CBPS (version 0.14)

vcov_outcome: Calculate Variance-Covariance Matrix for Outcome Model

Description

vcov_outcome Returns the variance-covariance matrix of the main parameters of a fitted CBPS object.

Usage

# S3 method for CBPSContinuous
vcov_outcome(object, Y, Z, delta, tol = 10^(-5), lambda = 0.01)

Arguments

object

A fitted CBPS object.

Y

The outcome.

Z

The covariates (including the treatment and an intercept term) that predict the outcome.

delta

The coefficients from regressing Y on Z, weighting by the cbpsfit$weights.

tol

Tolerance for choosing whether to improve conditioning of the "M" matrix prior to conversion. Equal to 1/(condition number), i.e. the smallest eigenvalue divided by the largest.

lambda

The amount to be added to the diagonal of M if the condition of the matrix is worse than tol.

Value

A matrix of the estimated covariances between the parameter estimates in the weighted outcome regression, adjusted for uncertainty in the weights.

Details

This adjusts the standard errors of the weighted regression of Y on Z for uncertainty in the weights.

References

Lunceford and Davididian 2004.

Examples

# NOT RUN {
###
### Example: Variance-Covariance Matrix
###

##Load the LaLonde data
data(LaLonde)
## Estimate CBPS via logistic regression
fit <- CBPS(treat ~ age + educ + re75 + re74 + I(re75==0) + I(re74==0), 
		    data = LaLonde, ATT = TRUE)
## Get the variance-covariance matrix.
vcov(fit)
# }