Professional Documents
Culture Documents
RESEARCH NOTE
transformational leadership and employee innova- team, as well as instances where observations of
tion at the individual and team levels. innovation for individual team members were
collected and then aggregated to the team level
Hypothesis 1: High-UA countries will show
during analysis. Each team-level observation
stronger relationships between supervisor trans-
includes data from a single team and a single
formational leadership and individual innovation
supervisor to whom the team reports.
than low-UA countries.
Hypothesis 2: High-UA countries will show Coding of Study Characteristics
stronger relationships between supervisor trans- All study characteristics were coded by two trained
formational leadership and team innovation than judges, obtaining an initial 91% agreement. Fol-
low-UA countries. lowing a consensus meeting, any remaining dis-
crepancies were resolved by the head judge (i.e., the
second author). The final agreement was 100% on
METHOD all variables.
Selection of Primary Studies Effect size and other statistical information
Our literature search concluded in September 2016. We collected the observed effect size (r) between
First, keyword searches were conducted in nine supervisor transformational leadership and
databases (e.g., PsycINFO, ProQuest Dissertations employee innovation, along with reliability infor-
and Theses). Second, going 10 years back, manual mation (i.e., Cronbach’s alpha). For the few samples
searches of 11 journals were conducted (e.g., Journal that did not report reliability information, estimates
of International Business Studies, The Leadership were imputed using the mean reliability across all
Quarterly). Third, the references of six major review samples in the meta-analysis. When studies reported
articles were examined (e.g., Anderson et al., 2014). multiple effect sizes for employee innovation or
Fourth, recent papers presented at major academic transformational leadership, a composite correla-
conferences were searched. Fifth, researchers who tion was computed (equations 10.6 and 10.7 in
published recently on these topics were emailed a Hunter & Schmidt, 2004) to maintain the assump-
request for unpublished work. tion of independence. Composite reliability esti-
Over 10,000 papers were initially identified. After mates were computed using Mosier’s reliability
removing duplicate papers, non-empirical papers, formula (equation 10.15). Sample size (n) refers to
and papers with abstracts that indicated our key number of individual employees or teams.
variables were absent, 929 papers remained. Seven
inclusion criteria were then applied: (1) conducted Supervisor transformational leadership
in a field setting, (2) contained information on Primary studies using construct labels such as
both transformational leadership and employee transformational leadership, charismatic leader-
innovation, (3) collected data at the individual- or ship, and inspirational motivation were coded as
team-level of analysis, (4) clearly described trans- transformational leadership. The average Cronbach’s
formational leadership behaviors or styles, rather alpha was .91 (SD = .05).
than potential antecedents to (e.g., traits) or out-
comes of (e.g., climate) transformational leader- Employee innovation
ship, (5) clearly described measures of innovation, Primary studies using construct labels such as
rather than related constructs (e.g., innovation employee creativity, idea implementation, and
adoption), (6) contained information needed to innovation were coded as employee innovation. The
compute an effect size, and (7) if multiple studies average Cronbach’s alpha was .90 (SD = .07) for
were published using the same dataset, only one individual-level innovation and .86 (SD = .07) for
was retained. These criteria resulted in a final team-level innovation.
database of 51 independent samples collected at
the individual level of analysis (n = 14,560 individ- Uncertainty avoidance
uals) and 30 independent samples collected at the UA scores for each country were drawn from
team level of analysis (n = 3129 teams), represent- Hofstede et al.’s (2010) index. While assigning
ing a total of 17 countries. Team-level samples country-level UA scores to our samples ignores
included instances where respondents were asked within-country variability (Kirkman, Lowe, & Gib-
to provide observations of innovation for an entire son, 2006), societal-level values data serve as robust
approximations of the cultural knowledge, SDq = .20) and team innovation (q = .31, SDq-
resources, structures, and norms held by a society = .17). The percentage of variance in effect sizes
(Peterson & Barreto, 2014, 2018). Hofstede’s UA explained by sampling error and statistical artifacts
index ranges from 0 to 100 based on data from was relatively small at the individual (9%) and team
multiple international studies over several decades. (28%) levels, suggesting a large degree of between-
Because between-country comparisons of culture sample heterogeneity that might be explained by
scores are relatively stable over time (Beugelsdijk, moderators.
