Sie sind auf Seite 1von 11

Journal of Applied Psychology

1995, Vol.80, No. 4. 468-478

Copyright 1995 by the American Psychological Association, Inc.


0021-9010/95/$3.00

Further Assessments of Bass's (1985) Conceptualization of


Transactional and Transformational Leadership
Peter Bycio

Rick D. Hackett

Xavier University

McMaster University

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Joyce S. Allen
Xavier University
B. M. Bass (1985) proposed that the Multifactor Leadership Questionnaire consists of 5
factors: 2 facets of transactional leadership (Contingent Reward and Management-byException) and 3 facets of transformational leadership (Charismatic Leadership, Individualized Consideration, and Intellectual Stimulation). A confirmatory factor analysis
involving hospital nurses revealed some support for this 5-factor representation, but a 2factor Active-Passive model was also tenable, because the transformational components
and Contingent Reward were all highly correlated. Alternatively, differential relationships
to a series of outcomes, including intent to leave and J. P. Meyer and N. J. Allen's (1991)
facets of organizational commitment, were observed as a function of the leader behaviors
involved.

Using ideas originally proposed by Burns (1978) for political settings, Bass (1985) applied the concepts of transactional and transformational leadership to business organizations. In particular, those leaders who identified the needs
of their followers and exchanged rewards for appropriate
levels of effort and performance were viewed as transactional leaders. In contrast, transformational leadership was
seen as moving beyond transactions to increase the level of
followers' awareness for valued outcomes by expanding and
elevating their needs and encouraging them to transcend
their self-interests. These expansions or shifts in values were

thought to result in superior, otherwise unexpected levels of


follower effort and performance. In a related manner, using
constructive/developmental theory, Kuhnert and Lewis
(1987) viewed the likelihood of transformations to be a
function of ongoing changes in how leaders and followers
process and organize information about the world. From
this perspective, transformations occur only when the personal standards and the value system of the leader have become organizing processes for subordinates.

Measurement of Transactional and


Transformational Leadership
In an attempt to identify the behaviors underlying the
transactional and transformational conceptualizations,
Bass (1985) developed the Multifactor Leadership Questionnaire (MLQ). A principal-components analysis of the
73-item MLQ with a sample of military officers revealed
five leadership factors. Three factors were seen as being
transformational: (a) Charismatic Leadership, the amount
of faith, respect, and inspiration engendered by the leader;
(b) Individualized Consideration, the degree of attention
and support given to individual followers; and (c) Intellectual Stimulation, the extent to which the leader enables followers to rethink the ways they do things. The remaining
two factors were seen as being transactional: (a) Contingent
Reward, the degree to which the leader provides reinforcement in return for appropriate follower behavior and (b)
Management-by-Exception, the extent to which subordinates hear from the leader only when failures or problems
occur.

Peter Bycio, Department of Management and Entrepreneurship, College of Business, Xavier University; Rick D. Hackett,
Michael G. Degroote School of Business, McMaster University,
Hamilton, Ontario, Canada; Joyce S. Allen, Accounting and
Information Systems, College of Business, Xavier University.
The confirmatory factor analysis portion of this article was
originally presented at the March 1994 Research Methods Division Conference on Causal Modeling held at the Krannert
School of Management, Purdue University.
This research was funded by the Xavier College of Business
D. J. O'Conor Fellowship and by the Social Sciences and Humanities Research Council of Canada (Grant 410-91 -0903).
We gratefully acknowledge the technical support of Peter
Hausdorf and Anami Bhargava.
Correspondence concerning this article should be addressed
to Peter Bycio, Department of Management and Entrepreneurship, College of Business, Xavier University, Cincinnati, Ohio
45207-5163. Electronic mail may be sent via Internet to
bycio@xavier.xu.edu.
468

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

TRANSACTIONAL AND TRANSFORMATIONAL LEADERSHIP

469

Following the initial analysis, several subsequent investigations have used different versions of the MLQ. For
example, Bass (1985) also referenced Form 2 (p. 220),
Form 4 (p. 220), a modified Form 4 (p. 225), and an
unnumbered 37-item MLQ (p. 225). More recent studies used Form 5 (Hater & Bass, 1988), Form 11R
(Waldman, Bass, & Yammarino, 1990), and Form 10
(Howell & Avolio, 1993). The number of items and their
specific content vary among these forms, as does the
target user population. Some versions also target more
than five constructs. For example, Forms 5 and 10 attempt to distinguish between Active and Passive Management-by-Exception. Regardless of the particular form involved, subsets of the MLQ facets have been differentially
related to leader performance, organizational outcomes,
and follower satisfaction (Hater & Bass, 1988; Keller,
1992), performance appraisal effectiveness (Waldman,
Bass, & Einstein, 1987), and success in a management
group simulation (Avolio, Waldman, & Einstein, 1988).
Various forms of the MLQ have also been used to assess
the augmentation hypothesis (cf. Bass, 1985), which
states that components of transformational leadership
should predict performance and satisfaction outcomes
beyond what can be accounted for by the transactional
scales but not vice versa (Hater & Bass, 1988; Waldman
etal., 1990).
Despite the relatively wide range of outcomes that have
been related to various forms of the MLQ, little research
has been devoted to the underlying psychometric characteristics of the questionnaire. As Bass (1985) noted, for
example, there is a need to replicate the MLQ factor
structure with diverse samples and occupations. Moreover, Bass's original analysis did not allow for the possibility that the factors were interrelated, despite his
contention that leaders are capable of being both transactional and transformational. Until this point, however,
only Howell and Avolio (1993) have studied the MLQ
factor structure in a way that allowed the facets to be correlated. Using MLQ-10 responses from a relatively small
group of managers (78 leaders as rated by 322 followers),
Howell and Avolio conducted a partial least squares analysis that revealed low to moderate correlations among the
six target dimensions. Obviously, the extent to which
transactional and transformational leadership are correlated is of theoretical interest, because Burns (1978) originally viewed them as representing opposite ends of the
same continuum. Accordingly, one aim of the present
study, which used hospital nurses (N = 1,376), was to use
a large sample-analytical technique, confirmatory factor
analysis, to further examine the underlying structure of
the MLQ-1.

