Professional Documents
Culture Documents
Executive Officers
Connections to Top
Management Team
Integration and Middle
Managers Responses
Abstract
In this article, we examine the concept of humility among chief executive officers (CEOs) and the process through which it is connected to integration in the
top management team (TMT) and middle managers responses. We develop
and validate a comprehensive measure of humility using multiple samples and
then test a multilevel model of how CEOs humility links to the processes of
top and middle managers. Our methodology involves survey data gathered
twice from 328 TMT members and 645 middle managers in 63 private companies in China. We find CEO humility to be positively associated with empowering leadership behaviors, which in turn correlates with TMT integration. TMT
integration then positively relates to middle managers perception of having an
empowering organizational climate, which is then associated with their work
engagement, affective commitment, and job performance. Findings confirm
our hypotheses based on social information processing theory: humble CEOs
connect to top and middle managers through collective perceptions of empowerment at both levels. Qualitative data from interviews with 51 CEOs provide
additional insight into the meaning of humility among CEOs and differences
between those with high and low humility.
Keywords: CEO humility, top management team integration, empowerment,
middle managers
Ou et al.
35
Upper echelon theorists have proposed and observed that CEOs characteristics significantly and broadly affect organizations (Finkelstein, Hambrick, and
Cannella, 2009). For example, self-aggrandizing CEOs tend to make riskier
investments (Malmendier and Tate, 2005), pay higher premiums for acquisitions (Hayward and Hambrick, 1997), pursue more dynamic and grandiose strategies, and create fluctuating organizational performance (Chatterjee and
Hambrick, 2007). These effects are often channeled through CEOs unconstrained exercise of power (House and Aditya, 1997), suggesting that a CEOs
high self-regard can be potentially detrimental to organizations. Alternatively,
scholars have suggested that leaders should drop airs of omniscience and
authority, avoid communicating in monologues (Drucker, 1992; Weick, 2001),
and learn along with others, or make up for their missing skills by relying on
others (Ancona et al., 2007: 92). These characteristics are often recognized as
typical manifestations of humility.
Humility has become a fast-growing research topic in psychology and management despite being mistakenly viewed as low self-esteem in the 1980s
(Weiss and Knight, 1980). Tangney (2002) systematically reviewed the construct in theological, philosophical, and psychological literatures, summarized
positive elements embedded in humility, and differentiated it from other related
constructs such as modesty and narcissism. Psychologists have developed various measures and tested the positive implications of humility (Exline and
Geyer, 2004; Peterson and Seligman, 2004; Rowatt et al., 2006; Davis et al.,
2011). Empirical studies on humility have surged since Lee and Ashton (2004)
expanded the classic Big-Five Personality Inventory to the Big-Six HEXACO
Inventory, which included a dimension labeled honesty-humility. Management
research on humility has shown similar momentum. Collins (2001) conducted
case studies on the humility of CEOs who managed high-performing organizations, and Vera and Rodriquez-Lopez (2004) and Morris, Brotheridge, and
Urbanski (2005) laid solid theoretical foundations for humility in management
and leadership. Humility has been conceptually associated with virtuous, moral,
ethical, participative, empowering, and servant leadership (Guillen and
Gonzalez, 2001; Morris, Brotheridge, and Urbanski, 2005; Hackett and Wang,
2012). Most recently, Owens and colleagues (Owens and Hekman, 2012;
Owens, Johnson, and Mitchell, 2013) developed a behavioral measure of humility and conducted rigorous empirical studies on humble leaders.
Despite theoretical discussions (Vera and Rodriguez-Lopez, 2004; Morris,
Brotheridge, and Urbanski, 2005), case studies (Collins, 2001; Hayes and
Comer, 2010), and qualitative interviews (Owens and Hekman, 2012) on humble CEOs, current research has still fallen short in providing empirical information on the nature and effects of humble CEOs in two ways. First, working
definitions and, correspondingly, measures of humility have not fully captured
coherent dimensions of humility. Various definitions include a wide range of
dimensions (from three to thirteen) (Vera and Rodriguez-Lopez, 2004; Owens,
Johnson, and Mitchell, 2013), without clear theoretical justifications for which
dimensions belong to its core content domain versus which ones are simply
correlates of humility. For example, some definitions include dimensions such
as frugality (Vera and Rodriguez-Lopez, 2004), sincerity, or fairness (Lee and
Ashton, 2004) that seem to lack face validity (Davis, Worthington, and Hook,
2010). Relatedly, some leadership researchers regard humility as a trait (e.g.,
Morris, Brotheridge, and Urbanski, 2005), while others treat it as a component
Downloaded from asq.sagepub.com by guest on January 20, 2015
36
Ou et al.
37
38
humble people accept their imperfections (Templeton, 1997). Such selfacceptance, however, does not imply self-abasement or self-degradation. To
the contrary, humble people are fully aware of their talents and abilities.
Knowing their limitations helps them put their strengths in perspective
(Tangney, 2002), allowing them to avoid arrogance or self-contempt. Selfacceptance allows them to willingly disclose themselves, admit their limitations
or mistakes, and actively seek feedback. Thus, through reflective consciousness, humility is associated with open-mindedness and willingness to learn
from others (Templeton, 1997).
Second, in their experience of interpersonal being, humble people appreciate
others as being like themselves, with strengths and weaknesses. As Newman
(1982: 283) pointed out, humility requires a severe appraisal of oneself combined with a reasonably generous appraisal of others. Humble people appreciate others positive worth, strengths, and contributions (Morris, Brotheridge,
and Urbanski, 2005). Such appreciation is grounded in the understanding of
their own strengths and thus generates no need for entitlement or dominance
over others (Peterson and Seligman, 2004).
Third, in their experience of executive functions, humble people are less
self-focused and more engaged in self-transcendent pursuits. Humility is
regarded as a virtue leading to growth (Owens and Hekman, 2012) and selfrealization (Newman, 1982). Being aware of something greater than the self,
humble people understand that they fall short of a standard that they are striving to reach. Their life pursuits are less about themselves than about the larger
community, the greater whole, moral principles, or ultimate universal truth
(Grenberg, 2005; Morris, Brotheridge, and Urbanski, 2005), causing them to forget the self (Tangney, 2002). Self-transcendence protects them from excessive
ego and pursuits of materialism or excessive luxury (Peterson and Seligman,
2004).
