This function provides a number of univariate balance metrics.
Generally, users should call MatchBalance and not this function
directly.
balanceUV(Tr, Co, weights = rep(1, length(Co)), exact = FALSE, ks=FALSE,
nboots = 1000, paired=TRUE, match=FALSE,
weights.Tr=rep(1,length(Tr)), weights.Co=rep(1,length(Co)),
estimand="ATT")A vector containing the treatment observations.
A vector containing the control observations.
A vector containing the observation specific weights. Only use this option when the treatment and control observations are paired (as they are after matching).
A logical flag indicating if the exact Wilcoxon test
should be used instead of the test with a correction. See
wilcox.test for details.
A logical flag for if the univariate bootstrap
Kolmogorov-Smirnov (KS) test should be calculated. If the ks option
is set to true, the univariate KS test is calculated for all
non-dichotomous variables. The bootstrap KS test is consistent even
for non-continuous variables. See ks.boot for more
details.
The number of bootstrap samples to be run for the
ks test. If zero, no bootstraps are done. Bootstrapping is
highly recommended because the bootstrapped Kolmogorov-Smirnov test
only provides correct coverage even for non-continuous covariates. At
least 500 nboots (preferably 1000) are
recommended for publication quality p-values.
A flag for if the paired t.test should be used.
A flag for if the Tr and Co objects are the result
of a call to Match.
A vector of weights for the treated observations.
A vector of weights for the control observations.
This determines if the standardized mean difference
returned by the sdiff object is standardized by the variance of
the treatment observations (which is done if the estimand is either
"ATE" or "ATT") or by the variance of the control observations (which
is done if the estimand is "ATC").
This is the standardized difference between the treated
and control units multiplied by 100. That is, 100 times the mean
difference between treatment and control units divided by the standard
deviation of the treatment
observations alone if the estimand is either ATT or
ATE. The variance of the control observations are used if
the estimand is ATC.
This is the standardized difference between the treated and control units multiplied by 100 using the pooled variance. That is, 100 times the mean difference between treatment and control units divided by the pooled standard deviation as in Rosenbaum and Rubin (1985).
The mean of the treatment group.
The mean of the control group.
The variance of the treatment group.
The variance of the control group.
The p-value from the two-sided weighted t.test.
var.Tr/var.Co.
The object returned by ks.boot.
The object returned by two-sided weighted
t.test.
The return object from a call to
qqstats with standardization---i.e., balance test
based on the empirical CDF.
The return object from a call to
qqstats without standardization--i.e., balance tests
based on the empirical QQ-plot which retain the scale of the
variable.
Sekhon, Jasjeet S. 2011. "Multivariate and Propensity Score Matching Software with Automated Balance Optimization.'' Journal of Statistical Software 42(7): 1-52. http://www.jstatsoft.org/v42/i07/
Diamond, Alexis and Jasjeet S. Sekhon. 2013. "Genetic Matching for Estimating Causal Effects: A General Multivariate Matching Method for Achieving Balance in Observational Studies.'' Review of Economics and Statistics. 95 (3): 932--945. http://sekhon.berkeley.edu/papers/GenMatch.pdf
Rosenbaum, Paul R. and Donald B. Rubin. 1985. ``Constructing a Control Group Using Multivariate Matched Sampling Methods That Incorporate the Propensity Score.'' The American Statistician 39:1 33-38.
Hollander, Myles and Douglas A. Wolfe. 1973. Nonparametric statistical inference. New York: John Wiley & Sons.
Also see summary.balanceUV, qqstats
ks.boot, Match, GenMatch,
MatchBalance,
GerberGreenImai, lalonde
# NOT RUN {
data(lalonde)
attach(lalonde)
foo <- balanceUV(re75[treat==1],re75[treat!=1])
summary(foo)
# }
Run the code above in your browser using DataLab