{)RGANIZATIONAI~ BE,ttAVIGR AND HUMAN PERFORMANCE 6, 4 1 4 - 4 4 0 (19,71)

Toward a Behavioral Theory of Leadership 1

GARY YUKL
Department of Psychology, The University oj Akron,,
Akron, Ohio ~804

A great deal of the apparent inconsistency in the leadership literature may

be due to semantic confusion about leader behavior and to the absence of a
eoneeptual framework which includes intermediate and situationM variables.
A system of three distinct leader behavior dimensions is proposed to reduce
this confusion. Two of the dimensions are similar to the familiar variables,
Consideration and Initiating Structure. The third dimension, Decision-Cen-
tralization, refers to the extent to which a leader allows his subordinates to
participate in decision-making. A discrepancy model is developed to explain
the relation between leader behavior and subordinate satisfaction with the
leader. A multiple linkage model is developed to explain how the leader be-
havior variables interact with situational variables to determine group pro-
ductivity. A review of the leadership literature revealed that the results of
previous research are generally consistent with the proposed models. The
compatibility of the linkage model with F~edler's Contingency Model is dis-
cussed, and suggestions for future research are offered.

Despite over two decades of extensive leadership research, the relation

of leader behavior to subordinate productivity and satisfaction with the
leader is still not very dear. The apparent absence of consistent rela-
tionships in the research literature (Sales, 1966; Norman, 1966; Lowin,
1968) m a y be due in part to several related problems. First, there is a
great deal of semantic confusion regarding the conceptual and opera-
tional definition of leadership behavior. Over the years there has been a
proliferation of leader behavior terms, and the same term is often
defined differently from one study to the next. Secondly, a great deal
of empirical data has been collected, but a theoretical framework which
adequately explains causal relationships and identifies limiting con-
ditions has not yet emerged. Finally, the research has often failed to
include intermediate and situational variables which are necessary in
order to understand how a leader's actions can affect his subordinates'
productivity.
The purpose of this article is to begin the development of a theory
which explains how leader behavior, situational variables, and inter-

1The author is grateful to Is2en Wexley and Alexis Anikeeff for their helpful com-
ments.
414
 BEHAVIORAL THEORY OF LEADERSHIP ~15

mediate variables interact to determine subordinate productivity and

satisfaction with the leader. In the first section of the article, a system
of three distinct and generally applicable leader behavior dimensions
will be proposed. In the next two sections, these leadership dimensions
will be used to develop a discrepancy model of subordinate satisfaction
and a multiple linkage model of leader effectiveness. Finally, the extent
to which the research literature supports these behavioral models wilt
be evaluated.

CLASSIFICATION OF LEADER BEHAVIOR

Consideration and Initiating Structure

Some early investigators began with a list of very specific leadership
activities (e.g., "inspection," "write reports, . . . . hear complaints") and
attempted to determine how performance of these activities or the
amount of time allocated to them related to leader success. Since the
number of specific leader activities that are possible is nearly endless,
several Ohio State University psychologists attempted to find a few
general behavior dimensions which would apply to all types of leaders.
Factor analyses of leadership behavior questionnaires were carried out,
and two orthogonal factors were found (Hemphill & Coons, 1957; Halpin
& Winer, 1957). These factors were called Consideration and Initiating
Structure. Consideration refers to the degree to which a leader acts in
a warm and supportive manner and shows concern and respect for his
subordinates. Initiating Structure refers to the degree to which a leader
defines and structures his own role and those of his subordinates toward
goal atta;nment.
The principal method for measuring these variables has been the use
of either the Leader Behavior Description Questionnaire (Hemphill &
Coons, 1957) or the Supervisory Behavior Description questionnaire
(Fleishman, 1957a). These questionnaires are administered to a leader's
subordinates. A related questionnaire, called the Leadership Opinion
Questionnaire (Fleishman, 1957b), is administered to the leader him-
self. This questionnaire is considered to be a measure of leader attitudes
rather than leader behavior. Occasionally other observers, such as peers
or superiors, are the source of leader behavior descriptions, and in some
studies Consideration and Initiating Structure are experimentally ma-
nipulated by having leaders play predetermined roles.

Decision-Centralization
A somewhat different approach to the classification of leaders was
initiated by Lewin's (1944) theoretical typology of democratic, auto-
416 ~AgY YU~L

eratie, and laissez-faire leaders. St,udies following in this tradition have

usually focused on the relative degree of leader and subordinate in-
fluence over the group's decisions. The various decision-making pro-
eedures used by a leader, such as delegation, joint decision-making,
consultation, and autocratic decision-making, can be ordered along a
continuum ranging from high subordinate influence to complete leader
influence. Although a leader will usually allow more subordinate par-
tieipation and influence for some decisions than for others, the average
degree of participation can be computed for any specified set of typical
decisions, tIeller and Yukl (1969) have used the term "Decision-Cen-
tralization" to refer to this average. A high Decision-Centralization score
means a low amount of subordinate participation. Naturally, a leader
is capable of voluntarily sharing decision-making with his subordinates
only to the extent that he has authority to make decisions.
Most methods that have been used to measure participation can also
be regarded as a measure of Decision-Centralization. Participation and
Decision-Centralization have been measured by subordinate ratings of
their perceived autonomy or influence in decision-making, by subordinate
responses to a questionnaire concerning the leader's decision behavior,
and by leader responses to a decision behavior questionnaire. In some
studies the leader's actual decision-making behavior has been experi-
mentally manipulated. The term Decision-Centralization was introduced
for two reasons. First, this term emphasizes the behavior of the leader
rather than the behavior of the subordinates. Second, the definition of
Decision-Centralization explicitly encompasses a greater variety of
leader decision procedures than does the typical definition of partici-
pation (Heller & Yukl, 1969).2

