DFFITS for brma Objects

Computes DFFITS (Difference in FITS, standardized) for a fitted brma object. DFFITS measures how much the fitted value for observation \(i\) changes if observation \(i\) is removed, standardized by the estimated standard error of the fit.

Usage

# S3 method for class 'brma'
dffits(model, ...)

Arguments

model

a fitted normal-outcome brma object without a weightfunction component.

...

additional arguments (currently ignored).

Value

A named numeric vector of DFFITS values, one for each observation.

Details

DFFITS values are computed as a PSIS leave-one-out deletion diagnostic. For each observation \(i\), the leave-one-out posterior mean fitted value at that observation is estimated with normalized PSIS weights and compared to the full-posterior fitted value: $$DFFITS_i = \frac{\hat{\mu}_i - \hat{\mu}_{i(-i)}}{SD_{(-i)}(\mu_i)}$$

This targets deletion influence on fitted values directly. It does not use LOO-PIT residuals, which are predictive outlier diagnostics rather than fitted-value deletion diagnostics.

Estimate-unit LOO must first be computed with model <- add_loo(model, unit = "estimate"). If the leave-one-out posterior SD of a fitted value is near zero, the corresponding DFFITS value is returned as NA.

See also

influence.brma, cooks.distance.brma, hatvalues.brma

Examples

if (FALSE) { # \dontrun{
if (requireNamespace("metadat", quietly = TRUE)) {
  data(dat.lehmann2018, package = "metadat")
  fit <- bPET(yi = yi, vi = vi, data = dat.lehmann2018, measure = "SMD")
  fit <- add_loo(fit)

  dffits(fit)
}
} # }