Overall, humility is grounded in a self-view of accepting that something is
greater than the self and manifests in self-awareness, openness to feedback,
appreciation of others, low self-focus, and self-transcendent pursuit. Those six
dimensions form a coherent domain of humility with the transcendent self-view
as the cognitive core and low self-focus and self-transcendent pursuit as deeplevel motivations. Both cognition and motivation drive behavioral manifestations
of self-awareness, openness to feedback, and appreciation of others. Although
individuals may exhibit humble behaviors without incorporating a transcendent
self-view, such behaviors may be inauthentic or inconsistent over time.
Defining humility from a self-experience framework also distinguishes it
from other related constructs such as narcissism and modesty. Narcissism
may appear to be an antonym of humility because it predominantly features a
strong desire for self-focus, attention, or continuous self-affirmation (Chatterjee
and Hambrick, 2007) and may relate negatively to humilitys low self-focus. But
humility encompasses other facets such as openness to feedback, appreciation
of others, self-transcendent pursuits, and a transcendent self-concept, all of
which do not conceptually overlap with narcissisms definition or behavioral
manifestations. Modesty is a moderate, non-boastful self-presentation (Tice
et al., 1995). Although humility should positively relate to modesty, the latter is
more about self-presentation and less about fundamental self-views (Peterson
and Seligman, 2004).
Ou et al.
39
40
procedures, and routines allow information sharing, autonomy, and teambased self-management (Blanchard, Carlos, and Randolph, 1995)the contextual conditions that enable empowerment in the workplace (Spreitzer, 2008).
Empowering organizational climate relates to, but is distinct from, psychological
empowerment, which involves individuals perceptions of meaning, competence, self-determination, and impact in their work roles (Spreitzer, 1995),
their psychological experience about the nature of the work itself.
An empowering climate fulfills needs for competence and autonomy and
stimulates intrinsic motivation to perform well (Spreitzer, 2008). Middle managers shared perceptions of an organizational empowering climate shape their
attitudes and behaviors (Salancik and Pfeffer, 1978). First, they may respond
with work engagement, a work-related state of mind characterized by vigor,
dedication, and absorption (Schaufeli, Bakker, and Salanova, 2006). A second
response is affective commitment, or emotional attachment to the organization
(Meyer, Allen, and Smith, 1993). Third, they may respond with enhanced job
performance and task completion (Tsui et al., 1997). Together, work engagement elevates middle managers psychological well-being; affective commitment enables organizations to keep loyal talent; and their enhanced job
performance contributes to strategy implementation. These three responses
reflect a comprehensive assessment of attitudes and behaviors that potentially
contribute to individual and organizational effectiveness (Harter, Schmidt, and
Hayes, 2002).
CEOs humility and empowering leadership. Empowering leadership
constitutes behaviors that enhance the meaningfulness of work, foster participation in decision making, express confidence in high performance, and
provide autonomy (Ahearne, Mathieu, and Rapp, 2005). Humble CEOs, recognizing their own limitations, are likely to use subordinates strengths to compensate and are eager to recruit capable employees who might even question
their own ideas or beliefs (Drucker, 1992). CEOs who communicate selftranscendent pursuits help subordinates understand the value of their work in
relation to larger organizational purposes. By acknowledging their limitations
and appreciating others strengths (Morris, Brotheridge, and Urbanski, 2005),
sharing power with others becomes the natural, behavioral self-expression of
humble CEOs. They admit their reliance on others (Weick, 2001), express a
propensity to treat subordinates as equals (Whitener et al., 1998), provide
autonomy, and include subordinates in decision making. These tendencies further demonstrate their confidence in subordinates. Although not empirically
tested previously, theoretical discussions about humility have also suggested
an association between leaders humility and empowerment (Morris,
Brotheridge, and Urbanski, 2005; Owens and Hekman, 2012). Hence, we
hypothesize:
Hypothesis 1: CEOs humility will relate positively to their empowering leadership
behaviors.
Ou et al.
41
42
organizational policies and practices (Pless, Maak, and Waldman, 2012). Third,
information from different middle managers is compared and negotiated. TMT
integration allows middle managers to receive consistent information and form
convergent perceptions that the organization values information sharing, autonomy, and self-management, all of which constitute behavioral norms consistent
with empowering organizational climates (Blanchard, Carlos, and Randolph,
1995). We thus propose:
Hypothesis 3: Top management team integration will relate positively to middle managers perceptions of an empowering organizational climate.
Figure 1 illustrates our multilevel model of CEO humility. To test the hypotheses, we conducted two quantitative studies supplemented by interviews with
CEOs. Study 1 developed items for three new dimensions of humilitylow
self-focus, self-transcendent pursuit, and transcendent self-conceptand validated them together with the other three dimensions of self-awareness, openness to feedback, and appreciation of others, which are similar to Owens
(2009) three behavioral dimensions: willingness to view oneself accurately,
teachability, and appreciation of others strengths and contributions. Study 2
used the validated comprehensive six-dimensional measure of humility to test
our proposed model with multilevel and multiphase data obtained from a sample of 63 organizations in China. Qualitative data from personal interviews with
Ou et al.
43
Figure 1. Hypothesized model of mechanisms linking CEO humility and middle managers
responses.
COMPANY LEVEL
H1 (+)
CEO Humility
Self-awareness
Openness to feedback
Appreciation of others
Low self-focus
Self-transcendent
pursuit
Transcendent selfconcept
H2 (+)
CEO Empowering
Leadership
Meaningfulness
Participation
Confidence
Autonomy
TMT integration
Collaborative
behavior
Joint decision
making
Information
sharing
Shared vision
INDIVIDUAL LEVEL
H3 (+)
Empowering
organizational
Climate
Information
sharing
Autonomy
Team
responsibility
H4 (+)
Middle Managers
Responses
Work engagement
Affective commitment
Job performance
44
sorted items into one of the three new dimensions of humility or a fourth
does not fit category. We discarded items when fewer than 75 percent of
the judges correctly assigned them to their a-priori dimensions (Hinkin, 1998).
Seventeen items remained after the procedure.
Phase 2: Scale Refinement through Factor Analysis
We adopted two steps with two samples to assess the reliability and factor
structure for scale refinement (Gerbing and Hamilton, 1996). In Step 1, we
used exploratory factor analysis (EFA) to refine the three new dimensions. In
Step 2, we included both the new dimensions and Owens (2009) dimensions
for confirmatory factor analysis (CFA) for further refinement of a sixdimensional scale.