est of parsimony, we began with the simplest possible


MLQ representation, a general factor. Next, we evaluated
three more complex models consisting of a pair of twofactor representations and a five-factor representation. As
noted below, each of these more complex structures had
some theoretical or empirical justification. Accordingly,
a specific hypothesis regarding the relative merits of these
models would be premature.
The first two-factor representation of interest involved
a Transformational factor consisting of charismatic leadership, individualized consideration, and intellectual
stimulation and a Transactional factor incorporating
contingent reward and management-by-exception. This
model implies support for the basic distinction between
the two forms of leadership but not for their specific components. It is consistent with studies in which the transformational facets augment the transactional facets in explaining variance related to leadership outcomes (cf.
Hater & Bass, 1988; Waldman etal., 1990).
The second two-factor representation of interest involved a dimension defined solely by Management-byException and a second dimension that incorporated the
transformational and contingent reward items. This
model is in line with the two-factor higher order analysis
in Bass (1985, p. 215), which revealed a Passive (defined
by management-by-exception) and an Active (the remaining four MLQ-1 scales) Leadership conceptualization. Support for this Active-Passive distinction was subsequently found in a study by Waldman et al. (1987) in
which management-by-exception was negatively related
to a series of performance appraisal outcomes, whereas
positive relationships involving each of the transformational scales and contingent reward were observed. Similarly, Avolio et al. (1988) found measures of financial
performance in a management game to be significantly
related to the transformational scales and to contingent
reward but not to management-by-exception. Medley
and LaRochelle (1994) also argued that the MLQ-1 was
best represented by Active and Passive dimensions.
Finally, a five-factor model consisting of three transformational facets and two transactional facets was examined. This representation is consistent with the exploratory first-order factor analysis in Bass (1985) and with
the MLQ-10 results of Howell and Avolio (1993). The
need to evaluate this more complex structure is also evident from the study by Avolio et al. (1988) in which more
than one transformational facet was required to account
for optimal variance in leadership outcomes.

Alternative MLQ-1 Factor Structures


We used a nested analysis (cf. Bollen, 1989, pp. 291296) to compare a series of MLQ-1 models. In the inter-

Another aim of this study was to examine the relationship of the MLQ-1 factors to a series of outcome
variables. Three groupings were studied: performance

Leadership Facets and Outcome Variables

470

P. BVCIO, R. HACKETT, AND J. ALLEN

and satisfaction, intent to leave, and organizational


commitment.

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Performance and Satisfaction


As originally constructed, the Bass (1985) scales allow
for follower assessments of (a) their levels of extra effort,
(b) their satisfaction with the leader, and (c) the
effectiveness of the leader. These variables are pertinent
to the augmentation hypothesis, namely, that transformational leadership generates enhanced levels of subordinates' effort and performance beyond what transactional processes do. For example, in both military and
Fortune 500 samples, Bass (1985, pp. 227-228) found
that transformational leadership predicted variance in
subordinates' extra effort and subordinate-rated leader
effectiveness beyond what could be accounted for by the
transactional scales alone. Similar findings for subordinate-rated leader effectiveness were reported with managers at an express air firm (Hater & Bass, 1988) and
U.S. Navy officers (Waldman et al., 1990). Therefore, we
expected the transformational scales to augment the
transactional scales in the prediction of subordinate-assessed performance and satisfaction.

transformational leadership is used (Bass, 1985, p. 32).


This is especially true since recent investigations have reconceptualized organizational commitment into several
facets, which might each be differentially related to Bass's
scales. Thus, Meyer and Allen (1991) viewed commitment in terms of three components: (a) affective commitment (AC), reflecting the employee's emotional attachment to and involvement with the organization; (b)
continuance commitment (CC), relating to the costs the
employee associates with leaving; and (c) normative
commitment (NC), the employee's feelings of obligation
to stay. When this three-component commitment model
is examined in the context of Bass's view of leadership,
some interesting possibilities emerge. For example, transformational leadership might exhibit strong positive relationships to AC, given the strong feelings of emotional
attachment it is supposed to foster. Furthermore, contingent reward might be positively related to CC, given its
focus on the accumulated benefits that would be lost by
leaving. Thus, an exploration of the ties between the Bass
(1985) and Meyer and Allen (1991) models is needed.
Method

Intent to Leave

Participants and Procedure

Intent to quit the nursing profession and intent to leave


the job were explored for their possible differential relationships to the leadership scales. Specifically, given the
inspirational, challenge-oriented nature of the transformational facets, they might have strong negative relationships with intent to quit the profession, provided that the
initial career choice was appropriate. In contrast, management-by-exception might have less to do with intentions regarding the profession, because employees wanting to avoid superiors who appear only when things go
wrong would typically consider alternative jobs, not alternative professions.
With regard to intent to leave the job, we expected a
strong negative association with contingent reward. This
follows from Waldman et al. (1990), who viewed contingent reward as a basic element of the employment contract that is necessary for the maintenance of minimally
acceptable performance. Moreover, continuing with
Waldman et al.'s logic, if appropriate rewards were combined with an inspirational, challenging work environment emanating from transformational leadership, an
augmentation effect for intent to leave would be observed. This effect is similar to the one anticipated for
performance outcomes, albeit in the opposite direction.

