Extract Posterior Mean Score Matrix from an hbb_fit

Extracts the posterior means of the Stan-generated score variables (score_ext, score_int, score_kappa) and assembles them into an \(N \times D\) matrix, where \(D = 2P + 1\): \((\alpha_1, \ldots, \alpha_P, \beta_1, \ldots, \beta_P, \log\kappa)\).

Usage

compute_score_matrix(fit)

Arguments

fit

An object of class "hbb_fit". Must be a weighted model variant ("weighted" or "svc_weighted").

Value

Numeric matrix of dimension \(N \times D\). Columns are ordered as (score_ext columns 1 to P, score_int columns 1 to P, score_kappa).

Details

Score variables are generated quantities in the weighted/svc_weighted Stan models. CmdStanR names matrix variables as var[1,1], var[1,2], ..., var[1,P], var[2,1], ... (first index varies slowest, row-major ordering). Posterior means are computed across all MCMC draws.

See also

sandwich_variance()

Other sandwich: compute_H_obs(), compute_J_cluster(), compute_der(), print.hbb_sandwich(), sandwich_variance()