Learn R Programming

matchMulti (version 1.0)

sdiff: Balance Measures

Description

Balance assessment for individual variables, before and after matching

Usage

sdiff(varname, treatment, orig.data, match.data = NULL, 
treat.wts = NULL, ctrl.wts = NULL, mt.wts = NULL, mc.wts = NULL)	
ttest.balance(varname, treatment, orig.data, match.data = NULL,
treat.wts = NULL, ctrl.wts = NULL, mt.wts = NULL, mc.wts = NULL)	
fisher.balance(varname, treatment, orig.data, match.data = NULL,
treat.wts = NULL, ctrl.wts = NULL, mt.wts = NULL, mc.wts = NULL)
wilc.balance(varname, treatment, orig.data, match.data = NULL,
treat.wts = NULL, ctrl.wts = NULL, mt.wts = NULL, mc.wts = NULL)

Arguments

varname
name of the variable on which to test balance
treatment
name of the binary indicator for treatment status
orig.data
a data frame containing the data before matching
match.data
an optional data frame containing the matched sample
treat.wts
optional weights for treated units in the original sample
ctrl.wts
optional weights for control units in the original sample
mt.wts
optional weights for treated units in the matched sample
mc.wts
optional weights for treated units in the matched sample

Value

  • a labeled vector. For sdiff, the vector has six elements if match.data is provided: treated and control means and standardized differences before and after matching. If match.data is not provided, the vector has only the three elements corresponding to the pre-match case. For the other functions, if match.data is provided, the vector contains p-values for the test before and after matching. Otherwise a single p-value is given for the pre-match data.

Details

The sdiff function computes the standardized difference in means. The other functions perform different kinds of balance tests: t.balance does the 2-sample t-test, fisher.balance does Fisher's exact test for binary variable, and wilc.balance does Wilcoxon's signed rank test.

References

Rosenbaum, Paul R. (2002). Observational Studies. Springer-Verlag.

Rosenbaum, Paul R. (2010). Design of Observational Studies. Springer-Verlag.