role theory approaches for effectiveness of marketing...
TRANSCRIPT
147
CHAPTER 6
ROLE THEORY APPROACHES FOR EFFECTIVENESS OF MARKETING-ORIENTED
BOUNDARY SPANNERS
Comparative Review, Configural Extension, and Potential Contributions
JAGDIP SINGH AND ARGUN SAATCIOGLU
Abstract
Role theory has proved remarkably promising in examining effectiveness of marketing-oriented boundary spanners. This paper reviews different approaches for examining role theory implica-tions for boundary spanners—namely universalistic and contingency approaches—and develops the configural approach by extending configurational theory principles to role theory. Neither the contingency nor the configural approach has received much attention in the marketing literature. We compare and contrast different approaches, outlining bodies of work that have remained less accessible to marketing researchers. By triangulating across the alternative approaches, we expose underlying assumptions and press for critical assessment of their ecological validity. We identify op-portunities for potential contributions by exploring promising but as yet uncharted approaches.
Marketing oriented boundary spanners such as salespeople, frontline, and customer contact em-ployees fill critical roles that influence organizational effectiveness and sustainability. Consider the following:
• Boundary spanners are strategically important because they represent the “face” of the or-ganization to customers and public, and are critical nodes where knowledge about markets and consumers is accumulated.
• Boundary spanning work is rarely routinized and involves significant people-oriented work. Boundary spanners are required to constantly interact with customers, undertake tasks that involve emotional labor, and provide discretion to tailor their behaviors to individual customer needs, problems, and demands.
• Boundary spanning work is sensitive to internal and external organizational environments. Variation in consumer demands (e.g., seasonal and/or economic variations in demand for products/services) and in internal operations (e.g., new product/service introductions or interface technologies) often affects boundary spanners unpredictably.
148 RUTH N. BOLTON AND CRINA O. TARASI
• Boundary spanners are organizationally monitored and controlled (e.g., via human supervi-sion, electronic, audio, and video devices). Organizations are increasingly concerned about the productivity of boundary spanners, while keeping the quality of customer service delivered in focus.
• Boundary spanning work is highly stressful. Such work is likened to a “three-cornered fight” with the customer (demanding attention and service) and the organization (demanding effi-ciency and productivity) at the two ends and the boundary spanner “caught in the middle.”
• Boundary spanning roles are profit centers. They are expected to cross-sell, up-sell, and more-sell while in the process of providing high-quality service/information. This dual ac-countability injects competing pressures on boundary spanners.
Of the various theories applied to study effectiveness of marketing-oriented boundary span-ners, role theory is arguably the most promising so far. With its roots in sociology dating back to the fifties (Merton 1949; Rommetveit 1954), early grounded research on work organizations can be traced to the sixties. The much-cited work of Kahn et al. (1964) and Belasco’s (1966) research with salespeople were important steps in translating sociological notions of role theory into meaningful and relevant constructs for the study of marketing-oriented boundary spanners. For instance, independent of Kahn et al., Belasco (1966) made some important observations on the different role demands experienced by salespeople: (a) intellectual demands that require intel-ligence, problem-solving skills, and job knowledge abilities, (b) emotional demands from dealing with issues such as “advocacy conflict”—internal conflict from being an advocate for the customer and the company at the same time, and (c) interactional demands that arise from the intensity and adaptability required in the diverse range of interactions. Without effective coping mechanisms, Belasco feared that salesperson effectiveness would be seriously undermined regardless of their intelligence, job knowledge, and/or skills. Although Kahn et al. and Belasco provided rich theory for probing boundary spanning roles, scientific progress lingered till Rizzo, House, and Lirtzman (1970) published validated scales of role conflict and ambiguity, and stirred up research interest in this topic. By 1985, Jackson and Schuler reported over 200 articles on role conflict and ambiguity in organizational settings that were published between 1970 and 1983; of these, 96 were original empirical studies that they meta-analyzed. A few years later, Brown and Peterson (1993) were able to locate 59 studies that focused specifically on salesperson role conflict and ambiguity and its influence on performance and satisfaction.
Despite this volume of research, academic and practitioner perspectives on boundary role stress are defined by convergence and contrasts. Both perspectives converge on the view that bound-ary role stressors incur heavy costs for the organization and individual alike because of lowered productivity, reduced motivation and commitment, and increased health costs (Cavanaugh et al. 2000; Maslach and Leiter 1997). In some professions, especially involving frontline and customer contact work, stressors have been described as reaching epidemic proportions (Marino 1997). Contrasting perspectives emerge when the influence of boundary role stressors is considered. In contrast to the convergence in the academic literature, practitioners have long argued about the potential of boundary role stressors to promote performance, enhance motivation, and spark cre-ativity (Newton 1995; Mohrman and Cohen 1995). With regularity, the popular press has fancied workplace mantras such as, “it is better to burn out than to rust out,” presumably to assure boundary spanners that they are not alone in facing stress and that stress can be turned into an opportunity to develop and enhance oneself. Paradoxically, while this notion of “eustress” has deep roots in the academic literature (Yerkes and Dodson 1908; Selye 1976), empirical studies have generally produced weak and mixed evidence. Thus, while much academic research suggests redesigning
MANAGING CUSTOMER RELATIONSHIPS 149
and reconfiguring boundary work in a way that reduces, if not eliminates, critical role stressors (Tubre and Collins 2000), practitioners view such recommendations with little relevance since they hold that the nature of customer interface (e.g., people-oriented, nonroutinized work) and its unpredictability (e.g., variability in internal and external conditions) make role stress an inherent aspect of boundary roles.
To bridge these perspectives, a promising approach has been proposed by Karasek (1979) and his colleagues that views a singular focus on role stressors as myopic and misguided. Instead, it argues that the study of role stressors must simultaneously consider the job scope—the degree of autonomy, feedback, and participation afforded to boundary role employees (Karasek 1979; Xie and Johns 1995). Noting that greater job scope may make all the difference between “eustress” and “distress,” and between “healthy” and “unhealthy” work, this approach is theoretically appealing because of its conceptual richness, and managerially attractive as evidenced by the popularity of empowerment programs. Unfortunately, such stressor–scope models have received limited empiri-cal attention in the marketing literature. As such, the potential of the stressor–scope framework to provide insights and bridge perspectives is largely unrealized.
The purpose of this review is to illuminate and strengthen the preceding bridge to germinate new research ideas and directions for understanding the effectiveness of marketing-oriented boundary spanners. Specifically, we provide a review of three different theoretical perspectives on boundary role stress and effectiveness, including universalistic, contingency, and configural perspectives. The universalistic perspective reflects much current research in marketing and is grounded in the role episode model of Kahn et al. (1964). The contingency perspective is based on contributions to Karasek’s model. Because many of these contributions have occurred outside the marketing literature, we provide a detailed discussion of the theory underlying this perspective, and review the associated empirical literature. Finally, we develop the configural perspective as a theoretical contribution of this paper. This perspective extends ideas from configurational approaches to posit nonlinear and higher-order effects of role theory that cannot be represented by contingency ap-proaches. While these perspectives have competing elements, our orientation is comparative and complementary. We compare these perspectives for their theoretical distinctiveness to encourage future research that approaches the study of boundary role stress and effectiveness from multiple, not singular, perspectives to uncover convergent and anomalous ideas. Focusing on a singular theoretical perspective, as reflected in much marketing literature, limits the vision of understanding. In addition, at their boundaries, these theoretical perspectives offer opportunities for interesting and creative work that has remained as yet untapped and unexploited. We provide an outline for future research directions to this end.
Theoretical Perspectives on Role Stressors and Boundary Spanning Roles
Universalistic Perspective
This perspective posits that role stressors invariably have dysfunctional consequences for boundary spanner outcomes including performance, satisfaction, and commitment, regardless of job context, scope, and/or nature of the organization.
Theory
Rooted in Kahn et al.’s work, this perspective posits that role stressors have a linear relationship with boundary spanner outcomes. The commonly examined role stressors include role conflict,
150 RUTH N. BOLTON AND CRINA O. TARASI
role ambiguity, and role overload (Behrman and Perrault 1984; Belasco 1966; Kahn et al. 1964). Role conflict occurs when a boundary spanner believes that the expectations and demands of two or more members of his or her role set are incompatible (e.g., boss and customer). Role ambigu-ity relates to the perceived lack of information needed by an employee to perform his or her role adequately and his or her uncertainty about the expectations of different role set members. Role overload occurs when the frontline employee perceives that the cumulative role demands exceed his or her abilities and motivation to perform a task. The influence of different role stressors on boundary spanner performance and well-being is supported conceptually by the role episode model of Kahn and colleagues (1964), which posits that (1) boundary spanners interact with different role senders (e.g., customers, boss, co-workers) in many episodes to obtain information, direction, task demands, and assistance; (2) role sender demands and expectations take the form of perceived stressors when a boundary spanner believes that there is conflict (e.g., among demands), ambiguity (e.g., about expectations), or overload (e.g., of demands and expectations); (3) perceived stressors are influenced by a person’s psychological, dispositional, and sociological characteristics; and (4) persistent stressors are likely to overwhelm the person’s resources and thereby have a dysfunctional impact on his or her behavioral and psychological outcomes (e.g., job performance, satisfaction). Hobfoll and Freedy (1993) have conceptualized the influence of role stressors within a conservation of resources framework. Boundary spanners are thought to regulate their behaviors to cope with role stressors in a way that conserves their valued resources; however, regulation failures occur when stressors overwhelm an individual’s coping resources, resulting in impaired performance and well-being. While it allows for the possibility that different boundary spanners may perceive disparate levels of role stressors in similar work situations, this perspective is universalistic in its predictions about the linear and direct effect of perceived role stressors on outcomes.
Empirical Findings and Assessment
This perspective has produced significant empirical work summarized in several meta-analyses and reviews (Jackson and Schuler 1985; Fisher and Gitelson 1983; King and King 1990; Brown and Peterson 1993). The preponderance of evidence suggests that the influence of role stressors is consistent, compelling, and invariably dysfunctional (Singh 1993; Brown and Peterson 1993; Rhoads, Singh, and Goodell 1994; Behrman and Perreault 1984; Fisher and Gitelson 1983; Spec-tor, Dwyer, and Jex 1988). For example, Brown and Peterson (1993), in their meta-study, found correlations of –.24, –.36, –.28, and .36 between role ambiguity and job performance, satisfaction, commitment, and propensity to leave, respectively. Likewise, the correlations between role conflict and job performance, satisfaction, commitment, and propensity to leave were –.07, –.33, –.34, and .28 respectively. Although role overload is not as frequently studied as other role stressors, in general the correlations between role overload and the different job outcomes parallel those obtained for role conflict and ambiguity in terms of both magnitude and direction. For instance, Singh et al. (1994) report correlations of –.14, –.39, .25, and .09 between role overload and job satisfaction, organizational commitment, turnover intentions, and job performance, respectively. As such, there is broad evidence supported by meta-analytical results that role stressors have significant linear and dysfunctional relationships with critical job outcomes (Fisher and Gitelson 1983; Jackson and Schuler 1985; Brown and Peterson 1993).
While researchers have called for exploring moderating variables, the majority of the empirical work has downplayed the effect of situational or contextual variability. For instance, in Brown and Peterson’s meta-analysis, the influence of supervisory behaviors and job/task variables accounted for less than 10 percent of explained variance. Consequently, Brown and Peterson (p. 68) claimed
MANAGING CUSTOMER RELATIONSHIPS 151
that their role stress model that excludes job context variables indicates “considerable robustness and generalizability . . . across relationships and study contexts.” Likewise, in another meta-analy-sis, Churchill et al. (1985, p. 109) found that, on average, organizational context variables explain “only 1% of the variation in performance” and that this influence was the “lowest . . . among the six categories of predictors studied.” These findings parallel Jackson and Schuler’s (1985) meta-analysis in that contextual variables have significant, negative but rather weak ( << 15 percent shared variance) direct effects on role stressors, with marginal direct effects on job outcomes.
Contingency Perspective
In accord with this perspective, the effect of role stressors on boundary spanner outcomes is con-tingent on a third variable, such that this effect may be dysfunctional, neutral, or even functional, depending on the level of the contingent factor. Within this perspective, different theoretical models have been proposed that specify the contingent variable and the mechanism involved in moderat-ing the effect of role stressors. The most commonly used theoretical frameworks are rooted in Karasek’s (1979) demand–latitude model displayed in Figure 6.1.
Theory
In Karasek’s model, the influence of psychological demands—or role stressors in the context of boundary spanners—on job outcomes is contingent on the availability of decision latitude to the individual (e.g., autonomy)—also referred to as job scope or job control. Karasek suggests that certain modalities of the demand–latitude interplay result in higher job outcomes than do other modalities. In particular, Karasek asserts that increasing levels of psychological job demands must be matched with increasing levels of decision latitude for maintaining or enhancing performance and psychological well-being. Utilizing “low” and “high” distinctions for demands and latitude (see Figure 6.1), Karasek developed the logic for the underlying mechanism for the differing influence
Figure 6.1 A Graphical Representation of Karasek’s Job Demands (Role Stressors)– Decision Latitude Model
Low-Strain Active
Passive High-Strain
Job Demands
Low High
High
JobLatitude
Low
3 4
1 2
152 RUTH N. BOLTON AND CRINA O. TARASI
of role stressors. When latitude is low and boundary spanners have little discretion in making task decisions, Karasek hypothesized that the job context would be either “passive” or “high-strain” corresponding to a “low” or “high” level of role stressors, respectively. Specifically, when role stressors are high, a “high-strain” work context is obtained because employees lack the necessary coping resources that come from decision latitude to deal with high levels of role demands, and are easily overwhelmed (see cell 2 in Figure 6.1). In a high-strain environment, boundary spanner performance is impaired and well-being undermined. By contrast, when role stressors are at a low level, with decision latitude also at a low level, Karasek suggested that a passive work context is obtained where both performance and well-being are suboptimal (see cell 1 in Figure 6.1). Why so? Drawing from alienation theory (Blood and Hulin 1967), Karasek reasoned that employees generally lack the stimulation to actively engage in tasks when role stressors are low. This passive orientation toward tasks is exacerbated by low levels of autonomy that fail to provide a sense of control over decisions that affect an individual’s job, thereby inhibiting employee efforts to insert challenge and growth in low-stress jobs. Consequently, boundary spanning jobs with low levels of stressors and latitude are posited to be passive and suboptimal.
Now consider the contingencies when decision latitude is high (see cells 3 and 4 in Figure 6.1). Karasek predicted that the job context would be either “active” or “low-strain” corresponding to a high or low level of role stressors respectively. Specifically, when role stressors are low, Karasek reasoned that boundary spanners possess significant resources that stem from autonomy that can be deployed to address challenging job demands. However, the low level of stressors offers little by way of challenges to channel individual resources. This abundance of underutilized resources makes for a low-strain job context. Karasek did not view such low-strain jobs favorably. Rather, he argued that such job contexts lack the potential to grow individual skills and enhance self-efficacy and, conse-quently, are suboptimal (cell 3, Figure 6.1). By contrast, Karasek posited a favorable perspective for job contexts with high levels of role stressors. Building from motivation theories (Csikszentmihalyi 1975; Hackman and Oldham 1976), Karasek noted that employees with high levels of task control are likely to enjoy resources needed to cope with challenging demands. When the job context sup-plies these demands under conditions of “high” role stressors, the boundary spanner can draw from available resources to effectively cope with the challenge. Because effectively dealing with chal-lenging job demands is efficacious, the boundary spanner is likely to grow from this experience by being more self-confident, resourceful, and energized to tackle future challenges (Csikszentmihalyi 1975). Karasek noted that such job contexts are “active” (cell 4, Figure 6.1).
