Dear Matchitlist,
Many thanks for that incredible and useful software that solves
(almost) all my matching problems ...
I would like to know if it is possible to use something similar to the
"exact" option in nearest neighbor matching for all the others
matching algorithms..
That is, for example, apply a genetic algorithm but being able to
obtain matches that have absolutely the same values for, lets say,
their date of birth or any other string or numeric variable...
Many thanks in advance,
Best
--
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/
Hi, MatchIt Gurus...
I am running a simulation using MatchIt as part of my dissertation. I am
trying to compile a dataset of some of the summary data from the MatchIt
procedure (e.g. summary(df, standardized=TRUE)), but I seem to be unable to
do that. I keep getting the message that that object cannot be coerced into
a dataframe. Is there anyway that I can export these statistics so that I
can compile the results over many repetitions?
Thanks!
Shane
First my complements to all contributors of MatchIt, you really made my life
easier!
Now to my question. I estimated a variety of propensity scores by playing
with several different models: GBM, Logit and then trying out several
matching techniques: Nearest Neighbour (with a variety of calipers), Full
Matching, Subclassification and Genetic Matching.
In the literature I found
<http://www.sciencedirect.com/science/article/pii/S1010794009005727>researchers
to assess the balance by comparing the standardized difference in means and
calculating the variance ratio.
MatchIt summary output shows the standardized difference in means being most
effectively reduced with 1:1 Matching with a Caliper of 0.25 (all covariates
below a 10% level).
Since the t-test was understandably removed in later versions I wonder if
the variance ratio suffered the same fate? Or in other words why is it not
included in the output? And would it be possible to generate it from the
output?
And do you perhaps have any other pointers on how to assess balance or
generally accepted values that indicate a good balance?
Greatly appreciated if someone could point me in the right direction.
Regards,
Paul
Hi,
If I understand well, if I use different distance measures with the distance argument in matchit(), then the genetic method will used those distance measures to do mathching within matchit() ? Am I right ? I tried that approach, but I can't be sure that the genetic method effectively use the distance argument from matchit(), since that method is a stochastic algorithm which doesn't give the same answers anyway with tha same distance measure.
As an aside, I read the documentation for Jas Sekhon’s GenMatch function and I used GenMatch() direcly , but there is no object produces by GenMatch() which gives a distance value for each observation. There is only a "Weight.matrix" object which gives a weight for each variable and a "matches" object which gives a weigh for each matches pair.
Thanks,
François Maurice, B. Sc., A. Stat.
Candidat à la maîtrise
Département de sociologie
Université de Montréal
De : "Stuart, Elizabeth A." <estuart(a)jhsph.edu>
À : Francois Maurice <maurice.francois(a)ymail.com>
Envoyé le : Lundi 15 Août 2011 9h14
Objet : Re: [matchit] Distance values from Genetic matching
Hi,
Genetic matching uses the same propensity score as the other methods (at least partially); the difference is that it uses a different algorithm (and some other information on the covariates) to actually pick the matches.
The genetic matching algorithm MatchIt uses is from Jas Sekhon’s GenMatch function. You can read more about it here, which explains the distance measure used and the algorithm:
http://sekhon.berkeley.edu/matching/
Here is a brief description from that website: “GenMatch finds optimal balance using multivariate matching where a genetic search algorithm determines the weight each covariate is given. The user can choose which function of covariate balance to optimize from a list or provide one of her own.”
So the “distance” in genetic matching is still just the propensity score, but because the algorithm is different from nearest neighbor matching you get different matches.
Liz
On 8/14/11 11:37 AM, "Francois Maurice" <maurice.francois(a)ymail.com> wrote:
Hi,
>
>I'm trying to implement the ideas in Harder, Stuart and Anthony (2010), to seperate the estimation step from the application step and that way to combine various mathcing methods with various distance measures.
>
>But I have a problem with genetic mathing. Is there a way to obtain distance values from genetic mathcing (method="genetic") ? The distance values for the genetic method obtain from the matchit() object are the same as those when choosing method="nearest" and distance="logit" (the final results are not the same).
>
>Thanks,
>
>François Maurice, B. Sc., A. Stat.
>Candidat à la maîtrise
>Département de sociologie
>Université de Montréal
>
>
Hi,
I'm trying to implement the ideas in Harder, Stuart and Anthony (2010), to seperate the estimation step from the application step and that way to combine various mathcing methods with various distance measures.
But I have a problem with genetic mathing. Is there a way to obtain distance values from genetic mathcing (method="genetic") ? The distance values for the genetic method obtain from the matchit() object are the same as those when choosing method="nearest" and distance="logit" (the final results are not the same).
Thanks,
François Maurice, B. Sc., A. Stat.
Candidat à la maîtrise
Département de sociologie
Université de Montréal
Hi there,
I've been using matchit function together with the caliper argument and I'm afraid that this argument is not programmed according to the "caliper theory". To my knowledge if caliper=0.2 this number has to be multiplied by the sqrt((var(psT) + var(psC))/2), while the results of matchit function are that 0.2 is multiplied only to the sd(psALL).
psT - propensity scores of treated group
psC - propensity scores of control group
psALL - propensity scores of treated and control group
Please correct me if I'm wrong otherwise I'm wondering if its possible that you correct this mistake.
I'm looking forward to hearing from you soon.
Best regards,
Ana
Hi there,
Could you please tell me how is the calculation of the std.mean.diff in the matchit output summary programmed- what is the exact formula used for this standardization (is it standardized according to the SDc and SDt group?)
Many thanks,
Best regards,
Ana
Hi there,
Is MatchIt package capable of producing a matrix with all the matched pairs? The thing is that when match.data function is used, we only get all the data that are matched, while I would be more interested in getting the matched pairs.
Many thanks in advance!
Best regards,
Ana