Jro-~ ‘f . :- . . . '- ~‘ 4’. tum '0 I6

 

 

 

 

THESIS

 

 

This is to certify that the
dissertation entitled
CUSTOMERS AS SUBSTITUTES FOR LEADERSHIP IN SERVICE ORGANIZATIONS:

THEIR ROLE AS NON-LEADER SOURCES OF GUIDANCE AND SUPPORT

presented by

David Earl Bowen

has been accepted towards fulfillment

ofthe requirements for

Ph.D. (mgnmin Management

 

 

Major farofessor

Dategluly 27; 1983

MS U is an A/firmun'w Action/Equal Opportunity Instilurmn Or 12771

 

MSU

LlBRARlES

y

 

 

 

RETURNING MATERIALS:
Place in book drop to
remove this checkout from
your record. FINES will
be charged if book is
returned after the date
stamped below.

 

 

M

0 NOT CIRCULATE

, D
AUG 2 A 200'»

/\§‘Vu‘}”éf I

 

Eu
OCl E b 3996

EP
/ .

 

NOV-i

M USE ONIY

I 1'84
’3 H??? "”1":

6; :99
i Q lQQ?

, A

3- 'L‘AZ i

II I
*{A

 

CUSTOMERS AS SUBSTITUTES FOR LEADERSHIP IN SERVICE
ORGANIZATIONS: THEIR ROLE AS NON-LEADER
SOURCES OF GUIDANCE AND SUPPORT

By

David Earl Bowen

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Management

1983

ABSTRACT
CUSTOMERS AS SUBSTITUTES FOR LEADERSHIP IN SERVICE
ORGANIZATIONS: THEIR ROLE AS NON-LEADER
SOURCES OF GUIDANCE AND SUPPORT
By

David Earl Bowen

This study examined the different sources of guidance and
support that influence the attitudes and performance of employees
in service organizations. The different sources studied were: (1)
leaders' instrumental behavior (i.e., telling employees what to do)
and supportive behavior (i.e., displaying concern for employees'
well—being) and, (2) several possible "substitutes for leadership"
(i.e., non—leader characteristics that can act in place of leader
behaviors): organizational formalization, professional
orientation, and—-most centrally——customers.

Customers were said to act as substitutes by displaying
instrumental and supportive behaviors and by enriching the
characteristics of tasks employees perform. Customers were
hypothesized to substitute for leader behaviors’ influence upon
employee job satisfaction, employee service quality views, the
performance of the department in which employees worked, and
customer views of the quality of service provided by the
department.

Survey data were collected from 525 employees and 2,680
customers of retail, eye-care departments. Also, management

provided rankings of department performance.

David Earl Bowen

Results revealed that when employee job satisfaction and
employee service quality views were the criteria, both perceived
customer and leader behaviors were significant correlates. Thus,
customers acted as supplements, rather than as substitutes, for
leadership. For department rank, customers did substitute for
leadership since employee perceptions of customer behaviors were
significant correlates but perceived leader behaviors were not.
However, for customer service quality views, employee perceptions
of how customers enriched their task characteristics was the only
aspect of customer influence to be a significant correlate.
Customers as a source of task enrichment was classified as only a
supplement since all leader behaviors were also significant
correlates. Finally, customer supportive behavior was positively
correlated with all four criteria, whereas customer instrumental
behavior was negatively correlated with all four criteria.

Organizational formalization acted mainly as a supplement
for leadership. Professional orientation was dropped from the
analyses due to measurement problems.

It was concluded that customers can be an important source
of influence in service organizations. This fact requires
attention in the design of service organizations and in future work

on leadership and its substitutes.

ACKNOWLEDGEMENTS

There are many people to whom I owe a great deal for the
completion of this dissertation. My dissertation committee heads
the list. Those individuals were Dr. Benjamin Schneider, Chairman,
and Drs. John Wanous and Mary Van Sell. They acted as sources of
guidance and support for which there could be no substitutes!

I want to thank Ben Schneider, particularly, for his
patience and generosity in mentoring my dissertation effort,
specifically, and my graduate education, more generally. He has
always provided me a model of excellence in both his life at work
and with his family. He remains for me the guiding example of what
an accomplished academic should be.

I want to thank John Wanous for more than assisting in the
dissertation. John more than anyone helped me through my early
anxious moments of the doctoral program through a combination of
humor, intellectual guidance, and——ever so important at that
point-~friendship. It grew to be a friendship that I sense-—
certainly hope-—will be a lasting one.

Mary I thank for sound methodological advice, particularly,
and for being a sympathetic listener down the homestretch.

I want to thank Mike Fitzgerald, Rand Gottschalk, Nancy
Moeller and Elaine Pulakos for their computer assistance. Elaine

ii

really helped me get underway. Near the end, Rand gave unselfishly
to his time when I was experiencing particular difficulty—-and
anxiety.

A number of colleagues cheered me and steered me during the
dissertation effort. I most want to thank Drs. Gareth Jones,
Janina Latack, Elaine Pulakos, Arnon Reichers, and Sara Rynes.
What good friends!

The secretaries of the Department of Management helped me
again and again. They are Marie Dumeney, Kathy Jegla, and Clair
Rammel. Thank you, ladies, for your incredible patience with my
endless questions about what I needed to do to submit my
dissertation, obtain my degree, etc.

My typist, David Turner, is to be thanked for responding so
well to numerous unreasonable deadlines and for offering useful
editorial adivce.

I thank my parents for understanding all those times I was
too busy to get together.

Most of all—~without slighting those already mentioned—-I
thank God. I thank Him for whatever ability I have. Most
fundamentally, I thank Him for those occasions when--feeling alone

or helpless——I knew with Him beside me I could be neither.

iii

TABLE OF CONTENTS

LIST OF TABLES . . .
Chapter
I. INTRODUCTION AND REVIEW OF THE LITERATURE. . . . .

LEADER AND NON—LEADER SOURCES OF INFLUENCE

UPON EMPLOYEES . . . . . . . . . . .

LEADERSHIP THEORY: ASSUMPTIONS, SHORTCOMINGS

AND SOME RECENT APPROACHES. . . . . . .

SUBSTITUTES FOR LEADERSHIP . . . . . . . . . . .
PREVIOUS RESEARCH ON SUBSTITUTES FOR
LEADERSHIP . . . . . . . . . . . . .
A SUMMARY VIEW OF LEADER AND NON—LEADER SOURCES
OF INFLUENCE UPON EMPLOYEES. . . . . . . .
ORGANIZATIONAL BEHAVIOR IN THE SERVICE SECTOR. .
THE BOUNDARIES OF THE SERVICE SECTOR . . .
THE UNIQUE ORGANIZATIONAL DYNAMICS OF
THE SERVICE SECTOR . . . . . . . . .

CUSTOMERS AS SUBSTITUTES FOR LEADERSHIP IN

SERVICE ORGANIZATIONS. . . . . . . . . . . .

HOW CUSTOMERS MAY ACT AS A SUBSTITUTE FOR
LEADERSHIP . . . . . . . . . . . . .
HYPOTHESES . . . . . . . . . . . . . . . .

iv

viii

12

32

II.

III.

METHODOLOGY. . . . . . . . .
RESEARCH DESIGN. . . . . . . . . . . . . . . .
SETTING AND PARTICIPANTS . . . . . . . . . . .
DATA COLLECTION PROCEDURES . . . . . . . . . .

EMPLOYEE AND CUSTOMER INTERVIEWS . . . .
SURVEY ADMINISTRATION. . . . . . . . . .

DATA FROM SENIOR MANAGEMENT. . . . . . .
MEASURES . . . . . . . . . . . . . . . . . . .
CUSTOMERS AS A SUBSTITUTE FOR LEADERSHIP:
ACTING AS AN INTERPERSONAL SOURCE
OF SUPPORTIVE AND INSTRUMENTAL
BEHAVIOR . . . . . . . . . . . . .
CUSTOMERS AS A SUBSTITUTE FOR LEADERSHIP:
ACTING AS A SOURCE OF TASK
ENRICHMENT . . . . . . . . . . . .
PROFESSIONAL ORIENTATION . . . . . . . .
ORGANIZATIONAL FORMALIZATION . . . . . .
LEADER BEHAVIORS . . ....... . . .
JOB SATISFACTION . . . . . . . . . . . .
EMPLOYEE AND CUSTOMER PERCEPTIONS OF
SERVICE QUALITY. . . . . . . . . .
DEPARTMENT RANK. . . . . . . . . . . . .
DATA ANALYSIS. . . . . . . . . . . . . . . .
RESULTS. . . . . . . . . . . . . . . . . . . . . .
INTERCORRELATIONS AMONG VARIABLES. . . . . . .

V

61
61
61
62

62
63

64

65

65

68
71
72
72

76

76
78
78
84

84

IV.

DISCUSSION. . . . . ...... .

CORRELATES OF EMPLOYEE JOB SATISFACTION:
INDIVIDUAL—LEVEL ANALYSIS WITH ALL
EMPLOYEES. . . . . . . . . . . . .

CORRELATES OF EMPLOYEE PERCEPTIONS OF
SERVICE QUALITY: INDIVIDUAL—LEVEL
ANALYSIS WITH ALL EMPLOYEES.

CORRELATES OF CUSTOMER PERCEPTIONS OF
SERVICE QUALITY: DEPARTMENT—LEVEL
ANALYSIS WITH AGGREGATED DISPENSER

DATA . . . . . . . . .

CORRELATES OF DEPARTMENT RANK: DEPARTMENT—

LEVEL ANALYSIS WITH AGGREGATED
DISPENSER DATA . .
SUMMARY COMPARISONS OF CUSTOMER VERSUS
LEADER INFLUENCE UPON EMPLOYEE
ATTITUDES AND PERFORMANCE. . . .

SUMMARY. . . . . . . . . . . . . . . .

THE IMPLICATIONS OF CUSTOMER INFLUENCE
FOR DESIGNING AND MANAGING SERVICE

ORGANIZATIONS. . . . . . . . . . .

o c o a o o o a I

THE ISSUE OF SELECTING AND SOCIALIZING

CUSTOMERS TO FILL PRESCRIBED
ORGANIZATIONAL ROLES . . . .
THE IMPLICATIONS FOR THEORY AND RESEARCH

vi

89

96

99

104

. 106

110

111

116

119

 

IN LEADERSHIP AND SUBSTITUTES FOR

LEADERSHIP ........ . . . . . . . . . . 123

EXPLAINING THE NEGATIVE CORRELATIONS

INVOLVING CUSTOMER INSTRUMENTAL

BEHAVIOR . . . . . . . . . . . . . . . . 123
LEADERSHIP, SUBSTITUTES, AND SUPPLEMENTS:

ALTERNATIVE METHODS OF INFLUENCING

EMPLOYEE ATTITUDES AND PERFORMANCE. . . 126
SOME SUGGESTIONS FOR THE CHOICE OF

METHODS AND VARIABLES IN FUTURE

RESEARCH ON SUBSTITUTES FOR

LEADERSHIP ..... . . . . ...... 135

CONCLUSION . .................... 141
APPENDICES

A. FORMULAS CITED ............... . . . . . 144

B. MEANS AND STANDARD DEVIATIONS OF VARIABLES

AT THE INDIVIDUAL—LEVEL WITH ALL EMPLOYEES . . . . . 146
C. MEANS AND STANDARD DEVIATIONS OF VARIABLES

AT THE DEPARTMENT—LEVEL WITH AGGREGATED

DISPENSER DATA ....... . . . . ........ 147

LIST OF REFERENCES ...........

vii

10.

11.

12.

13.

LIST OF TABLES

Potential Substitutes for Leadership
(Adapted from Kerr and Jermier, 1978). . . .

The Kerr and Jermier (1978) Study. .

The Howell and Dorfman (1981a) Study . .

The Howell and Dorfman (1981b) Study . . . .

The Sheridan, Vrdenburgh, and Abelson (1981)
Study. . . . . . . . . . . . . . . . .
Stages of Economic Activity (Adapted from
Foote and Hatt, 1953). . . . . . .
Customers as a Substitute for Leadership:
Acting as an Interpersonal Source of
Instrumental and Supportive Behaviors. . . .

Customers as a Substitute for Leadership:
Acting as a Source of Task Enrichment. . .

Professional Orientation and Organizational
Formalization. . . . . . . .

Leader Behavior Scales . . . . . . . . . . .
Data Analysis Outcomes for Determining If a
Characteristic Is a Substitute . . . . . . .
Intercorrelations among Variables at the
Individual Level . . . . . . . . . . . .
Intercorrelations among Variables at the
Department Level . . . . . . . . . . . .
Comparison of Leadership and Potential
Substitutes' Relationships with Employee Job
Satisfaction (Individual—Level Analysis with
All Employees) . . . . . . . . . . . . . .

viii

14
34
35
37

39

42

67

7O

73
74

8O

85

87

90

15.

16.

17.

19.

20.

21.

22.

23.

24.

Revised Comparisons of Leaders Behaviors
Versus Customer Behaviors Relationships with
Employee Job Satisfaction. . . . . . . . . .

Hierarchical Multiple Regressions of
Employee Job Satisfaction on Various
Potential Substitute and Leadership
Combinations . . . . . . . . . . . . . . . .

Comparison of Leadership and Potential
Substitutes' Relationships with Employee
Perceptions of Service Quality (Individual—
Level Analysis with All Employees) . . . .

Revised Comparisons of Leader Behaviors
Versus Customer Behaviors Relationships with
Employee Perceptions of Service Quality. . .

Hierarchical Multiple Regressions of
Employee Service Quality Perceptions on
Various Potential Substitute and Leadership
Combinations . . . . . . . . . . . . . . . .

Comparison of Leadership and Potential
Substitutes' Relationships with Customer
Perceptions of Service Quality (Department—
Level Analysis with Aggregated Dispenser
Data). . . . . . . . . . . . . .

Hierarchical Multiple Regressions of
Customer Perceptions of Service Quality on
Two Potential Substitute and Leadership
Combinations . . . . . . . . . . . . . . .

Comparison of Leadership and Potential
Substitutes' Relationships with Department
Rank (Department—Level Analysis with
Aggregated Dispenser Data). . . . . .

Comparison of Customers' Overall
Relationship with Criteria Versus Leaders'
Overall Relationship with Criteria . . . . .

Hierarchical Multiple Regressions of
Employee Job Satisfaction and Employee
Service Quality Perceptions on All the
Customer Scales and Total Leader Behavior. .

ix

0 O I 97
. . 99
C O O Q C 100

. . . . . . . 105

. . . . . . . 109

l
'fi' Ii

 

.4'

CHAPTER I
INTRODUCTION AND REVIEW OF THE LITERATURE

This study examines the different sources of guidance and
support that influence the attitudes and performance of employees
of service organizations. Based upon the literature describing the
processes by which employees are influenced (e.g., theories of
role—making and, most centrally, leadership) and the emerging
literature on the unique organizational dynamics of service
organizations, this research explores the degree to which customers
influence the employees of a retail, for—profit, eye—care chain by
acting as substitutes for leadership.

The various sources that influence employees in any
organization include individuals in positions of hierarchical
leadership, as well as various non-leader sources (Kerr and Slocum,
1981). Typically, it has been assumed that of the many possible
sources that gguld influence employees, some form of hierarchical
leadership always will be one important source. This assumption,
however, has recently been questioned by Kerr (1977; Kerr and
Jermier, 1978), who argued that various non—leader sources e.g.,
coworkers or task characteristics, may replace or "substitute" for
leadership—-actually making it impossible and unnecessary for
leadership to have an effect. This study uses the substitutes for

leadership conceptualization as a framework for assessing the

2

respective strength of various leader and non—leader sources of
influence. Central attention is paid to customers as a non-leader
source of influence that may act as a substitute for leadership.

That customers might be one of the substitutes for
leadership in service organizations is suggested by the developing
literature identifying differences in the organizational dynamics
of the service sector versus the manufacturing sector (e.g. Mills
and Moberg, 1982; Schneider, 1980; Snyder, Cox, Jesse Jr., 1982).
One difference, for example, is that a higher percentage of service
employees work in boundary roles than do their manufacturing
counterparts (Adams, 1976; Parkington and Schneider, 1979). Thus,
there is more face to face social interaction between customers and
the typical service employee than is true for the average
manufacturing employee. Additionally, customers often play an
important role in the actual creation of the service offered, e.g.,
doctors relying upon patients to accurately describe their ailment,
bank tellers expecting customers to fill out deposit tickets, etc.
(e.g. Eiglier and Langeard, 1977; Fitzsimmons and Sullivan, 1982;
Schneider and Bowen, 1983). Customer involvement is most evident
in quinary sector (Foote and Hatt, 1953) service organizations--
those that attempt to change the service recipient in some way,
e.g. health—care and educational organizations.

This emerging literature has not, however, been uniform in
specifying the consequences of the customers' involvement, i.e.,
whether customer involvement contributes to or detracts from

positive individual and organizational outcomes. Chase (1978;

1981) for example, believes that the less direct contact the
customer has with the service system, the greater the potential of
the system to operate at peak efficiency. Using Thompson's (1967)
framework, Chase argued that the service organization's core

technology should be sealed off from customers. In his view,
service systems with high customer contact are more difficult to
control and more difficult to rationalize than those with low
customer contact. On the other hand, Eiglier and Langeard (1977)
concluded that for a service organization to increase performance
it should obtain more active participation from its clientele in
the production of the service. In sum, recent writers have agreed
that customer involvement influences the organizational dynamics of
service organizations, but have disagreed about the form, strength,
and direction of that influence.

It is, then, from a joint interest in the sources of
influence upon employees and the unique dynamics of the service
sector that a focus on customers as a substitute for leadership in
service organizations emerges. This research assumes there may be
various non—leader sources that substitute for leadership and that
customer involvement in the service creation process may result in
customers acting as a substitute for leadership for the employee.
Indeed, it is hypothesized that in service organizations customers
may be the most influential of the potential substitutes for
leadership, compared to the professional orientation of employees

and the degree of organizational formalization. By testing this

. In; ntfitmfl

l
.Vy-“-«

2,-

 

4
hypothesis, this research clarifies both the importance of
leadership as a source of influence and the dynamics of
organizations within the service sector, now the employer of

seventy percent of the U.S. work force (Fitzsimmons and Sullivan,

1982).

Leader and Non—Leader Sources of

Influence upon Employees

The work on role—making can provide a theoretical starting
point for identifying the different sources of guidance and support
that influence the attitudes and performance of employees. As
described by Graen:

. . . the role making system . . . determines behavior in terms
of the direction and magnitude of energy expended at a given
point in time and the changes in expenditures that occur over
time (1976: p. 1202).
The determinants of a role—making system can be classified, at a
very general level, into three major categories (Graen, 1976): (1)
physical-technological systems—~representing the constraints and
demands that are imposed by the accepted beliefs about those
physical and technological systems, (2) social—cultural systems-—
representing the constraints and demands that are imposed by the
accepted beliefs about the social and cultural systems, and (3)
person systems——representing the constraints and demands that are
imposed by the accepted beliefs about the person systems that
highlight the role for the actor. Theoretically, these three sets

of systems should combine to determine the attitudes and

I v .me ..

i
m

5
performance of a person in a particular role at a given time (cf.
Graen, 1976).
Additional work in role theory specifies further the
sources that influence employees in their organizational role. For

example, Kahn, Wolfe, Quinn, Snoek, and Rosenthal (1964) proposed

an interpersonal role—making model in which role behavior is a
function of occupational identity, intrinsic satisfactions, and
the role pressures exerted on the role incombent, i.e., focal
person, by other persons, i.e., role set, who have a vested
interest in how the organizational role is performed. The
employees' immediate supervisor is presented as the role set member
with the most influence on employees—-particularly during
employees' early tenure with the organization.

Another example drawn from the role theory literature that
sheds light on the different sources that influence employees is
Graen's (1976) interdependent role—systems model. This model,
built largely upon the earlier work by Kahn et a1. (1964) as well
as by Katz and Kahn (1966), focused on how employees' supervisors
could influence the work role behaviors chosen by employees in
their attempt to conform to competing role pressures. More
specifically, Graen detailed how supervisors could influence
employee perceptions of the expected utilities of alternative
patterns of work behavior using an expectancy theory formulation.

In sum, this brief exposition of the role—making process
underscores the wide range of sources that may influence the

attitudes and performance of employees. However, the employees'

 

immediate supervisor is singled out from among the many possible
role determinants as frequently being the most influential, or at
least as being an important arbiter of competing role demands. As
will be clear in the following discussion, the leadership
literature has been even more constant in assuming that leadership
will always be an important source of influence upon employees.
Leadership Theory: Assumptions, Shortcomings, and
Some Recent Approaches
Assumptions. An apparent assumption in all current theories and
models of leadership is that some form of hierarchical leadership
will always be important in influencing subordinate attitudes
or performance (Kerr, 1977; Kerr and Jermier, 1978). As Kerr and
Jermier state it, ". . . Even situational approaches to leadership
share the assumption that while the stylg_of leadership likely to
be effective will vary according to the situation, §gmg leadership
style will always be effective regardless of the situation" (1978:
p. 375). They review how this assumption can be found, in
differing degrees of explicitness, in numerous leadership theories,
including the Vertical Dyad Linkage model (Dansereau, Cashman and
Graen, 1973; Graen, Dansereau, and Minami, 1972;) the Fiedler
Contingency Model (1964, 1967); and most models of decision
decentralization (e.g., Bass and Valenzi, 1974; Heller and Yukl,
1969; Tannenbaum and Schmidt; 1958; Vroom and Yetton, 1973).

The Path-Goal Theory (House, 1971; House and Dessler, 1974;

House and Mitchell, 1974) is the least explicit in assuming that

some form of hierarchical leadership is always important. It
maintains that leader attempts to clarify paths and goals may be
unnecessary and redundant in certain situations. However, even in

these situations, House and Mitchell (1974) note that, "although

such control may increase performance by preventing soldiering or
malingering, it will also result in decreased satisfaction." Thus,
in no situations are leader behaviors hypothesized by Path—Goal
Theory to be irrelevant (Kerr and Jermier, 1978).

In sum, although theories of leadership vary somewhat in
their prescriptions regarding the appropriateness of different
leadership styles in particular situations, all assume that the
effective leader provides some type of guidance and/or positive
feeling for subordinates as they carry out their job tasks (Howell

and Dorfman, 19818).

Shortcomings. Despite the assumption of leadership theory that
leader behaviors are always important, several researchers have
recently noted that leadership variables continue to account for
only a small portion of the criterion variance in most empirical
studies (Howell and Dorfman, 1981a; Kerr and Jermier, 1978; Osborn
and Hunt, 1975). True, strong relationships between the leadership
behaviors of supervisors and subordinate outcomes have sometimes
been reported. However, "conclusions have had to be based on
statistical rather than practical significance, and . . . the
researcher's ability to show that the trivially low correlations

obtained were not the result of chance" (Kerr and Jermier, 1978:

p. 375).

A related problem in leadership theory and research has
been a difficulty in distinguishing leadership effects from other
contextual influences occurring simultaneously that may also
produce effects. For example, Pfeffer (1977) maintained that
leadership has much less of an effect on the variation in
organizational outcomes than do external factors such as economic
conditions. He maintained that individuals continue, nevertheless,
to believe in leadership effects because doing so provides a simple
causal framework and justification for the structure of a social
collectivity. Thus, he noted, the importance of leadership in a
given social context is more an outcome of various social
processes——other than leadership——than leadership effects, per se.

Contextual variables have also been considered as both
covariates and moderators in leadership research. In a covariate
view, both the leader's behavior and the subordinate's outcome
measures are greatly influenced by social forces in the work
groups, by the broader organizational climate, by the work
technology and by the superior—subordinate role expectations in a
given work situation (Franklin, 1975; Osborn and Hunt, 1975;
Pfeffer and Salancik, 1975; Schriesheim, 1980).

Contextual variables, however, have most frequently been
investigated as moderators of the relationship between leader
behaviors and subordinate outcomes. As Osborn and Hunt stated:

Situational variables, such as task structure, do not influence
leader behavior but moderate the relationship between

9
leadership and criteria. This is essentially the dominant
argument implicit in current contingency approaches. . . .

Here, situational variables do not predict a criterion or

leadership; they only alter the impact of leader behavior
(1975: p. 32).

