Semi-asymptotic counterfactual distribution

distreg_cfa.sas takes input object from dr_asympar for counterfactual semi asymptotic bayesian distribution. This involves taking random draws from the normal approximation of the posterior at each threshold value.

distreg_cfa.sas(ind, drabj, data, cft, cfIND, vcovfn = "vcov",
  iter = 100)

Arguments

ind	index of object in list drabj (i.e. a threshold value) from which to take draws
drabj	object from dr_asympar
data	dataframe, first column is the outcome
cft	column vector of counterfactual treatment
cfIND	the column index(indices) of treatment variable(s) to replace with cft in data0
vcovfn	a string denoting the function to extract the variance-covariance. Defaults at "vcov". Other variance-covariance estimators in the sandwich package are usable.
iter	number of draws to simulate

Value

fitob vector of random draws from density of F(yo) using semi-asymptotic BDR

Examples

y = faithful$waiting
x = scale(cbind(faithful$eruptions,faithful$eruptions^2))
qtaus = quantile(y,c(0.05,0.25,0.5,0.75,0.95))
drabj<- dr_asympar(y=y,x=x,thresh = qtaus); data = data.frame(y,x)
cfIND=2 #Note: the first column is the outcome variable. 
cft=0.95*data[,cfIND] # a decrease by 5%
cfa.sasobj<- distreg_cfa.sas(ind=2,drabj,data,cft,cfIND,vcovfn="vcov")
par(mfrow=c(1,2)); plot(density(cfa.sasobj$original,.1),main="Original")
plot(density(cfa.sasobj$counterfactual,.1),main="Counterfactual"); par(mfrow=c(1,1))