Hey all-
I have the fortune of having around 40 pre-treatment variables that I
can match on. Should I use them all?
If not, is there a systematic way to choose which variables to
include? For example, should I just use the ones that correlate the
most with the treatment and the outcome?
Thanks!
--
Casey A. Klofstad
University of Miami
Department of Political Science
Coral Gables, FL
klofstad(a)gmail.com
http://moya.bus.miami.edu/~cklofstad
--
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,
I assume this error message means that I should first transform the
treatment variable to a 0 to 1 proportional form?
Thanks,
Anders
>
m.out.c1<-matchit(a_norm_total2*initiateinter~a_norm_total2*defendinter+a_norm_total2*initiatecivil+cinc.nmc/ter+gpconc+
+ dems6/num_stat+igoprop+open+diversity+major+polity2+disintegrate,
data=conflictdata$m1, method="full")
Error in eval(expr, envir, enclos) : y values must be 0 <= y <= 1
>
--
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/
Dear all,
I've really been enjoying using matchit --- it's much easier to
change options than to change my own code, and it's very clearly
documented. Thanks for writing it. I've read much of the
documentation, and have one question and one funny property to share:
1. It looks like the package is now entirely non-parametric, but is
there any way to get the old output for MatchIt including sd and t-
test results, as shown here?
http://gking.harvard.edu/matchit/docs/Propensity_Score_Match.html
The empirical QQ plots do seem more helpful than tests of central
tendency, but I find it hard to evaluate a few dozen covariates
visually. Are there any guidelines for interpreting the eQQ summary
statistics?
2. I discovered a funny property that I can't explain. It's not a
problem since I discovered a work-around, but I wanted to let you
know, and I'm curious if someone has an explanation.
The first set of commands do not work, and the second commands do,
although they do exactly the same thing. Presumably covar2 and
covar1 do not have exactly the same properties in spite of containing
the same data.
> covar1<-P1[,2:36]
> dim(covar1)
[1] 3440 35
> is.data.frame(covar1)
[1] TRUE
> is.matrix(covar1)
[1] FALSE
> test1<-matchit(formula = pledge2 ~ mom_rel1 + dad_rel1 + age1,
data=covar1, method="nearest", distance = "logit", mahvars = c
("mom_rel1", "dad_rel1"), caliper=0.25)
Error: subscript out of bounds
> covar2<-P1[1:3440,2:36]
> dim(covar2)
[1] 3440 35
> is.data.frame(covar2)
[1] TRUE
> is.matrix(covar2)
[1] FALSE
> test2<-matchit(formula = pledge2 ~ mom_rel1 + dad_rel1 + age1,
data=covar2, method="nearest", distance = "logit", mahvars = c
("mom_rel1", "dad_rel1"), caliper=0.25)
> [runs fine]
Thanks,
Janet
--
Janet Rosenbaum
Please start using my new email address now: "janet at
post.harvard.edu."
"jerosenb at fas.harvard.edu" will expire.
--
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/
