bs_bs_banner

British Journal of Management, Vol. 26, 126–138 (2015)

DOI: 10.1111/1467-8551.12048

Research Note

Empowering Leadership, Psychological

Empowerment and Employee Outcomes:
Testing a Multi-level Mediating Model
Kai Hung Fong and Ed Snape1
Tianda Group Limited, Level 24, CITIC Tower, No. 1 Tim Mei Avenue, Central, Hong Kong, and
1
Department of Management, Hong Kong Baptist University, Hong Kong Special Administrative Region,
China
Corresponding author email: [email protected]

This study examined the effects of empowering leadership on employees in a customer

service organization, using data from 266 employees and their supervisors from 41 work
teams in a division of a large Hong Kong telecommunications corporation. Empowering
leadership had acceptable levels of within-group agreement and between-group variabil-
ity, providing support for its analysis as a group-level construct. Testing a multi-level
model in which psychological empowerment was hypothesized to mediate the relationship
between both within-group and between-group empowering leadership and individual
outcomes, empowering leadership was associated with psychological empowerment at
both levels. There was evidence of significant mediation effects at both the individual and
group levels. These findings underline the importance of analysing both within-group and
between-group differences in empowering leadership and their effects on individual
attitudes and behaviours.

Scholars have suggested that leadership theory has basis. Studies of empowering leadership have
been grounded in a bureaucratic framework, been conducted mainly at the individual level of
emphasizing top-down control (Uhl-Bien, Marion analysis (e.g. Ahearne, Mathieu and Rapp, 2005;
and McKelvey, 2007). However, organizations are Arnold et al., 2000). There is theoretical support
moving away from hierarchical management, for this in that a leader may empower individual
emphasizing instead the need to empower employ- employees to a greater or lesser degree, depending
ees, requiring leaders concerned less with directing on individual characteristics (e.g. Yukl and Fu,
subordinates than with supporting and empower- 1999; Zhang and Bartol, 2010). However, it is
ing them to perform (Arnold et al., 2000), and increasingly being recognized that leadership may
there is evidence that such empowerment is posi- be analysed at the group level, with followers
tively associated with employees’ attitudinal and subject to the common effects of a shared
behavioural outcomes (e.g. Avolio et al., 2004). leader (e.g. Feinberg, Ostroff and Burke, 2005).
Empowering leadership has been seen as a dis- Studies have thus adopted either an individual- or
tinct type of leader behaviour (Arnold et al., group-level approach to empowering leadership
2000), providing support to subordinates through (Srivastava, Bartol and Locke, 2006; Zhang and
coaching, encouragement, training and emotional Bartol, 2010). However, it is important to address
support, and information (Bandura, 1986). both the between- and within-group processes
Employees are thus more likely to have a sense of linking empowering leadership and outcomes at
meaning in their jobs, and to feel capable of the same time. We address this, analysing effects
implementing job activities on a self-managed at both levels.

© 2013 British Academy of Management. Published by John Wiley & Sons Ltd, 9600 Garsington Road, Oxford OX4
2DQ, UK and 350 Main Street, Malden, MA, 02148, USA.
Leadership and Empowerment 127

UNIT LEVEL

Unit-level
empowering
leadership.

Individual-level Psychological Employee

empowering empowerment. attitudes and
leadership. behaviours.

INDIVIDUAL LEVEL

Figure 1. Hypothesized research model

Our contributions are as follows. First, we provide clearer insights into the nature of the
evaluate whether empowering leadership, repre- mediation process. In doing so, our aim is not to
sented by Arnold et al.’s (2000) Empowering resolve whether empowering leadership is to be
Leadership Questionnaire (ELQ), can be vali- conceptualized as either a group-level or an
dated as a group-level construct. Although the individual-level phenomenon, but to evaluate
ELQ was initially developed and validated at the both. We therefore assess the effects of empower-
individual level, the questionnaire items focus on ing leadership and psychological empowerment at
the behaviour of the leader towards the respond- both levels of analysis.
ent’s work group as a whole. While others have
analysed empowering leadership at the team level
(Chen et al., 2011; Srivastava, Bartol and Locke, Conceptual framework and hypotheses
2006), to the authors’ knowledge, we provide the
first evaluation of the group properties of the There is theory and evidence to suggest that
ELQ using the full scale. If the ELQ can be con- empowering leadership behaviour may be
ceptualized as a group-level construct, this makes regarded as either an individual-level or a group-
it possible to analyse between-group differences in level construct (Srivastava, Bartol and Locke,
the consequences of empowering leadership. 2006; Zhang and Bartol, 2010). Therefore, we
Second, we test a model in which psychological assess both the individual-level and group-level
empowerment mediates the relationships between effects of empowering leadership on employees’
empowering leadership and employee work out- psychological empowerment, and on individual-
comes (see Figure 1). Chen et al. (2007) have level attitudinal and behavioural outcomes,
shown the importance of taking a multi-level implying a multi-level mediating model.
approach to the analysis of leadership and
empowerment, but they analysed leadership
Empowering leadership and
climate and leader–member exchange, rather than
psychological empowerment
empowering leadership. Studies that have exam-
ined the process by which empowering leadership Psychological empowerment is concerned with
influences employee attitudes and behaviour have employees’ perceptions of their power to cope
analysed it at the individual, team or cross (team with events, situations and problems (Thomas
to individual) level (Chen et al., 2011; Srivastava, and Velthouse, 1990), and has been defined as
Bartol and Locke, 2006; Zhang and Bartol, 2010). individuals’ experience of intrinsic motivation,
However, there is a theoretical basis for both based on cognitions about themselves in relation
between- and within-group effects, so it is impor- to their work role (Spreitzer, 1995). Feelings of
tant to estimate effects at both levels simultane- empowerment are likely to be shaped by the
ously, not least to avoid conflating the two organizational context, and in particular by man-
(Zhang, Zyphur and Preacher, 2009), and to agement practices that delegate decision-making

© 2013 British Academy of Management.