Empirical Findings
A distinctive aspect of Karasek’s theory is that it depicts “interaction effects” of role stressors and decision latitude on job outcomes. As such, in most cases, Karasek’s hypothesis is tested by examining the significance of the interaction term involving appropriate measures for job demands and latitude, with performance or other job outcomes as the dependent variable. In empirical tests of Karasek’s (1979) theory across a wide range of work contexts using a variety of designs (see Table 6.1 for a summary), the results for the interaction effects have been mixed. For instance, studies by Landsbergis (1988), Abdel-Halim (1981), Daniels and Guppy (1994), and Kelloway and Barling (1991) both validated and extended Karasek’s model. Landsbergis (1988) found that job strain (e.g., dissatisfaction, depression, and psychosomatic symptoms) and burnout were higher in health care jobs that combined high job (workload) demands and job control (decision latitude). Likewise, Abdel-Halim (1981) found support for the interactive effects of role stressors
(text continues on page 171)
MANAGING CUSTOMER RELATIONSHIPS 153Ta
ble
6.1
A S
um
mar
y o
f K
ey E
mp
iric
al S
tud
ies
Usi
ng
a C
on
tin
gen
cy P
ersp
ecti
ve f
or
Exa
min
ing
Ro
le T
heo
ry E
ffec
ts f
or
B
ou
nd
ary
Sp
ann
ers
Aut
hors
Mod
el/V
aria
bles
Util
ized
Met
hodo
logi
cal A
ppro
ach
Sam
ple
Fin
ding
s
Abd
el-H
alim
D
epen
dent
var
iabl
es:
Mul
tiple
line
ar r
egre
ssio
n 89
ran
dom
ly
Inte
ract
ions
of
role
am
bigu
ity a
nd
(197
8)an
alys
is u
sing
inte
ract
ion
sele
cted
line
and
M
PS
(F
= 5
.33,
p <
. 05)
and
of
role
Jo
b sa
tisfa
ctio
nte
rms
staf
f m
anag
ers
over
load
and
MP
S (
F =
10.
93,
p <
fr
om s
ix u
nits
of
a .0
1) s
igni
fican
tly p
redi
cted
sat
isfa
ctio
n Jo
b in
volv
emen
tla
rge
heav
y-(R
2 =
.44
, F
= 5
.57,
p <
.05
). In
low
eq
uipm
ent
MP
S s
ituat
ions
(i.e
., lo
w jo
b Jo
b an
xiet
ym
anuf
actu
ring
enric
hmen
t),
incr
easi
ng le
vels
of
role
co
mpa
nyam
bigu
ity a
nd r
ole
over
load
red
uced
In
depe
nden
t va
riabl
es:
satis
fact
ion.
In
high
MP
S s
ituat
ions
, ro
le
ambi
guity
slig
htly
red
uced
sat
isfa
ctio
n S
kill
varie
ty,
task
iden
tity,
tas
k w
here
as r
ole
over
load
con
side
rabl
y si
gnifi
canc
e, a
uton
omy,
and
fee
dbac
k in
crea
sed
satis
fact
ion.
from
job
as m
oder
atin
g jo
b en
richm
ent
char
acte
ristic
s co
mbi
ned
into
T
he in
tera
ctio
n of
rol
e am
bigu
ity w
ith
Mot
ivat
ing
Pot
entia
l Sco
re (
MP
S)
MP
S (
F =
5.2
5, p
< .
05)
sign
ifica
ntly
pr
edic
ted
job
invo
lvem
ent
(R2 =
.24
, F
=
Rol
e co
nflic
t, ro
le a
mbi
guity
, an
d ro
le
5.20
, p
< .
05).
Rol
e am
bigu
ity
over
load
as
thre
e di
stin
ct m
easu
res
of
decr
ease
d in
volv
emen
t in
bot
h lo
w a
nd
role
str
ess
high
MP
S s
ituat
ions
, bu
t m
uch
less
in
the
latte
r th
an in
the
for
mer
.
Fin
ally
, th
e in
tera
ctio
n of
rol
e ov
erlo
ad
and
MP
S (
F =
6.6
2, p
< .
05)
sign
ifica
ntly
pr
edic
ted
job
anxi
ety
(R2 =
.26
, F
= 6
.57,
p
< .
05).
Rol
e ov
erlo
ad in
crea
sed
anxi
ety
in lo
w M
PS
situ
atio
ns b
ut
redu
ced
it in
hig
h M
PS
situ
atio
ns.
(con
tinue
d)
154 RUTH N. BOLTON AND CRINA O. TARASIA
bdel
-Hal
im
Dep
ende
nt v
aria
ble:
Thr
ee-w
ay a
naly
sis
of
170
rand
omly
T
hree
-way
inte
ract
ions
of
role
(1
981)
varia
nce
anal
ysis
with
mai
n,
sele
cted
man
ager
ial
ambi
guity
, jo
b ch
arac
teris
tics,
and
S
atis
fact
ion
with
wor
ktw
o-w
ay,
and
thre
e-w
ay
and
non-
man
ager
ial
tech
nolo
gy (
F =
5.
43,
p ≤
.02)
and
rol
e in
tera
ctio
n ef
fect
spe
rson
nel—
89 f
rom
ov
erlo
ad,
job
char
acte
ristic
s, a
nd
Inde
pend
ent
varia
bles
:a
heav
y eq
uipm
ent
tech
nolo
gy (
F =
5.7
3, p
≤ .
01)
wer
e m
anuf
actu
ring
firm
st
atis
tical
ly s
igni
fican
t. In
sim
ple,
low
-R
ole
confl
ict,
role
am
bigu
ity,
and
role
an
d 81
fro
m fi
ve
scop
e jo
bs in
long
-link
ed m
anuf
actu
ring
over
load
as
thre
e di
stin
ct m
easu
res
of
smal
l to
med
ium
te
chno
logi
es a
nd in
com
plex
, hi
gh-
role
str
ess
size
ban
kssc
ope
jobs
in m
edia
ting
serv
ice
tech
nolo
gies
, th
e in
tera
ctio
n of
tw
o ro
le
Ski
ll va
riety
, ta
sk id
entit
y, t
ask
stre
ssor
s (a
mbi
guity
and
ove
rload
) an
d si
gnifi
canc
e, a
uton
omy,
and
fee
dbac
k jo
b ch
arac
teris
tics
wer
e fo
und
to b
e as
five
dis
tinct
mea
sure
s of
job
desi
gn
sign
ifica
nt. T
he t
hree
-way
inte
ract
ions
ch
arac
teris
tics/
job
com
plex
ityin
volv
ing
role
con
flict
, an
d al
l tw
o-w
ay
inte
ract
ions
wer
e no
nsig
nific
ant.
Org
aniz
atio
nal t
echn
olog
y (n
omin
ally
m
easu
red
in t
wo
cate
gorie
s—lo
ng-
The
se r
esul
ts in
dica
te t
hat
linke
d m
anuf
actu
ring
tech
nolo
gy o
r or
gani
zatio
nal t
echn
olog
y de
term
ines
m
edia
ting
serv
ice
tech
nolo
gy)
the
man
ner
in w
hich
job
char
acte
ristic
s an
d ro
le s
tres
sors
inte
ract
.
Pay
ne a
nd
Kar
asek
’s (
1979
) de
man
d-co
ntro
l C
lust
er a
naly
sis
usin
g W
ard’
s 14
8 te
ache
rs f
rom
C
lust
er a
naly
sis
base
d on
the
F
letc
her
(198
3)(d
iscr
etio
n) m
odel
met
hod,
AN
OV
A,
and
mul
tiple
va
rious
tow
ns in
the
in
depe
nden
t va
riabl
es id
entifi
ed s
even
re
gres
sion
Mid
land
s of
Eng
land
grou
ps w
ith d
istin
ct c
ombi
natio
ns o
f D
epen
dent
var
iabl
es:
dem
ands
, co
ntro
l, an
d su
ppor
t. A
NO
VA
ac
ross
the
sev
en g
roup
s in
dica
ted
that
S
atis
fact
ion
with
wor
k,
ther
e w
ere
sign
ifica
nt d
iffer
ence
s on
he
adm
aste
r/he
adm
istr
ess,
col
leag
ues,
sa
tisfa
ctio
n w
ith t
he h
eadm
aste
r, an
d w
ith t
he f
elt
pres
sure
on
the
job
colle
ague
s, p
ay,
and
the
felt
pres
sure
on
the
job.
Gro
up 2
(lo
w d
isci
plin
ary
Fel
t ad
equa
cy o
f pa
yde
man
ds,
aver
age
on o
ther
dem
ands
, hi
gh s
uppo
rt,
and
high
dis
cret
ion)
was
D
epre
ssio
nsi
gnifi
cant
ly (
p <
.05
) m
ore
satis
fied
than
all
othe
r gr
oups
. Ove
rall,
the
A
nxie
tyde
man
ds,
supp
orts
, di
scre
tion
mod
el
acco
unte
d fo
r a
smal
l pro
port
ion
of t
he
Obs
essi
onva
rianc
e in
the
dep
ende
nt v
aria
bles
(b
etw
een
10%
and
21%
of
the
Som
atic
com
plai
nts
satis
fact
ion
mea
sure
s). T
he t
hree
-way
Aut
hors
Mod
el/V
aria
bles
Util
ized
Met
hodo
logi
cal A
ppro
ach
Sam
ple
Fin
ding
s
MANAGING CUSTOMER RELATIONSHIPS 155in
tera
ctio
n of
dem
ands
, di
scre
tion,
and
C
ogni
tive
impa
irm
ents
supp
ort
wer
e no
t in
clud
ed in
the
m
ultip
le r
egre
ssio
n an
alys
is. T
he
Inde
pend
ent
varia
bles
:de
man
d/di
scre
tion
inte
ract
ion
(as
orig
inal
ly p
ropo
sed
by K
aras
ek)
did
not
Dis
cipl
inar
y de
man
ds,
mai
ntai
ning
si
gnifi
cant
ly p
redi
ct a
ny o
f th
e st
anda
rds,
and
wor
kloa
d de
man
ds a
s de
pend
ent
varia
bles
.th
ree
dist
inct
mea
sure
s of
job
dem
ands
Inte
rper
sona
l sup
port
Job
disc
retio
n
Spe
ctor
(19
87)
Kar
asek
’s (
1979
) de
man
d-co
ntro
l M
ultip
le r
egre
ssio
n an
alys
is,
136
cler
ical
E
ach
depe
nden
t va
riabl
e w
as r
egre
ssed
(d
iscr
etio
n) m
odel
usin
g in
tera
ctio
n te
rms
empl
oyee
s dr
awn
on c
ontr
ol,
a st
ress
or,
and
the
from
diff
eren
t in
tera
ctio
n of
con
trol
and
tha
t st
ress
or.
Dep
ende
nt v
aria
bles
:po
sitio
ns a
t a
maj
or
Out
of
a to
tal o
f 30
reg
ress
ion
anal
ysis
, st
ate
univ
ersi
tyon
ly 2
(7%
) ha
d in
tera
ctio
n te
rms
that
To
tal s
atis
fact
ion
and
gene
ral
wer
e si
gnifi
cant
, pr
ovid
ing
wea
k su
ppor
t sa
tisfa
ctio
n as
tw
o di
stin
ct m
easu
res
of
to K
aras
ek’s
orig
inal
inte
ract
ion
satis
fact
ion
hypo
thes
is. T
he f
orm
of
thes
e in
tera
ctio
ns w
ere
foun
d no
t to
Fr
ustr
atio
nco
rres
pond
to
thos
e re
quire
d by
K
aras
ek’s
hyp
othe
sis
eith
er.
Anx
iety
Hea
lth s
ympt
oms
(hav
ing
them
and
go
ing
to t
he d
octo
r fo
r th
ese
sym
ptom
s as
tw
o di
stin
ct m
easu
res)
Inde
pend
ent
varia
bles
:D
iscr
etio
n (c
ontr
ol)
Wor
kloa
d, r
ole
ambi
guity
, ro
le c
onfli
ct,
and
inte
rper
sona
l con
flict
as
dist
inct
m
easu
res
of jo
b de
man
ds; s
um o
f th
e st
anda
rdiz
ed s
core
s of
the
se m
easu
res
as a
com
bine
d m
easu
re o
f jo
b de
man
ds
(con
tinue
d)
156 RUTH N. BOLTON AND CRINA O. TARASILa
ndbe
rgis
K
aras
ek’s
(19
79)
dem
and-
cont
rol
MA
NO
VA
, A
NC
OV
A
771
empl
oyee
s A
t S
tep
1, M
AN
OV
A in
dica
ted
that
, (1
988)
mod
el(R
oy/B
argm
ann
step
-dow
n co
mpo
sed
of 1
5 ac
ross
all
the
depe
nden
t m
easu
res,
an
alys
is),
hie
rarc
hica
l mul
tiple
di
ffere
nt jo
b tit
les
in
ther
e w
ere
sign
ifica
nt d
iffer
ence
s (p
<
Dep
ende
nt v
aria
bles
:re
gres
sion
usi
ng in
tera
ctio
n 4
barg
aini
ng u
nits
in
.001
, df
= 3
3/69
0) b
etw
een
the
four
te
rms,
can
onic
al c
orre
latio
ns,
2 ho
spita
ls a
nd 1
qu
adra
nts
of K
aras
ek’s
orig
inal
mod
el.
Job
diss
atis
fact
ion,
sle
epin
g pr
oble
ms,
an
d t-
test
snu
rsin
g ho
me
AN
CO
VA
indi
cate
d th
at jo
b de
pres
sion
/life
dis
satis
fact
ion,
and
di
ssat
isfa
ctio
n, e
mot
iona
l exh
aust
ion,
ph
ysic
al/p
sych
osom
atic
str
ain
as f
our
Ste
p 1:
Exa
min
ing
mea
n an
d pe
rson
al a
ccom
plis
hmen
t di
ffere
d di
stin
ct m
easu
res
of p
sych
olog
ical
di
ffere
nces
of
depe
nden
t ac
ross
the
fou
r qu
adra
nts
(p <
.05
; F-
stra
inva
riabl
es a
cros
s K
aras
ek’s
va
lues
ran
ged
from
3.1
1 to
28.
87; d
f or
igin
al f
our
confi
gura
tions
of
rang
ed f
rom
3/2
34 t
o 3/
244)
. In
Cor
onar
y he
art
dise
ase
(CH
D)
dem
ands
and
latit
ude
usin
g pa
rtic
ular
, jo
b di
ssat
isfa
ctio
n an
d sy
mpt
om in
dica
tor
MA
NO
VA
and
AN
CO
VA
em
otio
nal e
xhau
stio
n w
ere
sign
ifica
ntly
pr
oced
ures
high
er a
mon
g hi
gh-s
trai
n em
ploy
ees,
Fr
eque
ncy
and
inte
nsity
of
emot
iona
l an
d th
e ac
tive
and
pass
ive
quad
rant
s ex
haus
tion,
dep
erso
naliz
atio
n, a
nd
Ste
p 2:
Pre
dict
ing
stra
in a
nd
did
not
diffe
r in
ter
ms
of p
sych
olog
ical
pe
rson
al a
ccom
plis
hmen
t as
six
dis
tinct
oth
er o
utco
mes
with
wor
kloa
d st
rain
mea
sure
s—bu
t ac
tive
empl
oyee
s m
easu
res
of b
urno
ut m
easu
red
by
dem
ands
, de
cisi
on la
titud
e,
did
repo
rt h
ighe
r em
otio
nal e
xhau
stio
n M
asla
ch B
urno
ut I
nven
tory
(M
BI)
dem
and–
latit
ude
inte
ract
ion,
th
an d
id p
assi
ve o
nes.
