shuyang-stat / multilevelmatching Goto Github PK

correct links in roxygen are, e.g., \code{\link[lme4]{glmer}}

I typically use codecov

some s3 would be nice, e.g., print(). summary(), etc.

• Roll back initial vignette for a "legacy" version
• Create new vignette for v1.0 release that has new multiMatch() function as well as description of legacy functions

Need to setup Travis CI via: https://travis-ci.org/shuyang1987/multilevelMatching

• Travis will allow for easy unit testing, and codecov, etc.

Migrating this issue from BarkleyBG#8

the variance estimator for matching on estimated propensity scores from multinomial logistic regression was discussed in 2012, but the best reference is:

Abadie, Alberto, and Guido W. Imbens. "Matching on the estimated propensity score." Econometrica 84.2 (2016): 781-807.

http://onlinelibrary.wiley.com/doi/10.3982/ECTA11293/full

please add a DOI, arXiv or ISBN to your reference in your Description text in the form
authors (year) doi:...
authors (year) arXiv:...
authors (year, ISBN:...)
with no space after 'doi:', 'arXiv:' and angle brackets for auto-linking.

This repo will be in "maintenance mode" for a bit as it's going through some changes.

I'll be the new maintainer until further notice. I'll mark a change in the package description and (at least temporarily) in the README

Edit for clarity; explain difference between the new multiMatch() function and the "legacy" functions: multilevelGPSMatch() and multilevelMatchX()

Merge over the commits from my fork. This includes the new function multiMatch() which implements all 3 types of matching: on covariates, and existing or estimated propensity scores. (i.e., everything but GPS stratification)

The code report error when i run the function multiMatch. The dataset i use is simulated_data of the package. Could you please take a look at the code file, and repare. Thank you so much.
Error in length(class(X)) != 1 && class(X) != "matrix" :
'length = 2' in coercion to 'logical(1)'

As in BarkleyBG#9