128 K. H. Fong and E. Snape

authority (Seibert, Wang and Courtright, 2011), conceptualized at the group level (Srivastava,
and leadership behaviours aimed at empowering Bartol and Locke, 2006), the suggestion being
subordinates (Ahearne, Mathieu and Rapp, 2005; that leader behaviour may be perceived more or
Arnold et al., 2000). Empowering leadership less in common by subordinates, as an ambient
enhances individuals’ experience of intrinsic moti- stimulus. Individual subordinates may assess their
vation relating to their cognitions of competence, leader based not only on how the leader behaves
self-determination, impact and meaning (Bowen towards them personally, but also on the leader’s
and Lawler, 1992). Unlike broader concepts such behaviour towards the group. Such group-
as transformational leadership, empowering lead- level empowering leadership may explain variance
ership is more clearly focused on empowerment, in employee outcomes over and above that
so that the association with empowerment is likely explained by individual-level empowering leader-
to be stronger. ship, since followers of an empowering leader will
Questions remain about the appropriate level of be influenced not only by the dyadic relationship
analysis at which to assess empowering leadership. with the leader, but also by the dynamics of being
Zhang and Bartol (2010) argue that leaders differ- a member of an empowered group. Social influ-
entiate between subordinates in the degree to ence theory suggests that members of a group
which they empower them. Building on the tend to develop similar perceptions and attitudes,
insights of leader–member exchange (LMX) owing to their regular interaction and sharing
theory, which suggests that leaders adjust their of experiences and information (Salancik and
behaviour according to individual follower char- Pfeffer, 1978), and this may lead to the develop-
acteristics (Dansereau, Graen and Haga, 1975; ment of norms and habits concerning behaviours
Nielsen and Daniels, 2012), Wu, Tsui and Kinicki (Hausknecht, Hiller and Vance, 2008). Individu-
(2010) use the term ‘differentiated leadership’ to als may thus be influenced by the way in which
refer to the extent to which a leader ‘exhibits their co-workers respond to an empowering
varying levels of individual-focused leadership leader, giving rise to a group-level effect.
behavior to individual group members’ (p. 90). Consistent with a group-level perspective,
This will directly influence the quality of individu- Arnold et al.’s (2000) ELQ has a group focus, in
als’ exchanges with the leader. Leader–member terms of the definitions of sub-dimensions and in
exchange theory suggests that dyadic exchange the wording of specific items. Items either specifi-
relationships are formed based on personal rela- cally refer to ‘my work group’ as the focus of the
tionships, compatibility and follower characteris- leader’s behaviour (e.g. ‘Encourages work group
tics (Dansereau et al., 1975). Followers are divided members to express ideas/suggestions’) and/or
into in-groups and out-groups (Sherony and describe behaviour that is general rather than
Green, 2002), with the favoured in-group benefit- aimed at one person (e.g. ‘Leads by example’).
ting from the leader’s attention, which helps them This is true of all five sub-dimensions: leading by
create meaning from their work experience and example (e.g. ‘Sets high standards for perfor-
relate this to the overall goals of the group (Nielsen mance by his/her own behaviour’), participative
and Daniels, 2012). In-group members have a decision-making (e.g. ‘Listens to my work group’s
more positive exchange with their leaders, are ideas and suggestions’), coaching (e.g. ‘Helps my
likely to enjoy the leader’s respect and trust, and be work group see areas in which we need more
afforded greater autonomy (Graen and Uhl-Bien, training’), informing (e.g. ‘Explains rules and
1995). This suggests that empowering leadership expectations to my work group’), and showing
behaviours may be seen as part of the leader’s input concern/interacting with the team (e.g. ‘Cares
into a positive exchange, differentiated based on about work-group members’ personal problems’).
in-group membership. Consistent with this, Yukl This suggests that all five dimensions reflect a
and Fu (1999) found that managers delegated group-focus, although while the referent is the
more to subordinates they saw as being competent, work group, it is still possible that followers see
sharing their task objectives, having a favourable things differently based on the leader’s interaction
exchange relationship and longer dyad tenure, and with them individually.
who were themselves supervisors. Chen et al. (2011) analysed empowering leader-
Apart from this differentiated individual-level ship behaviours at the team level, finding empirical
perspective, empowering leadership has also been support for aggregation of followers’ ratings,

© 2013 British Academy of Management.

Leadership and Empowerment 129

using Kirkman and Rosen’s (1999) scale, with the elling group or individual-level empowering lead-
team leader as the referent. Similarly, Srivastava, ership as an antecedent of individual outcomes,
Bartol and Locke (2006) examined the team-level with an individual-level mediator, risks conflating
mechanisms linking empowering leadership and within-group and between-group mediation
team performance, providing empirical support effects. They recommend procedures to differen-
for aggregating subordinates’ ratings for their tiate the between-group and within-group effects.
supervisor and analysing the effect of this group- We explain this below, but the key point is that,
level rating on team performance. Using 15 (out of having identified the potential for effects at both
38) items from the ELQ, they found high within- the group and individual levels, it is essential to
group consistency of subordinate ratings for a estimate both simultaneously.
particular supervisor, as well as a between-group In taking this approach, we are not necessarily
mean square that was significantly greater than the expecting to resolve whether empowering leader-
within-group mean square. In contrast, Zhang and ship and the empowerment of subordinates is to be
Bartol (2010) found that between-group variance conceptualized as either a group-level or an
was not significantly greater than within-group individual-level phenomenon. Instead, we antici-
variance, supporting the analysis of empowering pate that there may be effects at both levels, and we
leadership behaviours as an individual-level phe- aim to provide a clear assessment of their relative
nomenon. However, they used Ahearne, Mathieu importance. In addition to conducting an aggrega-
and Rapp’s (2005) measure of ‘leadership empow- tion analysis on the ELQ to assess the degree of
erment behaviour’, which consists of items that between-group and within-group variance, we test
focus on individual supervisor–subordinate inter- the following hypotheses about the association
action (e.g. ‘My manager helps me understand between empowering leadership and individual
how my objectives and goals relate to that of the employees’ psychological empowerment:
company’). This contrasts with the ELQ, which
focuses on the subordinate’s impression of the H1: Both (a) individual-level and (b) group-
supervisor’s behaviour towards the work group in level empowering leadership behaviour
general (e.g. ‘Encourages work group members to will be positively related to psychological
express ideas/suggestions’). empowerment.
This paper adopts the Arnold et al. (2000)
ELQ, so that we anticipate group-level properties
Empowerment and employee outcomes
and effects for empowering leadership. Unlike
Srivastava, Bartol and Locke (2006), we use all 38 Empowering leadership is a motivating leadership
items so that, to our knowledge, we provide the style that is likely to be associated with employee
first evaluation of the group properties of the full performance (Ahearne, Mathieu and Rapp, 2005;
ELQ scale. However, we are mindful of sugges- Arnold et al., 2000). Analysed at the individual
tions that leaders may treat individual subordi- level, it has been shown to be positively associated
nates differently, leaving open the possibility of with job performance (Ahearne, Mathieu and
within-group differences in employee perceptions Rapp, 2005) and employee creativity (Zhang and
of empowering leadership behaviour. Bartol, 2010), and for team-level empowering
Previous studies have examined the process by leadership there is evidence of a positive associa-
which empowering leadership may influence atti- tion with team performance (Srivastava, Bartol
tudes and behaviour, conducted at the individual, and Locke, 2006), and a cross-level association
team or cross (team to individual) levels (Chen with individual innovative and teamwork behav-
et al., 2011; Srivastava, Bartol and Locke, 2006; iours and turnover intentions, with psychological
Zhang and Bartol, 2010). Since empowering lead- empowerment and affective commitment mediat-
ership is aimed specifically at empowering indi- ing some of these associations (Chen et al., 2011).
vidual employees (Ahearne, Mathieu and Rapp, Empowerment theorists view psychological
2005; Arnold et al., 2000), our focus is on indi- empowerment as a mechanism through which
vidual psychological empowerment as a mediator contextual factors such as leadership influence
in the association with employee attitudinal and employee attitudes and behaviours (Maynard,
performance outcomes. However, Zhang, Zyphur Gilson and Mathieu, 2012; Spreitzer, 1995, 1996;
and Preacher’s (2009) analysis suggests that mod- Spreitzer, Kizilos and Nason, 1997). The argu-

