seqMI: Inference using MI-based sequential based method, under J2R

In Merck/mmrm: Mixed model repeated measures and its extension

Description

Given a wide-form data, compute the estimate of the ATE under J2R using MI-based sequential method.

Usage

seqMI(
  seq_formula,
  data_wide,
  trt_name,
  M,
  fit_model,
  id_name,
  estimator = "reg with interaction",
  bootstrap = FALSE,
  B = 100
)

Arguments

seq_formula	a list of formulas for sequential regression
data_wide	wide-form of the data frame
trt_name	name of the treatment variable
M	imputation size
fit_model	type of the analysis model. Available: lm, Rfit::rfit (rank regression), MASS::rlm (robust regression)
id_name	name of the id variable
estimator	type of the estimator. Available: sample mean, reg with interaction, reg without interaction. Default: reg with interaction
bootstrap	Whether to use bootstrap variance
B	the number of bootstrap replicate

Value

A list of MI estimation of the ATE

• ate_estMI point estimation of the ATE

• var_rubinMI variance estimation by Rubin's rule

• var_bootMI variance estimation by nonparametric bootstrap

Merck/mmrm documentation built on Dec. 17, 2021, 4:12 a.m.