confint.ateRobust: Confidence Intervals for the Average Treatment Effect

In bozenne/riskRegressionLight: Light version of the package riskRegression

Description

Confidence intervals for the average treatment effect

Usage

## S3 method for class 'ateRobust'
confint(object, parm = NULL, level = 0.95,
  meanRisk.transform = "none", diffRisk.transform = "none",
  method.se.AIPTW = "orthogonality", ...)

Arguments

object	A ateRobust object, i.e. output of the ateRobust function.
parm	not used. For compatibility with the generic method.
level	[numeric, 0-1] Level of confidence.
meanRisk.transform	[character] the transformation used to improve coverage of the confidence intervals for the mean risk in small samples. Can be "none", "log", "loglog", "cloglog".
diffRisk.transform	[character] the transformation used to improve coverage of the confidence intervals for the risk difference in small samples. Can be "none", "atanh".
method.se.AIPTW	[character] Which type of standard error should be used for the AIPTW estimator? Can be "orthogonality": the incertainty related to the estimation of the nuisance parameters is ignored, or "full": only the incertainty related to the estimation of the nuisance parameters for the censoring model is ignored. Can also be "both" to output the results relative to "orthogonality" and "full".
...	not used.

Details

The standard errors obtained with method.se.AIPTW="orthogonality" and method.se.AIPTW="orthogonality" are asymptotically equal in correctly specified models.

Author(s)

Brice Ozenne

bozenne/riskRegressionLight documentation built on May 7, 2019, 12:52 a.m.