Step 1: Exploratory factor analysis. The sample consisted of 276 full-time
undergraduate students in China. Professors invited students to voluntarily participate in a survey administered after class. The response rate was 97.7 percent. On average, students were 20.7 years old (S.D. = 1.03), and 60 percent
were women. The items were translated into Chinese, and respondents indicated on a 6-point scale, from 1 = strongly disagree to 6 = strongly agree,
the extent to which they agreed that each of the 17 humility items described
them. We used a 6-point scale to increase variance in responses because East
Asians are more likely to choose the midpoint in odd-numbered scales (Si and
Cullen, 1998). We used principal-axis factor analyses with oblimin rotation to
impose a three-factor solution and deleted items if their factor loadings were
less than .40 or had high cross-loadings on other factors (Hinkin, 1998).
Results supported the proposed three-factor structure. We retained the 14
items that met the retention criteria. Cronbachs alphas were .81, .75, and .77,
respectively, thereby meeting standards for applied research (Nunnally, 1978).
These results provided preliminary evidence of the structure and internal consistency of the three new dimensions of humility.
Step 2: Confirmatory factor analysis. We submitted the items for all six
humility dimensions (14 new items and 8 from Owens 2009 measure) to a CFA
to further refine the scale and assess its psychometric properties. We collected
data from 286 M.B.A. students in China. Each student invited another person to
evaluate his or her respective humility because observers can assess more accurately personal characteristics that are either highly observable (self-awareness,
openness to feedback, and appreciation of others) or highly evaluative and subject
to self-serving biases (i.e., low self-focus, self-transcendent pursuit, and transcendent self-concept) (Vazire, 2010). On average, the students were 29.1 years old
(S.D. = 5.50) and had a work tenure of 7.49 years (S.D. = 5.30); 66 percent were
women. Those who evaluated the students humility had known the students for
an average of 8.56 years (S.D. = 8.36). The survey was presented in Chinese and
used the same 6-point scale. We used three criteria to assess overall model fit:
RMSEA lower than .06, and TLI and CFI higher than .90 (Hu and Bentler, 1999).
We examined whether each item significantly loaded on its a-priori factor and
deleted items with modification indices higher than 10 (Bollen, 1989).
Results showed four items with high modification indices that caused model
misfit, and we deleted them. The remaining 18 items demonstrated
Downloaded from asq.sagepub.com by guest on January 20, 2015
Ou et al.
45
46
Ou et al.
47
Table 1. Confirmatory Factor Analysis for Humility Dimensions Discriminant Validity with
Related Constructs*
Model
Humility dimensions with core self-evaluation
Model 0 (baseline model): 7 factors
Model 1 combining self-awareness and core self-evaluation
Model 2 combining appreciation of others and core self-evaluation
Model 3 combining openness to feedback and core self-evaluation
Model 4 combining low self-focus and core self-evaluation
Model 5 combining self-transcendent pursuit and core self-evaluation
Model 6 combining transcendent self-concept and core self-evaluation
Humility dimensions with learning goal orientation
Model 0 (baseline model): 7 factors
Model 1 combining self-awareness and learning goal orientation
Model 2 combining appreciation of others and learning goal orientation
Model 3 combining openness to feedback and learning goal orientation
Model 4 combining low self-focus and learning goal orientation
Model 5 combining self-transcendent pursuit and learning goal orientation
Model 6 combining transcendent self-concept and learning goal orientation
Humility dimensions with modesty
Model 0 (baseline model): 7 factors
Model 1 combining self-awareness and modesty
Model 2 combining appreciation of others and modesty
Model 3 combining openness to feedback and modesty
Model 4 combining low self-focus and modesty
Model 5 combining self-transcendent pursuit and modesty
Model 6 combining transcendent self-concept and modesty
Humility dimensions with narcissism
Model 0 (baseline model): 7 factors
Model 1 combining self-awareness and narcissism
Model 2 combining appreciation of others and narcissism
Model 3 combining openness to feedback and narcissism
Model 4 combining low self-focus and narcissism
Model 5 combining self-transcendent pursuit and narcissism
Model 6 combining transcendent self-concept and narcissism
w2
w2
250.25
517.26
593.94
435.97
499.83
411.24
724.00
267.01
343.69
185.72
249.58
160.99
473.75
228.06
472.82
493.93
380.50
487.68
405.69
557.85
244.76
265.87
152.44
259.62
177.63
329.79
250.42
546.92
651.96
477.20
414.12
465.19
698.96
296.50
401.54
226.78
163.70
214.77
448.54
240.78
530.32
636.58
470.10
460.24
468.40
858.68
289.54
395.80
229.32
219.46
227.62
617.90
* 2 = change in chi-square between the alternative model and the baseline model. All 2 are significant at the
p < .01 level.
w2
d.f.
RMSEA
CFI
TLI
w2
160.36
109
.04
.98
.98
678.51
491.20
113
113
.13
.11
.78
.85
.73
.82
518.15
330.84
832.54
689.28
113
113
.15
.13
.72
.78
.66
.73
672.18
528.92
* RMSEA = root-mean-square error of approximation; CFI = comparative fit index; TLI = Tucker-Lewis Index; and
2 = change in chi-square between the alternative model and the baseline model. All 2 and 2 are significant at
the p < .01 level.
48
managers averaged 35.2 (S.D. = 7.90). Most respondents were male (89 percent of CEOs, 70 percent of TMT members, and 61 percent of middle managers) and had college or above education (97 percent of CEOs, 86 percent of
TMT members, and 83 percent of middle managers). On average, the CEOs
had been in their positions for 7.91 years (S.D. = 4.22), and the TMT members
had been working with the CEOs for 6.11 years (S.D. = 4.96). Most CEOs were
founders (71 percent), 18 percent were internally promoted, and 11 percent
were external hires.
We collected data from multiple sources and at two time periods to minimize common method variance (Podsakoff et al., 2003). We obtained a list of
TMT members from the CEO and a list of middle managers from the companys human resource manager. We visited each company, interviewed CEOs
to understand their management experience and philosophy, and administered
the surveys to other participants in a conference room. We explained to the
participants the importance of truthful answers for scientific research and
ensured confidentiality by guaranteeing that only the researchers would see
individual answers and that the company would receive only a company-level
summary.3 TMT members completed a survey containing items for CEOs
humility, leadership style, TMT integration, and some control variables, while
the middle manager survey contained items for empowering organizational climate and items on their demographics. Two weeks later, the human resource
manager administrated the second surveys, in which TMT members evaluated
middle managers performance and provided a subjective measure of organizational performance (which we used as a control variable), while middle managers reported on work engagement, affective commitment, and job satisfaction.
Participants submitted surveys in self-sealed envelopes to ensure confidentiality. The human resource managers provided the data on objective measures of
organizational level controls and CEOs demographics.
