question
nb1234567
3.pdf
EFFECTIVENESS CORRELATES OF TRANSFORMATIONAL AND
TRANSACTIONAL LEADERSHIP: A META-ANALYTIC REVIEW
OF THE MLQ LITERATURE
Kevin B. Lowe* University of North Carolina-Greensboro
K. Calen Kroeck Florida international University
Nagaraj Sivasubramaniam State University of New York at Binghamton
A meta-analysis of the transformational leadership literature using the Mulfiflcfor Leadership
Questionnaire (MLQ) was conducted to (a) integrate the diverse findings, (b) compute an average effect
for different leadership scales, and (c) probe for certain moderators of the leadership style-effectiveness
relationship. Transformational leadership scales of the MLQ were found to be reliable and significantly
predicted work unit effectiveness across the set of studies examined. Moderator variables suggested by the
literature, including level of the leader (high or low), organizational setting (public or private), and
operationalization of the criterion measure (subordinate perceptions or organizational measures of
effectiveness), were empirically tested and found to have differential impacts on correlations between
leader style and effectiveness. The operationalization of the criterion variable emerged as a powerful
moderator. Unanticipated findings for type of organization and level of the leader are explored regarding
the frequency of transformational leader behavior and relationships with effectiveness.
INTRODUCTION
Bums (1978) identified two types of leadership styles, transformational and transactional leadership. The transformational leader construct was suggested by Bums based on a
* Direct all correspondence to: Kevin B. Lowe, Department of Business Adminstration, Joseph M. Bryan School
of Business and Economics, University of North Carolina-Greensboro, Greensboro, NC 27410.
Leadership Quarterly, 7(3), 385-425.
Copyright 0 1996 by JAI Press Inc.
All rights of reproduction in any form reserved.
ISSN: 1048-9843
386 LEADERSHIP QUARTERLY Vol. 7 No. 3 1996
qualitative analysis of the biographies of various political leaders. The notion of a
transformational leadership style as a construct has also been addressed in the works of
several scholars (Bass 1985; Conger & Kanungo, 1987, 1988; House, 1977; Podsakoff, McKenzie, Moorman, & Fetter 1990; Tichy & Devanna, 1986; Trite & Beyer, 1986; Yukl,
1989) with varying degrees of specificity and rigor. The transformational leader has been
characterized as one who articulates a vision of the future that can be shared with peers and subordinates, intellectually stimulates subordinates, and pays high attention to individual
differences among people (Yammarino & Bass, 1990a). This transformational leader was
posited as a contrast to the transactional leader who exchanges valent rewards contingent upon a display of desired behaviors (Burns, 1978; Waldman, Bass, & Einstein, 1987).
Bass (1985), viewing the transformational and transactional leadership constructs as
complementary constructs, developed the Multifactor Leadership Questionnaire (MLQ) to assess the different leadership styles. Even though a substantial research base exists using
the MLQ, relationships among the various components of transformational and
transactional leadership constructs and leader effectiveness, in different settings, is not well understood. A comprehensive review and analysis of the research using the MLQ is
necessary to better understand the nomological connections which summarize the validity evidence for these constructs. The purpose of this paper is to conduct a meta-analytic
review of the literature which uses the MLQ to describe the transformational and
transactional leadership constructs, and to analyze the research in which these constructs
have been empirically linked to leader effectiveness.
LITERATURE REVIEW
Leadership as a Dimensional Construct
In developing the construct, Burns (1978) drew from the literature on traits, leadership
styles, leader-member exchange research, as well as his own observations, and put forth the
idea of a transformational and transactional leadership style. Bums considered the transformational leader to be distinct from the transactional leader, where the latter is
viewed as a leader who initiates contact with subordinates in an effort to exchange something of value, such as rewards for performance, mutual support, or bilateral disclosure. At the other pole of the leadership style dimension, Bums viewed the
transformational leader as one who engages with others in such a way that the leader and the follower raise one another to a higher level of motivation and morality (cf. Kuhnert &
Lewis, 1987), not easily explained in traditional instrumental exchanges. Higher aspirations or goals of the collective group are expected to transcend the individual and result in the achievement of significant change in work unit effectiveness. Bums believed that all managers could be classified by leadership style according to their propensity for
transactions with versus transformation of subordinates.
Leadership as a Complementary Construct
Bass (1985) viewed the transformational/transactional leadership paradigm as being comprised of complementary rather than polar constructs. He integrated the transformational and transactional styles by recognizing that both styles may be linked to
Transformational vs. Transactional Leadership 387
the achievement of desired goals and objectives. In this view, the transformational leadership style is complementary to the transactional style and likely to be ineffective in the total absence of a transactional relationship between leader and subordinate (Bass, Avolio, & Goodheim, 1987). In line with this reasoning, a given manager may be both transformational and transactional. Tosi (1982) noted that supporting every charismatic leader is someone with the ability to manage the mundane, day-to-day events that consume the agendas of many leaders. Transformational leadership thus augments transactional management to achieve higher levels of subordinate performance with the primary difference residing in the process by which the leader motivates subordinates and in the types of goals set. The ability of the transformational leader to obtain performance beyond basic expectations of workers has been labelled the “augmentation hypothesis” (Waldman, Bass, & Yammarino, 1990).
Bass (1985) characterized the transactional leader as one who operates within the existing system or culture, has a preference for risk avoidance, pays attention to time constraints and efficiency, and generally prefers process over substance as a means for maintaining control. The skillful transactional leader is likely to be effective in stable, predictable environments where charting activity against prior performance is the most successful strategy. This leader prototype is consistent with an equitable leader-member exchange relationship where the leader fulfills the needs of followers in exchange for performance meeting basic expectations (Bass, 1985; Graen & Cashman, 1975).
Transformational leaders seek new ways of working, seek opportunities in the face of risk, prefer effective answers to efficient answers, and are less likely to support the status quo. Transformational leaders do not merely react to environmental circumstances-they attempt to shape and create them (Avolio & Bass, 1988). Transformational leaders may use transactional strategies when appropriate, but they also tend to utilize symbolism and imagery to solicit increased effort. The leader accomplishes this by raising the level of
intellectual awareness about the importance of valued outcomes, by raising or expanding individual needs, and by inducing a belief in transcending self-interest for the sake of the
team or organization (Bass, 1985, p. 20). Bass’s conceptualization of the transformational leader extended House’s (1977) idea of
the charismatic leader by incorporating the individualized consideration and intellectual stimulation aspects. Graham (1988) differentiated the ideas of these theorists by concluding that House’s conceptualization of Charisma characterizes followers as dependent on the leader-simply “automotors” responding to the leaders’ charismatic acts. Bass sees followers as those who demonstrate free choice behavior and develop follower autonomy within the overlay of the leader’s vision. Thus, true transformational leadership requires employee empowerment, not employee dependence, according to Bass’ conceptualization. It is also important to note that Avolio and Bass (1988) see these constructs as splitting into two dimensions the aspects of some widely utilized leadership scales (e.g. the Initiating Structure construct from the Ohio State studies). The transactional leader may clarify the task structure with the “right way” to do things in a way that maintains dependence on the leader for preferred problem solutions. The transformational leader on the other hand may provide a new strategy or vision to structure the way to tackle a problem, endowing the subordinate’s sovereignty in problem solving.
Despite the intuitively compelling articulation of transformational leadership in the early stages of conceptualization, little systematic evidence was available to validate the
388 LEADERSHIP QUARTERLY Vol. 7 No. 3 1996
construct. Unlike its complementary construct, transactional leadership, no instrument was
available to test the predictive validity, limitations, and applications of the theory. In the
following section we review the development of the MLQ and the research base that has accumulated using this instrument.
Development of the MLQ
Bass (1985) developed an instrument to measure both transactional and transformational leader behavior and to investigate the nature of the relationship between these leader styles
and work unit effectiveness and satisfaction. The resulting instrument, the Multifactor
Leadership Questionnaire (MLQ), was conceptually developed and empirically validated to reflect the complementary dimensions of transformational and transactional leadership
with sub-scales to further differentiate leader behavior. The initial 142 item pool for the
MLQ was developed by combining a review of the literature with an open-ended survey
asking 70 executives for their descriptions of attributes of transformational and transactional leaders. Factor analysis indicated five scales with acceptable reliabilities. The
final 73 items were factor analyzed again in a later study (Hater & Bass, 1988) with similar
results. The MLQ has since acquired a history of research as the primary quantitative
instrument to measure the transformational leadership construct. Three of the five scales were identified and defined as characteristic of transformational
leadership (Bass 1985; Bass, Avolio, & Goodheim, 1987). They are as follows: Charisma: The leader instills pride, faith and respect, has a gift for seeing what is
really important, and transmits a sense of mission which is effectively articulated.
Individualized Consideration: The leader delegates projects to stimulate learning
experiences, provides coaching and teaching, and treats each follower as a respected
individual. Intellectual Stimulation: The leader arouses followers to think in new ways and
emphasizes problem solving and the use of reasoning before taking action. Two scales were identified and defined as being characteristic of transactional leadership
(Bass, 1985; Bass, Avolio, & Goodheim, 1987): Contingent Reward: The leader provides rewards if followers perform in accordance
with contracts or expend the necessary effort. Management-by-Exception: The leader avoids giving directions if the old ways are
working and allows followers to continue doing their jobs as always if performance goals
are met. The anchors of the leadership style scales incorporated a magnitude-estimation ratio to
each other of 4:3:2: I:0 with “frequently, if not always “ and “not at all” serving as the endpoint anchors (Bass, Cascio, & O’Conner, 1974). The transformational factors of
Charisma, Individualized Consideration, and Intellectual Stimulation have been identified in earlier research as being highly correlated with Charisma accounting for roughly 60% of the variance in the transformational scale (Bass, 1988a). Our own meta-analysis of the intercorrelations (Appendix 1) indicate a high intercorrelation between transformational
scales and between the transformational factors and Contingent Reward. The MLQ has been examined in over 75 research studies, appearing in journals,
dissertations, book chapters, conference papers, and technical reports. The instrument has been used to study leaders in a variety of organizational settings such as manufacturing, the
A ut
ho r(
s)
So ur
ce o
f St
ud y
A vo
lio ,
Y am
-
m ar
in a,
& B
as s
(1 99
1)
A vo
lio ,
W al
dm an
,
& E
in st
ei n
(1 98
8)
Jo ur
na l
of
M af
la ge
m en
t
Fo rt
un e
50 0
hi gh
te
ch
G ro
up
an d
O rg
an iz
at io
ns
St ud
ie s
U ni
ve rs
ity
cl as
sr oo
m
B as
s &
A vo
lio
(1 98
9)
E du
ca tio
na l
an d
Ps yc
ho lo
gi ca
l
M ea
su re
m en
t
V ar
io us
em
pl oy
in g
M B
A
st ud
en ts
B as
s (1
98 %
) Fo
rt un
e 50
0 Fi
rm
B as
s (1
98 5b
) L
ea de
rs hi
p an
d
Pe rf
or m
an ce
B ey
on d
E xp
ec ta
-
tio ns
(B
oo k)
N ew
Z ea
la nd
pr
iv at
e
in du
st ry
B as
s (1
98 %
) N
ew Z
ea la
nd
go ve
rn m
en t
ag en
ci es
B as
s &
Y am
m ar
in o
(1 99
1)
H at
er
& B
as s
(1 98
8)
H ow
el l
& A
vo lio
(1 99
3)
T ab
le 1
A
P ub
lis he
d St
ud ie
s In
cl ud
ed i
n M
et a
A na
ly si
s --
-.- -
__
T yp
e of
O rg
an iz
at io
n L
ev el
o f
Le ad
er
R at
er (
N S
it e)
’ R
es ea
rc h
P ur
po se
(N
S iz
e) ”
s?
J --
3
D ep
en de
nt V
ar ia
bl es
3 s.
A pp
lie d
Ps yc
ho lo
gy :
U .S
. N
av y
A n
In te
rn at
io na
l
R ev
ie w
Jo ur
na l
of A
pp lie
d A
ir d
el iv
er y
se rv
ic e
Ps yc
ho lo
gy
co m
pa ny
Jo ur
na l
of A
pp lie
d C
an ad
ia n
fi na
nc ia
l
Ps yc
ho lo
gy
in st
itu tio
n
D iv
is io
n an
d Su
bo rd
jn at
es
of
de pa
~m en
t he
ad s
(4 7)
m
an ag
er s
(1 41
)
T ea
m
“P re
si de
nt ”
(2 7)
M
B A
st
ud en
ts
(1 63
)
Su pe
rv is
or s
of p
ar t-
Su
bo rd
in at
e M
B A
tim e
M B
A
st ud
en t’
s st
ud en
ts
(8 7)
w or
ki ng
fu
ll- tim
e (8
7)
U pp
er
le ve
l m
an ag
er s
(2 56
)
M an
ag er
s (4
5)
M id
-l ev
el
m an
ag er
s (2
56 )
Pr of
es si
on al
s &
m an
ag er
s( 45
)
C en
tr al
Se ni
or
le ve
l
ad m
in is
tr at
or s
(2 3)
Su rf
ac e
fl ee
t of
fi ce
rs
(l ie
ut en
an t
or l
ie ut
en -
ad m
in is
tr at
or s
(2 3
1 V al
id at
e M
L Q
In ve
st ig
at e
co m
m on
m et
ho ds
pr
ob le
m
R el
at io
ns hi
p be
tw ee
n
le ad
er sh
ip
an d
si m
ul at
io n
ga m
e pe
rf or
m an
ce
H ow
p ro
to ty
pe s,
le
ni en
cy ,
an d
sa tis
fa ct
io n
ef fe
ct
ra tin
gs
V al
id at
e M
L Q
V al
id at
e M
L Q
C om
m is
si on
ed
an d
no nc
om is
-
C on
gr ue
nc e
of s
el f
an d
ot he
r ra
tin gs
an t
ju ni
or
gr ad
e)
(1 55
) s io
ne d
of fi
ce rs
(6
20 ”
Is t,
2n d,
o r
3r d
le ve
l Su
bo rd
in at
es
fo r
T o
va lid
at e
th e
au gm
en ta
.
m an
ag er
s,
m in
im um
(3
12 )
tio n
ef fe
ct
of t
ra ns
fo rm
a-
of o
ne y
ea r
of e
xp er
i-
tio n;
d o
st ud
y w
ith ou
t
en ce
(5
4)
m et
ho d
bi as
T op
f ou
r le
ve ls
of
Su
bo rd
in at
es
(3 22
) R
el at
io ns
hi p
to l
oc us
of
m an
ag em
en t
(i .e
., co
nt ro
l an
d in
no va
tio n;
bu si
ne ss
un
it m
an ag
- fo
cu s
on c
on tin
ge nt
em ,
fi na
nc ia
l m
an ag
- re
w ar
d
er s)
( 78
)
M L
Q
ef fe
ct iv
en es
s 2 ”
M L
Q
sa tis
fa ct
io n
< ? $ Pe
rf or
m an
ce
on f
iv e
fi na
nc ia
l m
va ri
ab le
s 2 i.
L ea
de r
E ff
ec tiv
en es
s M
em be
r 2
Sa tis
fa ct
io n
5 k Su
pe rv
is or
y ap
pr ai
se d
pe rf
or m
an ce
S
M L
Q
ex tr
a ef
fo rt
“J
5’
M
L Q
ef
fe ct
iv en
es s
M L
Q
sa tis
fa ct
io n
M L
Q
ef fe
ct iv
en es
s
M L
Q
sa tis
fa ct
io n
E ar
ly
pr om
ot io
n
C on
~b ut
io n
to m
is si
on
M L
Q
sa tis
fa ct
io n,
M L
Q
ef fe
ct iv
en es
s
Su pe
rv is
or s
ra tin
gs
of o
ff ic
er s
Su pe
rv is
or s
ra tin
gs
of w
or k
gr ou
p
% of
g oa
ls
m et
(C on
tin ue
d)
K el
le r
(1 99
2)
K ir
by ,
K in
g,
&
Pa ra
di se
(1 99
1)
K om
iv es
(1 99
1)
M cD
an ie
l &
W ol
f
(1 99
2)
Se ltz
er
& B
as s
( 1 99
0)
Jo ur
na l
of
M an
ag em
en t
Se ltz
er ,
N um
er of
f.