Maseland, & Hoorn, 2015), Hofstede et al.’s (2010) Categorical meta-analysis procedures were used
index was used to code all samples. In order to to estimate whether high and low levels of UA
maximize power for subgroup analysis, countries explain the relationship between supervisor trans-
were split into low (0–50%) and high (51–100%) formational leadership and employee innovation at
categories for UA based on percentiles. the individual and team levels. In the hierarchical
categorical analyses, because the k-size is in some
Controls cases small, we focused on interpreting trends in
In studies where supervisors were the source of the mean corrected correlations, rather than null
both leadership and innovation data, or studies hypothesis significance tests (e.g., Aguinis, Stur-
where followers were the source of both leadership man, & Pierce, 2008). Nevertheless, p values reflect-
and innovation data, we anticipated that correla- ing the statistical significance of effect size
tions would be stronger due to the potential for contrasts are presented for interested readers. In
confirmation bias and social desirability bias. Thus, addition, an alternative presentation of the results
if leadership and innovation data were collected based on meta-regression procedures is available as
from the same source, the study was coded as 1 for supplementary materials.
yes; otherwise, it was coded as 0 for no when For individual-level innovation (see Table 1),
common source measurement was not present. effect sizes from high-UA countries (q = .37, SDq-
Criterion objectivity was also coded as a potential = .24) trended higher than those from low-UA
control, but only three studies used objective countries (q = .25, SDq = .15). This difference in
criteria (e.g., number of patents), thus this variable effect sizes was not statistically significant
was dropped. We sought to control for other (p value = .279). Studies that used common
potentially relevant context variables (e.g., indus- sources of measurement for leadership and inno-
try, profit vs. non-profit, leader tenure, organiza- vation variables showed stronger relationships
tional size), but they could not be included due to compared with studies that used non-common
insufficient reporting. sources. After separating common source from
non-common source studies, the general trend for
Analytic Procedures UA remained. For common source studies, high-
Psychometric meta-analysis (Hunter & Schmidt, UA countries (q = .43, SDq = .25) trended toward
2004) was used to analyze the data with the larger effect sizes than low-UA countries (q = .31,
‘‘psychmeta’’ package in R. Correlation coefficients SDq = .07), and these differences in effect sizes
were corrected for unreliability in the predictor and approached the level of statistical significance
criterion based on Cronbach’s alpha estimates. (p value = .077). For non-common source studies,
Both the uncorrected, weighted, average effect (r) high-UA countries (q = .28, SDq = .19) also trended
and the unreliability-corrected, weighted, average higher than low-UA countries (q = .23, SDq = .16),
effect (q) are presented. When the k-size was 5 or and this difference in effect sizes was statistically
greater, sensitivity analyses were conducted to significant (p value = .003). This pattern remained
determine the extent to which effect size estimates consistent after removing studies flagged as out-
may have been biased by outliers (Viechtbauer & liers, providing support for hypothesis 1.
Cheung, 2010). Further details on the method and For team innovation (see Table 2), estimates of
data are available as supplementary materials. effect sizes from high-UA countries (q = .36, SDq-
= .18) trended stronger than those from low-UA
countries (q = .21, SDq = .06). This difference in
RESULTS
effect sizes was statistically significant
Supervisor transformational leadership demon-
(p value \ .001). Once again, common source
strated a moderate-sized relationship with both
studies demonstrated stronger relationships. For
individual employee innovation (q = .30,
common source studies, too few studies were
Table 1 Individual-level meta-analysis of supervisor transformational leadership and employee innovation with robustness checks
Moderators k N r SDr q SDq CILL CIUL CVLL CVUL %Var Moderator tests
Contrast p value
Overall 51 14,560 .27 .18 .30 .20 .24 .35 .04 .55 8.55
Overall (wo) 48 12,609 .22 .14 .24 .14 .20 .28 .07 .42 22.93
Uncertainty avoidance (UA)
1. High 18 5658 .35 .21 .37 .24 .26 .49 .07 .68 5.05 2 vs. 4 .279
2. High (wo) 16 4055 .24 .14 .26 .14 .18 .33 .07 .44 22.20
3. Low 33 8902 .23 .15 .25 .15 .19 .30 .06 .44 15.53
4. Low (wo) 32 8554 .21 .13 .23 .13 .18 .29 .06 .40 27.87
Common source
1. UA High 9 3376 .40 .22 .43 .25 .26 .60 .11 .75 3.50 2 vs. 3 .077
2. UA High (wo) 7 1714 .35 .13 .38 .13 .27 .48 .21 .54 18.10
3. UA Low 8 2347 .29 .08 .31 .07 .25 .38 .23 .40 40.82
Non-common source
1. UA High 9 2282 .27 .16 .28 .19 .15 .41 .05 .52 11.03 2 vs. 4 .003
2. UA High (wo) 6 1435 .19 .02 .21 .00 .19 .23 .21 .21 100.00
3. UA Low 25 6555 .20 .16 .23 .16 .16 .29 .02 .43 14.08
4. UA Low (wo) 20 4753 .12 .08 .13 .06 .09 .17 .05 .21 56.17
Hunter and Schmidt’s (2004) moderator test was used as described in Aguinis et al. (2008). Where SDq = 0, a pooled SD was calculated in order to
conduct the t test (see Borenstein, Hedges, Higgins, & Rothstein, 2011, p. 172).