© 2013 British Academy of Management.

130 K. H. Fong and E. Snape

ment is that empowered employees experience zational commitment (Hui, Au and Fock, 2004;
meaningfulness in their work and feelings of self- Kirkman and Rosen, 1999). In considering OCB,
efficacy and competence (Spreitzer, 1995), are we distinguish between individually focused
intrinsically motivated and actively oriented organizational citizenship behaviour (OCBI) and
towards their work (Spreitzer, 1995), and so are organization-focused organizational citizenship
likely to perform well. At the individual level, behaviour (OCBO). The former may benefit spe-
empowerment has been found to mediate between cific individuals, through interpersonal altruism,
transformational leadership and organizational the latter benefitting the organization as a whole,
commitment (Avolio et al., 2004), while Bogler by showing high levels of conscientiousness over
and Somech (2005) found that empowerment and above job requirements (Williams and
mediated the relationship between teachers’ par- Anderson, 1991). We hypothesize as follows:
ticipative decision-making and organizational
citizenship behaviour. The literature suggests that H2: Psychological empowerment mediates the
psychological empowerment and intrinsic motiva- relationships between (a) individual-level and
tion mediate the relationship between leadership (b) group-level empowering leadership, on the
and performance outcomes at the individual one hand, and (i) job satisfaction, (ii) organiza-
(Ahearne, Mathieu and Rapp, 2005; Zhang and tional commitment, (iii) IRB, (iv) OCBI and (v)
Bartol, 2010) and group levels (Srivastava, Bartol OCBO, on the other.
and Locke, 2006). One study evaluated leadership
and empowerment effects at individual and team
levels, with psychological empowerment mediat- Method
ing the relationship between leader–member
Sample and data collection
exchange and individual performance, and team
empowerment mediating the relationship between Data were collected from 302 employees and their
team climate and team performance (Chen et al., supervisors in a call centre of a Hong Kong tel-
2007). ecommunications corporation. There were no
We suggest that individual psychological formal policies aimed specifically at encouraging
empowerment will mediate the relationship employee empowerment. Employees completed
between empowering leadership behaviour and surveys assessing their supervisors’ empowering
employee outcomes. As we have seen, there is leadership behaviours, and their own psychologi-
some evidence to support this (Chen et al., 2011). cal empowerment, organizational commitment
While we conceptualize psychological empower- and job satisfaction. Supervisors rated the IRB
ment as an individual-level construct, we antici- and citizenship behaviours of subordinates.
pate between-group as well as within-group Surveys were administered at scheduled times
variance, and we assess mediation at both levels. during work hours. All were briefed on the
While this involves aggregating psychological research prior to completing the questionnaire,
empowerment to the group level, using the group and were assured that responses would be treated
mean to estimate between-group effects, in doing in confidence. Respondents completed the ques-
this we are simply recognizing the possibility of tionnaires individually and returned them directly
between-group variance, reflecting differences in to the researcher. The questionnaire was admin-
the empowering leadership behaviours of differ- istered in Chinese, the English-language original
ent group leaders. having been translated into Chinese by the first
In assessing the individual outcomes of empow- author and back-translated into English by a pro-
ering leadership, we include both job attitudes fessional translator, with the translation then
and behaviours, specifically job satisfaction and checked for accuracy.
organizational commitment, and in-role (IRB) Some employees did not respond owing to
and organizational citizenship behaviours (OCB). absence, and 23 questionnaires were dropped
The literature has looked at both attitudes and because of incomplete data. The final sample
behaviours as consequences of leadership (e.g. included 266 employees from 41 work teams, an
Bass and Riggio, 2006), and work on psychologi- effective response rate of 88%. Teams were the
cal empowerment shows a positive association employees’ primary work units, with a single
with attitudes such as job satisfaction and organi- supervisor who provided the ratings for IRB and

© 2013 British Academy of Management.

