Posterior probability of consistent intervention direction across cases

For each fixed-effect term with a corresponding unit-level random effect, returns the posterior probability that the unit-specific effect (b plus the random effects of the unit and every enclosing grouping, as in summarize_per_case_effects()) points in the expected direction for every unit, from the joint posterior of the unit effects.

Usage

summarize_cross_case_consistency(
  model,
  case_var = "case",
  ci_level = 0.95,
  intervention_goal = c("increase", "decrease")
)

Arguments

model

A fitted brmsfit.

case_var

Name of the case-level grouping column in the model (default "case").

ci_level

Credible interval width (default 0.95).

intervention_goal

"increase" (default) or "decrease". The expected direction of a term is the goal direction, reversed for reversal withdrawal steps (see summarize_fixed_effect_decisions()).

Value

A data frame with columns term, prob_all_positive, prob_all_negative (both on the native coefficient scale), and prob_consistent, the probability that every case changes in the expected direction (prob_all_positive when that direction is an increase, prob_all_negative when it is a decrease).

Examples

if (interactive()) {
  dat <- simulate_scd_data()
  fit <- brms::brm(outcome ~ phase + (1 + phase | case),
                   data = dat, refresh = 0)
  summarize_cross_case_consistency(fit)
}