Dear Ajay,
I'm just trying to follow the recommendation of the paper "The
Essential Role of Pair Matching in Cluster-Randomized Experiments,
with Application to the Mexican Universal Health Insurance Evaluation"
by Kosuke Imai, Gary King, and Clayton Nall, Statist. Sci. Volume 24,
Number 1 (2009), 29-53." Another interesting reference they cite is
"Optimal multivariate matching before randomization." Greevy, R., Lu,
B., Silber, J. H. and Rosenbaum, P. (2004). Biostatistics 5 263–275.
I am conducting a cluster/group-randomized experiment with treatment
at the level of schools -- in order to increase power (as well as
address other issues mentioned in the Imai, King, Nall 2009 paper) --
I am trying to match schools in pairs on observable covariates that
would be more strongly related to the treatment and potential
outcomes. Also, within schools, students will be given treatment A, B,
or control (at the individual level)--blocking students into groups of
3 and then randomizing treatment status within each group of 3 would
also increase power if my understanding is correct.
best wishes,
Prashant
On Sat, Nov 6, 2010 at 11:39 AM, Ajay Shah <ajayshah(a)mayin.org> wrote:
>> On 11/4/10, Prashant <presearchwork(a)gmail.com> wrote:
>> > Dear all,
>> >
>> > I read a past matchit post from 2007 (Prof. Imai) that matchit can't
>> > conduct matching (say in pairs) before randomization.
>> > He introduced his R program "experiment" which can conduct matching in
>> > pairs before randomization using an optimal matching method.
>> > Is there any program available to conduct optimal matching in
>> > equal-size blocks where the units in each block is more than 2 (i.e.
>> > where there are more than two treatment groups)?
>> >
>> > If not, I would most appreciate any practical advice whatsoever on
>> > what would be the best way to match before randomization when there
>> > are more than two treatment groups and if there is any software
>> > available for this purpose.
>
> Dear Prashant,
>
> I find your question to be intruiging. I don't know the field. Can you
> explain the setting and what you are trying?
>
> BTW the package that he pointed to is "blockTools" and not "blocking".
>
> --
> Ajay Shah http://www.mayin.org/ajayshah
> ajayshah(a)mayin.org http://ajayshahblog.blogspot.com
> <*(:-? - wizard who doesn't know the answer.
>
--
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/
Dear all,
I read a past matchit post from 2007 (Prof. Imai) that matchit can't
conduct matching (say in pairs) before randomization.
He introduced his R program "experiment" which can conduct matching in
pairs before randomization using an optimal matching method.
Is there any program available to conduct optimal matching in
equal-size blocks where the units in each block is more than 2 (i.e.
where there are more than two treatment groups)?
If not, I would most appreciate any practical advice whatsoever on
what would be the best way to match before randomization when there
are more than two treatment groups and if there is any software
available for this purpose.
Thank you kindly,
Prashant
--
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/