Leadership and Empowerment 131

OCB and was the target for employee ratings of scale was 0.69. This is relatively low (e.g. com-
the supervisor’s leadership. The average group pared with Seibert, Silver and Randolph’s (2004)
size was 6.5; 32% of employees worked in the 0.88), and might be expected to result in an
Hong Kong Special Administrative Region and underestimation of correlations with other con-
the rest in Guangzhou in the People’s Republic of structs. In spite of this, significant findings
China; 95% were customer service staff; over 50% emerged for psychological empowerment (see
were aged 25 or below, with over 98% aged 45 or below), although we cannot rule out the possi-
below; 75% had been with the organization for bility that these are somewhat understated.
two years or less; and 59% were female.
Discriminant validity of ELQ and PE. To estab-
lish the discriminant validity of empowering lead-
Measurement
ership and psychological empowerment, we
A 7-point Likert response format was used for all compared an unconstrained model with one in
items (1 ‘strongly disagree’ to 7 ‘strongly agree’). which the correlation between the two second-
Scales were calculated as the mean of individual order constructs was constrained to one. The
items. unconstrained two-second-order-factor model
produced an acceptable fit (χ2 = 959.64, df = 455,
Empowering leadership behaviours. Empowering CFI = 0.93, RMSEA = 0.06). The change in chi-
leadership behaviours were operationalized with square test between this and the constrained
the ELQ (Arnold et al., 2000), with 38 items model was significant (Δχ2 = 120.70, Δdf = 1,
assessing the sub-dimensions of Leading By p < 0.01), suggesting that the unconstrained
Example, Participative Decision-Making, Coach- model provided a significantly better fit. Empow-
ing, Informing and Showing Concern/Interacting ering leadership and psychological empowerment
with the Team. We replicated Arnold et al.’s were moderately correlated (r = 0.41, p < 0.01),
(2000) procedure for assessing the dimensionality but the correlation was significantly less than one.
of the ELQ, finding support for the five-factor
model (details are available from the first author). Job satisfaction. We used a three-item scale
However, as in previous studies (Arnold et al., based on Cammann et al.’s (1983) job satisfaction
2000; Srivastava, Bartol and Locke, 2006), the scale: ‘I am satisfied with my job’, ‘I am satisfied
correlations between the sub-dimensions were with my pay’ and ‘I am satisfied with the recogni-
high, and a confirmatory factor analysis sug- tion I receive for a job well done’. Alpha was 0.77.
gested a single second-order factor with an
acceptable fit (χ2 = 400.35, df = 165, CFI = 0.96, Organizational commitment. Affective organiza-
RMSEA = 0.07). The fit of the hypothesized tional commitment was measured with four items
model was comparable to the original US-based from Meyer and Allen (1997): ‘I would be happy
customer service sample (Arnold et al., 2000). to spend the rest of my career with this organiza-
Alpha for the full scale was 0.94. tion’; ‘I do not feel a strong sense of belonging to
my organization’ (reverse coded); ‘I do not feel
Psychological empowerment. Spreitzer’s (1995) part of the family at my organization’ (reverse
12-item psychological empowerment scale was coded); and ‘This organization has a great deal of
used, with four subscales: meaning, competence, personal meaning to me’. Alpha was 0.75.
self-determination and impact. The four-factor
model showed a marginal fit (χ2 = 189.01, Employee behaviours. In-role and citizenship
df = 48, CFI = 0.88, RMSEA = 0.11), but supe- behaviours were assessed by supervisors. Four
rior to a single-factor model (χ2 = 575.78, df = items from Williams and Anderson (1991) were
54, CFI = 0.57, RMSEA = 0.19; Δχ2 = 386.77, used to measure IRBs: ‘Adequately completes
Δdf = 6, p < 0.01). Consistent with Spreitzer assigned duties’; ‘Perform tasks that are expected
(1995, 1996), Spreitzer, Janasz and Quinn (1999) of him/her’; ‘Meets formal performance require-
and recent studies (Avolio et al., 2004; Seibert, ments of the job’; and ‘Fails to perform essential
Silver and Randolph, 2004), responses were duties’. Alpha was 0.90. OCBI was measured with
averaged to form an overall psychological four altruism items from Podsakoff et al. (1990):
empowerment score. Reliability for this overall ‘Helps others who have been absent’; ‘Help others

© 2013 British Academy of Management.

132 K. H. Fong and E. Snape

who have heavy workloads’; ‘Willingly helps ‘referent-shift consensus model’ (Chan, 1998),
others who have work related problems’; and ‘Is with subordinates evaluating the supervisor’s
always ready to lend a helping hand to those behaviour towards the group. This is suggested by
around him/her’. Alpha was 0.83. OCBO was the ELQ items (e.g. ‘Helps my work group see
measured using four conscientiousness items from areas in which we need more training’). To assess
the same source: ‘Attendance at work is above the group-level properties of empowering leader-
the norm’; ‘Does not take extra breaks’; ‘Obeys ship, we examined within-group agreement (rwg),
company rules and regulations even when no one intraclass correlation (ICC(1)) and reliability of
is watching’; and ‘Is one of my most conscientious the mean (ICC(2)).
employees’, with an alpha of 0.87. Because of the nested structure of the data, we
used hierarchical liner modelling (HLM) to test the
Measurement model. Fit indexes for the meas- hypotheses. Given the focus on both individual-
urement model for all employee-rated constructs level and group-level empowering leadership, we
was marginally acceptable (χ2 = 1612.33, df = 687, group-mean centred the individual-level inde-
CFI = 0.89, RMSEA = 0.07), as was that for pendent variables (empowering leadership and
the supervisor-rated constructs (χ2 = 191.69, psychological empowerment) and included their
df = 51, CFI = 0.95, RMSEA = 0.10). The three group means at level 2. This allows us to estimate
supervisor-rated constructs (IRB, OCBI and separate level 1 and level 2 mediating effects for
OCBO) were highly correlated (all > 0.7 in the level 1 outcomes (MacKinnon, 2008), and avoids
CFA), so we also estimated a single-factor model. conflating the two (Zhang, Zyphur and Preacher,
This provided an inferior fit (χ2 = 605.99, df = 54, 2009).
CFI = 0.79, RMSEA = 0.20; Δχ2 = 414.30, Δdf =
3, p < 0.01), providing evidence of discriminant
validity despite the high correlations. Results
Control variables. Initially, we decided to Table 1 shows the descriptive statistics, reliabili-
control for gender (1 = male, 2 = female), age ties and correlations. The reliabilities were
(1 = 25 or below, up to 5 = 56 or above), organi- acceptable.
zational tenure (from 0 = ‘Less than 1 year’, 1 = 1 As a prelude to testing the hypotheses, we
year, 2 = 2 years and up to 11 = ‘Greater than 10 assessed whether there was significant variance
years’), education level (from 1 = Secondary in empowering leadership at the between-group
school or below up to 5 = Doctoral degree), job level. An ANOVA indicated significant between-
level (1 = managerial and 2 = non-managerial), group variance (F = 4.61, p < 0.001), with an
job nature (1 = customer service and 2 = other) ICC(1) of 0.36 suggesting that over a third of
and work location (1 = Hong Kong and variance was between-group, and an ICC(2) of
2 = Guangzhou). However, in general, the asso- 0.78, suggesting that the mean was a reliable indi-
ciation between demographic characteristics and cator of between-group differences. Furthermore,
our hypothesized outcomes has been weak or there was evidence of considerable within-group
inconsistent (e.g. Podsakoff et al., 2000, pp. 530– agreement in employee ratings of their leader’s
531; Meyer and Allen, 1997, pp. 43–44). Consist- behaviour, with a median rwg of 0.88. These find-
ent with this, our initial analyses showed these to ings are supportive of aggregating employee
be generally non-significant, and including them ratings of their leader’s behaviour.
did not affect the conclusions. In the interests of Hypotheses 1a and 1b suggested that
parsimony and clarity, the control variables have individual- and group-level empowering leader-
been omitted from the analyses reported below. ship would be positively associated with psycho-
logical empowerment. We tested this using HLM
(Table 2). A null model with no predictors sug-
Analysis gested significant between-group variance in psy-
chological empowerment (τ00 = 0.09, p < 0.001).
Employees worked in teams, with a unique The ICC suggested that 16% of variance in psy-
supervisor. We assume that leadership is to some chological empowerment was between groups.
degree a group-level phenomenon, representing a Along with the tests for empowering leadership,

© 2013 British Academy of Management.