Measures
The study involved four constructs at the organization levelCEO humility,
empowering leadership, TMT integration, and empowering organizational
climateand three measures at the individual levelmiddle managers work
engagement, affective commitment, and job performance. All measures,
except job performance, use 6-point scales ranging from 1 = strongly disagree
or almost never to 6 = strongly agree or always. Because we tested relations among latent variables, we created indicators from dimensional scores or
item parcels using the item-to-construct-balance method to reduce the number
of parameters to be estimated (Williams, Vandenberg, and Edwards, 2009).
The four organization-level measures were aggregated from individual-level
data. Aggregation was confirmed by using ANOVA, rwg(j), ICC(1), and ICC(2).
The aggregation of data is appropriate when the F statistic for ANOVA is significant, rwg(j) is higher than .70 (James, Demaree, and Wolf, 1984), and ICC(1) is
3
To express our appreciation to the company and managers for supporting the research, we provided a company report after the study and gave each manager a small gift for participation in data
collection. We also invited as many as three managers from each company to participate in a oneday seminar held within six months after the study was completed. In this seminar, we shared the
research findings, and two guest lecturers spoke on popular management topics. This gesture
partly facilitated the good response rate.
Ou et al.
49
non-zero (Bliese, 2000). Ideally, ICC(2) values should be higher than .70
(Kozlowski and Klein, 2000), but a low ICC(2) value does not prevent aggregation, although it may result in lower power in detecting relationships involving
Level 2 variables (Bliese, 2000).
We measured CEO humility using 18 items from the newly validated scale
and one additional item from Owens update on the openness to feedback
dimension (Owens, Johnson, and Mitchell, 2013) to ensure that we used the
most current measure of the behavioral dimensions. The scale had good psychometric properties (a for each dimension ranges from .78 to .81), and aggregation was supported: ANOVA F = 1.89, p < .01; median rwg(j) = .98; ICC(1) =
.14; ICC(2) = .47. The full measure of 19 items is shown in Online Appendix B.
We measured CEO empowering leadership behaviors with the Chinese version (Zhang and Bartol, 2010) of the 12-item scale by Ahearn, Mathieu, and
Rapp (2005). We added four items specific to executive-level leaders. This
scale has four dimensions: enhancing the meaningfulness of work, fostering
participation in decision making, expressing confidence in high performance,
and providing autonomy from bureaucratic constraints. Online Appendix B lists
the 16 items. The measure had good psychometric properties (a for each
dimension ranges from .81 to .88; ANOVA F = 1.50, p < .05; median rwg(j) =
.97; ICC[1] = .08; ICC[2] = .33).
We used a split sampling approach to avoid common method variance
because TMT members assessed both CEO humility and CEO empowering
leadership. We randomly selected half of a top management team to measure
CEO humility and the other half to assess CEO empowering leadership behaviors. This approach has been shown to remove the problem of common
method bias (Ostroff, Kinicki, and Clark, 2002).
We measured TMT integration with nine items adapted from Simsek et al.
(2005) and four items from Pearce and Ensley (2004) on shared vision (see
Online Appendix B). The measure had good psychometric properties (a for
each dimension ranges from .84 to .91; ANOVA F = 1.43, p < .05; median
rwg(j) = .96; ICC[1] = .07; ICC[2] = .30).
We measured empowering organizational climate with the Chinese version
(Chen, Lam, and Zhong, 2007) of Blanchard, Carlos, and Randolphs (1995)
30-item scale. It includes three dimensions: information sharing, autonomy
through boundaries, and team responsibility and accountability. A sample item
is We receive the information needed to help us understand the performance
of our organization. The measure displayed acceptable psychometric properties (a for each dimension ranges from .88 to .92; ANOVA F = 2.43, p < .01;
median rwg(j) = .97; ICC[1] = .12; ICC[2] = .59).
Managerial responses. We measured work engagement (a = .93) with the
Chinese version (Zhang and Gan, 2005) of the nine-item Utrecht work engagement scale (Schaufeli, Bakker, and Salanova, 2006). Affective commitment
(a = .89) was measured with a six-item Chinese version (Chen and Francesco,
2003) of the affective commitment scale (Meyer, Allen, and Smith, 1993). We
measured managerial performance (a = .90) with six items, adapted from Tsui
et al. (1997), measuring quality, efficiency, professional standards, ability, judgment, and job knowledge using a 5-point Likert-type scale, with 1 = below
average and 5 = above average. We created a latent variable of managerial
responses with work engagement, affective commitment, and job performance
as indicators. We used this latent variable throughout the hypothesis tests and
Downloaded from asq.sagepub.com by guest on January 20, 2015
50
performed robustness checks using each outcome as a separate latent dependent variable.
Control Variables
We controlled for firm performance, CEO education, TMT size, and average
team tenure to rule out alternative explanations of TMT integration. Firm performance affects TMT members perceptions of their internal dynamics (Simsek
et al., 2005). The measure was a two-item scale asking respondents to evaluate their companys sales and return on assets (ROA) relative to their major
competitors (Sully de Luque et al., 2008). TMT members assessed it at Time 2,
and we aggregated it to the company level after assessing within-group agreement and between-group variance (a = .87; median rwg(j) = .76; ICC[1] = .45;
ICC[2] = .81). We used a subjective measure because the socially constructed
reality, rather than the actual figures, may be more likely to affect TMT members perceptions and behaviors (Berger and Luckmann, 1966). Moreover, 12
companies refused to disclose objective performance data. The correlation
between subjective performance and available, objective same-year ROA was
.35 (p < .05), confirming the validity of the subjective measure. We omitted
other company and CEO demographics such as company age, size, industry,
and CEO age, gender, work tenure, organization tenure, tenure as a CEO, and
founder status because those variables did not significantly correlate with the
key variables in the research model. We controlled for TMT size and average
team tenure because larger teams tend to have more conflict (Amason and
Sapienza, 1997), and longer average team tenure is likely to generate better
coordination and communication (Smith et al., 1994). We also controlled for
CEOs narcissism to test whether humility provides predictive power beyond
narcissism. CEOs assessed their own narcissism using 14 items (a = .86) from
NPI-16 (Ames, Rose, and Anderson, 2006). We further controlled for CEO
transformational leadership (completed by TMT members) to test whether
empowering leadership provides predictive power beyond transformational
leadership. Transformational leadership was measured by 20 items (a = .93;
median rwg(j) = .98; ICC[1] = .19; ICC[2] = .56) from the Multifactor Leadership
Questionnaire 5X (Bass and Avolio, 1995).