& B
as s
(1 98
9)
Jo ur
na l
of H
ea lth
&
U
ni ve
rs ity
cl
as sr
oo m
St
ud en
ts
(2 77
)
H um
an
R es
ea rc
h
Si ng
er ,
M S.
Ps
yc ho
lo gi
ca l
(1 98
5)
R ep
or ts
Jo ur
na l
of M
an ag
e-
In du
st ri
aI
R &
D o
rg a-
Pr
oj ec
t m
an ag
er s
Pr of
es si
on al
ni za
tio n
(T im
e 1
= 68
) em
pl oy
ee s
(T im
e 2
= 48
) (T
im e
I =
46 2)
(T im
e 2
= 44
0)
Jo ur
na l
of E
du ca
-
tio na
l R
es ea
rc h
Se x
R ol
es
Jo ur
na l
of N
ur si
ng
A dm
in is
tr at
io n
Sc ho
ol
di st
ri ct
s T
ea ch
er s,
pr
in ci
pa ls
, Su
bo rd
in at
es
ad m
in is
tr at
or s
(5 8)
(N
= 1
30 )
U ni
ve rs
iti es
R
es id
en ce
ha
ll
di re
ct or
s (7
4)
R es
id en
t
as si
st an
ts
(6 05
)
H os
pi ta
l O
ne n
ur se
, C
hi ef
M
id -l
ev el
E xe
cu tiv
e O
ff ic
er s.
ad
m in
is tr
at or
s &
m id
-l ev
el
ad m
in is
tr a-
nu
rs es
(.
46 )
to ts
(9
)
U ni
ve rs
ity
cl as
sr oo
m
M B
A
st ud
en ts
(5
5)
N ew
Z ea
la nd
E
xe cu
- T
op
le ve
l co
m pa
ny
ri ve
s ex
ec ut
iv es
(3
8)
Sp an
gl er
&
B ra
io tta
G
ro up
an
d U
S.
co rp
or at
io ns
B
oa rd
au
di t
(1 99
0)
O rg
an iz
at io
n St
ud ie
s co
m m
itt ee
ch ai
rp er
so ns
(3
2)
W al
dm an
, B
as s.
Jo
ur na
l of
O
ne F
or tu
ne
50 0
m an
-U pp
er
m an
ag em
en t
& E
in st
ei n
O cc
up at
io na
l uf
ac tu
ri ng
fi
rm
(2 56
)
(1 98
7)
Ps yc
ho lo
gy
St ud
en t
te am
m em
be rs
( 13
8)
T es
t of
a u~
nl en
ta tio
n
hy po
th es
is
Pa rt
-t im
e M
B A
s
(2 77
)
R el
at io
ns hi
p to
s tr
es s,
bu rn
ou t,
ef fe
ct iv
en es
s,
ex tr
a ef
fo rt
C om
pa ny
ex ec
ut iv
es
(3 8)
C om
pa re
re
al
an d
id ea
l
le ad
er s
M L
Q
ex tr
a ef
fo rt
M L
Q
ef fe
ct iv
en es
s
M L
Q
sa tis
fa ct
io n
B ur
no ut
: st
re ss
M L
Q
ex tr
a ef
fo rt
M L
Q
ef fe
ct iv
en es
s
M L
Q
sa tis
fa ct
io n
M L
Q
w or
k un
it ef
fe ct
iv en
es s
M L
Q
L ea
de r
ef fe
ct iv
en es
s
IM L
Q s
at is
fa ct
io n
Si x
ite m
e ff
ec tiv
en es
s sc
al e
St ud
y ef
fe ct
of
l ea
de r
st yl
e on
a ud
it co
m m
itt ee
A ud
it co
m m
itt ee
m em
be rs
, C
FO ,
ch ie
f au
di to
r (6
7)
V ar
io us
Sh
ow
ho w
st
yl es
re
la te
to
Pe
rf or
m an
ce
ap pr
ai sa
l ev
al ua
tio n
su pe
rv is
or >
an d
pe rf
or m
an ce
ap
pr ai
sa l
V ar
io us
m
ea su
re s
of s
at is
fa ct
io n
m an
ag er
s (2
56 )
ou tc
om es
R el
at io
ns hi
p be
tw ee
n
tr an
sf or
m at
io na
i an
d
tr an
sa ct
io na
l le
ad er
sh ip
an d
pr oj
ec t
gr ou
p
pe rf
or m
an ce
L ea
de r
ch ar
ac te
ri st
ic s
an d
be ha
vi or
s as
so ci
at ed
w ith
ex
tr a-
or di
na ry
pe
r-
fo m
la nc
e
V al
id at
e M
L Q
T es
t of
t ra
ns fo
rm at
io na
l
le ad
er sh
ip
in n
ur si
ng
Pr oj
ec t
qu al
ity
B ud
ge t
sc he
du le
ef
fi ci
en cy
E ff
ec tiv
en es
s ev
al ua
tio n
M L
Q
ef fe
ct iv
en es
s
M L
Q
ex tr
a ef
fo rt
Sa tis
fa ct
io n
w ith
le
ad er
Sa tis
fa ct
io n
ov er
al l
M L
Q
ef fe
ct iv
en es
s
(C on
tin ue
d)
T ab
le
1A (
C on
ti nu
ed )
A ut
ho r(
s)
W al
dm an
, B
as s
& Y
am m
ar in
o
(1 99
0)
So ur
ce
of S
tu dy
T
yp e
of O
rg an
iz at
io n
L ev
el of
Le ud
er
R at
er
(N S
iz e)
R
es em
-h
P ur
po se
G ro
up
an d
U .S
. N
av al
A
ca de
m y
O ff
ic er
s (1
86 )
Im m
ed ia
te
T es
t of
a ug
m en
ta tio
n
O rg
an iz
at io
n St
ud ie
s su
bo rd
in at
es
(7 93
) hy
po th
es is
Y am
m ar
in o
& B
as s
M ea
su re
s of
N
av y
w ar
c ol
le ge
Su
pe ri
or s
of s
en io
r Se
ni or
of
fi ce
rs
L on
gi tu
di na
l re
la tio
ns hi
p
( 1 99
0a )
L ea
de rs
hi p
(B oo
k)
of fi
ce rs
(3
1 8)
(3
18 )
of s
ty le
s an
d sa
tis fa
ct io
n
Y am
m ar
in o
& B
as s
H um
an
R el
at io
ns
(1 99
0b )
Y am
m ar
in o
&
Pe rs
on ne
l
D ub
in sk
y (1
99 4)
Ps
yc ho
lo gy
U .S
. N
av al
A
ca de
m y
O ff
ic er
s (1
86 )
Su bo
rd in
at es
of
C
on ce
pt ua
l cl
as si
fi ca
tio n
na va
l of
fi ce
rs
of t
ra ns
fo rm
at io
na l
le ad
-
(7 93
) er
sh ip
L ar
ge
m ed
ic al
pr
od -
Sa le
s m
an ag
er s
(3 3)
Sa
le s
pe rs
on ne
l W
ith in
an
d be
tw ee
n an
a-
uc ts
t ir
m
(1 05
) ly
is o
f dy
ad ic
vs
. gr
ou p
ef fe
ct s
of t
ra ns
fo rm
a-
tio na
l le
ad er
sh in
D ep
en de
nt
V ur
ia bl
es
Su bo
rd in
at e
ra te
d ef
fe ct
iv en
es s
Su pe
rv is
or y
pe rf
or m
an ce
ev
al ua
tio n
fr eq
ue nc
y
Pr om
ot io
n re
co m
m en
da tio
n
M L
Q
ex tr
a ef
fo rt
M L
Q
ef fe
ct iv
en es
s
M L
Q
sa tis
fa ct
io n
A ca
de m
ic
pe rf
or m
an ce
M ili
ta ry
pe
rf or
m an
ce
M L
Q
ex tr
a ef
fo rt
M L
Q
ef fe
ct iv
en es
s
M L
Q
sa tis
fa ct
io n
M L
Q
ef fe
ct iv
en es
s
M L
Q
sa tis
fa ct
io n
Sa le
s pe
rf or
m an
ce
ra tin
g (M
D P
R )
N ot
e:
‘L ea
de r
N -s
iz e
us ed
fo
r m
et a-
an al
yt ic
ca
lc ul
at io
ns .
R at
er
N
si ze
pr
ov id
ed
fo r
m fo
rm at
io n
pu rp
os es
on
ly
T ab
le 1
B
U np
ub lis
he d
St ud
ie s
In cl
ud ed
i n
M et
a A
na ly
si s
A ut
ho r(
s)
So ur
ce
c$ St
ud y
T yp
e of
O rg
un iz
ui on
Lt
w l
c? f’ L
eu de
r’
R at
er
iN S
iz e)
’ R
es ea
rc h
P ur
po se
D
ep en
de nt
V
ar iu
hl es
~_
A
tw at
er
&
T ec
ni ca
l re
po rt
U
.S .
N av
y M
id sh
ip m
en
(s qu
ad
Pl eb
es
(1 23
5)
T o
as se
ss
th e
re la
ti% hi
p M
L Q
ef
fe ct
iv en
es s
Y an
m ar
in o
(1 98
9)
le ad
er s)
(1
07 )
be tw
ee n
le ad
er sh
ip
st yl
e M
L Q
sa
tis fa
ct io
n
an d
le ad
er
su cc
es s
M L
Q
ex tr
a ef
fo rt
A vo
lio
& H
ow el
l C
on fe
re nc
e pa
pe r
- C
an ad
ia n
fi na
nc ia
l Se
ni or
E
xe cu
tiv es
(7
6) S
ub or
di na
te s
(2 37
) E
ff ec
ts
of l
ea de
r-
fo llo
w er
M
L Q
sa
tis fa
ct io
n
(1 99
2)
. ce
nt er
fo
r C
re at
iv e
in st
itu tio
n
L ea
de rs
hi p
B ry
an t
(1 99
0)
B ry
ce
(1 98
9)
C ow
en
(1 99
0)
M as
te r’
s th
es is
SO
O +b
ed h
os pi
ta l
N ur
se
m an
ag er
s (7
)
N ur
si ng
(A kr
on
U .)
D is
se rt
at io
n Ja
pa ne
se
co rp
or at
io ns
1s
t. 2n
d L
ev el
(U .S
.I nt
er na
tio na
l U
,)
su pe
rv is
or s:
ex ec
ut iv
es
an d
m an
ag er
s ( 1
32 )
E d.
D
D is
se rt
at io
n U
ni ve
rs ity
Pr
es id
en ts
(1
53 )
(G on
za ga
U
.)
H oo
ve r,
Pe tr
oa ko
, A
m er
ic an
R
es ea
rc h
Pr iv
at e
hi gh
sc
ho ol
s H
ea dm
as te
r of
& S
ch ul
tz
A ss
oc ia
tio n
pr in
ci pa
ls
(4 5)
(1 99
1)
co nf
er en
ce
pr es
en ta
tio n
K in
g Ph
.D .
di ss
er ta
tio n
K -
12 s
ch oo
ls
Pr in
ci pa
ls ,
(1 98
9)
(N ew
O rl
ea ns
) &
A
dm in
is tr
at or
s:
N ur
se s
(4 2)
Su bo
rd in
at es
(3 26
)
C hi
ef
ac ad
em ic
of fi
ce r,
C hi
ef
B us
in es
s of
fi ce
r,
de an
s of
a ca
de m
ic
un its
(8
53 )
Su bo
rd in
at es
of
pr in
ci pa
ls
(1 46
)
Su bo
rd in
at e
ra t-
in gs
K oh
, St
ee rs
, &
A
ca de
m y
of
T er
bo rg
M
an ag
em en
t
(1 99
0)
co nf
er en
ce
co lle
ge
de an
s an
d de
pa rt
m en
t K
-1 2
(1 08
)
ch ai
rs
(2 08
) C
ol le
ge
(1 00
)
Si ng
ap or
e se
co nd
ar y
Pr in
ci pa
ls
(8 9)
T
ea ch
er s
(8 14
)
sc ho
ol s
co ng
ru en
ce
an d
% o
f ga
la s
m et
pe rs
on aI
ity
va ri
ab le
s as
pr ed
ic to
rs
of s
at is
fa ct
io n
an d
pe rf
or m
an ce
R el
at io
ns hi
p of
l ea
de rs
hi p
to t
ur no
ve r
M L
Q e
ff ec
tiv en
es s
M L
Q
sa tis
fa ct
io n
T ur
no ve
r
L ea
de rs
hi p
st yl
es
in
Ja pa
n,
re la
tio ns
hi p
w ith
ef fe
ct iv
en es
s an
d ex
tr a
ef fo
rt
T es
t of
r el
at io
ns hi
p
be tw
ee n
st yl
e an
d
en ro
llm en
t gr
ow th
R el
at io
ns
am on
g
lle ad
er sh
ip
st yl
es
an d
ef fe
ct iv
en ss
. sa
tis fa
ct io
n
T es
t of
t he
a ug
m en
ta tio
n
hy po
th es
is
in t
w o
di ff
er en
t se
tti ng
s
R el
at io
ns
be tw
ee n
le ad
er sh
ip
st yl
e w
ith
co m
m itm
en t.
sa tis
fa ct
io n
an d
ef fe
ct iv
in es
s
E ff
ec tiv
en es
s (r
at ed
by
s ub
or di
na te
)
E ff
ec tiv
en es
s (r
at ed
by
s up
er io
r)
E xt
ra e
ff or
t (r
at ed
by
s ub
or di
na te
1
E xt
ra e
ff or
t (r
at ed
by
s up
er io
r)
E nr
ol lm
en t
gr ow
th
M L
Q
sa tis
fa ct
io n
M L
Q
ef fe
ct iv
en es
s
M L
Q
Ss at
is fa
ct io
n
M L
Q
ef fe
ct iv
en es
s
M L
Q
sa tis
fa ct
io n
7~ o
f st
ud en
ts
ac hi
ev in
g sa
tis fa
ct or
y
sc or
e
St ud
en t
sa tis
fa ct
io n
L la
ng
(1 99
0)
Pe re
ir a
(1 98
6)
Pi lla
i &
M ei
nd el
(1 99
1)
R os
s
(1 99
0)
Sa lte
r
(1 98
9)
Se ltz
er
& B
as s
(1 98
7)
T uc
ke r,
B
as s,
&
D an
ie l
(1 99
2)
Y ou
ng
(1 99
0)
D is
se rt
at io
n
(U .
of M
in ne
so ta
)
In te
rn at
io na
l
C on
gr es
s of
A pp
lie d
Ps yc
ho lo
gy
A ca
de m
y of
M an
ag em
en t
co nf
er en
ce
pr es
en ta
tio n
D is
se rt
at io
n
(G eo
rg e
W as
hi ng
to n
U .)