Leadership and Empowerment 133

Table 1. Means, standard deviations and correlationsa

Variable Mean Standard 1 2 3 4 5 6 7 Mean Standard

deviation deviation

1. Psych empowerment 4.95 0.72 (0.69) 0.52** 0.52** 0.34* 0.44* 0.36* 0.30 4.94 0.45
2. Empowering leadership 4.84 1.03 0.38** (0.94) 0.32* 0.34* 0.31* 0.30 0.25 4.97 0.74
3. Job satisfaction 4.94 1.13 0.50** 0.27** (0.77) 0.65** 0.01 0.08 −0.16 4.87 0.61
4. Org commitment 4.30 1.09 0.37** 0.21** 0.65** (0.75) −0.15 −0.13 −0.22 4.28 0.50
5. IRB 4.78 1.13 0.25** 0.18** 0.04 −0.04 (0.90) 0.82** 0.79** 4.77 0.77
6. OCBI 4.97 1.02 0.19** 0.09 0.04 0.02 0.72** (0.83) 0.62** 4.96 0.61
7. OCBO 4.71 1.28 0.13* 0.14* −0.06 −0.05 0.68** 0.63** (0.87) 4.75 0.73

Note: a Group means, standard deviations, and correlations (n = 41) are above the diagonal; individual-level means, standard
deviations and correlations (n = 266) are below the diagonal. Individual-level Cronbach alphas in parentheses.
*p < 0.05; **p < 0.01.

Table 2. Hierarchical linear modelling results for psychological accounting for 20% of between-group variance,
empowermenta
supportive of Hypothesis 1b. The association
Variable Null model Model 1 between empowering leadership and psychologi-
Level 1 Intercept 4.95*** (0.09***) 3.49*** (0.07***)
cal empowerment was thus evident at both levels.
Empowering 0.32*** (0.04†) Hypotheses 2a and 2b suggested that psycho-
leadership logical empowerment would mediate the effects of
Level 2 (group individual-level (2a) and group-level (2b) empow-
means) ering leadership. In assessing mediation, four con-
Empowering 0.30***
leadership
ditions must be satisfied (Baron and Kenny,
Within-group 0.43 0.36 1986). First, the initial variable, empowering lead-
residual variance ership, must be significantly associated with the
ICC 0.16 mediator, psychological empowerment. This was
R2within-group 0.17 met at both level 1 and level 2, as shown by the
R2between-group 0.20
Model deviance 565.67 531.45
findings for Hypothesis 1. Second, the initial vari-
able must be associated with the outcome. Third,
Note: aEmployees n = 266, Groups n = 41. Entries are estimates there must be significant relationships between the
of the fixed effects (γs). Estimates of the random variance com- mediator and outcome. Finally, the effect of the
ponents (τs) are in parentheses. The analysis was repeated with
initial variable on the outcome must be reduced
individual-level control variables, but the pattern of results did
not differ substantially from those reported here. (to zero for full mediation) when the mediator is
†
p < 0.1; *p < 0.05; **p < 0.01; ***p < 0.001. entered. In the following, conditions 2–4 are
tested in a multi-level context.
As a preliminary, we examine whether there
this justified estimating how group-level empow- was significant between-group variance in out-
ering leadership influences empowerment. There comes. The null models shown in the first column
was within-group variance in empowering leader- of Table 3 suggest that there was significant
ship and psychological empowerment, so that between-group variance for job satisfaction
within-group analysis was also viable. (ICC = 0.15), IRB (ICC = 0.25), OCBI (ICC =
Model 1 evaluated between- and within-group 0.17) and OCBO (ICC = 0.12), but not for organi-
effects, adding group-mean centred individual- zational commitment (ICC = 0.05). Testing the
level empowering leadership at level 1 and the second mediation condition, model 1 added
group mean of empowering leadership at level 2. group-mean centred empowering leadership and
The results in Table 2 show a positive association the group mean of empowering leadership. Model
between level 1 empowering leadership and psy- 2 assesses conditions 3 and 4, by adding group-
chological empowerment (γ10 = 0.32, p < 0.001), mean centred psychological empowerment and
explaining 17% of the within-group variance. the group mean of psychological empowerment.
Hypothesis 1a was supported. Group-level The findings for job satisfaction in model 1
empowering leadership also had a significant rela- suggest that, at both the within-group and
tionship to empowerment (γ01 = 0.30, p < 0.001), between-group levels, empowering leadership

© 2013 British Academy of Management.

134 K. H. Fong and E. Snape

Table 3. Hierarchical linear modelling results for employee outcomesa

Variable Null model Model 1 Model 2

Job satisfaction
Level 1 intercept 4.91*** (0.19***) 3.02*** (0.15***) 0.34 (0.15***)
Empowering leadership 0.42*** (0.10) 0.19† (0.02†)
Psychological empowerment 0.73*** (0.07)
Level 2 (group means)
Empowering leadership 0.38*** 0.13
Psychological empowerment 0.77***
Within-group residual variance 1.05 1.00 0.79
ICC 0.15
R2within-group 0.09 0.28
R2between-group 0.20 0.23
Model deviance 808.37 790.38 734.12
Organizational commitment
Level 1 intercept 4.29*** (0.06) 4.29*** (0.07†) 4.29*** (0.09**)
Empowering leadership 0.20* (0.01) 0.02 (0.02)
Psychological empowerment 0.66*** (0.00)
Level 2 (group means)
Empowering leadership
Psychological empowerment
Within-group residual variance 1.13 1.10 0.95
ICC 0.05
R2within-group 0.03 0.16
R2between-group
Model deviance 800.68 799.84 765.02
IRB
Level 1 intercept 4.77*** (0.32***) 3.58*** (0.30***) 1.16 (0.28***)
Empowering leadership 0.16* (0.02) 0.12† (0.02)
Psychological empowerment 0.23* (0.09)
Level 2 (group means)
Empowering leadership 0.24† 0.10
Psychological empowerment 0.63*
Within-group residual variance 0.99 0.96 0.90
ICC 0.25
R2within-group 0.03 0.09
R2between-group 0.05 0.12
Model deviance 794.71 791.20 783.06
OCBI
Level 1
Intercept 4.96*** (0.18***) 4.29*** (0.19**) 2.73* (0.19***)
Empowering leadership 0.11 (0.03) 0.06 (0.03)
Psychological empowerment 0.24** (0.02)
Level 2 (group means)
Empowering leadership 0.14 0.09
Psychological empowerment 0.37
Within-group residual variance 0.89 0.86 0.83
ICC 0.17
R2within-group 0.03 0.06
R2between-group 0.00 0.00
Model deviance 754.98 757.04 751.54
OCBO
Level 1 intercept 4.74*** (0.20**) 3.71*** (0.19**) 2.34† (0.18**)
Empowering leadership 0.16 (0.03) 0.13 (0.01)
Psychological empowerment 0.09 (0.14)
Level 2 (group means)
Empowering leadership 0.21† 0.09
Psychological empowerment 0.40
Within-group residual variance 1.45 1.42 1.37
ICC 0.12
R2within-group 0.02 0.05
R2between-group 0.05 0.18
Model deviance 878.00 877.91 876.01

