Per-subject influence functions for misskappa estimates — influence.misskappa

`stats::influence()` returns the same per-subject influence-function matrix stored in `fit$psi`. Rows are subjects and columns are coefficients, with scaling $$\mathrm{vcov}(\hat\theta) = n^{-2} \Psi^\top \Psi.$$

This is the public S3 adapter for R's `stats::influence()` generic; `fit$psi` remains the stored component used internally by `joint_vcov()` and the equality-test helpers.

Usage

# S3 method for class 'misskappa_estimate'
influence(model, ...)

Arguments

model: A `misskappa_estimate` object.
...: Unused; present for S3 generic conformance.

Value

A numeric matrix of per-subject influence-function rows.

Examples

fit <- kappa(dat.gwet2014, estimator = "ipw")
#> Warning: rater pair(s) rater4-rater5 co-observed by only one subject; the corresponding pairwise covariance is degenerate and the standard error unreliable.
psi <- stats::influence(fit)
all.equal(vcov(fit), crossprod(psi) / nrow(psi)^2)
#> [1] TRUE