24 Sep
2018
24 Sep
'18
4:56 p.m.
Hello,
I have been using MatchIt for a project and have some categorical
covariates. (I know that continuous covariates are optimal, but
unfortunately we only have categorical for some of our data). I have
already coded them so that the categories are numeric.
When I perform the matching and ask for the summary, sometimes it gives
balance measures for all of the levels of the categorical variable and
other times it does not. For example, my variable Class Standing is 4
levels and I get balance measures for all 4 levels, but my variable Current
GPA is 4 levels and I only get balance measures for the second, third, and
fourth level of the variable.
Is there a way to correct this?
Thank you,
Maureen Bowers
24 Sep
24 Sep
5:31 p.m.
New subject: Question about balance measures for categorical covariates
Use the *cobalt *package, which interfaces directly with *MatchIt *output.
bal.tab(m.out), where m.out is the output of a call to matchit(), produces
a balance table like summary() but solves some of these issues. Binary
variables will have one level displayed, while multi-category variables
will have all levels displayed. Note the default balance statistic for
binary variables is the raw difference in proportion, but the option to
produce standardized difference is available.
I'm not sure why that glitch occurs in *MatchIt*. Nothing about the raw
code seems to imply that this might occur except if you have an unusual
variable name that uses spaces or other punctuations.
Noah
PS, full disclosure, I am the author of *cobalt*.
On Mon, Sep 24, 2018 at 4:56 PM Maureen Bowers <maureenbowers(a)gmail.com>
wrote:
Hello,
I have been using MatchIt for a project and have some categorical
covariates. (I know that continuous covariates are optimal, but
unfortunately we only have categorical for some of our data). I have
already coded them so that the categories are numeric.
When I perform the matching and ask for the summary, sometimes it gives
balance measures for all of the levels of the categorical variable and
other times it does not. For example, my variable Class Standing is 4
levels and I get balance measures for all 4 levels, but my variable Current
GPA is 4 levels and I only get balance measures for the second, third, and
fourth level of the variable.
Is there a way to correct this?
Thank you,
Maureen Bowers
-
---
MatchIt mailing list served by HUIT
List Address: matchit(a)lists.gking.harvard.edu
Subscribe/Unsubscribe: http://lists.gking.harvard.edu/mailman/listinfo/ei
MatchIt Software and Documentation: http://gking.harvard.edu/matchit/
Browse/Search <http://gking.harvard.edu/matchit/Browse/Search> List
Archive: http://lists.gking.harvard.edu/mailman/private/matchit/
Matchit mailing list
Matchit(a)lists.gking.harvard.edu
To unsubscribe from this list or get other information:
https://lists.gking.harvard.edu/mailman/listinfo/matchit
5:51 p.m.
New subject: Question about balance measures for categorical covariates
Actually, I did find the issue. The first factor variable provided in the
formula should have all levels represented, while the other factor
variables will have their first level removed. This has to do with how the
matchit code process the input formula (lines 46-49 in the matchit.R
script).
Noah
On Mon, Sep 24, 2018 at 5:31 PM Noah Greifer <noah.greifer(a)gmail.com> wrote:
Use the *cobalt *package, which interfaces directly
with *MatchIt *output.
bal.tab(m.out), where m.out is the output of a call to matchit(),
produces a balance table like summary() but solves some of these issues.
Binary variables will have one level displayed, while multi-category
variables will have all levels displayed. Note the default balance
statistic for binary variables is the raw difference in proportion, but the
option to produce standardized difference is available.
I'm not sure why that glitch occurs in *MatchIt*. Nothing about the raw
code seems to imply that this might occur except if you have an unusual
variable name that uses spaces or other punctuations.
Noah
PS, full disclosure, I am the author of *cobalt*.
On Mon, Sep 24, 2018 at 4:56 PM Maureen Bowers <maureenbowers(a)gmail.com>
wrote:
Hello,
I have been using MatchIt for a project and have some categorical
covariates. (I know that continuous covariates are optimal, but
unfortunately we only have categorical for some of our data). I have
already coded them so that the categories are numeric.
When I perform the matching and ask for the summary, sometimes it gives
balance measures for all of the levels of the categorical variable and
other times it does not. For example, my variable Class Standing is 4
levels and I get balance measures for all 4 levels, but my variable Current
GPA is 4 levels and I only get balance measures for the second, third, and
fourth level of the variable.
Is there a way to correct this?
Thank you,
Maureen Bowers
-
---
MatchIt mailing list served by HUIT
List Address: matchit(a)lists.gking.harvard.edu
Subscribe/Unsubscribe: http://lists.gking.harvard.edu/mailman/listinfo/ei
MatchIt Software and Documentation: http://gking.harvard.edu/matchit/
Browse/Search <http://gking.harvard.edu/matchit/Browse/Search> List
Archive: http://lists.gking.harvard.edu/mailman/private/matchit/
Matchit mailing list
Matchit(a)lists.gking.harvard.edu
To unsubscribe from this list or get other information:
https://lists.gking.harvard.edu/mailman/listinfo/matchit
2063
days inactive
2063
days old
matchit@lists.gking.harvard.edu
2 comments
2 participants
participants (2)
-
Maureen Bowers -
Noah Greifer