Reviews of the variables that may serve to moderate the
relationships between leader behaviors and subordinate outcomes
suggest a nearly unlimited list of possibilities e.g. task
predictability, subordinate locus of control, the leader's position
power, etc., (e.g., Bass, 1981; House and Baetz, 1979; Kerr,
Schriesheim, Murphy, and Stogdill, 1974).

In sum, research in leadership does not appear to support
the assumption that it will always be an important source in
influencing subordinate attitudes and performance. Other sources
apparently bear more directly on subordinate outcomes than does
leadership or, at a minimum, other sources moderate how leadership

influences subordinate outcomes.

Some Recent Approaches. Recently, a number of works in the
leadership literature have attempted to come to terms with
leadership's failure to be as strongly related to subordinate
outcomes as had been typically assumed. A brief review of three of
these approaches offers examples of the new directions being
pursued in leadership research.

Hunt and Osborn (1982; Hunt, Osborn, and Schuler, 1978;
Osborn and Hunt, 1975) have examined the linkages among macro
factors, leader behavior, and leader effectiveness. Their work,

which was initially presented as a "Reactive-Adaptive Approach" to

10

leadership and later as their "Multiple Influence Model of
Leadership" has demonstrated that environmental and organizational
conditions (e.g., task environment complexity and organization
size) alone, and in combination with leadership, appear important
in affecting both individual performance and satisfaction criteria.
Thus, macro variables may influence employee attitudes and
performance independent of any leadership effects.

Miner (1975) suggested the more extreme alternative of
doing away with the concept of leadership. Instead, he proposed
the concept of control as a replacement. Miner discussed four
types of control. First, hierarchical control refers to the
scaling of authority within the organization; this is where
"leadership" would fit, if anywhere. Second, there is professional
or ideological control, referring to the values, norms, and ethics
of the profession or some outside group. Third, there is the group
control exercised within a group of coworkers. Fourth, there is
task control. Miner maintained that for task control to be
effective there must be present both the "push" of work, e.g., the
pace of the assembly line, as well as the "pull" of work, e.g., as
provided by job enrichment. Miner concluded that all these types
of control are used in organizations, but not with the same
frequency.

T031 (1982) provides a third example of a recent approach
that suggests why leadership's importance may have been overstated.
Tosi proposed that leadership was only one of seven factors that

determine organizational behavior patterns. The seven were: (1)

 

i

‘Is‘
2%

‘ . prairies.

11

Organizational formalization, (2) Technology, (3) Socialization,
(4) Selection processes, (5) Reward systems, (6) Work
relationships, and (7) Leadership——defined as the interpersonal
influence of a higher level organizational official. Tosi
suggested that a consideration of the sources of influence upon
employees should be guided by three considerations: (1) How much
variance in behavior patterns is accounted for by each factor?, By
all combined?, (2) Do these factors have a simple additive or
interactive effect on behavior patterns? and (3) Are there
circumstances (and what are they) in which different patterns of
these factors would result in the similar level of productivity?
The above approaches all detail sources of influence that
may be as important, or more important, than leadership in
affecting employees. Although they do question the assumption that
leadership will always be an important source of influence, they do
not explicitly question the assumption that leadership will always
have ppm; influence upon employees regardless of the situation.
Thus, the contributions of these three approaches can be thought of
as mainly refocusing attention on an issue succinctly stated by
Dubin:
The central analytical problem is this: how to sort out the
influences of supervisory behavior from other influences that
play upon productivity.
Two general possibilities present themselves. (1) On the
assumption that supervision accounts for a fixed and sizeable
proportion of the variation in output, comparisons may be made
between different kinds of supervision to see which has the

higher correlation with productivity. This has been the
strategy followed in most studies of supervision. (2) An

-<
Va‘hi

U

V.

'r
' k
n.

.5;

4

i KHIML s

12

alternative approach is to ask under what circumstances does
supervision make more or less difference than do other factors

 

affecting productivity? There may very well be considerable

variability in the relative importance of supervision from one
organizational setting to another (1965: pp. 54—5).

It is only in the work on "substitutes for leadership"
(Kerr, 1977; Kerr and Jermier, 1978) that one finds a third
possibility——that there may be settings in which leadership is

altogether unimportant and unnecessary.
Substitutes for Leadership

Kerr (1977) described substitutes as characteristics which
replace or "act in place of" a leader's behavior that tend to
negate the leader's ability to affect subordinate attitudes and
performance. Thus, in situations where strong substitutes exist,
he argued that the choice of leadership style becomes largely
irrelevant. In these terms, the substitutes for leadership model
moves beyond simply detailing other variables that may be a source
of influence in addition to leadership or that may moderate
leadership's influence. Instead, the substitutes for leadership
are said to serve as non—leader sources of guidance and support
which can, if strong enough, negate a leader's ability to influence
subordinates' outcomes (Kerr and Slocum, 1981). The unique
contribution of the substitutes concept then, is, the suggestion
that it should not be assumed that leadership will always be a
source of influence and that the analytical task is to discover to
what degree. Rather, situations should be examined based upon the

assumption that leadership may not be necessary, depending on the

TI

.3 19‘7" 11,; . “age-’1? w’ 1.

13

mix of other non—leader sources of influence present for employees.

Kerr and Jermier (1978) proposed a number of individual,
task, and organizational characteristics which they suggested might
act as substitutes for leadership. The focal leader behaviors in
the typology consist of the two leader behavior styles that
dominate the research literature. Table 1 identifies which
individual, task or organizational characteristic will tend to
negate a leader's supportive or instrumental leader behavior-—thus
allowing the characteristic, rather than the leader behavior, to
affect employees' attitudes and performance.

A brief discussion of some of the characteristics in Table
1 may help clarify both its content, and the nature of the
substitutes construct itself. Employees' professional orientation,
for example, is presented as a potential substitute for both the
leader's supportive and instrumental behaviors. It is considered a
potential substitute for leadership because employees with such an
orientation typically cultivate horizontal rather than vertical
relationships, give greater credence to peer review processes than
to hierarchical evaluations and tend to develop important referents
external to the employing organization (Filley, House, and Kerr,
1976). Clearly, such attitudes and behaviors can sharply reduce
the influence of the hierarchical superior (Kerr and Jermier,

1978).

 

14

TABLE 1

POTENTIAL SUBSTITUTES FOR LEADERSHIP

(adapted from Kerr and Jermier, 1978)

 

 

May Substitute for the Leader's:

 

Characteristics Supportive Instrumental
Behaviors Behaviors
Of the Subordinate
1. Ability, experience, X
training, knowledge
2. Need for independence X X
3. Professional orientation X X
4. Indifference toward X X
organizational rewards
0f the Task
5. Unambiguous and routine X
6. Methodologically invariant X
7. Provides its own feedback X
concerning accomplishment
8. Intrinsically satisfying X
0f the Organization
9. Formalization (explicit X
plans, goals, and areas
of responsibility)
10. Inflexibility (rigid, X
unbending rules and
procedures)
11. Highly-specified and active X

advisory and staff functions

15

TABLE 1—-Continued

 

 

Max Substitute for the Leader's:

 

Characteristics Supportive Instrumental
Behaviors Behaviors
Of the Organization

12. Closely—knit, cohesive work X X
groups

13. Organizational rewards not X X
within the leader's control

14. Spatial distance between X X

superior and subordinates

 

Organizational formalization is presented in Table l as an
organizational characteristic that might serve as a substitute for
the leader's instrumental behaviors. It refers to the presence of
explicit plans, goals, and areas of responsibility. Based on the
work of Van de Ven, Delbecq, and Koenig (1976), Kerr and Jermier
(1978) proposed that under conditions of low—to-medium task
uncertainty and low task interdependence the existence of these
written guidelines could serve as substitutes for leader—provided
coordination.

Kerr and Jermier (1978) offered illustrative explanations
of how several of the remaining characteristics in Table 1 might
substitute for the leader's supportive and/or instrumental
behaviors. Additionally, they presented fifty—five questionnaire
items for the measurement of the fourteen potential substitutes
listed in Table 1. Sample items include "My job satisfaction
depends to a considerable extent on people in my occupational
specialty who are not members of my employing organization,"
(a Professional Orientation item); "My job responsibilities are
clearly specified in writing" (an Organizational Formalization
item). Kerr and Jermier (1978) stated the items were written so as
to permit a distinction between effects which are the results of
leadership and those which stem from substitutes for leadership.
In other words, the items were intended to distinguish leader—

provided autonomy, goal and role clarity, from autonomy and clarity

 

 

 

 

 

 

 

 

 

 

17

which stem from other sources.

Previous Research pp_Substitutes for Leadership
Only four studies (Kerr and Jermier, 1978; Howell and

Dorfman, 1981a, 1981b; Sheridan, Vrdenburgh and Abelson, 1981) have

focussed exclusively on the substitutes for leadership construct.
These four studies are each summarized in separate tables (See
Tables 2 — 5). The following discussion will be organized around
issues common to the four separate studies: (1) the dimensions of
leadership examined, (2) the varying conceptual and statistical
definitions of the substitutes construct as well as the different
variables tested as substitutes, (3) the use of attitudes and/or
performance as criteria, (4) the choice of data analysis
strategies, (5) the degree to which the results support the

substitutes construct, and (6) the directions suggested for future

research.

The dimensions pf leadership examined. Three of the four studies
(Howell and Dorfman, 1981a, 1981b; Kerr and Jermier, 1978) measured
instrumental and supportive behaviors of the leader using a set of
scales designed specifically for use in Path-Goal hypothesis
testing (Schriesheim, 1978). The three instrumental leader
behaviors that were measured were: (1) Role Clarification—~1eader
behaviors which clarify what is expected of subordinates in their
work roles; (2) Work Assignment--1eader behaviors which involve the

assignment of subordinates to specific tasks; and (3) Specification

of Procedures—~leader behaviors which specify rules, procedures,

 

18

and methods for subordinates to use or follow in the execution of
their jobs. It has been shown that these varieties of instrumental
leadership are perceived distinctly by subordinates, and that they
differentially relate to different kinds of criteria (Schriesheim
and Bish, 1974; Schriesheim, 1978). Lastly, supportive leader
behavior was assessed by items indicating warmth, friendship,
trust, and concern for the subordinates' personal welfare.

In Sheridan et a1. (1981), seven dimensions of leader
behavior were assessed (See Table 5) based on instruments developed
by Kruse and Stogdill (1973) and Sims (1977). In sum, the studies
of the substitute construct have looked primarily at the two most
commonly studied dimensions of leadership, instrumental and

support.

The different treatments pf Egg substitutes construct. Substitutes
were conceptualized by Kerr and Jermier (1978) as characteristics
which render leadership not only impossible, but also unnecessary,
as a source of influence upon employee attitudes or performance.
Kerr and Jermier stated that substitutes may be correlated with
both predictors and the criterion, but tend to improve the validity
coefficient when included in the predictor set. That is, they will
not only tend to affect which leader behaviors (if any) are
influential, but will also tend to impact upon the criterion
variable.

Substitutes were defined more broadly or liberally in the

Howell and Dorfman (1981a, b) studies:

 

 

19

Leadership substitutes replace or 'act in place of' a specific
leader behavior. They also may act a§_moderator pp suppressor
variables .pz, influencing the relationship between leader
behavior and subordinate attitudes and70r Aperformance.
‘HOwever, by acting in place of a specific leader behavior,
substitutes can play a much more important role than simply as
a moderator variable (1981a: p. 715; emphasis added).

 

 

In the Howell and Dorfman definition, then, a substitute does not
necessarily have to actually "act in place of" a leader behavior.
Instead, it need only influence the relationship between leadership
and the attitudes or performance of employees. Nevertheless, the
definition used by Howell and Dorfman is still very similar to Kerr
and Jermier's. In both cases, substitutes are described as playing
their most important role when they do, indeed, act in place of
leadership.

In Sheridan et al. (1981) an additional twist was added to
the original Kerr and Jermier definition of a substitute.
Specifically, Sheridan et a1. maintained that substitutes may also
affect the frequency with which a leader demonstrates certain

behaviors:

. . . substitute variables are posited to directly affect the
subordinate's attitudes and behavior and indirectly affect them
by influencing the frequency that the leader demonstrates
specific leadership activities. For example, it has been
suggested that the design of highly structured tasks for
subordinates would tend to reduce the leader's task direction
(Barrow, 1976; Lord, 1976) which in turn should have a positive
motivational effect on those subordinates (House, 1971). Thus,
the structure of work tasks can have both direct and indirect
effects on the subordinate's job performance (pp. 3—4).

This definition is like the others in claiming that substitute
variables may directly affect subordinate outcomes, i.e. substitute

variables may take the place of leadership. However, its

20

description of the indirect effects of substitutes is unique for
the following reason: it claims that substitutes may influence the
freguencz that certain leader behaviors are displayed, rather than
their just negating the influence of leader behaviors regardless of
how often they are displayed.

In sum, the conceptual and statistical definitions of the
substitutes construct vary across the four studies. However, the
definitions are consistent in saying that substitutes may directly
affect subordinate outcomes. Indeed, substitutes were generally
said to play their most important role when they take leadership's
place as a source of influence upon employee attitudes and
performance. In addition to that role, substitutes were said to
moderate the influence of leadership (Howell and Dorfman, 1981a;
1981b) and to affect the frequency with which leader behaviors are
demonstrated (Sheridan et al., 1981).

Turning to the substitute variables themselves, three of
the four studies (Howell and Dorfman, 1981a, 1981b; Kerr and
Jermier, 1978) focussed on the characteristics listed in Table 1.
The actual individual, task, or organizational characteristic
assessed in each study are indicated in Tables 2—4. The
substitutes in the Howell and Dorman studies were measured using
modified forms of the scales developed and used by Kerr and Jermier
(1978). In the Sheridan et al. (1981) substitutes were tested in
addition to those identified in Table 1. For example, locus of

control, nursing technology, and unit structure were tested as

21
potential substitutes. The Kerr and Jermier (1978) scales were not
used in Sheridan et al. Instead, they used measures previously
established in other research [e.g. Job Design was measured using
an instrument adapted from Hackman and Oldham (1976)].

In sum, research on variables that may act as substitutes
for leadership has been largely limited to the individual, task,
and organizational characteristics originally identified by Kerr
and Jermier (1978). However, some new variables and measurement

approaches appeared in the recent study by Sheridan et al. (1981).

The use pf employee attitudes and/or performance as criteria.

Employee attitudes were the sole criteria used in the studies by
Kerr and Jermier (1978) and Howell and Dorfman (1981a, 1981b).
Specifically, Kerr and Jermier used organizational commitment and
role ambiguity. Both Howell and Dorfman studies used
organizational commitment and general job satisfaction. Only the
Sheridan et al. study used employee job performance in addition to
the employee attitude criterion, job tension. The result for
employee attitudes (R2 = .327; p < .01) was much stronger than for

employee job performance (R2 = .133; p < .01).

Ihg choice 2§_g§£§ analysis strategies. Although each study shared
a common interest in identifying substitutes for leadership, they
used very different analysis strategies for classifying variables
as a substitute. The different strategies are described in detail
in order to provide a context for viewing the analysis strategy

used in the present research.

V .. 'v
o- n

3“th

. TILE» Ifih‘i

mu»;
.m‘

 

22

Kerr and Jermier analyzed their data by entering all leader
behaviors and potential substitutes (see Table 2) into a regression
equation simultaneously. They then examined the simple
correlations and beta weights of this full model as a basis for
selecting some leader behaviors and substitutes for inclusion in a
more parsimonious predictor set. Then a reduced model was tested
using the selected predictors. The multiple R and beta weights of
the reduced model were then compared to those in the full model.

Howell and Dorfman's data analysis strategy in both their
studies was developed as ". . . a logical strategy to assess the
degree pf_substitutability of a particular individual, task, or
organizational characteristic for a specific leader behavior"
(1981a: p. 718). They noted this assessment is not possible when
all leader behaviors and possible substitutes are simultaneously
included in a single multiple regression model, as was done in Kerr
and Jermier (1978). Instead, Howell and Dorfman proposed an
analytic procedure that is stepwise in nature, involving a series
of increasingly rigorous tests (i.e., differing hierarchical
regression equations) which result in classifying a particular
individual, task, or organizational characteristic as a
nonsubstitute, weak substitute, strong substitute, or a very strong
substitute.

There were four steps in the Howell and Dorfman analysis:
(1) there must be a logical explanation of how a variable could

possibly take the place of a specific leader behavior and therefore

..ll EeEDHWn

amn‘nwm

 

 

 

 

23
make the leader behavior unnecessary; (2) relationships are
examined to see if a specific leader behavior is even important in
terms of its predictive power over a given criterion. Two analytic

strategies are used: (a) if the bivariate correlation between the
leader behavior and the criterion is significant, it is considered
to be a relevant leader behavior; (b) in a second, less stringent,
analysis a subgroup moderator analysis is conducted. Here, the
sample is dichotomized based on a median separation on the
substitute being studied. Both bivariate correlation and multiple
regression analysis is conducted on the resulting subgroups. If
the leader behavior produces either a correlation coefficient or a
regression coefficient which is significant in the "low" substitute
subgroup, then it is considered to be potentially relevant.

The next step (3) is to determine if the potential
substitute is significantly correlated with the criterion. If this
bivariate correlation is not significant, then the variable is not
even a potential substitute (and is therefore classified a
nonsubstitute). If this bivariate correlation is significant, then
a multiple linear regression equation strategy is followed, where
the general form of the equation is:

criteri°n = bPS YPS + bLB YLB + bPSLB YPS YLB + K:
where PS = potential substitute; LB = leader behavior.

The potential substitute is entered first in a stepwise
hierarchical regression procedure and is "forced" to remain in the

predictor set. The leader behavior, on the other hand, is allowed

to drop out if it does not add significantly to the explained

24

criterion variance. Thus, the leadership variable always is
entered after the potential substitute to determine if the
significant substitute effect can make the significant leadership
effect impossible and unnecessary.

The following steps complete the analysis: if bPS is not
significant, than yPS is a nonsubstitute. If bPS is significant,
then yPS is a substitute of some form.

If bLB is significant, then yPS is a weak substitute. If
bLB is not significant, then yPS is either a strong or very strong
substitute.

If bPSLB is significant, then yPS is a strong substitute.

If bPSLB is insignificant, then yPS is a very strong substitute.

In sum, the Howell and Dorfman strategy represented a more
sophisticated method for analyzing the hypothesized relationships
among the leader and non—leader sources that Kerr and Jermier
(1978) presented in Table 1. Specifically, they developed a
strategy to assess the degree of substitutability of a particular
individual, task, or organizational characteristic (or set of
characteristics) for a specific leader behavior.

Finally, the data analysis strategy used by Sheridan et al.
(1981) assessed the direct and indirect effects of substitutes (as
they were defined by them above), as well as the effect of formal
leadership. In their analysis, all of the substitutes and leader

behaviors were entered in the regression equation simultaneously.

s

b a. ‘1- J‘-

- Mmfimhw '- :_‘

25
Actually, two regression equations were tested:

(1) Outcome = f (Substitutes)

(2) Outcome = f (Substitutes + Leadership)

From these two regression equations, the total effects of the
substitute variables were decomposed into their direct and indirect
effects, where indirect effects were estimated from the difference
in path coefficients for the two equations.

Again, indirect effects involved the substitutes affecting
the frequency with which a leader behavior was displayed. In turn,
this was said to affect employee attitudes or performance.

In sum, these three different approaches underscore that
there has been no one analysis strategy followed in the research on
substitutes for leadership. Instead, each researcher used an
analysis strategy customized to a particular conceptual and

statistical definition of the substitutes construct.

Egg degree £p_ppi 2 Egg results support Egg substitutes construct.
Although the limited amount of work on the substitutes construct
makes it difficult to state firm conclusions about what
characteristics most act as substitutes, the following summary
points can be offered: (1) Organizational formalization is an
organizational characteristic that consistently substituted, to
some degree, for instrumental leadership. In Kerr and Jermier
(1978), it was a substitute when role ambiguity was the criterion.
In Howell and Dorfman (1981a; 1981b), organizational formalization

showed varying degrees of substitutability for all three

. .‘"§3’:3:’fdx‘im-.i«

 

 

26

instrumental leader behaviors across two criteria, organizational
commitment and general job satisfaction. In Howell and Dorfman
(1981a), organizational formalization was a very strong substitute
for the work assignment dimension of instrumental leadership. In
Howell and Dorfman (1981b), organizational formalization was a
strong substitute for the specification of procedures dimension of
instrumental leadership. Organizational formalization appears,
then, to be a characteristic that may substitute for aspects of
instrumental leadership in a variety of organizational settings.

Task characteristics provided some degree of a substitute
effect in all four studies. In Kerr and Jermier (1978),
intrinsically satisfying tasks was the most consistent substitute.
Intrinsically satisfying tasks and task routinization were found to
be consistent, but weak substitutes in Howell and Dorfman (1981a,
1981b). In Sheridan et al. (1981) the characteristic termed job
design had both direct and indirect substitute effects when
employee job tension was the criterion. In sum, task
characteristics consistently had a substitute effect but generally
the magnitude of that effect was less than that of organizational
formalization.

The characteristics of the individual that have been tested
have generally not acted as substitutes. In Kerr and Jermier
(1978), none of the characteristics of the individual (ability,
experience, training, and knowledge; professional orientation; need
for independence) included in the full regression equation remained

in the reduced equation of significant substitutes and leader

41..

 

27

behaviors. Similarly, professional orientation and ability,
experience, etc. did not act as substitutes in Howell and Dorfman
(1981a; 1981b). Only in Sheridan et al. (1981) did the
"professionalism" of employees have a substitute effect. However,
in this study professional orientation, per se, was not measured,
but rather professional commitment. Sheridan et al. defined
professional commitment as representing the nurse's behavioral
intention to continue employment in the nursing profession. This
measures something quite different from whether professionalism or
a professional orientation is a characteristic of the employee.

The lack of empirical support for professional orientation
as a substitute is certainly inconsistent with the strong
theoretical case made for its substitute potential by Kerr (1977;
Kerr and Jermier, 1978). Indeed, a professional orientation was
one of the key characteristics that led Kerr (1977) to believe that
there were non—leader sources of guidance and support that could
make formal leadership redundant. Additionally, Howell and Dorfman
(1981a) conceded their suprise that a professional orientation did
not act as a substitute in their study, given the sample consisted
of employees of a community hospital that included nurses, etc.
Howell and Dorfman attempted to explain this result by noting the
oftentimes bureaucratic (mechanistic) nature of hospital
organizations (Georgopoulos, 1975) which may have increased the
relative importance of the leader behaviors and decreased the

importance of potential substitutes. In this vein, Kerr and Slocum

.tggllfifififififiwmr

 

28
(1981), in a recent theoretical paper on controlling employee
performance, hypothesized that substitutes will operate most
strongly in organic organizations. In any event, Howell and
Dorfman urged additional work with samples that include
professional employees in order to clarify the conditions under
which certain substitutes, including a professional orientation,

are most important.

Suggested directions for future research on the substitutes

construct. Kerr and Jermier (1978) recommended that future work on
substitutes for leadership should distinguish between "substitutes"
and "neutralizers". Neutralizers, they claimed, are
characteristics which make it effectively impossible for leadership
to make a difference. In contrast, they defined substitutes as
characteristics which render leadership not only impossible, but
also unnecessary. Neutralizers and substitutes are similar in that
both act to reduce the impact of leader behaviors upon subordinate
attitudes and performance. They differ, however, in that:
Substitutes do, but neutralizers do not, provide a 'person or
thing acting or used in place of' the formal leader's negated
influence. The effect of neutralizers is therefore to create
an 'influence vacuum,‘ from which a variety of dysfunctions may
emerge (Kerr and Jermier, 1978: p. 395).
With respect to their typology of characteristics, they note that
whereas all fourteen characteristics may clearly be termed
neutralizers, it is not clear all fourteen are substitutes.

Kerr and Jermier (1978) also drew a statistical distinction

between neutralizers and substitutes. Neutralizers were said to be

 

29
a type of moderator variable when uncorrelated with both predictors
and the criterion, and act as a suppressor variable when correlated
with predictors but not the criterion. Alternatively, Kerr and
Jermier stated that substitutes may be correlated with both
predictors and the criterion, but tend to improve the validity
coefficient when included in the predictor set.

