5

Meta-Analytic Research in
International Business and
International Management
Peter J. Buckley a, Timothy M. Devinney a, and Ryan W. Tang b
a
Centre for International Business, University of Leeds; and b University of Technology, Sydney

Over the past decade, international business and international management

researchers have utilized meta-analytic approaches to synthesizing findings in the
extant literature. This chapter reviews the studies published in the top five inter-
national business and management journals from 2004 to 2012. The review inves-
tigates major problems in the published meta-analyses by evaluating their overall
analyses as well as the approaches utilized. The findings of this review reveal
differences among the journals and improvements in the approaches applied in
recent years. The chapter ends by discussing why and how international business
and management researchers need to focus more on methodological fundamentals
in their applications of meta-analysis.

Introduction

Meta-analysis is a statistical analysis of extant studies on a research topic

(Glass, 1976). It refers to a technique of reviewing a literature by quasi-
standardized procedures for generating scientific conclusions (Hunter &
Schmidt, 2004). With a strong tradition of meta-analytic studies in education
(e.g., Glass, 1976) and psychology (e.g., Schmidt & Hunter, 1977), research-
ers in other disciplines have been critical to the increasing number of meta-
analyses being published. According to the ISI Web of Science (retrieved
5 October 2012), the number of meta-analytic articles published in manage-
ment rose from 28 in the 1980s to 220 in the 1990s. After 2000, ISI Web
of Science recorded an astounding 744 meta-analytic articles published in
management journals. However, as will be discussed in this chapter, meta-
analytic methodological applications have not been well developed in inter-
national business (IB) and international management (IM), both in terms of
their quantity or quality.

Reprinted with permission from Emerald Group Publishing Limited, All rights
reserved. Philosophy of Science and Meta-Knowledge in International Business and Manage-
ment Advances in International Management (2013) 26, 263–297.

100

P. J. Buckley, The Multinational Enterprise and the Emergence of the Global Factory
© Palgrave Macmillan, a division of Macmillan Publishers Limited 2014
 Meta-Analytic Research in International Business and International Management 101

Inherently, IB/IM research is multidisciplinary in scope and interdis-

ciplinary in terms of its content and method (Thomas, Cuervo-Cazurra, &
Brannen, 2011). The complex nature of this research requires IB/IM research-
ers to apply rigorous and systematic ways to understand clearly where we are,
what we know, in what way we can improve on what we know and what we
do, and in which direction we should go as a social science discipline. These
‘w’s call for an optimal way to synthesize the past achievements in the IB/
IM research domain.
In the more than half a century since IB/IM arose as a distinctive disci-
pline, the volume, complexity and orientation of research in the domain
have expanded dramatically. Across these studies we see variation in terms
of the samples used, methods applied, theoretical models and constructs
tested. This varied literature consistently creates a demand for reviews that
integrate, synthesize and identify the central issues in the extant literature
as it evolves.
Literature reviews have been widely applied by IB/IM researchers as
exemplified by qualitative reviews (e.g., Brouthers & Hennart, 2007) as
well as quantitative syntheses (i.e., meta-analyses such as Stahl, Maznevski,
Voigt, & Jonsen, 2010), with the narrative and objective review being more
common. For example, the literature on the entry mode choice of multi-
national companies (MNCs) has been reviewed by more qualitative studies
(e.g., Brouthers & Hennart, 2007; Canabal & White III, 2008; Malhotra,
Agarwal & Ulgado, 2003; Slangen & Hennart, 2007) than meta-analytic ones
(i.e., Morschett, Schramm-Klein & Swoboda, 2010; Zhao, Luo & Suh, 2004).
Although narrative syntheses provide valuable insights into IB/IM theories,
the meta-analytic reviews go one step further by statistically integrating
knowledge and methodically examining the findings of primary studies.
Furthermore, unlike narrative reviews, meta-analyses are capable of test-
ing generalizations by investigating moderating effects across studies. This
increases the reliability of the reviewed literature and provides guidance
on the future direction of IB/IM research more clearly (Cooper, Hedges, &
Valentine, 2009).
Furthermore, meta-analysis is a crucial approach to resolving the problem
of insufficient power (i.e., sample size) in many primary empirical studies
(Lipsey & Wilson, 2001). It combines a set of independent primary stud-
ies that bear on the same research topic into a simple pattern of results by
correcting distorted primary findings arising from artefacts that may lead
to conflicting conclusions (Hunter & Schmidt, 2004). This is particularly
important in the case of IB/IM research, where samples are rarely purpose-
ful and based on the availability of data from different countries at different
points in time. Thus, meta-analysis can benefit IB/IM researchers by specify-
ing the moderator variables, such as which countries were examined, and
future topics or sample domains, such as which types of countries need to
be examined in the future. In addition, meta-analytic reviews complement
102 Peter J. Buckley et al.

the narrative literature reviews and the primary studies by integrating extant
research with a particular focus or goal. Meta-analytic techniques can also be
powerful tools to improve IB/IM research by highlighting where construct
and model validity is weak (Farley & Lehmann, 2001). Hence, IB/IM is not
just an area where meta-analytic approaches can add great value, but meta-
analytic approaches are arguably critical to filling in the contextual modera-
tors that are the hallmark of IB/IM scholarship.
This chapter aims to (1) review meta-analyses in IB/IM discipline, (2) out-
line major issues for evaluating meta-analyses and (3) examine methodologi-
cal issues in the extant IB/IM meta-analyses. To achieve these goals, the next
section takes an overview of meta-analytic studies in the IB/IM discipline.

Meta-analysis in IB/IM Research

Meta-analysis arose as a reliable means of summarizing and integrating social

science studies in the 1970s. Since that time, meta-analysis has developed
as a tool, helping researchers in many disciplines (e.g., clinical research),
and served as an important complement to other approaches to reviewing a
literature review (Bohlin, 2012). The importance of meta-analysis in specific
disciplines is seen in Bausell, Li, Gau, and Soeken (1995), who examined
meta-analyses published between 1980 and 1993 in two academic domains
(i.e., social science and health science). Over this 13-year period, the number
of meta-analyses grew at a 14.7% annual rate, with 40 studies published in
1980 and 251 in 1992. Six years after that study, Lee, Bausell, and Berman
(2001) showed that this rate of growth had increased to 17% per year
between 1993 and 2000. Inspired by these findings, Kirca and Yaprak (2010)
searched the meta-analytic literature published between 1980 and 2009 in
marketing, management and IB disciplines in the top 25 journals of the
three disciplines. The authors found 104 meta-analyses in marketing and
414 in management, but only 24 in IB/IM. At the time they argued that IB/
IM researchers seemed not to favour meta-analytic methodologies.
Following similar principles of the previous studies, this chapter searched
the extant meta-analytic papers published in top IB/IM journals from 2000
through October 2012. Specifically, this study defines IB/IM journals in
accordance with the Journal Quality List by subject areas (Harzing, 2012)
and ranks the candidate journals by the total cites and the impact factor
from ISI Journal Citation Reports (ISI, 2012). Based on these criteria, the
top five IB/IM journals are Journal of International Business Studies ( JIBS),
Journal of World Business ( JWB), Journal of International Management ( JIM),
International Business Review (IBR) and Management International Review
(MIR). In the databases that contain these journals (i.e., Science Direct,
ProQuest), this study conducted a computerized search by retrieving titles
and abstracts containing the keyword ‘meta’. Among the initial results, this
chapter retains meta-analysis (e.g., meta-analytic review and meta-analytic
 Meta-Analytic Research in International Business and International Management 103

test), but excludes articles regarding meta-factor, meta-construct and articles

referring to but not conducting meta-analysis. In particular, we exclude
studies using the title ‘meta-analytic review’ that failed to meet the strict
definition of statistical meta-analysis (i.e., Takahashi, Ishikawa, & Kanai,
2012). This process resulted in a list of 15 meta-analyses (see the appendix).
The list might not exhaustively include all IB/IM meta-analyses, but it can
be viewed as representative of the current meta-analytic studies in leading
IB/IM journals to a great extent. These journal papers build a preliminary
basis for evaluating the current status of meta-analytic techniques and
reviewing common practices in IB/IM discipline as well. Figure 5.1 depicts
the amount and proportion of meta-analyses published in each journal. It
shows JIBS published the most meta-analysis studies in the past decade.
Figure 5.2 illustrates the number of published meta-analyses in every year
since 20041 and the proportion of meta-analyses to all articles in the five
journals in the corresponding years.
As illustrated by Figure 5.2, the findings from the top five IB/IM journals
suggest that the publication of meta-analytic articles has increased dramati-
cally in three of last four years, despite the fact that over a longer period meta-
analytic methods are still rarely applied. Even in the year when the largest
number of meta-analysis papers appear (e.g., 2012), they amounted to only
1.26% of the total number of studies published in the top five IB/IM journals.
Interestingly, Figure 5.2 also shows that the proportion of meta-analytic arti-
cles in IB/IM journals maintained a pattern of constantly low quantity and
proportion in most years, climbing above 1% only in 2009, 2010 and 2012.

