targetQg: targetQg

In benkeser/drinf: Computes doubly-robust estimators and doubly-robust confidence intervals

targetQg

Description

Function that targets estimates of Q and g to solve the EIF + remainder approximations that result from misspecification. Fits a single logistic fluctuation regression of c(A0, A1, Q2n, L2) on relevant offsets and so-called clever covariates.

Usage

targetQg(A0, A1, L2, Qn, gn, Qnr.gnr, abar, tolg, tolQ, return.models,
  offset.model = TRUE, allatonce = FALSE, ...)

Arguments

A0	A vector treatment delivered at baseline.
A1	A vector treatment deliver after L1 is measured.
L2	A vector outcome of interest.
abar	A vector of length 2 indicating the treatment assignment that is of interest.
tolg	A numeric indicating the truncation level for conditional treatment probabilities.
tolQ	A numeric
return.models	A boolean indicating whether the fluctuation model should be returned with the output.
offset.model	A boolean indicating whether to fit a fluctuation model with an offset term (if TRUE) or intercept-only with weights (if FALSE)
allatonce	Do all the targeting with a single fit; if FALSE, then target g, followed by Q
Q2n	A vector of estimates of Q_2,0
Q1n	A vector of estimates of Q_1,0
g1n	A vector of estimates of g_1,0
g0n	A vector of estimates of g_0,0
Q2nr.obsa	A vector of estimates of Q_2,0^r evaluated at observed A0 value
Q2nr.seta	A vector of estimates of Q_2,0^r evaluated at A0 = abar[1]
Q1nr	A vector of estimates of Q_1,0^r
g0nr	A vector of estimates of g_0,0^r
g1nr	A vector of estimates of g_0,0^r
h0nr	A vector of estimates of h_0,0^r
h1nr	A vector of estimates of h_1,0^r
hbarnr	A vector of estimates of h^r, the iterated reduced dimension regression.

Value

A list with named entries corresponding to the estimators of the fluctuated nuisance parameters evaluated at the observed data values. If return.models = TRUE output also includes the fitted fluctuation model.

benkeser/drinf documentation built on June 10, 2025, 11:25 p.m.