sigma_from_family

For a GLM family with variance function V(mu), link function g, and
dispersion parameter phi, the sampling standard deviation of the MLE
on the link scale is:
sigma(eta) = sqrt(phi * V(mu)) / |dmu/deta|
where mu = linkinv(eta + offset).

internal

Tool-set to support Bayesian evidence synthesis. This
includes meta-analysis, (robust) prior derivation from historical
data, operating characteristics and analysis (1 and 2 sample
cases). Please refer to Weber et al. (2021) <doi:10.18637/jss.v100.i19>
for details on applying this package while Neuenschwander et al. (2010)
<doi:10.1177/1740774509356002> and Schmidli et al. (2014)
<doi:10.1111/biom.12242> explain details on the methodology.

Sebastian Weber

RBesT

R Bayesian Evidence Synthesis Tools

Novartis Pharma AG

Beat Neuenschwander

Heinz Schmidli

Baldur Magnusson

Yue Li

Satrajit Roychoudhury

Lukas A. Widmer

Daniel Sabanés Bové

Trustees of Columbia University

sigma_from_family function

<dl><dt>eta</dt>
<dd>Numeric vector of values on the link scale (parameter of interest).</dd>
<dt>family</dt>
<dd>An R family object (e.g., binomial(), MASS::negative.binomial(0.5)).</dd>
<dt>offset</dt>
<dd>Numeric scalar. Shift applied before evaluating the variance
function. For NB/Poisson with log link, this is log(exposure). Default 0.</dd>
<dt>phi</dt>
<dd>Numeric scalar. Dispersion parameter. For gaussian family this is
sigma^2 (user-supplied). For binomial/NB/Poisson this is 1 (the default).</dd></dl>

Arguments

Compute sigma as a function of eta for a GLM family — sigma_from_family

<dl>

<dt>eta</dt>
<dd>Numeric vector of values on the link scale (parameter of interest).</dd>


<dt>family</dt>
<dd>An R family object (e.g., binomial(), MASS::negative.binomial(0.5)).</dd>


<dt>offset</dt>
<dd>Numeric scalar. Shift applied before evaluating the variance
function. For NB/Poisson with log link, this is log(exposure). Default 0.</dd>


<dt>phi</dt>
<dd>Numeric scalar. Dispersion parameter. For gaussian family this is
sigma^2 (user-supplied). For binomial/NB/Poisson this is 1 (the default).</dd>

</dl>

Compute sigma as a function of eta for a GLM family

sigma_from_family: Compute sigma as a function of eta for a GLM family

Description

Usage

Value

Arguments