Skip to contents

Recomputes cpaic_ranks() over a grid of target populations, so that the population dependence of the hierarchy is visible. Under population adjustment a component's rank is a function of the target, and this is the object that shows it.

Usage

rank_curve(
  object,
  em,
  values,
  at = NULL,
  what = c("treatment", "component"),
  lower_is_better = FALSE,
  ...
)

Arguments

object

A cmlnmr() fit.

em

Name of the effect modifier to vary.

values

Numeric vector of target values for em.

at

Optional named vector fixing the other effect modifiers. Defaults to 0 for each.

what, lower_is_better

See cpaic_ranks().

...

Unused.

Value

A data frame with one row per (element, target value), giving sucra, mean_rank, p_best and estimate, plus estimable.

See also

Examples

if (FALSE) {
rank_curve(fit, em = "x1", values = seq(-1, 1, by = 0.25), what = "component")
}