M as
te r’
s th
es is
M
ar in
e he
lic op
te r
N av
y po
st gr
ad ua
te
sq ua
dr on
s
A m
er ic
an
Ps yc
ho lo
gi ca
l
A ss
oc ia
tio n
co nf
er en
ce
C en
te r
fo r
C re
at i
L ea
de rs
hi p
co nf
er en
ce
pr es
en ta
tio n
D is
se rt
at io
n
(B ow
lin g
G re
en
St at
e U
.)
ve
U .S
. na
va l
ac ad
em y
U .S
. na
vy
of fi
ce rs
Pr iv
at e
in du
st ry
or ga
ni za
tio n
in I
nd ia
(2 82
) Fu
nc tio
na l
m an
ag er
s Su
bo rd
in at
es
of
(5 1)
m
an ag
er s
(2 55
)
U ni
ve rs
ity
cl as
sr oo
m
U .S
. ai
r fo
rc e
ac ad
em y
U ni
ve rs
ity
cl as
sr oo
m
U ni
ve rs
ity
U ni
ve rs
ity
Su bo
rd in
at es
(2
82 )
T es
t of
a ug
m en
ta tio
n
St ud
en t
gr ou
p le
ad er
s St
ud en
t gr
ou p
(1 6)
m
em be
r (8
0)
C om
m is
io ne
d O
ff ic
er s
C ad
et
(4 0)
su
bo rd
in at
es
(4 40
0)
C ol
on el
s,
de pa
rt m
en t
Su bo
rd in
at es
he ad
s,
di vi
si on
(1
21 )
of fi
ce rs
(5
6)
V ar
io us
su
pe rv
is or
s Su
bo rd
in at
e
of t
he M
B A
(2
94 )
M B
A s
(2 94
)
M em
be rs
of
d om
in an
t Su
bo rd
in at
es
co al
iti on
(c
ha nc
el lo
r, (1
06 )
de an
s,
de pa
rt m
en t
ch ai
rs ,
an d
se le
ct ed
fa cu
lty )
(1 06
)
D ea
ns
(1 07
) D
ep ar
tm en
t
ch ai
rs
(3 83
)
hy po
th es
is
R ep
lic at
io n
of B
as s
(1 98
5b )
in a
de ve
lo pi
ng
co un
tr y
L ab
ex
pe ri
m en
t;
pe rc
ei ve
d ef
fe ct
iv en
es s
of t
ra ns
fo rm
at io
na l
le ad
-
er sh
ip
in a
c ri
si s
E xa
m in
at io
n of
p er
so na
l-
ity a
ttr ib
ut es
an
d th
ei r
ef fe
ct
on p
er fo
rm an
ce
an d
sa tis
fa ct
io n
of t
ra ns
-
fo rm
at io
na l
m ili
ta ry
le ad
er s
H ow
st
yl es
re
la te
to
L ea
de r
ef fe
ct iv
en es
s
U ni
t ef
fe ct
iv en
es s
U pw
ar d
re la
tio ns
Jo b
pe rf
or m
an ce
ef
fe ct
iv en
es s
O rg
an iz
at io
na l
ef fe
ct iv
en es
s
Pe rc
ei ve
d ef
fe ct
iv en
ss
of l
ea de
r
(s ub
or di
na te
ra
tin g)
M ea
n to
ta l
G PA
; m
ea n
G Pa
; m
ea n
to ta
l gr
ad e
po in
t (a
ca de
m ic
on
ly );
in te
rc ol
le ga
ite ,
in tr
am ur
al ,
ae ro
bi c
sc or
es ;
cl ub
su
pp or
t; m
ar ch
in g
sc or
es ;
pr of
es si
on al
co
m pe
te nc
y;
sa tis
fa ct
io n
w ith
sq
ua dr
on s;
et
c.
M L
Q
ef fe
ct iv
en es
s
sa tis
fa ct
io n,
ef
fe ct
iv en
es s,
M
L Q
sa
tis fa
ct io
n
gr ou
p co
he si
on ,
un it
M L
Q
ex tr
a ef
fo rt
m or
al e,
ex
tr a
ef fo
rt
U ni
t m
or al
e an
d co
he si
on
R ep
lic at
io n
of t
he
M L
Q
ef fe
ct iv
en es
s
tr an
sf or
m at
io na
l au
gm en
- M
L Q
sa
tis fa
ct io
n
ta tio
n ar
gu m
en t
M L
Q
ex tr
a ef
fo rt
T es
t re
la tio
ns hi
ps
am on
g
tr an
sf or
m at
io na
l le
ad er
-
sh ip
a nd
sa
tis fa
ct io
n,
ef fe
ct iv
en es
s an
d ex
tr a
ef fo
rt
T es
t of
r el
at io
ns hi
p
be tw
ee n
le ad
er sh
ip
st yl
e
an d
ef fe
ct iv
en es
s an
d
sa tis
fa ct
io n
M L
Q
ef fe
ct iv
en es
s
M L
Q
sa tis
fa ct
io n
M L
Q
ex tr
a ef
fo rt
M L
Q
ef fe
ct iv
en es
s
M L
Q
sa tis
fa ct
io n
M L
Q
ex tr
a ef
fo rt
N ot
e:
‘L ea
de r
N -s
iz e
us ed
f or
m et
a- an
al yt
ic
ca lc
ul at
io ns
. K
at es
N s
ir e
pr ov
id ed
fo
r in
fo rm
at io
n pu
rp os
es
on ly
394 LEADERSHIP QUARTERLY Vol. 7 No. 3 7996
military, educational and religious institutions, and at various levels in the organization including first line supervisors, middle managers, and senior managers. MLQ scales have been related to a range of effectiveness criteria such as subordinate perceptions of effectiveness, as well as to a variety of organizational measures of performance such as supervisory ratings, number of promotion recommendations, military performance grades and such objective measures as percent of goals met, pass rate on educational competency exams, and financial perfo~ance of the work unit. Studies using subordinate measures of effectiveness as a criterion have occasionally been criticized on the basis of mono-method bias because they typically utilize the effectiveness measure embedded in the MLQ (Avolio, Yammarino, & Bass, 1991; Bass & Avolio, 1989). The effect of using subordinate perceptions of effectiveness rather than organizational measures is a point we shall return to later in the paper.
Research Using the MLQ
Bass (1985) stated that environmental and organizational characteristics are likely to impact the degree to which transformational leadership results in work-unit and organizational effectiveness. Table la and 1 b provide a listing of studies which have measured the association between the MLQ scales and leader effectiveness. The tables list all of the dependent variables that were measured in a given study (e.g., effectiveness, satisfaction, extra effort, upward relations) to acknowledge that the MLQ has been utilized to measure the relationship between leader behaviors and several different desired outcomes. For purposes of this study, otzly the measure o~~~ecti~~e~e.~.~ was coded and included in the meta-analysis, thus the listing of other dependent variables is for exposition
purposes only. The research literature in Tables la and lb is segregated by published (la) and unpublished (lb) studies. The tables show that the MLQ has been used in a variety of organization types, in several countries, and for leaders with both high and low levels of responsibility. The MLQ has been administered in both private fimrs {cf. Avolio, Yammarino, & Bass, 199 1; Bryce. 1989; Keller, 1992) and public organizations (cf. Waldman, Bass, & Yammarino, 1990; Cowen, 1990; Koh, Steers, & Terborg, 1990). The level of the leader studied has been high in some research studies (e.g., Bass, 1985; Avolio & Howell, 1992; Salter, 1989; Young, 1990) and low in other research studies (Komives, 1991; Ross, 1990; Bryant, 1990).
Review of Past Research Findings
Research findings using the MLQ have generally reported statistically significant relationships between leader effectiveness and the transfo~ational scales of Chor~sm~, r~d~vidual~zed consideration, and ~~te~~ectua~ ~tirn~lat~oF~. The transactional scale Corttingent Reward has also been associated with effectiveness though the magnitude of the association is less than that evidenced by the transformational scales. The transactional scale Management-by-Exception generally has low correlations with effectiveness and is often negative when significant.
While there appears to be some consistency in the direction and significance of the scale associations, the strength of the association is less consistent. The scale Charisma correlates r = .91 with effectiveness in a mihtary study by Atwater and Yamm~no (1989) which used the M;LQ measure of effectiveness. Hater and Bass (1988) report a correlation
Transformational vs. Transactional Leadership 395
of r = .36 using the same measure in a Fortune 500 organization. Bass (1985) reports an even lower r = .19 in a Fortune 500 organization when effectiveness is measured by performance appraisal instruments.
Bass and Avolio (1989) report a correlation of r = .77 between Individualized Consideration and effectiveness in a sample of MBA students employed full-time, while Bass and Yammarino (1991) report a correlation of r = .21 in a military sample for this scale when effectiveness was operationalized as supervisory ratings of contribution to the mission. Zntellectual Stimulation correlates r = .74 with effectiveness in a sample of resident assistants rating hall directors (Komives, 1991), while Spangler and Braiotta (1990) found a correlation of r = .25 between the leadership style of board audit committee chairmen and effectiveness.
Singer (1985), in a study of New Zealand executives, found a correlation of r = .71 between Contingent Reward and effectiveness, while Waldman, Bass, and Einstein (1987) report a correlation of zero for the same variables in a sample of U.S. managers. Kirby, King, and Paradise (1991) report a correlation of r = .17 between Management-by- Exception and effectiveness, whereas Bass (1985) reports -0.34 for the same MLQ sub- scale; both studies were conducted in educational settings using equivalent measures of effectiveness.
As evidenced by the above, the array of settings, sample characteristics, and methods for operationalizing measures of effectiveness has yielded a wide range of observed validity coefficients between the five scales of the MLQ and various measures of effectiveness. Though the results provide very general support for the construct of transformational leadership, much less is known about how the separate scales of the MLQ relate to effectiveness, how the choice of criterion measures might affect research outcomes, or how sample characteristics such as organization type or level of the leader may moderate the relationship between leadership style and effectiveness. Sufficient research using the MLQ has accumulated to warrant an integrative study which systematically combines these findings from a variety of research settings as a basis for discussing known relationships among components of leadership style and outcome variables of interest.
The objective of this study is test for overall effects and moderators of the relationship between leader behaviors and effectiveness using meta-analytic techniques. In the present study, a meta-analytic technique provided by Hunter and Schmidt (1990) was used: (a) to compute an average effect size across studies, (b) to compute an average effect size corrected for sampling error and measurement attenuation, and (c) to probe for certain moderators of the leadership style-effectiveness relationship.
HYPOTHESES
Specific hypotheses regarding prevalence and effectiveness of transformational and transactional leadership behaviors are described below. Hypotheses addressing the prevalence of behavior (e.g. what do leaders do) are tested by examining differences in mean frequency ratings of leader behavior as measured by the MLQ scale. Hypotheses addressing the effectiveness of leader behavior (e.g. how does what they do relate to performance) are tested by examining differences in the relationship between the mean frequency rating of leader behavior and a measure of effectiveness. It is expected that leader behavior and effectiveness are influenced by factors such as type of organization,
396 LEADERSHIP QUARTERLY Vol. 7 No. 3 1996
level of the leader in the organization, and the type of criterion used to measure effectiveness.
Type of Organization
Bass (1985) adopted the mechanistic/organic classification scheme of Bums and Stalker (1961) and argued that transformational leadership is more likely to emerge in
organizations with less constrictive (organic) environments. Similarly, he contended that transformational leadership is less likely to emerge in more constrictive (mechanistic) organizations. For example, American public universities, as state agencies, are “enmeshed with union rules and contracts, as well as departmental and faculty norms and traditions. Changes often occur mainly as a consequence of political tradeoffs among powerful coalitions.. . much leadership is substituted for by organizational processes such as committee reviews, collegial decision making, and tenure regulations” (Bass, 1985, p. 160). In such an environment, transformational leadership behaviors are less likely to emerge. Managers in mechanistic organizations are likely to engage in more transactional behaviors than their counterparts in private organizations. In contrast, private organizations, with their market-focus, may have much less institutionalized substitutes for leadership, and managers may engage in transformational behaviors to motivate their
subordinates to achieve “performance beyond expectations.” It is also expected for similar reasons that transformational leadership behaviors are
likely to be more effective in private organizations than in public organizations. To the extent that transformational leadership is more likely to emerge in private firms as compared to public organizations, it follows that one of the reasons that this occurs is
because such transformation of workers is more important in private firms due to the greater impact on work unit effectiveness. To the extent that individual behavior is less prescribed, formalized and defined in private firms as compared to public organizations, there should be greater opportunity for realizing performance beyond expectations by
subordinates resulting from the transformational activity of leaders in such firms. In accordance with these expectations, the following hypotheses were developed:
Hypothesis 1. Transformational leadership is more frequent in private organizations than in public organizations. In contrast, transactional leadership behaviors are more frequent in public organizations than in private
organizations.
Hypothesis 2. The relationship between effectiveness and both transformational and transactional leadership is moderated by the type of organization (public or private). The relationship between transformational leadership behavior and effectiveness will be stronger in private organizations than in public organizations.
level of the Leader
The literature creates a certain ambiguity as to whether transformational leadership has the same impact on effectiveness across all levels of the organizational hierarchy. The issue of the level of the leader moderating the leadership style-effectiveness relationship is,
Transformational vs. Transactional Leadership 397
hence, unclear. Sashkin and Fulmer (1988) asserted that a great deal of the leadership
literature has neglected the organizational perspective of leadership in favor of what they
called “supervisory management”-the role of the leader at middle and lower levels in the
organization. Etzioni (1961) believed that charismatic leadership would be most useful in
situations in which enactors were charged with concern for ends more than means. He
viewed middle managers as being more concerned with means and lower level line
managers as being instrumental performers. Avolio and Bass (1988) found
transformational leadership being practiced at all levels of the organization but observed to
a greater extent at the higher levels. Tichy and Uhich (1984) agree that transformational
leadership is likely to be more prevalent at higher organization levels. Hence, it is expected
that transformational leadership will be more prevalent at higher levels of the organization
than at the lower levels. In contrast, lower level managers are likely to be more
transactional in their leadership style than upper level managers. It is also expected that the relationship between leader behaviors and effectiveness will
be higher at upper levels of management and weaker at lower levels of management. It
follows that, if transformational leadership is more prevalent at higher levels of the
organization, it is because such behavior is more important and has a greater value due to
its impact on effectiveness at the higher cleave of the hierarchy. In accordance with these
assumptions, it could be expected that:
Hypothesis 3. Transformational leadership is more prevalent at upper levels of
management than at lower levels.
Hypothesis 4. The relationship between effectiveness and both transformational
and transactional leadership is moderated by the level of the leader (high vs.
low) in the organization. The relationship between transformational leadership
and effectiveness will be stronger for high level leaders than for low level
leaders.
Type of Criterion
Valid and accurate measures of human behavior and performance are a recurring
problem in the social sciences. The result is that several different measurement methods of
the criterion may be posited by different scientists in quest of the same goal. The MLQ-
based literature is no exception. The dichotomy suggested here is to consider as one group
those studies which use the effectiveness criterion measure embedded in the MLQ. Thus,
in this first group, the rating of leadership style and the rating of leader effectiveness are
completed by the same source, the subordinate, at a common point in time. We have
labelled these subordinate perceptions of leader eflectiveness. Such measures have been
criticized for creating common method bias and, as such, may have a strong impact on
findings regarding the leadership style-effectiveness relationship (Binning, Zaba, &
Whattam, 1986). The second grouping of criterion measures are those which are not subordinate-
determined, but rather are quasi-institutional measures of the leader’s effectiveness. Such
measures include both hard measures (e.g., profit or percent of goals met) and soft
398 LEADERSHIP QUARTERLY Vol. 7 No. 3 1996
measures such as supervisory performance appraisals. We have labelled this category of
criteria organizational measures of leader effectiveness. Organizational measures, while perhaps reducing the common method bias problem
(Avolio, Yammarino & Bass, 1991) may not be especially valid measures of the effectiveness of the transformational characteristics of the leader, as they are often designed
to capture primarily transactional outcomes. Avolio and Bass (1988) noted that many
performance appraisal systems do not emphasize performance beyond expectations and thus the relationship with transformational leadership may be attenuated (p. 47). This may be true
for many of the measures of what are here called organizational measures, the result being
a downward bias on observed validity coefficients for this category of effectiveness criterion. In this sense, subordinate perceptions of effectiveness may be a better indicator
of the impact of transformational leadership on motivation and performance. lf, however, transformational scales are more strongly related to organizational
measures than transactional scales, this is strong evidence that the effect of transformational leadership is much more than simple affective impressions about the
leader’s effectiveness which cannot be distinguished from the other scales of the MLQ. Comparison of the relationships found with subordinate perceptions of effectiveness and
organizational measures of effectiveness also can reveal the possible magnitude of
common method bias. Such comparisons may also provide a more accurate estimate of the true impact of transformational leadership on work unit effectiveness. In either case, there
are compelling reasons for studying these different types of effectiveness criteria
separately to ascertain the relationships they hold with measures of transactional and transformational leadership. On this basis the following hypothesis was developed:
Hypothesis 5. The relationship between transformational/transactional leadership scales and effectiveness will be moderated by the type of criterion (subordinate perceptions vs. organizational measures) used to measure effectiveness. The
relationships will be stronger for subordinate perceptions of effectiveness than
for organizational measures of effectiveness.