Note: aEmployees n = 266, Groups n = 41. Entries are estimates of the fixed effects (γs). Estimates of the random variance components (τs) are
in parentheses. The analysis was repeated with individual-level control variables, but the pattern of results did not differ substantially from
those reported here.
†
p < 0.1; *p < 0.05; **p < 0.01; ***p < 0.001.

© 2013 British Academy of Management.

Leadership and Empowerment 135

was positively associated with job satisfaction significant (γ10 = 0.06, p > 0.10, and γ01 = 0.09,
(γ10 = 0.42, p < 0.001, γ01 = 0.38, p < 0.001). In p > 0.10). There was no support for between-
model 2, psychological empowerment was signifi- group mediation (Hypothesis 2biv), with no sig-
cant at level 1 (γ20 = 0.73, p < 0.001) and level 2 nificant association between empowerment and
(γ02 = 0.77, p < 0.001), with the coefficient on level OCBI at this level. However, the within-group
1 empowering leadership reduced in size, but mar- analysis provides an example of what some have
ginally significant (γ10 = 0.19, p < 0.10) and that termed an ‘indirect effect’ (Kenny, Kashy and
on level 2 not significant (γ01 = 0.13, p > 0.10). Bolger, 1998), questioning the need for a direct
This suggests partial mediation at level 1 and full relationship between the initial variable and
mediation at level 2, with significant indirect outcome in mediation, on the grounds that such a
effects for both (Sobel test statistic within = 4.25; relationship may be distal and hence more diffi-
p < 0.001; Sobel test statistic between = 2.96; cult to detect than the indirect paths to and from
p < 0.01), supportive of Hypotheses 2ai and 2bi. the mediator. Consistent with this, condition 2
For organizational commitment, the null was not met, but empowering leadership was
model suggested that there was no significant associated with empowerment and empowerment
between-group variance for organizational com- with OCBI. This indirect effect was significant
mitment (τ00 = 0.06, n.s.) so we did not analyse the (Sobel test statistic = 2.84; p < 0.01), providing
level 2 effect, assessing level 1 relationships only. evidence for Hypothesis 2aiv.
Model 1 suggested that empowering leadership Finally, for OCBO, model 1 suggests that
was positively associated with commitment at the empowering leadership was not significantly asso-
within-group level (γ10 = 0.20, p < 0.05). In model ciated with OCBO at the within-group level
2, psychological empowerment was significant at (γ10 = 0.16, p > 0.10), but was at the between-
level 1 (γ20 = 0.66, p < 0.001), and level 1 empow- group level (γ01 = 0.21, p < 0.10). In model 2, psy-
ering leadership was non-significant (γ10 = 0.02, chological empowerment was significant at
p > 0.10). These results provide evidence of full neither level 1 (γ20 = 0.09, p > 0.10) nor level 2
mediation at the within-group level only, with a (γ02 = 0.40, p > 0.10). These results provide
significant indirect effect (Sobel test statis- support for neither within-group nor between-
tic = 4.08; p < 0.001), providing support for group mediation (Hypotheses 2av and 2bv).
Hypothesis 2aii, but not 2bii.
For IRB, in model 1 empowering leadership
was positively associated with IRB at the within- Discussion
group (γ10 = 0.16, p < 0.05) and between-group
levels (γ01 = 0.24, p < 0.10). In model 2, psycho- We sought to make two contributions: to evaluate
logical empowerment was significant at level 1 empowering leadership as a group-level construct,
(γ20 = 0.23, p < 0.05) and at level 2 (γ02 = 0.63, and to test a model in which psychological
p < 0.05). The coefficient on level 1 empowering empowerment mediates the relationships between
leadership was smaller, but still significant empowering leadership and employees’ work out-
(γ10 = 0.12, p < 0.10), suggesting partial mediation comes. On the first, consistent with Srivastava,
(Sobel test statistic = 1.90; p < 0.10), providing Bartol and Locke (2006), who used an abbrevi-
support for Hypothesis 2aiii. The coefficient on ated version of the ELQ, we found acceptable
level 2 empowering leadership was not significant levels of within-group agreement and between-
(γ01 = 0.10, p > 0.10) suggesting full mediation group variability, providing evidence for the
(Sobel test statistic = 2.09; p < 0.05), consistent validity of empowering leadership at the group
with Hypothesis 2biii. level. On the second, our findings suggest that
For OCBI, model 1 suggested that empowering empowering leadership was associated with psy-
leadership was not significantly associated with chological empowerment at both the individual
OCBI at the within-group (γ10 = 0.11, p > 0.10) or and group levels. At the individual level, psycho-
between-group level (γ01 = 0.14, p > 0.10). In logical empowerment played a role as a mediator
model 2, psychological empowerment was signifi- between empowering leadership and job satisfac-
cant at level 1 (γ20 = 0.24, p < 0.01), but not at level tion, organizational commitment, IRBs and
2 (γ02 = 0.37, p > 0.10). The coefficients on level 1 OCBI, but not OCBO. At the group level, there
and level 2 empowering leadership were still non- was also significant mediation, although only for

© 2013 British Academy of Management.