Questionnaires were sent to a random sample of 4,000 registered nurses who belonged to a 75,000-member nursing association. As was explained in an accompanying cover letter, the
project was conducted purely for research purposes wherein the
association would receive group-level findings only. Those who
did not respond were sent a follow-up reminder postcard. Two
weeks after this second mailing, nonrespondents were sent another copy of the questionnaire. This resulted in a 57% response
rate; however, after missing data were accounted for, a sample
of 1,376 resulted. This loss was not surprising, given that the
MLQ-1 directions specifically give respondents the option of
leaving an item blank if they regard it to be nonapplicable or if
they are uncertain about it. Ninety-seven percent of the participants were female, and 3% were male; 74% were married; the
mean age was 37 years; and the mean organizational tenure was
9 years.

Organizational Commitment
There is a need to know more about employee commitment levels as a function of whether transactional or

Measurement of Leadership Variables


In this study, we used the subset of MLQ-1 items that defined
the factors in the Bass (1985, pp. 209-212) first-order exploratory analysis. Thus, there were 13 transactional items: 7 for
Contingent Reward and 6, all of the passive variety, for Management-by-Exception. We used 27 transformational items, including 18 for Charismatic Leadership, 7 for Individualized
Consideration, and 3 for Intellectual Stimulation. Item 37, "inspires loyalty to him/her," which also originally helped define
Charismatic Leadership, was not included in the analysis because it was inadvertently omitted from our questionnaire. A 5point scale ranging from 0 to 4 was used for all items, with

TRANSACTIONAL AND TRANSFORMATIONAL LEADERSHIP

higher numbers representing greater perceived amounts of the


leadership attribute.

Measurement of Outcome Variables


We assessed extra effort (3 items), satisfaction with the leader
(2 items), and leader effectiveness (4 items) by using the scales
within the MLQ-1 (cf. Bass, 1985). Also, we used 3-item scales
to assess intent to quit one's current job and intent to leave the
profession. AC, CC, and NC were measured with the three 8item instruments presented in Allen and Meyer (1990).

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Data Analysis
Confirmatory factor analysis. We studied the factor structure of the MLQ-1 by using LISREL 7 (Joreskog & Sorbom,
1986) to determine if the data were represented best by (a) one
general factor, (b) two correlated factors (Transactional and
Transformational Leadership), (c) two other correlated factors
(Management-by-Exception vs. the remaining scales), or (d)
five correlated factors (Contingent Reward, Management-byException, Charismatic Leadership, Individualized Consideration, and Intellectual Stimulation). The item-level covariance
matrix was analyzed using maximum likelihood estimation,
and standardized results are presented.
LISREL 7 generates an estimated matrix by using an investigator-specified factor structure as a guide. If only small differences exist between the actual and estimated matrices, the hypothesized factor structure is viewed as a plausible one. Unfortunately, there is a lack of consensus regarding how to best
determine when a difference is small (Bentler, 1990; Mulaik et
al., 1989). As recommended by Bollen (1989, p. 281), we assessed the relative fit of the models by using several indices: the
nonnormed fit index, the comparative fit index (Bentler, 1990),
the parsimonious fit index (James, Mulaik, & Brett, 1982), the
goodness-of-fit index, the adjusted goodness-of-fix index, and
the root-mean-square residual (Joreskog & Sorbom, 1986).
Tests of the augmentation hypothesis. We used hierarchical
multiple regression to test the augmentation hypothesis (cf.
Bass, 1985; Hater & Bass, 1988). Specifically, first we entered
the transactional scales into the regression equation followed by
the transformational ones. We used an .Ftest to determine if the
transformational scales added significantly to the prediction of
the outcome variables.
Comparing the size of dependent correlations. The expected
differential relationships between the leadership scales and the
outcome variables (intent to quit the nursing profession, intent
to quit the job, as well as AC, CC, and NC) were assessed using
the formula given by Cohen and Cohen (1983, p. 56) for comparing the size of dependent correlations.

Results
Confirmatory Factor Analysis
Table 1 shows that as one progresses from the most
restricted model (one general factor) to the least restricted model (five correlated factors), the indices generally exhibit incremental improvements in overall fit.

