Suppose that my meta-analysis has several studies performed by the same author. There could be heterogeneity at the author level (because perhaps the author is not very attentive, or uses different tools, etc). How can I account for possible author effects in my meta analysis? Specifically, how might I adjust my computed effect size under fixed and random effects models?

  • 1
    $\begingroup$ You are probably looking for multi-level meta-analysis with a random effect for study and for author. $\endgroup$
    – mdewey
    Commented Apr 16, 2020 at 10:46

2 Answers 2


I would recommend fitting a model with random effects for authors and studies within authors. This is the 'three-level' model described by Konstantopoulos (2011). An example illustrating such an analysis (using R and the metafor package) is provided here:

Konstantopoulos (2011) | The metafor Package

(just think 'district = author').


Konstantopoulos, S. (2011). Fixed effects and variance components estimation in three-level meta-analysis. Research Synthesis Methods, 2(1), 61–76.


A couple of options including meta-regression that we present in this paper: Can authorship bias be detected in meta-analysis? by Abou-Setta AM et al. (2019).

One thing you will need to note is that the 'author' may not always be the first author, but maybe the last author or co-author as you'll want to create a unique ID for each other in each publication prior to running your analysis.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.