METHOD
Issues to Consider in Meta-Analysis
The central purpose of meta-analysis is to analyze statistically the results of many
empirical studies in order to reveal a set of summary findings. There are numerous methods
available for meta-analysis, and the most appropriate method for a particular study is determined by the characteristics of the included studies and the purposes of the researcher (Bangert-Drowns, 1986). The Hunter and Schmidt (1990) method was selected for this study as it best accomplishes the study objectives outlined earlier, is one of the more rigorous methods (Bangert-Drowns, 1986) and is the mostly widely used method for conducting meta-analysis in the Leadership/Organizational Behavior literature.
While meta-analysis corrects some of the problems with traditional reviews (e.g., too
much emphasis on favored studies), it creates a need for precision and consistency in the quantification of the empirical findings. Not only must the coding be precise and consistent. but the researcher must also have a sense of what variables should be coded.
Transformational vs. Transactional Leadership 399
Clear coding definitions, consensus among raters on what needs to be coded, and reliability
checks (to ensure coder consistency) are necessary to conduct a proper analytic review
(Schmitt & Klimoski, 1991). The development of a literature-driven list of variables of interest, a detailed and rigidly
adhered to coding format, and a high degree of inter-coder reliability do not reduce meta-
analysis to an exact science. Wanous, Sullivan and Malinak (1989) highlighted the role of
researcher judgment calls in meta-analysis and their combined effects on results. One such
judgment call is whether to include unpublished studies (e.g., doctoral dissertations,
technical reports) as part of the literature search. Research has tentatively concluded that
few differences exist between published and unpublished studies in various meta-analyses
(Hunter & Schmidt, 1990); however, the issue should always be explored for any particular
segment of the literature, and we therefore revisit the issue in the current study.
Selection of Studies
Five criteria were used for inclusion of studies in the meta-analysis. First, the study must
have used the iK!,Q to measure leadership style from the perspective of the subordinate.
Second, the study must have reported a measure of leader effectiveness. Third, the sample
size must have been reported. Fourth, a Pearson correlation coefficient (or some other type
of test statistic that could be converted into a correlation) between leadership style and
effectiveness must have been reported. Fifth, the leader rated must have been a direct
leader of the subordinate (not an idealized or hypothetical leader). Studies were located using a variety of methods including (a) computer searches of ABZ-
INFORM, PSYCHLIT, and ERIC; (b) manual searches of Social Science Citation Index,
Academy of Management Proceedings, Academy of Management meeting programs; (c)
reference lists of published and unpublished sources; and (d) correspondence with the
Center for Leadership Studies at the State University of New York at Binghamton. The
literature search yielded over 75 studies, of which 39 met the five criteria for inclusion.
Twenty-two of these studies were published in journals and books and yielded 29 usable
validity coefficients (five studies reported both subordinate perceptions and organizational
measures of effectiveness) for at least one scale of the MLQ. Seventeen unpublished
studies (dissertations, conference proceedings, working papers) yielded 23 usable validity
coefficients for at least one scale of the MLQ.
Unpublished Studies
Inclusion of unpublished studies is a subject that in the past has received considerable
debate. Inclusion of unpublished studies minimizes the “file drawer problem” (Wanous, Poland, Premack, & Davis, 1992; Rosenthal, 1979). The file drawer problem refers to an
alleged phenomenon that studies with significant findings are more likely to get published,
while those that have smaller effect sizes are destined for the file drawer of the researcher. Hunter and Schmidt (1990) suggest that there are two issues here, one of availability bias
and one of source bias. Availability bias is likely to occur because the average researcher
has greater access to published than unpublished sources. Source bias is likely to occur if journals are predisposed to publishing significant results or favoring certain methodology
types.
400 LEADERSHIP QUARTERLY Vol. 7 No. 3 1996
McNemar (1960) suggested that the inclusion of unpublished studies was likely to result
in lower validity coefficients, since studies that have been published are more likely to
have higher effect sizes and are more often statistically significant. A second argument is
that higher validity coefficients are found in published sources as a result of more rigorous
methodological practices. This latter point is of course an empirical question and one that
has been consistently resolved in favor of the conclusion that no practical or statistical
difference exists in the average validities between published and unpublished studies
(Glass, McGaw, & Smith, 198 1; Gaugler, Rosenthal, Thornton, & Bentson, 1987; Hunter
& Schmidt, 1990; Rosenthal, 1984; Wanous, et al., 1992). Hunter and Schmidt (1990) note
that if there were true differences in methodological quality, then a meta-analysis that
corrects for methodological weaknesses (i.e., measurement error) would correct for these
differences. Corrections for correlations suggested by Hunter and Schmidt’s (1990) were
made for correlations in this study, and both corrected and uncorrected correlations were
analyzed according to their recommended procedure. Given the relatively small absolute differences due to source in aggregated correlations,
the wide range in individual correlations for both published and unpublished studies, a
finding of significant differences by source for only one of the five MLQ scales, and
theoretical reasons for believing that moderators were likely to exist, published and
unpublished studies were combined for purposes of this meta-analytic review.
Coding of Information
After the studies meeting the five criteria were accumulated from the literature search,
they were reviewed, and relevant data were extracted and coded. The studies were coded
twice by two third-year doctoral students. Prior to any discussion between coders, intra-
coder (over the two codings) and inter-coder agreement for the various information taken
from the studies exceeded 95% for all study variables. The few cases where disagreement
did exist were a result of clerical error, except in one study for which different measures of
the criterion variable were selected by the coders as the most appropriate measure of
effectiveness. Subsequent discussion led to correction of the clerical errors and agreement
on the appropriate criterion variable. For studies in which multiple measures of effectiveness were reported, the coding policy
was to identify the most appropriate measure of effectiveness from the perspective of the
subordinate and the most appropriate measure of effectiveness from the perspective of the
organization. For those research studies which reported both a measure of effectiveness
from the perspective of the subordinate and an organizational measure of effectiveness,
average validity coefficients were computed and used for these studies in subsequent
analyses ( cf. Gaugler et al., 1987; Hunter & Schmidt, 1990) except when testing the
criterion hypotheses directly. Sample size was recorded as the number of observations used to compute the correlation
coefficient. Most studies used average leadership scores for the leaders and thus a typical
sample would be “793 subordinates rated 186 leaders on.. . an average subordinate score was taken for each leader.” Therefore, the sample size used in the present meta-analysis is
more representative of the number of leaders than the number of subordinates completing
the MLQ.
Transformational vs. Transactional Leadership 401
Analysis
To test hypotheses concerning the Type of Organization (Hypotheses 1 and 2), study samples were classified as being from either a Public or a Private organizational setting. Studies were coded as Private if the sample description clearly identified the entity as a
private enterprise (e.g., “‘managers at a Fortune 500 company,” “headmasters of secondary private schools”). Studies were classified as Public when the sample description clearly
specified a public sector organization (e.g., “Naval fleet officers,” “department chairs in large state funded institutions”). Four studies in which the composition of the sample
represented a mix of organizations (e.g., “MBA students working full-time”) were removed
from the Type of Organization analyses (Bass & Avolio, 1989; Seltzer & Bass, 1990; Seltzer, Numeroff, & Bass, 1989; and Pillai, 1991). Hypothesis 1 predicted the existence of a significant mean difference in leadership behavior between public vs. private sector
organizations (diference in what Zeuders do). For this analysis, overall means and standard deviations for each subgroup for each of the five scales were computed by weighting the
sample means and variances by the sample size. Since each of the studies provided an
independent sample, this ag~egation method yielded an unbiased estimate of the population standard deviation. Comparisons of the two means (Public vs Private) were
conducted for each MLQ scale by computing a z-test statistic for the difference between two independent sample means (Neter, Wasserman, & Kutner, 1985, pp. 13-14). Hypothesis 2 predicted higher correlations in Private Organizations than in Public Organizations for the
association between transformational leadership behavior and effectiveness (diSference in how does what they do relate tope~-$ormance). To test this moderator effect, a critical ratio z-test statistic was computed for each MLQ scale to assess the significance of differences between mean effect sizes (Hunter & Schmidt, 1990, p. 438).
To test hypotheses concerning the Level of the Leader (Hypotheses 3 and 4), study samples were classified as High Level Leader if the sample description clearly identified
the leader as upper-middle or top management (e.g., Colonels, board members, division
heads). Studies were classified as Low Level Leader if the sample description was clearly low to middle level supervisory (nursing supervisor, lieutenant junior grade, hall directors). Six studies in which the composition of the sample likely represented a mix of leaders were removed from the Level of Leader analyses (Bass, 1985 (two studies); Bass & Avolio,
1989; Seltzer & Bass, 1990; Seltzer, Numeroff & Bass, 1989; Seltzer & Bass, 1987). Hypothesis 3 predicted the existence of a significant mean difference in leadership
behavior between High vs. Low Level Leaders. To test this difference, a z-test of independent sample means as described above for Hypothesis 1 was conducted for each
scale of the MLQ. Hypothesis 4 predicted higher correlations between transformational behavior and effectiveness for High Level Leaders as compared to Low Level Leaders. In
other words, it was expected that across studies, Level of the Leader would moderate the impact of ~~sfo~ation~ leadership on work unit effectiveness. To test this m~erator effect, a critical ratio z-test statistic was computed to assess the significance of differences between mean effect sizes for each scale of the MLQ in much the same way as the test described for Hypothesis 2.
To test Hypothesis 5, the Type of Criterion moderator analysis, study results were classified according to how the work unit effectiveness variable was operationalized. When the effectiveness criterion measure was that which is embedded in the MLQ (i.e.,
402 LEADERSHIP QUARTERLY Vol. 7 No. 3 1996
when both leader description and evaluation of effectiveness was provided by the same source, the subordinate), we categorized the criterion as subordinate perceptions of leader
effectiveness. The second category of criterion measures were those which were not subordinate determined, but rather were quasi-institutional measures of the leader’s
effectiveness. Such measures included both hard measures (e.g., profit or percentage of goals met) and soft measures such as supervisory performance appraisals. We have
labelled this category of criteria organizational measures of leader eflectiveness. Studies which included both types of criteria were included in the Type of Criterion analysis for each type of criterion if all other rules for inclusion were met.
Fourteen studies yielding 15 coefficients utilized Organizational Measures of effectiveness. Nine were from published studies (Bass, 1985; Bass & Yammarino, 1991;
Hater & Bass, 1988; Howell & Avolio, 1993; Keller, 1992; Waldman, Bass, & Einstein, 1987; Waldm~, Bass, & Yammarino, 1990; Yammarino & Bass, 1990a, Yamma~no & Dubinsky, f994), and five were unpublished studies (Atwater & Yammarino, 1989; Avolio
& Howell, 1992; Koh, Steers & Terborg, 199 1; Ross, 1990; Salter, 1989). More detailed information regarding the different types of criterion measures used in each of these studies is provided in Tables la and I b. Further tests for moderators within the Organizational Measures classification was not possible due to the dissimilarity among the various
organizational measures used and the small number of studies that would be available for inclusion in a more refined categorization. While we could have separated studies by judgmental versus objective measures of effectiveness, this distinction is fuzzy and further
misconstrues the nature of the effectiveness construct, as it lumps together subordinate ML& ratings of effectiveness with supervisory judgments that are garnered for institutional
purposes. We therefore opted to distinguish subordinate perceptions from organizational
measures of effectiveness used for institutional purposes, as this was deemed the simplest, least contaminated moderator variable abstraction,
Meia-Analysis Procedure
Correlations were first subjected to a meta-analysis to eliminate the effects of sampling
error. Reliability data was reported for Charisma and Intellectual Stimulation in 23 studies,
for Individualized Consideration and Contingent Reward in 22 studies, and for Management-by-Exception in 21 studies. Since reliability data was not reported for every
study on every scale, correction for attenuation due to unreliability was performed across
all studies using an artifact distribution technique provided by Hunter and Schmidt (1990). Studies were weighted for sample size and corrected for measurement error using the
procedures suggested by Hunter and Schmidt ( 1990). Credibility and confidence intervals were constructed around the mean effect sizes to
make inferences about the extent to which moderators may be present and to determine the accuracy and significance of the meta-analytic estimates of mean effect size (a thorough review is found in Whitener, 1990). Credibility intervals provide information about the degree to which validity findings can be generalized. Based on the corrected standard deviation, credibility intervals also indicate the likely existence of moderators that can explain differences among studies, indicating that the population of studies should be grouped into sub-populations. If the interval is large (such that meaningful estimation is precluded) or includes zero, the mean corrected effect size may represent the mean of
Transformational vs. Transactional Leadership 403
several different sub-populations of studies. If the interval is smalf or does not include zero, the mean corrected effect size is assumed to estimate one ~pulation parameter, with no evidence of moderators. Credibility intervals indicate the possible existence of moderators but do not identify any specific moderator variable. If moderators can be identified using theory, studies can be grouped into sub-populations that can again be analyzed with credibility intervals to determine if further sub-grouping is required, or even possible, given an adequate number of studies (Arnold, 1982; Whitener, 1990).
Confidence intervals are generated using the standard error of the mean effect size to assess the accuracy of the estimate of the mean effect size. The standard error used to
generate the confidence interval is based on conclusions reached in creditability interval analysis (homogeneity or heterogeneity of sample). Some have criticized the use of credibility intervals in meta-analysis on grounds of low power when differences between population correlations are small (Sagie & Koswolsky, 1993). However, the problem of low power for detecting moderators when correlation differences are small is not unique to credibility intervals but a problem common to a number of decision rule techniques tested by Sagie and Koslowsky (1993). The use of credibility intervals results in lower
chances of a Type I error at the expense of reduced power. Finally, in accordance with the method provided by Hunter and Schmidt (1990, p. 438), one-tailed tests for differences in mean effect sizes were also conducted to test the different moderator hypotheses.
RESULTS
Figure 1 provides a graphic representation of the correlational findings between MLQ scales and Ieader effectiveness separately for published and unpublished studies. As can be seen, the pattern is highly similar, although a difference between the two sources of studies
was found for the InteEEectuaE Stimulation scale. Because the pattern of correlations was so
c 0
r 0.6 r e I
0.4 a t i o 0.2 n
0 MBE CR IS XC CH
MLQ Scales
Figure 1. The Moderating Effect of Published vs. Unpublished studies on the Relationship between MLQ Scales and Leader Effectiveness
T ab
le 2
O
ve ra
ll ze
ta -A
na ly
si s
P ub
lis he
d an
d U
np ub
lis he
d St
ud ie
s C
om bi
ne d
T ot
al
S am
pl e
C rc
~ ~
~ ~
,, cI
M
ea n
S ca
le
M om
S
ca lu
C
or ~
e~ {t
io rr
R
an ge
of
C
or re
la tim
M
ea n
R m
M
ea n
9. 5
5%
95 %
S iz
e S
co re
S
D
C or
re la
tio n
C hr
re ct
ed
C on
jd en
ce
C re
di bi
lit y
f o r Sr al
e C
0r jjk
irn t.s
C
0e flc
ie nt
.s
C or
re la
tio n
In te
r&
In te
rv al
C ha
ri sm
a
In di
vi du
al iz
ed
C on
si de
ra tio
n
In te
lle ct
ua l
St im
ul at
io n
C on
tin ge
nt
R ew
ar d
M an
ag em
en t-
by -
E xc
ep tio
n
6. 48
5 .9
2 2.
52
1. 04
47
.I
0 to
. 91
,6
23
.7 13
.6
9 to
. I4
.2
3 to
1 .0
0
6. 23
2 .8
8 2.