For managerial responses, we included both organization-level and
individual-level controls. For the former, we controlled for company age, size,
industry sector, and subjective firm performance. We measured company size
as the natural log of the number of employees. We coded industry as a dummy
variable with 1 = manufacturing and 0 = service or trading. Such dichotomization of industry helps reduce industry-based indicators (Gomez-Mejia, LarrazaKintana, and Makri, 2003). We controlled for middle managers demographics,
including gender, education level, work tenure, company tenure, and job satisfaction. We measured job satisfaction with three items (a = .89) from Bono
and Judge (2003).
Analyses
We tested the hypotheses using multilevel structural equation modeling
(MSEM) with Mplus 6.12 (Muthen and Muthen, 2007). This method allows
simultaneous investigation of multiple paths at different levels and accounts for
Downloaded from asq.sagepub.com by guest on January 20, 2015
Ou et al.
51
52
Mean S.D.
1. MM work
4.75 .70 (.94)
engagement
2. MM affective
4.66 .85 .66 (.91)
commitment
3. MM job
3.66 .68 .12 .12 (.90)
performance
.08
.05
4. CEO humility
4.47 .50 .09
5. CEO empowering
4.53 .46 .01
.04
.15
leadership
6. Top management
4.27 .43 .09
.11 .18
team integration
7. Empowering
4.03 .36 .16 .20 .18
organizational climate
.09
.04
8. Company size
5.57 1.42 .08
(Ln of no. of employees)
.09
.02
9. Company age
14.03 9.22 .13
10. Industry
.41 .50 .04
.04
.04
11. Company subjective
4.26 1.41 .02
.02
.05
performance
.06 .02
12. MM gender
0.61 .49 .10
13. MM work tenure
12.96 8.58 .18 .17 .01
14. MM company tenure
6.13 6.02 .08
.14 .05
15. MM education
3.33 .92 .12 .19 .05
16. MM job satisfaction
4.36 .89 .71 .64 .08
17. CEO narcissism
3.56 .68 .03
.00
.09
18. CEO transformational
4.49 .59 .19 .18 .21
leadership
19. CEO gender
.89 .32 .06
.04
.04
20. CEO company tenure
9.71 5.12 .01
.07
.08
21. CEO tenure as CEO
7.91 4.22 .08
.01
.01
22. CEO education
4.32 1.03 .10
.08
.11
23. CEO founder status
.71 .46 .15 .08 .04
24. TMT size
5.49 2.98 .04
.08
.05
25. TMT average
4.30 1.98 .03
.06
.06
team tenure
Variable
13. MM work tenure
14. MM company tenure
15. MM education
16. MM job satisfaction
17. CEO narcissism
18. CEO transformational
leadership
19. CEO gender
20. CEO company tenure
21. CEO tenure as CEO
22. CEO education
23. CEO founder status
24. TMT size
25. TMT average
team tenure
10
11
(.90)
.31
(.88)
.41
.57 (.95)
.31
.36
.23
.02
.06
.07
.14
.02
.05
.04
.14
.11
.17
.16
.26
.03
.01
.15
.41
.09 .18
(.87)
.05
.07
.06
.07
.06
.24
.26
.02
.23
.07
.04
.01
.06
.31
.08
.01
.04
.28
.09
.01
.24
.47 (.96)
.06
.05
.02
.25
.07
.09
.25
12
13
14
15
16
17
18
.15
.06
.01
.11
.10
.05
.51
.32
.18
.07
.02
.36
.06
.04
.01
.11
.06
.04
(.89)
.02
.13
(.86)
.14
(.97)
.13
.03
.13
.02
.16
.04
.05
.10
.07
.06
.07
.24
.06
.16
.05
.21
.02
.04
.29
.12
.41
.06
.16
.13
.22
.10
.17
.13
.10
.06
.03
.12
.09
.02
.02
.01
.08
.02
.01
.08
.10
.03
.07
.05
.17
.21
.11
.09
.05
.35
.37
.11
.11
.05
.06
.16
.06
.07
.07
.17
.13
.20
.23
.21
.10
.01 .01 .01
.16 .05 .13
.27 .54 .18
.47 .18
.30
.29
.47 .08
.32
.11
.12
.17
.04
.17
.10
19
20
21
22
23
24
.02
.06
.04
.22
.14
.01
.66
.11
.09
.25
.28
.14
.45
.33
.18
.08
.01
.01
.08
.15
.15
* All variables were presented at their appropriate levels. Therefore, for correlations of middle-manager-level
variables, N = 645; for correlations of organization-level variables, N = 63; for cross-level correlations, organizationlevel data were disaggregated, and N = 645. MM = middle manager; TMT = top management team; and CEO =
Downloaded from asq.sagepub.com by guest on January 20, 2015
chief executive officer.
Ou et al.
53
Model
1. 5-factor model: distinct factors for CEO humility, CEO
empowering leadership, TMT integration, empowering
organizational climate, and middle manager responses
2. One-factor model
3. 4-factor model: CEO humility and empowering leadership
combined
4. 4-factor model: CEO humility and TMT integration combined
5. 4-factor model: CEO empowering leadership and TMT integration
combined
6. 4-factor model: empowering organizational climate and TMT
integration combined
w2
70.79
56
.02
.98 .97
237.30
118.96
65
60
.06
.04
10
4
112.83
123.98
60
60
.04
.04
.92 .89
.90 .87
34.57
25.50
4
4
162.41
60
.05
.84 .78
43.92
d.f.
* 2 = Satorra-Bentler scaled 2; RMSEA = root-mean-square error of approximation; CFI = comparative fit index;
TLI = Tucker-Lewis index; and 2 = Satorra-Bentler scaled adjusted 2 difference.
CEO humility on empowering organizational climate (.16, p < .10) and on middle managers responses (.11, p < .10) were only marginally significant, rendering weaker support for hypotheses 3 and 4. As this method has a low
power to detect mediation (MacKinnon et al., 2002), and all mediation paths
were significant, we concluded that hypotheses 3 and 4 were still supported.
The results confirm that CEO humility is associated with middle managers
responses through empowering leadership behaviors, TMT integration, and
middle managers perceptions of an organizational empowering climate.
Robustness Checks
We conducted five robustness tests using a series of structural models to rule
out alternative explanations of the findings. The results are shown in Online
Appendix C. First, we reran the research model using each middle manager
outcome as a dependent variable. Models 2, 3, and 4 illustrate that all three
models fit the data well, and the path coefficients demonstrated the same pattern found for our baseline structural model (model 1).