JIBS JWB JIM IBR MIR

MIR, 3

JIBS, 6

IBR, 2

JIM, 2
JWB, 2

Figure 5.1 Meta-analyses published by IB/IM Journals

1
No meta-analyses were published between 2000 and 2003 in the top five journals.
Before 2000, there was only one published meta-analysis: Peterson and Jolibert (1995).
104 Peter J. Buckley et al.

Number of studies Percentage of total

4 1.40%

1.20%

3
1.00%

0.80%
2
0.60%

0.40%
1

0.20%

0 0.00%
2004 2005 2006 2007 2008 2009 2010 2011 2012

Figure 5.2 IB/IM meta-analyses published by year

However, the retrieval results also disclose problems in IB/IM meta-analyses.

For example, IB/IM meta-analysts tend to select similar research topics.
Specifically, four meta-analyses investigated spillover, corruption, leader-
ship, and financial analyst’s predictions, respectively, while eleven others are
related to three topics only: culture, firm performance and entry strategy.
Culture attracted six meta-analyses, as did firm performance. Meta-analyses
concerning the MNC’s market entry decision are relatively less than those
about the preceding two topics but were still significant in accounting for
four of the studies. There are two meta-analytic studies that each accounted
for the three topics simultaneously (i.e., Magnusson, Baack, Zdravkovic,
Staub, & Amine, 2008; Tihanyi, Griffith, & Russell, 2005). The sections that
follow outline the findings of the meta-analytic studies to date in IB/IM.

Culture
The effect of cultural factors is not only a prevalent IB/IM research topic but
also a crucial meta-construct in IB/IM research per se, reflecting as it does
the international environment in which MNCs reside.
Tihanyi et al. (2005) and Magnusson et al. (2008) examined how culture
influenced the MNC’s entry mode choice and performance. These two
studies found similar results – the effects of cultural factors on entry mode
and performance are not statistically significant, but MNC’s home country,
industry sector, time and operationalization of cultural difference/distance
significantly moderated the effective relationship. For example, the patterns
of US-based and Europe-based MNCs’ entry strategies showed a substantial
 Meta-Analytic Research in International Business and International Management 105

relationship with cultural factors, while MNCs from other countries or

regions did not.
At a more micro level, Fischer and Mansell (2009) and Stahl et al. (2010)
applied meta-analytic methods to synthesize the research on cross-culture
work teams by integrating the findings about cultural effects in the MNC’s
working environment. The former found cultural factors impacted differ-
ent types of commitment and the latter suggested that a variety of cultural
effects across contexts and research designs.
In contrast to the preceding four meta-analyses, Steel and Taras (2010)
considered the culture as the consequence (or dependent variable). They
found both micro features (i.e., age, gender, education and socio-economic
status) and macro characteristics (i.e., wealth and freedom) might deter-
mine an individual’s cultural values. Analytically, Steel and Taras (2010)
meta-analysed the mean differences of cultural scores (i.e., survey results on
cultural values) and the moderating effects of variables such as age, gender,
wealth and freedom. The authors examined hypotheses by regressions with
the mean differences as the dependent variables and the moderators as inde-
pendent variables. However, the results demonstrate only how the factors
(i.e., age, gender, etc.) relate to deviations in the scores of previous surveys
on cultural values. Regarding cultural values, Taras, Steel, and Kirkman
(2012) did not investigate the determinants of cultural values, but applied
meta-analytic techniques to review the pre-dominant cultural framework
(i.e., Hofstede’s cultural value dimensions) in the IB/IM discipline. That
meta-analytic study validated Hofstede’s cultural values and found a trend
implying that Hofstede’s scores have been declining in relevance, due, they
argue, to cultural change, and transmission gradually made current cultural
values different to those calibrated in past decades.

Multinational Performance
MNC performance has been investigated in relationship to other variables.
Besides Tihanyi et al. (2005) and Magnusson et al. (2008), who examined
the relationship between performance and cultural values, Bausch and Krist
(2007) and Yang and Driffield (2012) integrated findings about the relation-
ship between an MNC’s globalization and its performance via testing various
moderator effects. Among the moderators, MNC’s country of origin is seen as
a significant factor influencing the globalization-performance relationship.
Likewise, van Essen, Heugens, Otten, and van Oosterhout (2012) examined
the relationship between MNC performance and executive compensation.
These authors found a moderately positive but considerately significant
relationship between MNC performance and compensation, although this
relationship varied dramatically across countries, arguably due to the dif-
ferent institutional structures operating in those countries. Reus and Rottig
(2009) analysed factors that may determine an MNC’s performance in
different host countries and found unique features of MNCs in China, with
106 Peter J. Buckley et al.

a significant negative effect of hierarchical control on partner conflict seen

in the China samples when compared to the non-China samples.

Entry Mode
Tihanyi et al. (2005) and Magnusson et al. (2008) analysed how entry
mode choice was impacted by cultural factors, while Zhao et al. (2004) and
Morschett et al. (2010) studied the determination of entry strategies. Zhao
et al. (2004) conducted a meta-analysis based on transaction cost theory
that helps the study construct six dimensions of determinants (e.g., cultural
distance, country risk, international experience and advertising intensity).
Morschett et al. (2010) was not constrained by a single theoretical paradigm
and examined 13 external determinants of entry mode choice (e.g., cultural
distance, country risk, market growth and volatility of demand). However,
they did not include internal factors (e.g., international experience and
advertising intensity) and thus any generalization of the results is limited.
Obviously, a strategic decision related to international entry is impacted
by its experience (e.g., Delios & Beamish, 1999) and constrained by its
marketing ability (e.g., Dikova & van Witteloostuijn, 2007).

Remaining Studies
Except for the preceding topics, the three remaining meta-analytic stud-
ies selected different research focuses. García-Meca and Sánchez-Ballesta
(2006) found significant moderator effects of country, measurement, and
time by synthesizing studies on factors that influence the accuracy of
financial analysts’ forecasts. Meyer and Sinani (2009) statistically reviewed
the literature regarding positive spillovers of foreign direct investment.
The authors revealed a curvilinear relationship between spillovers and the
development levels of host countries. Judge, McNatt, and Xu (2011) inte-
grated previous research on causes as well as effects of national corruption.
The results suggest the political/legal effects had the strongest relationship
with corruption.
The 15 meta-analytic studies published in the top 5 IB/IM journals in the
21st century reveal a limited overview and synthesis of the domain to date.
Nevertheless, these studies are a significant improvement in the application
of meta-analytic approaches compared to earlier work. For instance, the first
published meta-analysis in our review (i.e., Zhao et al., 2004) aggregated
data from only 38 primary studies, whereas Steel and Taras (2010) inte-
grated data from 508 primary studies and Taras et al. (2012) summarized
findings from 451 primary studies. Furthermore, in more recent years, IB/
IM meta-analysts have applied more deliberative and sophisticated methods
to model the data compiled from more sizeable literatures. For example,
Fischer and Mansell (2009) and van Essen et al. (2012) applied multivariate
modelling techniques to synthesize the findings, while the studies published
 Meta-Analytic Research in International Business and International Management 107

in the early years of the last decade (e.g., García-Meca & Sanchez-Ballesta,
2006) employed only very basic statistical approaches (e.g., distribution and
relationship) for disclosing patterns of integrated data. The preceding exam-
ples hopefully underscore the fact that meta-analytic methods are gradually
being applied to answer more complex questions and explain more deep-
seated phenomena requiring more sophisticated and nuanced approaches.
Yet, one must acknowledge that larger sample sizes and/or more com-
plicated modelling do not pari passu lead to a better meta-analytic study.
In particular, as mentioned previously and discussed subsequently, meta-
analyses in IB/IM still have considerable space in which to make progress.
Furthermore, IB/IM researchers need to harken back to the criteria for
conducting a valid meta-analysis if these studies are going to be useful in
truly synthesizing and integrating what we do and do not know. The next
section provides nonstatistical guidelines for assessing meta-analytic studies
in efforts to outline the evaluation of meta-analytic studies. These guidelines
are synthesized from major methodological literature with considerable
references to widely applied and commonly accepted practices published in
major journals. Since the statistical details regarding the calibrations (e.g.,
formula, calculation, critical value, etc.) are not in the scope of this chapter,
IB/IM researchers are referred to meta-analytic literature – such as Cooper et al.
(2009), Hunter and Schmidt (2004), Lipsey and Wilson (2001), Rosenthal
(1991) and Hedges and Olkin (1985) – for more detailed general discussion.

