Publication Bias:A call for improved
meta-analytic practice in the
organizational sciences
George C. Banks, Sven Kepes and Michael A. McDaniel
Virginia Commonwealth University, 301 West Main Street, PO Box 844000, Richmond, VA 23284, USA.
banksgc@vcu.edu
Previous research has introduced the threat of publication bias to meta-analytic reviews in
management and industrial/organizational (I/O) psychology research. However, a compre-
hensive review of top-tier journals demonstrates that more than two thirds of meta-analytic
studies in management and I/O psychology ignore the issue. Of the studies that do empiri-
cally evaluate publication bias, almost all use methods that are based on problematic
assumptions (e.g., the Failsafe N and subgroup comparisons by source of data).The current
paper reviews the issue of publication bias and introduces to management and I/O psychol-
ogy new methodological techniques to assess this bias.To illustrate the methods, multiple
publication bias methods are demonstrated in a meta-analytic review of conditional reason-
ing tests for aggression.We offer specific recommendations that address both design and
analysis issues to mitigate the existence and influence of publication bias.
1. Introduction
P
ublication bias exists to the extent that available
research results are unrepresentative of all research
results.The typical consequence of publication bias is an
overestimation of effect sizes (Dickersin, 2005; McDan-
iel, Rothstein, & Whetzel, 2006). It has been proposed
that publication bias is one of the greatest threats to the
validity of meta-analytic reviews (Banks & McDaniel,
2011; Rothstein, Sutton, & Borenstein, 2005b), and that
such reviews are one of our most important tools for
advancing science and evidence-based management
(Briner & Rousseau, 2011). Therefore, confidence in the
validity and robustness of our meta-analytic results is
contingent upon the extent to which publication bias
influences our research.
Journal articles, books, and tutorials provide guidance
on how to address the issue of publication bias in meta-
analytic reviews (e.g., Berlin & Ghersi, 2005; Borenstein,
Hedges, Higgins, & Rothstein, 2009; Hunter & Schmidt,
2004; McDaniel et al., 2006; Rothstein, Sutton, & Boren-
stein, 2005a; Song et al., 2010). Despite this, the majority
of meta-analytic reviews published in the top-tier man-
agement and industrial/organizational (I/O) psychology
journals continue to ignore this type of bias. In this
respect, our literature lags behind related psychological,
educational, and medical research.
To demonstrate the degree to which our literature is
methodologically behind other fields of research, we
examined meta-analytic reviews published in top man-
agement and I/O psychology journals. This review com-
plements previous reviews of publication bias (e.g.,
Aguinis, Dalton, Bosco, Pierce, & Dalton, 2011a; Gey-
skens, Krishnan, Steenkamp, & Cunha, 2009) by providing
data from other fields of research as a comparison
point.Additionally, our review complements and extends
the work that was conducted by Ferguson and Brannick
(2012), who obtained similar results in regard to general
psychology. However, our review takes an additional step
by illustrating that management and I/O psychology lags
behind several fields in the evaluation and assessment of
publication bias.
We reviewed the literature from 2005 to 2010 (we
choose 2005 as our starting point as it coincides with
the release of a key publication, Publication bias in meta-
analysis : Prevention, assessment, and adjustments by Roth-
stein, Sutton, & Borenstein [2005a]). This period was
thought to provide a representative sample of meta-
International Journal of Selection and Assessment Volume 20 Number 2 June 2012
© 2012 Blackwell Publishing Ltd.,
9600 Garsington Road, Oxford, OX4 2DQ, UK and 350 Main St., Malden, MA, 02148, USA