MatchIt documentation or the archives.
Thanks,
Bill
**************************************************************************
William F. Mabe, Jr., PhD
Project Director
Heldrich Center for Workforce Development
30 Livingston Ave.
Rutgers University
New Brunswick, NJ 08901
p: (732)932-4100 ext. 6213
f: (732)932-3454
--
MatchIt mailing list served by Harvard-MIT Data Center
List Address: matchit(a)lists.gking.harvard.edu
Subscribe/Unsubscribe: http://lists.gking.harvard.edu/?info=matchit
MatchIt Software and Documentation: http://gking.harvard.edu/matchit/