Review of Meta-analytic Studies on International Business

Users of meta-analysis are confronted with a number of process issues when

conducting a meta-analytic study. Procedures of different meta-analyses
may vary, but a meta-analytic study normally has five sections: research
specification, literature retrieval, dataset preparation, data analysis and
results report (Devinney & Tang, this volume). Each section has its major
queries and essential options, which are summarized in Table 5.1.
Using this structure we review the 15 IB/IM meta-analyses published since
2000 by evaluating their alignment with the major queries in the five sec-
tions. This evaluation focuses on published information only. The absence
of essential items may be due to either the meta-analysts not considering the
items or the published study not including them despite the fact that they
were conducted. Also, this chapter does not purposely rank these studies
in terms of their quality. It instead attempts to examine the status of meta-
analysis in IB/IM using these studies as illustrative examples. In addition,
several of the articles report more than one meta-analysis. For instance,
Fischer and Mansell (2009) meta-analysed more than one relationships by
different metrics of effect size. In that case, all the descriptive information
pertains to the article rather than to the individual analyses.
108

Table 5.1 Major issues in meta-analytic studies

Section Major Query Essential Options and Items

Research Definition of Stated research topic; well-defined

specification research questions variables; applicable to
meta-analysis; theoretical and
practical context
Literature Data sources Single/multiple database; specific
retrieval journal(s); reference section of other
article; authors; a combination of
the sources
Key terms and Key terms representing research
combinations topic; combinations of key terms
Other retrieval E-mail listservers; directly contacting
strategies authors; browse of potential
literature (e.g., reviews and eligible
studies); non-English publications;
Dataset Effect size metric Correlation; mean; (log) odds ratio;
preparation multiple
Inclusion criteria Research topic; empirical studies;
independence; statistics for effect
size; other specifications (e.g., time
period)
Coding procedure Coding protocol (operational
measures to effect size, study
descriptors); reliability (single/
multiple coders)
Correction Transformation; alleviation;
unreliability; validity;
dichotomization;- range
Outliers Identifier (means of graphs,
residuals, homogeneity statistics,
sample-adjusted meta-analytic
deviancy); handle (trim, recode)
Data Statistical Mean and standard deviation of effect
Analysis information size; confidence intervals; others
Homogeneity analysis Q-test; 75%-rule or similar
Moderator analysis Fixed-, random- and mixed-effects
models
Missing data Contacting original authors; likewise
deletion; pairwise analysis; regression
Publication bias Identifier (funnel plot, statistical test);
evaluation (fail-safe N, trim and fill)
Additional data Definition and proxy of external
variable; external sources
Report Results Statements about findings, generality,
and limitations; illustrating by
figures, graphs, and tables
Included studies Reference list of included studies

Compiled according to Devinney and Tang (this volume).

 Meta-Analytic Research in International Business and International Management 109

Research Specification
Similar to any empirical study, meta-analysis requires clearly defined vari-
ables that can represent the questions that the study will answer, the phe-
nomena the study will explain or/and the research topic that the study
will address. The variables should be able to reflect the theoretical and
conceptual frameworks that underlie the arguments of the meta-analytic
study. Also, the research topic should be stated explicitly to validate the
variables defined in the study. All of the 15 IB/IM meta-analyses addressed
the 4 aspects: explicitly stated research topic, the explanation of concept in
the theoretical and/or practical context, well-defined variable, and the appli-
cability of the research question to meta-analysis. For example, Taras et al.
(2012) meta-analysed national cultural indices within Hofstede’s framework
(e.g., Hofstede, 1980; Hofstede, Hofstede, & Minkov, 2010) by focusing on
the four culture dimensions (i.e., power distance, individualism, masculinity
and uncertainty avoidance) that had been investigated for three decades and
where the work appeared in around 100 journals.
However, 2 of the 15 meta-analyses did not set up the theoretical and
practical contexts (i.e. García-Meca & Sánchez-Ballesta, 2006; Yang &
Driffield, 2012). The theoretical and practical context that serves as the
basis of the meta-analysis has to be verified in order to establish meaning-
ful research questions and build connections between concepts (i.e., vari-
ables) and concrete events (i.e., the proxies or operationalizations of the
variables). Nevertheless, García-Meca and Sánchez-Ballesta (2006) failed to
embed their meta-analysis into a context that demonstrates the rationale
for searching effect sizes (i.e., six categories of correlation coefficients)
and moderators (e.g., measurements of variables), although they briefly
mentioned the importance of their research and reviewed extant literature
related to variables of interest. Yang and Driffield (2012) applied an attrac-
tive effect size metric (i.e., estimate of coefficient in a linear model) that
may disclose causality. However, these authors did not verify the logic for
choosing account-and market-based firm performance and controlling firm
characteristics and business cycle effects. They also fail to mention how
they define the degree of multinationality, which is a notable exclusion
given that their research was aimed at examining the multinationality–
performance relationship.

Literature Retrieval
The first concern regarding literature retrieval is the sources of primary
studies. As the availability of studies varies across databases, a well-
conducted meta-analysis often searches potential studies in multiple data-
bases for broader and more complementary literature. Other important
sources include journals that concentrate on a certain area, references cited
by articles reviewing previous literature, and researchers who are expert in
a related research field. Multiple-source searching is always prior to single
110 Peter J. Buckley et al.

one, because any one or two sources may not provide comprehensive and
exhaustive primary studies. An unrepresentative literature pool resulting
from limited sources may bias the conclusions of a meta-analysis (Devinney &
Tang, this volume).
The studies examined were quite limited in terms of their breadth and
depth of literature examined. Only three articles (i.e., Stahl et al., 2010;
Steel & Taras, 2010; van Essen et al., 2012) took into account multiple
databases, major journals and published reviews simultaneously. Of the 12
remaining meta-analyses, 5 took advantage of two data sources (i.e., mul-
tiple databases and major journals, or both journals and published reviews
in a research area), while four drew from only one of the three major data
sources (i.e., multiple databases, specific journals or published reviews).
Three meta-analyses searched potential extant literature only in one single
individual database but not multiple databases (i.e., Judge et al., 2011; Reus &
Rottig, 2009; Yang & Driffield, 2012).
Similar to data sources, key terms and their combinations (e.g., AND,
OR) are important for literature retrieval. The implication of this impor-
tance is threefold. First, a good meta-analysis reports what key terms are
utilized. Second, the terms precisely represent the research topic of inter-
est. Third, the terms and their combinations are related to the inclusion
of primary studies. In IB/IM meta-analyses, both pieces of information are
reported simultaneously in only one study (i.e., Stahl et al., 2010). Key
terms but not combinations are provided in four other meta-analyses,
while the remaining 10 IB/IM meta-analyses did not mention what key
terms were searched in the data sources. Authors of the 10 articles might
have employed some keywords in literature searching, but the fact that
they did not disclose the key terms used makes it difficult for readers to
know how the eligible primary studies were retrieved. In particular, other
researchers who want to examine those meta-analytic studies will not be
able to replicate them.
In addition, other strategies for searching previous studies are employed
by many authors of meta-analyses because of the complementarity of com-
mon searching approaches. For example, e-mail listservers and contacting
authors are ways for collecting unpublished studies from potential research-
ers. Browsing candidate literature selected in the previous steps may find
missing studies in eligible bibliography. In addition, although the majority
of academic research is published or written in English, non-English studies
may also investigate research topics of meta-analyst’s interest.
These strategies were applied in meta-analytic studies published by one
journal more frequently than the other. Specifically, in the JIBS-published
meta-analyses all were comprehensive in concentrating on the authors of
publication. On the other hand, they never mentioned e-mail listservers,
and all the top five IB/IM journals published meta-analytic articles that
 Meta-Analytic Research in International Business and International Management 111

browsed potentially selected literature. Furthermore, no extant IB/IM meta-

analysis collected data from non-English studies. This leaves the IB/IM meta-
analyses a potential bias in primary data collection, particularly in the case
of examining unpublished but potentially relevant doctoral theses.

Dataset Preparation
Meta-analysis requires a common effect size (i.e., effect size metric, effect size
estimation) to represent the quantitative findings that can be standardized
across studies for meaningful and numerical analyses. Three types of effect
size metrics are commonly used: standardized mean difference (i.e., Hedges’
g, Cohen’s d, and Glass’ gGlass), correlation (e.g., Pearson correlation r, partial
correlation rxy,z) and odds ratio (i.e., i or o) (Devinney & Tang, this volume).
Judgement on the selection of effect size metric is linked to the research
questions specified initially. In other words, a good meta-analysis selects the
effect size metric that can demonstrate the essential concept of research topic
(i.e., the phenomenon, association or causality). The 15 IB/IM meta-analyses
applied typical effect size metrics (i.e., mean, correlation and odds ratio) as well
as some uncommon ones (e.g., t-statistic in Meyer & Sinani, 2009). These effect
size metrics were generally well-justified for answering the research questions.
However, inclusion criteria were not explicitly disclosed in all IB/IM
meta-analyses. Four studies did not explain what criteria they used to screen
potential literature. Inclusion criteria are important because they define
the eligible studies for meta-analysis and explain what the meta-analysis is
about. The criteria are also important for other researchers who may want to
replicate the meta-analysis and understand the analysing scope.
Coding procedure was more frequently explained by IB/IM meta-analysts
than inclusion criteria, because the procedure collects data from eligible
studies according to coding protocol that illustrates the definition and
scope of effect size and moderator. This coding information helps users of
the meta-analysis understand what data are included as well as excluded
(Devinney & Tang, this volume). An entire protocol may be too long to
be enclosed in a journal article, but a brief introduction on how the meta-
analyst codes the variables (i.e., what measurements or operationalizations
represent a variable) is helpful for understanding the results, as well as
assessing quality, of the meta-analysis. In addition, multiple-coder cod-
ing is an important feature of a good meta-analysis because it reduces the
bias associated with a single coder. The reliability (e.g., agreement rate and
Cohen’s ϕ) of coders for the same literature pool is a crucial value to evaluate
a meta-analytic study.
In IB/IM meta-analyses, authors of 12 of the meta-analyses gave coding
information and half of these articles employed multiple coders to collect
data from primary studies. In particular, the method of multiple coding was
applied more often in studies published after 2009. Specifically, six of the
112 Peter J. Buckley et al.

