multi_estimate

Computes the domain-specific pseudo-weighted Hájek mean and its
Taylor-linearized variance for the multi-reference raking estimator.

internal

Provides pseudo-weighted estimates of means and
prevalences for finite population inference from nonprobability
samples using auxiliary information from one or multiple probability
reference surveys. The package supports estimation with multiple
reference surveys, allowing auxiliary information to be combined
when no single survey contains all variables relevant to
participation. Optional cumulative precalibration can be applied to
align weighted totals of shared variables across surveys. Methods
are based on the generalized estimating equations framework of
Landsman et al. (2026) <doi:10.1002/sim.70403> for correcting
participation bias. For a single reference survey, the package
implements the raking ratio calibration method and includes the
adjusted logistic propensity (ALP) method of Wang, Valliant, and Li
(2021) <doi:10.1002/sim.9122>, as well as the Chen-Li-Wu (CLW)
method of Chen, Li, and Wu (2020) <doi:10.1080/01621459.2019.1677241>.
Analytic variance estimation uses Taylor linearization and accounts
for complex sampling designs in the reference surveys via integration
with the 'survey' package.

Jiakun Lin

nonprobsampling

Inference for Nonprobability Samples Using Multiple Reference
Surveys

Victoria Landsman

Aya A. Mitani

multi_estimate function

<dl><dt>Y</dt>
<dd>Outcome vector for the nonprobability sample.</dd>
<dt>Z</dt>
<dd>Domain indicator vector. Use `rep(1, length(Y))` for the
overall mean.</dd>
<dt>w</dt>
<dd>Estimated pseudo-weights from the multi-reference raking
build step.</dd>
<dt>X</dt>
<dd>Design matrix for the nonprobability sample.</dd>
<dt>D</dt>
<dd>Block-diagonal design-based variance-covariance matrix of the
estimated auxiliary totals from the reference surveys.</dd>
<dt>S_beta</dt>
<dd>Sensitivity matrix for the multi-reference raking
estimating equations, typically <code>t(w * X) %*% X</code>.</dd></dl>

Arguments

Estimate step for multi-reference raking — multi_estimate

<dl>

<dt>Y</dt>
<dd>Outcome vector for the nonprobability sample.</dd>


<dt>Z</dt>
<dd>Domain indicator vector. Use `rep(1, length(Y))` for the
overall mean.</dd>


<dt>w</dt>
<dd>Estimated pseudo-weights from the multi-reference raking
build step.</dd>


<dt>X</dt>
<dd>Design matrix for the nonprobability sample.</dd>


<dt>D</dt>
<dd>Block-diagonal design-based variance-covariance matrix of the
estimated auxiliary totals from the reference surveys.</dd>


<dt>S_beta</dt>
<dd>Sensitivity matrix for the multi-reference raking
estimating equations, typically <code>t(w * X) %*% X</code>.</dd>

</dl>

multi_estimate: Estimate step for multi-reference raking

Description

Usage

Value

Arguments