The need for future researchers to distinguish between
substitutes and neutralizers was also mentioned by Howell and
Dorfman (19813). However, they—~like Kerr and Jermier (1978)——
chose not to try, themselves, to differentiate between the two in
their own analyses. Actually Howell and Dorfman complicated the
task of distinguishing between the two by defining substitutes in
terms that also included the Kerr and Jermier definition of a
neutralizer. That is, the Howell and Dorfman description of the
substitutes role included the possibility that substitutes may also
act as moderator or suppressor variables. This role is the one
Kerr and Jermier said was the basis for labelling a characteristic
a neutralizer.

Sheridan et al. (1981) was the only study of the four that
attempted to distinguish between substitutes and neutralizers in
its data analysis. Sheridan et al. stated that neutralizer
variables tend to moderate the relationship between leader
behaviors and subordinate outcomes, but do not directly influence
the leader's behavior. Additionally, they posited that
neutralizers may directly affect subordinate outcomes. The data

analysis test for this neutralizer's role is shown by adding the

1:2:

.4: A:
M‘

an.
J

3O

neutralizer variable to the regression equation from Sheridan et
al. presented earlier:

Outcome = f(Substitutes + Leadership + Neutralizer +

Neutralizer x Leadership)
This equation, then, tests for the various influence effects
hypothesized by Sheridan et al.: a direct leadership effect,
direct and indirect substitute effects, and direct and moderating
neutralizer effects. They found that administrative climate (See
Table 5) had a neutralizer effect when job performance was the
criterion, but not when job tension was the criterion.

To summarize how neutralizers have been discussed, it can
be said that each of the four studies described neutralizers
differently. Furthermore, three out of four studies did not
attempt to distinguish between neutralizers and substitutes in
analyzing the data. This indicates that there is lack of agreement
about how substitutes and neutralizers are conceptually different,
as well as lack of confidence that any conceptual difference can be
readily identified in the data analysis.

In addition to the possible existence of substitutes and
neutralizers, Howell and Dorfman (1981a; 1981b) suggested that
certain characteristics may act as "supplements" to leadership.
Specifically, they claimed that the leader's behavior and one or
more of the potential substitutes may at times coexist—-"filling in
for one another as the situation dictates" (p. 728). In these

instances, Howell and Dorfman concluded that the individual, task,

31

or organizational characteristics might best be termed
"supplements" to leadership. This conclusion fits with a recent
definition of substitutes offered by Kerr and Slocum (1981) that
described them as sources of information about tasks and the
motivation to perform tasks which can supplement—-or, if strong
enough, negate—-a leader's ability to influence subordinates'
outcomes (emphasis added).

Finally, an additional elaboration of the substitutes
construct suggested by Kerr and Jermier (1978) would be to specify
other leader behaviors and other characteristics which may act as
substitutes for leader behavior. Progress on specifying other
leader behaviors was made in Sheridan et al. (1981). Recall that
their study consisted of seven dimensions of leader behavior,
rather than the customary two-dimensional view of leadership.

Specifying other characteristics that may act as
substitutes could proceed in two directons, as implied in Kerr and
Jermier (1978) and Kerr and Slocum (1981). First, Kerr and Jermier
(1978) emphasized that their list of characteristics of the
individual, task, and organization was only representative—~that
other characteristics could be identified within each of these
three categories. Secondly, Kerr and Slocum (1981) noted that
Kerr's previous work on substitutes was limited to sources of
influence upon employees that were internal to the organization.
They noted that other sources might be identified if one assumed a
more open—systems model or external control of organizations

perspective. In sum, these works suggested that

 

32
additional characteristics could be found either by looking more
closely within the organization or by starting to look outside of

the organization.

A_ Summary IIE!.2£ Leader and Non—Leader Sources p£_Influence pppp
Employees

This section presented an overview of the different leader
and non-leader sources of guidance and support that influence the
attitudes and performance of employees. The role-making literature
presented as an introduction identified numerous sources that
influence employees in a particular role at a given time, principal
among these being supervision——or leadership. Then it was shown
that all conventional theories of leadership have assumed that some
form of hierarchical leadership will always be important in
influencing subordinate attitudes and performance. However,
against a backdrop of weak results linking leadership and
subordinate outcomes, recent works in leadership have begun
detailing the numerous influence sources other than leadership.
These works (e.g., Miner, 1975; Osborn and Hunt, 1982; Tosi, 1982)
question leadership's importance, but do not deny that it will be a
source of influence, to some degree.

The substitutes for leadership concept, on the other hand,
maintains that there may be situations in which formal leadership
is both impossible and unnecessary. Borrowing from Kerr and
Jermier (1978) and Howell and Dorfman (1981a, 1981b) substitutes
for leadership can be defined as non—leader characteristics that

may, when strong enough, replace the influence of leader behavior

33

upon subordinate outcomes; when less strong, non—leader
characteristics coexist with leadership as a source of influence
and can be termed supplements to leadership.

A few studies have explored different characteristics of
the individual, task, and organization that may substitute for
leadership's impact on subordinate outcomes. The results have
indicated that intrinsically satisfying tasks and organizational
formalization may substitute for either the leader's supportive or
instrumental behaviors. A professional orientation, however, does
not appear to be a substitute——despite the strong theoretical case
to the contrary. The results from these few studies have provided
only mixed support for the different substitutes suggested by Kerr
and Jermier (1978). However, the value of the substitutes concept
to the present research lies in its providing an explicit
conceptual and analytical scheme for assessing the different
sources that influence employees, without taking for granted

leadership's influence.

6

 

**mh. u m

**mm.| aoADAUAMAnaHu

mHom
**om.l coaumufiamanom
**mm.u weflsmmfiuam
MHHmuHmcfiuuaH
**om.i :oaumuflnfiuzom
munwflmz

mums m.amcoa couscmm

zufiswwnam oaou mom

 

mmsouw xuo3 uflax zaomoao
wmmum scum oua>c<
suflfiaafixofiweH
coaumNHHmauom

2

Ho. v a*#
mo. v m*

 

**mo. u m "coaumNHcmwuo ecu mo
**NN. :oaumofimaumao wcflzwmfiumm maamowmawuucH
OHom xomnvmmm
**mq. meflswmfiumm Aosofiv eoflnaunaauaom Amman .aflmamaanaomv
mHHmoncfinumH .Hm um oNNwm "xmmh ecu mo mousuoooum
*ofi. Mumnwomm xmmH mafiswfinam mHom mo coaum0flmflomam
munwwms oocmucmmmccfl new vmmz uqoaawfimm< xuoz
mama m.Hmwoa woosuom Awnofiv .Hm :OHumuGOHHo HHGOflmmowowm coauMOHMHHmao oaom
um Hmuuom mweoazoax .wcficfimuu "amuaoasuumcH
unmaufiaaoo unmauflsaoo xuflfifin<
Hm:0Humsflcmwuo mom Hmn0fiuMNHmmmHo Hmswfi>flccH mzu mo o>Huuoma=m
muasmom mfiumufluo commoa mmusuflumnsm muoa>mamm nmwmmq

 

 

soasm Awsofiv mmHzmmw aza meme may

N mqm<H

coaueNflHmauom wuo

<3 How
eusuaumcsm wcouum mue>

mxmeu weNflqflusom
<3 uom eunufiumcsm xmez

aoHumNfiHmauow wuo
mxmeu ueuficfiusom
ommm How meusuflumcsm xeez

masouw xuos ufinx haemOHo
mxmeu veuflcfiusom

 

Aommm .a: .qom nouv
coaueufiaeauow
Hm:0fiumwflcmwuo
Aqua .mem Houv
mmsoum xuos uwcx haewoao
coaumuficmwuo ecu mo

Ampm Hoev
xuaaeawu Emma
Aommm .<3 Home
mcmeu mo coaumuficflusom

 

com new mausuuumasm saws Asoouv Acpm nouv
.Hm ue mmHeB mcflmmmflumm hHHmoflmcfluucH Ammofi .Eflecmeflucomv
E maaouw xuoz uflcx haemoao :ofiuommmflumm "came ecu wo Aommmv menseeooum
wCHzmmHumw com Hmueqeu mo coflum0flmaoemm
suuaoumeuuueu Aqua Houv Aazv Deadwumma Enos
cam now mausnuumaam use: Aesouv auuam .ooa .eueaunaaxa .suuuuaa Anomv eouumuuuunaHo auom
wee ueuuom Aqua .mbm Howv "HmuceasuuecH
usesuflaaoo uceauflasoo :oaueucefiuo Hecofimmemoum
HmGOHumNHammuo mom Hmc0flumwficmmuo “Henwfl>flv:H ecu mo Amwa e>fluuoam=m
muaamem eflueufluo eeumeH meunuflumcsm wuoa>ecem neweec

 

 

smpem Amuwouv zazmmom oza cameos use

m mqm<H

 

36

coaueNfiHeaHom wuo
40m how esufiumcaw xmez

xoecweem xmeH
wcwmwmflumm
mfiamoflmafluucH
mam How meusuflumcam xmez

:owueewmfiumm cow Hmuecew uom

 

muasmem maneufiuo veumeH meusuflumcsm

muoa>ecem hemmed

 

 

easeuueoolum mumaa

37

madouw cue: ufiax haemoao
mxmmu ceNfiafiusom
.oue .eocefluemxe .zufiafic<

ommm
How meusuflumcsm wcouum

a co .m ecoamme oumcoq Ho
H . m m

:ouuenwaeauow who
ommm How eusufiumcsm waouum

mafizwwwumm
maamuwmcwuuuH
mam Mom eusufiumcnm xmeB

Assauv
.He ue mmaez
noauoewwflumm
coh aeuecew

Aommm .nom nous
xeecceem xmeH
Amem nouv
wufizwmflumm adamoflmcfiuucH
"xmmH ecu we

Aommm .qom nouv
uue .eoceflueqxe .muflafic<

Awsou .auecmeflucomv
Aommmv menseeooum
mo :ofiumoawuoeam

 

mHmaonmemoumco: Ambm nowV ecumzeu A<3v usesdwfimm< xuoz
use mHec0flmmemoum cuoc Mom monoav cuflam wuo co eeueaa eocmuuomaH Acomv coaumUHwHumHU eaom
use ueuuom Hmzwfl>fivnH ecu mo "HeuceasuumcH
uneaufiaaoo uneauflaaoo
HecOAuNNflcmwuo uom HmGOAumuflcmmuo meexoaaam Hm:0fimmemoum mom Ambmv e>fiuuomnsm
Aweuoeuumcmv wuasmem eflueuwuo ueumee meusufiumcnm muoa>ecem Heceec

 

 

unpam Aaumouv

z<zmmom Qz< aqmzom mmH

q mqm<B

 

<3 How
eusuflumcsm wqouum e can
com Mom eusuflumcsm cue:
e mm3 :ofiumNfiHeauom muo

Aommm .as .40“ nouv

mace .mamnoammewouncoc mom coaumuflamauow wuo
Aommm .«3 .qom .mam nouv
coauenwamauom who mqsouw cue: uflcx zaemOHo
com How eusufiumcam xmez "coaueuficmwuo ecu we
zfico .wHeGOAmmewoua Mom Aommm .<3 .Aom HowV
mcmmu veNHaflusom
w waflmmmflumm maaeeflmafluuaH "cmeH ecu mo

mbm How eusufiumcsm xmez
Ambm Howv mwueseu

 

mHeGOHmmewoumco: wuo co weeeaa euamuuanH
cam mam:oammewoum cuoc uom "Henefi>flw:H ecu mo
coauoemmfiumm cow Hmuecew nom meemoaaae Hmc0fimmewouaao: Mom
Aweuoeuumcev mufismem maneufluu weumeh meusuflumcsm muoa>ecem ueweec

 

 

wescfluGOOIlq mqm<H

 

 

 

39

**5Nm. n Am

uoewme HeNflHmuuse:
m e>mc uo: vac
eueafiao e>fiumuumficfiac<

Alv :owumwefiem
AIV coaufiqwouem
uoemme
:e cufia muofi>mcec Heweec

Alv awumae new
Alv cowmecoe msouu
Alv uneauflaaoo
Hemoflmmemoum
muoewme uoeufiwcfi uce
ueeuwu cuflz meuauflumcsm

A+V Houuqoo mo msooq
Alv eecefluemxe cow
A+v coaueoswe Hmauom
mace .muoemwe
uueuflc cufl3 meusufiumcsm

scamaeu cow pom

mmam an
wwhflwmma mGOHm

Iceafle e>flm
eocmEuomuem cow

Aaomuv
.Hm ue acem
:owmceu cow

eueEHHo e>flumuumflcflac<
“ueufifieuuse: m m<

acumecoo msouo

cwflmec cow

eusuosuum uwca
sweaoccoeu wcwmusz
Houucoo mo msooc
uceauwsaoo Hmcowmmewoum
eocefluenxe cow
:ofiumosve Hesuom

Assauv asum
sea Amsouv uuuewoom
use emsux scum veumec<

comfich
>ua>flufimcem
camafluaem
coaufiawoeem
:oaumweaem
mmece>fluuemm<
coaueuoaa<

chfimceafie uou>mcec ce>em

 

muasmem

eflueufluo

weumeH meusuflumcsm

muou>mcem ueveec

 

 

wmbhm AH MHOzv zomcmm< Qz< .mom:mzmmm> z<QHmmmm mm&

m mum<e

40

**mmH. n «m

uoewme
ueNHHeuuse: e e>ec vat
eumawao e>fiumuumflnfiaw<

AI\+V mmece>fluuemm<
uoemme
:e cufl3 muofi>ecec neweec

Alv mmoaoccoeu wcflmusz
A+v :oflmecoo macaw
muoewwe uoeufiucfi use

uueufiu cuH3 weuauflumcsm

A+v euceflueaxe cow
maco .muoemwe
uoeufic cuwz meusuflumcam

eocmauomuem cow mom

 

muasmem eflueufluu weumeh meusuflumcsm muofi>ecem seemed

 

 

wescfiuGQOIlm mqmck

41

Organizational Behavior ip_the Service Sector

 

The preceding discussion dealt with sources that could be
expected to influence the employees of all organizations.
Similarly, the characteristics suggested as potential substitutes
by Kerr and Jermier (1978) were purposefully developed at a level
of abstraction to be applicable to all organization (Jermier,
personal communication). The interest in the present research,
however, is service organizations and, more specifically, how
customers can act as substitutes for leadership in them.
Therefore, the organizational dynamics that set the service sector
apart from the manufacturing sector are now examined. These
differences suggest the likelihood that in service organizations

customers may influence employees by acting as a substitute for

leadership.

The Boundaries pf the Service Sector

The range and number of organizations that can be
considered to be service organizations is large enough that it is
useful to briefly mention efforts to categorize them. The service
sector is often referred to as the tertiary sector, based on
Clark's (1957) three-fold division of the economy. Here, the
primary sector was principally agriculture; the secondary,
manufacturing or industrial; and then the tertiary, services.

The services sector was subdivided more finely by Foote and
Hatt (1953). Their categories/stages of economic activity appear

in Table 6. The service sector is said to be composed of three

42
TABLE 6

STAGES OF ECONOMIC ACTIVITY
(adapted from Foote and Hatt, 1953)

 

 

 

Primary (Extractive): Quartenary (Trade and commerce):*
Agriculture Transportation
Mining Retailing
Fishing Communications
Forestry Finance and insurance
Real estate I:
w
Secondary (Goods—producing): Government .3
Manufacturing I',
Processing Quinary (Refining and extending
human capacities):* ,/
Tertiary (Domestic services):* Health
Restaurants and hotels Education
Barber and beauty shops Research
Laundry and dry cleaning Recreation
Maintenance and repair Arts

*Services

43

categories: domestic services; trade and commerce; and refining
and extending human capacities. According to Foote and Hatt (1953)
and Gersuny and Rosengren (1973), tertiary and quartenary services
are intended to maintain the recipient "as is," whereas quinary
services are designed to change and improve the recipient in some
way.

The service organization in this study is a retail eye—care
chain. Consequently, it can be considered as offering both
quartenary and quinary services. That is, it is a for—profit
retail organization (quartenary) that attempts to improve the eye
care of its customers (quinary). As mentioned in the introduction,
it is when quinary services are offered that customer involvement

with employees is most pronounced.

The Unique Organizational Dynamics pf the Service Sector

There has been a tendency in the literature on
organizations to focus nearly exclusively on the nature of
manufacturing organizations and to assume that the principles that
have emerged from studying those organizations fully capture the
dynamics of all organizations (Miller and Rice, 1967; Shamir,
1978). However, analogies between services and manufacturing
(e.g., a hospital is like a job shop) can be superficial and
misleading because they tend to ignore the unique people-processing
and people-changing nature of many services (c.f. Fitzsimmons and
Sullivan, 1982). Similarly, Chase (1981) has emphasized that it is

inappropriate to matter of factly apply operations manangement

44

principles used in manufacturing systems to service organizations.

A number of recent works have helped reverse this tendency
to overgeneralize across organizations by identifying the ways in
which service organizations are unique (e.g., Chase, 1978; Eiglier
and Langeard, 1977; Fitzsimmons and Sullivan,' 1982; Gersuny and
Rosengren, 1973; Schneider, 1980; Schneider and Bowen, 1983; and
Shostack, 1977a, 1977b). Although these works differ in important
ways, they essentially agree upon three fundamental ways in which
service organizations differ from manufacturing organizations.
These three differences underscore that the social interaction
between employees and customers is more complex in service
organizations than is true in manufacturing organizations. That
is, the opportunity for mutual interpersonal influence between
employees and customers is pronounced in service organizations.

(1) Th; Intangibility pf Services. Whereas the output of
manufacturing organizations are tangible goods that exist in
time and space, services consist solely of intangible acts or
processes that exist in time only (Shostack, 1981). Bateson
(1977) maintains that services are "doubly intangible," that
they are characterized by "palpable" intangibility and "mental"
intangibility. The former refers to the fact that services
cannot be touched, the latter refers to the customer's
difficulty in envisioning what has been obtained when receiving
a service, e.g., what does the customer purchase when buying
insurance?

Since services are intangible their true "reality" can only

is a}? -

45

be defined experientially. The lack of a concrete basis upon
which to define services results in service employees often
being evaluated as attributes of the service itself (Lovelock,
1981; Shostack, 1977b). Furthermore, more than one version of
reality may be found in a service market. That is, the reality
of a service is in the eye of the beholder, i.e., the customer
(Shostack, 1977b), or in other words, the reality of a service
is what the customer claims it to be.

(2) Customer Participation in the Production pf, Services.
In service organizations, the customer plays a key role in the
division of labor involved in the creation of many services
(cf. Gersuny and Rosengren, 1973). This stands in marked
contrast to manufacturing organizations, where the customer has
little or no actual involvement in the production process.
However, in many service organizations employee productivity is
in part dependent upon the knowledge, experience, motivation,
behavior, and cooperation of the customer (e.g., Gersuny and
Rosengren, 1973; Lovelock, 1981). For example, the reliability
of a doctor's diagnosis may depend upon the patients' ability
and willingness to describe their illness. As mentioned,
customer participation is most pronounced in service
organizations comprising the quinary sector-—those
organizations that attempt to change or improve the customer.
In these service organizations at least some measure of
cooperation by customers is a prerequisite for the

implementation of decisions about their problems (Danet, 1981).

46
(3) .222. Simultaneous Production and Consumption .2:
Services. In manufacturing organizations there is typically a
delay between the production of a good and its consumption by a
customer. In service organizations, however, the provision and

receipt of a service takes place nearly simultaneously. In

other words, whereas the production, distribution, and sale of
a product can be uncoupled, this is not possible with services
(Thomas, 1978). Again, services are more experienced at a
given point in time than they are exchanged and possessed.
Thus, the typical service employee is more exposed to the
customers outside the organization than their manufacturing
counterparts. Indeed, as noted in the introduction, service
employees often occupy boundary roles and have frequent and
salient face-to-face exchanges with customers (Adams, 1976).
These three characteristics (the intangibility of services,
customer participation in the production of services, and the
simultaneous production and consumption of services) set service
organizations apart from manufacturing organizations. These
characteristics illustrate that the social interaction between
buyer and seller is much more elaborate in the production of
services than with material goods. In other words, the economic
market and division of labor is more clean for manufacturing
organizations that it is for service organizations (c.f. Gersuny
and Rosengren, 1973). In service organizations, customers and
employees interact and "work together" in the production of service

and the definition of its reality.

47

It is not clear, however, how this extensive social
interaction between customer and employee influences individual and
organizational outcomes. Parsons maintained that a view in which
clients play a role in the organization recognizes ". . . the
possibility and/or necessity of client contribution to the
solidarity of the service system" (1970: p. 9; emphasis added). On
the other hand, Gersuny and Rosengren (1973) concluded that
customer participation in the production process is a possible
source of both "new vistas of organic solidarity" app. "a growing
arena of conflict." Conflict between employee and customer was
seen as stemming from disagreement over the terms upon which
cooperation is to take place——not unlike the conflict between
employer and employee, in general (Gersuny and Rosengren, 1973).
In sum, research is needed to specify further the consequences
arising from the interactions that take place between the customers
and employees of service organizations.

Customers as Substitutes for Leadership
ip_Service Organizations

 

The substitutes for leadership concept provides a useful
framework for examining the influence that customers may exercise
within service organizations. That is, customers may serve as a
source of support and guidance for employees by acting in place of
the instrumental and supportive behaviors of those in positions of
hierarchical leadership—-a possibility clearly suggested by the
above description of the interaction between employees and

customers in the service sector.

48

Customers were not suggested as potential substitutes for
leadership in the Kerr and Jermier (1978) typology. That the Kerr
and Jermier (1978) typology was intentionally limited to internal

characteristics of the organization made it unlikely that customers

would be considered as a potential substitute for leadership.
However, when the conceptual domain of the substitutes construct is
extended "outside of" the organization, a host of new potential
substitutes can be considered. One can think of adding to the
three category typology of Kerr and Jermier (1978) a fourth
category of extra—organizational characteristics. Examples could
include suppliers, other organizations, state and federal laws
and—~of course——customers, all of which may act as substitutes for
leadership. These characteristics, although primarily outside the
organization, nevertheless can be expected to influence the
attitudes and performance of employees inside the organization.

The consideration of customers as an external
characteristic, rather than internal, is at least arguable because
it depends on the larger unresolved question in organizational
theory as to how one defines the boundary of the organization
(e.g., Starbuck, 1976). As Starbuck noted, the problem lies in
determining what measures or decision rules are to be used in
deciding who or what belongs in or out of the organization.

There are several decision rules to choose from in the
literature for locating the customer as inside or outside the
organization boundary. This research accepts Aldrich's (1979)

notion that it is an organization's ability to control an

49

individual's organizational entry and exit that determines whether
an individual should be viewed as essentially inside or outside the
organizational boundary. Similarly, Pfeffer and Salancik (1978)
have argued that the organization ends and the environment begins
at that point where the organization's control over activities
diminishes and the control of other organizations or individuals
begins.

Control over individuals' movements (Aldrich; Pfeffer and
Salancik) is a boundary—decision rule well suited to a
consideration of whether customers should be thought of as inside
or outside the service organization. As Shamir (1978) observed,
the customers of service organizations typically participate in the
organization voluntarily and can withdraw their participation at
any moment they choose. This freedom of movement is especially
true in for—profit retail service organizations, i.e., the type of
organization in the present research.

In sum, customers represent a potential substitute for
leadership that does not "fit" in Kerr and Jermier's (1978)
typology of characteristics internal to the organization. Instead,
customers are more appropriately placed in an ggppayorganizational
category of possible substitutes for leadership. Thus, considering
the customers' role in service organizations not only identifies
one new characteristic that may act as a substitute, it focusses
attention on a whole new category of characteristics that may serve

as non-leader sources of support and guidance for employees.

50
How Customers May Act as a_Substitute for Leadership_
There are two possible ways to think about how customers
may act as substitutes for leadership. First, customers may simply
take the place of leaders as an interpersonal source of

instrumental and supportive behaviors. That is, employees may

rely upon the customers' display of instrumental behaviors (e.g.,
telling the employee what needs to be done and how to do it) and
supportive behaviors (e.g., telling employees things that are
warming and self—worth enhancing) moreso than they rely upon the
leaders' behaviors. In other words, customers may substitute for
leaders as the individuals whose instrumental and supportive
"leadership" matters to employees. Second, customers may define
and shape the characteristics of the jobs performed by employees.
Specifically, customers may say or do things that affect employees'
perceptions of enriched task characteristics. In turn, enriched
tasks (e.g., those that are intrinsically satisfying, provide
employees feedback, etc.) have been suggested to be substitutes
for leadership (Kerr and Jermier, 1978). Finally, it is most
likely that customers exercise their influence in both these ways
simultaneously.