50
0. 99
41
-.o
i to
. 84
,5
28
‘6 1.
5 .5
9 to
. 64
.1
6 to
1 .0
0
6, 36
0 .X
6 2.
48
0. 85
45
.0
8 to
. 89
,5
12
,6 02
,1
58 to
. 63
.I
2 to
1 .0
0
7, 16
3 .8
2 1.
83
0. 90
43
-.2
5 to
, 72
.3
38
,4 08
.3
x to
. 43
-.
I2
to .
94
6. 94
8 .6
5 2.
32
0. 74
41
-.s
4 to
. 47
.0
40
,0 54
.0
4 to
. 07
-.3
2 to
. 43
Transformational vs. Transactional Leadership 405
similar and because no significant difference was found for four of the five MLQ scales, the
samples were combined as is common practice. Results of the overall meta-analysis are provided in Table 2. Measurement of leadership
styles displayed sufficient internal consistency reliability for four of the five scales with
only Management-by-Exception below the normally accepted value of .70 (Nunnally,
1978, p. 245). The transformational scales were similar across studies in regard to their
sample size weighted means and sample size weighted variances. The means for the
transformational scales of Charisma (2.52), Individualized Consideration (2.50), and
Intellectual Stimulation (2.48) were found to be higher than the means of the transactional
scales of Contingent Reward (1.83) and Management-by-Exception (2.32) indicating that
transformational behaviors were more frequently observed than transactional behaviors
across all studies. Coefficients for the association between leadership style and
effectiveness were higher for transformational scales (.71, .61, .60) than for transactional
scales (.4 1, .05), with Charisma correlating most highly with leader effectiveness for all
types of criteria. Management-by-Exception, on the other hand, exhibited the lowest
correlation with effectiveness. Table 2 shows that a substantial range of correlations was found for all five scales across
the set of studies included in the research sample. Credibility intervals for the
transformational scales did not include zero, providing an indication that a positive effect
exists across studies between transformational leadership and effectiveness. However, due
to the range of effect sizes and the range of the credibility intervals, the presence of
moderators is suggested, and the necessity to break the studies into subgroups was clearly
indicated. Credibility interval analysis for the transactional scales included zero. This
finding suggests that transactional leadership does not have a clear positive or negative
impact on effectiveness across studies. The range of the credibility intervals suggest that
the obtained effect sizes may be moderated by some variable which could be identified by
further differentiating the studies into subpopulations. Accordingly, moderators suggested
by our review of the literature were investigated.
Effects of Type of Organization
Hypothesis 1 predicted that transformational leadership behaviors would be more
prevalent in Private as compared to Public Organizations. Transactional leadership
behaviors were expected to be more common in Public Organizations than in Private
Organizations. Table 3 presents the means and standard deviations for each MLQ scale in
studies of both Public and Private Organizations. Directly contrary to expectations,
transformational leadership behaviors were more commonly observed in Public
Organizations than in Private Organizations. For all three transformational scales, the
mean score of leaders in Public Organizations was significantly greater than the mean
score of leaders in Private firms (p < .Ol). For example, the mean Charisma score for leaders in Public Organizations was 2.61 as compared to a mean of 2.37 for leaders in
Private Organizations (z = 8.69, p < ,001). While there was no significant difference in the
frequency of Contingent Reward leadership behaviors, Public Organization leaders were
perceived by their subordinates as practicing significantly more frequent Management-by-
Exception behavior than leaders in Private firms (2.41 vs. 2.10, z = 16.25, p < .OOl).
T ab
le
3 T
he M
od er
at in
g E
ff ec
t of
T yp
e of
O rg
an iz
at io
n on
t he
R el
at io
ns hi
p B
et w
ee n
M L
Q
Sc al
es
an d
L ea
de r
E ff
ec ti
ve ne
ss
M ea
n Sc
al e
SD
C ha
ri sm
a
Pu bl
ic
3, 67
0
Pr iv
at e
2, 25
7
D Q
j% re
nc r
T es
t ’
In di
vi du
al iz
ed
C on
si de
ra tio
n
Pu bl
ic
3, 48
3
Pr iv
at e
2, 27
1
D if
fe re
nc e
T es
t
In te
lle ct
ua l
St im
ul at
io n
Pu bl
ic
3, 48
3
Pr iv
at e
2. 39
8
D if
fe re
nc e
T es
t
C on
tin ge
nt
R ew
ar d
Pu bl
ic
4, 56
0
Pr iv
at e
2, 34
0
D if
fe re
nc e
T es
t
M an
ag em
en t-
by -E
xc ep
tio n
Pu bl
ic
4, 24
2
Pr iv
at e
2, 27
1
D if
fe re
nc e
T es
t
.9 4
.9 0
.8 6
.8 8
.8 8
.8 6
.8 3
.8 1
.6 4
.6 6
2. 61
2. 37
2= 8.
69 *“
*
2. 58
2. 36
z= g.
og **
*
2. 52
2. 43
:= 4.
03 **
*
I . 85
1. 83
~ 0.
89
2. 41
2. 10
;= 16
,2 5*
**
1. 09
22
.0
6 to
. 9
1
0. 93
23
.1
9 t o
.8 3
0. 93
20
-.
O l
to .
84
1. 12
20
.1
6 to
. 77
0. 90
20
.0
4 to
.8 0
0. 75
24
.0
8 to
. 89
0. 96
0. 72
0. 70
0. 59
24
-. 05
t o
.7 0
20
-. 25
t o
.7 1
19
-.3 4
to .
31
20
-.5 3
to .
41
N o
f R
an ge
of
M
eu n
C or
re la
ti on
C
or re
la ti
on
M ea
n R
an
C oe
ff X
en u
C oe
fi ci
en ts
C
or re
la ti
on
C or
re ct
ed
C or
re la
ti on
.6 7
.7 4
.5 1
.5 9
-= 2.
22 *
.5 5
.6 3
,4 45
.5
4
;= 1
.4 8
.5 7
.6 5
.3 9
.4 7
~ 2.
94 ~
~
.3 5
.4 1
.2 6
.3 2
?= 1.
22
.0 8
.lO
-.0 2
-. 03
< =
2. 98
**
95 %
95
c /c
C on
fi de
nc e
C re
di hi
li Q
In te
rv al
In
te tw
l
.6 5
to .
82
.3 1
to 1
.0 0
.5 0
to .
69
.0 8
to 1
.0 0
.5 4
to .
72
.I 6
to 1
.0 0
.4 5
to .
62
.I 4
to .
93
.5 7
to .7
2 .2
7 to
1.
00
.3 7
to .
56
-.0 5
to .
98
.3 2
to .
50
-.l O
t o
.9 1
.2 3
to .
42
-.1 8
to .
83
.0 4
to .1
7 -.2
2 to
. 43
-.0 9
to .
03
-.3 2
to .
27
N ot
es ,
‘F or
c om
pa ri
so n
of m
ea ns
, co
m pu
te d
as z
-t es
t of
di ffe
re nc
e be tw
ee n
tw o
In de
pe nd
en t
sa m
pl e
m ea
ns (
N et
er ,
W as
se rm
an
& K
ut ne
r,
19 90
, p.
13 ).
F or
co rr
el at
to na
l di
ff er
en ce
s,
co m
pu te
d as
z -
te st
o f
cr iti
ca l
ra tlo
b et
w ee
n t\*
o m
ea n
ef fe
ct
si ze
s (H
un te
r &
S ch
m id
t, 19
90 ,
p. 4
38 ).
*/ 1<
.0 5.
** p<
.o 1.
** *<
.0 01
.
Transformational vs. Transactional Leadership 407
Hypothesis 2 calls for a test of the moderating effect of Type of Organization (Public vs. Private) on the correlational relationship between MLQ scales and leader effectiveness. Table 3 shows results for this comparison. Significant differences in mean effect size were
found between Public and Private Organizations for the scales Charisma (z = 2.22, p < .05),
Intellectual Stimulation (z = 2.94, p < .Ol) and Management-by-Exception (z = 2.98, p < .Ol). Thus, the hypothesis that Type of Organization moderates the relationship between leadership style and effectiveness was partially supported. Significantly higher positive
relationships were found in Public as compared to Private Organizations for each of the three scales, a finding which contradicted our expectation of stronger positive relationships
in Private Organizations. The 95% confidence intervals for both Public and Private organizations did not include zero for the transformational MLQ scales, and for the
transactional scale Contingent Reward. For the transactional scale Management-by- Exception, the confidence interval included zero in studies of Private Organizations but
was nonzero for Public Organizations. Taken alone, these results could be interpreted as an indication of a positive relationship between leadership styles and effectiveness for at least
four of the five MLQ scales. However, credibility intervals were large for all five scales and included zero for both moderator categories of the two transactional scales. Additionally, the credibility interval for the Intellectual Stimulation scale included zero for the moderator
category Private Organization but did not include zero for the moderator category Public Organization. These results suggest that other moderators may explain the differences in effect sizes across studies and that the probe for moderators should continue.
Effects of Level of the leader
Hypothesis 3 stated that Higher-Level Leaders would be perceived to practice more frequent transformational leadership behavior than Lower Level Leaders. Lower level
leaders were expected to practice more transactional leadership behaviors than Higher Level Leaders. The results for the hypothesis were the reverse of what was expected. Table
4 presents results which indicate that mean scores on all three transformational scales and
one of the two transactional leadership scales were higher for Low Level Leaders than for High Level of Leaders. For instance, the mean Individualized Consideration score for High
Level Leaders was 2.39, while Low Level Leaders had a mean of 2.66 (z = 10.85,~ < ,001). As found earlier, there were no differences in the mean frequency of Contingent Reward behaviors but Low Level Leaders were rated as exhibiting significantly more Munagement-
By-Exception behaviors than High Level of Leaders (2.48 vs 2.12, z = 17.81,~ < .Ol). Hypothesis 4 tested the moderating effect of Level of the Leader (High Level vs. Low
Level) on the correlational relationship between MLQ scales and leader effectiveness.
Table 4 presents the results for this comparison. No significant differences in mean effect sizes were found for any of the MLQ scales differentiated by Level of the Leader. Thus, the
hypothesis that Level of the Leader moderates the relationship between leadership style and effectiveness was not supported. The 95% confidence intervals surrounding these effect sizes were nonzero, indicating a positive relationship between leadership styles, as measured by each of the five scales of the MLQ and effectiveness for both Low and High Level Leaders. Credibility intervals were large and excluded zero for both High and Low Level Leaders for the three transformational scales and the transactional scale Management-by-Exception. For the transactional scale Contingent Reward, the credibility
T ot
al S
am pl
e
Si ze
C ha
ri sm
a
H ig
h L
ev el
L
ea de
rs
2, 03
3
L ow
L ev
el
L ea
de rs
3,
35 6
D if
fe re
nc e
T es
t ’
In di
vi du
al iz
ed
C on
si de
ra tio
n
H ig
h L
ev el
L
ea de
rs
2, 17
4
L ow
L
ev el
L
ea de
rs
2, 96
2
D if
fe re
nc e
T es
t
In te
lle ct
ua l
St im
ul at
io n
H ig
h L
ev el
L
ea de
rs
2, 17
4
L ow
L
ev el
L
ea de
rs
3. 09
0
D if
fe re
nc e
T es
t
C on
tin ge
nt
R ew
ar d
H ig
h L
ev el
L
ea de
rs
2, 25
0
L ow
L
ev el
L
ea de
rs
3. 95
5
D (f
fe re
nc e
T es
t
M an
ag em
en t-
by -E
xc ep
tio n
H ig
h L
ev el
L
ea de
rs
2. 17
4
L ow
L
ev el
L
ea de
rs
3, 81
6
D if
fe re
nc e
T es
t
T ab
le
4 T
he M
od er
at in
g E
ff ec
t of
t he
L ev
el
of t
he L
ea de
r
on t
he R
el at
io ns
hi p
B et
w ee
n M
L Q
Sc
al es
an
d L
ea de
r E
ff ec
ti ve
ne ss
M ea
n
C ro
nh ac
h C
(
fo r
Sc al
e
.9 3
.9 0
.8 8
.8 7
.8 8
.8 5
.8 1
.8 2
.6 7
.6 5
M ea
n S
cu le
Sc or
e
2. 50
2. 60
< =
3. 39
** *
2. 39
2. 66
~= 10
.8 5*
**
2. 40
2. 56
:= 6.
87 **
*
1. 83
1. 85
;= 0.
85
2. 12
2. 48
:= 17
.8 1*
**
M ea
n Sc
al e
SD
1. 07
9
.4 0
to .
86
.6 2
1. 04
26
.3
6 to
. 91
.6
1
0. 81
10
.3
0 to
. 84
.5
3
0. 93
19
.3
7 to
. 83
.5
1
0. 76
10
.2
5 to
. 78
.5
3
0. 88
23
.1
9 to
. 89
.4
8
0. 82
11
.2
5 to
. 71
.3
9
0. 93
21
.0
5 to
.7
2 .2
9
0. 72
0. 77
10
-.1 2
to .
47
.0 9
20
-.5 4
to .
39
.0 7
N of
R
an ge
c$
M
ea n
95 7
c
C or
re la
ti on
C
or re
la ti
on
M ea
n R
aw
C or
re la
ti on
C
or re
ct ed
C
on ji
de nc
e
C oy
fi ci
en ts
C
o& f?
ci en
t. s
C or
re la
ti on
In
te rv
al
.6 9
.7 0
z= o.
1 2
.6 0
.6 0
;= 0.
04
.6 1
.5 7
:= 0.
46 **
.4 7
.3 5
PI .5
9
.1 2
.0 9
:= 0.
48
95 %
C re
di bi
li ty
In te
rv al
.5 5
to .
84
.2 1
to 1
.0 0
.1 8
to
1. 00
.0
8 to
1 .0
0
.4 8
to .
72
.1 7
to 1
.0 0
.5 1
to .
69
.1 4
to 1
.0 0
.4 8
to .
74
.1 5
to 1
.0 0
.4 8
to
.6 7
.0 6
to 1
.0 0
.3 6
to
.5 7
.0 7
to .
87
.2 5
to
.4 5
-.2 0
to .
90
.0 3
to .
20
.0 3
to .
15
N ot
es : ‘
Fo r c
om pa
ri so
n of
m ea
ns ,
co m
pu te
d ab
:- te
bt o
fd if
te re
nc e
be tw
ee n
t& o
in de
pe nd
en t
sa m
pl e
m ea
ns (
N et
er ,
W as
se rm
an
& K
ut ne
r,
19 90
, p.
13 ).
F or
c or
re la
tio na
l di
ff er
en ce
s.
co m
pu te
d as
z -
te st
o f
cr iti
ca l
ra tio
b et
w ee
n tw
o m
ea n
ef fe
ct
su es
(H
un te
r &
S ch
m dt
. 19
90 ,
p. 4
38 ).
*/ x0
5.
** p<
.o 1.
** *<
m l.
Transformational vs. Transactional Leadership 409
interval was nonzero for the moderator category High Level Leader, but included zero for
the moderator category Low Level Leader. These results suggested that additional moderators may explain the differences in effect sizes across studies and that a probe for
moderators should extend beyond type of organization and level of leader.
Effects of Type of Criterion
Hypothesis 5 tested the moderating effect of Type of Criterion measure, subordinate perceptions vs. organizational measures on the co~e~ational relationship between M!,Q
scales and leader effectiveness. Table 5 shows results for this comparison. Large and
significant differences @ c .OOl) in mean effect size were found between subordinate perceptions of effectiveness and organizational measures of effectiveness for all five MLQ
scales. Thus, the hypothesis that Type of Criterion moderates the relationship between
leadership style and effectiveness was fully supported. Significantly higher positive relationships were found for subordinate perceptions of
effectiveness as compared to organizational measures of effectiveness @ < .Of). The 95% confidence intervals for both subordinate perceptions and organizational measures did not
include zero for the transfo~ation~ scales and the transactional scale C~~~~~genr Reward.