136 K. H. Fong and E. Snape

job satisfaction and IRB. The individual-level 2008), although samples such as ours’ may have
findings provides support for those who argue utility in meta-analyses incorporating a cross-
that leaders treat individual subordinates differ- cultural analysis.
ently in terms of empowerment (Zhang and These findings have practical implications. The
Bartol, 2010), but the group-level findings suggest attitudes and behaviours of customer-service
that it is also important to look at between-group employees are crucial, as customers’ perceptions
differences in empowering leadership. of service may be influenced by their interactions
There are implications for debates on ‘differen- with employees, while high-quality customer
tiated’ leadership (Wu, Tsui and Kinicki, 2010). service may provide a competitive advantage
We found individual-level effects suggesting that (Chebat and Kollias, 2000). Our results suggest
leaders differentiate in their treatment of follow- that encouraging supervisors to practise empow-
ers, which has positive effects on followers’ ering leadership may underpin employee empow-
attitudes and behaviour. We did not examine erment and result in positive employee attitudes
group-level outcomes, but Wu, Tsui and Kinicki and behaviour, and ultimately improve custom-
(2010) found that differentiated leadership ers’ service experience.
resulted in within-group divergence in identifica- Our findings suggest that empowering leader-
tion with the leader, and ultimately undermined ship contributes positively to the psychological
group collective efficacy and effectiveness. This is empowerment, attitudes and work behaviours of
an area where further research on empowering followers. Employees responded to their leader’s
leadership would be valuable. In addition, it empowering leadership as an ambient stimulus,
would be useful to consider the possibility of and also to the extent to which their leader
alternative mediating mechanisms. We examined empowered them relative to others in their group.
two levels of analysis and several attitudinal and These findings add to our knowledge about the
behavioural outcomes of empowering leadership. effects of empowering leadership, showing that it
A primary aim was to assess the impact of is important to differentiate between effects at the
empowering leadership on the empowerment of individual and group levels.
followers, and the mediating role that this may
play with respect to outcomes. Nonetheless, our
findings include instances of partial mediation,
References
which leaves open the possibility that there are Ahearne, M., J. Mathieu and A. Rapp (2005). ‘To empower or
other mediating processes at work. not to empower your sales force? An empirical examination
We do not consider culture directly, so we of influence of leadership empowerment behavior on cus-
cannot draw firm conclusions about the implica- tomer satisfaction and performance’, Journal of Applied Psy-
tions of cultural differences. Nonetheless, our chology, 90, pp. 945–955.
Arnold, J. A., S. Arad, J. A. Rhoades and F. Drasgow (2000).
findings on the efficacy of empowering leadership ‘The empowering leadership questionnaire: the construction
in a Chinese sample are significant, because ques- and validation of a new scale for measuring leader behaviors’,
tions have been raised as to whether Western lead- Journal of Organizational Behavior, 21, pp. 249–269.
ership models are applicable in Asian cultures Avolio, B. J., W. Zhu, W. Koh and P. Bhatia (2004). ‘Transfor-
(Hui, Law and Chen, 1999). Some argue that mational leadership and organizational commitment: mediat-
ing role of psychological empowerment and moderating role
employees from a high power distance or uncer- of structural distance’, Journal of Organizational Behavior, 25,
tainty avoidance culture may react to attempts at pp. 951–968.
empowerment with stress and withdrawal rather Bandura, A. (1986). Social Foundations of Thought and Action:
than a sense of empowerment (Hui, Au and Fock, A Social Cognitive Theory. Englewood Cliffs, NJ: Prentice-
2004; Robert et al., 2000), although others have Hall.
Baron, R. and D. Kenny (1986). ‘The moderator–mediator vari-
found a positive association between empower- able distinction in social psychological research: conceptual,
ment and employee attitudes among Asian strategic and statistical considerations’, Journal of Personality
workers (Avolio et al., 2004). Our sample and Social Psychology, 51, pp. 1173–1182.
responded positively to empowering leadership, Bass, B. M. and R. E. Riggio (2006). Transformational Leader-
suggesting that empowering leadership is general- ship, 2nd edn. Mahwah, NJ: Lawrence Erlbaum Associates.
Bogler, R. and A. Somech (2005). ‘Organizational citizenship
izable to the Chinese context. Stronger insights on behavior in school: how does it relate to participation in
the efficacy of empowering leadership await true decision making?’ Journal of Educational Administration, 43,
cross-cultural studies (e.g. Felfe, Yan and Six, pp. 420–438.

© 2013 British Academy of Management.

