Confidence intervals for plqr_result (experimental)

Confidence intervals for plqr_result (experimental)

Usage

# S3 method for class 'plqr_result'
confint(object, parm = NULL, level = 0.95, target = NULL, ...)

Arguments

object

A plqr_result object from plqr_dnn().

parm

Optional coefficient names or indices.

level

Confidence level.

target

Optional target indices.

...

Not used.

Value

A data frame with confidence intervals and a theta_source attribute.

Details

Coefficient-level standard errors and intervals require the direct PLQR sandwich (fit with density or density_col); the moment-level jackknife SE available on fits without a density input has no proven translation into a coefficient-level interval in this package's source theory — see markdown_notes/design/15_plqr_variance_estimation_rationale.md.

Intervals are centered at the coefficient value the object represents: for current fits the supplied theta_eval (meta$theta_source = "supplied"), not an estimated local root. The source is attached as attr(., "theta_source").

Lifecycle

Experimental. PLQR functionality is outside the CATE and regression workflow focus of the initial public release. Its API and inference interfaces may change in later releases.

See also

plqr_dnn().