That customers may act as a substitute in the first way
(i.e., by simply taking the place of the leader), is a possibility
given the frequent face—to—face interactions employees and
customers have. Furthermore, customers can be expected to attempt
to lead employees in the exchanges they have. As Schneider,

Parkington, and Buxton (1980) have noted, customers can be viewed

 

51
as one of the service organization's multiple constituencies which
try to dictate how the organization is to function effectively
(Pennings and Goodman, 1977). In other words, customers have an
interest in shaping the attitudes and performance of employees in a
way that will result in the employees providing service in
accordance with customers' desires.

Furthermore, employees may value guidance and support from
their customers more than from their superiors. Service employees
in boundary roles are, after all, often as close psychologically,
or even closer, to their customers than to other employees of the
organization (Aldrich and Herker, 1977; Parkington and Schneider,
1979). Also, Schneider et al. (1980) proposed, based on the
vocational choice literature (Holland, 1973), that boundary role
positions in service organizations attract socially-enterprising
types who want to give good service. Indeed, it has been shown
that employees may intend to quit their job if they think customers
view the quality of service they receive as poor (Schneider and
Bowen, 1983). These factors (the psychological bond between the
typical service employee and the customer; the fact that it is
important to give good service) combine to suggest that employees
might rely upon customers as non-leader sources of guidance and
support.

Employees might also rely upon customers as substitutes for
leadership simply because doing so represents the path of least
resistance. As Adams (1976) noted, boundary-role employees'

ability to control and manipulate extra-organizational expectations

52
and pressures is, in many cases, more limited than their ability to
control and manipulate intra-organizational expectations and
pressures. Also, a formal expectation of many service employees is
to act as though the customers were always right (Shamir, 1980).

This can result in employees being particularly sensitive to
guidance and support offered by customers.

Empirical work consistent with a conceptualization of
customers as a substitute for leadership can be found in Ouchi's
(1977) study on the controlling aspects of clients. He found that
the nature of the clientele was a significant predictor of the
control mechanisms used by the organization. Ouchi speculated that
higher income clients will, in general, be more likely than lower—
income clients to impose their will on the salesclerk.
Furthermore:

With the clients thus providing direct control over a large
range of non—output measured goals, the store can safely place
a great deal of emphasis on sales volume without worrying that
employees will ignore other performance areas (Ouchi, 1977: p.
108).
Thus, control is exercised via a combination of formal
organizational controls and clientele influence. Ouchi concluded
by saying this is a view of control that "easily admits the actors
in the environment into consideration" (1977: p. 111).

The second way the customers of service organizations were
suggested to act as non—leader sources of support and guidance was
by enriching the characteristics of jobs performed by service
employees. In turn, enriched tasks may substitute for leadership.

More specifically, the general hypothesis is that customers may

vm'

{Iii-Ir an

, ,atii

53

shape how employees perceive the following five core job dimensions

of the Job Characteristics Model (Hackman and Lawler, 1971; Hackman

and Oldham, 1975):

1. Skill variety. The degree to which a job requires a
variety of activities, involving use of a number of
different individual skills and talents.

2. Task identity. The degree to which a job requires
completion of a whole, identifiable piece of work—-that is,

doing a job from beginning to end, with visible results at
the end.

3. Task significance. The degree to which a job has a
substantial impact on the lives or work of other people,

whether within the organization or in the external
environment.

4. Autonomy. The degree to which a job provides freedom,
independence, and discretion to the individual in

scheduling the work and determining the procedures to be
used in carrying it out.

5. Task feedback. The degree to which carrying out the

activities required by a job results in direct and clear

information about the effectiveness of the employee's
performance.

Tasks that have these five characteristics are considered
to be enriched. Enriched tasks provide sources of both support and
guidance for employees. That is, enriched jobs possess both the
"push of work" and "pull of work" (Miner, 1975) that is necessary
for tasks to be effective in controlling employee attitudes and
performance. It was this logic that led Kerr and Jermier (1978) to
propose that task characteristics such as task—provided feedback
and intrinsically satisfying work might serve as substitutes for

leadership.

The role of customers in shaping how employees perceive

54

these task characteristics was given only limited attention in the
development of the job characteristics model. This likely reflects
that the model was initially developed to guide the redesign of
jobs in which employees dealt mainly with things, e.g.
manufacturing jobs, clerical jobs involving typing and filing etc.,
rather than with persons. Hackman and his associates did, however,
add the dimension "dealing with others" to supplement the five gpgg
job dimensions. The "dealing with others" dimension assessed the
degree to which the job required employees to deal with other
people (either customers, other company employees, or both) in
order to complete the work. This supplemental dimension was
included to see how an opportunity to satisfy social needs was
related to employee satisfaction and motivation.

A central point in the present research is that in service
organizations "dealing with others", i.e. customers, largely
defines the very core of the job. In service organizations,
customer—employee interactions are not just a supplemental job
dimension, they can in large measure shape how employees perceive
the other characteristics of their job. In this vein, Bell (1973)
has described work in a post-industrial, services—oriented society
as a "game between persons." In contrast, he described work in an
industrial society as a "game against fabricated nature," e.g.
machines. Similarly, several authors have noted that whereas the
employees of manufacturing organizations work on inanimate raw

materials, the employees of service organizations work on reactive,

55
animated customers (Perrow,1967; Shamir, 1978; Schneider and Bowen,
1983). Although these observations may not apply to all service
organizations, they are particularly valid when looking at quinary

service organizations that attempt to change the client in some

way——the type of service organization in this research.

The point is that the customer plays a unique role in
shaping the characteristics of jobs in service organizations.
Shamir's descriptions of the customers of hotels suggests how this

might be the case:

Human beings are heterogeneous in their nature, their
demands, their tastes and their expectations. To a greater
extent than inputs to most production organizations, they
present the hotel with many exceptions. . . (1978: p. 287).

And,
In addition to their heterogeneity and unpredictability, their
inflow into the system is much less steady or regular than the
inflow of material to most production organizations. . . .
Some of this variation can be predicted and planned for, but
not always can the variation in usage be predicted by the
organizations, since it is sometimes the coincidental result
of many independent decisions. (1978: p. 287).
Certainly these observations suggest how customers can serve to
define the task characteristics of skill variety and job autonomy,
respectively. Furthermore, the relevance of customers for
determining task significance is implied in the very definition of
that job dimension, i.e., "impact on the lives or work of other
people, whether within the organization or in the external
environment." These points illustrate that the complex social

interactions that frequently take place between customers and

service employees can enrich the jobs these employees perform.

 

56
Indeed, establishing client relationships has been suggested as an
effective strategy for implementing job enrichment (Hackman, 1977;
Slocum and Sims, 1980).

In sum, customers may be an important substitute for
leadership for employees of service organizations. Customers can
be thought of as a non—leader source of influence essentially
outside the organization that influences employees inside the
organization. The manner in which customers may act as a
substitute for leadership can be captured in each of two ways: (1)
Employees may rely upon the customers' display of instrumental and
supportive behaviors moreso than they rely upon leaders as an
interpersonal source of these behaviors, and (2) Customers may
enrich the tasks employees perform which, in turn, can substitute
for leadership. Together, these two perspectives suggest customers
are an important substitute for leadership in service

organizations. Indeed, they may be the most important substitute

of all.

Hypotheses

 

The central interest of this research concerns how strongly
the customers of service organizations influence employee
attitudes and performance compared to other sources. This interest
is examined by testing a number of exploratory hypotheses about
which leader or non—leader source of influence relates most
strongly to employee: attitudes (their job satisfaction and service

quality views) and performance (customer views of the quality of

 

57

service offered by the unit in which employees work and management
rankings of unit performance):
H1. Customer supportive behavior will substitute for leader

supportive behavior's influence upon employee attitudes and

performance.

H2. Customer instrumental behavior will substitute for leader
instrumental behavior's influence upon employee attitudes
and performance.

H3. Customers as task enrichment will substitute for both leader
supportive and instrumental behavior's influence upon employee
attitudes and performance.

H4. Organizational formalization will substitute for leader
instrumental behavior's influence upon employee attitudes and
performance.

H5. Professional orientation will substitute for leader
instrumental behavior's influence upon employee attitudes and
performance.

H6. Customers will be a stronger substitute than either
organizational formalization or professional orientation.

The attitude and performance criteria were chosen for the
following reasons: Employee job satisfaction has been speculated
to be a particularly important attitude in service organizations.
This is because it may be that employees need to be satisified
themselves before being concerned about satisfying customers

(Schneider, 1976). Next, employees' own attitudes about service

58

quality have been described as a useful but overlooked diagnostic
available to service organizations (Fitzsimmons and Sullivan,
1982). The use of management rankings in this study helped remedy
the lack of attention performance criteria have received in other
studies of the substitute construct. Further, several authors,
e.g. Schneider (1980), have emphasized that measures of service
employee performance should include an assessment of the quality of
service offered; thus, the use of customer service quality views as
a measure of employee job performance.

Organizational formalization and a professional
orientation, from the Kerr and Jermier (1978) typology, were
included to provide a benchmark against which to measure the
strength of customers as a newly proposed substitute. More
specifically, organizational formalization was chosen because it
has consistently been shown to substitute for instrumental leader
behaviors (Kerr and Jermier, 1978; Howell and Dorfman, 1981a,
1981b). Furthermore, the employees in the present sample made
frequent references to the rules, budgets, etc. that they must deal
with in performing their jobs (See Methods below). PrOfessional
orientation was chosen because it, too, was alluded to in the
interviews with employees. Although a strong conceptual case has
been made for why professionalism should act as a substitute (Kerr,
1977; Kerr and Jermier, 1978), it has not proven to be a substitute
in most studies (Kerr and Jermier, 1978; Howell and Dorfman,
1981a). This study responds to the suggestion by Howell and

Dorfman (1981a) to search for situations in which professionalism

 

59

may act as a substitute.
Finally, two issues that are not being addressed in this
research deserve explanation. First, the hypotheses do not specify

different relationships for each of the different attitude and

performance criteria. For example, no hypothesis was made about
whether a given substitute effect would be stronger with, say,
customer service quality views as the criterion versus the
management rankings. This was not done because the real interest in
substitutes research is in whether a non—leader source can replace

a leader source's influence upon the criterion, regardless pf_the

 

magnitude of that influence. Furthermore, it was felt that: (1)
for employee and customer service quality views as criteria there
was insufficient prior research to guide specific hypotheses about
how different aspects of leader and non—leader influence would
affect them, and (2) for employee job satisfaction and performance
criteria there is abundant research on how different dimensions of
leadership affect them. However, the research results are
contradictory and qualified enough that they, too, would not be
particularly helpful in framing more specific hypotheses.

Lastly, the present research is not concerned with
assessing neutralizers. This is consistent with the majority of
the earlier work (Howell and Dorfman, 1981a, 1981b; Kerr and
Jermier, 1978) which avoided an empirical test of the neutralizer
construct. As discussed earlier, the one study (Sheridan et al.,

1981) that did measure neutralizers did so in a way that was

60
inconsistent with how the construct was defined by Kerr and Jermier
(1978).

The neutralizer construct is, nonetheless, interesting.
However, it seems impossible, even unrealistic, to separate
neutralizers from substitutes in "real life." First, recall the
definition of substitutes by Howell and Dorfman (1981a) stated
earlier. That definition of a substitute appeared to subsume Kerr
and Jermier's (1978) statistical definition of a neutralizer.
Secondly, it seems impossible to keep apart the leadership vacuum
(created by neutralizers) and the characteristics which rush in to
fill the vacuum (the substitutes). For these reasons, the present
research does not attempt to identify characteristics that might

act as neutralizers.

CHAPTER II

METHODOLOGY

Research Design

The design was a correlational field study. Survey data
were collected from the employees and customers of eye—care
departments. Rankings of departments by senior management were

also obtained. Data were collected at one point in time.

Setting and Participants

Data for this study were collected from the employees,
customers, and management of a national, retail, eye—care
organization. The organization was composed of a large number of
relatively autonomous departments. These departments were grouped
by geographic region, with a Group Manager assigned to each. The
departments were also classified as one of three different
department types. The type of department depended upon the range
of services provided and whether or not there was an eye doctor
available. (Further description of these department types is not
presented in order to preserve the organization's anonymity.)
Department employees consisted of Department Managers, Dispensers
(who performed duties such as fitting customers with glasses

frames, adjustments, etc.), assorted other help, e.g. clerical, and

61

.,,

n; 6.11;}. a

, -v-
a»;

mix

62

Surveys were mailed to employees and customers of these

departments. Of the 972 employee surveys mailed, 525 surveys were
returned for a response rate of 54 percent. Of the 17,866 customer

surveys mailed, 2,680 surveys were returned for a response rate of

15 percent. These response rates are similar to those obtained in
recent survey research studies (Schneider et al., 1980; Schneider
and Bowen, 1983) of the employees and customers of banks, another

for-profit service organization.

Data Collection Procedures

Employee apd Customer Interviews.

Semi—structured interviews were conducted and tape recorded
with samples of small groups of employees and customers. The
interviews were used to help determine the variables and issues
most appropriate for consideration in this study, as well as in the
larger research project of which this study was a part. The larger
project had as its focus the quality of service in the different
departments as perceived by both employees and customers. The
entire research effort was enthusiastically endorsed by the
organization's management who notified employees of the forthcoming
interviews and surveys and encouraged them to participate.

The researchers used the following questions to guide the
employee interviews: (1) Describe your (the employee's) job and
what makes for your good days and bad days on the job, (2) What is

your definition of good customer service? and (3) What things

 

 

 

 

63
interfere with (or promote) your ability to provide the best
service possible? The employees' answers helped reveal what the
employees considered to be good performance in their service
organization, as well as their perceptions of the sources that
influenced their performance and attitudes at work.

The customer interviews involved a discussion of the
following topics: (1) descriptions of their experiences within the
department, (2) explanations of what good service meant to them,
and (3) summary evaluations of the qaulity of service they had
received in the past. The customers' answers indicated how they
viewed the department's service quality overall, as well as what
specific department practices and procedures they considered
related to service quality.

The employee and customer answers, in combination with
issues identified by the researcher in other studies of service
organizations (Schneider and Bowen, 1983), were then used to guide
the development of the employee and customer surveys. Only
portions of these surveys were directly relevant to the present

hypotheses.

Survey Administration.

Surveys were mailed to all employees and to a sample of 50
customers from the departments that had provided the researchers
the necessary customer address labels by the time the surveys had
to be mailed. The surveys were mailed approximately two to three

months after the interviews were completed. Both the employee and

64
customer surveys were accompanied by a return business reply
envelope.
The employee survey was also accompanied by a letter from

the company's president requesting participation. In this letter,

employees were assured that their individual responses would never
be seen by management and that company management would receive no
survey feedback identified by department number. To facilitate
follow-up research, the employee survey had a stick-on name and
address label attached, but employees were told to remove the label
if they desired (approximately 25 did so). The letter made clear
that participation was voluntary. In addition, the mailing to
employees contained a copy of the survey sent to customers. This
was done to relieve any anxiety employees may have had about what
questions were being asked of their customers. It also enabled
employees to respond knowledgeably to questions customers might
ask them about the survey they had received.

Finally, two reminders were sent to employees by the

company president encouraging those who had not yet responded to

return their survey.

Data from Senior Management.

The departments within each group were ranked from "best"
to "worst" by their Group Manager. The Group Managers were told to
use any basis they wanted in making the rankings. The best
department was assigned the number one and the worst department was

assigned the number equalling the number of departments within the

65
group. These rankings were collected at the same time the surveys
were administered.
These rankings provided a tool for at least partially

controlling for factors other than leadership and its substitutes

that might affect an exclusively dollar-based department
performance criterion, e.g. profitability. The interviews with
employees, coupled with common business sense, had suggested a
limitless number of variables that it would be desirable to control
for if department profit was the criterion, e.g., department type,
the relationship with the host store, location, age of department,
etc. It was infeasible to a priori specifiy all relevant
variables, collect data on each, and then subsequently control for
their effects. Instead, it was decided that the rankings could
provide a "natural" standardizing of the performance data, i.e.,
Group Managers would likely weigh the effects of relevant control
variables before ranking the departments.

Subsequent analyses between the rankings and department net
profit revealed r = .53 (p < .001), suggesting that Group Managers
at least considered profit in their rankings but also considered
(i.e., controlled for) other issues as well. When asked what these
other issues were, Group Managers indicated several common
considerations: quality of staff, location, achievement versus

potential, and progress compared to past performance.
w

Customers as a Substitute for Leadership: Acting as an

 

 

66

Interpersonal Source sf Supportive and Instrumental Behaviors.
Two three-item scales were used to assess the degree to
which customers take the place of leaders as an interpersonal

source of instrumental and supportive behaviors. In Table 7, both

the "Customer Supportive Behavior" items and the "Customer
Instrumental Behavior" items are shown. The items in both scales
are identical to the Schriesheim (1978) items describing the
leader's behaviors except that customers substitute for the leader
as the subject/focus of each item. For example, the leader
supportive behavior item, "my immediate supervisor acts rudely
towards me (R)" became "customers act rudely towards me (R)." This
wording appeared to provide the most straightforward approach to
measuring whether the instrumental and supportive behaviors of

customers or leaders were the most strongly related to employee

outcomes 0

The Customer Instrumental Behavior scale, as originally
developed, included all three aspects of instrumental leadership
described by Schriesheim (1978). In other words, it contained one
item for each of the three instrumental behavior dimensions, i.e.,
role clarification, specification of procedures, and work
assignment. There were two reasons for this approach: (1) it was
assumed that employees would perceive "leadership behaviors"
displayed by customers in less differentiated terms than they
viewed the behavior of their supervisor. That is, because

employees see many customers, but only one leader, it appeared

 

67

TABLE 7

CUSTOMERS AS A SUBSTITUTE FOR LEADERSHIP: ACTING AS AN
INTERPERSONAL SOURCE OF INSTRUMENTAL AND
SUPPORTIVE BEHAVIORS

 

 

CUSTOMER SUPPORTIVE BEHAVIOR

1. Patients do things to make my job more pleasant.
2. Patients act rudely towards me (R).

3. Patients treat me without considering my feelings (R).

CUSTOMER INSTRUMENTAL BEHAVIOR

1. Patients explain what is expected of me on my job.
2. Patients tell me how I am to go about doing my job.

3. Patients let me decide what specific things to do for them (R).

(dropped)
1 = Very True 3 = Neither True 4 = False
nor False
2 = True 5 = Very False

68
likely that the three dimensions of instrumental behaviors would
blur with customers as the focus. (2) It was neither possible nor
desirable to use all fifteen of the instrumental leadership items

in customer form. It was not possible because of space constraints

in the survey. It was not desirable because it was felt that to
essentially repeat all the instrumental items would sensitize
respondents to the researcher's interest in contrasting supervisor
and customer behavior. This could have resulted in respondents
trying to answer the items so as to make their supervisor look
relatively "good" or "had," rather than merely answering the
questions as actual descriptions of their supervisor's behavior.
The Customer Instrumental Behavior scale originally had an
alpha coefficient of .43. An examination of the scale's item—total
correlations resulted in dropping the work assignment item,
"Patients let me decide what specific things to do for them" in
order to have a more acceptably reliable measure of customer's
instrumental behavior. The Customer Instrumental Behavior scale
actually used, then, assessed only role clarification and
specification of procedure behaviors and had an alpha coefficient

of .66. For the Customer Supportive Behavior scale, the alpha

coefficient was .62.

Customers ‘as .a Substitute for Leadership: Acting asna Source .2;

Task Enrichment.

This measure consisted of five items written to assess how

customers may influence employees' perceptions of the five core job

 

69

dimensions of the job characteristics model (Hackman and Lawler,
1971; Hackman and Oldham, 1976). For example, the item, "Patients
say or do things that make me feel that what I do on my job is
important," was written to assess how customers may be a source of
perceived task significance. These items are shown in Table 8. A
Customers as Task Enrichment score was created by summing each
employee's responses to the five items. Missing data was handled
by deleting any case in which all five items were not answered.

There were several reasons why the Customers as Task
Enrichment items were simply summed rather than inserted into the
Motivation Potential Score (MPS) algorithm proposed by Hackman and
Oldham (1975). Principally, Roberts and Glick (1981) have
summarized persuasive evidence that indicates a simple additive
combinatorial strategy has the same predictive ability as the

Hackman and Oldham multiplicative model, while at the same time

being more parsimonious. Secondly, there was Schmidt's (1973)
concern with multiplying scales that are at best interval rather
than ratio. This concern was particularly telling in the present
study where the job characteristic dimensions were assessed by
single items. Finally, since this study was not really intended to
be a test of Hackman and Oldham's model, it was not imperative to
use their more complicated combinatorial strategy. The Customers
as Task Enrichment scale had an alpha of .28. Since the five core
job dimensions of the job characteristics model were claimed to be

theoretically independent (Hackman and Lawler, 1971; Hackman and

70
TABLE 8

CUSTOMERS AS A SUBSTITUTE FOR LEADERSHIP: ACTING AS
A SOURCE OF TASK ENRICHMENT

 

 

The requests patients make of me re uire me to use a wide
variety of skills in serving them (ski l variety).

Patients say or do things that make me feel that what I do on
my job is important (task significance).

The work I do with patients normally involves finishing
whatever I start with them (e.g., order their glasses/lenses
and then dispense them later on) (task identity).

After I've done something for a patient, I know I'll find out
from them whether they're satisfied with what I did (feedback).

The nature and flow of the patients' requests (in the
Department or by phone) allows me little freedom in scheduling
my work and in deciding how to carry it out (autonomy). (R)

1

Very True 3 = Neither true 4 = False
nor False

2 True 5

Very False

71

Oldham, 1975), the low alpha of the present scale is consistent

with the model's propositions.

Professional Orientation.

A three item scale adapted from Kerr and Jermier (1978) was
developed (Table 9). Items (1) and (2) of this study's scale were
identical to the original Kerr and Jermier (1978) items. The third
Kerr and Jermier item, "My job satisfaction depends to a
considerable extent on people in my occupational specialty who are
not members of my employing organization," was not used. It was
felt that because the item contained a non leader source .EEH. a
subordinate outcome that it was more a measure of a substitute for
providing employee outcomes than of a substitute for leadership,
per se. It was replaced by the item, "I rely upon what I believe
to be the professional standards of the optical field to guide my
work." A third item was desirable in order to at least maintain
the already short length of the original Kerr and Jermier scale.
It should be noted that the way the items are worded in this scale,
Professional Orientation is reduced to being only a potential
substitute for instrumental leadership, only. That is, all the
items have a guiding, rather than supporting influence.

Unfortunately, this three item Professional Orientation
scale had an alpha of only .38. Examination of item—total
correlations revealed that, at best, the scale's alpha would only
be .44. This was an unacceptable level of reliability, given the

scale was intended to be a uni-dimensional measure composed of

72

internally consistent items. This, in contrast to the Customers as

Task Enrichment scale where the low alpha (.28) confirmed the
intended theoretical independence of the items.

The Professional

Orientation scale was dropped from subsequent analyses.

Organizational Formalization.

A seven item scale adapted from Kerr and Jermier (1978) was
used (Table 9). Six of the seven items were identical to the
original Kerr and Jermier items. The seventh item, "This
organization has a policy in writing for every situation one can
encounter" was added because the employee interviews indicated many
employees viewed the formalization within the organization in terms

of this issue. The alpha for the Organizational Formalization

scale was .72.

Leader Behaviors.

Items from the leadership scales developed by Schriesheim
(1978) were used to create three different leader behavior scales
(see Table 10). One was Leader Supportive Behavior, which
consisted of seven of Schriesheim's original eleven
support items. Four items were dropped due to space constraints in
the survey and the assumption that seven items with an average
intercorrelation of .63 would yield a sufficiently reliable scale
score for analyses. Specifically, given the reliability
coefficient of .95 for Schriesheim's eleven items reported in Kerr
and Jermier (1978), it was assumed (based on the Spearman—Brown

Prophecy Formula) that the average item intercorrelation .was 63.