Leadership and Empowerment 137

Bowen, D. E. and E. E. Lawler III (1992). ‘The empowerment of Maynard, M. T., L. L. Gilson, and J. E. Mathieu (2012).
service workers: what, why, how, and when’, Sloan Manage- ‘Empowerment – fad or fab? A multilevel review of the past
ment Review, 33, pp. 31–39. two decades of research’, Journal of Management, 38, pp.
Cammann, C., M. Fichman, D. Jenkins and J. Klesh (1983). 1231–1281.
‘Assessing the attitudes and perceptions of organizational Meyer, J. P. and NJ. Allen (1997). Commitment in the Workplace:
members’. In S. Seashore et al. (eds), Assessing Organiza- Theory, Research, and Application. Thousand Oaks, CA: Sage.
tional Change: A Guide to Methods, Measures and Practices, Nielsen, K. and K. Daniels (2012). Does shared and differenti-
pp. 71–138. New York, NY: John Wiley. ated transformational leadership predict followers’ working
Chan, D. (1998). ‘Functional relations among constructs in the conditions and well-being? Leadership Quarterly, 23, pp. 383–
same content domain at different levels of analysis: a typology 397.
of composition models’, Journal of Applied Psychology, 83, Podsakoff, P. M., S. B. MacKenzie, R. H. Moorman and R.
pp. 234–246. Fetter (1990). ‘Transformational leader behaviors and their
Chebat, J.-C. and P. Kollias (2000). ‘The impact of empower- effects on followers’ trust in leader, satisfaction, and organi-
ment on customer contact employees’ role in service organi- zational citizenship behaviors’, Leadership Quarterly, 1, pp.
zations’, Journal of Service Research, 3, pp. 66–81. 107–142.
Chen, G., B. L. Kirkman, R. Kanfer and D. Allen (2007). ‘A Podsakoff, P. M., S. B. MacKenzie, J. B. Paine and D. J.
multilevel study of leadership, empowerment, and perfor- Bachrach, (2000). ‘Organizational citizenship behaviors: a
mance in teams’, Journal of Applied Psychology, 92, pp. 331– critical review of the theoretical and empirical literature and
346. suggestions for future research’, Journal of Management, 26,
Chen, G., P. N. Sharma, S. K. Edinger, D. L. Shapiro and J.-L. pp. 513–563.
Farh (2011). ‘Motivating and demotivating forces in teams: Robert, C., T. M. Probst, J. J. Martocchio, F. Drasgow and J.
cross-level influences of empowering leadership and relation- Lawler (2000). ‘Empowerment and continuous improvement
ship conflict’, Journal of Applied Psychology, 96, pp. 541–557. in the United States, Mexico, Poland, and India: predicting fit
Dansereau, F., G. B. Graen and W. Haga, (1975). ‘A vertical on the basis of the dimensions of power distance and indi-
dyad linkage approach to leadership in formal organizations’, vidualism’, Journal of Applied Psychology, 85, pp. 643–658.
Organizational Behavior and Human Performance, 13, pp. Salancik, G. R. and J. Pfeffer (1978). ‘A social information
46–78. processing approach to job attitudes and task design’, Admin-
Feinberg, B. J., C. Ostroff and W. W. Burke (2005). ‘The role of istrative Science Quarterly, 23, pp. 224–253.
within-group agreement in understanding transformational Seibert, S. E., S. R. Silver and W. A. Randolph (2004). ‘Taking
leadership’, Journal of Occupational and Organizational Psy- empowerment to the next level: a multiple-level model of
chology, 78, p. 471. empowerment, performance, and satisfaction’, Academy of
Felfe, J., W. Yan and B. Six (2008). ‘The impact of individual Management Journal, 47, pp. 332–349.
collectivism on commitment and its influence on organiza- Seibert, S. E., G. Wang and S. H. Courtright (2011). ‘Anteced-
tional citizenship behavior and turnover in three countries’, ents and consequences of psychological and team empower-
International Journal of Cross-Cultural Management, 8, pp. ment: a meta-analytic review’, Journal of Applied Psychology,
211–237. 96, pp. 981–1003.
Graen, G. B. and M. Uhl-Bien (1995). ‘Relationship-based Sherony, K. M. and S. G. Green (2002). ‘Coworker exchange:
approach to leadership: development of leader–member relationships between coworkers, leader–member exchange,
exchange (LMX) theory of leadership over 25 years: applying and work attitudes’, Journal of Applied Psychology, 87, pp.
a multi-level multi-domain perspective’, Leadership Quar- 542–548.
terly, 6, pp. 219–247. Spreitzer, G. M. (1995). ‘Psychological empowerment in the
Hausknecht, J. P., N. J. Hiller and R. J. Vance (2008). ‘Work workplace: dimensions, measurement, and validation’,
unit absenteeism: effects of satisfaction, commitment, labor Academy of Management Journal, 38, pp. 1442–1465.
market conditions and time’, Academy of Management Spreitzer, G. M. (1996). ‘Social structural characteristics of psy-
Journal, 15, pp. 1223–1245. chological empowerment’, Academy of Management Journal,
Hui, C., K. S. Law and Z. X. Chen (1999). ‘A structural equa- 39, pp. 483–504.
tion model of the effects of negative affectivity, leader– Spreitzer, G. M., M. A. Kizilos and S. W. Nason (1997). ‘A
member exchange, and perceived job mobility on in-role and dimensional analysis of the relationship between psychologi-
extra-role performance: a Chinese case’, Organizational cal empowerment and effectiveness, satisfaction, and strain’,
Behavior and Human Decision Processes, 77, pp. 3–21. Journal of Management, 23, pp. 679–704.
Hui, M. K., K. Au and H. Fock (2004). ‘Empowerment effects Spreitzer, G. M., S. C. De Janasz and R. E. Quinn (1999).
across cultures’, Journal of International Business Studies, 35, ‘Empowered to lead: the role of psychological empowerment
pp. 46–60. in leadership’, Journal of Organizational Behavior, 20, pp.
Kenny, D. A., D. A. Kashy and N. Bolger (1998). ‘Data analysis 511–526.
in social psychology.’ In D. T. Gilbert, S. T. Fiske and G. Srivastava, A., K. M. Bartol and Locke, E. A. (2006). ‘Empow-
Lindsey (eds), The Handbook of Social Psychology, 4th edn, ering leadership in management teams: effects on knowledge
pp. 233–265. New York, NY: McGraw-Hill. sharing, efficacy, and performance’, Academy of Management
Kirkman, B. I. and B. Rosen (1999). ‘Beyond self management: Journal, 49, pp. 1239–1251.
antecedents and consequences of team empowerment’, Thomas, K. W. and B. A. Velthouse (1990). ‘Cognitive elements
Academy of Management Journal, 42, pp. 58–74. of empowerment: an “interpretive” model of intrinsic task
MacKinnon, D. P. (2008). Introduction to Statistical Mediation motivation’, Academy of Management Review, 15, pp. 666–
Analysis. New York, NY: Lawrence Erlbaum Associates. 681.

© 2013 British Academy of Management.

138 K. H. Fong and E. Snape

Uhl-Bien, M., R. Marion and B. McKelvey (2007). ‘Complexity Yukl, G. and P.P. Fu (1999). ‘Determinants of delegation and
leadership theory: shifting leadership from the industrial consultation by managers’, Journal of Organizational Behav-
age to the knowledge era’, Leadership Quarterly, 18, pp. 298– ior, 20, pp. 219–232.
318. Zhang, X. and K. M. Bartol (2010). ‘Linking empowering lead-
Williams, L. J. and S. E. Anderson (1991). ‘Job-satisfaction and ership and employee creativity: the influence of psychological
organizational commitment as predictors of organizational empowerment, intrinsic motivation, and creative process
citizenship and in-role behaviors’, Journal of Management, engagement’, Academy of Management Journal, 53, pp. 107–
17, pp. 601–617. 128.
Wu, J. B., A. S. Tsui and A. J. Kinicki (2010). ‘Consequences of Zhang, Z., M. J. Zyphur and K. J. Preacher (2009). ‘Testing
differentiated leadership in groups’, Academy of Management multilevel mediation using hierarchical linear models’,
Journal, 53, pp. 90–106. Organizational Research Methods, 12, pp. 695–719.

Kai Hung Fong is currently Group Human Resources and Administration Director with a publicly
listed company in Hong Kong. He was awarded his Doctor of Business Administration by the
Hong Kong Polytechnic University. His teaching and research interests focus on leadership and
human resource management.

Ed Snape is a Professor in the Department of Management, Hong Kong Baptist University. Prior to
joining HKBU, he taught management at the Hong Kong Polytechnic University, University of
Bradford School of Management and at the University of Strathclyde. His teaching and research
interests are in human resource management and employee relations.

© 2013 British Academy of Management.