Next, we included control variables for TMT integration and middle managers responses. We separately included control variables for TMT integration
(model 5) and for middle managers responses (model 6) to accommodate the
modest sample size. Results reveal that model 5 (2 [135] = 232.58, p < .01;
CFI = .87; TLI = .85; RMSEA = .03) fits the data worse than model 1, but path
patterns are similar across the two models (see figure 2b). The estimated parameters were slightly smaller but remained significant. Among the control variables, only CEO transformational leadership related significantly to TMT
integration ( = .33, p < .05). These results provided strong evidence on the
predictive power of CEO humility and empowering leadership, net of the effect
of narcissism and transformational leadership.
Model 6 included both organization-level and individual-level controls for middle manager responses. This model fit the data less well (2 [132] = 192.95,
p < .01; CFI = .95; TLI = .94; RMSEA = .03) than model 1. As shown in figure
Downloaded from asq.sagepub.com by guest on January 20, 2015
H2
H1
.45
H3
.68
.73
TMT
integration
TI2
.96
.48
.95
EC1
CEO
empowering
leadership
.52
TMT
integration
.45
Empowering
organizational
climate
.69
Middle
managers
responses
.79
EC3
.71
CEO
humility
.45
CEO
empowering
leadership
.73
TMT
integration
.50
Empowering
organizational
climate
.37
Middle
managers
responses
CEO
humility
.43
.77
EC2
Empowering
organizational
climate
EL2
.76
EL1
.80
CEO
empowering
leadership
TI1
.90
CEO
humility
.89
.70
.99
OC
Middle
managers
responses
.54
TP
.82
WE
54
Administrative Science Quarterly 59 (2014)
Ou et al.
55
2c, the path patterns are similar to the baseline model except that the linkage
between empowering organizational climate and middle managers responses
drops from .71 to .37 (p < .01). Among the organization-level controls, only
company size was significant ( = .22, p < .05). At the individual level, only
managers education ( = .07, p < .05) and job satisfaction ( = .84, p < .05)
had significant effects on managerial responses; other control variables had no
effect.
Third, we examined potential reverse causality in the research model. We
tested four possible alternative sequences and reported the results in models 7
through 10.We compared the fit of these models with model 1 using Akaikes
(1987) information criterion (AIC), which is appropriate for non-nested model
comparisons; the model with the lowest AIC is the best-fitting model (Akaike,
1987). Online Appendix C shows that the baseline model (model 1) has the
lowest AIC (4262.43), supporting the original mediation sequence.
We further examined whether the proposed variables fully mediated the
effect of CEO humility and whether subjective firm performance reduced the
effects of CEO humility and empowering leadership. Based on model 1, model
11 added a path from empowering leadership to empowering organizational climate ( = .13, p > .10) and a path from TMT integration to middle managers
responses ( = .03, p > .10). Neither path was significant, and model 11 was
no better than model 1 (2[2] = .59, p > .10). Similarly, model 12 added
direct paths from CEO humility ( = .04, p > .10) and empowering leadership
( = .05, p > .10) to middle managers responses. The model was no better
than model 1 (2[2] = .11, p > .10), and neither path was significant.
Following the parsimony principle (Bollen, 1989), the full mediation sequence in
model 1 represented a better model than models 11 and 12. Model 13 tested
whether subjective firm performance affected TMTs assessment of CEO
humility and empowering leadership by adding paths from firm performance to
CEO humility ( = .06, p > .10) and empowering leadership ( = .22, p > .10).
Again, neither path was significant; other path coefficients did not change, and
AIC was higher than that for model 1.
Finally, we tested for common method bias, despite using a split sample
approach to measure CEO humility and CEO empowering leadership, because
the two latent variables partially shared the sample with TMT integration, and
all three variables were measured at Time 1. As Podsakoff et al. (2003) suggested, we followed Williams, Cote, and Buckley (1989) to examine the influence of common method variance among the three variables. We compared
the baseline measurement model of the three constructs with a null model
whereby all items were independent, a model of two method factors, and a
bi-factor model of both measure and method factors. The method model was
better than the null model (2[9] = 195.33, p < .05), and the bi-factor model
did improve model fit from the baseline model (2[7] = 12.7, p < .05), suggesting method bias in our results. That said, the method effects accounted for
only 12 percent of the variance, which was much lower than the 25 percent
benchmark reported by Williams, Cote, and Buckley (1989). In addition, parcels
for TMT integration had no significant loadings on the method factor, demonstrating that the method model was suboptimal. Therefore we concluded that
common method variance was not a major concern for this study.
Using the same method, we tested the impact of potential social desirability
bias, because all participants were gathered in the same room to complete
Downloaded from asq.sagepub.com by guest on January 20, 2015
56
surveys in the presence of other colleagues for the Time 1 survey. We constructed a method factor that combined all four latent variables measured at
Time 1, namely, CEO humility, CEO empowering leadership, TMT integration,
and empowering organizational climate. The method model was worse than
the null model (2[24] = 69.45, p < .05), and the bi-factor model, including
the baseline latent variables and the method factor, was worse than the baseline model (2[7]= 24.42, p < .05). These results suggest that social desirability bias was not a concern.
Qualitative Evidence on CEO Humility
To substantiate our quantitative assessment of CEO humility, we adopted Fu
et al.s (2010) method to identify CEO humility qualitatively by asking CEOs
questions indirectly related to humility. To capture transcendent self-views, we
asked CEOs whether they believed that man can conquer nature. In Chinese,
nature broadly means something that is greater than any individual, whether
the power comes from universal natural laws or social concepts of heaven or
god(s). To delve into self-awareness and openness to feedback, we asked
CEOs to assess their strengths and weaknesses and their views of their
organizations problems and future constraints. We probed their appreciation
of others by asking them to assess the TMT members and describe their
decision-making styles. We examined CEOs self-focus and self-transcendent
pursuit by asking why they wanted to be CEOs. Twelve CEOs declined to be
audio-taped, so we included 51 interviews in the content analysis.
Data coding. The 51 interviews yielded 518 single-spaced pages; on average, each interview engendered 48.8 paragraphs and 9,814.8 words. The first
author trained two coders to ensure that they understood the definitions and
items of humility and its dimensions. The two coders were Masters students
in management and were blind to the hypotheses. Both coders independently
coded all transcripts, compared their codes, and discussed discrepancies. They
could change or retain their original coding after discussion. The first author
met with the coders to clarify ambiguities and answer questions. For each
interview transcript, they coded each paragraph as strong, weak, or neutral
reflections of each of the six humility dimensions. Cohens Kappa for these
two coders was .95, indicating acceptable interrater agreement (Landis and
Koch, 1977; Sun, 2011). The coders also provided an overall rating of the
CEOs humility: 1 = strong, 0 = neutral, and 1 = weak. We averaged this
rating of CEO humility by the two coders and compared it with TMT members
ratings.