73

TABLE 9
PROFESSIONAL ORIENTATION AND ORGANIZATIONAL FORMALIZATION

 

 

PROFESSIONAL ORIENTATION (Dropped)

1.

For feedback about how well I am doing my job, I rely on people
in the optical field, whether or not they work for this
organization.

I receive very useful information and guidance from people in
the optical field who do not work for this organization.

I rely on what I believe to be the professional standards of
the optical field to guide my work.

ORGANIZATIONAL FORMALIZATION

1.

2.

My job responsibilities are clearly specified in writing.

This company has a policy in writing for every situation one
can encounter.

Written schedules and work specifications are available to
guide me on my job.

There are contradictions and inconsistencies among the written
goals and objectives of this company(R).

Performance appraisals at this company are based on written
standards.

My duties, authority, and accountability are documented in
organizational policies, procedures, and job descriptions.

Written documents (e.g., budgets, schedules, etc.) are used as
an essential part of my job.

l--'
ll

Very True 3 = Neither True 4
nor False
True 5

False

M
II

Very False

74
TABLE 10

LEADER BEHAVIOR SCALES

 

 

l) LEADER SUPPORTIVE BEHAVIOR

1.

O

O O

\lO‘U‘lb (JON

Maintains a friendly working relationship with one.

Looks out for my personal welfare.

Behaves in a manner which is thoughtful of my personal
needs.

Does things to make my job less pleasant (R).

Treats me without considering my feelings (R).

Shows respect for my personal feelings.

Acts rudely towards me (R).

2) LEADER INSTRUMENTAL BEHAVIOR

Role Clarification

1.

LJ'l-L‘LJJN

Gives vague explanations of what is expected of me on
the job (R).

Gives me unclear goals to reach on my job (R).

Explains the level of performance that is expected of me.
Explains what is expected of me on my job.

Explains the quality of work that is expected of me.

Specification of Procedures

6.
7.

Tells me how I am to go about doing my job.

Permits me to ignore rules and regulations which affect
how I do my job (R).

Lets me develop my own methods of doing my job (R).
Gives me instructions on how to do my job.

Decides how I am to do my job.

3) TOTAL LEADER BEHAVIOR [includes all the above items]

Very True 3 = Neither True 4

nor False
True 5

False

Very False

75
This was the figure used here to estimate the probable alpha of a
scale with seven items. This Leader Supportive Behavior scale had
an alpha coefficient of .90.

The second scale was Leader Instrumental Behavior,
which was originally to have combined all three aspects of
instrumental behaviors described by Schriesheim: role
clarification, work assignment, and specification of procedures.
However, since the work assignment dimension was dropped from the
Customer Instrumental Behavior scale, it was also dropped from the
Leader Instrumental Behavior scale. This was done so that the same
set of instrumental behaviors would be the focus when comparing
relationships between customer and leader instrumental behavior
with employee outcomes. The Leader Instrumental Behavior scale
actually used, then, was composed of only role clarification and
specification of procedures items.

There is both conceptual and empirical justification for
combining these two aspects of instrumental leadership.
Conceptually, as noted earlier, it was felt that customers could
substitute for leader instrumental behavior only in a global sense,
not as a substitute for each specific aspect of leader instrumental
behavior. Thus, one measure of leader instrumental behavior was
required for which the one measure of customer instrumental
behavior might be tested as a substitute. The empirical
justification is provided by the .78 alpha computed for this leader
instrumental behavior scale. This indicated that although

Schriesheim (1978) has treated the dimensions of instrumental

76

leadership distinctly, role clarification and specification of
procedures can, nonetheless, be scaled together with good internal
consistency. (In one analysis, described later, Leader Role

Clarification and Leader Specification of Procedures were used

separately. Their alphas were .82 and .56, respectively.)

The final leader behavior scale was Total Leader Behavior.
This scale consisted of all the items from the Leader Supportive
Behavior and Leader Instrumental Behavior scales. The alpha
coefficient for the Total Leader Behavior scale was .85. The Total
Leader Behavior scale was used in analyses involving customers as
task enrichment, which was hypothesized to substitute for both the

leader's instrumental and supportive behavior.

All respondents were asked to describe their "immediate
supervisor" when describing leader behaviors. Use of this response
set allowed each respondent to describe the relevant person without
having to provide separate foci for employees in different roles.
Note that because Doctors were autonomous members of the
departments they did not respond to these items and are not
included in the analyses.

Job Satisfaction.

Job satisfaction was measured by a five-point faces scale
(Kunin, 1955).

Employee and Customer Perceptions 2: Service Quality.

Employees were asked to describe the overall quality of
service in their department by responding to the following item:

"Indicate with a check mark (V) how you think the customers of your

77

department view the general quality of the service they receive in
your department." Six alternatives were provided: outstanding,
excellent, good, not so good, bad, terrible. This item provided a
measure of employees' attitudes about the quality of service
provided by their department to customers.

Customers also were asked their service quality views by
using a similar item that read: "Check the box E? that best
describes the general quality of the service you receive in your
department." The same six—point scale used in the employee survey
was used for this item. The mean of the customers' service quality
perceptions was computed for each department and used as one
measure of department workgroup performance.

The interrater reliability of department customers' service
quality perceptions was assessed to determine whether the
department mean represented customer consensus on the quality of
service provided by the department. Interrater reliability was
assessed in two ways: (1) A between—groups one-way ANOVA was run on
customer service quality perceptions (F = 1.89, p <. 001, df, 132).
A significant F, of course, suggested that variation among
individuals was associated more with differences between
departments than within departments. That is, it is an indication
of within department agreement. Second, a procedure developed by
James, Wolf, and Demaree (1981) for estimating interrater
reliability in incomplete designs was used (see Note 1 in Appendix

A). This procedure provides direct estimates of within—group

78
interrater reliability. Therefore, it requires no assumptions
about mean differences between departments, as does the ANOVA
model. Applying the James et al. formula yielded a .68 estimate of

interrater reliability for customer perceptions of service quality.

(Estimates can range from 0 to 1.00; the procedure does not provide
significance levels for the values obtained). Both of these
estimates of interrater reliability indicated that the mean of
department customers' service quality perceptions was a reliable

measure of one aspect of department performance.

Department Rank.

 

In addition to the mean of the customers' service quality
perceptions, the Group Managers' department rankings were used as a
measure of department performance. Given there were different
numbers of departments within groups, it was necessary to somehow
make the rankings comparable across groups. This was done by using
a procedure suggested by Guilford (1954). In this procedure (See
Note 2 in Appendix A), a centile value is computed for each ranking
that accounts for the number of units ranked within each group.
These centile values are then converted to Z—values, thus making

the rankings comparable across groups.

Data Analysis

 

As mentioned in the review of studies on substitutes for
leadership, no one data analysis strategy has emerged as most

appropriate for studying the substitutes construct. In this study,

79

the data were analyzed using correlation and regression analyses.
The analysis primarily involved a series of comparisons within
pairs of correlations. Each pair of correlations consisted of a
leadership scale's correlation with one of the criteria versus its
hypothesized substitute's correlation with the same criterion. For
example, the difference between I (leader support-job satisfaction)
and r (customer support—job satisfaction) was tested. These
comparisons were made for each possible pairing of a leadership
dimension and its hypothesized substitute across all four dependent
variables.

The classification strategy presented in Table 11 was then
used to describe to what degree, if any, a given characteristic
acted as a substitute for leadership. The table shows four
different sets of outcomes (A, B, C, D) that can occur when
computing zero—order correlations and testing for the significance
of the difference between them. The outcomes are arranged from A
to D based upon their demonstrating decreasing evidence that a
characteristic acts as a substitute. Thus, the top—line outcome in
A (The potential substitute is significantly correlated with the
criterion, but leadership is not. The difference between
correlations is significant) indicated that a characteristic acts
as a strong substitute for leadership. At the other extreme, the
last—line outcome in D (Leadership is significantly correlated with
the criterion but the potential substitute is not. The difference
between correlations is significant) indicated that a

characteristic clearly does not act as a substitute for leadership.

 

80
TABLE 11

DATA ANALYSIS OUTCOMES FOR DETERMINING IF
A CHARACTERISTIC IS A SUBSTITUTE

 

 

Zero—Order Correlations
S = significant

NS nonsignificant
Potential Significant Strength of Potential
Possible Substitute Leadership Difference Substitute's Effect
Outcome Criterion Criterion (*)a (Descending Order)
A) S > NS * Strong substitute
S > NS Substitute
B) S > S * Could be a substitute
S > S or b
S < S Only is a supplement
S < S *
C) NS > NS * Not a substitute;
NS > NS no leadership effect
NS < NS either
NS < NS *
D) NS < S Not a substitute;
NS < S * Leadership has an
effect

 

aHotelling's (1940) formula for difference between correlated
coefficients of correlation (See Note 3 in Appendix A).

bDetermined by outcome of hierarchical regression analysis.

81

Outcomes in set B are the other results——in addition to A--
that can indicate that a characteristic acts as a substitute for
leadership. Here, both the potential substitute and leadership are
significantly correlated with the criterion. When this occurred,
an additional analysis was run to see if the potential substitute
actually EQHIE. take the place of leadership, thus making it
impossible or unnecessary for leadership to continue to have a
significant effect on the criterion. This possibility was tested
using the hierarchical regression strategy suggested by Howell and
Dorfman (1981a; 1981b). In their procedure, the potential
substitute was entered first in a hierarchical regression equation,
followed by leadership. If the beta weight for leadership was
nonsignificant, then it was concluded that the potential substitute
could substitute for leadership. If, however, the beta weight for
leadership remained significant, then the potential substitute
could not be considered to "take the place of" leadership. In
these situations, Howell and Dorfman said the potential substitute
was more appropriately termed a supplement to leadership. Finally,
to complete the explanation of Table 11, neither type C nor D
outcomes were considered to be substitutes for leadership.

This classification strategy was applied not only to the
customers as substitutes scales, but to organizational
formalization, as well. Additionally, in those situations where
both a customer scale and organizational formalization proved to be
substitutes, the difference between their respective correlations

with a common criterion was examined to identify the stronger

82

substitute.

A final analysis step was performed to test the central,
general hypothesis of the research: that customers may be a

stronger source of influence in service organizations than leaders.

Here, the overall influence of customers was compared to the
overall influence of leaders against each of the four criteria.
This was done by first regressing a criterion upon the three
customer scales (Customer Supportive Behavior, Customer
Instrumental Behavior, and Customers as Task Enrichment), then
regressing the same criterion on three leader behavior scales
(leader supportive behavior, leader specification of procedures,
and leader role clarification) The leader and customer multiple Rs
were compared in all four cases and the classification strategy in
Table 11 was used to describe the overall effect of customer
influence. The significance of the difference between multiple Rs
was tested using Sympson's (1979) procedure (see Note 4 in Appendix
A).

This analysis strategy was followed twice-~first at the
individual level of analysis and, second, at the department level.
In the individual level analysis, the data from all employees
(Department Managers and Dispensers) were in included (N = 500).
The criteria were each individual employees' job satisfaction and
service quality perceptions. The focus of this analysis was how
these criteria were related to the employees' perceptions of their

immediate supervisors' leadership, as well as to their perceptions

83
of the various substitutes.

The second analysis was done at the department level. Data
from Dispensers, only, was used. The Dispensers' perceptions of
their Department Managers' leadership, as well as the possible
substitutes for leadership, were aggregated within departments (N =
166). It was not possible to assess the interrater agreement of
the Dispensers' perceptions given the small size of the departments
(none larger than eight employees) and the requirement of the James
et al. (1981) procedure to have at least ten raters per unit in
order to yield unbiased estimates. The criteria used were the two
measures of department performance, i.e., department rank and the
mean of the customers' perceptions of the quality of service
provided by the departments. The focus of this analysis, then, was
how these department performance criteria were related to the
Department Managers' leadership of the department (as perceived by

the Dispensers) as well as the substitutes for leadership.

CHAPTER III
RESULTS

Intercorrelations Among Variables

Individual-level intercorrelations are reported in Table 12
for the leadership and potential substitute variables and the two
attitude criteria used at this level of analysis: employee job
satisfaction and employee perceptions of service quality. Scale
reliabilities appear in parentheses on the diagonal and are
Cronbach alphas unless noted otherwise. Means and standard
deviations for these variables at the individual level appear in
Appendix B.

Table 13 contains the department—level intercorrelations
for, again, the leadership and potential substitute variables and
the two performance criteria used at this level of analysis:
customer perceptions of service quality and department rank.
Reliability estimates for the criteria are noted on the diagonal.
Means and standard deviations for these variables at the
department—level appear in Appendix C. Note that at both levels of
analysis the correlation between leader instrumental behavior and
leader supportive behavior is positive, but the correlation between
customer instrumental behavior and customer supportive behavior is

negative.

84

85

uceacufiuam
some

mamm.v *AANH.- ***em. ***ou. ***HN. AAHH. ma mumaonmso

uoa>ecem
HmuceasuumaH
Aoc.v ***oN.I ***mH.I wo.l ***wH.I heaoumao

How>ecem
e>fiuuoma=m
AN©.v **mH. **HH. **NH. Heaoumso

u0fl>ecem
Aom.v ***om. ***ms. nausea Hence

uow>ecem
HmcesduumcH
Awm.v ***wm. seemed

uoa>mcem
e>fiuuoma=m
Aoo.v acumen

.o

 

o m q m N H eacmfiue>

 

 

qm>mq A<DQH>HQZH mmH H<
mmqm<Hm<> wzoz< monH<qmmmoumm92H

N H mcmflw

86

.eusmeea
seuflleawaflm .maeufi ucecnemeccfi NHHeuwuoeuoecue ”memecuceuem cu eue mecmae coeccouo .woq ou
oaq scum mewamu eufim eHmaem .euew eueamaooafi mo enamoem "euoz .Hoo. v m*** “Ho. v a** “mo. v m*

 

 

:oauuewmwuem
Aav***om. ***um. ***om. ***0u.- ***Nm. ***sm. *AIGN. ***mm. nos mmsouasm .o
suuumeo
eow>uem mo mcowu
Aav***om. ***om. *OH.- ***om. ***mm. *AAON. ***mm. leagues mmsouaam .m
aoaumwfiamauom
ANN.V ***wm. 00.1 ***ma. ***om. ***Nq. ***qN. HmGOHumNflamwuo .m
a w A o m a m N H auaaunm>

 

 

weaawucoollNH mqm<B

87

mAmN.v Amu.l ***mm. **HN. ***om.

***oq.l **ON.I *wH.I

NH. *mH.

***mo.

HH.

*mH.I

OH.

***mn.

***Nm.

uceacufiucm

xmea
me mueaoumao

uoa>ecem
HeuceaauumcH
Heaoumso

uoa>mcem
e>fluuoam=m
Heaoumao

uow>ecem
ueeeec deuce

uow>ecem
HeceanuumcH
ueveeq

uofl>ecem
e>wuuoaaam
ueueec

.o

 

essences

 

 

qm>mq HZHZHm<mmm mme H<
mmqm<Hm<> uzoz< monH<AmmmoumMHzH

mH mqm<H

88

.wemwemmm uoc
moH Eouw mewaeu eNHm eamamm .mumw eueamaoocfl mo emseoem

c .muflaficmflaeu ueumuueucfl
"euoz .uoo. v a*** “Ho. v a** Mmo. v a*

a "muu on

 

 

Acv mo. mo. NH. **mN.I **HN. NH. NH. Nu. xcem uaesuueaem .0
m:0Humeouem
suuumso

mawo.v *OH. *mu. mo. mo. *ou. *om. *mu. aou>nam naeoumao .w
coHueNHHeauom

***mm. «o. ***mN. ***oq. ***Hq. **qN. HecOHumNficmwuo .n
m w n o m a m N H eHceHue>

 

 

wescfiunoollmfi mam<H

89
Correlates 2: Employee Job Satisfaction:
Individual—Level Analysis
with All Employees

These results concern whether customers, as well as

organizational formalization, can substitute for different leader

behaviors as correlates of employee job satisfaction. Recall the
data analysis strategy involves examining which alternative source
of influence, i.e. customers versus leaders, correlates more
strongly with the criterion——in this case, employee job
satisfaction.

Table 14 summarizes the results from Table 12 with respect
to: (1) the zero—order correlation between leadership and the job
satisfaction criterion, (2) the zero—order correlation between the
hypothesized substitute for leadership and the job satisfaction
criterion, and (3) which of the two correlations was larger and the
significance or nonsignificance of the differences between them.
These three outcomes were then used to describe the effect of a
potential substitute using the classification strategy presented in
Table 11.

All leader and substitute scales correlated significantly
with the employee job satisfaction criterion. For both leaders and
customers, supportive behavior was the strongest correlate. All
significant correlations were positive except for customer
instrumental behavior. It correlated negatively with employee job
satisfaction (r = —.l6, p < .001). Thus, whereas the supportive

behaviors of both leaders and customers related positively to

90

uneaeaaasm e we muo<

uneaeaamsm e we muo<
Aaowuwwue e>fluewea

e we muee ..e.Hv
uneaeammsm e we muo<

uceaeaamam m we muo<

mz

mz

mz

***HN.
coauewfiaeauom

descaueufiummuo

***©N.
uceacONHQM
xmeH me mueaoumso

***©H.I
uoa>mcem
HeuceasuumcH Heaoumno

***Nm.
uoa>ecem
e>fiuuoma=m Heaoumso

A ***©N.
uow>ecem
HeuceasuumcH neceec

A *ssmm.
uoa>ecem

ueueeq Heuoe

A ***©N.
uoa>ecem
HeuceasuumcH neceec

A ***mm.
u0H>ecem
e>fluuoma=m ueveec

HOG. v Q***

 

coeuommmfiuem
cow eexoamam

 

uoemmm
.meusufiumcnm Hefluaeuom
mo aowumfluomen

eoceuemmam mo
eoqeOAMNGme

eusuflumcsm Hefiuceuom

aficmueveeq

efiueufiuo

 

 

Ammmwoqmzm AA< mHHZ mHmwA<z< A>mAIA<DQH>HmzHV
ZOHHU<mmHH<m mow mmwoqmzm mHHB mmHmmonH<Amm
.mmHDHHHmmDm A<HHZMHom Qz< mHmmmmm<mc mo zomHm<mzoo

«H mqm<H

91
employee job satisfaction, for instrumental behavior only leader
behavior had a positive relationship with job satisfaction.

The size of the leadership correlation always exceeded that
of the customer correlation. However, these results should be
viewed in the context of two qualifications. First, in only one
pair was the leadership correlation significantly greater than the
customer correlation. This was where the leader instrumental
behavior positive correlation was significantly greater than the
customer instrumental behavior negative correlation (t = 6.23; p <
.01; two—tailed test).

Second, the fact that the leader scales correlated more
strongly with employee job satisfaction than did the customer
scales certainly in part reflected the higher reliability of the
leader scales. The leader scale alphas were .90 for leader
supportive behavior and .78 for leader instrumental behavior. On
the other hand, the customer scale alphas were .62 for customer
supportive behavior and .66 for customer instrumental behavior.

A "fair" test (in measurement terms) of whether customers
or leaders are a stronger influence on employee job satisfaction
would require that the scales for each have identical
reliabilities. A fair test was produced by using Nunnally's (1978)
formula (see Note 5 in Appendix A) for estimating the correlation
between two variables when the reliability of one of the variables
has been increased by a particular amount. Thus, the pairwise
comparisons were reassessed with the alpha of the Customer Support

scale increased to .90 (the alpha for leader support) and the alpha

92

of the Customer Instrumental scale increased to.78 (the alpha for
leader instrumental). (Note: In this approach, the customer
scales were not fully corrected for attenuation. That is, they

were not corrected to some mythical 5.: 1.00; they were only

corrected to the level of reliability of the leaders scales, to
which they were compared. Finally, it was not appropriate to
correct for internal consistency in the Customers as Task
Enrichment scale, given it was composed of theoretically
independent items).

Table 15 shows the revised pairwise comparisons involving
employee job satisfaction when the leader and customer scales have
equal reliabilities. These results show that if we were able to
measure customer supportive behavior as reliably as we measure
leader supportive behavior, then customer support would correlate
more strongly (but not significantly) with employee job
satisfaction than does leader support. Indeed, in absolute terms
customer supportive behavior would be the strongest correlate of
employee job satisfaction.

Turning to the actual classification of the potential
substitutes' effects, since all the correlations in the pairwise
comparisons were significant, this indicated (based on the
classification strategy in Table 11) that each potential substitute
could either be a substitute or only a supplement. To determine
which of these two roles best described the potential substitute's

effect, the hierarchical regression strategy described in the

93

TABLE 15

REVISED COMPARISONS OF LEADER BEHAVIORS VERSUS
CUSTOMER BEHAVIORS RELATIONSHIPS WITH
EMPLOYEE JOB SATISFACTION

 

 

Potential Substitute

Reliability Reliability

 

 

Criteria Leadership, Unadjusted Adjusted
Employee Job Leader Supportive Customer Supportive
Satisfaction Behavior Behavior

.35*** .32*** .39***

Leader Instrumental Customer Instrumental
Behavior Behavior

.26*** -.l6*** -.17***

 

***p < .001

94
Methods section was used. Recall that this is a strategy adapted
from Howell and Dorfman (1981a; 1981b) in which the potential
substitute is entered first in a hierarchical regression equation,

followed by leadership. If the leadership beta weight is then

nonsignificant, it is concluded that the potential substitute
indeed could act as a substitute for leadership. This is true
because the potential substitute makes the effect of
leadership impossible and unnecessary, i.e., nonsignificant. If,
however, the leadership beta weight is significant, then the effect
of leadership is both possible and necessary. In that case, the
significant effect of the potential substitute only adds to, i.e.,
supplements, the leadership effect.

In the hierarchical regression analyses against the job
satisfaction criterion, the leadership beta weight was always
significant (See Table 16). Consequently, the potential
substitutes can best be described as supplements to leadership.
However, in the case of customer instrumental behavior this
description is somewhat misleading, because although it adds a
significant effect to leadership's, it is a negative addition. The
nature of the contribution of customer instrumental behavior will
be considered further in the Discussion section.

Since the analysis did not reveal any substitutes, it was
not necessary to determine whether customers were a stronger
substitute than organizational formalization. It can be noted,
however, that the effect of organizational formalization as a

supplement is significantly stronger (and in an opposite direction

95

TABLE 16

HIERARCHICAL MULTIPLE REGRESSIONS OF EMPLOYEE JOB
SATISFACTION ON VARIOUS POTENTIAL SUBSTITUTE
AND LEADERSHIP COMBINATIONS

 

 

 

Cumulative
Step Variable Beta R
1 Customer Supportive Behavior .27*** .31
2 Leader Supportive Behavior .31*** .44
1 Customer Instrumental Behavior —.13** .14
2 Leader Instrumental Behavior .24*** .27
1 Customers as Task Enrichment .l8*** .24
2 Total Leader Behavior .28*** .36
1 Organizational Formalization .18*** .25
2 Leader Instrumental Behavior .16** .29

 

**p < .01; ***p < .001

96

from) than the effect of customer instrumental behavior as a

supplement.

Correlates sf Employee Perceptions_p£
Service Quality: Individual—Level
Analysis with All Employees

Table 17 reports the results concerning whether leader
behaviors or the potential substitutes correlated more strongly
with the criterion of employee perceptions of service quality. The
same pattern of results found for employee job satisfaction was
obtained: All leader and substitute scales Icorrelated
significantly with employee perceptions of service quality.
However, the correlations were systematically lower than when job
satisfaction was the criterion. Supportive behavior was again the
highest correlate for both the leadership measures and substitute
measures. Also, the significant correlations were positive with
the exception, again, of customer instrumental behavior (r = —.10).

