SVMMatch (version 1.1)

treatment.overlap: Exploring hard-to-match treated observations.

Description

treatment overlap characterizes treated observations which have no natural matches in the data because they fall outside the common support region.

Usage

treatment.overlap(obj, color=TRUE, thresh=.95)

Arguments

obj

A fitted SVMMatch object.

color

Whether to plot in color or black and white. TRUE or FALSE.

thresh

Proportion of time a treated observation must fall outside the common support region in order to be considered unmatched.

Value

no.overlap

A binary vector the length of the number of treated observations taking a value of one if the observation falls outside the region of common support at some rate above the threshold.

logit

A logistic regression using the original covariates in the SVMMatch object to predict whether an observation falls outside the common support region.

Details

SVMMatch estimates a region of common support between treated and untreated observations. This function identifies and characterizes the treated observations that regularly fall outside the common support region. Having too many of these observations makes estimation of the treatment effect casts doubt on whether the average treatment effect is identified in the data.

References

Ratkovic, Marc. 2014. "Balancing within the Margin: Causal Effect Estimation with Support Vector Machines." Working paper.

See Also

svmmatch

Examples

## Not run: 
# ##See svmmatch() for a full implementation
# ##Load data
# 	data("LaLonde")
# 	Data1<-LaLonde
# 	Data1<-Data1[Data1$exper==0|Data1$treat==1,]
# 	attach(Data1)	
# 
# ##Format X matrix
# 	varnames<-c("age","educ","black","married","nodegr","hisp",
# 		"re75","re74")
# 	X<-cbind(Data1[,varnames],Data1$re75==0,Data1$re74==0)
# 	X<-as.matrix(X)
# 
# ##Fit model
# 	set.seed(1)
# 	m1.param<-svmmatch(treat, X, dv=re78, burnin=100, gibbs=100, thin=5)
# 
# 	#Assess treatment overlap--are there treated 
# 	#values that can't be matched?
# 		treat1<-treatment.overlap(m1.param)
# 
# 	## End(Not run)