seven multi-coded meta-analyses were published after 2009, but only one
before 2009. It seems to suggest that the reliability of data collection has
gradually improved in recent IB/IM meta-analyses.
Furthermore, correction of artefacts aims to adjust the imperfect raw
data collected in primary studies for desirable statistic power (e.g., from
correlation r to Fisher’s z) and accurate estimation (e.g., correcting artificial
dichotomization by the numbers and proportions). Only five meta-analyses
published after 2009 corrected for effect sizes, while none did so before. In
particular, the meta-analyses published in JIBS are more likely to correct for
effect sizes (i.e., three out of six did this).
In addition, the published meta-analytic studies in the top five IB/IM jour-
nals indicate that most IB/IM meta-analysts were not aware of the outliers
in their datasets. Those values potentially result in a skewed distribution of
effect size and potentially bias meta-analytic findings. However, only one
IB/IM meta-analysis discussed outliers. There might be three reasons for the
majority of authors not doing so. First, their meta-analytic datasets did not
have any outliers. Nevertheless, a perfect raw-dataset hardly exists. Despite
having a perfect dataset it is still practical to let readers know there are no
outliers. Second, the outliers had been trimmed or recoded, but not men-
tioned in the articles. The adjustment of original datasets by trimming or
recoding is important information for other researchers to judge and evalu-
ate the data collection process, and therefore should be disclosed. Lastly,
some IB/IM meta-analysts might think it was not necessary to take outliers
into account. In fact, the distribution of an original dataset is always skewed
by extreme values, but the typical purpose of meta-analysis is to summarize
the overall findings in a research domain (Lipsey & Wilson, 2001).

Data Analysis
The 15 IB/IM meta-analyses described effect sizes by the mean, standard devi-
ation and confidence intervals. These statistics can demonstrate the distribu-
tion of effect size, but only mean effect size was presented in every article.
Standard deviations were available in 10 meta-analytic studies. Confidence
intervals were only provided by four meta-analyses. In contrast to the
authors who did not provide these essential statistics, some meta-analysts
computed extra statistical data to facilitate their discussions. For example,
Fischer and Mansell (2009) and Steel and Taras (2010) provide the correlation
matrices of effect sizes and moderators to infer the relationships of interest.
An important item of meta-analytic data analysis is homogeneity analysis.
It provides evidence of the variability of studies by computing the prob-
ability that the variance of effect size is only due to sampling error. This is
crucial for meta-analysts to figure out whether significant moderator effects
exist in effect sizes (Devinney & Tang, this volume). However, only eight
IB/IM meta-analyses present this item. Most of them (i.e., five out of eight)
were published in JIBS, with only 1 IB/IM meta-analysis not presenting
 Meta-Analytic Research in International Business and International Management 113

homogeneity analysis (i.e., Tihanyi et al., 2005). In contrast, this crucial

analysis did not appear in any of the meta-analytic studies published in
JWB and JIM.
Although homogeneity analysis was absent in some of the 15 IB/IM
meta-analyses, moderator analyses were provided by all but 1 IB/IM meta-
analysis (i.e., Tihanyi et al., 2005). As a follow-up step to the homogeneity
analysis, moderator analysis gives insights into the discrepancy between
primary studies by finding whether features of the primary studies (e.g.,
year of publication, research design and sample range), contextual factors
(e.g., national economic status and institutional environment) or other
substantive differences account for variations in effect sizes. Thus, the six
meta-analyses that conducted moderator analysis without homogeneity
analysis may confuse readers on how the studies justified the need to exam-
ine moderating effects.
In addition, missing data are always inevitable in a meta-analytic dataset.
It is impossible that all the eligible primary studies provided perfect data
and these data were arranged in a form that the meta-analysts expected.
In 6 of the 15 IB/IM meta-analyses, information on missing data was not
reported. One reason for the absence of this key piece of information might
be that there was no missing data. However, similar to the information on
outliers, it is always practical to inform readers explicitly whether or not
missing data exist. If there were missing data, the way of handling them
is important for users of the meta-analysis to judge and assess the research
and its results.
Likewise, publication availability bias is another issue that was not widely
recognized. Only four articles among the 15 IB/IM meta-analyses identified
and handled the problem of publication bias. Although moderator analy-
sis can explain how publication features moderate effect sizes, the extent
to which the distribution of available studies differ from the population
distribution can only be disclosed by meta-analytic approaches regarding
publication bias, such as ‘fail safe N’ (Rosenthal, 1979). If there is a big dif-
ference between the distributions of available literature and the population,
the representativeness or generalizability of meta-analytic findings will be
weakened.
Lastly, additional data (i.e., external data that are not collected from meta-
analysis included literature) are gradually being incorporated by IB/IM meta-
analysts. Before 2009, no additional data were used in the published IB/IM
meta-analyses; however, 5 of the 10 meta-analyses published after 2009 took
advantage of additional data in their moderator analyses. The authors of
these meta-analyses employed additional data to investigate research ques-
tions that were not addressed by extant studies, for example, Taras et al.
(2012) and van Essen et al. (2012). With additional data, both meta-analytic
studies not only analysed previous findings but also expanded their analyses
to reach conclusions that a single primary study may be able to achieve.
114 Peter J. Buckley et al.

In Taras et al. (2012), additional data were collected from public databases
(e.g., the World Bank and the United Nations) and provided extra evidence
on how national cultural values evolved over the decades. Likewise, when
integrating the institution-based view into the meta-analysis of contracting-
theory-based literature, van Essen et al. (2012) collected external data on
institutional variables that were not available in the examined literature.
Their meta-analytic findings extended the previous understanding on
firm performance–executive compensation relationship from a theoretical
framework to another paradigm.

Analytic Report
Major results of meta-analytic studies are presented by stating major
findings, generality and limitations of the meta-analysis. If the statements
are facilitated by figures, graphs and tables, the meta-analytic report can be
more efficient and effective. Another critical item for meta-analyst’s report-
ing is the included studies. Users of meta-analysis need to know which
primary studies led to final conclusions. If inappropriate primary studies
are included, the meta-analysis based on them may not present to valid or
generalizable findings. In particular, the reference list of included studies is
evidence of the validity of the inclusion criteria.
In the top 5 IB/IM journals, all meta-analyses reported findings, generality of
conclusions, and included studies by illustrating them in tables, while only 8 of
15 applied figures and graphs. Among the eight studies, seven were published
after 2009. Furthermore, three among the 15 meta-analyses did not explicitly
report the methodological limitations of their meta-analytic procedures (i.e.,
Magnusson et al., 2008; van Essen et al., 2012; Yang & Driffield, 2012).
The preceding discussion is summarized in Table 5.2. It is not intended
to form an exclusive checklist for calibrating all facets of these IB/IM meta-
analyses, because this review evaluates the meta-analytic methodologies
used rather than the specifics of the IB/IM implications. In addition, both
the IB/IM discipline and meta-analytic methodologies are still developing.
Progress may update the criteria for judgement. Finally, the procedures
underlying meta-analyses vary. This chapter is not able to cover all the
details of all types of meta-analyses, but only capable of highlighting the
important and general issues.

Findings

Understanding the past achievements is crucial for IB/IM researchers to

develop our discipline (Buckley, 2002). Meta-analysis is one of the best ways
to summarize previous studies and search empirical generalizations. The
soundness of summary and generalization depends on the fitness to which
meta-analytic approaches are applied. Negligence of any critical issue may
result in crucial implications for the quality of meta-analysis. From this
 Meta-Analytic Research in International Business and International Management 115