The results for employee perceptions of service quality
also differed from those for employee job satisfaction in that
there were two comparisons where the leadership correlations did
not exceed the potential substitutes' correlations. First, the
correlation for customer supportive behavior was stronger, but not
significantly, than the correlation for leader supportive behavior
(r = .30; p < .001 and r = .28; p < .001, respectively). Second,
the correlation between organizational formalization and the

service quality criterion equalled the correlation between leader

instrumental behavior and perceptions of service quality (I = .20,

97

Hoe. v a*** “3. v a*

 

 

*ssoN. n ***ON.
coauesfiaeauom uoa>ecem
uceaeamasm e we muo< mz HecOHueNflaewuo HeucesduumcH ueeeec
***ON. A ***wN.
uceacuwucm uoa>ecem
uaeaeaamsm e we muo< m2 xmee me mueaoumso necmec Heuoe
Aaounueea m>uuawee *0u.u A ***om.
e we muoe ..e.fiv uoa>ecem uoa>ecem
uneaeammam m we muu< m HeuaeasuumcH neaoumso HeuceaduumcH ueceec
suuumso
***om. v ***wN. eufi>uem mo
How>ecem uow>ecem mGONuaeouem
uaeaeaasm m we muoc mz e>wuuoqm=m Heaoumno e>fiuuomasm ueueec eemoamam
uoewmm eoqeuemmwn mo eusufiumcsm Hewuaeuom mflcmueeeec efiueufiuo

.meusuwumcnm Hmwuaeuom
mo aowumwuumem

eocmOHmacwfim

 

 

Ammmuonmzm qua muuz mumunaza um>mauq<amu>umsz
sauuaac mousmmm
mo monummommm mmsoumzm muuz mmummzouuaumm
.mmusuuummsm qauezmuom mza mummemnamu co zomumamzoo

NH mcm<e

98
p < .001). Finally, the only pair of correlations in which there
was a significant difference was between leader instrumental
behavior (r = .20, p < .001) versus customer instrumental behavior

(r = -.10, p < .001; t = 4.39, p < .01, two—tailed test).

To further clarify the relative strength of customers
versus leaders as sources of influence, the reliabilities of the
Customer Instrumental and Supportive Behavior scales were again
adjusted to equal the reliabilities of the comparable leader
scales. The revised comparisons using the corrected customer
scales appear in Table 18. After the correction, customer
supportive behavior appears a bit more clearly as the strongest
correlate of employee perceptions of service quality but the
difference between the adjusted correlation for customer supportive
behavior and leader supportive behavior was also not significant.

With respect to classifying the potential substitutes'
effects, the fact that both leadership and its hypothesized
substitute were significantly correlated with the criterion in all
comparisons required that the hierarchical regression strategy be
followed again. In all cases, the beta weight for leadership was
significant (See Table 19). Thus, all the potential substitutes
were described as supplements. Again, the negative addition made
by customer instrumental behavior will be considered in the

Discussion section.

99
TABLE 18

REVISED COMPARISONS OF LEADER BEHAVIORS VERSUS
CUSTOMER BEHAVIORS RELATIONSHIPS WITH
EMPLOYEE PERCEPTIONS OF
SERVICE QUALITY

 

 

Potential Substitute

Reliability Reliability

 

 

 

Criteria Leadership Unadjusted Adjusted
Employee Leader Supportive Customer Supportive
Perceptions of Behavior Behavior
Service Quality .28*** .30*** .36***

Leader Instrumental Customer Instrumental
Behavior Behavior
.20*** —.10* -.11*
***p < . 001

 

Correlates p: Customer Perceptions a: Service
Quality: Department—Level Analysis with
Aggregated Dispenser Data
Table 20 contains results for the relationships between
customer perceptions of service quality and aggregated Dispensers'
perceptions of the Department Managers' leadership and the
substitutes for leadership. All three leader scales correlated
significantly with the customer perceptions of service quality
criterion. However, of the three customer scales, only customers
as task enrichment correlated significantly with the criterion (r =
.15; p < .05); neither customer supportive behavior (r = .08, us)

nor customer instrumental behavior (I = .03, us) were significant

100

TABLE 19

HIERARCHICAL MULTIPLE REGRESSIONS OF EMPLOYEE SERVICE
QUALITY PERCEPTIONS ON VARIOUS POTENTIAL
SUBSTITUTE AND LEADERSHIP COMBINATIONS

 

 

 

Cumulative
Step Variable Beta R
1 Customer Supportive Behavior .26*** .28
2 Leader Supportive Behavior .22*** .36
1 Customer Instrumental Behavior —.10* .11
2 Leader Instrumental Behavior .19*** .22
1 Customers as Task Enrichment .17*** .20
2 Total Leader Behavior .23*** .29
1 Organizational Formalization .15** .21
2 Leader Instrumental Behavior .13* .24

 

*p < .05; **p < .01; ***p < .001

101

no. v m*

 

 

sou. A *oN.
cowueuwamauom uoa>ecem
uceaeammsm m we muo< mz HemONueNflaemuo HeuceaauumcH penned
*mH. A *aa.
unenc0fiucm uoa>mcem
uceaefimmsm m we muo< mz came an mueaoumao ueweeu Hmuoh
uoc meow
euauflumcsm Hefiuceuom mo.| A *oN.
usc uoemwe uoa>mcem uoa>ecem
:e mec mflcmueveec m HeucesduumaH uesoumso HeuceanuumcH uecmec
Annamao
uo: meow euauwumcsm mo. A *mfi. euw>uem mo
Hewuqeuom usc .uoemme uoa>ecem uow>mcem mcowumeouem
:e mmc awcmueveeq mz e>fluuommsm weaoumno e>fluuoaasm ueveec neaoumsu
uoemmm eoceueMMHQ mo eusuflumcsm Heuuceuom aflcmueeeec maneufluu

.meusufiumcam Hewuaeuom
mo coaumwuumen

eOGeOAMHawfim

 

 

NHHA<DO m0H>mmm

A<B<m mmmzmmmHQ nme<ummwu< mHHB mHqu<z< Am>mqlezm29m<mmmv

mo mZOHHmmummm mmZOHmDO mHHB mmHmmonH<Amm
.mMHDHHBmmDm A<HBZMHOA Qz< mHmmmmm<mA mo zomHm<mzou

ON mAm<H

102
correlates. Also, the leadership correlation exceeded (but not
significantly) the customer correlation in all three pairwise
comparisons. Finally, even after correcting for their relative

unreliability, the Customer Supportive Behavior and Customer

Instrumental Behavior scales did not correlate significantly with
customer perceptions of service qualtiy. These adjusted
correlations were r = .10, ns and r = —.04, ns, respectively.

In addition to customers as task enrichment, organizational
formalization was the only other hypothesized substitute to
correlate significantly with customer perceptions of service
quality (I = .16, p < .05). The leader behaviors for which these
two characteristics were hypothesized as substitutes also
correlated significantly with the criterion; namely, total leader
behavior (r = .19, p < .05) and leader instrumental behavior (r =
.20, p < .05), respectively. Therefore, to determine whether
customers as task enrichment and organizational formalization were
substitutes, or only supplements, the hierarchical regression
strategy was again followed. In both cases, the leadership beta
weights were not significant (See Table 21). However, the beta
weights for the two hypothesized substitutes were also not
significant. Consequently, these two potential substitutes were
best described as supplements, not substitutes. (That there were
significant zero—order correlations between the hypothesized
substitutes and the criterion, but the beta weights were
nonsignificant, occurred because the correlations were based on

more cases. That is, given there is pairwise deletion of cases in

 

 

103

TABLE 21

HIERARCHICAL MULTIPLE REGRESSIONS OF CUSTOMER PERCEPTIONS

OF SERVICE QUALITY ON TWO POTENTIAL SUBSTITUTE
AND LEADERSHIP COMBINATIONS

 

 

 

Cumulative
Step Variables Beta R
1 Customers as Task Enrichment .12 .16
2. Total Leader Behavior .15 .22
1. Organizational Formalization .11 .17
2. Leader Instrumental Behavior .13 .20

 

NOTE: Beta weights are not significant

104
the correlation analysis, but listwise deletion in the regresson
analysis, the regression analysis had a smaller number of cases.
Since the correlations themselves were barely significant, the loss
of additional cases in the regression analysis yielded
nonsignificant beta weights).
Correlates pf_Department Rank:
Department—Level Analysis with
Aggregated Dispenser Data

Table 22 contains results for the relationships between
department rank and aggregated Dispensers' perceptions of the
Department Managers' leadership and the substitutes for leadership.
None of the three leadership scales correlated significantly with
the department rank criterion. However, two of the three customer
scales were significant correlates: Customer Supportive Behavior
(r = .21, p < .01) and Customer Instrumental Behavior (r = -.23, p
< .01). Consequently, based on the classification strategy in
Table 11, customer supportive behavior and customer instrumental
behavior were said to act as substitutes for leaders' supportive
and instrumental behavior when department rank was the criterion.
Since the customer instrumental behavior correlation was
significantly greater than the leader instrumental behavior
correlation (t = 2.48; p < .05; two—tailed test), customer
instrumental behavior was classified as a strong substitute.
(Again, the implications of the negative sign for the customer
instrumental behavior correlation will be considered in the

Discussion Section). The difference between customer supportive

105

uoemme

 

E. was EEALSSH 8.: A 2.
Hon eusuflumcsm :owumwflfimauom uoa>ecem
Hmfluneuoa ecu uecuflez mz HenceueNNcmeO HeuaeasuumaH neveec
uoemwe
:e mmc nwcmueveec NH. u NH.
you eusufiumcsm uneacONucm uofi>ecem
HmHuceuom ecu necuwez mz came we mueaoumao weaned HeuOH
uaeaeomameu e>Huewec
waouum e we muoe *smN.I v NH.
..e.w .eusufiucnm uoa>ecem uoa>ecem
wqouum e we muu< m HeuceasuumaH Hesoumau deuceasuumcH uecmec
**HN. v NH.
uow>ecem uoa>ecem mwcwxcem
esuwumcam e we muo< mz e>fluuomasm Heaoumsu e>fluuoamsm ueweec uceauuemem
uoemmm eoqeuemmwn mo eusuflumcam Hefiuceuom nflcmueweec ewueufluo

.meusuwumcam Hewuceuom

mo coaumwuomem

eoneuamacwwm

 

 

A<H<Q mmmzmmmHm QMH<Ummww< mHHz mHmWA<z< qm>mqlhzm25m<mmmv
mz<m Hzmsz<mmm mHHB mmHmmonH<Amm
.mMHDHHHmmDm A<HHzmecm Qz< mHmmmmm<mq mo zomHm<mzcu

NN mqm<H

106
behavior's correlation with the criterion and the leader supportive

behavior correlation was not significant. Lastly, neither

customers as task enrichment nor organizational formalization

correlated significantly with the criterion (r = 12, ns, and r = —

.03, ns, respectively).

Summary Comparisons pf_Customer versus
Leader Influence upon Employee
Attitudes and Performance

The last data analysis step outlined in the Methods
involved comparing overall customer influence versus overall leader
influence for each of the four criteria. This provided a test of
the central, general hypothesis of the research: that customers
may be a stronger source of influence in service organizations than
leaders.

In this analysis, each criterion was regressed on two
equations and the multiple Rs of the two equations were then
compared. The one equation included all three customer scales
(Customer Supportive Behavior, Customer Instrumental Behavior, and
Customers as Task Enrichment) and, thus, provided a measure of
customer influence overall. The second equation included the three
aspects of leadership subsumed by the leader supportive behavior
and leader instrumental behavior scales (support, role
clarification, and specification of procedures). This equation,
then, provided a measure of leader influence, overall.

The results of these overall comparisons are reported in

107

Hooo. v EAAA .uo. v AAA "mo. v a.

.meusceooum mo GOHueOHHHoeam
Heweeq can cowueOHHHueHo eHom uecmec .u0H>mcem e>wuuomasm neceec no :OHmmeuweu eHaEHmc

.uceacuHunm xmeH me
mueaoumsu cam .u0H>mcem HmumesduumcH Heaoumso .u0H>ecem e>Huuomaum ueEoumsu co :OHmmeuweu quEHmu

 

 

eusuflumcnm
e we muo< NH. A *mN. xcem uceauuemea
uuemmm me
mmc u0H>ecem
HeceeH
no: Heaoumsu HuHHeso eOH>uem
uecuwez wH. v NH. Ho mGOHuaeouem nesoumso
uqeaeHmmsm zuHHeDO eOH>uem
e we muo< ***NN. A ***om. mo quHuaeouem eeHOHmam
uceaeHmaam
e we muo< ***wm. v ***qm. :OHuoemmHuem cow
uoemmm Heaoumso m eHaHuHaz .mueveec m eHQHuHsz .mueaoumso eHueuHuo
HHmue>o

mo :OHumHuomen

 

 

<Hmm5Hmu mHHB mHmmZOHH<HmM HH<mm>o .mmmm4mq mammm>
<HmMHHmo mHHz mHmmZOHH<Hmm HH<mm>o .mmmzoembo mo zomHm<mzoo

MN mqm<e

108
Table 23. For two of the four criteria (employee perceptions of
service quality and department rank) the multiple R for the
customer equation exceeded the multiple R for the leader equation.
The difference between the multiple R for leaders versus the
multiple R for customers was insignificant in all four comparisons.

The overall effect of customers as a source of influence
upon each criterion was classified according to the scheme
presented in Table 11. For employee job satisfaction and employee
perceptions of service quality, both the customer and leadership
equations had significant multiple Rs. Consequently, the
hierarchical regression strategy was followed to determine if
customers, overall, could act as a substitute for the total
influence of leadership, or only supplement it. For both criteria,
then, the three customer scales were entered first in a regression
equation, followed by the Total Leader Behavior scale. For both
criteria, the beta weight for total leader behavior was significant
(See Table 24). Thus, the overall effect of customers on employee
job satisfaction and employee perceptions of service quality is
best described as a supplement to leadership.

For customer perceptions of service quality, neither the
customer scales, overall, nor the leaders, overall, produced
significant multiple Rs. Again, that there would be significant
zero-order correlations but nonsignificant multiple Rs reflects two
factors: the fewer cases available in the regression analysis and
the marginal significance of the zero-order correlations.

Finally, for department rank, the multiple R for customers

109

TABLE 24

HIERARCHICAL MULTIPLE REGRESSIONS OF EMPLOYEE JOB
SATISFACTION AND EMPLOYEE SERVICE QUALITY

PERCEPTIONS ON ALL THE CUSTOMER AND

LEADER BEHAVIOR SCALES

 

 

 

Criterion Step Variables Beta Cumulative
Employee 1 Customer Supportive Behavior .25*** .34
Job
Satisfaction Customer Instrumental Behavior —.02 .34
Customers as Task Enrichment .09 .37
2 Total Leader Behavior .32*** .48
Employee 1 Customer Supportive Behavior .23*** .28
Service
Quality Customer Instrumental Behavior -.01 .28
Perceptions
Customer as Task Enrichment .05 .30
2 Total Leader Behavior .24*** .38

 

***P < .001

110
was significant, whereas the multiple R for leaders was not. (The
difference between the two multiple Rs was not significant.) Thus,
customers, overall, do act as a substitute for leadership when

department rank is the criterion.

Summary

In this section, the intercorrelations among variables were
reported, as were results of correlational and regression analyses
testing the hypotheses presented in the Introduction. Mixed
support was found for the hypotheses. Generally, the results
indicated that customers do seem to influence employee attitudes
and performance, but they do not, for three of the four criteria,

actually take the place of formal leaders' influence.

CHAPTER 4
DISCUSSION

The literature on organizations, from economics to
organizational behavior, has paid little attention to the role of
the customer in organizational behavior. Typically, the customer
has been viewed as a relatively passive recipient of the outcomes
of organizational behavior-~filling the role of an independent,
fleeting partner to certain market transactions with the firm.
Certainly, some work suggests a more assertive role for customers;
for example, viewing them as one of the organization's
constituencies (Pennings and Goodman, 1977) whose interests must be
respected. However, such a perspective fails to illuminate how
customers influence the attitudes and performance of individual
employeese—particularly those in lower—level boundary—role
positions. When the influence of customers upon these employees is
considered, it is often depicted as a problem for the organization.
In this vein, Aldrich and Herker (1977) warn of the negative
consequences of employees overidentifying with customers. In sum,
the literature has either ignored customers, described them as
passive, or cautioned about the negative effects they may have upon
employees when they do exercise influence.

In contrast to the above, this research examined whether

customers might fill a more active and positive organizational role

111

112

by acting as "substitutes for leadership." The unique
characteristics of service organizations, particularly, suggested
that their customers might fill this role. In service
organizations, customers frequently work with the boundary-role
employees in the creation of the service. Furthermore, the
simultaneous production and consumption of services that typifies
many service organizations is the basis of salient and frequent
complex social interactions between employees and customers.
Together, these characteristics suggested a more active role for
customers in the organizational behavior of the service sector than
in the manufacturing sector.

Customers were considered to act as leaders in two possible
ways. One, customers could simply take the place of leaders as the
interpersonal source of instrumental and supportive behaviors that
matters most to employees. Two, customers could shape the
employees' perceptions of the presence of enriched task
characteristics. In turn, enriched task characteristics have been
hypothesized to be substitutes for leadership (Kerr and Jermier,
1978). Together, these possibilities suggested the general
hypothesis that customers may substitute directly for the
instrumental and supportive behaviors of formal leaders and

indirectly by affecting the nature of tasks performed by employees.
In this role, customers could be a strong source of influence upon
the attitudes and performance of service organization employees.

Specifically, the present research tested a number of hypotheses

concerning whether formal leaders, customers, or organizational

113

formalization were the stronger source of influence upon boundary—
role employees in a retail, eye-care chain.

The results revealed that employees' perceptions of the
instrumental and supportive behaviors of both formal leaders and
customers were significantly related to their attitudes, i.e.,
their job satisfaction and own service quality views. More
precisely, perceptions of both customer supportive behavior and
leader supportive behavior were positively related to employee
attitudes, with the reports of customer behaviors being the
stronger correlate for both attitudes (after correction for
unreliability in the customer scales). For instrumental
leadership, perceived formal leader behaviors were positively
related to employee attitudes, but perceived customer behaviors
were negatively related. Lastly, employee perceptions that
customers are a source of enriched task characteristics were
positively related to both employee attitudes.

The pattern of results involving customer perceptions of
service quality and department rank differed depending upon which
of the two criteria was considered. When customer perceptions of
service quality was the criterion, employees' perceptions of formal
leader behaviors correlated more consistently with it than did the
employees' perceptions of customer behaviors. Indeed, Customers as-
Task Enrichment was the only customer scale to correlate
significantly with the customer perceptions of service quality and

it acted as a supplement to total leader behavior.

114
When department rank was the criterion, employees'
perceptions of customer behaviors correlated more consistently with
it than did the employees' perceptions of leader behaviors.
Specifically, customer supportive behavior and customer

instrumental behavior correlated significantly with" department

rank; but none of the formal leaders' behaviors did. Thus, these
two customer behaviors were classified as substitutes for
leadership. Finally, it should be noted that customer instrumental
behavior was negatively correlated with department rank, as it was
with the two employee attitude criteria.

In the comparison of overall customer influence versus
overall leader influence (Table 23), the multiple R for customers
exceeded (but not significantly) the multiple R for leaders for two
of the four criteria: employee perceptions of service quality and
department rank. For department rank, the multiple R for customers
was significant and the multiple R for leaders was not. Thus, in
this case customers, overall, were classified as a substitute for
leadership.

These results can be summarized as follows: (1) When the
measures of customers as a source of influence were significantly
related to the criteria, they were more frequently classified as
supplements than as substitutes for leadership. The one notable
exception was customer supportive behavior and customer
instrumental behavior acting as substitutes when department rank
was the criterion. (2) In general, the customer scales were more

consistently related to the individual—level attitude criteria,

115

i.e., employee job satisfaction and their service quality views,
than it was to the department-level performance criteria, i.e.,
customer perceptions of service quality and department rank. At
the department—level two of the three customer scales correlated
significantly with the department rank; one with customer
perceptions of service quality. At the individual—level all three
customer scales correlated significantly with both employee
attitudes. (3) Unlike the formal leaders' influence, not all of
the customers' influence was positive. Specifically, customer
instrumental behavior was negatively related to three 0f the four
criteria.

Turning to the remaining tested substitute, organizational
formalization, the results indicated that the conclusions drawn
from the customer data were applicable to it, as well. That is,
organizational formalization acted more as a supplement than as a
substitute and was more consistently related to employee attitudes
than to the department performance criteria.

The results can be discussed further by considering (1)
what the customer—focused results suggest for designing and
managing service organizations in which boundary—role employees are
in contact with customers, and (2) the implications of the results
for theory and research in leadership and substitutes for
leadership. Part of the discussion here will consider the role
played by organizational formalization and professional orientation

in the present research.

116

The Implications pf Customer Influence for Designing
and Managing Service Organizations

The results involving the nature of customer influence can

help inform emerging debate on managing customer participation in

the service sector. This is evident in the following analyses of
suggested service system designs.

One approach to managing customer involvement in the
service sector is found in Chase's (1978) work in high versus low
customer contact service operations. Chase's view was that the
less direct contact the customer has with the service system, the
greater the potential of the system to operate at peak efficiency.
He defined customer contact as the physical presence of the
customer in the system. Following Thompson's (1967) logic, Chase
maintained that system efficiency would be enhanced if the
"technical core" could be sealed off from environmental forces,
i.e., customers. He advocated a system design that would minimize
customer-contact in all, or at least some, of a service
organization's operations.

Chase's theoretical work builds on a view of customers as a
source of primarily negative outcomes for the organization.
However, as the present results for customers as task enrichment
and customer supportive behavior show, customer influence can be
positively related to both employee job satisfaction and
performance. Therefore, to seal employees off from customers and

the supportive behaviors and task enriching qualities they offer

117
may be to remove an important and beneficial source of influence
upon them.

Chase's argument to buffer the organization from the
customer does, however, make sense when the focus is the
instrumental behaviors of the customers, i.e., customers telling
employees how to go about doing their job. Again, the instrumental
dimension of customer behavior was negatively related to employee
job satisfaction, employee perceptions of service quality, and
department rank. Apparently, then, although it may not be
desirable to totally seal off employees from customers, it may be
desirable to at least screen them from certain customer behaviors,
i.e., instrumental. Perhaps the organization could undertake
interventions that would minimize the likelihood that customers
would tell employees how to go about doing their jobs. The
organization could, for example, better manage the impression given
by employees that they do, indeed, know how to do their job——
without the customer telling them what to do. This could possibly
be done by having employees wear professional—looking smocks,
displaying their Opticianry degrees, printing signs emphasizing
that employees of this organization are skilled professionals with
excellent training, etc.

Another approach to designing the service system is what
Levitt (1972; 1976) has termed the "industrialization of service."
This refers to transferring modes of production developed by
manufacturing operations to service operations. More specifically,

Levitt argued for the use of machines, standardized procedures, and

118

narrow functional specialization in the design of service
operations. He maintained this is the key to achieving the same
low—cost reliable abundance in the service sector that has been

realized in the manufacturing sector.

The present results indicate that following Levitt's
proposals might also undermine the positive relationships found
between customers enriching job characteristics and employee job
satisfaction, employee perceptions of service quality, and customer
perceptions of service quality. For example, service employees who
perform only a narrow segment of the total service provided may be
less likely to interact with customers who comment on their task
significance, make requests that require a wide variety of skills
to serve them, etc. To "industrialize service" may reduce cost,
but it may also destroy a naturally occurring situation in which
customers act as a source of enriched task characteristics that, in
turn, are associated with important outcomes such as employee job
satisfaction.

In sum, the above two approaches can be viewed as attempts
to design away the naturally occurring and varied exchanges that
may take place between employees and customers in the service
system. The objective in both is to minimize costs/increase the
efficiency of the service system. That particular outcome was not
measured in the present research, so it is not possible to say how
it would be related to the present customer measures. However, it

can be said that if the system designs suggested by Chase (1978)

119
and by Levitt (1972; 1976) make it less possible for customers to
display supportive behaviors toward employees or to enrich the
tasks they perform, these may be systems in which both employee
attitudes and performance suffer. In other words, they may be
system designs that do not, in Parson's words, recognize ". . . the

possibility and/or necessity of client contribution to the
solidarity of the service system" (1970: p. 9).

The Issue pf_Selecting and Socializing Customers

£p_ Fill Prescribed Organizational Roles.