k number of independent samples; N total sample size; r sample-size-weighted mean observed correlation; SDr standard deviation of observed
correlations; mean true-score correlation in the Hunter and Schmidt meta-analytic tradition (corrected for unreliability in the independent and
dependent variables); SDq standard deviation of corrected correlations; CILL and CIUL lower and upper bounds of the 95% confidence interval around
the corrected correlations; CVLL and CVUL lower and upper bounds of the 80% credibility interval around the corrected correlations; %Var percentage
of variance attributable to sampling error and statistical artifacts; wo without outliers; UA uncertainty avoidance.
Table 2 Team-level meta-analysis of supervisor transformational leadership and employee innovation with robustness checks
Moderators k N r SDr q SDq CILL CIUL CVLL CVUL %Var Moderator tests
Contrast p value
Overall 30 3129 .27 .17 .31 .17 .24 .38 .09 .52 27.84
Overall (wo) 27 2752 .23 .12 .26 .09 .21 .31 .14 .37 58.67
Uncertainty avoidance (UA)
1. High 16 2047 .32 .18 .36 .18 .26 .46 .12 .59 19.94 2 vs. 4 \ .001
2. High (wo) 14 1788 .26 .11 .30 .09 .23 .37 .18 .41 50.80
3. Low 14 1082 .19 .13 .21 .06 .14 .29 .13 .29 79.68
4. Low (wo) 13 964 .17 .11 .19 .00 .12 .26 .19 .19 100.00
Common source
1. UA High 5 816 .34 .15 .39 .15 .24 .54 .20 .58 22.48 – –
2. UA High (wo) 4 664 .28 .09 .32 .05 .23 .42 .26 .39 71.90
3. UA Low 1 43 – – – – – – – – –
Non-common source
1. UA High 11 1231 .30 .19 .33 .20 .21 .46 .08 .58 19.62 2 vs. 4 .019
2. UA High (wo) 10 1124 .26 .13 .28 .11 .19 .37 .15 .42 48.88
3. UA Low 13 1039 .20 .11 .23 .04 .16 .30 .18 .28 90.64
4. UA Low (wo) 12 921 .18 .10 .20 .00 .14 .26 .20 .20 100.00
Hunter and Schmidt’s (2004) moderator test was used as described in Aguinis et al. (2008). Where SDq = 0, a pooled SD was calculated in order to
conduct the t-test (see Borenstein et al., 2011, p. 172).
k number of independent samples; N total sample size; r sample-size-weighted mean observed correlation; SDr standard deviation of observed
correlations; mean true-score correlation in the Hunter and Schmidt meta-analytic tradition (corrected for unreliability in the independent and
dependent variables); SDq standard deviation of corrected correlations; CILL and CIUL lower and upper bounds of the 95% confidence interval around
the corrected correlations; CVLL and CVUL lower and upper bounds of the 80% credibility interval around the corrected correlations; %Var percentage of
variance attributable to sampling error and statistical artifacts; wo without outliers; UA uncertainty avoidance.
we could not examine relationships for subdimen- leadership by concentrating these investments in
sions. Thus, which particular transformational high-UA countries where it helps employee inno-
leadership behaviors serve as the most important vation the most.
resources for innovation in different cultures forms
a topic for future research.
ACKNOWLEDGEMENTS
CONCLUSION We would like to thank Kelsey Medeiros, Tyler
Using a meta-analytic framework, we examined the Mulhearn, and Tristan McIntosh who contributed as
boundary conditions of the relationship between coders, Michael Mumford who inspired the initial
transformational leadership and employee innova- literature review, and the many scholars who
tion within a cross-cultural context. We found that responded to our calls for primary data. We also thank
supervisor transformational leadership relates pos- our Editor, Dr. Mark Peterson, and three anonymous
itively to employee innovation in most countries reviewers, whose thoughtful recommendations signif-
and may serve as a particularly beneficial resource icantly improved the quality of this manuscript. An
for employee innovation in cultures marked by earlier version of this work was presented at the 33rd
higher levels of UA. These findings suggest that annual conference of the Society for Industrial and
organizations might achieve a higher return on Organizational Psychology, Chicago, IL.