The
act
ive
soci
al s
uppo
rt,
job
inse
curit
y,
quad
rant
was
als
o ch
arac
teriz
ed b
y In
depe
nden
t va
riabl
es:
phys
ical
exe
rtio
n, a
nd h
azar
d hi
gh p
erso
nal a
ccom
plis
hmen
t ex
posu
reco
mpa
red
to a
ll ot
her
quad
rant
s.Jo
b ch
arac
teris
tics
(JC
S S
urve
y)S
tep
3: C
anon
ical
cor
rela
tion
At
Ste
p 2,
hie
rarc
hica
l reg
ress
ion
Ski
ll di
scre
tion,
dec
isio
n au
thor
ity,
with
dem
ogra
phic
s an
d jo
b in
dica
ted
that
the
dem
and–
latit
ude
unio
n in
fluen
ce,
and
part
icip
atio
n as
ch
arac
teris
tics
as in
depe
nden
t in
tera
ctio
n di
d no
t si
gnifi
cant
ly p
redi
ct
four
dis
tinct
mea
sure
s of
dec
isio
n va
riabl
es,
and
the
stra
in
any
of t
he f
our
mea
sure
s of
la
titud
esc
ales
, C
HD
indi
cato
r, ps
ycho
logi
cal s
trai
n. T
he m
ain
effe
cts
of
smok
ing,
and
bur
nout
de
man
d an
d la
titud
e, a
s w
ell a
s of
W
orkl
oad
dem
ands
mea
sure
s as
dep
ende
nt
soci
al s
uppo
rt,
haza
rd e
xpos
ure,
va
riabl
es
phys
ical
exe
rtio
n, a
nd jo
b se
curit
y S
uper
viso
r su
ppor
t an
d co
-wor
ker
sign
ifica
ntly
(p
< .
05)
pred
icte
d al
l of
the
supp
ort
as t
wo
dist
inct
mea
sure
s of
fo
ur s
trai
n m
easu
res
and
acco
unte
d fo
r so
cial
sup
port
sign
ifica
nt p
ortio
ns o
f va
riatio
n. F
or jo
b sa
tisfa
ctio
n, ∆
R2 r
ange
d fr
om .
046
to
Job
inse
curit
y.1
52 a
nd t
he r
esul
ting
R2 w
as .
433
(F =
6.
28,
p <
.00
1). F
or d
epre
ssio
n, ∆
R2
Aut
hors
Mod
el/V
aria
bles
Util
ized
Met
hodo
logi
cal A
ppro
ach
Sam
ple
Fin
ding
s
MANAGING CUSTOMER RELATIONSHIPS 157
(con
tinue
d)
Phy
sica
l exe
rtio
nra
nged
fro
m .
020
to .
065
and
the
resu
lting
R2 w
as .
163
(F =
4.2
2, p
<
Haz
ard
expo
sure
.01)
. For
phy
sica
l str
ain,
∆R
2 ran
ged
from
.01
8 to
.07
3, t
he r
esul
ting
R2 w
as
Job
Str
ain
(JC
S S
urve
y).1
95 (
F =
4.8
8, p
< .
001)
. For
sle
epin
g pr
oble
ms,
∆R
2 ran
ged
from
.01
1 to
.04
6,
Dep
ress
ion/
life
diss
atis
fact
ion
the
resu
lting
R2 w
as .
133
(F =
2.5
8, p
<
.05)
. The
R2 f
or C
HD
was
.13
and
the
Jo
b di
ssat
isfa
ctio
nde
man
d–la
titud
e in
tera
ctio
n di
d si
gnifi
cant
ly in
crea
se t
he R
2 .P
hysi
cal/p
sych
osom
atic
str
ain
Sle
epin
g pr
oble
ms
Sim
ilar
resu
lts w
ere
obta
ined
for
the
bu
rnou
t sc
ales
as
depe
nden
t va
riabl
es.
The
dem
and–
latit
ude
inte
ract
ion
did
not
sign
ifica
ntly
pre
dict
any
of
the
stra
in
mea
sure
s. W
orkl
oad
dem
ands
si
gnifi
cant
ly in
crea
sed
R2 f
or e
mot
iona
l ex
haus
tion
and
depe
rson
aliz
atio
n bu
t no
t pe
rson
al a
ccom
plis
hmen
t. La
titud
e in
crea
sed
R2 f
or a
ll bu
rnou
t sc
ales
, an
d un
ion
activ
ity s
igni
fican
tly in
crea
sed
R2
exce
pt f
or p
erso
nal a
ccom
plis
hmen
t. T
he r
esul
ting
R2 s
for
emot
iona
l ex
haus
tion,
dep
erso
naliz
atio
n, a
nd
pers
onal
acc
ompl
ishm
ent
wer
e .4
0, .
22
to .
23,
and
.25
to .
38,
resp
ectiv
ely.
Per
rew
e an
d G
anst
er (
1989
)K
aras
ek’s
(19
79)
dem
and-
cont
rol
mod
el,
and
a w
orkl
oad-
cont
rol m
odel
Mul
tiple
reg
ress
ion
anal
ysis
us-
ing
inte
ract
ion
term
s12
5 vo
lunt
eer
unde
r-gr
adua
te
Obj
ectiv
e de
man
ds h
ad a
sig
nific
ant
mai
n ef
fect
onl
y on
sat
isfa
ctio
n (β
= –
stud
ents
enr
olle
d in
.9
4, F
= 1
4.52
, p
< .
001,
R2 =
.12)
usi
ng
Dep
ende
nt v
aria
bles
:an
intr
oduc
tory
on
ly t
he F
aces
sca
le. O
bjec
tive
cont
rol
man
agem
ent
cour
seha
d no
sig
nific
ant
mai
n ef
fect
s on
any
of
Task
sat
isfa
ctio
n, m
easu
red
usin
g G
M
the
depe
nden
t va
riabl
es.
Face
s S
cale
and
Sem
antic
Diff
eren
tial
scal
eP
erce
ived
dem
ands
had
sig
nific
ant
nega
tive
mai
n ef
fect
s on
sat
isfa
ctio
n P
erce
ived
anx
iety
mea
sure
d us
ing
usin
g bo
th t
he S
eman
tic D
iffer
entia
l A
ffect
Adj
ectiv
e C
heck
Lis
t (A
AC
L) a
nd
scal
e (β
= –
.29,
F =
9.4
6, p
< .
01,
R2 =
158 RUTH N. BOLTON AND CRINA O. TARASIS
ubje
ctiv
e S
tres
s sc
ale
.08)
and
the
Fac
es s
cale
(β
= –
.50,
F =
24
.45,
p <
.00
1, R
2 = .
18).
Per
ceiv
ed
Pul
se r
ate
and
skin
tem
pera
ture
as
two
dem
ands
als
o ha
d si
gnifi
cant
pos
itive
di
stin
ct m
easu
res
of p
sych
olog
ical
ef
fect
s on
per
ceiv
ed a
nxie
ty u
sing
bot
h ar
ousa
lth
e S
ubje
ctiv
e S
tres
s sc
ale
(β =
.20
, F
=
15.5
6, p
< .
001,
R2 =
.13
) an
d A
AC
L (β
In
depe
nden
t va
riabl
es:
= .
52,
F =
45.
26,
p <
.00
1, R
2 = .
29).
Per
ceiv
ed c
ontr
ol h
ad a
sig
nific
ant
mai
n O
bjec
tive
and
perc
eive
d be
havi
oral
ef
fect
onl
y on
anx
iety
usi
ng t
he A
AC
L (β
co
ntro
l=
–.0
2, F
= 4
.06,
p <
.00
1, ∆
R2
= .
03,
R2
= .
32).
The
find
ings
on
the
mai
n ef
fect
s O
bjec
tive
and
perc
eive
d w
orkl
oad
of d
eman
d an
d co
ntro
l wer
e la
rgel
y de
man
dsco
nsis
tent
with
Kar
asek
’s o
rigin
al
post
ulat
ions
.
As
far
as t
he in
tera
ctio
n ef
fect
s w
ere
conc
erne
d, o
nly
the
perc
eive
d de
man
d/pe
rcei
ved
cont
rol i
nter
actio
n w
as f
ound
to
have
a s
igni
fican
t ef
fect
. T
his
effe
ct w
as o
n pe
rcei
ved
anxi
ety
(β
= –
.07,
F =
2.8
7, p
< .
10,
∆R2 =
.02
, R
2 =
.17)
. Con
trol
ling
for
the
leve
l of
perc
eive
d co
ntro
l (lo
w,
med
ium
, hi
gh),
it
was
fou
nd t
hat
the
rela
tions
hip
betw
een
perc
eive
d de
man
ds a
nd p
erce
ived
an
xiet
y be
cam
e w
eake
r as
per
ceiv
ed
cont
rol g
ot h
ighe
r.
Aut
hors
Mod
el/V
aria
bles
Util
ized
Met
hodo
logi
cal A
ppro
ach
Sam
ple
Fin
ding
s
MANAGING CUSTOMER RELATIONSHIPS 159
(con
tinue
d)
War
r (1
990)
Kar
asek
’s (
1979
) de
man
d-co
ntro
l M
ultip
le r
egre
ssio
n an
alys
es
1,68
6 em
ploy
ees
of
Bot
h jo
b de
man
ds a
nd d
ecis
ion
latit
ude
(dec
isio
n la
titud
e) m
odel
and
War
r’s
usin
g qu
adra
tic a
nd in
tera
ctio
n re
sear
ch c
ompa
ny
wer
e fo
und
to h
ave
sign
ifica
nt li
near
(1
987)
tw
o-di
men
sion
al a
ffect
ive
wel
l-te
rms
in 7
5 lo
catio
ns in
the
m
ain
effe
cts
on a
ll th
ree
mea
sure
s of
be
ing
mod
el t
hat
incl
udes
anx
iety
, U
Kw
ell-b
eing
(βs
ran
ged
from
.60
to
1.37
, de
pres
sion
, pl
easu
re,
and
arou
sal
p <
.01
), e
xcep
t fo
r th
e ef
fect
of
deci
sion
la
titud
e on
anx
iety
-con
tent
men
t.D
epen
dent
var
iabl
es:
Job
dem
ands
was
fou
nd t
o ha
ve a
A
nxie
ty-c
onte
ntm
ent,
depr
essi
on-
nonl
inea
r (q
uadr
atic
) ef
fect
on
job
enth
usia
sm,
plea
sure
-dis
plea
sure
(jo
b sa
tisfa
ctio
n (β
= –
1.28
, si
gnifi
canc
e of
sa
tisfa
ctio
n) a
s th
ree
dist
inct
mea
sure
∆R
2 <
.01
, R
2 =
.19
), a
nxie
ty-c
onte
ntm
ent
of a
ffect
ive
wel
l-bei
ng(β
= –
.99,
sig
nific
ance
of
∆R2
< .
01,
R2
=
.29)
, an
d de
pres
sion
-ent
husi
asm
(β
= –
Inde
pend
ent
varia
bles
:.8
5, s
igni
fican
ce o
f ∆R
2 <
.01
, R
2 =
.12
). Jo
b de
man
dsD
ecis
ion
latit
ude
had
a no
nlin
ear
effe
ct
Dec
isio
n la
titud
eon
ly o
n jo
b sa
tisfa
ctio
n (β
= –
.46,
si
gnifi
canc
e of
∆R
2 <
.01
, R
2 =
.42
). F
inal
ly,
the
inte
ract
ion
of jo
b de
man
ds
and
deci
sion
latit
ude
did
not
have
a
sign
ifica
nt e
ffect
on
any
of t
he t
hree
m
easu
res
of w
ell-b
eing
.
The
se r
esul
ts p
rovi
de s
ubst
antia
l su
ppor
t fo
r K
aras
ek’s
orig
inal
idea
s an
d fin
ding
s re
gard
ing
the
nonl
inea
r ef
fect
s of
dem
and
and
latit
ude
on k
ey jo
b ou
tcom
es. H
owev
er,
they
pro
vide
no
supp
ort
for
his
hypo
thes
is r
egar
ding
the
in
tera
ctiv
e ef
fect
s of
dem
ands
and
co
ntro
l.
160 RUTH N. BOLTON AND CRINA O. TARASID
wye
r an
d K
aras
ek’s
(19
79)
dem
and-
cont
rol
Mul
tiple
reg
ress
ion
anal
ysis
, 11
5 fu
ll-tim
e F
our
sign
ifica
nt in
tera
ctio
n ef
fect
s w
ere
Gan
ster
(19
91)
mod
elus
ing
inte
ract
ion
term
sw
orke
rs (
blue
-col
lar
foun
d. T
he in
tera
ctio
n be
twee
n co
ntro
l an
d tr
ade
and
perc
eive
d w
orkl
oad
dem
ands
had
a
Dep
ende
nt v
aria
bles
:oc
cupa
tions
) in
a
nega
tive
effe
ct o
n ab
senc
e (β
= –
1.00
, la
rge
man
ufac
turin
g F
(1,8
6) =
16
.12,
p <
.01
, ∆R
2 = .
15)
and
a A
bsen
ce,
tard
ines
s, a
nd s
ick
days
as
plan
tpo
sitiv
e ef
fect
on
satis
fact
ion
(β =
.43
, th
ree
dist
inct
mea
sure
s of
em
ploy
ee
F(1
,86)
= 4
.05,
p <
.05
, ∆R
2 = .
04),
w
ithdr
awal
indi
catin
g th
at a
t hi
gh le
vels
of
cont
rol,
perc
eive
d w
orkl
oad
is a
ssoc
iate
d w
ith
Job
satis
fact
ion
(ove
rall)
and
wor
k lo
w a
bsen
teei
sm a
nd h
igh
wor
k sa
tisfa
ctio
n (a
spec
ts o
f th
e w
ork)
as
satis
fact
ion.
Und
er lo
w c
ondi
tions
of
two
dist
inct
mea
sure
s of
sat
isfa
ctio
nco
ntro
l, ho
wev
er,
wor
kloa
d te
nds
to
incr
ease
abs
ente
eism
and
dec
reas
e In
depe
nden
t va
riabl
es:
satis
fact
ion.
Wor
kloa
d de
man
ds (
subj
ectiv
e),
men
tal
The
inte
ract
ion
betw
een
cont
rol a
nd
dem
ands
(ob
ject
ive)
, an
d ph
ysic
al
obje
ctiv
e m
enta
l dem
ands
had
a
dem
ands
(ob
ject
ive)
as
thre
e di
stin
ct
nega
tive
effe
ct o
n bo
th t
ardi
ness
(β
= –
job
dem
ands
3.50
, F
(1,8
6) =
31.
49,
p <
.01
, ∆R
2 = .
26)
and
sick
day
s (β
= –
3.98
, F
(1,8
6) =
4.6
2, p
C
ontr
ol (
subj
ectiv
e)<
.05
, ∆R
2 = .
04),
indi
catin
g th
at a
t hi
gh
leve
ls o
f co
ntro
l, m
enta
l dem
ands
are
as
soci
ated
with
low
er le
vels
of
tard
ines
s an
d si
ck d
ays.
Whe
n co
ntro
l is
low
, m
enta
l dem
ands
lead
to
high
tar
dine
ss
and
sick
day
s. T
he in
tera
ctio
n of
con
trol
an
d ph
ysic
al d
eman
ds d
id n
ot h
ave
a si
gnifi
cant
effe
ct o
n an
y of
the
de
pend
ent
varia
bles
.
Aut
hors
Mod
el/V
aria
bles
Util
ized
Met
hodo
logi
cal A
ppro
ach
Sam
ple
Fin
ding
s
MANAGING CUSTOMER RELATIONSHIPS 161
(con
tinue
d)
Kel
low
ay a
nd
War
r’s (
1987
) m
odel
of
empl
oym
ent
Str
uctu
ral e
quat
ion
mod
elin
g.A
ll 2,
300
empl
oyee
s T
he s
truc
tura
l equ
atio
n m
odel
fit
the
Bar
ling
(199
1)an
d m
enta
l hea
lth,
whi
ch d
istin
guis
hes
A p
roce
ssua
l mod
el w
here
of
a la
rge
hosp
ital
data
wel
l (Q
= 8
4, W
19,7
20 =
35.
