Cited by 588Meta-analysis and systematic reviewsPsychology Research and BibliometricsStatistical Methods and Bayesian Inference

Correcting for Bias in Psychology: A Comparison of Meta-Analytic Methods

Open access
Carter, Evan C., Schönbrodt, Felix D., Gervais, Will M., Hilgard, Joseph·2019·Advances in Methods and Practices in Psychological Science

Published version of a preprint

Abstract

Publication bias and questionable research practices in primary research can lead to badly overestimated effects in meta-analysis. Methodologists have proposed a variety of statistical approaches to correct for such overestimation. However, it is not clear which methods work best for data typically seen in psychology. Here, we present a comprehensive simulation study in which we examined how some of the most promising meta-analytic methods perform on data that might realistically be produced by research in psychology. We simulated several levels of questionable research practices, publication bias, and heterogeneity, and used study sample sizes empirically derived from the literature. Our results clearly indicated that no single meta-analytic method consistently outperformed all the others. Therefore, we recommend that meta-analysts in psychology focus on sensitivity analyses—that is, report on a variety of methods, consider the conditions under which these methods fail (as indicated by simulation studies such as ours), and then report how conclusions might change depending on which conditions are most plausible. Moreover, given the dependence of meta-analytic methods on untestable assumptions, we strongly recommend that researchers in psychology continue their efforts to improve the primary literature and conduct large-scale, preregistered replications. We provide detailed results and simulation code at https://osf.io/rf3ys and interactive figures at http://www.shinyapps.org/apps/metaExplorer/ .

Integrity signals

Registry

Crossref · 10.1177/2515245919847196

Cites

77 of 77 references verified0 citation errors · 0 not checkable

Statistics

Recomputing p-values needs the paper’s full text.

Claims the paper makes

We couldn’t retrieve this paper’s full text, so its claims and self-check weren’t run. The registry and citation checks still apply.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • The absence of publication bias is highly unlikely unless a set of preregistered reports is examined.

    “The absence of publication bias is highly unlikely unless a set of preregistered reports is examined (Carter et al. 2019).”

    From The gamification of learning: a meta-analysis (2020) · cites this paper as Carter et al. 2019

  • The trim-and-fill method tends to adjust too little when bias is strong, and Fail-Safe N neither tests nor adjusts for the presence of bias.

    “The trim-and-fill method (Duval & Tweedie, 2000) tends to adjust too little when bias is strong (Carter et al., 2019), and Fail-Safe N (Rosenthal, 1979) neither tests nor adjusts for the presence of bias.”
  • Bias adjustments can improve performance relative to a meta-analysis that does not adjust for publication bias.

    “Bias adjustments can improve performance relative to a meta-analysis that does not adjust for publication bias (Carter et al., 2019; McShane et al., 2016; Moreno et al., 2009; Simonsohn et al., 2014).”
  • Negative and statistically significant SPET estimates have been observed in both simulation studies and meta-analysis.

    “negative and statistically significant SPET estimates have been observed in both simulation studies (Carter et al., 2019) and meta-analysis (Hilgard et al., 2019).”
  • The p-curve estimate of average study power may be an underestimate because p-hacking adds large p values to the p-curve, canceling out the right skew and biasing the estimate downward.

    “On the one hand, this may be an underestimate of average study power—in simulation studies, p-hacking adds large p values to the p-curve, canceling out the right skew and biasing the p-curve estimate downward (Carter et al., 2019).”

Cites

Flagged first, then by citations

Cited by

Sorted by citations

Authors
Carter, E. C., Schönbrodt, F. D., Gervais, W. M., & Hilgard, J.
Affiliations
  • DEVCOM Army Research Laboratory (US)
  • Ludwig-Maximilians-Universität München (DE)
  • University of Kentucky (US)
  • Illinois State University (US)
Corresponding author
Carter, Evan C.
Venue
Advances in Methods and Practices in Psychological Science
Published
2019-06
Volume
2
Issue
2
Pages
115-144
ISSN
  • 2515-2459 (ISSN-L)
  • 2515-2467
OpenAlex
W2781957993
Cited by
588
References
76
Citations by year
2026: 43 · 2025: 60 · 2024: 69 · 2023: 86 · 2022: 83 · 2021: 87 · 2020: 79 · 2019: 54 · 2018: 23 · 2017: 3 · 2016: 1
FWCI
50.69
Type
journal article
Language
en
Open access
Locations
  • Advances in Methods and Practices in Psychological Science (publishedVersion)
  • PsyArXiv (OSF Preprints) (submittedVersion, cc-by)
  • PsyArXiv (OSF Preprints) (submittedVersion)
Topics
  • Social Sciences › Decision Sciences › Statistics, Probability and Uncertainty › Meta-analysis and systematic reviews
  • Social Sciences › Psychology › Experimental and Cognitive Psychology › Psychology Research and Bibliometrics
  • Physical Sciences › Mathematics › Statistics and Probability › Statistical Methods and Bayesian Inference
Keywords
publication bias, questionable research practices, effect size overestimation, meta-analytic methods, sensitivity analysis, heterogeneity, simulation study, pre-registered replication

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review