471

This, in itself, is not surprising, because models with a


greater number of free parameters (i.e., less restricted)
will by definition fit better. Still, improvements are apparent even among the indices (e.g., the adjusted goodness-of-fit index and the parsimonious fix index) that
control for increases in the number of parameters estimated. In any case, the nonnormed fit index of .89 for
the five-factor model is very close to .90, a benchmark
of good overall fit (cf. Bentler & Bonett, 1980). With a
nonnormed fit index of .86, the two-factor Active-Passive
Leadership representation can also be regarded as a reasonable fit, especially because it has somewhat different
implications than Bass's (1985) original conceptualization. Therefore, we examined both models more closely.
Results pertaining to alternative representations can be
obtained from Peter Bycio.
Table 2 shows the MLQ-1 items and their means, standard deviations, and alpha scale reliabilities. Squared factor loadings, reflecting the proportion of leader and error
variance associated with each item, are also presented for
both the Active-Passive model and the five-factor model.
The means for the transformational scales ranged between 1.32 and 2.08, which (given the scale labels used)
implies that these behaviors were observed only once in a
while or, at most, sometimes. Thus, even though the large
standard deviations combined with the relatively normal
distributions imply that at least some head nurses were
typically transformational, the mean values are still a
contrast to other MLQ-1 and MLQ-5 samples (Bass,
1985; Bass & Avolio, 1990). For example, among followers of U.S. Army personnel attending War College, those
responding to the MLQ-1 indicated that transformational leadership typically occurred sometimes or fairly
often (cf. Bass, 1985, pp. 200-201). Our averages are
also generally lower than those in other studies involving
various subsets of MLQ-1 items. In manufacturing
(Waldman et al., 1987) and in research and development
(Keller, 1992) environments, the means for charismatic
leadership and intellectual stimulation ranged between
1.98 and 3.20.
With regard to the squared factor loadings, note that
the pattern of findings was very similar for both models.
First, although the large sample helped make all the
leader-based loadings statistically significant, each of
the Contingent Reward and Management-by-Exception
items had very large error components. In fact, the average error variance across these items was .72 for the twofactor Active-Passive model and .66 for the five-factor
model. Thus, both scales have a high degree of nonleadership content, despite their satisfactory alphas of .71 for
Management-by-Exception and .80 for Contingent Reward (see Table 2). The transformational scales, in contrast, had lower error components, which averaged .43 for

472

P. BYCIO, R. HACKETT, AND J. ALLEN

Table 1
Overall Fit Indices for the Multifactor Leadership Questionnaire-1 Scales
Model

x2

df

NNFI

CFI

PFI

GFI

AGFI

RMSR

Null model
One general factor
Two correlated factors"
Two correlated factors'1
Five correlated factors

39,928.92
6,857.76
6,516.21
6,085.21
4,683.07

780
740
739
739
730

.84
.84
.86
.89

.84
.85
.86
.90

.79
.79
.80
.83

.73
.75
.76
.82

.70
.72
.73

.09
.09
.10
.10

.80

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Note. N = 1,376. NNFI = nonnormed fit index; CFI = comparative fit index; PFI = parsimonious fit
index; GFI = goodness-of-fit index; AGFI = adjusted GFI; RMSR = root-mean-square residual.
" One transactional factor and one transformational factor. bManagement-by-Exception versus the remaining four scales.

the two-factor Active-Passive model and .39 for the fivefactor model.
The intercorrelations associated with the five-factor
model are shown in Table 3. Note that the correlations
among the transformational factors were uniformly high
(.81-.91) and that Contingent Reward was strongly related (.79-.83) to the transformational facets. Only
Management-by-Exception was easily distinguishable, as
reflected by its negative relationships with all of the other
factors. This pattern is consistent with the .43 correlation observed among the factors in the two-factor ActivePassive model. Thus, except for Management-by-Exception, both models imply that these factors are highly
correlated.
Relationships to Outcome Variables
Given the questions raised in the confirmatory factor
analysis concerning the dimensionality of the MLQ-1, we
not only examined the relationship of the five individual
scales to various outcomes, but we also created an additional predictor, active leadership, by combining the
transformational scales with contingent reward. Obviously, it was of interest to compare the outcomes' relationships that emerged from using a single active leadership dimension as opposed to the four individual scales.
Descriptive statistics for the leadership scales and the outcome variables are shown in Table 4. Multiple regression
results pertaining to the augmentation hypothesis are
presented in Table 5.
Performance and satisfaction. Consistent with previous findings, Table 4 shows that the transformational
scales had strong positive relationships with subordinates' extra effort, satisfaction with the leader, and subordinate-rated leader effectiveness (cf. Bass, 1985; Bass &
Avolio, 1990). Contingent reward was also positively related to these outcomes, but with few exceptions, t tests
for dependent correlations (dfs = 1, 357) revealed the
relationships to be smaller in magnitude (ps < .01) than
those involving the transformational scales. The signifi-

cant negative associations involving management-by-exception and the performance and satisfaction measures
were also in line with previous studies (cf. Bass, 1985;
Bass & Avolio, 1990) in which either zero or negative
findings have typically been observed.
With regard to the augmentation effect, F tests associated with the hierarchical multiple regressions (see Table
5) revealed that when the transformational scales were
added as predictors to the transactional facets, significant
proportions of additional variance were accounted for in
the performance and satisfaction variables, as we hypothesized. Alternatively, contrary to Bass's (1985) model, it
is clear that charismatic leadership, by itself, was the dominant predictor. For example, the multiple correlation of
.71 involving extra effort and the transactional scales
showed a significant rise, F(3, 1370) = 337.60, p < .01,
to .85 when the transformational scales were added to the
equation, but this increase was not impressive considering that the simple bivariate correlation between charismatic leadership and extra effort was .82. Even when all
of the transformational scales were combined with contingent reward to form active leadership, the level of prediction (R = .83; see Table 5) was virtually identical to
the predictive ability of charismatic leadership alone. Together, these results imply a lack of discriminant validity
among the scales comprising active leadership and suggest that the two-factor Active-Passive model may be
more suitable.
Intent to leave. As we hypothesized, each transformational facet had a significant (p < .01) relationship with
intent to leave the nursing profession; however, the correlations were modest in magnitude (rs ranged from .23
to -.27; see Table 4). In fact, AC, an organization-based
attitude, shared significantly more variance with intent
to leave the profession (r = -.37; see Table 4) than did
any of the transformational facets. In contrast, as expected, greater degrees of transformational leadership
were associated with reductions in the intent to leave the
profession, whereas management-by-exception had a significantly smaller association with this intention.