01,
p <
be
twee
n jo
b-re
late
d an
d co
ntex
t-fr
ee
subj
ectiv
e co
mpe
tenc
e an
d w
ere
incl
uded
in t
he
.01)
.m
enta
l hea
lthaf
fect
ive
wel
l-bei
ng m
edia
te
stud
y. 7
20 u
sabl
e th
e ef
fect
s of
job
resp
onse
s w
ere
Con
text
-fre
e m
enta
l hea
lth w
as
Dep
ende
nt v
aria
ble:
char
acte
ristic
s an
d ro
le
obta
ined
. The
si
gnifi
cant
ly p
redi
cted
(p
< .
01)
by
Con
text
-fre
e m
enta
l hea
lth (
Gen
eral
st
ress
ors
on c
onte
xt-f
ree
nurs
ing
staf
f m
ade
subj
ectiv
e co
mpe
tenc
e (–
.14)
. It
was
H
ealth
Que
stio
nnai
re-1
2 m
enta
l hea
lthup
43%
of
the
also
sig
nific
antly
pre
dict
ed (
p <
.00
1) b
y un
idim
ensi
onal
mea
sure
)gr
oup.
all t
hree
dim
ensi
ons
of a
ffect
ive
wel
l-be
ing—
satis
fact
ion
(–.2
2),
exha
ustio
n Jo
b-re
late
d su
bjec
tive
com
pete
nce
(.37
), d
eper
sona
lizat
ion
(.09
). (p
erso
nal a
ccom
plis
hmen
ts a
t w
ork)
as
Sub
ject
ive
com
pete
nce
was
sig
nific
antly
m
edia
ting
varia
ble
pred
icte
d (p
< .
001)
by
thre
e jo
b ch
arac
teris
tics—
varie
ty (
.27)
, au
tono
my
Wor
k sa
tisfa
ctio
n, e
mot
iona
l (.
23),
fee
dbac
k fr
om jo
b (.
13).
exha
ustio
n, d
eper
sona
lizat
ion
as t
hree
di
stin
ct m
edia
ting
mea
sure
s of
job-
Dim
ensi
ons
of jo
b-re
late
d w
ell-b
eing
re
late
d af
fect
ive
wel
l-bei
ngw
ere
sign
ifica
ntly
pre
dict
ed (
p <
.00
1)
by b
oth
role
str
esso
rs: r
ole
ambi
guity
→
Inde
pend
ent
varia
bles
:sa
tisfa
ctio
n, e
xhau
stio
n,
Aut
onom
y, t
ask
varie
ty,
task
iden
tity,
de
pers
onal
izat
ion
(–.1
7, .
10,
.25)
; rol
e fe
edba
ck f
rom
job,
fee
dbac
k fr
om c
o-co
nflic
t →
sat
isfa
ctio
n, e
xhau
stio
n,
wor
kers
as
five
dist
inct
mea
sure
s of
job
depe
rson
aliz
atio
n (–
.32,
.13
, –.
24).
In
char
acte
ristic
sad
ditio
n, s
atis
fact
ion
was
sig
nific
antly
R
ole
ambi
guity
and
rol
e co
nflic
t as
tw
o pr
edic
ted
(p <
.01
) by
aut
onom
y (.
17),
di
stin
ct m
easu
res
of r
ole
stre
ssta
sk v
arie
ty (
.37)
, an
d fe
edba
ck f
rom
co
-wor
kers
(.1
5).
The
se r
esul
ts n
ot o
nly
confi
rm W
arr’s
(1
987)
orig
inal
con
tent
ion
that
job
and
role
cha
ract
eris
tics
pred
ict
job-
rela
ted
men
tal h
ealth
, bu
t al
so e
xten
d W
arr’s
pr
opos
ition
s by
dem
onst
ratin
g th
at jo
b-re
late
d m
enta
l hea
lth m
edia
tes
the
effe
cts
of r
ole
stre
ssor
s an
d jo
b ch
arac
teris
tics
on c
onte
xt-f
ree
men
tal
heal
th.
162 RUTH N. BOLTON AND CRINA O. TARASIF
letc
her
and
Kar
asek
’s (
1979
) de
man
ds-d
iscr
etio
n M
ultip
le r
egre
ssio
n an
alys
is
3,08
6 pa
tient
s M
AN
OV
A in
dica
ted
that
, fo
r bo
th m
en
Jone
s (1
993)
(con
trol
) m
odel
.us
ing
inte
ract
ion
and
quad
ratic
be
twee
n th
e ag
es o
f an
d w
omen
, th
ere
wer
e si
gnifi
cant
te
rms
(for
effe
cts
anal
ysis
),
30 a
nd 6
0 (1
678
diffe
renc
es (
p <
.01
) in
tot
al d
eman
d D
epen
dent
var
iabl
es:
and
MA
NO
VA
(fo
r m
eans
w
omen
, 14
08 m
en)
and
disc
retio
n sc
ores
of
the
two
anal
ysis
)on
the
list
of
an
occu
patio
nal g
roup
s (m
anua
l and
Jo
b sa
tisfa
ctio
nN
HS
Hea
lth C
ente
r no
nman
ual w
orke
rs),
bot
h be
ing
Ste
p 1:
Rep
licat
ing
Kar
asek
’s
Pra
ctic
e in
si
gnifi
cant
ly h
ighe
r fo
r th
e no
nman
ual
Life
sat
isfa
ctio
nm
odel
usi
ng m
ultip
le
Her
tford
shire
, U
Kw
orke
rs (
tota
l diff
eren
ces
rang
ed f
rom
re
gres
sion
with
mai
n ef
fect
s 1.
2 to
4.5
; F-v
alue
s ra
nged
fro
m 4
6.2
to
Anx
iety
(thr
ee d
epen
dent
var
iabl
es:
285.
6). M
ultip
le r
egre
ssio
ns c
ontr
ollin
g de
pres
sion
, jo
b sa
tisfa
ctio
n,
for
occu
patio
nal s
tatu
s in
dica
ted
that
D
epre
ssio
nlif
e sa
tisfa
ctio
n)m
ain
effe
cts
of d
eman
ds a
nd d
iscr
etio
n w
ere
sign
ifica
nt (
p <
.05
) bu
t ag
ain
with
In
depe
nden
t va
riabl
es:
Ste
p 2:
Tes
ting
Kar
asek
’s
wea
k ex
plan
ator
y po
wer
(R
2 s ra
nged
m
odel
for
occ
upat
iona
l sta
tus
from
.02
7 to
.04
5). A
sig
nific
ant
effe
ct (
p Jo
b de
man
ds (
com
bina
tion
of p
ace
of
usin
g M
AN
OV
A<
.05
) w
as o
btai
ned
only
on
job
wor
k, w
orkl
oad,
job
over
load
, an
d sa
tisfa
ctio
n (β
= .
46)
and
life
satis
fact
ion
confl
ictin
g de
man
ds)
Ste
p 3:
Tes
ting
Kar
asek
’s
for
men
, bu
t no
sig
nific
ant
effe
ct o
n an
y m
odel
usi
ng m
ultip
le
of t
he d
epen
dent
var
iabl
es in
the
cas
e Jo
b di
scre
tion
(com
bina
tion
of s
kill
regr
essi
on w
ith in
tera
ctio
n of
wom
en. I
n ge
nera
l, fo
r bo
th m
en a
nd
disc
retio
n an
d de
cisi
on a
utho
rity)
term
s an
d co
ntro
lling
for
w
omen
, th
e am
ount
of
varia
nce
in e
ach
occu
patio
n st
atus
depe
nden
t va
riabl
e ac
coun
ted
for
by t
he
Inte
rper
sona
l sup
port
s (c
ombi
natio
n of
in
depe
nden
t va
riabl
es w
as v
ery
smal
l pr
ofes
sion
al c
ompe
tenc
e of
col
leag
ues
Ste
p 4:
Tes
ting
curv
iline
ar
(R2 s
rang
ed f
rom
.01
8 to
.09
5). T
here
an
d su
ppor
t fr
om b
oss,
co-
wor
kers
, re
latio
nshi
ps u
sing
w
as n
o ev
iden
ce o
f in
tera
ctio
n an
d lit
tle
and
spou
se/p
artn
er/fa
mily
)hi
erar
chic
al m
ultip
le
evid
ence
of
curv
iline
ar r
elat
ions
hips
.re
gres
sion
ana
lysi
s
Ste
p 5:
Tes
ting
for
the
effe
ct
of in
terp
erso
nal s
uppo
rt u
sing
hi
erar
chic
al m
ultip
le
regr
essi
on
Aut
hors
Mod
el/V
aria
bles
Util
ized
Met
hodo
logi
cal A
ppro
ach
Sam
ple
Fin
ding
s
MANAGING CUSTOMER RELATIONSHIPS 163
(con
tinue
d)
Fox
, D
wye
r, K
aras
ek’s
(19
79)
dem
ands
-con
trol
H
iera
rchi
cal m
ultip
le
All
of t
he 1
98 n
urse
s T
he in
tera
ctio
n of
per
ceiv
ed c
ontr
ol a
nd
Gan
ster
(19
93)
mod
elre
gres
sion
ana
lysi
s w
ith
empl
oyed
in a
pe
rcei
ved
wor
kloa
d si
gnifi
cant
ly
inte
ract
ion
term
sm
ediu
m-s
ized
pr
edic
ted
job
satis
fact
ion,
sys
tolic
blo
od
Dep
ened
ent
varia
bles
:pr
ivat
e ho
spita
lpr
essu
re a
t ho
me
and
at w
ork,
dia
stol
ic
bloo
d pr
essu
re a
t ho
me,
and
cor
tisol
Jo
b pe
rfor
man
ce (
com
bina
tion
of
leve
l at
wor
k (R
2 s ra
nged
fro
m .
11 t
o pa
tient
ass
essm
ent,
plan
ning
, .5
9, F
-val
ues
rang
ed f
rom
4.4
3, d
f =
de
velo
ping
pat
ient
car
e pl
ans,
etc
.)11
28,
p <
.05
to
6.29
, df
= 1
128,
p <
.0
5). I
n lo
w-c
ontr
ol s
ituat
ions
, w
orkl
oad
Job
satis
fact
ion
and
com
plai
nts
(illn
ess
redu
ced
satis
fact
ion,
incr
ease
d sy
stol
ic
and
som
atic
) as
tw
o di
stin
ct m
easu
res
bloo
d pr
essu
re a
t w
ork
and
at h
ome,
of
affe
ctiv
e ou
tcom
esan
d co
rtis
ol le
vels
at
wor
k. I
n hi
gh-
cont
rol s
ituat
ions
, w
orkl
oad
incr
ease
d A
rter
ial s
ysto
lic a
nd d
iast
olic
blo
od
satis
fact
ion,
red
uced
sys
tolic
blo
od
pres
sure
at
hom
e an
d at
wor
k, a
nd
pres
sure
at
wor
k, a
nd d
iast
olic
blo
od
saliv
ary
cort
isol
at
hom
e an
d at
wor
k)
pres
sure
at
hom
e, a
nd d
id n
ot a
ffect
as
eig
ht d
istin
ct m
easu
res
of
syst
olic
blo
od p
ress
ure
at h
ome
and
phys
iolo
gica
l out
com
es (
phys
iolo
gica
l co
rtis
ol le
vels
at
wor
k.da
ta w
as c
olle
cted
afte
r qu
estio
nnai
res
wer
e co
mpl
eted
)T
he in
tera
ctio
ns o
f pe
rcei
ved
cont
rol
and
obje
ctiv
e w
ork
load
sig
nific
antly
In
depe
nden
t va
riabl
es:
pred
icte
d jo
b sa
tisfa
ctio
n an
d co
rtis
ol
leve
ls a
t ho
me
(R2 s
rang
ed f
rom
.13
to
Pat
ient
load
, pa
tient
con
tact
hou
rs a
s .2
4, F
-val
ues
rang
ed f
rom
3.9
8, d
f =
pe
rcen
tage
of
tota
l wor
k, a
nd n
umbe
r 11
46,
p <
.05
to
8.05
, df
= 1
147,
p <
of
dea
ths
witn
esse
d as
thr
ee d
istin
ct
.01)
.Tw
o of
the
se in
tera
ctio
ns,
obje
ctiv
e m
easu
res
of w
ork
dem
ands
cont
rol/s
tres
sful
eve
nts
and
cont
rol/p
atie
nt c
onta
ct t
ime,
sup
port
the
W
ork
load
and
inve
ntor
y of
str
essf
ul
Kar
asek
mod
el. I
n lo
w-c
ontr
ol
even
ts a
s tw
o di
stin
ct s
ubje
ctiv
e si
tuat
ions
, st
ress
eve
nts
redu
ced
mea
sure
s of
wor
k de
man
dssa
tisfa
ctio
n an
d pa
tient
con
tact
tim
e in
crea
sed
cort
isol
leve
ls a
t ho
me.
In
Con
trol
(co
ntro
l ove
r ta
sk v
arie
ty,
task
hi
gh-c
ontr
ol s
ituat
ions
, st
ress
eve
nts
did
orde
r, pa
cing
, sc
hedu
ling,
pro
cedu
res,
no
t af
fect
sat
isfa
ctio
n an
d pa
tient
et
c.)
cont
act
time
only
slig
htly
red
uced
co
rtis
ol le
vels
at
hom
e.
164 RUTH N. BOLTON AND CRINA O. TARASID
anie
ls a
nd
War
r’s (
1987
) m
odel
of
empl
oym
ent
Long
itudi
nal r
epea
ted
244
rand
omly
N
o si
gnifi
cant
tw
o-w
ay in
tera
ctio
ns w
ere
Gup
py (
1994
)an
d m
enta
l hea
lth,
whi
ch d
istin
guis
hes
mea
sure
s de
sign
usi
ng
sele
cted
ob
tain
ed f
or p
artic
ipat
ion/
stre
ssor
s,
betw
een
job-
rela
ted
and
cont
ext-
free
m
ultip
le m
oder
ated
reg
ress
ion
acco
unta
nts
from
au
tono
my/
stre
ssor
s, a
nd s
ocia
l m
enta
l hea
lthan
alys
is w
ith in
tera
ctio
n te
rms
the
Inst
itute
of
supp
ort/s
tres
sors
. Sig
nific
ant
thre
e-w
ay
Cha
rter
ed
inte
ract
ions
obt
aine
d fo
r D
epen
dent
var
iabl
es:
All
cons
truc
ts m
easu
red
twic
e A
ccou
ntan
ts o
f st
ress
ors/
WC
LS/a
uton
omy
(p <
.05
) w
ith a
mon
th g
apE
ngla
nd a
nd W
ales
and
for
stre
ssor
s/W
CLS
/soc
ial s
uppo
rt
Job-
rela
ted
plea
sure
-dis
plea
sure
, jo
b-(p
< .
10).
rela
ted
anxi
ety-
cont
entm
ent,
and
job-
Mod
el 1
: with
par
ticip
atio
nre
late
d de
pres
sion
-ent
husi
asm
as
thre
e T
he s
igni
fican
t th
ree-
way
inte
ract
ions
di
stin
ct c
onte
xt-d
epen
dent
mea
sure
s of
M
odel
2: w
ith a
uton
omy
indi
cate
tha
t W
CLS
and
job
auto
nom
y ps
ycho
logi
cal w
ell-b
eing
buffe
r th
e ef
fect
s of
str
esso
rs o
n M
odel
3: w
ith s
ocia
l sup
port
psyc
holo
gica
l wel
l-bei
ng s
uch
that
G
ener
al H
ealth
Que
stio
nnai
re-1
2 as
st
ress
com
prom
ises
wel
l-bei
ng t
he le
ast
the
cont
ext-
free
uni
dim
ensi
onal
In
all
mod
els:
whe
n lo
cus
of c
ontr
ol is
inte
rnal
and
m
easu
re o
f w
ell-b
eing
auto
nom
y is
hig
h, a
nd m
ost
whe
n lo
cus
Ste
p 1:
initi
al w
ell-b
eing
of
con
trol
is e
xter
nal a
nd a
uton
omy
is
Inde
pend
ent
varia
bles
:sc
ores
low
.
