Check plausibility of a reported mean with the GRIM test (Brown & Heathers 2007)

Performs the GRIM (Granularity-Related Inconsistency of Means) test to assess whether a reported mean is numerically possible given the sample size.

Usage

check_grim(n, target_mean, decimals, tol.r = .Machine$double.eps^0.5)

Arguments

n

Integer. Sample size; a positive whole number.

target_mean

Numeric. Reported mean to be tested for plausibility.

decimals

Integer. Number of decimal places in the reported mean.

tol.r

Numeric. Tolerance for rounding errors; a non-negative value. Default is the square root of machine double precision epsilon.

Value

A list with components:

test

Logical. TRUE if the reported mean is plausible.

grim_mean

Numeric. The adjusted mean that is numerically plausible (rounded to `decimals`).

Examples

if (FALSE) { # \dontrun{
check_grim(10, 3.7, 1)
} # }