473

TRANSACTIONAL AND TRANSFORMATIONAL LEADERSHIP


Table 2
Statistics Associated With the Two-Factor Active-Passive Model and the Five-Factor Model
Leadership
variance
SD

Ml

M2

Ml

M2

1.25
1.36
1.42
1.42
1.29
1.31
1.36
1.26
1.24
1.33
1.36

.56
.41
.80
.79
.67
.82
.82
.70
.66
.55
.74

.56
.40
.79
.77
.66
.80
.81
.71
.66
.54
.71

.44
.59
.20
.21
.33
.18
.18
.30
.34
.45
.26

.44
.60
.21
.23
.34
.20
.19
.29
.34
.46
.29

1.24
1.31
1.21
1.26
1.25
1.13

.67
.50
.49
.63
.56
.68

.67
.51
.50
.64
.56
.68

.33
.50
.51
.37
.44
.32

.33
.49
.50
.36
.44
.32

.10
.25
.35
.32

.39
.08
.51
.64

.31
.05
.41
.55

.61
.92
.49
.36

.69
.95
.59
.45

.34
.21
.30

.70
.64
.43

.58
.61
.42

.30
.36
.57

.42
.39
.58

1.27

.75

.66

.25

.34

1.13
1.17

.58
.73

.46
.63

.42
.27

.54
.37

1.26
0.77
0.98

.30
.10
.12

.47
.42
.46

.32
.36
.36

.53
.58
.54

.68
.64
.64

1.33
0.75
0.78

.12
.00
.04

.49
.41
.17

.32
.24
.07

.51
.59
.83

.68
.76
.93

1.49

1.20

.21

.16

.79

.84

Management-by-Exceptionc
25. Is content to let me continue doing my job in the same way as always
2.49
54. Asks no more of me than what is absolutely essential to get the work done
1.51
58. Only tells me what I have to know to do my job
1.54
6 1 . As long as things are going all right, he/she does not try to change anything
2.24
69. As long as the old ways work, he/she is satisfied with my performance
1.91
71. It is all right if I take initiatives, but he/she does not encourage me to do so
1.55

1.13
1.22
1.30
1.30
1.26
1.32

.15
.15
.29
.43
.43
.32

.15
.15
.29
.42
.43
.33

.85
.85
.71
.57
.57
.68

.85
.85
.71
.58
.57
.67

MLQ-1 item

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Error variance

Charismatic Leadership"
2.10
1 . Makes me feel good to be around him/her
2.04
12. Commands respect from everyone
.51
17. Is a model for me to follow
.51
18. In my mind, he/she is a symbol of success and accomplishment
.52
22. I am ready to trust his capacity and judgment to overcome any obstacle
.42
26. Is an inspiration to us
.75
27. Makes me proud to be associated with him/her
.34
29. Has a special gift for seeing what is really important for me to consider
.22
38. Increases my optimism for the future
.65
40. Inspires loyalty to the organization
.68
4 1 . 1 have complete faith in him/her
42. Excites us with his/her visions of what we may be able to accomplish if we
1.37
work together
2.22
50. Encourages me to express my ideas and opinions
2.04
60. Encourages understanding of points of view of other members
1.55
62. Gives me a sense of overall purpose
1.36
66. Has a sense of mission which he/she transmits to me
68. Makes everyone around him/her enthusiastic about assignments
1.26
3.
5.
6.
10.
1 1.
1 5.
43.

Individualized Consideration1"
Is satisfied when I meet the agreed-upon standards for good work
2.96
Makes me feel we can reach our goals without him/her if we have to
2.45
2.17
I earn credit with him/her for doing my job well
Finds out what I want and tries to help me get it
1.65
You can count on him/her to express his/her appreciation when you do a
good job
1.85
Gives personal attention to members who seem neglected
.35
Treats each subordinate individually
2.15

Intellectual Stimulation"
19. Has provided me with new ways of looking at things which used to be a
puzzle for me
1.30
30. His/her ideas have forced me to rethink some of my own ideas which I had
never questioned before
1.25
32. Enables me to think about old problems in new ways
1.42
7.
21.
48.
53.
63.
65.
72.

Contingent Rewardd
Assures me I can get what I personally want in exchange for my efforts
Talks a lot about special commendations and promotions for good work
I decide what I want; he/she shows me how to get it
Whenever I feel it necessary, I can negotiate with him/her about what I
can get for what I accomplish
Tells me what I should do if I want to be rewarded for my efforts
Gives me what I want in exchange for showing my support for him/her
There is close agreement between what I am expected to put into the
group effort and what I can get out of it

Note. N = 1,376. Each item was rated on the following 5-point scale: 0 (not all all), 1 (once in a while), 2 (sometimes), 3 (fairly often), and 4
(frequently). M LQ-1 items are from Leadership and Performance Beyond Expectations (pp. 210 - 212), by B. M. Bass, 1985, New York: Free Press.
Copyright 1985 by The Free Press, a division of Simon & Schuster Inc. Reprinted with permission of the publisher. Active Leadership consisted of
the items from Charismatic Leadership, Individualized Consideration, Intellectual Stimulation, and Contingent Reward. Passive Leadership consisted of the items from Management-by-Exception. For Active Leadership, M = 1.57, SD = 0.90, and a = .97. MLQ-1 = Multifactor Leadership
Questionnaire; M1 = two-factor Active-Passive model; M2 = five-factor model.
'M= 1.62,50 = 1.06, a = .97. bM= 2.08, SD = 0.93, a = .85. c A/= 1.32,SD = 1.06, a = .87. "A/= 1.05,SD = 0.78, a = .80. c A/= 1.87,

474

P. BYCIO, R. HACKETT, AND J. ALLEN


Table 3
Factor Intercorrelationsfor the Five-Factor Model
Factor

1 . Charismatic Leadership
2. Individualized Consideration
3. Intellectual Stimulation
4. Contingent Reward
5. Management-by-Exception

_
.90
.91
.79
-.43

.83
.83
-.39

.81
-.44

-.26

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Note. N= 1,376.