The fact that customers do appear to influence employee
attitudes and performance raises the question of whether some
"types" of customers will make more of a contribution than others.
More precisely, are there certain characteristics that distinguish
"contributing" customers from "non-contributing" customers? If so,
the organization could then decide upon what sort of role it wanted
customers to play in the service system and then select and/or
socialize them accordingly. For example, one could say that the
present organization should be interested in selecting and/or
socializing customers (as substitutes) who display a 1 — 9
leadership style (Blake and Mouton, 1964)—-low task—orientation,
high person orientation.

One guide to customer selection is the study of Ouchi
(1977) in which the income level of the customers of retail
department stores was found to be a significant predictor of the
control methods used by the organization. Ouchi speculated that

higher income clients will, in general, be more likely than low—

120

income clients to impose their will on the salesclerk. Perhaps,
then, lower—income clientele would be less likely than higher
income clientele to try to exercise instrumental leadership over

the employees of the present organization.

A recent study by Langeard, Bateson, Lovelock and Eiglier
(1981) offers another suggestion as to how to identify customer
"types" who are willing to perform on the organization's terms.
They attempted to identify segments of the market population
according to its willingness, or non—willingness, to participate in
the service production process. Customers were asked to complete
surveys in which they were confronted with a number of "service

scenarios."

These scenarios represented service usage at a bank, a
gasoline service station, a hotel, a restaurant, and an outlet
selling traveler's checks; in each instance, customers were
presented a choice between less participative (full—service) and
more participative (self—service) alternatives. For example, the
bank scenario included the following question:

It is 10 A.M. and you wish to withdraw $50 from your checking

account. You have a credit card which would enable you to use

an automatic teller machine or you could go to a human teller

with your checkbook. So your choices are:

Either use the automatic teller machine; p£_ use the human
teller.

There are equally short lines of people waiting to use the
machine and at the teller window.

Similarly worded questions were offered for the other scenarios.

Their results showed that customers can be segmented into

121

groups that have varying propensity to participate in the
production of a service. Those individuals who were generally
participative in all scenarios were somewhat more likely to be
younger, male, and more educated, to be impatient and to dislike
waiting in line, and to like to play with machines more than those
,in the non—participative segment (Langeard et al., 1981). The
researchers urged continued research on finding people who have a
preference for a high level of participation and then identifying
their demographic or psychological characteristics.

The market segmentation approach of Langeard et al. (1981)
could be used in the present case to identify customers who are not
interested in exercising instrumental leadership over employees.
For example, the following type of question could be included as
part of a more far—reaching customer survey:

You are in the department to have your eyes checked and to get
glasses and/or contacts if need be. In dealing with the

employees of the department, would you be more comfortable:

Having the person working with you make most of the decisions
about what your needs are and how to meet them

or

Making most of the decisions, yourself, and informing the
employees of what you think needs to be done and how to go
about doing it.

Customer responses to this question could be pas basis upon which
the organization selected the customers they were more interested
in retaining as repeat business.

Of course, many for—profit service organizations will not

be able to afford selecting customers in so discriminating a

122
manner. It may be more feasible for the organization to select
nearly all customers attracted to it and to then attempt to
socialize them to become the type of customer desired. The
organization might use behavior modelling (Goldstein and Sorcher,

1974) to demonstrate what kinds of customer behavior, e.g.,

supportive, leads to customer rewards, e.g. prompt, courteous
attention, and also what kinds of customer behavior, e.g.
instrumental, leads to negative outcomes for the customer, e.g.
being ignored. For a model, the organization could use either an
employee, posing as a customer, or it could hire a "trained"
customer to display the desired behaviors. This socializing
effort, in combination with the earlier suggestion to better manage
the impression that employees know what to do without being told,
may help restrain customers from displaying instrumental behavior.
In sum, different approaches to selecting and socializing
customers can be attempted to better manage the customers'
contribution to the solidarity of the service system. In this
context, managers can be viewed as simultaneously managing employee
aaa customer behavior. Thus, the organization can make strategic
choices between selecting and socializing (training) employees
and/or customers as a means to individual and organizational
effectiveness. For example, the organization can choose between
socializing customers not to engage in instrumental behavior and/or

training employees to ignore or cope with customers when they do.

123

The Implications for Theory and Research 12 Leadership
and Substitutes for Leadership

 

A number of implications for research in leadership and on

the substitutes construct are suggested by the present results.

Explaining, Epa_ Negative Correlations Involving Customer
Instrumental Behavior.

That the significant correlations between customer
instrumental behavior and employee attitudes were negative, while
the comparable correlations for leader instrumental behavior were
positive, cannot be described within the present conceptual bounds
of the substitutes construct. Customer instrumental behavior did
not make it effectively impossible for leadership to have an
effect--the definition of a neutralizer offered by Kerr and
Jermier. Neither did it make it impossible 22H. unnecessary for
leadership to have an effect——the Kerr and Jermier definition of a
substitute. Finally, neither does it seem to be properly labeled
as a "supplement" to leadership——a characteristic that fills in for
leadership as the situation dictates (Howell and Dorfman, 19818);
the supplement definition implies a positive contribution.
Instead, customer instrumental behavior had a significant negative
effect on employee attitudes without preventing a positive
leadership effect. In brief, both leader and customer instrumental
behavior effectively "led" the employee, but in opposite
directions.

The negative correlations between customer instrumental

124
behavior and employee attitudes emphasizes that there are non-
leader characteristics which affect employee outcomes, but that
still allow leadership to have an effect and, indeed, make it all
the more necessary that it does. In these terms, leader

instrumental behavior must provide a countervailing effect to that

produced by customer instrumental behavior if employees are to be
satisfied. Thus, total influence on employees is not some
combination of the positive contributions of leadership,
substitutes, and supplements as might be assumed in reading works
on substitutes for leadership. Instead, one must also take into
account the negative countervailing effects of characteristics such
as customer instrumental behavior.

The present results also call into question what can be
termed the "influence is a zero—sum game" assumption of the
substitutes construct. That is, the substitute construct seems to
assume that employees will rely upon one or another source of
influence, but not both. The question here is why not both? Why
must one "take the place of" the other? The fact that more
supplements than substitutes were identified in both this research
and by Howell and Dorfman (19813; 1981b) suggests that influence is
not a zero-sum game; that employees do not necessarily choose
between sources, but rather may be influenced by both.

The negative role played by customer instrumental behavior,
however, suggests a finer distinction be made regarding this
assumption. Perhaps supportive influence is not a zero-sum game,

but instrumental influence is. Perhaps employees cannot get enough

125

support, and thus it is more likley that research will yield more
supplements (positive) to supportive leadership than substitutes
for it. That is, there may be no limit to the amount of
supportive behavior subordinates find rewarding or to the number of
sources that can provide it. On the other hand, perhaps employees
consider instrumental behavior a zero—sum game, and consequently,
it is more likely that substitutes, neutralizers, and
countervailing effects will be found for leader instrumental
behavior-—with the nonleader sources of instrumental leadership
being negatively correlated with the employee outcomes. That is,
there is likely a fixed amount of direction subordinates can
experience without also experiencing overload or conflict. Thus,
as employees attempt to keep their degree of instrumental influence
within fixed limits, different possible sources may take one
another's place, cancel one another out, or balance the positive
and negative effects of one another.

In the case of department rank, these conclusions need to
be considered against a pattern of results different from those for
employee attitudes. Here, both customer behaviors were
significantly related to the criterion, but the leader behaviors
were not. These findings might suggest that employees made clear
zero-sum choices between sources for both instrumental apg_
supportive leadership. However, note that the signs of the
customer instrumental behavior and leader instrumental correlations

were Opposite, and the difference between the positive correlations

 

126
for both customer supportive behavior and leader supportive
behavior was not significant. Thus, these results might also
suggest that instrumental influence is a zero-sum game and
supportive influence is not.

Finally, one can look within any pas source of influence-—
rather than between two different sources-to see if for that
particular source exerting influence is a zero—sum game. The
present results indicated that formal leaders could simultaneously
display both supportive and instrumental behavior (r = .38, p <
.001); thus, for leaders influence is ‘ppp_ a zero-sum game.
Customers, however, apparently can only exercise leadership as a
zero—sum game. For them, supportive and instrumental behavior was
negatively correlated (r = -.29, p < .001), i.e., the more
customers display of one dimension, the less they can display of

the other.

Leadership, Substitutes, and Supplements: Alternative Methods pf
Influencing_Employee Attitudes and Performance

 

Rather than looking at substitutes as sources that negate a
leader's influence, it may be more appropriate to view them as
alternative methods of providing employees guidance and support
that leaders can deliberately choose to use in place of their
interpersonal influence. In this view, substitutes for leadership

would make leadership unnecessary, but not "impossible and

 

unnecessary" (or, if impossible, by choice). When substitutes are

thought of in this way, they represent additional, rather than

fewer, degrees of freedom available to formal leaders in choosing

127
how to influence employee attitudes and performance.

The significant correlation between leader instrumental
behavior and organizational formalization (r — .42; p. < .001) is
evidence, possibly, of this revised view of the substitutes
construct. One possible explanation of this relatinship is that
leaders in this organization initiate task structure by developing
written guidelines, schedules, etc. to guide the employees' work.
If this explanation is valid, then even if organizational
formalization had been significantly correlated with a criterion
and instrumental leader behavior had not, it would be inappropriate
to think of organizational formalization as making instrumental
leadership "impossible and unnecessary." Instead, leaders are more
more properly viewed as choosing to relinquish interpersonal
instrumental leadership, developing organizational formalization as
a supplement, or even substitute, for it. In this sense, the
effective leader is one who personally guides the employee's work
or develops non-leader characteristics that do so. Or, put in
perhaps a better way, the good superior/manager is one who guides

the employee's work through personal leadership and/or substitutes

for leadership.

Leadership and substitutes for leadership present
management a mix of alternative methods to choose from in
attempting to influence employee attitudes and performance. In the
present organization, this suggests management should consider the
following mixes: what combination of leader instrumental behavior

and organizational formalization is most effective, and what

 

 

 

128

combination of leader supportive behavior, customer supportive
behavior, and customers as task enrichment is most effective. To
the degree the organization chooses to rely upon customers as

substitutes or supplements for leadership, effective management in

this organization will require building in the desired amount of
customer contact and selecting and/or socializing the desired
"type" of customers.

A scenario that would capture this mix of influence
considerations would be when management felt it wanted to respond
to the negative correlation between customer instrumental behavior
and department rank. Management could consider the following
interventions: organizational and task redesign to seal employees
off from customer instrumental behavior, socializing customers to
limit their instrumental behavior, and/or training leaders to
exercise their instrumental leadership more effectively so as to
provide a contervailing effect to customer instrumental behavior.
Again, effective management consists of choosing between leadership
and possible substitutes of leadership.

The mix of leadership and substitutes for leadership that
management chooses to use as sources of influence will be effective
only if the employees, themselves, agree to make similar choices
among the possible sources of influence to which they will respond.
That is, the employees, as well as management, have degrees of
freedom to play with in deciding what mix of leadership and

substitutes for leadership they will use as sources of support and

129

guidance. Some possible sources of influence, e.g. customer
instrumental leadership, may not be viewed as acceptable or
satisfying to employees. It is in attempting to produce a match
between the choices of management and employees that the role of
"neutralizers" (factors that make leadership impossible) becomes
particularly important. For example, an employee's indifference
toward organizational rewards (see Table 1) is a characteristic
that Kerr and Jermier (1978) said likely acts as a neutralizer. In
the present context, it can be viewed as an employee characteristic
that may make it impossible—-or certainly hinder—-management's
attempts to exercise influence through either leadership or
substitutes for leadership.

Hay gas an Organization Decide Upon the Best Mix pi
Leadership and Possible Substitutes for Leadership? Since in the
present study both leadership and non—leader characteristics were
related to employee attitudes and performance, it would be helpful
to the organization if it had some guidelines to follow in deciding
to what degree one should be emphasized relative to the other.
More generally, how can an organization systematically consider the
sources of influence on which it should rely?

One consideration that should guide the organization's
choices involves essentially an effectiveness versus efficiency
question. A source of influence is effective if it is positively
related to employee attitudes and/or performance. A source of
influence is efficient to the degree it realizes a positive outcome

with a minimal amount of input, e.g. if customers only had to

130

display a small amount of supportive behavior for employees to be
satisfied. The principal focus of this study and the earlier work
by Howell and Dorfman (1981a, 1981b) was in finding what leadership

dimensions and non—leader characteristics were significantly

related to certain employee outcomes; i.e., the effectiveness of
different sources. This focus should be broadened to also include
an analysis of which sources, leader or non-leader, produce their
effects most efficiently. This would help guide managers in
choosing the most parsimonious combination of sources for
influencing employees.

There are a number of ways the organization can approach
the efficiency question. One possibility would be to individually
look at what could be termed the "influence ratio" of each source.
The influence ratio (IR) can be thought of as essentially an
input/output ratio consisting of the influence source mean divided
by the employee outcome mean (assuming sources are measured by
scales with an identical number of response points, thus making
their means comparable). In this study, for example, where both
leader and customer supportive behavior were effective in
influencing job satisfaction, the IR of each source could be
compared to give some indication as to which source was more
efficient. Specifically, the IR for leader supportive behavior
would be 1.00 (2.08 / 2.09; from Appendix B) and the IR for
customer supportive behavior would be 1.12 (2.33 / 2.09; from

Appendix B). The lower value for customer supportive behavior

 

131
suggests that it is a relatively more efficient source of influence
than leader supportive behavior.

Rather than simply examining the efficiency of one source
versus another, one could also empirically compare the efficiency
of one set of sources versus various subsets of those sources.
That is, the organization may want to examine a series of "full"
versus "reduced" regression equations to determine which
combination of sources is the most efficient in producing a desired
significant effect. This data analysis strategy was used by Kerr
and Jermier (1978) in their test of the substitutes construct.

Examining the relative efficiency of different sources
would also require that the costs associated with the "use" of one
source or the other also be considered. For example, is it less
costly to train a leader in instrumental behaviors or to develop
written guidelines, etc.? Is it less costly to train leaders in
supportive behaviors or to design the organization in a way that
employees can be in contact with the supportive behaviors and
enriching job characteristic qualities of customers?

Finally, the ultimate key to understanding both the
relative effectiveness and efficiency of different sources lies in
achieving a better understanding of why employees choose among
different sources the way they do. The research focus, then,
should be less on developing criteria by which we, as researchers,
label a source as leadership, substitute, supplement, etc. and more
on trying to identify the criteria employees, themselves, use, as

the basis for their different responses to a variety of influence

132

sources. Ideally, models could be developed that could
simultaneously explain why any given source, leader or non-leader,
is likely to have a leadership, substitute, neutralizer,

supplement, or countervailing effect. If the effects of different

sources could be assessed within a common model, this could help
managers determine the relative effectiveness of alternative
methods and what mix of methods would be most successful.

Model—building of this kind can start with conventional
models of leadership effectiveness. These models, naturally,
attempt to explain how leaders can be a source of influence.
However, the point here is that one can have some other source,
e.g. customers, substitute for the leader in the model, but still
use the model's propositions to explain what influence, if any, the
substituted source might exercise. For example, path-goal theory
(e.g. House and Mitchell, 1974) can be used to offer a post—hoe
explanation of the differing effects of leader behavior versus
customer behavior upon employee job satisfaction. Recall that the
supportive behaviors of both leaders and customers were positively
related to employee job satisfaction whereas leader instrumental
behavior was positively related but customer instrumental behavior
was negatively related (See Table 14).

Turning to path-goal theory, itself, two of its principles
are particularly relevant to an analysis of these results. First,
the theory proposes that leader behavior is acceptable and

satisfying to subordinates to the extent that they see it as either

133

an immediate source of satisfaction or as instrumental to future
satisfaction (House and Baetz, 1980). Second, leader
supportiveness can be, in and of itself, immediately rewarding
(Evans, 1970). It seems reasonable that both leaders as well as
customers can be the source of support that the second principle
states employees find immediately satisfying. Thus, the positive
relationships for both leader and customer supportive behavior.

The different relationships for leader instrumental
behavior versus customer instrumental behavior can be explained in
terms of the first principle. Leader instrumental behavior may, if
allowed by employees, lead to organizational rewards, e.g. pay, and
thus be a source of extrinsic satisfaction. Customer instrumental
behavior, however, is less likely to be perceived as potentially
leading to this type of extrinsic satisfaction nor to intrinsic
satisfaction. Therefore, customer instrumental behavior is likely
to be neither acceptable nor satisfying to employees. In sum,
path—goal theory suggests that whereas leaders may be perceived as
a source of both intrinsic and extrinsic satisfaction (with
positive contributions made by both their supportive and
instrumental behaviors), customers may be perceived as a source of
only intrinsic satisfaction (with a positive contribution made by
only their supportive behavior).

Fiedler's Contingency Theory (1967) is another leadership
model that can be used to explain both leader and customer effects.
The dimensions of situation favorableness can be examined to see if

the situation makes the exercise of influence easier for the formal

134

leader or the customer. For example, the customer's ability to
influence employee job satisfaction was likely enhanced by
favorable "leader (customer)—member" relations reported by both

department employees and customers in the interviews. These good

relations likely reflect the earlier—stated assumptions that

employees entering face-to—face service roles are psychologically
close to their customers. However, since customers lack "position

"
power,

or what Fiedler (1971) terms "fate control" over employees,
the situation, overall, favors the influence of formal leaders over
customers. This is evident, again, in the differing outcomes for
leader instrumental versus customer instrumental behavior. Since
only formal leaders, not customers, are in a position to dispense
performance contingent organizational rewards, employees may find
satisfaction in doing as they are told by leaders, but not by
customers.

In sum, path—goal theory and Contingency Theory demonstrate
how conventional models of leadership can help explain the effects
of both leadership and possible substitutes for leadership. In the
present case, they were particularly useful for suggesting that
only the instrumental behavior of the leader will be acceptable and
satisfying to employees because only leaders control organizational
rewards. Interestingly, Kerr and Jermier (1978) proposed the
"organizational rewards not within the leader's control" will tend
to neutralize .pggp, supportive and instrumental leadership (see

Table 1). The results, however, indicate that when customers are

135
the source, supportive behavior remains a positive influence—-even
though customers do not control organizational rewards. The path—
goal theory principle that support can be an immediate source of
satisfaction apparently overrode Kerr and Jermier's belief that
lack of control over organizational rewards will neutralize both
dimensions of leadership. The situation for customers probably
holds for leaders, as well. That is, a formal leader's supportive
behavior would asp tend to be neutralized by lack of leader control
over organizational rewards, provided good leader-member relations
exist (as was true with customers!).

Finally, this section of the discussion emphasized the
value of using common frameworks to assess the relative effects of
leader and non—leader sources of influence. It demonstrated that
an answer to House and Baetz's (1979) call for a better
understanding of how factors operate as substitutes for, or
neutralizers of, leadership is to rely upon conventional models of
leadership, themselves, to explain the effects of a variety of
leader and non—leader sources.

Some Suggestions for the Choice pf Methods
and Variables 32 Future Research as
Substitutes for Leadership

Future research on the substitutes construct should attempt
to assess leadership, the potential substitutes, and the criteria
using different methods for each. This would overcome the two ways
in which common method bias affected the present results. First,

common method bias likely increased how strongly leadership and the

 

 

 

 

 

 

 

 

 

 

136
substitutes were related to employee attitudes, relative to the
strength of their relationship with the department performance
criteria. For the correlations involving employee attitudes, the

measures of leadership, substitutes for leadership, and attitudes

all came from a single source, the self—reports of employees.
Thus, the consistency effect typically found in such percept—
percept measurements likely played some role in producing the
consistently significant relationships found for employee
attitudes. On the other hand, common method bias obviously could
not inflate the relationships with the department criteria. Here,
measures external to the employee were used——customer perceptions
of service quality drawn from the customer data and department rank
provided by the management of the company. When these performance
measures were the criteria, there were fewer significant
correlations and the correlations, overall, were lower.

The second way the percept—percept measures may have
affected the results involves the correlations between the
leadership and potential substitute scales, themselves. That is,
not only does percept—percept bias affect how leadership and/or
potential substitutes will be related to different criteria, it
also increases the likelihood that leadership and the substitutes
will be related to one another. Consequently, this consistency
effect increases the probability that when perceived leader
behaviors are present, perceived substitute influence will be
present, as well. In other words, percept—percept bias works

against finding that a given characteristic actually "takes the

137

place of," i.e., substitutes, for leadership. This helps explain
why in this study and the earlier studies by Howell and Dorfman
(1981a; 1981b) supplements were more readily found than
substitutes.

Another important direction for future research is to
follow Kerr and Jermier's (1978) recommendation to identify other
non-leader sources and other dimensions of leadership that
influence employees. The present research identified a possible
new substitute, customers, but relied upon the two dimensions of
leadership most frequently studied, i.e., supportive and
instrumental behaviors. These two leader behaviors were the
central focus of the observation by researchers that leadership
variables continue to account for only a small portion of the
criterion variance in most empirical studies (Howell and Dorfman,
1981a; Kerr and Jermier, 1978; Osborn and Hunt, 1975). The lack of
strong leader—criteria relationships is particularly evident when
instrumental behavior is the leadership variable and/or performance
measures are the criterion.

In a sense, then, the relatively modest results of this
empirical study——the relatively low magnitude of the correlations—-
very likely reflect the theoretical beginnings of the substitutes
construct. That is, the interest in substitutes was spurred by
weak results for the supportive and instrumental behaviors of
leaders and the present study, in retrospect, could be described as

trying to find substitutes for two behaviors that do not seem to

138

matter a great deal to begin with. Consequently, it may be that to
whatever degree customers do act as substitutes for supportive and
instrumental leadership, they may be expected to only weakly relate

to employee outcomes.

Clearly, other dimensions of leadership, and substitutes
for them, must be identified that more strongly relate to criteria
of interest. The limited research on substitutes has focused on
whether instrumental and supportive leadership or their substitutes
bear the stronger relationship to criteria. This focus on which is
the stronger, i.e., do non-leader characteristics neutralize,
substitute, or supplement leadership, should not obscure the more
important need to focus on finding the strongest correlates of
employee outcomes.

One possible starting point for identifying other
dimensions of leadership is Yukl's (1981) identification of
nineteen categories of leader behavior, in addition to the
supportive and instrumental dimensions. Some of them, e.g. goal
setting and work facilitation, may relate more strongly to
performance than do the supportive and instrumental dimensions
typically studied. Certainly, the way leaders set goals has been
shown to be consistently related to employee performance (Locke,
Shaw, Saari, and Latham, 1981). Also, the leader facilitating the
work of employees, e.g. making certain employees have the resources
they need to reach their work goals, has been found to be
positively related to how customers view the quality of service

they receive (Schneider and Bowen, 1983).

139

If leader behaviors such as goal—setting and work
facilitation are the dimensions of leadership that really matter
(i.e., that relate to performance), then they deserve the same
extensive research effort that has been focused on instrumental and
supportive leadership. More comprehensively, substitutes for these
significant dimensions of leadership need to be identified, as
well. For example, in the present context customers could be a
source of work facilitation by providing employees the necessary
information about the condition of their eyes, their tastes in
frames, etc. that employees need to do their job. If management
could find ways of encouraging customers to share this information,
then they would not have to concentrate so much, themselves, on
trying to describe to employees the different needs and desires of
different market segments of customers.

Finally, more work is needed to identify characteristics of
the individual that may be sources of influence in addition to the
task, organizational, and extra—organizational characteristics
already identified as influential. That is, the preceding studies
(Sheridan et al., 1981; Howell and Dorfman, 1981, 1981b; Kerr and
Jermier, 1978) and the present research have established as sources
of influence: task characteristics, including the degree to which
they are intrinsically satisfying and/or routinized; organizational
characteristics such as organizational formalization; and extra-
organizational characteristics, including customers. However, the

preceding studies and the present research have not found either

140
"professional orientation" or "ability, experience, training,
knowledge" (characteristics of the individual in Table 1) to be a
substitute or even supplement for leadership.