Wor
k lo
cus
of c
ontr
ol (
WC
LS)
Ste
p 2:
mai
n ef
fect
s
Job
stre
ssor
sS
tep
3: a
ll tw
o-w
ay
inte
ract
ions
Soc
ial s
uppo
rtS
tep
4: t
hree
-way
inte
ract
ions
Job
auto
nom
y (c
ombi
natio
n of
tas
k in
depe
nden
ce,
clos
enes
s of
su
perv
isio
n)P
artic
ipat
ion
in d
ecis
ion
mak
ing
Aut
hors
Mod
el/V
aria
bles
Util
ized
Met
hodo
logi
cal A
ppro
ach
Sam
ple
Fin
ding
s
MANAGING CUSTOMER RELATIONSHIPS 165
(con
tinue
d)
Xie
and
Joh
ns
Dep
ende
nt v
aria
ble:
Cur
vilin
ear
hier
arch
ical
41
5 fu
ll-tim
e S
ubje
ctiv
e jo
b sc
ope,
DO
T, O
P, t
ask
(199
5)m
oder
ated
mul
tiple
reg
ress
ion
empl
oyee
s si
gnifi
canc
e, a
nd t
ask
iden
tity
Em
otio
nal e
xhau
stio
n (c
hron
ic s
tres
s)
anal
ysis
usi
ng in
tera
ctio
n (p
rofe
ssio
nals
, si
gnifi
cant
ly p
redi
cted
exh
aust
ion
in a
an
d an
xiet
y (lo
ng-t
erm
str
ess)
as
two
term
sm
anag
ers,
sal
es
U-s
hape
d cu
rvili
near
fas
hion
(∆R
2 s di
stin
ct m
easu
res
of s
tres
sw
orke
rs,
cler
ical
ra
nged
fro
m .
01 [
R2 =
.11
] to
.04
[R
2 =
wor
kers
, bl
ue-c
olla
r .1
1],
F-v
alue
s ra
nged
fro
m F
1,40
3 =
4.5
5,
Inde
pend
ent
varia
bles
:w
orke
rs)
from
a
p <
.05
to
F1,
409
= 1
9.09
, p
< .
001)
. Low
ra
ndom
sel
ectio
n of
an
d hi
gh le
vels
of
thes
e pr
edic
tors
wer
e Jo
b co
mpl
exity
fro
m D
ictio
nary
of
65 o
rgan
izat
ions
as
soci
ated
with
hig
h le
vels
of
Occ
upat
iona
l Titl
es (
DO
T)
and
incl
udin
g ba
nks,
ex
haus
tion,
whe
reas
at
mod
erat
e le
vels
O
ccup
atio
nal P
rest
ige
Inde
x (O
P)
as
insu
ranc
e th
ey p
redi
cted
low
leve
ls o
f ex
haus
tion.
two
dist
inct
obj
ectiv
e m
easu
res
of jo
b co
mpa
nies
, hi
gh
scop
esc
hool
s, h
ospi
tals
, A
nxie
ty w
as s
igni
fican
tly p
redi
cted
in U
-et
c.sh
aped
cur
vilin
ear
fash
ion
only
by
task
S
ubje
ctiv
e jo
b sc
ope
(com
bina
tion
of
iden
tity
(∆R
2 = .
01,
[R2 =
.03
], F
1,40
0 =
sk
ill v
arie
ty,
task
iden
tity,
tas
k 5.
08,
p <
.05
), in
dica
ting
that
low
and
si
gnifi
canc
e, a
uton
omy,
and
fee
dbac
k)hi
gh le
vels
of
iden
tity
wer
e as
soci
ated
w
ith h
igh
anxi
ety
whe
reas
mod
erat
e
Per
ceiv
ed d
eman
ds-a
bilit
y fit
leve
ls o
f id
entit
y pr
edic
ted
low
anx
iety
.
DO
T/d
eman
d-ab
ility
fit
and
OP
/dem
and-
abili
ty fi
t in
tera
ctio
ns s
igni
fican
tly
pred
icte
d bo
th e
xhau
stio
n an
d an
xiet
y (∆
R2 s
rang
ed f
rom
.00
9 [R
2 = .
09]
to
.015
[R
2 = .
10],
F-v
alue
s ra
nged
fro
m
F1,
409
= 4
.03,
p <
.05
to
F1,
401
= 6
.86,
p <
.0
1).
The
se in
tera
ctio
ns w
ere
slig
htly
and
ne
gativ
ely
asso
ciat
ed w
ith e
xhau
stio
n an
d an
xiet
y.
Wal
l , J
acks
on,
Kar
asek
’s (
1979
) de
man
d-co
ntro
l M
oder
ated
reg
ress
ion
anal
ysis
A
ll 1,
451
prod
uctio
n T
he in
tera
ctio
n ef
fect
s of
job
dem
ands
M
ulla
rkey
, an
d m
odel
, an
d a
dem
and–
latit
ude
mod
elus
ing
inte
ract
ion
term
sem
ploy
ees
and
and
job
cont
rol w
ere
sign
ifica
nt in
P
arke
r (1
996)
supp
ort
and
pred
ictin
g ea
ch d
epen
dent
var
iabl
e. F
-
166 RUTH N. BOLTON AND CRINA O. TARASID
epen
dent
var
iabl
es:
supe
rvis
ory
staf
f va
lues
ran
ged
from
7.8
1 to
13.
04 (
p <
fr
om f
our
Brit
ish
.01)
. The
inte
ract
ion
effe
cts
of jo
b Jo
b sa
tisfa
ctio
n, d
epre
ssio
n, a
nd
man
ufac
turin
g de
man
ds a
nd d
ecis
ion
latit
ude
wer
e,
anxi
ety
as t
hree
dis
tinct
mea
sure
s of
co
mpa
nies
how
ever
, no
nsig
nific
ant
in p
redi
ctin
g th
e jo
b st
rain
depe
nden
t va
riabl
es. F
-val
ues
rang
ed
from
.95
to
2.37
(p
< .
01).
Inde
pend
ent
varia
bles
:R
esul
ts in
dica
te t
hat
(psy
chol
ogic
al)
job
Job
dem
ands
(co
mbi
natio
n of
st
rain
is c
ause
d by
the
inte
ract
ion
effe
ct
mon
itorin
g de
man
d an
d pr
oble
m-
of jo
b de
man
ds a
nd jo
b co
ntro
l, no
t of
so
lvin
g de
man
d)jo
b de
man
ds a
nd d
ecis
ion
latit
ude
as
orig
inal
ly d
emon
stra
ted
by K
aras
ek
Job
cont
rol (
com
bina
tion
of t
imin
g (1
979)
.co
ntro
l and
met
hod
cont
rol)
Dec
isio
n la
titud
e (P
erce
ived
Int
rinsi
c Jo
b C
hara
cter
istic
s S
cale
, w
hich
in
clud
es m
easu
res
of f
reed
om t
o ch
oose
met
hod,
opp
ortu
nity
to
use
abili
ty,
amou
nt o
f va
riety
, et
c.)
Sch
aubr
oeck
K
aras
ek’s
(19
79)
dem
and-
cont
rol
Mul
tiple
reg
ress
ion
anal
ysis
11
0 ra
ndom
ly
Mai
n ef
fect
s an
d tw
o-w
ay in
tera
ctio
n an
d M
errit
t m
odel
usin
g in
tera
ctio
n te
rms
sele
cted
ful
l-tim
e ef
fect
s of
job
dem
ands
, jo
b co
ntro
l, an
d (1
997)
heal
th p
rofe
ssio
nals
se
lf-ef
ficac
y w
ere
nons
igni
fican
t in
D
epen
dent
var
iabl
es:
Mod
el 1
: dem
ogra
phic
at
a la
rge
pred
ictin
g di
asto
lic a
nd s
ysto
lic b
lood
va
riabl
esre
habi
litat
ion
pres
sure
s. T
he t
hree
-way
inte
ract
ion
Dia
stol
ic b
lood
pre
ssur
e an
d sy
stol
ic
hosp
ital
effe
ct w
as,
how
ever
, si
gnifi
cant
in
bloo
d pr
essu
re a
s tw
o di
stin
ct
Mod
el 2
: mai
n ef
fect
spr
edic
ting
both
dia
stol
ic (
p <
.05
) bl
ood
mea
sure
s of
job
stre
sspr
essu
re (
β =
–14
.6,
∆R2 =
.07
, F
1,66
=
Mod
el 3
: all
two-
way
5.
90,
p <
.01
8, T
otal
R2 =
.24
) an
d In
depe
nden
t va
riabl
es:
inte
ract
ions
syst
olic
(p
< .
01)
bloo
d pr
essu
re (
β =
–24
.50,
∆R
2 = .
11,
F1,
66 =
9.6
8, p
< .
001,
Jo
b de
man
dsM
odel
4: t
hree
-way
inte
ract
ion
Tota
l R2 =
.28
).
Job
cont
rol
Sel
f-ef
ficac
y as
mod
erat
ing
varia
ble
Aut
hors
Mod
el/V
aria
bles
Util
ized
Met
hodo
logi
cal A
ppro
ach
Sam
ple
Fin
ding
s
MANAGING CUSTOMER RELATIONSHIPS 167
(con
tinue
d)
Dec
isio
n la
titud
e (f
reed
om t
o ch
oose
m
etho
d, o
ppor
tuni
ty t
o us
e ab
ility
, am
ount
of
varie
ty,
etc.
)
Sch
aubr
oeck
K
aras
ek’s
(19
79)
dem
ands
-con
trol
H
iera
rchi
cal m
ultip
le
214
serv
ice
The
key
res
ults
foc
used
on
sign
ifica
nt
and
Fin
k (1
998)
mod
el;
regr
essi
on a
naly
sis
usin
g co
unse
lors
, sa
les
thre
e-w
ay in
tera
ctio
ns. C
onsi
sten
t in
tera
ctio
n te
rms
coun
selo
rs,
and
supp
ort
was
obt
aine
d fo
r th
e no
tion
that
Jo
hnso
n an
d H
all’s
(19
88)
dem
ands
-un
derw
riter
s in
tw
o bo
th jo
b co
ntro
l and
soc
ial s
uppo
rt a
re
cont
rol-s
uppo
rt m
odel
Mod
el 1
: mai
n ef
fect
sla
rge
insu
ranc
e ne
eded
to
faci
litat
e jo
b de
man
ds a
nd
com
pani
esre
sult
in p
ositi
ve o
utco
mes
for
D
epen
dent
var
iabl
es:
Mod
el 2
: 2-w
ay in
tera
ctio
n or
gani
zatio
nal c
omm
itmen
t (F
= 2
.28,
p
(sup
port
/con
trol
)<
.06
) an
d sa
tisfa
ctio
n (F
= 3
.79,
p <
Jo
b pe
rfor
man
ce.0
5). C
opin
g di
fficu
lties
wer
e re
port
ed
Mod
el 3
: 3-w
ay in
tera
ctio
n fo
r w
orke
rs f
acin
g jo
b de
man
ds c
oupl
ed
Pro
soci
al o
rgan
izat
iona
l beh
avio
r (s
uppo
rt/c
ontr
ol/d
eman
ds)
with
hig
h co
ntro
l and
low
sup
port
or
low
(P
SO
B)
(altr
uism
, at
tend
ance
, co
ntro
l and
hig
h su
ppor
t. B
oth
cont
rol
cons
cien
tious
ness
)an
d su
ppor
t ne
ed t
o be
hig
h fo
r ef
fect
ive
copi
ng. H
owev
er,
coun
ter
Phy
sica
l sym
ptom
atol
ogy
resu
lts a
re o
btai
ned
for
a si
ngle
st
ress
or—
resp
onsi
bilit
y fo
r ot
hers
.S
uper
viso
r an
d in
trin
sic
job
satis
fact
ion
Org
aniz
atio
nal c
omm
itmen
t
Sic
k le
ave
(hou
rs)
Inde
pend
ent
varia
bles
:
Soc
ial s
uppo
rt (
supe
rvis
or
cons
ider
atio
n);
Em
ploy
ee c
ontr
ol (
cont
rol o
ver
qual
ity,
met
hods
);
Wor
k (d
eman
ds)
stre
ssor
s (q
uant
itativ
e w
orkl
oad,
ski
ll un
deru
tiliz
atio
n, r
ole
confl
ict,
role
am
bigu
ity)
168 RUTH N. BOLTON AND CRINA O. TARASIS
ingh
(19
98)
Dep
ende
nt v
aria
bles
:P
ath
anal
ysis
usi
ng q
uadr
atic
51
8 sa
les
and
Res
ults
var
ied
by t
he jo
b ou
tcom
es
and
inte
ract
ion
term
s, a
nd
mar
ketin
g pe
rson
nel
exam
ined
. Lin
ear
effe
cts
wer
e Jo
b pe
rfor
man
ceco
rrec
ting
for
mea
sure
men
t dr
awn
from
the
su
ppor
ted
for
job
satis
fact
ion
and
erro
rm
embe
rs o
f th
e or
gani
zatio
nal c
omm
itmen
t. Tw
o-w
ay
Job
tens
ion
Ass
ocia
tion
of S
ales
in
tera
ctio
ns in
volv
ing
role
con
flict
and
an
d M
arke
ting
task
var
iety
, an
d ro
le a
mbi
guity
and
Jo
b sa
tisfa
ctio
nex
ecut
ives
, an
d au
tono
my
wer
e su
ppor
ted
for
job
from
tw
o di
ffere
nt
perf
orm
ance
(p
< .
05 a
nd p
< .
01).
Org
aniz
atio
nal c
omm
itmen
tun
its o
f a
For
tune
-C
opin
g w
as m
ore
effe
ctiv
e w
hen
high
50
0 in
dust
rial
role
con
flict
and
hig
h ta
sk v
arie
ty,
or
Turn
over
inte
ntio
nssu
pplie
r fir
mhi
gh r
ole
ambi
guity
and
low
aut
onom
y co
-occ
ured
. Sim
ilar
two-
way
In
depe
nden
t va
riabl
es:
inte
ract
ions
, in
volv
ing
role
ove
rload
/task
va
riety
and
rol
e am
bigu
ity/ta
sk v
arie
ty,
Rol
e co
nflic
t, ro
le a
mbi
guity
, an
d ro
le
wer
e si
gnifi
cant
for
tur
nove
r in
tent
ions
ov
erlo
ad a
s th
ree
dist
inct
mea
sure
s of
(p
< .
05).
Task
var
iety
fac
ilita
ted
copi
ng
role
str
ess
with
rol
e ov
erlo
ad b
ut h
inde
red
it w
hen
role
am
bigu
ity w
as h
igh.
Aut
onom
y, p
artic
ipat
ion,
fee
dbac
k, a
nd
task
var
iety
as
four
dis
tinct
mea
sure
s of
C
urvi
linea
r ef
fect
s in
volv
ing
two-
way
jo
b ch
arac
teris
tics
inte
ract
ions
and
qua
drat
ic t
erm
s w
ere
obta
ined
with
job
tens
ion
as t
he
depe
nden
t va
riabl
e. F
eedb
ack
faci
litat
ed c
opin
g w
ith r
ole
confl
ict
but
hind
ered
cop
ing
with
rol
e am
bigu
ity
(p <
.05
).