With regard to intent to leave the job, a significant negative relationship (r = -.21, p < .01) involving contingent reward was observed, as hypothesized. However, as
with intent to leave the profession, the organizationbased AC, by itself, predicted intent to leave the job to a
higher degree (r = -.42) than did the transactional and
transformational scales combined (R = .33; see Tables 4
and 5). Finally, support for the augmentation effect was
also obtained when the transformational scales were
added as predictors to the transactional scales, F(3,
1259) = 28.55, p < .00. Note that charismatic leadership
was again the major source of augmentation. It shared
almost the same proportion of variance with the intent
to leave variables as did all five of the leadership scales
combined (see Table 5).
Organizational commitment. As expected, AC had
strong positive relationships with each of the transformational scales (see Table 4). Furthermore, t tests for dependent correlations (dfs = 1,257) revealed that the AC
transformational values were all significantly larger (p <
.01) than those involving CC and NC. However, contrary
to expectations, CC was not positively associated with
contingent reward. Unanticipated significant, but small,
positive correlations were also observed between NC and
the transformational scales.

Discussion
Psychometric Properties of the MLQ
Although it is possible to view our results as being consistent with Bass's (1985) five-factor model of leadership,
the evidence cannot be regarded as exclusively positive.
First, there were high proportions of error variance associated with both the contingent reward and managementby-exception scales. Second, although the overall confirmatory factor analysis fit indices tended to support the
existence of five leadership components, the transformational factors were highly correlated, and more important, they generally did not have strong differential relationships with the outcome variables. Accordingly, despite the differences in overall fit indices, one could argue
that the simpler two-factor Active-Passive model is the
best reflection of these data.

With regard to the error variance and the discriminant


validity of the MLQ-1, it is important to reemphasize
that multiple forms of the questionnaire exist. In fact,
some versions reflect revisions or replacements of approximately half of the MLQ-1 items. Thus, the comparatively low average correlation of .54 that Howell and
Avolio (1993) found among the transformational factors
using Form 10 may be partly attributable to item differences. Of course, as noted earlier, Howell and Avolio used
partial least squares analysis as opposed to confirmatory
factor analysis, which also may account for some of the
discrepancy.
Substantial sample-based dissimilarities also exist between our study and that of Howell and Avolio (1993).
We used female head nurses, whereas they investigated
male senior managers at a financial institution. This
might have implications for both the mean levels of leadership observed and the patterns of leadership practiced.
One could argue, for example, that the unity of command principle is more likely to be violated in nursing
than it is in a large corporate setting. Specifically, in our
sample, the leadership function might somehow have
been split between the physicians and the head nurses,
which could account for the relatively low means. Moreover, the pattern of factor correlations we obtained might
be attributable to the fact that some of the managementby-exception originated with physicians, whereas much
of the transformational leadership and contingent reward
(to the extent that these were present) originated with
head nurses. Finally, it should be noted that our sample
was almost exclusively female, whereas Howell and Avolio's sample was virtually all male. Obviously a large-sample confirmatory factor analysis of the MLQ-10 that uses
a demographically diverse sample is needed.

Relationships to Outcome Variables


Performance and satisfaction. Consistent with our
findings, a series of studies have shown that ratings of
leader effectiveness relate positively to transformational
leadership and negatively to management-by-exception
(e.g., Bass & Avolio, 1990; Hater & Bass, 1988). Accordingly, future investigators of the MLQ performance links

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

TRANSACTIONAL AND TRANSFORMATIONAL LEADERSHIP

rr 'r

I " (N t~- (N *O V> OO

r r r
r^ fi r^i (N ^ rs ^ o

r r r i' ' ' r ' r

"B

1
-I

l
1
<3

t?