For the transactional scale Management-by-Exception the confidence intervat included
zero in studies utilizing organizational measures of effectiveness but did not include zero for studies which measured subordinate perceptions of effectiveness.
Although not as strongly associated with work unit effectiveness as measured by
subordinate perceptions, it should be noted that significant positive relationships were observed across studies for the three ~ansfo~ational leadership scales regardless of the
Type of Criterion @ < .Ol ). Confidence intervals narrowed considerably relative to other moderator analyses, suggesting that Type of Criterion is a powerful moderator of the
relationship between MLQ scales and effectiveness, and that it accurately partitions a portion of the variance in effect size across studies. Consideration of this moderator in
subsequent analyses was clearly indicated, and results suggest that prior moderator analyses (i.e., Type of Organization, Level of the Leader) should be revisited in light of this
moderator finding. Differences identi~ed by Type of Organization and Level of the Leader analyses could
be confounded by over-representation of one criterion inst~ment type or the other
(subordinate perceptions or organizational measures). The Type of Organization and Level
of the Leader meta-analyses were reanalyzed within criterion type. These results are presented graphically rather than in tabular form because the complex effects are more starkly displayed in a graph as compared to a table. Figures 2 and 3 graphically depict how the Type of Criterion moderates the leadership style-effectiveness relationship for the two
Types of Organization (Figure 2) and for the two Levels of Leader (Figure 3). This
conjecture, that meta-analysis should be conducted within the criterion type, is similar to the conclusion reached by Wofford and Liska (1993) that meta-analysis of path-goal research should only be conducted within instrument type.
Figure 2 compares four patterns of relationships (two Types of Organization x two Types of Criterion) between MLQ scores and effectiveness. Within the Subordinate Perceptions criterion type, studies of Public Organizations yielded significantly higher correlations between effectiveness and the scales Intellectual Stimulation, Contingent Reward, and
T ab
le
5 T
he M
od er
at in
g E
ff ec
t of
T yp
e of
C ri
te ri
on
M ea
su re
m en
t on
t he
R el
at io
ns hi
p B
et w
ee n
M L
Q
Sc al
es
an d
L ea
de r
E ff
ec ti
ve ne
ss
R an
ge o
f C
or re
la tio
n M
ea n
R U
N
M ea
il 95
c/
r 95
%
C oq
jfi ci
en ts
C
or re
lu tio
n C
or re
ct ed
C
on fid
ew e
C re
di bi
liQ
C or
re la
tio n
In te
iW d
In te
rv al
C ha
ri sm
a
Su bo
rd in
at e
R at
in gs
O rg
an iz
at io
na l
M ea
su re
s
D iff
er en
ce
in
Q fe
ct
S te
’
In di
vi du
al iz
ed
C on
si de
ra tio
n
Su bo
rd in
at e
R at
in gs
O rg
an iz
at io
na l
M ea
su re
s
D Q
fe re
nc e
in
E fS
ee cr
S iz
e
In te
lle ct
ua l
St im
ul at
io n
Su bo
rd in
at e
R at
in gs
O rg
an iz
at io
na l
M ea
su re
s
D @
re nc
e in
E
fS rc
r S
iz e
C on
tin ge
nt
R ew
ar d
Su bo
rd in
at e
R at
in gs
O rg
an iz
at io
na l
M ea
su re
r
D ife
re nc
r in
IZ
ffe cr
S
iz e
M an
ag em
en t-
by -E
xc ep
tio n
Su bo
rd in
at e
R at
in gs
O rg
an iz
at io
na l
M ea
su re
s
D iff
er en
ce
in
E fe
ct
S iz
e
5. 47
5 .9
3 2.
53
I .0
6 32
.3
6 to
. 9
1 .7
1
1. 60
8 .9
1 2.
62
0. 9
1 15
.I
0 to
. x3
.3
0
5, 28
6 .8
9 2.
50
0. 88
29
.3
0 to
. x4
.5
9
1, 29
5 .X
7 2.
70
0. 74
12
-.O
l to
. 6S
.2
4
5. 35
0 .X
7 2.
48
0. 87
31
.I9
to .
x9
.5 X
1. 42
2 .8
7 2.
63
0. 69
14
.0
X to
. 67
.2
2
5. 32
7 .X
4 I
.9 9
0. 96
28
.O
O to
.7 2
2. 37
1 .X
2 1.
68
0. 73
15
-.?
S to
. 4x
5, 22
8 .6
2 2 .
3 3
0. 76
29
-.5
4 to
. 37
2. 06
9 .6
7 2.
40
0. 63
12
m
.2 1
to .
16
.4 6
.0 7
.0 7
-.0 3
.X 1
.3 5
,- =1
6. 01
** *
.7 5
to
.8 6
.4 x
to 1.
00
.3 3
to .
37
.0 2
to
.6 X
.6 9
.2 8
;= 13
.3 4*
**
.6 3
to .
75
.3 4
to 1
.0 0
.2 6
to .
29
.0 2
to .
54
.6 X
.2 6
:= 13
,X 5*
“*
.6 3
to .
74
.3 3
to .
8X
.2 5
to .
2X
-.0 2
to .
55
.5 6
.0 X
% ,6
,3 5*
“*
50
to .
61
.2 3
to .
8X
.0 7
to .
09
-. 16
t o
.3 2
.I 0
-.0 4
:= 4.
38 **
”
.0 4
to .
I6
-. 32
to
. 52
-.o s
to .
03
-. I3
to
. 05
N ot
ea : ‘F
or
co m
pa ris
on
ot
m ea
n\ .
co m
pu ed
a\
: -t
at
ot
cr ik
al
m (io
be
tu se
n ~M
O m
ea n
ef fe
ct
si re
s (s
ee
H un
te r
&
S ch
m id
t, 19
90 ,
p. 4
38 ).
*/ ‘<
S E
** p<
.il 1.
** *<
.0 01
.
Transformational vs. Transactional Leadership 411
4-
4- a----_*-
_ _ $_ +-_ _ _ _ _ _ _ - - 5-r
9,’ - -.&- - - - _ _ _ _ _ _ _ _ _ _ _r I*G $@ , , =-. #* _&‘_ ___;“_ -&_?_____ , d p* e- i\on -. /- _&ii@_____________ ___ Public Organization - Private Organization
MBE
I
CR IC 1s CH
MLQ Scales
Figure 2. The Moderating Effect of Types of Measures (Subordinate Perceptions vs. Organizational Measures) on the Relationship between MLQ
Scales and Leader Effectiveness in Public and Private Organizations
-0.2 ! I MBE CR 1s IC CH
MLQ Scales
Figure 3. The Moderating Effect of Types of Criterion Measures (Subordinate Perceptions vs. Organizational Measures) on the Relationship between MLQ
Scales and Leader Effectiveness of High and Low Levels (HL vs. LL) of Leaders
Management-by-Exception. For organizational measures of effectiveness, studies of Public
Organizations had significantly higher correlations between effectiveness and the
Intellectual Stimulation scale. These findings for the Type of Organization by Type of Criterion analyses can be
compared with the findings described earlier and shown in Table 3 for the combined
criterion types. The prior finding of a difference for the Charisma scale across-criterion types was no longer significant when studies were separated by the Type of Criterion used
412 LEADERSHIP QUARTERLY Vol. 7 No. 3 1996
to measure effectiveness. In contrast, the significant differences between Public and
Private organizations across-criterion types for Intellectual Stimulation (p < .Ol) identified in Table 3 are consistent with the findings of significantly higher positive relationships for Public organizations within criterion types (Figure 2), whether the measure was subordinate perceptions of effectiveness (z = 2.03, p < .05) or organizational measures of effectiveness (z = 1.86, p < .05). While the across-criterion analyses (Table 3) yielded no significant difference by the Type of Organization for Contingent Reward, the within- criterion analyses (Figure 2) detected a significant difference between Public and Private Organizations (Z = 2.12, p < .05) for subordinate perceptions of effectiveness. Contingent Reward was more highly associated with subordinate perceptions of effectiveness in
Public Organizations. The across-criterion type analyses (Table 3) identified a significant difference between Public and Private Organizations for Management-by-Exception (z = 3.20, p < .Ol). However, the within-criterion analyses showed a significant difference only when effectiveness was measured by subordinate perceptions (Z = 3.20, p < .Ol) but showed no difference when effectiveness was determined by organizational measures.
Similarly, the findings for the Level of the Leader by criterion type analyses (within- criterion) can be compared with the findings described earlier and shown in Table 4 for the combined criterion types (across-criterion). However, the finding of no significant differences between High and Low Level Leaders for any of the MLQ scales within- criterion types (Figure 3) is consistent with the findings of no significant differences across-criterion types provided in Table 4. These findings of no significant differences between High and Low Level Leaders across-criterion or within-criterion types provide considerable evidence that Hypothesis 2 is not supported. The pattern of relationships between MLQ scales and effectiveness is similar for both High and Low Level Leaders regardless of whether effectiveness is measured by Subordinate Perceptions or by Organizational Measures of performance.
SUMMARY AND DISCUSSION
Results of the meta-analysis support the belief that transformational leadership is associated with work unit effectiveness. All hypotheses tested show higher associations between transformational scales and effectiveness than between transactional scales and effectiveness. Credibility intervals generally excluded zero for transformational scales and included zero for transactional scales, suggesting the existence of a positive relationship between transformational leadership and effectiveness across different contexts. Relationships between transactional scales and effectiveness were more ambiguous. Moderator analyses indicated that the type of criterion used to measure effectiveness is a powerful moderator of the relationship between MLQ scales and leader effectiveness. Several differences were found between studies conducted in the public and private sectors in the strength of the relationships between MLQ scales and effectiveness. No differences in the pattern of relationships between MLQ scales and effectiveness were observed in comparisons of studies of high level and low level leaders. These, and other results of the meta-analysis require further elaboration and discussion of their implications for understanding the effects of the transformational leadership construct. In an effort to clearly summarize the results, we have chosen to first present the limitations of this study and then organize the discussion around each of the ML& scales separately.
Transformational vs. Transactional Leadership 413
limitations of the Study
First, consistent with findings in other literature domains (cf. Gaugler et. al., 1987), the
present study suggests that source bias concerns regarding the published/unpublished dichotomy as a potential moderator is unwarranted for this literature domain. However, our finding of a difference for one MLQ scale suggests that the issue of source bias should
almost always be broached when applying meta-analytic techniques to a particular domain
of literature. Second, we tested the potential impact of the “file drawer problem” using a technique
provided by Hunter and Schmidt (1990) that estimates the number of studies with a
correlation coefficient of zero that would have to be relegated to the confines of researchers’ file drawers for the meta-analysis estimate of effect sizes to approach zero.
Because this is not really a result, we chose to include these calculations as part of the limitations section in order to describe the degree to which this potential limitation is relevant to the present study. The number of additional study correlation coefficients of
zero required to cause the meta-analysis effect size of the present study to approach zero would be Charisma, 3111 studies; Individualized Consideration, 2392 studies; Intellectual
Stimulation, 2500 studies; Contingent Reward, 1650 studies; and Management-by- Exception, 135 studies. Simply stated, this limitation for all practical purposes is not of concern based on the magnitude of the effect sizes and the substantial number of studies
used in this meta-analysis. Third, tests of mean differences in the amount of transformational and transactional
leadership exhibited by high vs. low level leaders and by leaders in public vs. private
organizations were based solely on studies which met specific criteria for inclusion in this study. A substantial number of other studies which have examined the transformational- transactional constructs were eliminated from our analyses, primarily because they did not include a measure of effectiveness. Many of these studies provided mean ratings on MLQ
scales, and it is possible that a different finding would emerge concerning level of the
leader and type of organization if all studies that have been conducted were included in the analyses. To the extent that our criteria for inclusion creates a nonrepresentative sampling
of the population of studies, the aggregated means are biased. Although it could be argued
that the studies including an effectiveness measure are more rigorous, and hence more accurately estimate the true distribution parameters, this is an empirical question that
cannot be answered at this juncture. However, the fact that our cross-study analyses are based on thousands of respondents suggests that the sampling error is not likely to be a leading concern. It is also worth noting that our “public” sample is primarily composed of
research in the military and in educational institutions. Thus the generalizability of these
findings to other prominent types of public agencies such as local, state, and federal government agencies is not known. Clearly, there is a research gap in assessing the impact of transformational leadership in these settings, information that would be relevant in light of recent discussions regarding the need for ways to “reinvent government” both in process and in leadership.
Fourth, perhaps the greatest limitation of this research study is that a single measure of the relevant constructs forms the basis of our conclusions. Avolio and Gibbons (1988) note that, while much of the prior research on the construct of transformational leadership has been conducted with top organizational leaders, the MLQ is the only instrument in
414 LEADERSHIP QUARTERLY Vol. 7 No. 3 1996
widespread use that attempts to assess transformational leadership in a quantitative way
across organizational levels. Nevertheless, mono-method bias may limit the
generalizability of conclusions beyond the specific items comprised by the scales of the B&Q_ Relationships found between effectiveness and transformational leadership are thus
bounded by the extent to which the &EQ accurately captures the constructs. Fifth, large sample sizes result in statistically significant differences between means and
correlation coefficients that may not be practically significant when viewed as absolutes.
As the number of studies of the construct increases, so will the opportunity to analyze these effects with regard to more precise moderator categories that will explicate the practical
importance of differences among the effect sizes. Sixth, though we bebeve the moderators chosen to be of theoretical and practical
interest, we have not empirically exhausted the potential moderators of the relationship between t~nsfo~ational leadership and effectiveness. Other moderators of theoretical
interest include the size of the work unit (large vs. small), the nature of organizational roles within the work unit (well-defined vs. fluid), the organizational role of the leader
(hierarchical vs. faciliatory), gender of the leader and the subordinate, and the version of the MLQ utilized. Though each of these limitations should be considered when interpreting
the results, the strengths of the relationships found in different types of organizations, at different levels of the leader, and utilizing different operationalizations of the criterion
variable provide compeIling evidence for the tr~sfo~ational construct.
leadership and Effectiveness
Across studies, Charisma was consistently the variable most strongly related to leader effectiveness among the MLQ scales. This finding emerged across studies regardless of type of organization, level of the leader, or in how effectiveness was measured. It should
be noted, however, that a much stronger association between Charisma and effectiveness was found for subordinate perceptions of effectiveness than for organizational measures. In addition, Chffris~~~ was more strongly related to effectiveness in public org~izations than in private firms. However, this effect was mitigated when studies were separated by the
type of criterion used to measure effectiveness. Among the MLQ scales, Charisma is perhaps the scale most closely associated with a generalized impression of transformational leadership, due to the nature of the items which make up the scale and
because the construct itself tends to represent affective reactions of subordinates to the leader. In this sense it could be expected that the most encompassing measure of
transfo~ational leader behavior would be more highly associated with perceptions as well as other measures of effectiveness.
Much like Charisma, the individualized Consideration scale was found to be much more strongly associated with subordinate perceptions of effectiveness as compared with organizational measures of effectiveness. A positive association between Individualized Consideration and effectiveness was consistent across studies. No differences were found in public vs. private organizations or for different levels of leader in how Individualized Consideruti~n relates to effectiveness. Through the aggregation of studies, ~ntellectl~a~ ~tirn~zation was revealed to be perhaps the most interesting among the five scales of the MLQ. It was related to subordinate measures of effectiveness as well as organizational measures, although it was more highly related to the subordinate measures. A significant
Transformational vs. Transactional Leadership 415
difference was noted in how Intellectual Stimulation relates to effectiveness in public vs. private organizations 0) < .Ol). Regardless of how it was measured, Intellectual Stimulation was more highly related to effectiveness in public organizations than in private
firms. The public versus private organization findings for Intellectual Stimulation are
somewhat counter-intuitive. Intellectual Stimulation is generally associated with encouraging subordinates to think about problems in new ways, which should be of particular importance in private organizations seeking competitive advantage. Public institutions are often thought to function within a more bureaucratic framework that may serve to suppress the impact of a transformational leadership style (Bass, 1985). The procedural nature of bureaucracies would tend to provide substitutes for leadership in the form of structures and procedures rather than creating leadership opportunities. But it may be the very nature of the mechanistic organization that propels transformational leadership in the form of Intellectual Stimulation leader behavior to be highly salient to individuals. In organizations where innovation is the norm, perceptions of Intellectual Stimulation may not be as prominently associated with effectiveness as it is in more mechanistic firms. To the extent that Intellectual Stimulation involves cognitive reappraisal of the status quo and the questioning of long held assumptions, it appears that this construct is more highly associated with performance in the public sector than it is in private industry (see Podsakoff et. al., 1990). Our findings support this conclusion, particularly for military and educational institutions.