their investments in supervisor transformational
REFERENCES
Aguinis, H., Sturman, M. C., & Pierce, C. A. 2008. Comparison Elenkov, D. S., & Manev, I. M. 2005. Top management
of three meta-analytic procedures for estimating moderating leadership and influence on innovation: The role of sociocul-
effects of categorical variables. Organizational Research Meth- tural context. Journal of Management, 31(3): 381–402.
ods, 11(1): 9–34. Engelen, A., Schmidt, S., Strenger, L., & Brettel, M. 2014. Top
Amabile, T. M. 1988. A model of creativity and innovation in management’s transformational leader behaviors and innova-
organizations. Research in Organizational Behavior, 10(1): tion orientation: A cross-cultural perspective in eight coun-
123–167. tries. Journal of International Management, 20: 124–136.
Anderson, N., Potočnik, K., & Zhou, J. 2014. Innovation and Ergeneli, A., Gohar, R., & Temirbekova, Z. 2007. Transforma-
creativity in organizations: A state-of-the-science review, tional leadership: Its relationship to culture value dimensions.
prospective commentary, and guiding framework. Journal of International Journal of Intercultural Relations, 31(6): 703–724.
Management, 40(5): 1297–1333. Fahr, J. L., Podsakoff, P. M., & Cheng, B. S. 1987. Culture-free
Avolio, B. J., & Bass, B. M. 1995. Individual consideration viewed leadership effectiveness versus moderators of leadership
at multiple levels of analysis: A multi-level framework for behaviors: An extension and test of Kerr and Jermier’s
examining the diffusion of transformational leadership. The ‘substitutes for leadership’ model in Taiwan. Journal of Inter-
Leadership Quarterly, 6(2): 199–218. national Business Studies, 18(3): 43–60.
Bass, B. M. 1997. Does the transactional–transformational Fischer, R., et al. 2019. Does organizational formalization
leadership paradigm transcend organizational and national facilitate voice and helping organizational citizenship behav-
boundaries? American Psychologist, 52(2): 130–139. iors? It depends on (national) uncertainty norms. Journal of
Bass, B. M., & Riggio, R. E. 2006. Transformational leadership. International Business Studies, 50(1): 125–134.
London: Psychology Press. Gong, Y., Huang, J. C., & Farh, J. L. 2009. Employee learning
Beugelsdijk, S., Ambos, B., & Nell, P. C. 2018. Conceptualizing orientation, transformational leadership, and employee cre-
and measuring distance in international business research: ativity: The mediating role of employee creative self-efficacy.
Recurring questions and best practice guidelines. Journal of Academy of Management Journal, 52(4): 765–778.
International Business Studies, 49(9): 1113–1137. Hammond, M. M., Neff, N. L., Farr, J. L., Schwall, A. R., & Zhao,
Beugelsdijk, S., Maseland, R., & Hoorn, A. 2015. Are scores on X. 2011. Predictors of individual-level innovation at work: A
Hofstede’s dimensions of national culture stable over time? A meta-analysis. Psychology of Aesthetics, Creativity, and the Arts,
cohort analysis. Global Strategy Journal, 5(3): 223–240. 5(1): 90–105.
Borenstein, M., Hedges, L. V., Higgins, J. P., & Rothstein, H. R. Hofstede, G. 2001. Culture’s consequences: Comparing values,
2011. Introduction to meta-analysis. Chichester, West Sussex: behaviors, institutions and organizations across nations. Thou-
Wiley. sand Oaks, CA: Sage.
De Vries, R. E., Roe, R. A., & Taillieu, T. C. 2002. Need for Hofstede, G., Hofstede, G. J., & Minkov, M. 2010. Cultures and
leadership as a moderator of the relationships between organizations: Software of the mind (3rd ed.). New York:
leadership and individual outcomes. The Leadership Quarterly, McGraw-Hill.
13(2): 121–137. Hunter, J. E., & Schmidt, F. L. 2004. Methods of meta-analysis:
Den Hartog, D. N., et al. 1999. Culture specific and cross- Correcting error and bias in research findings (2nd ed.).
culturally generalizable implicit leadership theories: Are Thousdand Oaks, CA: Sage.
attributes of charismatic/transformational leadership univer- Kelloway, E. K., Barling, J., & Helleur, J. 2000. Enhancing
sally endorsed? The Leadership Quarterly, 10(2): 219–256. transformational leadership: The roles of training and feed-
Dickson, M. W., Den Hartog, D. N., & Mitchelson, J. K. 2003. back. Leadership & Organization Development Journal, 21(3):
Research on leadership in a cross-cultural context: Making 145–149.
progress, and raising new questions. The Leadership Quarterly, Kirkman, B. L., Lowe, K. B., & Gibson, C. B. 2006. A quarter
14(6): 729–768. century of culture’s consequences: A review of empirical
research incorporating Hofstede’s cultural values framework. know, and where we are headed. Journal of International
Journal of International Business Studies, 37(3): 285–320. Business Studies, 49(9): 1167–1189.