P*
is P p <a a -a

475

should be sensitive to method variance issues (cf. Hunt,


1991) and should use independently collected indices
more closely tied to business objectives that are not subject to single-source attributional bias (e.g., Howell &
Avolio, 1993; Keller, 1992). Finally, although there was
support for the augmentation hypothesis among these
outcomes, charismatic leadership by itself was responsible for the effect, which, as noted earlier, raises doubts
about the appropriateness of a five-factor MLQ-1 model.
Intent to leave. As expected, management-by-exception had little to do with one's intent to leave the profession, whereas transformational leadership was associated
with modest decreases in this intention. Thus, a positive,
active leader might be reassuring to employees who are
experiencing doubts about their profession, but the relatively small size of the relationships involved implies limits regarding the degree to which even transformational
leaders can influence professional choices. Indeed, AC,
an organization-based attitude, shared significantly more
variance with intent to leave the profession than did any
of the transformational factors.
The pattern of results pertaining to intent to leave the
job was more in line with our expectations. In particular,
contingent reward was associated with a reduction in
one's intent to leave the job. Furthermore, additional reductions in this intention were obtained, given the presence of transformational leadership. However, as with intent to leave the profession, the organization-based AC,
by itself, predicted intent to leave the job to a higher degree than did the transactional and transformational
scales combined. Accordingly, despite the relatively
broad range of positive outcomes that Bass (1985) associated with transformational leadership, our results are
consistent with a meta-analysis conducted by Mathieu
and Zajac (1990), which suggests that the affective component of commitment has a much stronger association
with turnover intent.
Organizational commitment. As expected, the transformational scales had strong positive relationships with
AC that were significantly larger than those involving CC
or NC. This finding is consistent with the view that the
inspirational aspects of transformational leadership enhance AC but not the other less emotion-based facets of
commitment. The strong correlations are also notable
because they suggest that transformational leadership is
a significantly better predictor of AC than is leader consideration, which has an uncorrected meta-analytic
mean of .24 with attitudinal commitment (cf. Mathieu
& Zajac, 1990).
The strong relationship expected between contingent
reward and CC did not materialize. Conceptually, it is
difficult to explain why leaders who rely heavily on
differential rewards should not engender high levels of CC
in their subordinates. A retrospective explanation relates

476

P. BYCIO, R. HACKETT, AND J. ALLEN

.8

u c
> CJ

r* f*) t~~- OO

O t^~ O

?3

(NOfN

& *~

c E

- -^ ^

<N(^

Z
o
u

r- oo oo o

' fN

r ' * "*

"

5 c

rt rt
ft> o

nc co
O

a 'a
o c

11

S..S

S-2 SSSS 2S
i
i i i
-

SC '3s

(_>
U o

8*!

o" "

Pi 3
rn" ^ <N

.1 1
Affect;
commit!

~7

"" i/1} *<O ^?

-p.-

I*

'

" Os
00
(N

l.f

*"* QJ
G -^

__

t*s

i*

s
.2

i* r r

'06
(N

u c

2<- *ou

S
OB

S 'S

-0<N
<N <N

1-^fNI<N(N<N(N

t-?f
(NOO

rrr

'^

i-5 *

bi
c
o

^
W5

(5
fe c
*o >

uS '-5

's;

'

oo
S oo

f~

|
li

"5
^
(U
O

' f ' f*^


o

oo

"5 . ^

'

. ^ .

. .

i*-

t> .S "g
"5 2 ed

Tt

**

c o

oi)5 c _C "53 M
*~* "*" ?>

III
^ o S
.2P J3
w~) (N (N 1^'
fN NO ^O <?N
i' ' " <>

O
c
o

3
00
3
<

Pi

'c "

c' o_, r^
.(N, c^
^ ON
^.
a
Ic
S2
u
a
n
^
(U
>

j3 *o ta

5/1

-T^

c
O

^ ^> *2
C

^ c^

!-*!
K ?i "

i/-, (N O
m oo oo rn

l;i

rj
^

c ^*^ S
2 S c

oo

2 .^
"o <uS
^
soc^r^
. . .
1

Sc

S^

oor-i^-oo
. . . .

oovo
. .
m

UJ d>

-c u

2 IP

(N

^
1
k.

- Sg

eT^o.

^2 -a

liilliijlif
H

""

^ gfn
cd O

^i7
if^
S *o <S

t,

Passive leadership (r)


Active leadership (r)
Passive + active leadership
f*

Leadership scale

"5

-c

oa

*
f-) QQ QQ (^

K c

" Tested if adding the trans


profession where df= 3, 1 3
the variance that was accou

*:

o
c
o

j 1 1
3 jo

2 A "*"

O
cO

s ^* .s

feli
i{g
|."i

CO (N <N ^^ r^i fi r^

J>
>

"* O

*J5

e
1
1)

C*3

4l~H
/-, ^O
rrt

Table 5
Multiple Regression Resi

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

TRANSACTIONS, AND TRANSFORMATIONAL LEADERSHIP

to the specific nature of the CC item content. In particular, as noted elsewhere (cf. Hackett, Bycio, & Hausdorf,
1994; McGee & Ford, 1987), this scale contains items
reflecting both the accumulated benefits that would be
lost by leaving and one's perception of the number of alternative employment options that exist. Transactional
leadership has little to do with the latter but should be
related to the former. In other words, CC, as measured by
Meyer and Allen (1991), does not deal exclusively with
the potential loss of rewards that one might have expected
to accumulate from a leader who relies heavily on contingent reward. Of course, nurses in our sample perceived
contingent reward as being typically practiced only once
in a while, and this may have put restrictions on the magnitude of the relationship with CC.
Although they were not expected, significant but small
correlations were observed between NC and all the leadership scales, except for management-by-exception. One
post hoc explanation, consistent with the perspective of
Kuhnert and Lewis (1987), is that transformational
leaders typically hold a sense of moral obligation to the
organization as an end value, which in turn is adopted by
subordinates. Perhaps the modest size of the relationships involving NC and these active facets reflects the influence of less positive models in the organization whose
behavior is not indicative of the justice and the integrity
often ascribed to transformational leaders (cf. Kuhnert
& Lewis, 1987). Ethical lapses by coworkers, for example, might effectively counteract some of the feelings of
obligation that would otherwise be associated with transformational leadership.

