wDefficiency {docopulae}

Weighted D Efficiency

Description

wDefficiency computes the weighted D-, D_s or D_A-efficiency measure for a design with respect to a reference design.

Usage

wDefficiency(des, ref, mods, modw, A = NULL, parNames = NULL)

Arguments

des

a design.

ref

a design, the reference.

mods

a list of models.

modw

a vector of weights.

A

for

• D-efficiency: NULL

• D_s-efficiency: a vector of names or indices, the subset of parameters of interest.

• D_A-efficiency: either

  • directly: a matrix without row names.

  • indirectly: a matrix with row names corresponding to the parameters.

parNames

a vector of names or indices, the subset of parameters to use. Defaults to the parameters for which the Fisher information is available.

Details

Indices supplied to argument A correspond to the subset of parameters defined by argument parNames.

Weighted D efficiency is defined as

\left(\frac{\exp\int_{\mathcal{B}}\log\left|M(\xi,\bar{\boldsymbol{\theta}})\right|\mathrm{d}B}{\exp\int_{\mathcal{B}}\log\left|M(\xi^{*},\bar{\boldsymbol{\theta}})\right|\mathrm{d}B}\right)^{1/n}

and weighted D_A efficiency as

\left(\frac{\exp\int_{\mathcal{B}}\log\left|A^{T}M(\xi,\bar{\boldsymbol{\theta}})^{-1}A\right|^{-1}\mathrm{d}B}{\exp\int_{\mathcal{B}}\log\left|A^{T}M(\xi^{*},\bar{\boldsymbol{\theta}})^{-1}A\right|^{-1}\mathrm{d}B}\right)^{1/s}

Value

wDefficiency returns a single numeric.

See Also

design, param, Defficiency