Future work on individual characteristics might: (1)

Develop a more internally consistent measure of professional
orientation. Its reliability of .44 in this study and that of .51
in the Howell and Dorfman (19813) study makes it clear the
construct requires further specification. (2) Consider assessing
professional orientation more as a neutralizer than as a
substitute. Again neutralizers are a type of moderator variable
when uncorrelated with both predictors and the criterion, and act
as a suppressor variable when correlated with predictors, but not
the criterion. In both this study, and in Kerr and Jermier (1978),
professional orientation was more strongly significantly correlated
with the other predictors than it was with the criteria.
Empirically, then, this suggests a neutralizer role for
professional orientation——although further analysis and
specification of its role was precluded by its low scale
reliability. Conceptually, it would make sense that a professional
orientation might make leadership more impossible than unnecessary.
Indeed, only if the organization's top management, itself, had a
professional orientation would an individual employee's
professional orientation act as a substitute for leadership.
(3) Additional individual-level characteristics that might act as
substitutes for leadership need to be tested. Possibilities could

include the role of self-management (Manz and Sims, 1980) as a

141

substitute and the importance of a Protestant Work Ethic, i.e.,
finding hard work, in and of itself, satisfying.

Two final directions for future research can be mentioned
that build directly upon the present results. One would be to
identify substitutes, in addition to customers, that can be a
source of both positive and negative influence upon employees.
Second, would be to explore in what kinds of service organizations
the present results would differ. That is, are there service
organizations in which customer instrumental behavior may be
positively related to employee outcomes? Are there service
organizations in which the magnitude of customer influence may be
greater than that found for the present organization? Answering
these quetions can help managers better influence employee and

customer behavior in the service sector.

Conclusion

 

This study examined the different leader and non-leader
sources of influence upon the attitudes and performance of
boundary-role employees of service organizations. A central focus
was how the unique characteristics of service organizations, e.g.
customer involvement in the creation of a service, might result in
customers acting as a substitute for leadership. The results
indicated that although customers did not act as substitutes for
leadership, they did influence employees to a similar or greater

degree than formal leadership, organizational formalization, or a

142
professional orientation.
Two concluding observations can help summarize the
research. The first involves the relatively modest magnitude of

the results. The real significance of the results lies in their

directing attention to how customers can affect the quality of the
employees' work experience. That is, they emphasize the importance
of viewing customers not just as recipients of organizational
outcomes, but as contributors to employees' work experiences. This
represents an immodest change from how the customer's role has
typically been studied. Furthermore, the organization, itself, can
pursue strategies to decrease or increase the magnitude of the
significant results attributable to customers as a source of
influence. For example, organizations might diminish customer
influence by "buffering" (Chase, 1978) employees from them. On the
other hand, organizations might try to increase the magnitude of
the customer's influence by using them as a source of information
in appraising employee performance. In effect, this would place in
the customers' hands some control over organizational rewards. In
turn, this might alter both the strength and direction of the
influence of customer instrumental behavior. In sum, the results
can be interpreted as indicating that customers do, indeed, play a
significant role in service organizations and it is partly
management's choice to decide upon the magnitude of that role.

The second summary observation is that the results
underscore the idea that both employees and the organization have

considerable degrees of freedom available for choosing the sources

143

of influence on which they will rely. The different sources of
influence do not so much take the place of one another as they
offer supplemental and countervailing effects to one another. It
becomes management's task to decide which combination of leadership
and potential substitutes for leadership will influence employees
most effectively and efficiently. In turn, this decision can
benefit from understanding how employees, themselves, choose among
alternative sources. Finally, this process of trying to produce a
match between the organization's and employee's mix of sources can
be guided by using conventional leadership models to assess the

effects not only of leadership, but of possible substitutes for

leadership as well.

APPENDICES

APPENDIX A

FORMULAS CITED

Note 1: the James, Wolf, and Demaree (1981) formula for assessing

within—group interrater reliability:

rwg = 1 _ (82x Hfze)

where:

r
W8

S2
x

a2.

within-group interrater reliability for a single
group of raters who have rated the same target

the observed (error) variance on variable X among
the raters within a given group

the variance on X that would be expected if the
raters responded randomly, which is estimated by
(A2 — 1) / 12 for a discrete, uniform distribution
and A = number of response points on the rating

scale.

Note 2: Guilford's (1954) centile position formula used in making

rankings comparable across groups, given differences in numbers of

things ranked within groups:

 

(Ri - .5) 100
P:
n
P = centile value
R1 = rank value
n = number of things ranked

144

Note 3:

Note 4:

145
Hotelling's (1940) formula for the significance of the

difference between correlated coefficients of correlation

 

tdr = (r12 - r13 ) (N - 3) (1 + r23)
2(1 ' r232 ‘ r122 ' r132 + 2r23r12rl3)
where:
tdr = t—test value of the difference
r12 = correlation between criterion and one prediction
r13 = correlation between criterion and a second
predictor
r23 = correlation between the two predictors
N = lowest number of cases for any of the three
correlations

Sympson's (1979) formula for the significance of the
difference of two multiple Rs, where the predictors in the

two regression equations are mutually exclusive.

 

 

 

 

 

 

 

 

Zd = [MSE — MSE ]
I~SSE1 + SSEZ //A- 1 2r(e211, 8212) 1/2
2 + —
2(N—l)-P1-P2 EV—Pl-l N—P2-l [(N—P1-1)(N-P2’1)] ,
L .4- .J
Zd = test statistic for standardized normal distribution
MSE = Mean Squared Error
SSE = Sum of Squared Criterion Residuals
N = Sample size

Note 5:

146

P = number of predictors

r(e2i1, e212) = correlation between squared residuals

 

 

Nunnally's (1978) formula for estimating the correlation
between two variables after adjusting the reliability of
one of the variables by a certain amount:
I r

_ _ r11
r12 * r12

\l r11
where:
T12 = estimated correlation after adjusting reliability of

one of the variables

observed correlation before adjustment

H
;_.|
M

II

- adjusted reliability of the variable

01

I-J

|._|
I

observed, unadjusted reliability of the variable

"I
H
I-‘

ll

147

APPENDIX B

MEANS AND STANDARD DEVIATIONS OF VARIABLES AT THE
INDIVIDUAL—LEVEL WITH ALL EMPLOYEES

 

 

 

Variable Mean Standard Deviation

1. Leader Supportive 2.08 .79
Behavior

2. Leader Instrumental 2.58 .55
Behavior

3. Total Leader Behavior 2.52 .48

4. Customer Supportive 2.33 .64
Behavior

5. Customer Instrumental 3.85 .69
Behavior

6. Customers as Task 11.05 2.25
Enrichment

7. Professional Orientation 3.15 .80

8. Organizational 2.75 .61
Formalization

9. Employee Perceptions 2.61 .74
of Service Quality

10. Employee Job Satisfaction 2.09 .54

LIST OF REFERENCES

148

APPENDIX C

MEANS AND STANDARD DEVIATIONS OF ALL VARIABLES AT THE
DEPARTMENT—LEVEL WITH AGGREGATED DISPENSER DATA

 

 

Variable

Mean

Standard Deviation

 

Leader Supportive
Behavior

Leader Instrumental
Behavior

Total Leader Behavior

Customer Supportive
Behavior

Customers as Task
Enrichment

Professional Orientation

Organizational
Formalization

Customer Perception
of Service Quality

Department Rank

1.90

2.64

2.48

2.40

3.77

11.28

2.92

2.55

.25

.74

.69

.55

.61

.63

2.17

.53

.49

1.02

LIST OF REFERENCES

Adams, J. S. The structure and dynamics of behavior in
organizational boundary roles. In M. D. Dunnette (Ed.),

Handbook pf organizational and industrial psychology. Chicago:
Rand McNally, 19761

Aldrich, H. E. Organizations and environments. Englewood Cliffs,
N.J.: Prentice-Hall, 1979.

Adlrich, H. E. and D. Herker. Boundary spanning roles and

organizational structure. Academy 9; Management Review, 1977,
2, 217—230.

Barrow, J. C. Worker performance and task complexity as causal
determinants of leader behavior style and flexibility. ,gpprpal
pf_Appliengsychology, 1976, or, 433—440.

Bass, B. Stogdills handbook a: leadership. New York: Free Press,
1981.

 

Bass, B. and E. Valenzi. Contingent aspects of effective
management styles. In J. G. Hunt and L. L. Larson (Eds.),

Contingency approaches .Ep leadership. Carbondale, 111.:
Southern Illinois University Press, 1974.

Bateson, J. E. G. Do we need service marketing. In Marketing
consumer services: New insights, report no. 77—115.
Cambridge, Mass.: Marketing Science Institute, 1977.

 

 

Bell, D. The coming s£_post-industrial society: a;venture .12
social forecasting. New York: Basic Books, Inc., 1973

Blake, R. S. and J. S. Mouton. The managerial grid. Houston:
Gulf Publishing, 1964.

Chase, R. D. Where does the customer fit in a service operation?
Harvard Business Review, November - December, 1978, 137—142.

Chase, R. B. The customer contact approach to services:
Theoretical bases and practical extensions. Operations
Research, July—August, 1981,.g9, 698-706.

 

149

150

Clark, C. The conditions .9: economic advancement (third ed.).
London: The MacMillan Co., 1957.

Danet, B. Client-organization relationship. In P. C. Nystrom and
W. H. Starbuck (Eds.), Handbook 9: organizational design. New
York: Oxford University Press, 1981.

 

Dansereau, F., J. Cashman, and G. Graen. Instrumentalit .theory
and equity theory as complementary approaches in predic ing the

relationship of leadership and turnover among managers.

Organizational Behavior and Human Performance, 1973, '19, 184—
200.

Dubin, R. Supervision and productivity: Empirical findings and
theoretical considerations. In R. Dubin, G. C. Homans, F. C.
Mann and D. C. Miller (Eds.), Leadership and productivity:
Some facts .2: industrial life. San Francisco: Chandler
Publishing, 1965, 1—50.

Eiglier, P. and E. Langeard. A new approach to service marketing.
In Marketing consumer services: New insights, report 77-115.
Boston: Marketing Science Institute, 1977.

 

Evans, M. G. The effects of supervisory behavior on the path-goal

relationship. Organizational Behavior and Human Performance
1970, p, 277-298.

Fiedler, F. E. A contingency model of leadership effectiveness.
In L. Berkowitz (Ed.), Advances .1p_ experimental social
psychology. New York: Academic Press, 1964.

Fiedler, F. E. .a_theory.p£_leadership effectiveness. New York:
McGraw—Hill, 1967.

Fiedler, F. E. Leadership. Morristown, N. J.: General Learning
Press, 1971.

Filley, A. C., R. J. House and S. Kerr. Managerial process and
organizatinal behavior (2nd ed.). Glenview, 111.: Scott
Foresman, 1976.

 

 

Fitzsimmons, J. A. and R. S. Sullivan. Service operations
management. New York: McGraw—Hill, 1982.

 

Foote, N. N. and P. K. Hatt. Social mobility and economic
advancement. American Economic Review, May, 1953, 364-378.

Franklin, J. L. Relations among four social-psychology aspects of
organizations. Administrative Science Quarterly, 1975, 29 422—
434.

151

Georgopolous, B. A. Hospital organization research: Review and
source book. Philadelphia: W. B. Saunders Company, 1975.

Gersuny, C. and W. R. Rosengren. The service society. New York:
Schoken, 1970.

Goldstein, P. A. and M. Sorcher. Changing supervisory behavior.
New York: Pergammon, 1974.

Graen, G. Role-making processes within complex organizations. In

M. D. Dunnette (Ed.), Handbook pf_organizational and industrial
psychology Chicago: Rand McNally, 1976.

Graen, G., F. Dansereau, Jr. and T. Minami. Dysfunctional

leadership styles. Organizational Behavior and Human
Performance, 1972, Z 216—236.

Guilford, J. P. Psychometric methods. New York: McGraw-Hill Book
Company, 1954.

Hackman, J. R. Work design. In J. R. Hackman and J. L. Suttle

(Eds.), Improving life a; work. Santa Monica, Ca.: Goodyear
Publishing Company, 1977, 96-162.

 

Hackman, J. R. and E. E. Lawler III. Employee reactions to job

characteristics. Journal .pf, Applied Psychology Monograph,
1971, 2;, 259-286.

Hackman, J. R. and G. R. Oldham. Development of the job diagnostic
survey. Journal pf_Applied Psychology, 1975, pp, 159-170.

Hackman, J. R. and G. R. Oldham. Motivation through the design of

work: Test of a theory. Organizational Behavior and Human
Performance, 1976, la 250—279.

Heller, F. A. and G. Yukl. Participation, managerial decision

making, and situational variables. Organizational Behavior and
Human Performance, 1969, a, 227—241.

Holland, J. L. The psychology pf vocational choice (rev. ed.).
Waltham, Mass.: Blaisdell, 1973.

Hotelling, H. The selection of variates for use in prediction,
with some comments on the general problem of nuisance
parameters. Annual Mathematic Statistics, 1940, 11, 271—283.

House, R. J. A path—goal theory of leader effectiveness,
Administrative Science Quarterly, 1971, lg, 321—338.

152

House, R. J. and M. L. Baetz. Leadership: Some empirical
generalizations and new research directions. In B. M. Staw
(Ed.), Research is organizational behavior, Greenwich, Conn.:
JAI Press, 1979, 341—423.

House, R. J. and G. Dressler. The path—goal theory of leadership:
Some post-hoe and a—priori tests. In J. G. Hunt and L. L.
Larson (Eds. Contin enc a roaches to leadershi
Carbondale, Ill.: ou ern linois Universif7'Press, 1972:
29-550

House, R. J. and T. R. Mitchell. Path—goal theory of leadership.
Journal pf_Contemporary Business, 1974, En 81-97.

Howell, J. P. and P. W. Dorfman. Substitutes for leadership: Test

of a construct. Academy pf_Management Journal, 1981a, 23, 714-
728.

Howell, J. P. and P. W. Dorfman. Worker professionalism and
substitutes pf leadership. Paper presented at the Academy of
Management Annual Meeting, San Diego, 1981b.

Hunt, J. G. and R. N. Osborn. An adaptive—reactive theory of
leadership: The role of macro variables in leadership
research. In J. G. Hunt and L. L. Larson (Eds.), Leadership
frontiers. Carbondale, Ill.: Southern Illinois University
Press, 1975.

Hunt, J. G., R. N. Osborn and R. S. Schuler. Relations of
discretionary and nondiscretionary leadership to performance

and satisfaction in a complex organization. Human Relations,
1978, 91: 507—523.

 

James. L. R., G. Wolf, and R. G. Demaree. Estimating interrater
reliability .ia incomplete desigps. Office of Naval Research
Technical Report, Georgia Institute of Techology, 1981.

Katz, D. and R. L. Kahn. The social_psychology_pf organizations.
New York: Wiley, 1966.

 

Kahn, R. L., D. M. Wolfe, R. P. Quinn, J. D. Snoeck, and R. A.
Rosenthal. Organizational stress: Studies is role conflict
and ambiguity. New York: Wiley, 1964.

 

Kerr, S. Substitutes for leadership: Some implications for

organizational design. Organization and Administrative
Sciences, 1977, 8, 135.

 

Kerr, S. and J. Jermier. Substitutes for leadership: Their meaning

and measurement. Organizational Behavior and Human Performance,
1978, 375—403.

153

Kerr, S., C. A. Schriesheim, C. J. Murphy, and R. M. Stogdill.
Toward a contingency theory of leadership based upon the
consideration and initiating structure literature.
Organizational Behavior and Human Performance, 1974, 12, 62—82.

Kerr, S. and J. W. Slocum, Jr. Controlling the performance of
people in organizations. In P. C. Nystrom and W. H. Starbuck
(Eds. ), Handbook of organizational design (vol. 2Q_. New York:
Oxford University Press, 1981,116—134.

Kruse, L. C. and R. M. Stogdill. The leadership role of the nurse.
Columbus: The Ohio State University Research Foundation, 1973.

Kunin, T. The construction of a new type of attitude measure.
Personnel Psychology, 1955, 8, 65-78.

Langeard, E. J., C. H. Bateson, C. H. Lovelock, and P. Eiglier.
Services marketing: New insights from consumers and managers,
81—104. Cambridge, Mass.: Marketing Science Institute, 1981.

Levitt, T. Production line approach to service. Harvard Business
Review, September - October, 1972, 41-52.

 

Levitt, T. The industrialization of service. Harvard Business
Review, September — October, 1976, 63-74.

 

Lord, R. C. Group performance as a function of leadership behavior
and task structure: Toward an explanatory theory.
Organizational Behavior and Human Performance, 1976, ll, 76-96.

Locke, E. A., K. N. Shaw, L. M. Saari, and G. P. Latham. Goal

setting and task performance: 1969—1980. Psychological
Bulletin, 1981, 29, 125—152.

 

Lovelock. C. H. Why marketing management needs to be different for
services. In J. H. Donnelly and W. R. George (Eds.), Marketing
pf_services. Chicago: American Marketing Association, 1981.

Manz, C. C. and H. D. Sims, Jr. Self—management as a substitute

for leadership: A social learning theory perspective. Academy
pf Management Review, 1980, s, 361—368.

Miller, E. J. and A. K. Rice. Systems 2: organization. London:
Tavistock, 1967.

Mills, P. K. and D. J. Moberg. Perspectives on the technology of

service operations. Academy pf Management Review, 1982, _Z,
467—478.

154

Miner, J. The uncertain future of the leadership concept: An
overview. In J. G. Hunt and L. L. Larson (Eds.), Leadership

frontiers. Carbondale, Ill.: Southern Illinois University
Press, 1975.

 

Nunnally, J. C. _Psychometric pheory_(second edition). New York:
McGraw—Hill Book Company, 1978.

Osborn, R. N. and J. G. Hunt. An adaptive—reactive theory of
leaderhsip: The role of macro variables in leadership
research. In J. G. Hunt and L. L. Larson (Eds.)., Leadership

frontiers. Kent, Ohio: Comparative Administrative Research
Institute, 1975.

 

Ouchi, W. The relationship between organizational structure and

organizational control. Administrative Science Quarterly,
1977,-22,95—113.

 

Parkington, J. J. and B. Schneider. Some correlates of experienced

job stress: A boundary-role study. Academy 'p£_ Management
Journal, 1979,_22, 270—281.

 

Parsons, T. How are clients integrated in service organizations?
In W. R. Rosengren and M. Lefton (Eds.) Organizations and
clients: Essays ip_the sociology sf service. Columbus, Oh.:
Charles E. Merrill Publishing Company, 1970.

 

Pennings, J. M. and P. S. Goodman. Toward a workable framework.
In P. S. Goodman and J. M. Pennings (Eds.), New perspectives as

organizational effectiveness. San Francisco: Jossey—Bass,
1977.

 

Perrow, C. A framework for comparative organizational analysis.
American Sociological Review, 1967, 32, 194-208.

Pfeffer, J. The ambiguity of leadership. Academy a; Management
Review, 1977,_2, 104—111.

 

Pfeffer, J. and G. Salancik. Determinants of supervisory behavior:
A role set analysis. Human Relations, 1975, 2§_139-154.

Pfeffer, J. and G. Salancik, The external control .QI
organizations. New York: Wiley, 1978.

 

Porter, L. W. and F. J. Smith. The etiology pf organizational

commitment. Unpublished manuscript, University of California—
Irvine, 1970.

 

155

Porter, L. W., R. M. Steers, R. T. Mowday, and P. V. Boulian.
Organizational commitment, job satisfaction, and turnover among

psychiatric technicians. Journal pf_Applied Psychology, 1974,
_5_9_, 603-609 .

Rizzo, J. R., R. House, and S. Lirtzman. Role conflict and

ambiguity in complex organizations. Administrative Science
Quarterly, 1970,_l§, 150—163.

Roberts, K. H. and W. Glick. The job characteristics approach to

task design: A critical review. Journal ,p£_ Applied
Psychology, 1981,_§§, 193—217.

Schmidt, F. Implications of a measurement problem for expectancy
theory research. Organizational Behavior and Human
Performance, 1981, lg, 243—251.

Schneider, B. Staffing, organizations. Santa Monica, Ca.:
Goodyear Publishing Company, 1976.

Schneider, B. The service organization: Climate is crucial.
Organizational Dynamics, Autumn, 1980, 52—65.

Schneider, B. and D. E. Bowen. Employee and customer perceptions
,pf, service ,ip, banks: Replicaton and extension with
implications for integrating consumer and organizational
behavior. Working Paper, University of Maryland, 1983.

 

Schneider, B., J. J. Parkington, and V. M. Buxton. Employee and
customer perceptions of service in banks. Administrative
Science Quarterly, 1980, 22, 252—267.

 

Schriesheim, C. A. The development and validation pi instrumental
and supportive leadership scales and their application 52 some
tests pf path—goal theory sf leadership hypotheses Unpublished
doctoral dissertation, The Ohio State University, 1978.

 

Schriesheim, J. F. The social context of leader-subordinate
relations: An investigation of the effects of group

cohesiveness. Journal.p£ Applied Psychology, 1980, ,§§, 183—
194.

Schriesheim, C. A. and J. Bish. The content validity of Form XII
of the Ohio State leadership scales. Paper presented at the
National Academy of Management Conference, Seattle, 1974.

Schriesheim, C. and S. Kerr, Psychometric properties of the Ohio

State leadership scales. Psychological Bulletin, 1974, ,§l,
756-765.

156

Schriesheim, C. A. and S. Kerr, Theories and measures of
leadership: A critical appraisal of current and future
directions. In J. G. Hunt and L. L. Larson (Eds.), Leadership;

The cutting, edge. Carbondale, 111.: Southern Illinois
University Press, 1977, 9—45.

 

Shamir, B. Between bureaucracy and hospitality-some
organizational characteristics of hotels. Journal ,gf

Management Studies, 1978, lg, 285—307.

Shamir, B. Between service and servility: Role conflict in
subordinate service roles. Human Relations, 1980, pp, 741-756.

Sheridan, J. E., D. J. Vrdenburgh, and M. A. Abelson. Contextual
model ,9; leadership influence is hospital units. Unpublished
manuscript, The Pennsylvania State University, 1981.

 

 

Shostack, G. L. Breaking free from product marketing. Journal sf
Marketing, April, 1977(a), 73-80.

Shostack, G. L. Banks sell services-—not things. Bankers
Magazine, 1977(b), 40-45.

Shostack. G. L. How to design a service. In J. H. Donnelly and W.
R. George (Eds.), Marketing pf_services. Chicago: American
Marketing Association, 1981.

Sims, H. P. The leader as a manager of reinforcement
contingencies: An empirical example and a model. In J. G.
Hunt and L. L. Larson (Eds.), Leadership ap_the cutting adge.
Carbondale, Ill.: Southern Illinois University Press, 1977.

 

 

Slocum, J. W. Jr. and H. P. Sims, Jr. A typology for integrating

technology, organization, and job design. Human Relations,
1980, pp, 193—212.

 

Snyder, C. A., J. R. Cox, and J. Jesse. Dependent demand approach
to service organization planning and control. Academy .QI
Management Review, 1982,_Z, 455-466.

Starbuck, W. Organizations and their environments. In M. D.
Dunnette (Ed.), Handbook ,2: organizational and industrial
,psychology, Chicago: Rand McNally, 1976}

 

Sympsom, J. B. Testing differences between multiple correlations.
Research report. Princeton, N.J.: Educational Testing
Service, December, 1979.

Tannenbaum, R. and W. Schmidt. How to choose a leadership pattern.
Harvard Business Review, 1958, pp, 95-101.

 

 

157

Thomas, D. Strategy is different in service businesses. Harvard
Business Review, 1978, 29: 158—165.

Thompson, J. D. Organizations ip_action. New York: McGraw-Hill,
1967.

Tosi, H. J. Toward a paradigm shift in the study of leadership.
In J. G. Hunt, V. Sekaran, and C. A. Schriesheim (Eds.),
Leadership: Beyond establishment views. Carbondale, 111.:
Southern Illinois University Press, 1982.

Van de Ven, A. Delbecq, and R. Koenig. Determinants of

coordination modes within organizations. American Sociological
Review, 1976, a}, 322—338.

 

Vroom, V. and P. Yetton. Leadership and decision making.
Pittsburg: University of Pittsburgh Press, 1973.

 

Weiss, D. J., R. V. Dawis, G. W. England, and L. H. Lofquist.
Manual for the Minnesota satisfaction questionnaire.
Minneapolis: University of Minnesota, 1967.

 

 

Yukl, G. Leadership is organizations. Englewood Cliffs, N.J.:
Prentice—Hall, Inc., 1981.

 

MICHIGAN STATE UNIV. LIBRARIES
'. llllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllll
31293006716074