adpt.res.gen: Length Generation for One-sided CIs

In koohyun-kwon/rdadapt:

Length Generation for One-sided CIs

Description

Calculate excess lengths of one-sided CIs for different values of true Lipschitz coefficients.

Usage

adpt.res.gen(
  eps,
  CI.method,
  th,
  spec.etc,
  which.fun,
  n,
  sig.scale,
  which.spec = 1
)

Arguments

eps	N(0,1) random noises.
CI.method	method to be used to generate the CI.
th	true RD parameter value.
spec.etc	list of other specification parameters.
which.fun	which regression function to evaluate; supports 1:6.
n	sample size.
sig.scale	scaling parameter for the standard deviation.
which.spec	which design is used; the default is which.spec = 1.

Details

Modification for multi-dim hasn't been made yet.

Value

vector of excess lengths and coverage indicators corresponding to true Lipschitz coefficient values contained in spec.etc$C.true.vec.

Examples

spec.etc <- list(alpha = 0.05, mon_ind = 1,
C.small = 1, C.large = 3, se.method.RDH = "nn", se.initial.RDH = "nn", M.RDH = 1,
se.method = "nn", se.init = "Silverman", t.dir = "left", C.l = 1/2, C.u = 2, C = 3,
C.true.vec = seq(1/2, 3, length.out = 2), adpt.orc = TRUE)
n <- 100
th <- 1
eps <- rnorm(n)
which.fun <- 1
sig.scale <- 1/2
CI.method <- "adpt.one"
adpt.res.gen(eps, CI.method, th, spec.etc, which.fun, n, sig.scale)
spec.etc$adpt.orc <- FALSE
adpt.res.gen(eps, CI.method, th, spec.etc, which.fun, n, sig.scale)

koohyun-kwon/rdadapt documentation built on May 8, 2022, 8:49 p.m.