Results. The coders ratings and the TMT members ratings of CEO humility
correlated at .28 (p < .05), suggesting a moderate agreement. A closer comparison revealed some interesting findings. Using a mean split, we recoded the
TMT members ratings to 1 = above-average humility and 1 = below-average
humility, similar to the coders overall rating of humility. TMT members rated
27 CEOs with above-average humility, among which 19 were also in the
coders list, suggesting a 70-percent agreement level. In contrast, among the
24 CEOs rated by TMT members as below-average humility, nine were also in
Downloaded from asq.sagepub.com by guest on January 20, 2015
Ou et al.
57
Table 5. Coding Summary of Interviews with CEOs on Their Humility and Management
Philosophy
CEOs with Lower Humility
Interview
Questions
Observations
Sample quotes
Sample quotes
More emphasis
on strengths
and less on
weaknesses.
Weaknesses
identified are
more like
strengths in
disguise.
58
Table 5. (continued)
CEOs with Lower Humility
Interview
Questions
2.2 Views on the
organizations
problems /
constraints on
future
Observations
More likely to
attribute
problems to
external forces.
Less likely to see
constraints on
organizations
future.
Sample quotes
Sample quotes
In the foreseeable future, I will
only work on one business.
The society has a clear
division of labor; if you can do
one thing successfully, it is
already an accomplishment
(CEO 130: 14).
Our company grows rapidly,
and many companies have
asked us to distribute their
brands; it is very tempting, but
we think that we are not yet
ready (CEO 504: 4).
See more
weaknesses.
Less appreciation
of the TMTs
capability.
3.2 Decision
making styles
Tend to withdraw
I went through several
Usually I come up with
Tend to make
from making
stages. Before 2004, I was
some ideas, then I let them
decisions by
decisions alone and
the main one making
themselves and
discuss them. Most of the
decisions and they
time, they execute what Ive to empower TMT
ask the TMT
members to make
implemented; during 2004
decided (CEO 410: 13).
members to
decisions together.
2008, 23 people made all
implement.
What is the use of
the decisions; now it was a
discussion? You make your
group discussion model. The
decision, inform them, and
TMT meets once a month to
then let them implement it. I
figure out future strategy
think many Chinese are not
(CEO 504: 27).
good at discussion; most of
I have to withdraw, let go, force
the time they wait for the
bosss decision, and they
them to take responsibility. . . .
dont bring up their ideas
Starting last May, I forced
during discussions, so
myself not to come to the
sometimes you spend a lot of
office everyday. Gradually, I
time discussing with people
found several of them grew
below you, and it ends up
quickly and became more
wasting time (CEO 124: 37).
mature in making decisions
(CEO 201: 21).
Downloaded from asq.sagepub.com by guest on January 20, 2015
(continued)
Ou et al.
59
Table 5. (continued)
CEOs with Lower Humility
Interview
Questions
Observations
Sample quotes
Sample quotes
the coders list, only a 38-percent agreement. If we assume that TMT members are more able to detect true humility as they more frequently interact with
their CEOs, low agreement for CEOs with low humility suggests that CEOs
with low humility may be more inclined to feign humility in their answers to
interviewers. We then focused on the 28 CEOs with agreement in humility
level between TMT members and coders ratings19 CEOs with relatively
higher humility (average TMT rating = 4.86, average coder rating = 0.76) and
nine with relatively lower humility (average TMT rating = 4.00, average coder
rating = 0.81)in identifying their cognitive, motivational, and behavioral
manifestations of humility.
Table 5 summarizes the representative observations and quotes from the
interviews. The two groups differ in transcendent self-concept. CEOs with
lower humility tended to believe that man can conquer nature and to believe in
personal capability. CEOs with higher humility tended to believe that man proposes and nature disposes (CEO 130) and that mankind exists under certain
constraints. In terms of self-awareness and openness to feedback, the more
humble CEOs acknowledged their strengths but were also open to discussing
their weaknesses. For example, CEO 504, from a very successful apparel company, said, I have a lot of experience in selling jeans, but I need a lot of effort
to figure out . . . a direction for future innovation. Humble CEOs also showed
eagerness to learn. As CEO 413 said, I regard my job as a happy learning
vacation with pay. On appreciation of others, the more humble CEOs tended
to appreciate their TMT members capability and to treat them as complementing the CEO. For example, CEO 503 said, I appreciate my team members
passion and identification with this company. CEO 506 treated TMT members
Downloaded from asq.sagepub.com by guest on January 20, 2015
60
as partners and said that he purposefully chose those who complemented him
in capability and personality. They also tended to withdraw from making decisions solely by themselves. For example, a CEO managing a supermarket chain
said that he forced himself to avoid coming to the office daily because his
absence allowed TMT members to mature more rapidly.
When asked about why they wanted to be CEOs, the more-humble ones
mentioned goals beyond individual gain such as responsibility to society and
employees. CEO 130 said, The drive that led me to entrepreneurship is that I
think people must work diligently for societys sake. CEO 102 said, A persons value does not equal how much he or she earns. Those statements
reflect a low self-focus and self-transcendent pursuit. Such a perspective is in
line with what has been characterized as socialized charisma (Brown and
Trevino, 2006).
DISCUSSION
Our two empirical studies provide strong support for our conceptualization of
humility and our hypotheses on CEO humilitys links to top management teams
and middle managers. Study 1 produced six humility dimensions that constitute
a coherent multidimensional construct with good content, convergent, and discriminant validity and reliabilities. Study 2 confirmed the criterion-related validity
of CEO humility by its positive correlations with empowering leadership, TMT
integration, and empowering organizational climate. Findings from Study 2 support our conceptualization based on social information processing theory that
humble CEOs are connected to TMTs and middle managers responses by
shared perceptions that cascade across hierarchical levels.
Theoretical Contributions
Our study helps clarify the conceptual foundation of CEO humility and provides
additional insight into executive leadership and influence. First, our conceptualization alters the common misunderstanding that humble CEOs may lack confidence or may not be able to motivate others. We theorize that humble CEOs
willingly admit both their strengths and weaknesses, have self-confidence, and
are appreciative of others strengths. We further posit that humble CEOs have
self-transcendent pursuits, and our qualitative findings suggest that they often
have a broad perspective of service to society that others would find inspiring.
As our qualitative analysis illustrates, such perspectives are often socialized for
the collective good rather than for individual glory.