De
Jong
e, v
an
Kar
asek
’s d
eman
ds-c
ontr
ol m
odel
Mul
tilev
el r
egre
ssio
n an
alys
is
1,48
9 ra
ndom
ly
In p
redi
ctin
g em
otio
nal e
xhau
stio
n an
d B
reuk
elen
, (V
AR
CL)
usi
ng in
tera
ctio
n se
lect
ed h
ealth
car
e jo
b-re
late
d an
xiet
y, a
ggre
gate
-leve
l La
ndew
eerd
, D
epen
dent
var
iabl
es:
term
sw
orke
rs f
rom
m
easu
res
of jo
b de
man
ds a
nd jo
b N
ijhui
s (1
999)
diffe
rent
fun
ctio
nal
auto
nom
y (in
clud
ed in
Mod
el 3
) di
d no
t E
mot
iona
l exh
aust
ion
Mod
el 1
: Ful
ly u
ncon
ditio
nal
units
(in
tens
ive
expl
ain
addi
tiona
l var
ianc
e. I
n bo
th
mod
el o
nly
com
pris
ing
of t
he
care
, su
rgic
al u
nit,
case
s, M
odel
2 h
ad t
he h
ighe
st
Job-
rela
ted
anxi
ety
rand
om e
ffect
s of
uni
ts a
nd
etc.
) fr
om e
ight
pr
edic
tive
pow
er (
R2 =
.25
, R
2 = .
09,
inst
itutio
nsho
spita
ls a
nd e
ight
re
spec
tivel
y),
and
only
the
mai
n ef
fect
W
ork
mot
ivat
ion
nurs
ing
hom
esof
job
dem
ands
was
sig
nific
ant
(p <
M
odel
2: A
ll de
mog
raph
ics
.05)
. The
hig
her
the
dem
ands
, th
e
Aut
hors
Mod
el/V
aria
bles
Util
ized
Met
hodo
logi
cal A
ppro
ach
Sam
ple
Fin
ding
s
MANAGING CUSTOMER RELATIONSHIPS 169
(con
tinue
d)
Job
satis
fact
ion
and
indi
vidu
al-le
vel
high
er w
ere
exha
ustio
n an
d an
xiet
y.m
easu
rem
ents
of
job
Inde
pend
ent
varia
bles
:de
man
ds a
nd a
uton
omy
In p
redi
ctin
g w
ork
mot
ivat
ion
and
job
satis
fact
ion,
agg
rega
te-le
vel
Job
dem
ands
Mod
el 3
: Mod
el 2
plu
s m
easu
rem
ents
of
job
dem
ands
and
job
aggr
egat
e-le
vel
auto
nom
y (in
clud
ed in
Mod
el 3
) Jo
b au
tono
my
mea
sure
men
ts o
f jo
b ex
plai
ned
a si
gnifi
cant
am
ount
of
dem
ands
and
job
auto
nom
yad
ditio
nal v
aria
nce
(R2 =
.13
with
∆D
=
Dem
ands
and
aut
onom
y ar
e m
easu
red
11.7
1, p
< .
05.,
and
R2 =
.13
.2 w
ith ∆
D =
at
bot
h th
e in
divi
dual
and
agg
rega
te
A li
kelih
ood
ratio
tes
t (D
) is
12
.14,
p <
.05
, re
spec
tivel
y).
leve
ls (
mea
n gr
oup
scor
es o
f di
ffere
nt
used
to
dete
rmin
e th
e un
its in
clud
ed in
the
stu
dy).
Agg
rega
te-
diffe
renc
e of
pre
dict
ive
pow
er
The
indi
vidu
al-le
vel i
nter
actio
n ef
fect
of
leve
l mea
sure
men
ts c
ombi
ne
acro
ss t
he m
odel
sde
man
ds a
nd a
uton
omy
and
the
judg
men
ts a
cros
s in
divi
dual
jobs
, th
us
aggr
egat
e-le
vel m
ain
effe
ct o
f de
man
ds
rem
ovin
g va
rianc
e du
e to
indi
vidu
al
sign
ifica
ntly
pre
dict
ed m
otiv
atio
n (p
<
diffe
renc
es a
nd id
iosy
ncra
tic
.05)
. The
hig
her
the
inte
ract
ion,
the
re
spon
ses.
high
er w
as m
otiv
atio
n. T
he h
ighe
r th
e ag
greg
ated
dem
and,
the
low
er w
as
mot
ivat
ion.
Indi
vidu
al-le
vel m
ain
effe
cts
of jo
b de
man
d an
d jo
b au
tono
my,
and
the
ir ag
greg
ate-
leve
l int
erac
tion
effe
ct,
sign
ifica
ntly
pre
dict
ed jo
b sa
tisfa
ctio
n (p
<
.05
). T
he h
ighe
r th
e in
divi
dual
-leve
l au
tono
my
and
the
high
er t
he a
ggre
gate
-le
vel i
nter
actio
n, t
he h
ighe
r w
as
satis
fact
ion.
The
hig
her
the
indi
vidu
al-
leve
l job
dem
ands
the
low
er w
as
satis
fact
ion.
170 RUTH N. BOLTON AND CRINA O. TARASIJa
nsse
n (2
001)
Kar
asek
’s d
eman
ds-c
ontr
ol m
odel
Hie
rarc
hica
l mul
tiple
13
4 ra
ndom
ly
Job
dem
ands
squ
ared
inte
ract
ed w
ith
regr
essi
on a
naly
sis
usin
g se
lect
ed lo
w-le
vel
effo
rt-r
ewar
d fa
irne
ss t
o pr
edic
t bo
th
Dep
ende
nt v
aria
bles
:qu
adra
tic t
erm
san
d m
id-le
vel
perf
orm
ance
and
sat
isfa
ctio
n. T
he e
ffect
m
anag
ers
from
a
on s
tand
ard
job
perf
orm
ance
was
Jo
b pe
rfor
man
ce (
stan
dard
In
tera
ctio
n te
rms
anal
ysis
is
Dut
ch f
ood
prod
ucer
sign
ifica
nt (
p <
.05
) w
ith R
2 = .
30. T
he
perf
orm
ance
and
inno
vativ
e us
ed t
o de
tect
mod
erat
ion
and
effe
ct o
n le
ader
-rat
ed in
nova
tive
job
perf
orm
ance
)no
nlin
ear
effe
cts
perf
orm
ance
was
sig
nific
ant
(p <
.05
) w
ith R
2 = .
39. F
or s
elf-
rate
d in
nova
tive
Job
satis
fact
ion
job
perf
orm
ance
the
effe
ct w
as a
lso
sign
ifica
nt (
p <
.01
) w
ith R
2 = .
25. A
s fo
r B
oth
perf
orm
ance
and
sat
isfa
ctio
n sa
tisfa
ctio
n, t
he e
ffect
was
eve
n m
ore
wer
e m
easu
red
by m
eans
of
self-
rate
d si
gnifi
cant
(p
< .
001)
with
an
R2 =
.28
. an
d le
ader
-rat
ed s
cale
s in
add
ition
to
Pos
t-ho
c pr
obin
g of
the
se in
tera
ctio
n st
anda
rd m
easu
res
effe
cts
indi
cate
d th
at t
he m
anag
ers
who
pe
rcei
ve e
ffort
-rew
ard
fair
ness
per
form
In
depe
nden
t va
riabl
es:
bette
r an
d fe
el m
ore
satis
fied
in
resp
onse
to
inte
rmed
iate
leve
ls o
f jo
b Jo
b de
man
dsde
man
ds t
han
man
ager
s w
ho p
erce
ive
“und
er-r
ewar
d” u
nfai
rnes
s.E
ffort
-rew
ard
fair
ness
per
cept
ions
van
Ype
ren
K
aras
ek’s
dem
ands
-con
trol
mod
elH
iera
rchi
cal m
ultip
le
555
nurs
es w
ho
Job
dem
and–
job
cont
rol i
nter
actio
n an
dre
gres
sion
ana
lysi
s us
ing
wor
ked
at
sign
ifica
ntly
pre
dict
ed f
atig
ue (
p <
.00
1)
Hag
edoo
rn
Dep
ende
nt v
aria
bles
:in
tera
ctio
n te
rms
spec
ializ
ed u
nits
for
w
ith R
2 = .
27,
but
not
intr
insi
c (2
003)
patie
nts
with
men
tal
mot
ivat
ion.
The
dem
and-
soci
al s
uppo
rt
Intr
insi
c m
otiv
atio
nIn
tera
ctio
n te
rms
anal
ysis
is
defic
ienc
yin
tera
ctio
n, o
n th
e ot
her
hand
, us
ed t
o de
tect
mod
erat
ion
sign
ifica
ntly
pre
dict
ed in
trin
sic
Fatig
ueef
fect
sm
otiv
atio
n (p
< .
01),
but
not
fat
igue
. F
inal
ly,
the
dem
and-
cont
rol-s
uppo
rt
Inde
pend
ent
varia
bles
:in
tera
ctio
n pr
edic
ted
mot
ivat
ion
(0 <
.0
1),
but
not
fatig
ue. U
nder
con
ditio
ns o
f Jo
b de
man
dslo
w s
uppo
rt,
dem
ands
incr
ease
d m
otiv
atio
n w
hen
cont
rol w
as h
igh,
and
Jo
b co
ntro
lde
crea
sed
it w
hen
cont
rol w
as lo
w.
Und
er c
ondi
tions
of
high
sup
port
, S
ocia
l sup
port
dem
ands
dec
reas
ed m
otiv
atio
n w
hen
cont
rol w
as h
igh.
The
re w
as n
o ef
fect
w
hen
cont
rol w
as lo
w.
Aut
hors
Mod
el/V
aria
bles
Util
ized
Met
hodo
logi
cal A
ppro
ach
Sam
ple
Fin
ding
s
MANAGING CUSTOMER RELATIONSHIPS 171
(e.g., role conflict, ambiguity, overload) and job scope (e.g., skill variety, autonomy, feedback) on satisfaction across different technological contexts. Several other studies (Fox, Dwyer, and Ganster 1993; Dwyer and Ganster 1991; De Jonge et al. 1999) have supported Karasek’s model by differentiating between the subjective and objective aspects of job demands and job control. In general, results are consistent with Dwyer and Ganster’s (1991) findings that the interaction effects are consistently significant for the subjective or perceived measures of job demands and control (however, see De Jonge et. al. 1999).
Other studies have provided evidence that either fails to support or refutes Karasek’s contingency model. For instance, Spector (1987) studied the effects of demands and control on the satisfac-tion, frustration, anxiety, and health symptoms of clerical employees at a major university. He found that, out of thirty regressions, the interaction effect of demands and control was significant only in two, casting doubt on the validity of the demand–control model. Moreover, in these two cases of significant interaction effects, the direction of effects did not correspond to Karasek’s hypothesis. Likewise, Fletcher and Jones (1993) were unable to find any significant interaction effect of demands and control on satisfaction, anxiety, and depression among patients at a health care center. Schaubroeck and Fink’s (1998) study provided support for the three-way interactions of demands, control, and support, rather than the two-way interaction of demands and control as originally proposed by Karasek. Coping difficulties were found among employees facing high job demands coupled with high-control and low-support jobs. Likewise, low control coupled with high support produced coping difficulties in employees in high-demand jobs. As such, both control and support appear necessary for effective coping. In another study exploring three way effects involving demands, control, and support, Van Yperen and Hagedoorn (2003) reported that, while demand–control interaction influenced employee fatigue (but not motivation), the demand–support interaction had a significant effect on employees’ intrinsic motivation (but not fatigue). In addition, a three-way interaction between demands, control, and support was significant for motivation but not fatigue. Depending on the level of job demands, control, and support, Van Yperen and Hage-doorn noted that some jobs or roles may evidence divergence between fatigue and motivation, such as when employees are highly motivated and fatigued at the same time.
Recently, researchers have modeled the curvilinear effects of job demands and job control on outcomes. For example, although he failed to find significant interaction effects of demands and control (e.g., decision latitude), Warr (1990) found that demands were nonlinearly (U-shaped) related to three separate dimensions of well-being among research employees—anxiety, depres-sion, and displeasure. He also found that control was nonlinearly (increasing slope) related to job satisfaction. Curvilinear effects were more rigorously tested by Xie and Johns (1995), who found that job scope (e.g., task identity, task variety, autonomy, feedback)—a measure of job control—had a U-shaped relationship with job stress, measured by exhaustion and anxiety for a wide range of respondents (e.g., managers, sales workers, blue-collar workers). In other words, beyond an intermediate level, increasing levels of job scope enhance the stress level of boundary spanners instead of buffering the effects of role stressors as hypothesized by Karasek. These findings have been replicated and extended by Singh (1998).
Assessment
Two themes emerge from our review of the empirical research rooted in Karasek’s model. First, while the support for Karasek’s model is mixed and uneven, this should be interpreted in light of the general lack of support for contingency models in organizational behavior (Tosi 1992). In many theories of organizational and human processes, interaction effects appear theoretically
172 RUTH N. BOLTON AND CRINA O. TARASI
plausible but fail to show up in empirical results. Researchers have indicated that this could occur for several reasons including (a) sampling, (b) measurement error, and (c) unnecessary complex-ity. In terms of sampling, the problem lies in the lack of heterogeneity in studies (e.g., samples from a single division/organization). Without the heterogeneity in perceived levels of job scope and role stressors, theoretical interaction effects are severely limited by the resulting restriction of range and fail to be empirically detected (Tosi 1992). In terms of measurement error, readers will note that most of the studies in Table 6.1 utilize regression analysis to examine interaction effects (Xie and Johns 1995; Champoux 1992). This is problematic because regression analysis ignores measurement error in dependent and independent variables. Because for most social science re-search, measurement error can be significant, ranging from 32 percent (Cote and Buckley 1988) to 50 percent (Schmidt and Hunter 1996), the regression coefficients are likely biased downward, making it difficult to empirically detect small but interesting contingency effects (Busemeyer and Jones 1983; MacCallum and Mar 1995). Finally, some researchers have argued that, despite the theoretical complexity of social science models, the empirical phenomena are governed by pragmatically simple and linear relationships. Because of random variation or unmodeled effects (e.g., contextual factors), it is plausible that some empirical studies could support interaction ef-fects while many others refute such complexity resulting in a “mixed” pattern of results. In the context of Karasek’s model, because researchers have not systematically tackled concerns due to sampling and measurement error, it is probably inappropriate to view the “mixed” evidence as support for the “unnecessary complexity” explanation at this time.
Second, empirical findings from the more recent studies suggest that Karasek’s model needs to be modified to account for the curvilinear effects of job scope and role stressors on job outcomes. Both effects are supported by strong theoretical foundations. In the context of job scope, researchers draw from activation theory to posit an “overstimulation” effect so that excessive levels of job scope including feedback, participation, variety, and autonomy hinder rather help in one’s performance (Singh 1998; Schwab and Cummings 1976; Champoux 1978, 1992). In turn, this overstimulation effect is based on three interrelated propositions: (1) job scope acts as a motivational force that stimulates an individual to increase effort or expend energy in task performance; (2) each individual has a “characteristic” level of stimulation that represents an optimal point of motivation; and (3) if experienced stimulation level substantially exceeds this “characteristic” level, the individual becomes overwhelmed, resulting in increased anxiety and reduced performance (Gardner and Cummings 1988; Kahn and Byosiere 1992). Consequently, both high (or excessive) and low (or inadequate) levels of job scope result in lowered performance and heightened anxiety (Xie and Johns 1995). Intermediate levels of job scope that are closer to the individual’s characteristic level result in an optimal job context. Studies by Xie and Johns (1995), Champoux (1992), and others provide support for the curvilinear effects of job scope. In the case of role stressors, researchers evoke the Yerkes-Dodson law (Yerkes and Dodson 1908) drawn from early clinical and labora-tory studies to posit that role stressors hold the potential for both “distress” and “eustress” (Selye 1976). According to this law, both at low and high levels of role stressors, an individual’s job context is suboptimal. This is because performance is undermined by a lack of challenge in the “low” condition and by overactivation in the “high” condition. Moreover, both of the preceding conditions are characterized by passive coping driven by either a low level of motivation/resource activation or lack of sufficient resources to deal with overwhelming role stressors (Schaubroeck and Ganster 1993). This passive coping interferes with the individual’s adaptivity to environmen-tal demands, further deteriorating performance and eventually leading to a “distress” condition. However, performance is thought to be optimal in the intermediate role stress condition as the individual is energized/activated to respond and actively cope with environmental demands but is
MANAGING CUSTOMER RELATIONSHIPS 173
not overwhelmed by them. Selye (1976) referred to this condition as “eustress” or “good” stress. Studies by Singh (1998), Schaubroeck and Ganster (1993), and Jamal (1984) provide support for the curvilinear effects of role stressors.