perspective, this chapter comprehensively reviews meta-analytic approaches

as well as practices in the IB/IM discipline. The findings suggest that: (1)
meta-analytic methodology quality varied among IB/IM journals, as some
publications paid more attention on meta-analytic issues than the others;
and, (2) IB/IM researchers have been aware of more meta-analytic tech-
niques for accumulating and synthesizing previous literature to generate
new knowledge, since the quality of published meta-analyses has improved
after 2009, particularly regarding dataset preparation. However, some
methodological issues still need improvement.
In the 15 IB/IM meta-analyses, all studies published in JIM took into
account specific journals and published literature review when retriev-
ing potential literature. The two articles in JIM also provided standard
deviation (or variance) when describing effect sizes and presented the
meta-analytic results by figures and graphs. Likewise, more meta-analytic
studies published in JIBS contacted specific authors, investigated outliers
and conducted homogeneity analysis than those published in other IB/IM
journals. However, no meta-analysis in JIM disclosed major information
such as key terms, combination of key terms, correction of effect size, outli-
ers, confidence intervals, homogeneity analysis and publication bias, while
meta-analytic studies in JIBS seldom left out these.
Meta-analyses published in the three other IB/IM journals also show-
outstanding points, for example, multiple coders utilized by meta-analytic
studies in JWB, confidence intervals provided in MIR and missing data
uncovered in IBR. Yet, meta-analyses in JIBS and JIM are overall better than
the average, as they addressed 22 and 21 meta-analytic issues respectively
(Table 5.3).
Furthermore, this review of the 15 IB/IM meta-analyses suggests consid-
erable room for IB/IM meta-analysts to improve, although the period after
2009 witnessed big improvements in meta-analytic methodology (Table 5.4).
Specifically, IB/IM meta-analyses published after 2009 employed approaches
that were never used before 2009 for collecting data (e.g., published lit-
erature review and e-mail listservers), but fewer meta-analysts took the
advantages multiple databases after 2009. In addition, some methodological
limitations still arose after 2009. For example, homogeneity analysis and
publication bias are not as well accounted for as need be.
A simply defined ‘correct’ or ‘perfect’ way might not exist for conducting
and reporting a meta-analysis, but some approaches are superior to others.
IB/IM meta-analysts need to account for these to provide more consistent
and comprehensive analyses. In addition, there is variation in the simplicity
versus complexity as well as in the clarity versus confusedness among meta-
analytic procedures. But underlying this diversity of methods is a coherent
set of fundamentals that this chapter has discussed and with which we hope
to stimulate IB/IM researchers into conducting more and more effective
meta-analyses in the future.
 116

Table 5.2 Review of meta-analyses published in the Top-Five IB/IM Journals

Meta-Analytic Study Zhao et al. Tihanyi García-Meca

(2004) et al. (2005) & Sánchez-
Ballesta (2006)

Research topic Ownership- The effect The factors

based entry of cultural that influence
mode choice distance on the accuracy
entry mode of financial
choice, analysts’
international predictions
diversification,
and performance

Section 1: Research Stated research topic Y Y Y

specification
Theoretical and practical Y Y
context
Well-defined variable Y Y Y
Applicable to meta-analysis Y Y Y
Section 2: Literature Multiple databases Y Y
retrieval
Search in specific journals Y Y Y
Published reviews
Key terms
Combination of key terms
E-mail listserver
Contacting author Y Y
Selected literature
Non-English works
Section 3: Dataset Appropriate effect size metric Y Y Y
preparation
Inclusion criteria Y Y
Coding information Y Y
Multiple coders Y
Correcting for effect size
Identifying & handling
outliers
Section 4: Data Mean of effect sizes Y Y Y
analysis
Standard deviation/variance Y Y Y
Confidence intervals Y
Other statistics Y
Homogeneity analysis Y Y
Moderator analysis Y Y
Missing data Y
Publication availability bias Y
Additional data
Section 5: Findings Y Y Y
Meta-analytic report
Generality Y Y Y
Limitation Y Y Y
Figures/Graphs
Tables Y Y Y
Included studies Y Y Y
Count of Ys 18 17 20

Note: Y, Yes; a.r., available upon request from the meta-analysts.

 117

Bausch & Magnusson Fischer & Meyer & Reus & Rottig
Krist (2007) et al. (2008) Mansell (2009) Sinani (2009) (2009)

The relationship The effects Employee’s The local The determinants

between of cultural commitment spillovers of FDI of IJV’s
internationalization differences across cultures performance
and firm on MNE’s entry
performance strategy &
performance

Y Y Y Y Y

Y Y Y Y Y

Y Y Y Y Y
Y Y Y Y Y
Y Y

Y
Y Y
Y Y

Y
Y
Y Y

Y Y Y Y Y

Y Y Y Y
Y Y Y Y Y
Y
Y Y
Y

Y Y Y Y Y

Y Y
Y Y
Y Y Y
Y Y Y Y
Y Y Y Y Y
Y Y Y Y
Y
Y Y
Y Y Y Y Y

Y Y Y Y Y
Y Y Y Y
Y Y Y
Y Y Y Y Y
Y Y Y (a.r.) Y Y
20 17 23 23 21

(continued)
 118

Table 5.2 Continued

Meta-Analytic Study Stahl et al. Morschett et al. Steel &Taras

(2010) (2010) (2010)

Research topic The cultural The external The factors

diversity in antecedents of shaping
work teams entry mode culture, the
choice explanation
of cultural
variation,
and the
relationship
between
individual
and national
cultural values

Section 1: Research Stated research topic Y Y Y

specification Theoretical and practical Y Y Y
context
Well-defined variable Y Y Y
Applicable to meta-analysis Y Y Y
Section 2: Literature Multiple databases Y Y
retrieval Search in specific journals Y Y Y
Published reviews Y Y Y
Key terms Y
Combination of key terms Y
E-mail listserver Y
Contacting author Y
Selected literature Y
Non-English works
Section 3: Dataset Appropriate effect size metric Y Y Y
preparation Inclusion criteria Y Y Y
Coding information Y Y
Multiple coders Y Y
Correcting for effect size Y
Identifying & handling
outliers
Section 4: Data Mean of effect sizes Y Y Y
analysis Standard deviation/variance Y Y Y
Confidence intervals
Other statistics Y
Homogeneity analysis Y
Moderator analysis Y Y Y
Missing data Y Y
Publication availability bias Y
Additional data Y
Section 5: Findings Y Y Y
Meta-analytic report Generality Y Y Y
Limitation Y Y Y
Figures/Graphs Y Y Y
Tables Y Y Y
Included studies Y Y Y (a.r.)
Count of Ys 27 18 24
 119

Judge et al. (2011) van Essen et al. (2012) Taras et al. (2012) Yang & Driffield (2012)

The antecedents The relationship The national cultural The relationship between
and effects between firm indices multinationality and
of national performance performance
corruption and executive
compensation

Y Y Y Y
Y Y Y

Y Y Y Y
Y Y Y Y
Y Y
Y Y
Y
Y Y

Y
Y Y

Y Y Y Y
Y Y
Y Y Y
Y Y Y
Y Y

Y Y Y Y
Y Y
Y

Y
Y Y Y Y
Y Y
Y
Y Y
Y Y Y Y
Y Y Y Y
Y Y
Y Y
Y Y Y Y
Y Y Y Y
19 24 20 13
Table 5.3 Summary of major issues in meta-analyses published in the Top-Five IB/IM Journals
120

Total (15) JIBS (6) JWB (2) JIM (2) IBR (2) MIR (3)

Count Ratio Count Ratio Count Ratio Count Ratio Count Ratio Count Ratio

Section 1: Stated research topic 15 100% 6 100% 2 100% 2 100% 2 100% 3 100%
Research Theoretical and 13 87% 6 100% 2 100% 2 100% 1 50% 2 67%
specificaion practical context
Well-defined variable 15 100% 6 100% 2 100% 2 100% 2 100% 3 100%
Applicable to 15 100% 6 100% 2 100% 2 100% 2 100% 3 100%
meta-analysis
Section 2: Multiple databases 8 53% 3 50% 1 50% 1 50% 2 100% 1 33%
Literature Search in specific 9 60% 4 67% 1 50% 2 100% 1 50% 1 33%
retrieval journals
Published reviews 6 40% 4 67% 0 0% 2 100% 0 0% 0 0%
Key terms 5 33% 4 67% 1 50% 0 0% 0 0% 0 0%
Combination of 1 7% 1 17% 0 0% 0 0% 0 0% 0 0%
key terms
E-mail listserver 2 13% 0 0% 0 0% 1 50% 0 0% 1 33%
Contacting author 5 33% 5 83% 0 0% 0 0% 0 0% 0 0%
Selected literature 5 33% 1 17% 1 50% 1 50% 1 50% 1 33%
Non-English works 0 0% 0 0% 0 0% 0 0% 0 0% 0 0%
Section 3: Appropriate 15 100% 6 100% 2 100% 2 100% 2 100% 3 100%
Dataset effect size metric
preparation Inclusion criteria 11 73% 4 67% 2 100% 2 100% 2 100% 1 33%
Coding information 12 80% 5 83% 2 100% 1 50% 2 100% 2 67%
Multiple coders 7 47% 4 67% 2 100% 1 50% 0 0% 0 0%

