bootsens.md: Sensitivity analysis for missing data

In qingyuanzhao/bootsens: Bootstrapping sensitivity analysis

Description

Sensitivity analysis for missing data

Usage

bootsens.md(
  A,
  X,
  Y,
  gamma = 0,
  alpha = 0.05,
  estimand = c("all", "missing"),
  reg.adjust = FALSE,
  parallel = TRUE,
  B = 1000,
  warm.start = FALSE
)

extrema.md(
  A,
  X,
  Y,
  gamma = 0,
  estimand = c("all", "missing"),
  reg.adjust = FALSE,
  start = NULL
)

Arguments

A	Indicator of missingness
X	A matrix of covariates
Y	Outcome
gamma	Sensitivity parameter (log odds ratio)
alpha	Significance level
estimand	Target estimand, either E[Y] ("all") or E[Y|A=0] ("missing")
reg.adjust	Should regression adjustment (augmented IPW) be used?
parallel	Should parallel computing be used?
B	Number of Bootstrap resamples.
warm.start	Warm start the refitting of propensity score model (doesn't seem to help).
start	Starting values for the propensity score model to warm start glm.

Value

A (1 - alpha) confidence interval.

Extrema (an interval).

Functions

• extrema.md: Obtain extrema of IPW estimator for missing data

qingyuanzhao/bootsens documentation built on Jan. 17, 2021, 7:40 p.m.