Another unexpected finding was that Intellectual Stimulation was equally important in its relationship with effectiveness for low and high level leaders. This finding is in contrast to the commonly held assumption that intellectual stimulation of subordinates is more important at higher levels of the organization (Bennis & Nanus, 1985; Tichy & Devanna, 1986). This finding implies that the affective and cognitive appraisal of leader behavior is similar across organizational levels. Leaders promote effectiveness, or at least are perceived to enhance performance, at both the higher and lower levels of the organization when they display transformational behavior. This enhanced performance is achieved
through all those actions that characterize the transformational style including the intellectual stimulation of employees and concern with the human potential of each individual subordinate. The process of arousing the subordinate interests and dormant capabilities is often more important than to promise, threaten, or engage them in nondirective participatory decision making (Bass, 1988b). Whether it is simply that transformational leaders are listened to more attentively in delineating transactional benefits to employees, or whether some other type of motivational challenge is issued by such leadership, transformational behavior appears to have a real impact on performance throughout the organization. Those who have asserted that the transformational construct has been embraced because of the affective allure of its implications-rather than on empirical, practical or rational grounds-are impeached by the consistency of this result across studies.
Intellectual stimulation of subordinates is a relatively unexplored aspect of leadership behavior, the “third child’ of transformational leadership. It now seems quite clear that the leader who is able to intellectually stimulate subordinates will not only foster the perception of effectiveness among subordinates, but will also amplify performance itself as gauged by independent measures of productivity. Inducing employees to appreciate,
416 LEADERSHIP QUARTERLY Vol. 7 No. 3 1996
dissect, ponder and discover what they would not otherwise discern is perhaps the basis of
behavior that comes closest to our prototypical abstractions of “true leadership.” The leader who intellectually stimulates subordinates is teaching subordinates “how to fish for themselves rather than simply giving them the fish’ (Bass, 1988b). When leaders actually engage in such behavior, they appear to engender not only subordinate acclamations, but productive ardor as well. The transformational leader through intellectual stimulation instills feelings of power in followers to attain higher goals through socialized power rather than the “pure” charismatic leader who attempts to exert dominance and subjugate followers through personalized power (Waldman, 1987). Because intellectual stimulation has more than simply a subjective impact on perceptions of effectiveness, this critical leadership construct should be comprehensively investigated.
Credibility analysis revealed that the two transactional scales, Contingent Reward and Management-by-Exception, were inconsistent in their relationships with effectiveness across studies. Some research evidenced positive associations, while other findings showed a negative association between these transactional measures and effectiveness. Contingent Reward, however, appears to have a general positive association with subordinate
perceptions of work unit effectiveness, particularly in public organizations. The ability to successfully achieve transformational leader outcomes is believed to require building on an effective transactional base. Thus, Contingent Reward behaviors may be especially important in public organizations characterized by systematic and political reward granting. The leader who is able to obtain rewards and distribute them in meaningful increments in spite of systematic constraints may enhance unit effectiveness from the perspective of the subordinate. It may also be that the stronger relationship observed in public sector organizations is due to greater stability in the outcomes desired, rendering specific behavior shaping more appropriate. No positive impact of Contingent Reward on organizational measures of effectiveness was found across the studies considered in this meta-analysis. While it is not clear why Contingent Reward behaviors would not have a correspondingly positive impact on organizational measures of effectiveness in public organizations, we can speculate that it may be that these measures are rewarding compliance and quantity of outcomes rather than creativity and quality of outcomes.
Management-by-Exception was found to have no relationship with effectiveness except with subordinate perceptions of effectiveness in public organizations, where a low but perhaps meaningful relation was observed across studies. The implications of these findings suggest that the transactional scales of the MLQ should probably undergo some revision, as the constructs measured by these scales do not seem to achieve results similar to the relationships between effectiveness and transactional behavior found in other research (cf. Wofford & Liska, 1993). While the Contingent Reward scale appears to be associated with subordinate perceptions of effective leadership, the Management-by- Exception provides mixed results, precluding meaningful interpretation. This construct could be considered as representing nonleader behaviors that have zero or negative-effect relationships with effectiveness. Our results indicate that the correlations between Management-by-Exception and both subordinate measures of effectiveness are insignificant. Clearly, this provides strong support for the central assertion in this study that leadership behaviors (as opposed to nonleader behaviors) have a strong effect on organizational effectiveness, irrespective of whether effectiveness is determined from subordinate perceptions or from organizational measures. However, it should be noted that
Transformational vs. Transactional Leadership 417
the Management-by-exception scale has been refined over time into the two subscales active Management-by-exception and passive Management-by-exception. In this study we coded the aggregated scale for consistency purposes and thus predictor contamination may
be a problem. While we lack sufficient studies to make a definitive quantitative statement, our ad hoc qualitative review of the studies which provided these subscales supports the
contention that active management-by-exception may be positively related to effectiveness while passive management-by-exception has no relationship or is negatively related to
effectiveness.
Differences in leader Behavior
Across studies, subordinates in public organizations reported more frequent transformational behaviors by their leaders than was reported by subordinates in private
firms. Also, leaders in public organizations were described by their subordinates as exhibiting more Management-by-Exception behavior as compared to leaders in private
firms. No differences were indicated in the frequency of Contingent Reward behavior demonstrated in public or private org~izations. Across studies, subordinates of low level
leaders reported more frequent transfo~ational behaviors by their leaders than did subordinates of high level leaders. In addition, low level leaders were described by their
subordinates as exhibiting more Management-by-Exception behavior as compared to high level leaders. No differences were indicated in the frequency of Contingent Reward
behavior demonstrated by low level and high level leaders. Avolio and Bass (1988) contend that transformational leadership, though wide spread in
organizations, is more likely at higher levels of organizations. Similarly, Etzioni (1961)
suggested that ch~smatic leadership behaviors are more likely among top managers than among lower-level supervisors. However, cumulative findings in leadership research spanning a decade using MLQ scales indicate that, contrary to widely-held beliefs about
the emergence of transformational leadership, such behaviors are more prevalent among
lower-level leaders. Lower-level leaders were rated higher than higher-level leaders on all three transformational leadership scales. Similarly, results of this integrative review
indicate that public-sector leaders are rated as demonstrating more frequent transfo~ational behavior than leaders in private organizations. Several possible explanations are tentatively offered for these counter-intuitive findings.
One perspective on managerial selection and promotion suggests that these processes are
highly political in nature (Ferris & Judge, 1991). The ambiguous work context at the top management levels contributes to conformity in beliefs as individuals actively seek consensus in their opinions of the ambiguous information processed at the top management
levels (Nemeth & Staw, 1989). These efforts to facilitate homogeneity in beliefs is noticed in the internal staffing and promotion systems of many organizations. Kanter (1977)
coined the term “homosocial reproduction” to characterize organizations in which key decision makers favorably evaluate and promote people just like themselves. It is possible, then, that top managers promote only those whose leadership styles are similar to their own, thus leading to observed differences between higher and lower levels of management.
The information processing perspective on rating leader behaviors suggests that ratings by subordinates are subject to errors induced by several biases, including recall of prototypical behaviors (Lord & Alliger, 1985; Lord, Foti, & DeVader, 1984), judgments
418 LEADERSHIP QUARTERLY Vol. 7 No. 3 1996
based on availability heuristic (Tversky & Kahnemann, 1973), and attributions of extreme
outcomes to leaders (Meindl, 1990). For instance, subordinates of low-level leaders may
hold “naive” theories of leadership, and when relying on the availability heuristic to recall leader-relevant information, may judge prototypical behaviors to occur more frequently. Such heuristics may have had a biasing effect on the frequency rating scales (Lord, Foti, &
DeVader, 1984) utilized in the MLQ, thus resulting in the observed differences between low and high level leaders. It is also possible that transformational behaviors are more common and are thus institutionalized at the higher levels of management.
Transformational leadership may also be expected more at higher levels in the organizations and hence the exhibition of a transformational behavior may be less salient
to observers at these higher levels. Subordinates may not detect/notice such behaviors
unless they are truly exceptional, thereby leading to a downward bias on the mean scores reported for higher level leaders. A similar ~gument could be made for the lower mean scores on transformational leadership scales reported for leaders in private organizations,
where such behaviors may have been taken for granted or come to take on a “rule-like”
status (Meyer & Rowan, 1977; Scott, 1987). Aside from these explanations, it is possible, and we believe more plausible, that low
level leaders and public sector managers may indeed be more transformational in their leadership styies. Coupled with the finding that leadership effects are higher for leaders in
public org~izations, the study results suggest that leadership at the top and in private organizations may not have utilized the oppo~unity to elevate the petiomlance of their subordinates using transfo~ational leadership. Alternatively, it may be that these differences reflect real effects due to the functional demands made on organizational
leaders. Lower-level leaders through their day-to-day contact with subordinates may have a greater opportunity to effect work unit outcomes through frequent displays of charisma,
individualized consideration, and intellectual stimulation of subordinates. Higher-level leaders whose functional duties are more abstract-oriented towards long term policy may
have less need or fewer oppo~unities to exhibit these behaviors as frequently. Finally, it may be that the complex relationship between transfo~ational leadership and performance as a function of criterion type and sector may reflect the operational
evaluation standards being applied in each sector.
Type of Criterion
Moderator analyses indicated that the type of criterion used to measure effectiveness is a powerful moderator of the relationship between MLQ scales and leader effectiveness. This
large difference likely occurs due to a combination of mono-method bias and a fundamental difference in the aspects of effectiveness being measured. Mono-method bias likely contributes to inflated correlations between subordinate ratings of leader behavior and subordinate ratings of leader effectiveness as raters strive for cognitive consistency in their responses to the dependent and independent variables. However, it is also likely that organizational measures tend to attenuate the relationship between subordinate ratings of leader behavior and leader effectiveness by focusing the dependent variable on a more
narrow perspective of pe~o~~ce (score on a test, percent of goals met, financial indicators) than the constellation of outcomes that might be included in subordinate perceptions (individual development, organizational learning, more ethical practices).
Transformational vs. Transactional Leadership 419
Recent interest in 360 degree type performance review techniques that incorporate su~rdinate, customer, and supplier inte~~tations of effectiveness is one such overt
acknowledgement of the deficiency of traditional hierarchical and accounting based
measures of performance. Our position is that the tendency for mono-method bias to inflate
the relationship between reported behaviors and effectiveness and the tendency for myopic
organizational measures to attenuate this same relationship suggests that the “true”
relationship lies between that indicated by the study results for subordinate perceptions and
that for organizational measures. The results of the overall meta-analysis (Table 2) provide
our best estimate of the relationship between MLQ scales and effectiveness, but it must be
noted that this estimate contains much larger sample size for the subordinate perception as
criterion and thus is likely to be biased upward from the “true” relationship that would be
obtained with balanced sample size. Though the size of the “true” relationship is of interest,
the critical outcome is that a consistent relationship exists between transformational ratings
and effectiveness regardless of criterion type, while a similar claim cannot be made for the
transactional scales.
Implications for Practice
The development of effective and sustained leadership is a central concern for most
organizations. While much of the attention in the past has been on the development of
leaders at the upper levels of the organization, new organizational paradigms that include
the sharing of information, decentralization of decision-making authority, and widespread
use of teams have made the development of leaders across organizational levels
increasingly important. The findings of the present study have corresponding implications for the selection,
training, and development of all types of management and supervisory personnel for these
new organization structures. Interviewing practices might prove more predictive of
effectiveness if structured around situational questions about transformational experiences
in individualized consideration and intellectual stimulation. Traditionally, interviews at
lower levels have focused on assessing technical expertise, with considerably less
emphasis given to the interpersonal abilities of lower level leaders. While perhaps
approp~ate to mechanistic organizational structures where line leaders are given relatively
little autonomy in process and task design, the ability of line leaders to stimulate
subordinates in downsized, more fluid organizational designs may be a crucial ingredient
of organizational success. Interviews should strive to not only assess the ability of lower
level leaders to “oversee” technical aspects of the work, but to use multiple techniques to
assess leader ability to stimulate work teams. Situational questions about transformational
experiences in individualized consideration and intellectual stimulation might be
developed and scored. mini-assessment center type exercises might also be utilized to
assess how the potential leader probes and prods subordinates to solve a workplace problem, redesign a workplace process, or improve an internal or external customer
interface. Thus, the interview process might be recast not as an exercise in determining if
the supervisor understands and is qualified to oversee the “job,” but as one where the
determination is made if the supervisor has the ability to coach work unit members to
achieve breakthrough performance.
420 LEADERSHIP QUARTERLY Vol. 7 No. 3 1996
Training modules could be developed to point out the importance of and techniques for
enhancing intellectual stimulation as a fundamental aspect of sound supervisory practice
due to the impact it seems to have on perceptions of effective leadership. Most
organizations place emphasis on the acquisition of technical skills in the training of lower
level leaders and place increasing emphasis on inte~ersonal and strategic planning skills at higher organizational levels. This differential emphasis in training may be partially
attributed to a series of studies which sought to determine what managers do (Luthans,
Hodgetts, & Rosenkrantz, 1988; Mintzberg 1975). Our results indicate that the strength of
the relationship between transformational leadership and effectiveness is the same at both
lower- and higher-levels of the organization. This suggests that more emphasis should be
given to making these lower level leaders aware of the importance of giving subordinates
individual consideration and understand the processes that can be followed to intellectually
stimulate their work unit. The degree of loader-subordinate interaction is likely to be
greater at lower levels in the organization than at higher levels in the organization. At
higher organizational levels, leaders have greater requirements for interaction with
constituencies external to the organizational boundaries, thus reducing the time allowed for
leader-subordinate interaction. Thus, transformational leadership theory development
would be aided by a consideration of the task and level of interaction between subordinate
and leader. It follows that training should likely utilize situational and interactive exercises
in developing transfo~ational lower level leaders, while training at the upper levels would focus on enhancing leader ability to stimulate subordinates through clear written
communication and formal speech-making processes that are more characteristic of mass
and relatively less frequent communication processes.
CONCLUSIONS
In sum, this meta-analysis has integrated a wide range of findings regarding the
relationships between leadership styles and effectiveness. Transfo~ational leadership
behavior that elicits second-order changes in employee efforts is more highly associated
with effectiveness than the traditional first order changes resulting from transactional
behaviors. Tests for moderators in leadership research often provide mixed results (cf.
Howell, Dorfman, & Kerr, 1986; Kemery, Mossholder, & Dunlap,1989), but those of this
study clearly suggest that Intellectual Stimulation of subordinates may be differentially
effective across organization types. The moderator tests also suggest that leader behavior may be more inlpo~ant at lower
organizational levels than has been generally assumed by those who view transfo~ational leadership as primarily the vision inculcated by top managers in their call to action.
Transformational scales were related to effectiveness in highly similar patterns for both
high and low level leaders. The effects of Contingent Reward on organizational effectiveness confirm Tosi’s (1982) view that transactional leadership is a necessary
component of effective management. It is possible that the enduring importance of transactional leadership at higher organizational levels may have been overlooked in the
ardor that has accompanied our contemplation of the transfo~ational leadership cons~uct.
These findings provide considerable evidence regarding the role of transfo~ational leadership in different types of organizations and at different levels of the leader.
