correct_glass_bias: Correct for small-sample bias in Glass' Delta values

In psychmeta: Psychometric Meta-Analysis Toolkit

Correct for small-sample bias in Glass' Δ values

Description

Correct for small-sample bias in Glass' Δ values.

Usage

correct_glass_bias(delta, nc, ne, use_pooled_sd = rep(FALSE, length(delta)))

Arguments

delta	Vector of Glass' Δ values.
nc	Vector of control-group sample sizes.
ne	Vector of experimental-group sample sizes.
use_pooled_sd	Logical vector determining whether the pooled standard deviation was used (TRUE) or not (FALSE; default).

Details

The bias correction is estimated as:

delta_c = delta * gamma((nc - 1) / 2) / (sqrt((nc - 1) / 2) * gamma((nc - 2) / 2))

where Δ is the observed effect size, Δ_c is the corrected estimate of Δ, nc is the control-group sample size, and gamma() is the gamma function.

Value

Vector of d values corrected for small-sample bias.

References

Hedges, L. V. (1981). Distribution theory for Glass’s estimator of effect size and related estimators. Journal of Educational Statistics, 6(2), 107–128. doi: 10.2307/1164588

Examples

correct_glass_bias(delta = .3, nc = 30, ne = 30)

psychmeta documentation built on Aug. 26, 2022, 5:14 p.m.