To simultaneously account for the curvilinear effects of role stressors and job scope, Karasek’s model can be adapted by including an intermediate condition along both of its dimensions (see Figure 6.2). The resulting framework is a 3 x 3 model that represents nine distinct job contexts. For “low” levels of job scope, the job context is likely to be “passive,” “distress,” or “overwhelm-ing” corresponding to “low,” “intermediate,” or “high” levels of role stressors (cells 1, 2, and 3). This is because a “low” job scope provides little to no coping resources to the boundary spanner and the distress level increases directly with role stressors, except in the event that role stressors are at a “low” level. In this instance, the job context lacks any stimulation whatsoever, resulting in a “passive” condition in accord with activation theory and Yerkes and Dodson’s law. With “intermediate” level of job scope, the job context is either balanced (cell 5, scope = stressors) or unbalanced (cells 4 and 6, scope ≠ stressors). In the balanced condition, stressors and scope are at their optimal level and cohere with Karasek’s notion of an “active” job context where stress-ors are high enough to provide the challenge without overwhelming the individual, and scope is high enough to aid coping with role demands without overstimulating the employee. This also accords with the inverted U-hypothesis. In the unbalanced conditions, the boundary spanner is either underwhelmed (scope > stressors) or overwhelmed (stressors > scope). In the former job context, role stressors are too low to challenge the individual, while in the latter condition, scope is probably insufficient to aid effective coping with role demands. Finally, under conditions of “high” scope, job contexts vary significantly depending on the level of role stressors ranging from a munificent and underwhelming job context (cell 7, scope >> stressors) to a highly charged, over-stimulated context where both scope and stressors are high (cell 9). Both are less than optimal as
Figure 6.2 A Graphical Representation of the Contingency Model for the Influence of Role Stressors and Job Scope on Boundary Spanner Effectiveness
7
Underwhelming(Insipid)
8
Eustress
9
Overstimulation(Hyperactive)
4
Insipid
5
Balance(Eustress)
6
Distress
1
Understimulation(Passive)
2
Distress
3
Overstimulation(Distress)
Role Stressors
Low Intermediate High
High
Intermediate
Low
JobScope
Axis ofcongruence
Axis ofincongruence
174 RUTH N. BOLTON AND CRINA O. TARASI
per overstimulation and activation theories. However, for “intermediate” levels of role stressors, the job context is likely to indicate “eustress” as the job is both challenging and resourceful to allow effective coping with job demands (cell 8). Note that the model in Figure 6.2 is bisected by two diagonals: (a) an axis of congruence along which the job contexts are balanced with congru-ent levels of job scope and role stressors (i.e., cells 1, 5, and 9); and (b) an axis of incongruence depicting job contexts that are completely unbalanced with opposite levels of job scope and role stressors (i.e., cells 3 and 7). Interestingly, only a single optimal cell (cell 5) exists where both scope and stressors are at an intermediate level. While the preceding adaptation of Karasek’s model is consistent with emerging empirical evidence and grounded in strong theory, it has not been subjected to empirical investigation.
Configural Perspective
We propose a configural perspective that also examines the effects of role stressors by explicitly considering the simultaneous and interactive influence of job scope. However, compared to the contingency perspective, a configural perspective is more complex because of several reasons. First, it accounts for unique patterns of role stressors and job scope (referred to as “configurations”) that represent nonlinear and higher-order interactions that cannot be represented within the traditional contingency models (Miller and Friesen 1984). Second, the configurations are defined at a group rather than individual level such that they are shared representations of boundary roles. Third, it allows consideration of equifinality, or the notion that different configurations may be equally effective (e.g., in terms of performance, satisfaction). Fourth, and finally, a configural approach affords flexibility in modeling the phenomenon using approaches that distinguish between logical plausibility (e.g., ideal types) and empirical viability (Doty and Glick 1994). Below we discuss these ideas and review the limited empirical research to date.
Theory
In accord with configural theory, we posit that a job context can be defined by any specific combination of perceived role stressors (e.g., “high”) and job scope (e.g., “moderate”) that is a valid representation of boundary roles. Specifically, it posits that only a few, dominant combina-tions—termed “configurations”—are plausible that represent shared interpretive schemas of job contexts (Meyer, Tsui, and Hinings 1993). That is, a configurational perspective rejects the no-tion that infinite combinations of role stressors and job scope are empirically plausible as if these factors could be varied independently. Instead, it accepts the view that factors are interdependent and often can change only discretely. For instance, boundary roles involving extreme combina-tions of contrasts—such as high stress, low scope (cell 3, Figure 6.2) and high scope, low stress (cell 7, Figure 6.2)—are unlikely to be obtained empirically because they would be either so unbearable or unviable to be sustained as reasonable boundary roles in modern organizations. Likewise, it may be pragmatically difficult, if not impossible, to vary job conditions finely to obtain a continuous range of gradations. As we noted at the outset, practitioner views differ from academic thought on whether boundary roles can be redesigned as low-stress jobs. Practitioners’ assertions that such redesign is difficult at best due to the stress inherent in boundary roles indicate empirical limits on the viable range of role stressors in boundary jobs. However, practitioners are not passive principals (e.g., managers) unconcerned about agent (e.g., boundary spanner) stress and effectiveness. Rather, as active mangers, practitioners actively design jobs to allow boundary spanners to cope with their role stressors and effectively serve organizational goals. Such design
MANAGING CUSTOMER RELATIONSHIPS 175
efforts often involve supplying job scope to match role stressors, resulting in job contexts that gravitate toward the diagonal noted as the “axis of congruence” in Figure 6.2. Taken together, the preceding considerations suggest that only a few configurations of role stressors and job scope are likely to be empirically plausible.
Moreover, each configuration represents a schema about job context that is shared by collectives of boundary spanners. That is, a configuration is not defined at the individual level of analysis. Rather, it represents characteristics of the job that are shared by groups of boundary spanners. In this sense, configurations of job context are less sensitive to individual variability than operational-izations of job context in contingency or universalistic approaches. Debates about the appropriate level of analysis for conceptualizing job contexts have focused on two competing positions. On one hand, proponents of subjective experiences of work argue that an individual-level analysis is appropriate because it captures how boundary roles are perceived by individuals who fill them and any effort to aggregate experiences denies the fundamental place of individuals in organiza-tions. On the other hand, job design researchers take a managerial perspective to argue for an organizational unit as the appropriate level of analysis to capture largely “objective” aspects of boundary roles populated in the unit and reject subjective experiences as noisy data that is less useful for managerial efforts in designing jobs. As the subjective-objective debate continues in the literature, the configural perspective offers an intermediate position that bridges these perspectives. The configural perspective does not deny the relevance of individual perceptions of boundary roles. The starting point for a configural perspective is the boundary role occupant’s perceptions of role stressors and job scope inherent in his/her job. However, in construing job contexts as combinations of role stressors and job scope, the configural approach moves forward to identify configurations that are shared by collectives of boundary spanners. Why would cohesive collective schemas emerge? We posit that characteristics of boundary roles are interpreted by individuals through a process of interaction with job design practices, cognitive appraisals, and sense-mak-ing, and the resulting interpretations are shared, refined, and updated through a process of social interactions among boundary spanners resulting in shared collective schemas (James, Joyce, and Slocum 1988; Young and Parker 1999). Moreover, these collective schemas need not faithfully reproduce the objective work design features nor adhere to department/unit boundaries; yet, they meaningfully capture the patterns of boundary spanner interactions and systematically relate to critical employee and organizational outcomes (Young and Parker 1999). In this sense, configu-rations of boundary roles are jointly determined by objective work design efforts and subjective interpretations of boundary spanners.
Using a configural perspective to study job contexts has several advantages including testing for equifinality and modeling flexibility (Meyer, Tsui, and Hinings 1993; Miller and Friesen 1984; Doty, Glick, and Huber 1993). The notion of equifinality contrasts with the linear postulate that is common to most previous studies. The linear postulate posits that, given any two job contexts, it is possible to identify a single configuration that is ideally more favorable than the other because it provides more of the desirable characteristics (e.g., satisfaction, performance), and fewer of the undesirable characteristics (e.g., burnout, turnover intentions). In contrast, the equifinality proposition argues that it is highly unlikely that any configuration is significantly superior to other plausible configurations across all dependent variables considered. Rather, the more likely sce-nario is that two or more configurations are equally effective for some dependent variables (e.g., performance) and differentially effective for other variables (e.g., satisfaction). In other words, an equifinal view denies the presumed superiority of any specific configuration. In this sense, the proposition of equifinality embodies the notion of nonlinear effects. Such equifinality notions are, at best, difficult to test under universalistic and contingency approaches.
176 RUTH N. BOLTON AND CRINA O. TARASI
Finally, a configural perspective allows considerable flexibility in obtaining valid configurations including (a) deductive, theory-driven typology, and (b) inductive, empirical-driven taxonomy (Doty and Glick 1994). Typology-based configurations draw from a strong theoretical framework that specifies the number of plausible configurations, how each configuration is defined, and the rationale for the differential influence of different configurations. Because deductive configura-tions may or may not be empirically viable, they are often referred to as “ideal” types. By contrast, taxonomical configurations begin with data and extract viable configurations using aggregation procedures (e.g., cluster or latent class analysis). In order to obtain valid configurations, typical aggregation procedures are flexible to allow for (a) theory-based “centroids” to guide inductive procedures, and (b) split-sample designs that reduce the confounding effect of sampling variability. In this sense, not all logically plausible configurations can be empirically viable (Meyer, Tsui, and Hinings 1993). Readers will note that this concern with obtaining valid configurations when combined with tests for equifinality allows a finer-grained study of job contexts and their influence on boundary spanner processes than possible under universalistic or contingency perspectives.
Empirical Findings
Few, if any, studies have utilized a configural perspective to study role theory effects for boundary spanners. One such exceptional study is by Payne and Fletcher (1983). Using a sample of 148 teachers in UK, Payne and Fletcher utilized inductive procedures to extract empirically viable configurations. Foreshadowing later work, Payne and Fletcher utilized multiple dimensions of demands (e.g., disciplinary demands, maintaining standards, workload demands) and discretion (e.g., interpersonal support, job discretion) to faithfully capture the richness of teachers’ job con-text. In all, seven distinct configurations of job contexts were obtained that differed in terms of the demands, discretion, and constraints. Unfortunately, Payne and Fletcher did not compare their taxonomical configurations with theoretically developed “ideal” job contexts. Nevertheless, they found that all of the obtained configurations were equifinal—that is, there were no differences in terms of outcomes, with a single exception. One configuration with low levels of demands and high levels of discretion produced a significantly higher level of job satisfaction than any other job context. Payne and Fletcher did not thoroughly investigate these counterintuitive results. In addi-tion, while Payne and Fletcher pointed out the need for studies that sample heterogeneously from a well-defined sampling frame, they appear to be unaware of the empirical problems in detecting nonlinear and interaction effects with homogenous samples. No other study could be traced that had utilized a configural perspective to examine role theory effects.
Because the configural perspective approaches the phenomenon differently than contingency and universalistic approaches, it is likely that such a pursuit will yield new insights into the influ-ence of role stressors and job scope in boundary spanning positions. Despite its shortcomings, the Payne and Fletcher study provides an initial indication that a configural perspective may be rewarding due to its potential to reveal equifinal effects. At the minimum, a configural perspective is likely to provide findings that would be useful to triangulate with contingency and universalistic findings to yield a holistic understanding of the stress processes among boundary spanners.
Potential Contributions and Concluding Notes
Looking at the body of literature on boundary role stress and its multiple meta-analyses, one is prone to conclude that room for potential contributions is limited and the hurdle steep. Our review is intended to dispel this view. Beyond the universalistic approach, vast areas remain unexplored.
MANAGING CUSTOMER RELATIONSHIPS 177
Although the contingency perspective has received some attention, the study of contingencies that modulate the effect of role stressors for marketing-oriented boundary spanners has lagged. Market-ing researchers appear to be more interested in identifying the different and emergent sources of role stress (e.g., work–family conflict) than in guiding boundary role design efforts that facilitate coping with role stressors or, more boldly, transforming role stressors into an energizing force for active and fulfilling engagement in boundary roles. Possibly, the terminology is a stumbling block. Socialized interpretations of stress appear to favor negative and avoidance representations that restrict construals of stress to dysfunctional mechanisms for boundary role effectiveness. Karasek’s (1979) efforts to construe stress as role demands to blunt socialized representations inseparable in the former appear to have had little impact in the marketing literature as evident from Table 6.1, where marketing studies are conspicuously absent. Alternatively, marketing researchers accept Karasek’s theoretical arguments but are unmoved by his mixed empirical evidence. Lack of clear support for interaction effects does render the complexity of Karasek’s contingency model unattractive. Complexity in theorizing with uncertainty in empirical payoffs poses another stumbling block.
Our review favors a view of the preceding stumbling blocks that suggests opportunities waiting to be exploited. Overriding socialized representation of stress that allows for its functional and eustress effects has considerable practitioner and theoretical appeal. To exploit this opportunity, marketing researchers might find it useful to coalesce around common terminologies that reframe the notion of stress in boundary-spanning positions. One option is to rid stress of its presumed negative or positive connotations, and to posit theoretically driven contingencies that result in its positive or negative effects. In this sense, whether stress is negative or positive is determined contextually subject to coping resources afforded by contexts and coping capabilities deployed by individuals operating in their boundary contexts.
Taking a totally different tack, another option is to develop alternative terminologies and redefine terms more precisely. A possible approach is to explicitly distinguish between role stressors—the degree to which boundary role characteristics pose demands on individual capabilities and re-sources, and require effortful coping, and role stress—the degree to which boundary spanners experience psychological and physiological symptoms indicating that role stressors are exceeding individual coping capabilities and resources (e.g., anxiety). As such, the concept of role stress retains its negative connotation and is indicative of job contexts that are dysfunctional for bound-ary role effectiveness because resources afforded by job contexts are not sufficient to facilitate boundary spanner coping with the inherent role stressors. By contrast, role stressors can take on either a positive or negative meaning—as Karasek’s original notion of role demands does—given the potential for distress or eustress depending on job scope or other contextual contingencies. We see promise in this approach and, indeed, have used it throughout this paper. We suggest its serious consideration in future studies to overcome terminological blocks.
Despite the mixed results of contingency hypotheses, several reasons suggest that this is a fruit-ful avenue for studying effectiveness of marketing-oriented boundary spanners. First, contingency effects are contextually dependent such that what works for organizational employees may or may not work for marketing-oriented boundary spanners. Customers represent a uniquely different boundary spanning problem compared to other boundary spanning roles involving working with either internal employees in other departments or suppliers. Customers introduce considerable heterogeneity and unpredictability to boundary exchanges. As such, boundary roles involving customers can present a unique profile of role stressors, and require considerable job scope to cope with inherent challenges. What is less salient in other boundary roles may well be quite prominent in marketing-oriented boundary spanners. Second, recent methodological advances have made it
178 RUTH N. BOLTON AND CRINA O. TARASI
feasible to empirically detect interaction effects by providing tractable approaches for controlling measurement error and testing nonlinear effects (Marsh, Wen, and Hau 2004; Cortina, Chen, and Dunlap 2001; Ganzach 1997). Current experience with these approaches suggests that disregard-ing measurement error severely hinders the detection of contingent effects, thereby providing biased results and rendering much previous research problematic. Third, more recent studies have reported support for the contingency hypotheses indicating that the influence of role stressors is not only contingent on job scope but also on other job factors including support (Van Yperen and Hagedoorn 2003). Together, the preceding developments point to potential payoffs from future studies that posit theory-based contingency hypotheses and examine them using methodological approaches that control for confounding influences including measurement error.