Transformational vs. Transactional Leadership 421
APPENDIX I Me&Analysis of the Intercorrelations
of the MLQ Scales
CH IC IS CR MBE
CH .*5,s, %o) .70(Y) .‘0(7,
IC .6*(,, .70(Y) .05(8,
IS ,630, .09,x,
CR ,210,
MBE
Note: Subscript numbers m brackets denote the number of studies providing intercorrelations for the two MLQ scales.
REFERENCES
Arnold, H.J. (1982). Moderator variables: A clarification of conceptual, analytic, and psychometric
issues. Organizational Behavior and Human Performance, 29, 143-174. Atwater. L., & Yammarino, F.J. (1989). Transformational leadership among midshipmen leaders at
the United States Naval Academy. Technical report #ONR-TR6, Office of Naval Research. Avolio, B.J., & Bass B.M. (1988). Transformational leadership, charisma, and beyond. In J.G. Hunt,
B.R. Baglia, H.P. Dachler, & C.A. Schriescheim (eds.), Emerging leadership vistas (pp. 29-
50). Lexington, MA: Lexington Books. Avolio, B.J., & Gibbons, T.C. (1988). Developing transformational leaders: A life span approach. In
J.A. Conger & R.N. Kanungo (eds.), Charismatic leadership: The elusive factor in
organizational effectiveness (pp. 276-307). San Francisco: Jossey-Bass. Avolio, B.J., & Howell, J.M. (1992). The effects of leadership behavior and leader-follower
personality congruence on predicting follower satisfaction and consolidated business unit
performance. Centerfor Creative Leaderships’ Second Research Conference on Leadership,
Colorado Springs, CO. Avolio, B.J., Waldman, D.A., & Einstein, W. 0. (1988). Transformational leadership in a
management game simulation. Group & Organization Studies, 13(l), 59-80. Avolio, B.J., Yammarino, F.J., & Bass, B.M. (1991). Identifying common methods variance with
data collected from a single source: An unresolved sticky issue. Journal of Management,
17(3), 571-587.
Bangert-Drowns, R.L. (1986). Review of developments in meta-analytic method. Psychological
Bulletin, 99(3), 388-399.
Bass, B.M. (1985). Leadership and performance beyond expectations. New York: The Free Press. Bass, B.M. (1988a). Evolving perspectives of charismatic leadership. In J.A. Conger and R.N.
Kanungo (eds.), Charismatic leadership: The elusive factor in organizational effectiveness
(pp. 56-84). San Francisco: Jossey-Bass. Bass, B.M. (1988b). The inspirational process of leadership. Journal of Management Development,
7,21-31.
Bass, B.M., & Avolio, B.J. (1989). Potential biases in leadership measures: How prototypes,
leniency, and general satisfaction relate to ratings and rankings of transformational and
transactional leadership constructs. Educational and Psychological Measurement, 49, 509-
527.
Bass, B.M., Avolio, B.J., & Goodheim, L. (1987). Biography and the assessment of transformational
leadership at the world class level. Journal of Management, 13,7-19.
422 LEADERSHIP QUARTERLY Vol. 7 No. 3 1996
Bass, B.M., & Yammarino, F.J. (1991). Congruence of self and others’ leadership ratings of naval
officers for understanding successful performance. Applied Psychology: An International Review, 40(4), 437454.
Bass, B.M., Cascio, W.F., & O’Connor, E. (1974). Magnitude estimations of frequency and amount. Journal of Applied Psychology, 59,3 13-320.
Bennis, W., & Nanus, B. (1985). Leaders: The strategies for taking charge. New York: Harper and Row.
Binning, J.F., Zaba, A.J., & Whattam, J.C. (1986). Explaining the biasing effects of performance cues in terms of cognitive categorization. Academy ofManagement Journal, 29(3), 521-535.
Bryant, M.A. (1990). Relationship between nurse managers perceived transformational versus transactional leadership styles and staff nurse turnover. Unpublished masters thesis, University of Akron, Akron, OH.
Bryce, N.Y. (1989). Leadership styles of Japanese business executives and managers: Transformational and transactional (Doctoral dissertation, U.S. International University). Dissertation Abstracts International, 5006B.
Bums, J.M. (1978). Leadership. New York: Harper & Row. Bums, T., & Stalker, GM. (1961). The management of innovation. Chicago: Quadrangle Books. Conger, J.A., & Kanungo, R.N. (1988). The empowerment process: Integrating theory and practice.
Academy of Management Review, 13,47 l-482.
Conger, J.A., & Kanungo, R.N. (1987). Toward a behavioral theory of charismatic leadership in organizational settings. Academy of Management Review, 12,637-647.
Cowen, S.S. (1990). A study of relationships between perceived leader behaviors of presidents at
public four-year institutions of higher education in the United States and the changes in FTE enrollment, perceptions of effectiveness, subordinate satisfaction, and other factors of the presidency (Doctoral dissertation, Gonzaga University). Dissertation Abstracts International,
5301A. Etzioni, A. (1961). A comparative analysis of complex organizations. New York: Free Press. Ferris, G.R., & Judge, T.A. (1991). Personnel/human resources management: A political influence
perspective. Journal of Management, 17(2), 447-488. Gaugler, B.B., Rosenthal, D.B., Thornton III, G.C., & Bentson, C. (1987). Meta-analysis of
assessment center validity. Journal of Applied Psychology, 72(3), 493-5 11. Glass, G.V., McGaw, B., & Smith, M.L., (1981). Meta-analysis in social research. Beverly Hills,
CA: Sage. Graen, G., & Cashman, J.F. (1975). A role making model of leadership in formal organizations: A
developmental approach. In J. G. Hunt & L. L. Larson (eds.), Leadership frontiers. Carbondale, IL: Southern Illinois University Press.
Graham, J.W. (1988). Transformational leadership: Fostering follower autonomy, not automatic followership. In J.G. Hunt, B.R. Baliga, H.P. Dachler, & C.A. Schriescheim (eds.), Emerging leadership vistas. Lexington, MA.: Lexington Books.
Hater, J.J., & Bass, B.M. (1988). Superiors evaluations and subordinates perceptions of transformational and transactional leadership. Journal ofApplied Psychology, 73(4), 695-702.
Hoover, N.J., Petrovska, L., & Schultz, D. (1991). Transformational leadership and transactional leadership: An empirical test of theory. ERIC document reproduction service, No. ED 33 1177.
House, R.J. (1977). A theory of charismatic leadership. In J.G. Hunt & L.L. Larson (eds.), Leadership: The cutting edge. Carbondale, IL: Southern Illinois University.
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, 89 l-902. Howell, J.P., Dorfman, P.W., & Kerr, S. (1986). Moderator variables in leadership. Academy of
Management Review, II(l), 88-102.
Transformational vs. Transactional Leadership 423
Hunter, J.E., & Schmidt, F.L. (1990). Methods of metu-analysis: Correcting error and bias in
researchfindings. Newbury Park, CA: Sage Publications. Kanter, R.M. (1977). Men and women of the corporation. New York: Basic Books. Kemery, E.R., Mossholder, K.W., & Dunlap, W.P. (1989). Meta-analysis and moderator variables:
A cautionary note. Journal of Applied Psychology, 74, 168-170. Keller, R.T. (1992). Transformational leadership and the performance of research and development
project groups. Journal ofManagement, 18(3), 489-501. King, M.I. (1989). Extraordinary leadership in education: Transformational and transactional
leadership as predictors of effectiveness, satisfaction, and organizational climate in K-12 and
higher education (Doctoral dissertation, University of New Orleans). Dissertation Abstracts
International, 5008A. Kirby, P.C., King, M.I., & Paradise, L.V. (1991). Extraordinary leaders in education: Understanding
transformational leadership. Journal ofEducational Research, May-June, 303-3 11. Koh, W.L., Steers, R.M., & Terborg, J.R. (1991). The effects of transformational leadership on
teachers and students in Singapore. Paper presented at the annual meeting of the Academy of
Management, Miami Beach, FL. Komives, S.R. (1991). The relationship of same- and cross- gender work pairs to staff performance
and supervisor leadership in residence hall units. Sex Roles, 24(5/6), 355-363. Kuhnert, K.W., & Lewis, P. (1987). Transactional and transformational leadership: A constructive/
developmental analysis. Academy of Management Review, 12,648.657. Liang, T. (1990). Relationships among leader attributes, behaviors, and effectiveness of vocational
administration administrators (Doctoral dissertation, University of Minnesota). Dissertation
Abstracts International, 51, 11A. Lord, R.G., & Alliger, G.M. (1985). A comparison of four information processing models of
leadership and social perceptions. Human Relations, 38,47-65. Lord, R.G., Foti, R.J., & De Vader, C.L. (1984). A test of leadership categorization theory:Intemal
structure, information processing, and leadership perceptions. Organizational Behavior and
Human Performance, 34,343-378. Luthans, F., Hodgetts, R.M., & Rosenkrantz, S.A. (1988). Real managers. Cambridge, MA:
Ballinger.
McDaniel, C., & Wolf, G.A. (1992). Transformational leadership in nursing service. Journal of
Nursing Administration, 22(2), 60-65. McNemar, Q. (1960). At random: Sense and nonsense. American Psychologist, 1.5, 295-300. Meindel, J. (1990). On leadership: An alternative to conventional wisdom. In B.M. Staw & L.L.
Cummings (eds.), Research in organizational behavior (vol. 12). Greenwich, CT: JAI Press. Meyer, J.W., & Rowan, E. (1977). Institutionalized organizations: Formal structure as myth and
ceremony. American Journal of Sociology, 83, 340-363. Mintzberg, H. (1975). The manager’s job: Folklore or fact. Harvard Business Review, July-August,
49-61. Nemeth, C.J., & Staw, B.M. (1989). The trade offs of social control and innovations in groups and
organizations. In L. Berkowitz (ed.), Advances in experimental social psychology, 12 (175-
210). New York: Academic Press. Neter, J., Wasserman, W., & Kutner, M.H. (1985). Applied linear statistical models. Homewood, IL:
Irwin. Nunnally, J.C. (1978). Psychometric theory. New York: McGraw Hill. Pereira, D.F. (1986). Factors associated with transformational leadership in an Indian engineering
firm. Presented at the 21st International Congress of Applied Psychology, Jerusalem, Israel. Pillai, R., & Meindel, J.R. (1991). The effect of a crisis on the emergence of charismatic leadership:
A laboratory study. Best Paper Proceedings, Academy of Management.
424 LEADERSHIP QUARTERLY Vol. 7 No. 3 1996
Pod&off, P.M., MacKenzie, S.B., Moorman, R.H., & Fetter, R. (1990). Transformational leader
behaviors and their effects on followers’ trust in leader, satisfaction, and organizational
citizenship behaviors. Leadership Quarterly, l(2), 107-142. Rosenthal, R. (1979). The “file drawer problem” and tolerance for null results. Psychological
Bulletin, 86(3), 638-641.
Rosenthal, R. (1984). Meta-analysis procedures for social research. Beverly Hills, CA.: Sage. Ross, S.M. (1990). Transformational leadership: Measurement of personality attributes and
performance effects on work groups (Doctoral dissertation, George Washington University).
Dissertation Abstracts International, 5203B.
Sagie, A., & Koslowsky, M. (1993). Detecting moderators with meta-analysis: An evaluation and
comparison of techniques. Personnel Psychology, 46,629-640. Salter, D.J. (1989). Leadership styles in the United States marine corps transport helicopter
squadrons. Naval Postgraduate School working paper. Sashkin, R., & Fulmer, S. (1988). Toward an organizational theory of leadership. In J.G. Hunt, B.R.
Baglia, H.P. Dachler, & C.A. Schriescheim (eds.), Emerging leadership vistas (pp. 29-50).
Lexington, MA: Lexington Books. Schmitt, N.W., & Klimoski, R.J. (1991). Research methods in human resources management.
Cincinnati: Southwestern Publishing. Scott, W.R. (1987). The adolescence of institutional theory. Administrative Science Quarterly, 32,
493-s I 1. Seltzer, J., & Bass, B.M. (1987). Transformational leadership: Beyond initiation and consideration.
Annual meeting of the American Psychological Association, New York. Seltzer, J., & Bass, B.M. (1990). Transformational leadership: Beyond initiation and consideration.
Journal of Management, 16(4), 693-703. Seltzer, J., Numeroff,M., & Bass, B.M. (1989). Transformational leadership: Is it a source of more
burnout and stress? Journal of Health and Human Research, Fall, 175-185. Singer, M.S. (1985). Transformational versus transactional leadership: A study of New Zealand
company managers. Psychological Reports, 57, 143-146. Spangler, W.D., & Braiotta Jr., L. (1990). Leadership and corporate audit committee effectiveness.
Group and Organization Studies, 15(2), 134- 157. Tichy, N.M., & Devanna, M.A. (I 986). The transformational leader. New York: Wiley. Tichy, N.M., & Uhich, D. ( 1984). The leadership challenge: A call for the transformational leader.
Sloan Management Review, 26,59-68. Tosi, H.J., Jr. (1982). Toward a paradigm shift in the study of leadership. In J.G. Hunt, U. Sekaran,
& C.A. Schriescheim (eds.), Leadership: Beyond establishment views (pp. 222-223).
Carbondale, IL: Southern Illinois University Press. Trite, H.M., & Beyer, J.M. (1986). Charisma and its routinization in two social movement
organizations. Research in Organizational Behavior, 8, I 13-164. Tucker, M.L., Bass, B.M., & Daniel Jr., L.G. (1992). Transformational leadership’s impact on
higher education satisfaction, effectiveness. and extra effort. Paper presented at the Center of
Creative Leadership’s Second Annual Conference on Leadership, Colorado Springs, CO. Tversky, A., & Kahnemann, D. (1973). Availability: A heuristic for judging frequency and
probability. Cognitive Psychology, 5,207-232. Vroom, V.H., (1964). Work and motivation. New York: Wiley. Waldman, D.A. (1987). Reducing the incongruence between formal organization and the
development of individuals: A case for transformational leadership. Paper presented at the
47th meeting of the Academy of Management, New Orleans, LA. Waldman, D.A., Bass, B.M., & Einstein, W.O. (1987). Leadership and outcomes of performance
appraisal processes. Journal of Occupational Psychology, 60, 177- 186.
Transformational vs. Transactional Leadership 425
Waldman, D.A., Bass, B.M., & Yammarino, F.J. (1990). Adding to contingent-reward behavior: The augmenting effects of charismatic leadership. Group & Organization Studies, Z5(4), 381-394.
Wanous, J.P., Sullivan, S.E., & Malinak, J. (1989). The role of judgment calls in meta-analysis. Journal of Applied Psychology, 74,259-264.
Wanous, J.P., Poland, T.D., Premack, S.L., & Davis, KS. (1992). The effects of met expectations on newcomer attitudes and behaviors: A review and meta-analysis. Journal of Applied Psychology, 77,288-297.
Whitener, E.M. (1990). Confusion of confidence intervals and credibility intervals in meta-analysis. Journal of Applied Psychology, 75(3), 3 15-32 1.
Wofford, J.C., & Liska, L.Z. (1993). Path goal theories of leadership: A meta-analysis. Journalof Management, 19(4), 857-876.
Yammarino, F.J., & Bass, B.M. (1990a). Long-term forecasting and of transformational leadership and its effects among naval officers: Some preliminary findings. In K.E. Clark and M.B. Clark (eds.), Measures of leadership (pp. 151-169). West Orange, NY: Leadership Library of America.
Yammarino, F.J., & Bass, B.M. (1990b). Transformational leadership and multiple levels of analysis. Human Relations, 43( lo), 975995.
Yammarino, F.J., & Dublinsky, A.J. (1994). Transformational leadership theory: using levels of analysis to determine boundary conditions. Personnel Psychology, 47, 787-811.
Young, M.N. (1990). Transformational leadership behaviors of male and female academic deans (Doctoral dissertation, Bowling Green State University). Dissertation Abstracts International, 52,082lA.
Yukl, G.A. (1989). Managerial leadership: A review of theory and research. Journal of Management, Z5(2), 251-289.
