Extract Fixed-Effect Coefficients from an hbb_fit

Returns the posterior mean of the fixed-effect parameter vector \(\hat\theta = (\hat\alpha_1, \ldots, \hat\alpha_P,\; \hat\beta_1, \ldots, \hat\beta_P,\; \widehat{\log\kappa})\), optionally restricted to a single margin.

Usage

# S3 method for class 'hbb_fit'
coef(object, margin = c("both", "extensive", "intensive"), ...)

Arguments

object

An object of class "hbb_fit" returned by hbb.

margin

Character string specifying which parameters to return:

"both": (Default.) The full \(D = 2P + 1\) parameter vector.
"extensive": Only \(\hat\alpha_{1:P}\) (P-vector).
"intensive": Only \(\hat\beta_{1:P}\) (P-vector).

...

Currently unused; included for S3 method consistency.

Value

A named numeric vector of posterior means. Length depends on margin: \(2P + 1\) for "both", \(P\) for "extensive" or "intensive".

Mathematical note

The posterior mean is the Bayes estimator under squared-error loss. For survey-weighted pseudo-posteriors, the posterior mean converges to the pseudo-true parameter under standard regularity conditions (Williams and Savitsky, 2021).

Examples

if (FALSE) { # \dontrun{
fit <- hbb(y | trials(n_trial) ~ poverty + urban, data = my_data)
coef(fit)                       # full D-vector
coef(fit, margin = "extensive") # alpha only
coef(fit, margin = "intensive") # beta only
} # }