genRCT: Generalizing the average treatment effect (ATE) from the...

In idasomm/genRCT: Generalizing the average treatment effect (ATE) from the trial using observational studies (OS)

Generalizing the average treatment effect (ATE) from the trial using observational studies (OS)

Description

Provides estimation and inference for the average treatment effect (ATE), generalizing from a randomized controlled trial (RCT) to a target population leveraging observational studies. The method of sieves can be used for the estimation of sampling score and outcome models. The ATE can be estimated with up to 6 methods among the following: Naive, IPSW, AIPSW, CW, ACW-t, ACW-b.

Usage

genRCT(
  Y.trial,
  A.trial,
  X.trial,
  Y.rwe,
  A.rwe,
  X.rwe,
  family = "gaussian",
  estimators = c("Naive", "IPSW", "AIPSW", "CW", "ACW-t", "ACW-b"),
  sieve = TRUE,
  inference = TRUE,
  n.boot = 100,
  conf.level = 0.05,
  seed = NULL,
  plot.boot = TRUE,
  verbose = TRUE
)

Arguments

Y.trial	Observed outcome from a trial; vector of size n (trial sample size).
A.trial	Treatment received from a trial; vector of size n.
X.trial	Matrix of p baseline covariates from a trial; dimension n by p.
Y.rwe	Observed outcome from OS; if obtained, vector of size m (OS sample size); otherwise, set Y.rwe = NULL.
A.rwe	Treatment received from OS; if obtained, vector of size m; otherwise, set A.rwe = NULL.
X.rwe	Matrix of p baseline covariates from OS; dimension m by p.
family	The type of outcome; 'gaussian' for gaussian regression or 'binomial' for logistic regression. Default is "gaussian".
estimators	A vector of one or multiple methods to estimate the ATE. Allowed values are 'Naive', 'IPSW', 'AIPSW', 'CW', 'ACW-t', 'ACW-b'. The 'ACW-b' is allowed only when both Y.rwe and A.rwe are obtained. Default specifies all 6 methods.
sieve	A logical value indicating whether the method of sieves are used for estimating sampling score and outcome models. Used only if estimators = 'AIPSW' or 'ACW-t' or 'ACW-b'. Default is TRUE.
inference	A logical value indicating whether inference for the ATE via bootstrap should be provided. Default it TRUE.
n.boot	A numeric value indicating the number of bootstrap samples used. This is only relevant if inference = TRUE. Default is 100.
conf.level	The level of bootstrap confidence interval; Default is 0.05.
seed	An optional integer specifying an initial randomization seed for reproducibility. Default is NULL, corresponding to no seed.
plot.boot	A logical value indicating whether histograms of the bootstrap samples should be produced. Default is TRUE.
verbose	A logical value indicating whether intermediate progress messages should be printed. Default is TRUE.

Value

Return a list containing:

fit

idasomm/genRCT documentation built on April 15, 2023, 7:16 a.m.