Future Directions
Further study of the causal links among intent to leave,
the facets of organizational commitment, and the MLQ
is obviously needed. Although our research was not designed to allow strong causal inferences, the findings are
at least consistent with the view that transformational
leadership influences AC, which in turn reduces the
probability of leaving. Of course, as noted in other MLQ
studies (e.g., Avolio et al., 1988; Waldman et al, 1990),
the possibilities of alternative, reverse, or reciprocal causation remain.
Efforts to bridge the gap between personnel selection
and leadership theory (cf. Kuhnert & Russell, 1990)
should continue as well. Because the MLQ has been recommended as a possible selection tool (Waldman et al.,
1990), it is worth examining the degree to which the
questionnaire augments more traditional selection methods, including personality inventories (cf. Hogan & Hogan, 1994) and cognitive ability tests. As was the case
with the outcome variables in this study, the chances for
an augmentation effect in the selection realm would ob-

477

viously be enhanced to the degree that the MLQ accurately assesses relatively independent leadership constructs. The discriminant validity of the questionnaire
has implications for leadership training as well, because
there is little point in emphasizing the differences among
the various transformational facets if followers perceive
them as being part of the same leadership domain.
Clearly, strong assumptions about the factorial nature of
the MLQ across forms, populations, or both are
premature.

References
Allen, N. J., & Meyer, J. P. (1990). The measurement and antecedents of affective, continuance, and normative commitment to the organization. Journal of Applied Psychology, 63,
1-18.
Avolio, B. J., Waldman, D. A., & Einstein, W. O. (1988). Transformational leadership in a management game simulation.
Group & Organizational Studies, 13, 59-80.
Bass, B. M. (1985). Leadership and performance beyond expectations. New York: Free Press.
Bass, B. M., & Avolio, B. J. (1990). Transformational leadership development: Manual for the Multifactor Leadership
Questionnaire. Palo Alto, CA: Consulting Psychologists
Press.
Bentler, P. M. (1990). Comparative fix indices in structural
models. Psychological Bulletin, 107, 238-246.
Bentler, P. M., & Bonett, D. G. (1980). Significance tests and
goodness of fit in the analysis of covariance structures. Psychological Bulletin, 88, 588-606.
Bollen, K. A. (1989). Structural equations with latent variables. New York: Wiley.
Burns, J. M. (1978). Leadership. New York: Harper & Row.
Cohen, J., & Cohen, P. (1983). Applied multiple regression/
correlation analysis for the behavioral sciences. Hillsdale, NJ:
Erlbaum.
Hackett, R. D., Bycio, P., & Hausdorf, P. (1994). Further assessments of Meyer and Allen's (1991) three-component
model of organizational commitment. Journal of Applied
Psychology, 79, 15-23.
Hater, J. J., & Bass, B. M. (1988). Superiors' evaluations and
subordinates' perceptions of transformational and transactional leadership. Journal of Applied Psychology, 73, 695702.
Hogan, R., & Hogan, J. (1994). What we know about leadership: Effectiveness and personality. American Psychologist,
49, 493-504.
Howell, J. M., & Avolio, B. J. (1993). Transformational leadership, transactional leadership, locus of control, and support
for innovation: Key predictors of consolidated-business-unit
performance. Journal of Applied Psychology, 78, 891-902.
Hunt, J. G. (1991). Leadership: A new synthesis. Newbury
Park, CA: Sage.
James, L. R., Mulaik, S. A., & Brett, J. M. (1982). Causal analysis: Assumptions, models, and data. Beverly Hills, CA: Sage.
Joreskog, K. G., & Sorbom, D. (1986). LISREL: Analysis of
linear structural relationships by maximum likelihood, in-

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

478

P. BYCIO, R. HACKETT, AND J. ALLEN

strumental variables, and least squares methods (4th ed.)Mooresville, IN: Scientific Software.
Keller, R. T. (1992). Transformational leadership and the performance of research and development project groups.
Journal of Management, 18, 489-501.
Kuhnert, K. W., & Lewis, P. (1987). Transactional and transformational leadership: A constructive/developmental analysis. Academy of Management Review, 12, 648-657.
Kuhnert, K. W., & Russell, C. J. (1990). Using constructive
developmental theory and biodata to bridge the gap between
personnel selection and leadership. Journal of Management,
16, 595-607.
Mathieu, J. E., & Zajac, D. M. (1990). A review and metaanalysis of the antecedents, correlates, and consequences of
organizational commitment. Psychological Bulletin, 108,
171-194.
McGee, G. W., & Ford, R. C. (1987). Two (or more?) dimensions of organizational commitment: Reexamination of the
affective and continuance commitment scales. Journal of Applied Psychology, 72, 638-642.

Medley, F, & LaRochelle, D. R. (1994). Transformational


leadership and job satisfaction. Unpublished manuscript,
University of Florida, College of Nursing.
Meyer, J. R, & Allen, N. J. (1991). A three-component conceptualization of organizational commitment. Human Resource
Management Review, 1, 61-89.
Mulaik, S. A., James, L. R., Van Alstine, J., Bennett, N., Lind,
S., & Stilwell, C. D. (1989). Evaluation of goodness-of-fit indices for structural equation models. Psychological Bulletin,
105, 430-445.
Waldman, D. A., Bass, B. M., & Einstein, W. O. (1987). Leadership and outcomes of performance appraisal processes.
Journal of Occupational Psychology, 60, 177-186.
Waldman, D. A., Bass, B. M., & Yammarino, F. J. (1990). Adding to contingent-reward behavior: The augmenting effect of
charismatic leadership. Group & Organizational Studies, 15,
381-394.

Received April 14, 1994


Revision received December 20, 1994
Accepted December 20, 1994

Das könnte Ihnen auch gefallen