Accounting for multiple testing, is there a group with statistically significant specified sign effect

test_any_sign_effect(
  obj,
  check_negative = T,
  method = "fdr",
  alpha = 0.05,
  n_sim = 500
)

Arguments

obj

an estimated_partition object

check_negative

If true, check for a negative. If false, check for positive.

method

one of c("fdr", "sim_mom_ineq"). fdr is conservative. sim_mom_ineq Need samples sizes to sufficiently large so that the effects are normally distributed

alpha

alpha

n_sim

n_sim

Value

list(are_any= boolean of whether effect is negative)