While the contingency approach requires reconsideration by marketing researchers, the con-figural approach represents a new frontier of unexplored possibilities. The configural approach begins from different assumptions about boundary role contexts and construes the study of boundary role effectiveness from a different lens, one that inherently accounts for nonlinear and higher-order interactions, allows collective-level perceptions of role dynamics, incorporates the notion of equifinality—meaning that different configurations may have similar job outcomes—and affords flexibility in distinguishing between “ideal type” work contexts and “empirically viable” ones. We could not trace a single study in marketing that had utilized this approach for examining role theory effects for boundary spanners. Even in the extant body of work, empirical work on configural approach is lacking. We had traced a study by Payne and Fletcher (1983) that appeared consistent with this approach but neither explicitly nor fully considered the configural perspec-tive. We view this gap as a significant opportunity for providing insights into role theory effects for marketing-oriented boundary spanners. To the extent that these insights compare and contrast with those obtained from universalistic and contingency perspectives, there opens a dialogue on triangulation efforts and on the ecological validity of assumptions underlying differing approaches. Exposing assumptions and pressing for a critical assessment of their empirical reasonableness holds considerable promise in advancing our understanding of role theory effects for marketing-oriented boundary spanners. Current studies locked in largely universalistic approach are incapable of raising such fundamental questions.
In closing, there is little disagreement about the burden and magnitude of individual, organiza-tional, and societal cost of stress, especially on the boundaries that define an organization’s interfaces with its customers and society at large. Disagreements arise both in identifying when role stressors exceed functional levels, and in selecting the “best” strategy for combating the consequences of boundary role stress. To bridge these disagreements and provide new insights, we provide a com-parative review of different approaches, propose a configural approach that views job contexts as organizational or situational characteristics defined by specific stressor-scope combinations, and outline directions for future research that take advantage of the plurality of approaches. Overall, our review challenges marketing researchers to explicitly consider what boundary spanners do when they face role stressors when they aim to examine what role stressors boundary spanners face and what are their consequences. Examining role stressors without considering boundary spanner coping efforts is like studying water level in a bucket by observing the inflows without considering the outflows. Coping with role stressors is effortful and demands resources that either come from job scope supplied by the context or from an internal reservoir, which results in deple-tion afterward. It can be argued that organizations are responsible for providing a balanced job context where job scope provides sufficient resources to cope with role stressors without depleting internal reservoirs of boundary role occupants. Toward this end, our review implores marketing researchers to adopt a role design perspective that considers how job context can be configured for
MANAGING CUSTOMER RELATIONSHIPS 179
optimal effectiveness. Boundary role stress is not an individual problem, nor is it just a matter to be addressed organizational design. Rather, it requires a simultaneous consideration of job context and individual factors to develop active, challenging, and resourceful jobs that fulfill Karasek’s ambition for healthy work. We hope our study provides the impetus for such pursuit.
References
Abdel-Halim, Ahmed A. 1981. “Effects of Role Stress-Job Design-Technology Interaction on Employee Work Satisfaction.” Academy of Management Journal 24: 260–273.
Behrman, Douglas N., and William D. Perrault. 1984. “A Role Stress Model of the Performance and Satis-faction of Industrial Salespersons.” Journal of Marketing 48: 9–21.
Belasco, James E. 1966. “The Salesman’s Role Revisited.” Journal of Marketing 30: 6–11.Blood, Milton R., and Charles L. Hulin. 1967. “Alienation, Environmental Characteristics, and Workers’
Responses.” Journal of Applied Psychology 51: 284–306.Brown, Steven P., and Robert A. Peterson. 1993. “Antecedents and Consequences of Salesperson Job Satisfac-
tion: Meta-Analysis and Assessment of Causal Effects.” Journal of Marketing Research 30: 63–77.Busemeyer, Jerome R., and Lawrence E. Jones. 1983. “Analysis of Multiplicative Combination Rules When
the Casual Variables Are Measures with Error.” Psychological Bulletin 93: 549–562.Cavanaugh, M.A., W.R. Boswell, M.V. Roehling, and J.W. Boudreau. 2000. “An Empirical Examination of
Self-Reported Work Stress among U.S. Managers.” Journal of Applied Psychology 85: 65–74.Champoux, Joseph E. 1978. “A Preliminary Examination of Some Complex Job Scope-Growth Need Strength
Interactions.” Academy of Management Proceedings 38: 59–63.———. 1992. “A Multivariate Analysis of Curvilinear Relationships among Scope Work Context Satisfac-
tion and Affective Outcomes.” Human Relations 45: 87–111.Churchill, Gilbert A., Neill M. Ford, Steven W. Hartley, Orville C. Walker. 1985. “Determinants of Sales
Performance.” Journal of Marketing Research 22: 103–118.Cortina, Jose M., Gilad Chen, and William P. Dunlap. 2001. “Testing Interaction Effects in LISREL: Exami-
nation and Illustration of Available Procedures.” Organizational Research Methods 4: 324–360.Cote, Joseph A., and M. Ronald Buckley. 1988. “Measurement Error and Theory Testing in Consumer Research:
An Illustration of the Importance of Construct Validation.” Journal of Consumer Research 14: 579–582.Csikszentmihalyi, Mihaly. 1975. Beyond Boredom and Anxiety. San Francisco, CA: Jossey-Bass.Daniels, Kevin, and Andrew Guppy. 1994. “Occupational Distress, Social Support, Job Control, and Psy-
chological Well-Being.” Human Relations 47: 1523–1544.De Jonge, Jan, Gerard J.P. van Breukelen, Jan A. Landeweerd, and Frans J.N. Nijhuis. 1999. “Comparing
Group and Individual Level Assessments of Job Demand-Control Model: A Multilevel Approach.” Hu-man Relations 52: 95–122.
Doty, D. Harold, and William H. Glick. 1994. “Typologies as a Unique Form of Theory Building: Toward Improved Understanding and Modeling.” Academy of Management Review 19: 230–251.
Doty, D. Harold, William H. Glick, and George P. Huber. 1993. “Fit, Equifinality, and Organizational Ef-fectiveness: A Test.” Academy of Management Journal 36: 1196–1250.
Dwyer, Deborah J., and Daniel C. Ganster. 1991. “The Effects of Job Demands and Control on Employee Attendance and Satisfaction.” Journal of Organizational Behavior 12: 595–608.
Fisher, Cynthia D., and Richard J. Gitelson. 1983. “A Meta-Analysis of the Correlates of Role Conflict and Ambiguity.” Journal of Applied Psychology 68: 320–333.
Fletcher, Ben C., and Fiona Jones. 1993. “A Refutation of Karasek’s Demand-Discretion Model of Oc-cupational Stress with a Range of Development Measures.” Journal of Organizational Behavior 14: 319–330.
Fox, Marilyn L., Deborah J. Dwyer, and Daniel C. Ganster. 1993. “Effects of Stressful Job Demands and Control on Psychological and Attitudinal Outcomes in a Hospital Setting.” Academy of Management Journal 36: 289–312.
Ganzach, Y. 1997. “Misleading Interaction and Curvilinear Terms.” Psychological Methods 3: 235–247.Gardner, Donald G., and Larry L. Cummings. 1988. “Activation Theory and Job Design: Review and Re-
conceptualizaiton.” In Research in Organizational Behavior, ed. B.M. Staw and L.L Cummings, 81–122. Greenwhich, CT: JAI Press.
180 RUTH N. BOLTON AND CRINA O. TARASI
Hackman, J. Richard, and Greg R. Oldham. 1976. “Motivation through the Design of Work: Test of a Theory.” Journal of Applied Psychology 59: 638–642.
Hobfoll, Stevan E., and John Freedy. 1993. “Conservation of Resources: A General Stress Theory Applied to Burnout.” In Professional Burnout: Recent Developments in Theory and Research, ed. Christina Maslach, Wilmar B. Schaufeli, and Tadeusz Marek, 115–133. Washington, DC: Taylor & Francis.
Jackson, Susan, and Randall Schuler. 1985. “A Meta-Analysis and Conceptual Critique of Research on Role Ambiguity and Role Conflict in Work Setting.” Organizational Behavior and Human Decision Process 36: 16–78.
Jamal, Muhammad. 1984. “Job Stress and Job Performance Controversy: An Empirical Assessment.” Orga-nizational Behavior and Human Decision Processes 33: 1–21.
James, Lawrence R., William F. Joyce, and John W. Slocum. 1989. “Comment: Organizations Do Not Cog-nize.” Academy of Management Review 13: 129–132.
Kahn, Robert L., and Philippe Byosiere. 1992. “Stress in Organizations.” In Handbook of Industrial and Organizational Psychology, ed. Marvin D. Dunnette and Leaetta M. Hough. Palo Alto, CA: Consulting Psychologists Press.
Kahn, Robert L., Don Wolfe, Robert Quinn, and J. Diedrick Snoek. 1964. Organizational Stress: Studies in Role Conflict and Ambiguity. New York, NY: John Wiley & Sons.
Karasek, Robert A. 1979. “Job Demands, Job Decision Latitude, and Mental Strain: Implications for Job Redesign.” Administrative Science Quarterly 24: 285–308.
Kelloway, E. Kevin, and Julian Barling. 1991. “Job Characteristics, Role Stress and Mental Health.” Journal of Occupational Psychology 64: 291–304.
King, L.A., and D.W. King. 1990. “Role Conflict and Role Ambiguity: A Critical Assessment of Construct Validity.” Psychological Bulletin, 107: 48–64.
Landsbergis, Peter A. 1988. “Occupational Stress among Health Care Workers: A Test of the Job Demands-Control Model.” Journal of Organizational Behavior 9: 217–239.
MacCallum, Robert C., and Corinne M. Mar. 1995. “Distinguishing between Moderator and Quadratic Ef-fects in Multiple Regression.” Psychological Bulletin 118: 405–421.
Marino, S. 1997. “The Stress Epidemic.” Industry Week 246: 14.Marsh, Herbert W., Zhonglin Wen, and Kit-Tai Hau. 2004. “Structural Equation Models of Latent Interactions: Evalu-
ation of Alternative Estimation Strategies and Indicator Construction.” Psychological Methods 9: 275–396.Maslach, Christina, and Michael P. Leiter. 1997. The Truth about Burnout. San Francisco, CA: Jossey-
Bass.Merton, Robert K. 1949. Social Theory and Social Structure: Toward the Codification of Theory and Re-
search. Glencoe, IL: Free Press.Meyer, Alan D., Anne S. Tsui, and C. Roy Hinings. 1993. “Configurational Approaches to Organizational
Analysis.” Academy of Management Journal 36: 1175–1195.Miller, Danny, and Peter H. Friesen. 1984. Organizations: A Quantum View. Englewood Cliffs, NJ: Pren-
tice-Hall.Mohrman, Susan A., and Susan G. Cohen. 1995. “When People Get Out of the Box: New Relationships, New
Systems.” In The Changing Nature of Work, ed. A. Howard. San Francisco, CA: Jossey-Bass.Newton, Tim. 1995. Managing Stress: Emotion and Power at Work. Thousand Oaks, CA: Sage.Payne, Roy, and Ben C. Fletcher. 1983. “Job Demands, Support, and Constraints as Predictors of Psychologi-
cal Strain among Schoolteachers.” Journal of Vocational Behavior 22: 136–147.Rhoads, Gary K., Jagdip Singh, and Phillips Goodell. 1994. “The Multiple Dimensions of Role Ambiguity
and Their Impact on Psychological and Behavioral Outcomes Industrial Salespeople.” Journal of Personal Selling & Sales Management 14: 1–24.
Rizzo, John R., Robert House, and Stephen I. Lirtzman. 1970. “Role Conflict and Ambiguity in Complex Organizations.” Administrative Science Quarterly 15: 150–163.
Rommetveit, Ragnar. 1954. Social Norms and Roles. Minneapolis: University of Minnesota Press.
Schaubroeck, John, and Laurence S. Fink. 1998. “Facilitating and Inhibiting Effects of Job Control and So-cial Support on Stress Outcomes and Role Behavior: A Contingency Model.” Journal of Organizational Behavior 19: 167–195.
Schaubroeck, John, and Daniel C. Ganster. 1993. “Chronic Demands and Responsivity to Challenge.” Journal of Applied Psychology 78: 73–85.
MANAGING CUSTOMER RELATIONSHIPS 181
Schmidt, Frank L., and John E. Hunter. 1996. “Measurement Error in Psychological Research: Lessons from 26 Research Scenarios.” Psychological Methods 1: 199–223.
Schwab, Donald P., and Larry L. Cummings. 1976. “A Theoretical Analysis of the Impact of Task Scope on Employee Performance.” Academy of Management Review 1: 23–35.
Selye, Hans. 1976. The Stress of Life. New York: McGraw-Hill.Singh, Jagdip. 1993. “Boundary Role Ambiguity: Facets, Determinants and Impacts.” Journal of Market-
ing, 57: 11–31———. 1998. “Striking a Balance in Boundary-Spanning Positions: An Investigation of Some Unconven-
tional Influences of Role Stressors and Job Characteristics on Job Outcomes of Salespeople.” Journal of Marketing, 62: 69–87.
Singh, Jagdip, Jerry R. Goolsby, and Gary K. Rhoads. 1994. “Behavioral and Psychological Consequences of Boundary Spanning Burnout for Customer Service Representatives.” Journal of Marketing Research, 31: 558–569
Spector, Paul E. 1987. “Interactive Effects of Perceived Control and Job Stressors on Affective Reactions and Health Outcomes for Clerical Workers.” Work and Stress 1: 155–162.
Spector, Paul E., Deborah J. Dwyer, and Steve M. Jex. 1988. “Relation of Job Stressors to Affective, Health, and Performance Outcomes: A Comparison of Multiple Data Sources.” Journal of Applied Psychology 73: 11–19.
Tosi, Henry. 1992. The Environment/Organization/Person Contingency Model: A Meso Approach to the Study of Organizations. Greenwich, CT: JAI Press.
Tubre, Travis C., and Judith M. Collins. 2000. “Jackson and Schuler (1985) Revisited: A Meta-Analysis of the Relationships between Role Ambiguity, Role Conflict and Job Performance.” Journal of Manage-ment 26: 155–169.
Van Yperen, Nico W., and Mariet Hagedoorn. 2003. “Do High Job Demands Increase Motivation or Fa-tigue or Both? The Role of Job Control and Job Social Support.” Academy of Management Journal 46: 339–348.
Warr, Peter B. 1990. “Decision Latitude, Job Demands, and Employee Well-Being.” Work and Stress 4: 285–294.
Xie, Jia L., and Gary Johns. 1995. “Job Scope and Stress: Can Job Scope Be Too High?” Academy of Man-agement Journal 38: 1288–1309.
Yerkes, Robert M., and John Dodson. 1908. “The Relation of Strength of Stimulus to Rapidity of Habit Formation.” Journal of Comparative Neurology and Psychology 10: 459–82.
Young, Scott A., and Christopher P. Parker. 1999. “Predicting Collective Climates: Assessing the Role of Shared Work Values, Needs, Employee Interaction, and Work Group Membership.” Journal of Organi-zational Behavior 20: 1199–1218.