Skip to main content

Table 2 Parameters or factors varied in previous simulation studies

From: Heterogeneity estimation in meta-analysis of standardized mean differences when the distribution of random effects departs from normal: A Monte Carlo simulation study

Study

Heterogeneity estimators

Overall effect

(\({\mu }_{\theta }\))

Amount of heterogeneity

(\({\tau }^{2}\))

Number of studies

(\(k\))

Average sample size (\(N\))

Replicat

Viechtbauer [21]

CA, DL, HSiv, ML, REML

0, 0.2, 0.5, 0.8

0, 0.01, 0.025, 0.05, 0.1

5, 10, 20, 40, 80

20, 40, 80, 160, 320

100,000

Novianti et al. [43]

CA, MP, DL, REML, SJ, SJ(CA), CA2, DL2

0, 0.5

From 0 to 0.0366

10, 15, 20, 30, 50

From 40 to 400

10,000

Petropoulou and Mavridis [44]

CA, MP, DL, HSiv, ML, AREML, REML, SJ, SJ(CA), CA2, DL2, HM, MBH, DLp, DLb, BM, RB, RBp, FBinformative, FBvague

0, 0.3, 0.5, 0.8

0, 0.01, 0.05, 0.5

10, 20, 30, 50

From 40 to 400

1,000

Langan et al. [45]

CA, MP, DL, CA2, DL2, HM, SJ, SJ(CA), REML

0.5

From 0 to 2.44

2, 3, 5, 10, 20, 30, 50, 100

40, 220, 400, 1520, 3000

5,000

Boedeker and Henson [30]

CA, CA2, DL, DLp, DL2, HM, HSiv, SJ, MBH, PM, ML, and REML, FBmean, FBmedian, FBmode, RB, RBp, BM

0.5

0, 0.01, 0.025, 0.05, 0.075, 0.1, 0.25, 0.5, 1

5, 10, 20, 40, 60, 100

40, 80, 120

1,000

  1. Simulation factors included in previous simulation studies assessing the performance of different heterogeneity estimators under normal random-effects conditions. Within the fully Bayesian framework, FBinformative corresponds to the prior specification \({\tau }^{2}\sim logN(-2.56, {1.74}^{2})\); FBvague corresponds to the prior specification \(\mathrm{log}({\tau }^{2})\sim t(-3.44, {2.59}^{2}, 5)\); and FBmode, FBmean, FBmedian correspond to the fully Bayesian procedures also evaluated in the present simulation study to model a function of heterogeneity (five different priors for \(\tau\), and other two for \(1/{\tau }^{2}\)) centered on the posterior mean, median and mode, respectively