bias_test: Test of Large Bias

In koohyun-kwon/OptACI: Methods to Construct Optimal Average Coverage Confidence Intervals

Description

Tests whether the bias of the average coverage probability function estimated by the series approximation is too large.

Usage

bias_test(lam, xvec, Jn, alpha, tol = alpha/2)

Arguments

lam	a vector of shrinkage factors λ.
xvec	normalized observed outcome vector, corresponding to z_i in the paper.
Jn	the order of polynomial approximation.
alpha	the desired level of non-coverage probability
tol	a threshold value to test for the large bias; the default is alpha/2.

Details

The bias is deemed large if the upper bound bias estimate computed by gnB_bias is large than tol.

Value

a vector of logical values, each value taking TRUE if estimated bias is smaller than tol, with each value corresponding to each value of λ in lam.

Examples

xvec <- rnorm(50)
lam <- c(0.3, 0.35, 0.4, 0.45)
bias_test(lam, xvec, 2, 0.05)

koohyun-kwon/OptACI documentation built on Oct. 6, 2020, 8:09 a.m.