Skip to contents

Compute only the empirical-null distribution for split-half

Source: R/rel_split_half.R
rel_split_half_null.Rd

Standalone access to the null-distribution simulation used by rel_split_half(null = ...). Useful when the user wants to precompute and reuse a null across multiple observed analyses (e.g., when comparing across conditions with the same producer count).

Usage

rel_split_half_null(
  n_participants,
  n_pixels,
  null = c("permutation", "random_responders"),
  noise_matrix = NULL,
  n_permutations = 2000L,
  seed = NULL,
  pid = NULL
)

Arguments

n_participants

Integer.

n_pixels

Integer.

null

Either "permutation" or "random_responders".

noise_matrix

Required for "random_responders".

n_permutations

Integer. Default 2000.

seed

Optional integer.

pid

Optional internal progress-bar id; end users should leave at NULL.

Value

Numeric vector of length n_permutations.

Examples

if (FALSE) { # \dontrun{
null_dist <- rel_split_half_null(
  n_participants = 20L,
  n_pixels       = 32L * 32L,
  null           = "permutation",
  n_permutations = 200L,
  seed           = 1
)
quantile(null_dist, c(0.025, 0.975))
} # }