I am conducting a meta-analysis of meta-analyses, and one thing we want to do is compare the degree of publication bias between meta-analyses. Meta-analyses report publication bias in dozens of different ways (Egger Z statistic, Egger T tests, Egger Alpha, Egger D, Egger Beta (intercept), Egger's test (p), G2, Moderator Analysis (slope), Moderator Analysis (Q), Moderator I^2, Moderator t, Fail Safe N, P-curve value Z, Bayers Factor, Fisher's exact test, PET (p) and PEESE (p), PET (B) and PEESE (B), Begg's (z), Trim and Fill, Funnel Plot Regression Slope, Begg Rank Correlation Ranked Variances t, Funnel Symmetry/Asymmetry, Rank test p, Begg's Kendall's t, Significant Likelihood Ratio Test X^2, Peters' Significant, Regression test t).
Is there a way to convert these various metrics to a common metric for comparison in the same way various effect sizes can be converted to an r for comparison? If not all of them, can at least some of them be converted? For example, are Egger Z and Begg's z equivalent to a Z-score? And is an Egger D equivalent to a Cohen's d?
Are there any good papers that discuss and compare various publication bias metrics?
Thanks, Michael Anthony Woodley of Menie ! We have done this coding, but we were hoping for something more precise than that (if possible... but it is seeming like it might not be possible).
You pointed an interesting and challenging topic. I also agree that it would be a huge challenge to regroup different indicators especially because of their difference in calculation methods and assumptions. An interesting to do/think would be the feasability of assessing the correlation between these different tests when multiple ways of publication bias are assessed.
In case it would help you (but maybe you already look at these):
Lin L, Chu H, Murad MH, Hong C, Qu Z, Cole SR, Chen Y. Empirical Comparison of Publication Bias Tests in Meta-Analysis. J Gen Intern Med. 2018 Aug;33(8):1260-1267. doi: 10.1007/s11606-018-4425-7. Epub 2018 Apr 16. PMID: 29663281; PMCID: PMC6082203.
Macaskill P, Walter SD, Irwig L. A comparison of methods to detect publication bias in meta-analysis. Stat Med. 2001 Feb 28;20(4):641-54. doi: 10.1002/sim.698. PMID: 11223905.
- Badges
- Science topic
- Similar topics
- Phytochemistry
- Extracts