Correcting for 5 33% 3 50% 1 50% 0 0% 0 0% 1 33%

effect size
Identifying & 1 7% 1 17% 0 0% 0 0% 0 0% 0 0%
handling outliers
Section 4: Mean of effect sizes 15 100% 6 100% 2 100% 2 100% 2 100% 3 100%
Data analysis Standard deviation / 10 67% 5 83% 0 0% 2 100% 1 50% 2 67%
variance
Confidence intervals 4 27% 0 0% 1 50% 0 0% 1 50% 2 67%
Other statistics 5 33% 3 50% 0 0% 1 50% 0 0% 1 33%
Homogeneity analysis 8 53% 5 83% 0 0% 0 0% 1 50% 2 67%
Moderator analysis 14 93% 5 83% 2 100% 2 100% 2 100% 3 100%
Missing data 9 60% 4 67% 1 50% 1 50% 2 100% 1 33%
Publication 4 27% 1 17% 0 0% 0 0% 1 50% 2 67%
availability bias
Additional data 5 33% 3 50% 1 50% 1 50% 0 0% 0 0%
Section 5: Findings 15 100% 6 100% 2 100% 2 100% 2 100% 3 100%
Meta-analytic Generality 15 100% 6 100% 2 100% 2 100% 2 100% 3 100%
report Limitation 12 80% 5 83% 2 100% 2 100% 1 50% 2 67%
Figures/Graphs 8 53% 2 33% 1 50% 2 100% 1 50% 2 67%
Tables 15 100% 6 100% 2 100% 2 100% 2 100% 3 100%
Included studies 15 100% 6 100% 2 100% 2 100% 2 100% 3 100%
Average per meta-analytic study 20.27 – 22.00 – 19.50 – 21.00 – 18.50 – 18.00 –
121
Table 5.4 Summary of major issues across two time periods 122

Total (15) 2004–2008 (5) 2009–2012 (10) Changes

between
Count Ratio Count Ratio Count Ratio Periods

Section 1: Stated research topic 15 100% 5 100% 10 100% –

Research Theoretical and practical context 13 87% 4 80% 9 90% Improved
specification Well-defined variable 15 100% 5 100% 10 100% –
Applicable to meta-analysis 15 100% 5 100% 10 100% –
Section 2: Multiple databases 8 53% 4 80% 4 40% Worsen
Literature Search in specific journals 9 60% 4 80% 5 50% Worsen
retrieval Published reviews 6 40% 0 0% 6 60% Improved
Key terms 5 33% 0 0% 5 50% Improved
Combination of key terms 1 7% 0 0% 1 10% Improved
E-mail listserver 2 13% 0 0% 2 20% Improved
Contacting author 5 33% 2 40% 3 30% Worsen
Selected literature 5 33% 2 40% 3 30% Worsen
Non-English works 0 0% 0 0% 0 0% No change
Section 3: Appropriate effect size metric 15 100% 5 100% 10 100% –
Dataset Inclusion criteria 11 73% 3 60% 8 80% Improved
preparation Coding information 12 80% 4 80% 8 80% No change
Multiple coders 7 47% 1 20% 6 60% Improved
Correcting for effect size 5 33% 0 0% 5 50% Improved
Identifying & handling outliers 1 7% 0 0% 1 10% Improved
Section 4: Mean of effect sizes 15 100% 5 100% 10 100% –
Data analysis Standard deviation/variance 10 67% 3 60% 7 70% Improved
Confidence intervals 4 27% 2 40% 2 20% Worsen
Other statistics 5 33% 2 40% 3 30% Worsen
Homogeneity analysis 8 53% 3 60% 5 50% Worsen
Moderator analysis 14 93% 4 80% 10 100% Improved
Missing data 9 60% 2 40% 7 70% Improved
Publication availability bias 4 27% 2 40% 2 20% Worsen
Additional data 5 33% 0 0% 5 50% Improved
Section 5: Findings 15 100% 5 100% 10 100% –
Meta-analytic Generality 15 100% 5 100% 10 100% –
report Limitation 12 80% 4 80% 8 80% No change
Figures/graphs 8 53% 1 20% 7 70% Improved
Tables 15 100% 5 100% 10 100% –
Included studies 15 100% 5 100% 10 100% –
 Meta-Analytic Research in International Business and International Management 123

References
Bausch, A., & Krist, M. (2007). The effect of context-related moderators on the interna-
tionalization-performance relationship: Evidence from meta-analysis. Management
International Review, 47, 319–347.
Bausell, R. B., Li, Y.-F., Gau, M.-L., & Soeken, K. L. (1995). The growth of meta-analytic
literature from 1980 to 1993. Evaluation & the Health Professions, 18, 238–251.
Bohlin, I. (2012). Formalizing syntheses of medical knowledge: The rise of meta-
analysis and systematic reviews. Perspectives on Science, 20, 273–309.
Brouthers, K. D., & Hennart, J.-F. (2007). Boundaries of the firm: Insights from inter-
national entry mode research. Journal of Management, 33, 395–425.
Buckley, P. J. (2002). Is the international business research agenda running out of
steam? Journal of International Business Studies, 33, 365–373.
Canabal, A., & White, G. O., III. (2008). Entry mode research: Past and future.
International Business Review, 17, 267–284.
Card, D., & Krueger, A. B. (1995). Time-series minimum-wage studies: A meta-analysis.
The American Economic Review, 85, 238–243.
Cooper, H. M., & Hedges, L. V. (1994). The handbook of research synthesis. New York,
NY: Russell Sage Foundation.
Cooper, H. M., Hedges, L. V., & Valentine, J. C. (2009). The handbook of research
synthesis and meta-analysis. New York, NY: Russell Sage Foundation.
Delios, A., & Beamish, P. W. (1999). Ownership strategy of Japanese firms:
Transactional, institutional, and experience influences. Strategic Management
Journal, 20, 915–933.
Dikova, D., & van Witteloostuijn, A. (2007). Foreign direct investment mode choice:
Entry and establishment modes in transition economies. Journal of International
Business Studies, 38, 1013–1033.
Farley, J. U., & Lehmann, D. R. (2001). The important role of meta-analysis in inter-
national research in marketing. International Marketing Review, 18, 70–79.
Fischer, R., & Mansell, A. (2009). Commitment across cultures: A meta-analytical
approach. Journal of International Business Studies, 40, 1339–1358.
García-Meca, E., & Sánchez-Ballesta, J. P. (2006). Influences on financial analyst fore-
cast errors: A meta-analysis. International Business Review, 15, 29–52.
Glass, G. V. (1976). Primary, secondary, and meta-analysis of research. Educational
Researcher, 5, 3–8.
Harzing, A.-W. (2012). Journal quality list, 2012. Retrieved from http://www.harzing.
com/jql.htm
Hedges, L. V., & Olkin, I. (1985). Statistical methods for meta-analysis. Orlando, FL:
Academic Press.
Hofstede, G. H. (1980). Cultures consequences: International differences in work-related
values. Beverly Hills, CA: Sage.
Hofstede, G. H., Hofstede, G. J., & Minkov, M. (2010). Cultures and organizations:
Software of the mind: intercultural cooperation and its importance for survival (3rd ed.).
New York, NY: McGraw-Hill.
Hunter, J. E., & Schmidt, F. L. (1990). Methods of meta-analysis: correcting error and bias
in research findings. Newbury Park, CA: Sage.
Hunter, J. E., & Schmidt, F. L. (2004). Methods of meta-analysis: Correcting error and bias
in research findings. Beverly Hills, CA: Sage.
ISI (2012). Journal citation reports. Thomson Reuters, 2012. Retrieved from http://admin-
apps.webofknowledge.com.ezproxy.lib.uts.edu.au/JCR/JCR?SID=U1j22FJ1D56
cj4CcJlb&locale=en_US
124 Peter J. Buckley et al.

Judge, W. Q., McNatt, D. B., & Xu, W. (2011). The antecedents and effects of national
corruption: A meta-analysis. Journal of World Business, 46, 93–103.
Kirca, A. H., & Yaprak, A. (2010). The use of meta-analysis in international business
research: Its current status and suggestions for better practice. International Business
Review, 19, 306–314.
Kulik, J. A., Cohen, P. A., & Ebeling, B. J. (1980). Effectiveness of programmed instruc-
tion in higher education: A meta-analysis of findings. Educational Evaluation and
Policy Analysis, 2, 51–64.
Kulik, J. A., Kulik, C.-L. C., & Cohen, P. A. (1980). Effectiveness of computer-based
college teaching: A meta-analysis of findings. Review of Educational Research, 50,
525–544.
Lau, J., Antman, E. M., Jimenez-Silva, J., Kupelnick, B., Mosteller, F., & Chalmers, T. C.
(1992). Cumulative meta-analysis of therapeutic trials for myocardial infarction.
New England Journal of Medicine, 327, 248–254.
Lee, W.-L., Bausell, R. B., & Berman, B. M. (2001). The growth of health-related meta-
analyses published from 1980 to 2000. Evaluation and the Health Professions, 24,
327–335.
Lipsey, M. W., & Wilson, D. B. (2001). Practical meta-Analysis. Thousand Oaks, CA:
Sage.
Magnusson, P., Baack, D. W., Zdravkovic, S., Staub, K. M., & Amine, L. S. (2008). Meta-
analysis of cultural differences: Another slice at the apple. International Business
Review, 17, 520–532.
Malhotra, N. K., Agarwal, J., & Ulgado, F. M. (2003). Internationalization and
entry modes: A multi-theoretical framework and research propositions. Journal of
International Marketing, 11, 1–31.
Meyer, K. E., & Sinani, E. (2009). When and where does foreign direct investment
generate positive spillovers? A meta-analysis. Journal of International Business Studies,
40, 1075–1094.
Morschett, D., Schramm-Klein., H., & Swoboda, B. (2010). Decades of research on
market entry modes: What do we really know about external antecedents of entry
mode choice? Journal of International Management, 16, 60–77.
Peterson, R. A., & Jolibert, A. J. P. (1995). A meta-analysis of country-of-origin effects.
Journal of International Business Studies, 26, 883–900.
Reus, T. H., & Rottig, D. (2009). Meta-analyses of international joint venture perfor-
mance determinants: Evidence for theory, methodological artifacts and the unique
context of China. Management International Review, 49, 607–640.
Rosenthal, R. (1979). The file drawer problem and tolerance for null results.
Psychological Bulletin, 86, 638–641.
Rosenthal, R. (1984). Meta-analytic procedures for social research. Beverly Hills, CA: Sage.
Rosenthal, R. (1991). Meta-analytic procedures for social research (Rev. ed.). Newbury
Park, CA: Sage.
Schmidt, F. L., & Hunter, J. E. (1977). Development of a general solution to the prob-
lem of validity generalization. Journal of Applied Psychology, 62, 529–540.
Slangen, A., & Hennart, J.-F. (2007). Greenfield or acquisition entry: A review
of the empirical foreign establishment mode literature. Journal of International
Management, 13, 403–429.
Stahl, G. K., Maznevski, M. L., Voigt, A., & Jonsen, K. (2010). Unraveling the effects
of cultural diversity in teams: A meta-analysis of research on multicultural work
groups. Journal of International Business Studies, 41, 690–709.
Stanley, T. D., & Jarrell, S. B. (1989). Meta-regression analysis: A quantitative method
of literature surveys. Journal of Economic Surveys, 3, 161–170.
 Meta-Analytic Research in International Business and International Management 125

