AT: Average Effect

Description

AT can be used to calculate the unconditional average treatment effect (ATE) and average treatment effect on the treated (ATT) of a binary predictor/treatment, with corresponding `confidence' intervals conveniently calculated via posterior simulation.

Usage

AT(x,eq,nm.bin="",sig.lev=0.05,n.sim=1000,s.meth="svd",E=TRUE,treat=TRUE)

Arguments

A fitted SemiParBIVProbit object as produced by SemiParBIVProbit().

The equation containing the binary predictor of interest.

nm.bin

The name of the binary variable.

sig.lev

Significance level.

n.sim

The number of simulated coefficient vectors from the posterior distribution of the estimated model parameters.

s.meth

Matrix decomposition used to determine the matrix root of the covariance matrix. See the documentation of the mvtnorm package for further details.

If TRUE, then AT calculates the ATE. If FALSE, then it calculates the ATT.

treat

If TRUE, then AT calculates the ATT. If FALSE, then it calculates the average treatment effect on the control group. This only makes sense if used jointly with E=FALSE.

Value

resIt returns three values, namely the lower confidence interval limit, estimated AT and upper confidence interval limit, respectively.
sig.levThe significance level used.

Details

AT measures the average causal difference in outcomes under the treatment (the binary predictor/treatment assumes value 1) and under the control (the binary treatment assumes value 0). The corresponding `confidence' intervals are conveniently calculated via simulation from the posterior distribution of the estimated model parameters.

References

Marra G. and Radice R. (2011), Estimation of a Semiparametric Recursive Bivariate Probit in the Presence of Endogeneity. Canadian Journal of Statistics, 39(2), 259-279.

Examples

Run this code

## see examples for SemiParBIVProbit

Run the code above in your browser using DataLab