Matchit January 2011

matchit@lists.gking.harvard.edu

7 participants
4 discussions

by Evann Smith

I've created a matched mulitply imputed dataset (created in Amelia) via cem, using the option "datalist." Now that I've done so, I'm trying to figure out how to apply the match.data() command in MatchIt, so that I may analyze the data in Zelig. MatchIt does not seem to want to process the output object from cem, and yet I can't find a command in cem similar to match.data(). What's the best way to proceed? > matt4 <- cem(treatment="t4", drop=c("numcode", "group", "ccode", "country", "region", "rcode", "year"), datalist=datprep, data=data) > class(matt4) [1] "cem.match.list" "list" > t4data <- match.data(matt4) > class(t4data) [1] "NULL" > t4data NULL > modt4.1 <- zelig(prot ~ lgpop + gpro + poldis + gorg + repact + polity2 + secfil + toolfil + lgdp + lgdppc, model=ls, data=t4data) Error in paste("zelig2", model, sep = "") : cannot coerce type 'closure' to vector of type 'character' ==================================== Evann Smith Ph.D. Candidate Harvard University, Department of Government egsmith(a)fas.harvard.edu ==================================== -- 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/ Browse/Search List Archive: http://lists.hmdc.harvard.edu/lists/matchit/

13 years, 3 months

Weights when ratio>1

by Chris Hane

Hello, I am using nearest matching with ratio of 20 or 50 and a caliper. When I do this I get the message that not every treated observation has the requested number of matches. This is OK for my application, but the weights of the controls are all 1. I would like the weights to reflect the number of controls for each treated observation, That is, control weight=1/K where K is the number of non-NAs in the match.matrix in the row where the control obsID occurs. Otherwise the weights do not properly reflect the balance achieved by matching. Is there a simple way to compute this weight vector from the match matrix? Thanks, Chris

13 years, 3 months

Mahalanobis distance values

by Karen Johnson

Hello, I'm very new to R and to MatchIt, so I'm not really sure what I should be seeing...but when I run it using a Mahalanobis distance, all the distance values in the resulting output are "NA". This happens both with my "real" data and with the LALONDE data. Is this correct? (Note: I'm running the latest versions of R and MatchIt). Thanks. --Karen -- 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/ Browse/Search List Archive: http://lists.hmdc.harvard.edu/lists/matchit/

13 years, 3 months

"variable lengths differ (found for '(weights)')" error in Zelig library

by Sotiris Adamakis

Hi, I posted the following on the R discussion group, but this might be relevant here as well. I am trying to estimate the average treatmen effect on the treated (ATT) using first the MatchIt software to weight the data set and, after this, the Zelig software as shown in Ho et al. (2007). This is my source: http://imai.princeton.edu/research/files/matchit.pdf I encounter a slight problem when I apply the weights that are produced in the stage of preprocessing the data. The idea of this is to use the MatchIt software to preprocess the data and then use the Zelig software to generate the distribution of ATT. I believe that the main reason for preprocessing the data is to create weights (depending on the matching technique you use) so that balance would be achieved for the matching variables between the treatment and the control group. Then you use these weights in the regressions that follow in the Zelig library. Copied from the matchit article, whose link I provide above, the authors say: "If one chooses options that allow matching with replacement, or any solution that has different numbers of controls (or treateds) within each subclass or strata (such as full matching), then the parametric analysis following matching must accomodate these procedures, such as by using fixed effects or weights, as appropriate. (Similar procedures can also be used to estimate various other quantities of interest such as the average treatment effect by computing it for all observations, but then one must be aware that the quantity of interest may change during the matching procedure as some control units may be dropped.)" The following code is for the "lalonde" data set, where I get an error message in the end. I use the more recent R version (2.12.1) and have updated the Zelig and MatchIt libraries. > update.packages("MatchIt") > update.packages("Zelig") > library(Zelig) > library(MatchIt) > data(lalonde) > m.out1 = matchit(treat ~ age + educ + black + hispan + nodegree + married + re74 + re75, method = "subclass", subclass=6, data = lalonde) > z.out1 = zelig(re78 ~ age + educ + black + hispan + nodegree + married + re74 + re75, data = match.data(m.out1, "control"), model = "ls", weights="weights") > x.out1 = setx(z.out1, data = match.data(m.out1, "treat"), cond = TRUE) > s.out1 = sim(z.out1, x = x.out1) Error in model.frame.default(formula = re78 ~ age + educ + black + hispan + : variable lengths differ (found for '(weights)') I was wondering if somebody could tell me how to get around with this problem? Also, I have seen people adding the propensity scores in the regression analysis applied in the Zelig package, i.e. > z.out1 = zelig(re78 ~ age + educ + black + hispan + nodegree + married + re74 + re75 + distance, data = match.data(m.out1, "control"), model = "ls", weights="weights") Does anyone have a clue of why this can happen? Kind regards, Sotiris

13 years, 3 months

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

Matchit January 2011