Steel, P., & Taras, V. (2010). Culture as a consequence: A multi-level multivariate meta-
analysis of the effects of individual and country characteristics on work-related
cultural values. Journal of International Management, 16, 211–233.
Takahashi, K., Ishikawa, J., & Kanai, T. (2012). Qualitative and quantitative studies
of leadership in multinational settings: Meta-analytic and cross-cultural reviews.
Journal of World Business, 1–9. In press.
Taras, V., Steel, P., & Kirkman, B. L. (2012). Improving national cultural indices using
a longitudinal meta-analysis of Hofstedes dimensions. Journal of World Business, 47,
329–341.
Thomas, D. C., Cuervo-Cazurra, A., & Brannen, M. Y. (Eds.). (2011). Journal of
International Business Studies, (42, pp. 1073–1078).
Thompson, S. G., & Sharp, S. J. (1999). Explaining heterogeneity in meta-analysis:
A comparison of methods. Statistics in Medicine, 18, 2693–2708.
Tihanyi, L., Griffith, D. A., & Russell, C. J. (2005). The effect of cultural distance on
entry mode choice, international diversification, and MNE performance: A meta-
analysis. Journal of International Business Studies, 36, 270–283.
van Essen, M., Heugens, P. P., Otten, J., & van Oosterhout, J. H. (2012). An institution-
based view of executive compensation: A multilevel meta-analytic test. Journal of
International Business Studies, 43, 396–423.
Yang, Y., & Driffield, N. (2012). Multinationality-performance relationship.
Management International Review, 52, 23–47.
Zhao, H., Luo, Y., & Suh, T. (2004). Transaction cost determinants and ownership-
based entry mode choice: A meta-analytical review. Journal of International Business
Studies, 35, 524–544.
Appendix: Meta-analyses in top 5 IB Journals between 2004 and 2012
126

Panel A

Reference Journal Research Specification Literature Retrieval

Research Topic Theoretical paradigm Databases Journal searched Restriction Key terms

Zhao et al. JIBS Ownership-based Transaction ABI/Inform JIBS, MIR, AMJ, No n.r.
(2004) entry mode choice Cost Theory SMJ
Tihanyi et al. JIBS The effect of cultural n.r. ABI/Inform; JSTOR AMJ, ASQ, JIBS, No n.r.
(2005) distance on entry mode JM, JMS, MIR,
choice, international MS, SMJ
diversification, and MNE
Performance
García-Meca & IBR The factors that n.r. ScienceDirect; EJS JAR, JAE, TAR, No n.r.
Sánchez influence the accuracy Ebsco; SSRN; ABI JBR, JB, FA, AQS,
Ballesta (2006) of financial analysts’ Inform IBR, JFE
predictions
Bausch & MIR the relationship between Internalization; Business Source Premier; SMJ, AMJ, JIBS, No n.r.
Krist, 2007 internationalization and Learning; RBV; EconLit; ABI/Inform JBV, MIR, IBR
firm performance Eclectic
Magnusson IBR The effects of cultural Internationalization; ABI/Inform; Business n.r. No n.r.
et al. (2008) differences on MNE’s transaction cost Source Premier
entry strategy and theory
performance
Fischer & JIBS Employee’s commitment Commitment and PsycINFO n.r. No ‘Organizational
Mansell (2009) across cultures cultural theory commitment’
Meyer & JIBS The local spillovers Competitive Dynamics EconLit n.r. No ‘spillovers from
Sinani (2009) of FDI Theory (awareness- technology
motivation-capability transfer’,
framework) productivity
FDI spillovers’
Reus & MIR The determinants The Agency Theory, the ABI/Inform n.r. No n.r.
Rottig (2009) of IJV’s performance behavioural perspective
Stahl JIBS The cultural diversity Similarity-attraction, ABI/INFORM, Business Relevant books No ‘Team’, ‘group’;
et al. (2010) in work teams social identify & Source Premier, EconLit, and research ‘culture’,
categorization, PsychInfo, Science journals ‘diversity’;
information- Direct, and the Social ‘multicultural’,
processing Science Citation Index ‘international’,
‘multinational’
Morschett JIM The external antecedents Learning perspective Business Source JIBS, JIM, IBR, No n.r.
et al. (2010) entry mode choice of organizational Premier JBR, MIR
capabilities;
Transaction Cost
Theory
Steel & JIM The factors shaping Ecological Inference; EBSCO; PsycINFO; All relevant in No n.r.
Taras (2010) culture, the explanation Divergence & ERIC; ProQuest; and 1980–2006
of cultural variation, and Convergence Theory; ProQuest Digital
the relationship between Hofstede’s paradigm Dissertations;
individual and national
cultural values
Judge JWB The antecedents and Institutional choice ABI/Inform n.r. No ‘corruption’;
et al. (2011) effects of national perspective ‘many
corruption countries’
van Essen JIBS The relationship Institutional Based ABI/INFORM Global, 25 journals No ‘compensation’,
et al. (2012) between firm View EconLit, Google Scholar, ‘incentives’,
performance and JSTOR, SSRN, and ISI ‘pay’,
executive compensation Web of Knowledge ‘remuneration’,
‘salary’, ‘stock
option’
Taras JWB The national cultural Hofstede’s dimensions All major electronic Almost 100 No n.r.
et al. (2012) indices databases journals of
1980–2010
Yang & MIR The relationship between n.r. n.r. n.r. No n.r.
Driffield multinationality and
127

(2012) performance

(continued)
Appendix Continued
Panel A
128

Reference Journal Literature Retrieval Dataset Preparation

Terms Other efforts Effect size Criteria for Coding Correcting for Identification
Combination metric inclusion reliability effect size of outliers

Zhao JIBS n.r. Consulting Partial Ownership-based n.r. n.r. n.r.

et al. (2004) other correlation entry mode; firm
researchers level; variables of
transaction cost
theory; 1986–2002
Tihanyi JIBS n.r. Communicating Correlation r n.r. Two coders; n.r. n.r.
et al. (2005) with researchers Cohen’s κ
García-Meca & IBR n.r. Including Correlation r Empirical; statistical n.r. n.r. n.r.
Sánchez-Ballesta unpublished information;
(2006) studies independence
Bausch MIR n.r. reference part correlation r n.r. n.r. n.r. n.r.
& Krist, 2007 search of selected
literature;
searching on
the home pages
of researchers
Magnusson IBR n.r. Reference part Correlation r Correlations about n.r. n.r. n.r.
et al. (2008) of selected variables of interest;
articles sample size
Fischer & JIBS n.r. Consulting Percent of Non-experimental Two coders; Fisher’s z n.r.
Mansell (2009) published maximum studies; commitment the percentage
meta-analyses possible among working agreement
and reviews score (POMP team; after 1990
about the mean);
research topic; correlation r
contacting 25
researchers for
unpublished
data
Meyer & JIBS n.r. Internet; t-statistics Empirical; particular n.r. n.r. 10 times larger
Sinani (2009) review papers; estimations than the mean
including
unpublished
studies
Reus & MIR n.r. Listserves of Correlation r Correlation matrices n.r. By reliabilities, n.r.
Rottig (2009) AIB and AOB; or t-statistics; error variance, and
relevant books independence; correlation between
and articles variables of perfectly measured
interest; the variable
operationalizations
of constructs; the
same constructs
defined in another
study
Stahl et al. JIBS AND, OR Bibliographies, Correlation r Statistical Two coders; By sampling n.
(2010) conference (point- information; Cohen’s φ (point-biserial),
proceedings, biserial and independence reliability (selfreported
Internet searches; product- data), and Fisher’s z r.
including moment (product-moment)
unpublished correlation)
studies;
consulting other
researchers;
Morschett JIM n.r. Published Regression Overseas value-added n.r. n.r. n.r.
et al. (2010) reviews coefficient process (i.e.; no
(log odds export)
ratio)
Steel & JIM n.r. Major books Standardized Hofstede’s model Double coded; n.r. n.r.
Taras (2010) and reviews; Cohen’s d with original data; inconsistencies
reference part (mean) codable data; resolved by
of selected list studies; commensurability comparison
servers of mailing with Hofstede’s
the AIB and instruments
AOM; including
unpublished
129

studies

(continued)
Appendix Continued
Panel B
130

Reference Journal Literature Retrieval Dataset Preparation

Terms Other efforts Effect size Criteria for Coding Correcting for Identification
Combination metric inclusion reliability effect size of outliers

