.perm_eval_one

Computes the test statistic for a single permutation of weights.

internal

Provides diagnostic tests for assessing the informativeness of survey weights in regression models. Implements difference-in-coefficients tests (Hausman 1978 <doi:10.2307/1913827>; Pfeffermann 1993 <doi:10.2307/1403631>), weight-association tests (DuMouchel and Duncan 1983 <doi:10.2307/2288185>; Pfeffermann and Sverchkov 1999 <https://www.jstor.org/stable/25051118>; Pfeffermann and Sverchkov 2003 <ISBN:9780470845672>; Wu and Fuller 2005 <https://www.jstor.org/stable/27590461>), estimating equations tests (Pfeffermann and Sverchkov 2003 <ISBN:9780470845672>), and non-parametric permutation tests. Includes simulation utilities replicating Wang et al. (2023 <doi:10.1111/insr.12509>) and extensions.

Corbin Lubianski

svytest

Survey Weight Diagnostic Tests

.perm_eval_one function

<dl><dt>idx</dt>
<dd>Integer index vector for permuted weights.</dd>
<dt>X</dt>
<dd>Numeric design matrix.</dd>
<dt>y</dt>
<dd>Numeric response vector.</dd>
<dt>w_use</dt>
<dd>Normalized weight vector.</dd>
<dt>fit_null_beta</dt>
<dd>Baseline coefficient vector.</dd>
<dt>fit_null_XtX</dt>
<dd>Baseline precision matrix.</dd>
<dt>stat</dt>
<dd>Character string, statistic type.</dd></dl>

Arguments

Internal helper: Evaluate one permutation — .perm_eval_one

<dl>

<dt>idx</dt>
<dd>Integer index vector for permuted weights.</dd>


<dt>X</dt>
<dd>Numeric design matrix.</dd>


<dt>y</dt>
<dd>Numeric response vector.</dd>


<dt>w_use</dt>
<dd>Normalized weight vector.</dd>


<dt>fit_null_beta</dt>
<dd>Baseline coefficient vector.</dd>


<dt>fit_null_XtX</dt>
<dd>Baseline precision matrix.</dd>


<dt>stat</dt>
<dd>Character string, statistic type.</dd>

</dl>

.perm_eval_one: Internal helper: Evaluate one permutation

Description

Usage

Value

Arguments