search

guido-s / meta

Official Git repository of R package meta
http://cran.r-project.org/web/packages/meta/index.html
GNU General Public License v2.0
82 stars 32 forks source link

fix seTE calc bugs when method.smd == Glass #30

Closed jbkang closed 3 years ago

jbkang commented 3 years ago

Jiang Yingda and me found a small issue of seTE in metacont api, while calculate seTE of individual studies when method.smd = Glass. which TE^2 / (2 n.g - 1) should be TE^2 / (2 (n.g - 1)) we believe this is a bug of R with below reference.

The first reference is: Marfo, Philomena, and G. A. Okyere. "The accuracy of effect-size estimates under normals and contaminated normals in meta-analysis." Heliyon 5.6 (2019): e01838. image

The second reference is the original paper: Hedges, Larry V. "Distribution theory for Glass's estimator of effect size and related estimators." journal of Educational Statistics 6.2 (1981): 107-128. image

Here k = 1 for a single study; 1/ñ is the first term; m in the second term is actually n2-1.

The 3rd reference is from "STATA META-ANALYSIS REFERENCE MANUAL" RELEASE 16, on page 87, it state the formula of Variance is as below image

guido-s commented 3 years ago

Thank you for pointing out this bug.