Second, the positive effects of CEO humility provoke new thinking about
traits of effective leaders. Besides confirming the value of studying leaders
traits (DeRue et al., 2011), our study challenges previous reviews that have
consistently found influential leaders to be masculine, dominant, and authoritarian (Lord, De Vader, and Alliger, 1986; Koenig et al., 2011). Those views suggest that leadership requires a strong personal desire for influence. But the
reviews overlook the second piece of the influence puzzle: leadership also
depends on whether subordinates accept leaders influence (Blau, 1964).
Research shows that bold and assertive leaders who lack genuine concern for
others undermine their influence on subordinates (Fu et al., 2010; Galvin,
Waldman, and Balthazard, 2010). Our study is consistent with that research in
Downloaded from asq.sagepub.com by guest on January 20, 2015
Ou et al.
61
finding that CEO humility wins acceptance from top and middle managers by
appealing to collective interests and downplaying strong ego dominance. By
putting collective interests over self-interest, humility becomes a powerful
source of socialized charisma (House and Aditya, 1997). Relatedly, leaders who
lack strong egos are able to reduce status differences and gain trust among
status-seeking groups such as managers.
Third, our results contribute to understanding the multilevel effects of leadership, which the strategic leadership literature has heretofore not adequately
addressed (Cannella and Holcomb, 2005). Derived from social information processing theory, we view leaders as meaning managers (Smircich and Morgan,
1982) or cultural architects (Schein, 2010). Our findings suggest that CEOs can
form collective meanings across hierarchical levels, which in turn affect the attitudes and behaviors of employees at lower levels. Our research model reveals
several novel multilevel processes underlying strategic leadership. For example,
finding CEOs roles in cascading meaning to lower levels through TMT integration adds new understanding to indirect leadership theory (Yammarino, 1994).
Our results also demonstrate humble CEOs strategic value in building competitive advantage through cultivating shared perceptions across hierarchical levels
(Wong, Ormiston, and Tetlock, 2011). Furthermore, our finding of a significant
connection between TMT integration and empowering organizational climate
directly responds to a recent call for studies clarifying how a TMTs coordinated
actions or teamness may influence middle managers (Raes et al., 2011). This
finding also supports Griffin and Mathieus (1997) proposition of the connection
between higher-level group interactions and lower-level employees collective
perceptions.
Limitations and Future Research Directions
Our study has several limitations that future research should address. First, our
findings are based on a moderate sample size, despite our effort to secure
high-quality surveys and a high response rate from respondents at different
hierarchical levels from multiple companies. A moderate sample size rendered
lower power to detect mediation using the product term method, resulting in
weaker support for hypotheses 3 and 4. It also constrained our ability to completely dispute the concern about common method variance. We could have
directly controlled for common method variances by including method factors
in the multilevel structural equation models if the sample size were large
enough to do so. We tested for this bias as best we could, however, and our
results suggested that the extent of method bias was small compared with
past research.
Second, we focus only on TMT integration linking CEO humility to middle
managers responses, but other mediators might be important as well. For
example, the process by which CEOs communicate goals and create vertical
and horizontal alignments of goals across organizational levels has been posited
to mediate CEOs leadership and outcomes (Kinicki et al., 2011). Dean and
Sharfman (1996: 378) also recommended studying the impact of CEOs on the
quality of decision implementation, defined as the competence with which
steps are taken to execute the strategic decision. Carmeli and Schaubroek
(2006) have found TMT behavior integration to improve the quality of strategic
decisions. Our study suggests that humble CEOs could improve decisions
Downloaded from asq.sagepub.com by guest on January 20, 2015
62
Ou et al.
63
shown that traits continue to change throughout adulthood, and scholars propose that with some life changes and role transitions, individuals can become
functionally more mature, as indicated by increases in conscientiousness and
emotional stability (Caspi, Roberts, and Shiner, 2005). Future research can
explore whether certain life experiences or systematic training increase leaders
humility. In addition, human beings are holistic organisms constituted of many
attributes, and the effects of one attribute can be better understood by considering it in connection with other attributes (Allport, 1961). For example, our
study found only a weak correlation between humility and narcissism, implying
a possibility that these two traits may co-exist in some people. Can some leaders possess contradictory attributes simultaneously? Can they be narcissistic
when pursuing status and become humble when they reach the top? These
exciting and meaningful questions are among many to explore about humility.
We hope our study serves as a foundation for future inquiry on the roles of humble CEOs in organizational outcomes, economic and humanistic, for lower-level
employees, customers, shareholders, and the broader society.
Acknowledgments
We thank Don Lange, David Zhu, Peter Hom, Zhen Zhang, Donald Hambrick, Alan
Randolph, Yan Ling, Bradley Owens, and Michael Lubatkin for sharing their insights in
conceptualizing and measuring the key constructs in this paper. We appreciate the help
of Yanling Lian, Liangding Jia, Duanxu Huang, and Jianwu Jiang for data collection. This
work was supported by Singapore Ministry of Education Tier 1 Academic Research
Fund (grant number R-317-000-097-133), National Natural Science Foundation of China
(grant numbers 71032001 and 71072142), Arizona State University GPSA dissertation
grant (2009), and the International Association for Chinese Management Research /
Li-Ning Dissertation Proposal Grant (2010).
REFERENCES
Ahearne, M., J. Mathieu, and A. Rapp
2005 To empower or not to empower your sales force? An empirical examination of
the influence of leadership empowerment behavior on customer satisfaction and performance. Journal of Applied Psychology, 90: 945955.
Akaike, H.
1987 Factor analysis and AIC. Psychometrika, 52: 317332.
Allport, G. W.
1961 Pattern and Growth in Personality. New York: Holt, Rinehart and Winston.
Alon, I., J. Child, S. Li, and J. R. McIntyre
2011 Globalization of Chinese firms: Theoretical universalism or particularism.
Management and Organization Review, 7: 191200.
Amason, A. C., and H. J. Sapienza
1997 The effects of top management team size and interaction norms on cognitive
and affective conflict. Journal of Management, 23: 495516.
Ames, D. R., P. Rose, and C. P. Anderson
2006 The NPI-16 as a short measure of narcissism. Journal of Research in Personality, 40: 440450.
Ancona, D., T. W. Malone, W. J. Orlikowski, and P. M. Senge
2007 In praise of the incomplete leader. Harvard Business Review, 85 (2): 92103.
Anderson, J. C., and D. W. Gerbing
1988 Structural equation modeling in practice: A review and recommended twostep approach. Psychological Bulletin, 103: 411423.
Downloaded from asq.sagepub.com by guest on January 20, 2015
64
Ou et al.
65
66
Ou et al.
67
68
Ou et al.
69
70
Ou et al.
71
72