Lachman, R., Nedd, A., & Hinings, B. 1994. Analyzing cross- Viechtbauer, W., & Cheung, M. W. L. 2010. Outlier and
national management and organizations: A theoretical frame- influence diagnostics for meta-analysis. Research Synthesis
work. Management Science, 40(1): 40–55. Methods, 1(2): 112–125.
Lynn, R., & Martin, T. 1995. National differences for thirty-seven Wang, G., Oh, I. S., Courtright, S. H., & Colbert, A. E. 2011.
nations in extraversion, neuroticism, psychoticism and eco- Transformational leadership and performance across criteria
nomic, demographic and other correlates. Personality and and levels: A meta-analytic review of 25 years of research.
Individual Differences, 19(3): 403–406. Group and Organization Management, 36(2): 223–270.
Medeiros, K. E., Watts, L. L., & Mumford, M. D. 2017. Thinking Watts, L. L., Mulhearn, T. J., Todd, E. M., & Mumford, M. D.
inside the box: Educating leaders to manage constraints. In C. 2017. Leader idea evaluation and follower creativity: Chal-
Zhou (Ed.): Handbook of research on creative problem-solving lenges, constraints, and capabilities. In M. D. Mumford & S.
skill development in higher education (pp. 25–50). Hershey, PA: Hemlin (Eds.), Handbook of research on leadership and creativity
IGI Global. (pp. 82–99). Cheltenham: Elgar.
Minkov, M. 2017. A revision of Hofstede’s individualism-collec-
tivism dimension: A new national index from a 56-country
study. Cross Cultural & Strategic Management, 24(3): 386–404.
Mumford, M. D., & Gustafson, S. B. 1988. Creativity syndrome:
Integration, application, and innovation. Psychological Bulletin, ABOUT THE AUTHORS
103(1): 27–43. Logan L Watts is an assistant professor in the
Mumford, M. D., Scott, G. M., Gaddis, B., & Strange, J. M. Department of Psychology at Baruch College and
2002. Leading creative people: Orchestrating expertise and
relationships. The Leadership Quarterly, 13(6): 705–750. The Graduate Center, The City University of New
Peterson, M. F., & Barreto, T. S. 2014. The like it or not York. His research interests include leadership,
proposition: Implications of societal characteristics for the ethics, creativity, and stories.
cultural expertise and personal values of organization mem-
bers. Journal of Organizational Behavior, 35(8): 1134–1152.
Peterson, M. F., & Barreto, T. S. 2018. Interpreting societal Logan M Steele is an assistant professor in the
culture value dimensions. Journal of International Business Muma College of Business at the University of
Studies, 49(9): 1190–1207.
Ronen, S., & Shenkar, O. 2013. Mapping world cultures: Cluster South Florida. His research interests include lead-
formation, sources and implications. Journal of International ership, ethical decision making, creativity, and
Business Studies, 44(9): 867–897. innovation.
Sarros, J. C., Cooper, B. K., & Santora, J. C. 2008. Building a
climate for innovation through transformational leadership
and organizational culture. Journal of Leadership & Organiza- Deanne N Den Hartog is full professor of Organi-
tional Studies, 15(2): 145–158. zational Behavior at the University of Amsterdam
Shane, S. 1993. Cultural influences on national rates of innova-
tion. Journal of Business Venturing, 8(1): 59–73. Business School (ABS) in the Netherlands. She is
Shane, S., Venkataraman, S., & MacMillan, I. 1995. Cultural Research Director at ABS and heads its Leadership
differences in innovation championing strategies. Journal of and Management section. Her research interests
Management, 21(5): 931–952.
Strange, J. M., & Mumford, M. D. 2002. The origins of vision: include leadership, culture, innovative and proac-
Charismatic versus ideological leadership. The Leadership tive work behavior, dark triad traits, trust, and
Quarterly, 13(4): 343–377. HRM.
Tung, R. L., & Stahl, G. K. 2018. The tortuous evolution of the
role of culture in IB research: What we know, what we don’t
Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional
affiliations.
Supplementary information accompanies this article on the Journal of International Business Studies website (www.palgrave.com/journals).
Accepted by Mark F. Peterson, Area Editor, 2 April 2019. This article has been with the authors for four revisions.