R: Extract the H-decomposition of the Estimator

getIid {BuyseTest}

R Documentation

Extract the H-decomposition of the Estimator

Description

Extract the H-decomposition of the GPC estimator.

Usage

getIid(
  object,
  endpoint = NULL,
  statistic = NULL,
  strata = FALSE,
  cumulative = TRUE,
  center = TRUE,
  scale = TRUE,
  type = "all",
  cluster = NULL,
  simplify = FALSE
)

## S4 method for signature 'S4BuyseTest'
getIid(
  object,
  endpoint = NULL,
  statistic = NULL,
  strata = FALSE,
  cumulative = TRUE,
  center = TRUE,
  scale = TRUE,
  type = "all",
  cluster = NULL,
  simplify = TRUE
)

Arguments

`object`	an R object of class `S4BuyseTest`, i.e., output of `BuyseTest`
`endpoint`	[character] for which endpoint(s) the H-decomposition should be output? If `NULL` returns the sum of the H-decomposition over all endpoints.
`statistic`	[character] statistic relative to which the H-decomposition should be output.
`strata`	[character vector] the strata relative to which the H-decomposition of the statistic should be output. Can also be `"global"` or `FALSE` to output the H-decompositon of the pooled statistic. or `TRUE` to output the H-decompositon of each strata-specific statistic.
`cumulative`	[logical] should the H-decomposition be cumulated over endpoints? Otherwise display the contribution of each endpoint.
`center`	[logical] if `TRUE` the H-decomposition is centered around 0 (estimated statistic is substracted).
`scale`	[logical] if `TRUE` the H-decomposition is rescaled (by the sample size in the corresponding arm) such that its sums of squares approximate the variance of the estimator.
`type`	[character] type of H-decomposition to be output. Can be only for the nuisance parameters (`"nuisance"`), or for the u-statistic given the nuisance parameters (`"u-statistic"`), or both.
`cluster`	[numeric vector] return the H-decomposition aggregated by cluster.
`simplify`	[logical] should the result be coerced to the lowest possible dimension?

Details

WARNING: argument scale and center should be used with care as when set to FALSE they may not lead to a meaningful decomposition.

Value

A list of matrices, each element of the list correspond to a statistic (global or strata-specific) and each matrix has as many columns as endpoints and rows as observations.

Author(s)

Brice Ozenne