Judge JWB n.r. n.r. Correlation r Conceptualized Two coders; By the mean n.r.
et al. (2011) corruption; national interrater reliability;
level; multiple reliability
countries; empirical
studies; direct
relationship
van Essen JIBS n.r. Reviews and Pearson n.r. Two coders; Fisher’s z n.r.
et al. (2012) meta-analyses; correlations; subsample;
reference part partial Cohen’s κ
search (forward- correlation
tracing) of coefficients
selected articles;
contacting
authors;
including
unpublished
Taras JWB n.r. Reference parts Standardized Commensurability; Multiple n.r. n.r.
et al. (2012) of selected Cohen’s d instruments based coders;
studies; citation (mean) on Hofstede; test-retest
checking by empirical support for reliability
Google Scholar; convergent validity
mailing list-serves
of AIB and AOM;
unpublished
papers;
contacting
authors
Yang & Driffield MIR n.r. n.r. Estimate n.r. n.r. n.r. n.r.
(2012) of linear
coefficient
 Panel B

Reference Journal Data Analysis Other

Information

Basic statistical Homogeneity Moderator Missing data Publication # of primary

information Test analysis availability bias studies (n)

Zhao JIBS Sample size weighted mean; Q statistics Category analyses n.r. n.r. 38
et al. (2004) standard error; z-value;
p-value; % of variance;
Tihanyi JIBS Sample size weighted mean; n.r. n.r. n.r. n.r. 66
et al. (2005) variances
García-Meca & IBR Sample size weighted mean; Q statistics Category analysis If other statistics Fail-safe N 38
Sánchez-Ballesta variances; confidence are available, then
(2006) intervals transfer
Bausch & MIR Sample size weighted mean; 75% rule Category analysis n.r. File-drawer 41
Krist (2007) confidence & credibility analysis;
intervals; z-statistics; calculation of
average residual variance fail-safe N(x)
Magnusson IBR Sample size weighted n.r. Category analysis Contacting authors n.r. 74
et al. (2008) mean
Fischer & JIBS Sample size weighted; Q statistics Category analysis; Missing standard n.r. 164 (means)
Mansell (2009) intercorrelation and three-level deviation is handled and 37
rankorder correlation hierarchical linear by the means; if (correlations)
modelling missing country
data, take region
Meyer & JIBS Mean, standard deviation, Heterogeneity Meta-analytic Missing data of a n.r. 66
Sinani (2009) correlation matrix test regression year replaced by the
nearest year
Reus & MIR Sample size weighted mean; 75% rule; Category analysis Missing reliability n.r. 66
Rottig (2009) variances; chi-square Q statistics handled by an
artifact distribution
method
131

(continued)
Appendix Continued
Panel B
132

Reference Journal Data Analysis Other

Information

Basic statistical Homogeneity Moderator Missing data Publication # of primary

information Test analysis availability bias studies (n)

Stahl JIBS Sample size, mean, Q statistics Sub-sample Eliminating the Fail-safe N 108
et al. (2010) confidence interval, range, comparison by missing categories
variance Z-statistics
Morschett et al. JIM Variance and sample-size Q statistics; Sub-sample n.r. n.r. 72
(2010) weighted mean forest plot comparison on
vote counting
scale
Steel & JIM Sample size weighted n.r. Hierarchical Missing GDP data n.r. 508
Taras (2010) mean, standard deviation, linear model were assumed
correlation matrix by that prior to
publication
Judge JWB Sample size weighted mean; n.r. Category analysis Contacting author; n.r. 42
et al. (2011) confidence and credibility missing reliability
intervals replaced by
estimate of National
Academy of Science
van Essen JIBS Inverse variance weighted Q statistics; Hierarchical linear Contacting author n.r. 332
et al. (2012) mean, standard error I index model, weighted
least squares-
based regression
Taras JWB Mean n.r. Category analysis n.r. n.r. 451
et al. (2012)
Yang & MIR Mean, standard deviation, n.r. Meta-analytic n.r. Regressing 67
Driffield (2012) sample size regression t-ratio of each
estimate
Reference Journal Other Information

Cumulative # of effect # of Number of Key references of meta-analytic Number of

sample Size (N) sizes (k) variables moderators procedure hypotheses
from
literature From Additional
literature sources

Zhao JIBS 24,111 106 5 5 n.r. Hunter & Schmidt (1990); No

et al. (2004) Hedges & Olkin (1985)
Tihanyi JIBS 9,135 n.r. 6 n.r. n.r. Hunter & Schmidt (1990) 5
et al. (2005)
García-Meca & IBR 1,245,098 59 7 3 n.r. Lipsey & Wilson (2001); No
Sánchez- Rosenthal (1991); Hunter &
Ballesta (2006) Schmidt (1990)
Bausch & MIR 7,792 146 6 5 n.r. Hunter & Schmidt (1990); 7
Krist (2007) Rosenthal (1979)
Magnusson IBR 69,849 36 5 3 n.r. Hunter & Schmidt (1990) No
et al. (2008)
Fischer & JIBS 44,424 n.r. 3 10 8 Lipsey & Wilson (2001); 8
Mansell (2009) (means) Hedges & Olkin (1985)
and 10,533
(correlations)
Meyer & JIBS 121 n.r. 1 8 7 Stanley & Jarrell (1989); 4
Sinani (2009) Lau et al. (1992);
Thompson & Sharp (1999)
Reus & Rottig (2009) MIR 26,927 165 5 7 n.r. Hunter & Schmidt (2004) 8
Stahl JIBS 10,632 135 7 6 n.r. Hunter & Schmidt (1990); 34
et al. (2010) Hedges & Olkin (1985); Lipsey &
Wilson (2001), Rosenthal (1984)
Morschett JIM 50,974 156 14 2 n.r. Cooper & Hedges (1994); 14
et al. (2010) Lipsey & Wilson (2001); Kulik,
Cohen & Ebeling (1980); Kulik,
Kulik & Cohen (1980)
133

(continued)
Appendix Continued
Panel B
134

Reference Journal Other Information

Cumulative # of effect # of Number of Key references of meta-analytic Number of

sample Size (N) sizes (k) variables moderators procedure hypotheses
from
literature

Steel & JIM 2,115 36 8 4 3 Hunter & Schmidt (1990) 9

Taras (2010)
Judge JWB 3,170 511 3 17 n.r. Hunter & Schmidt (2004) 8 propositions
et al. (2011)
van Essen JIBS 659,810 (r) & 592 (r) & 2 53 11 Lipsey & Wilson (2001); 4
et al. (2012) 4,107,639 2415 (partial-r) Hedges & Olkin (1985)
(partial r)
Taras JWB 225,177 914 4 1 22 Hunter & Schmidt (2004) No
et al. (2012)
Yang & MIR 906,480 370 1 11 n.r. Card & Krueger (1995) No
Driffield (2012)

Journals in IB are defined by Journal Qualtity List (47th ed.) (Harzing, 2012).
The rank is made according to Total Cites and Impact Factor of Journal Citation Reports (Web of Knowledge, 2011).
The above listed are published between 2000 and 2012; only one before 1995 by Peterson & Jolibert in JIBS; no meta-analyses were published between
2000 and 2003.
‘n.r.’ means not reported.
AIB – The Academy of International Business; AOM – the International Management Division of the Academy of Management; JIBS – the Journal of Inter-
national Business Studies; MIR –Management International Review; AMJ – Academy of Management Journal; SMJ – Strategic Management Journal; ASQ – Admin-
istrative Science Quarterly; AOS – Accounting, Organization and Society; JM – Journal of Management; FA – Finance and Accounting; JB – Journal of Business; JMS
– Journal of Management Studies; MS – Management Science; JAR – Journal of Accounting Research; JAE – Journal of Accounting and Economics; TAR – The
Accounting Review; JBR – Journal of Business Research; IBR – International Business Review; JFE – Journal of Financial Economics; JBV – Journal of Business Venturing.
n is the number of studies (i.e., the sample size of a meta-analysis).
k is the number of effect sizes, which may or may not equal n. When using correlation r, for instance, if a study reports 2 r’s regarding interested
variables, the n will be smaller than k.
N is the number of cumulative samples size, which is the total amount of sample sizes in the primary studies included in a meta-analysis.