Reconciling the Two Approaches to Leader Behavior Classification

Is Decision-Centralization equivalent to Consideration and Initiating
Structure, or is it a distinct leadership dimension? The degree to which
the three dimensions are independent depends upon the precise definitions
given them. Since the definitions vary from study to study, it is not sur-
prising that there is some disagreement regarding the relation between
these dimensions. For example, Lowin (1968) has suggested that Initi-
ating Structure is conceptually similar to autocratic supervision, Sales
(1966) has suggested that "employee orientation" (which includes high
Consideration) is usually associated with democratic leadership, and
Newport (1962) has suggested that Consideration and Initiating Strue-
Despite my preference for the term Decision-Centralization, the more familiar
term participation will usually be used when discussing the direction of correlations
in order to avoid confusion.
 BEHAVIORAL THEORY OF LEADERSHIP 417

ture are similar, respectively, to democratic and autocratic leadership.

On the other hand, Gomberg (1966), MeMurry (1958), Schoenfeld
(1959), and Stanton (t962) have claimed that high Consideration and
autocratic leadership are not incompatible, or in other words, that Con-
sideration and Decision-Centralization are separate dimensions.
There are several sound theoretical arguments for treating Decision-
Centralization as a separate dimension of leader behavior. Let us look
first at the relation between Consideration and Decision-Centralization.
The Consideration scale in the Ohio State questionnaires includes sev-
eral items pertaining to the decision-making participation of subordi-
nates, and Consideration is sometimes defined as including the sharing
of decision-making with subordinates. However, one can argue that this
sharing is only considerate of subordinates when they clearly desire
participation, and the desire for participation can vary substantially
from person to person and from situation to situation. Inclusion of par-
ticipation items in a Consideration scale results in scores which are not
comparable across persons unless first adiusted for differences in partici-
pation preferences. It is more practical to define Consideration as simply
the degree to which a leader's behavior expresses a positive attitude
rather than an indifferent or negative attitude toward subordinates. When
defined in this manner, Consideration can be regarded as conceptually
distinct from Decision-Centralization. In general, a high Consideration
leader is friendly, supportive, and considerate; a low Consideration
leader is hostile, punitive, and inconsiderate. A leader who acts in-
different and aloof is between these extremes but is closer to the low
end of the continuum. The specific behaviors used in sealing Considera-
tion should be generally applicable to all types of leadership situations.
What about the relation between Decision-Centralization and Initi-
ating Structure? Although Initiating Structure is defined broadly as task-
oriented behavior, it appears to include at least three types of task be-
havior: (1) Behavior indicating the leader's concern about productivity
(e.g., goal-oriented comments to subordinates, and use of various rewards
and punishments to encourage productivity), (2) behavior insuring that
necessary task decisions are made, and (3) behavior insuring that these
decisions and directives from higher levels in the organization are
carried out (e.g., training and supervision). Note that this definition does
not specify who will actually make the decisions. The task orientation
of the leader does not appear to be very closely related to the amount
of influence he will allow subordinates in the making of task or mainte-
nance decisions. Even very autocratic leaders can differ considerably
with respect to their task orientation and concern about group per-
formanee. Therefore, it seems reasonable to treat Initiating Structure
418 OARY YC~L

and Decision-Centralization as separate dimensions of leader behavior.

The empirical evidence on the relation of Decision-Centralization to
Consideration and Initiating Structure is scanty, and the research which
will be cited should be regarded as suggestive rather than conclusive.
Most of these studies use the Consideration scale of the Leader Behavior
Description Questionnaire, which includes some participation items.
Naturally these items increase the likelihood of finding a significant cor-
relation between Consideration and Decision-Centralization.
In a study of 67 second-line supervisors in three companies, this
author found a low but significant correlation (r = - - . 2 4 ; p < .05) be-
tween Consideration and Decision-Centralization. Decision-Centraliza-
tion was measured by means of leader responses on the decision pro-
cedure questionnaire (Form C) described in Heller and YukI (1968).
There was no significant correlation between Decision-Centralization
and Initiating Structure.
Other evidence is provided by analyses of a more recent version of
the Leader Behavior Description Questionnaire, which has ten new
subscales in addition to the original scales for Consideration and Initi-
ating Structure. One of the new scales, called "Tolerance of Member
Freedom," can be regarded as a measure of participation or Decision-
Centralization. Stogdill, Goode, and Day (1962, 1963, 1964) admin-
istered this questionnaire to "subordinates" of corporation presidents,
labor union presidents, community leaders, and ministers. The correla-
tions between Consideration and Tolerance of Member Freedom for
the four samples, respectively, were .41, .42, .40, and .49. For a sample
of office supervisors rated by female subordinates on this questionnaire,
the correlation was .50 (Beer, 1966). 3 Decision-Centralization and
Initiating Structure were not significantly correlated in any of the five
samples just described.
Argyle, Gardner, and Cioffi (1957) analyzed the relation among
leadership dimensions as measured by questionnaires administered to
managers in England. Demo.eratie (vs authoritarian) leadership oor-
related .41 with nonpunitive (vs punitive) leadership. Democratic leader-
ship was not significantly correlated with pressure for production, a
component of Initiating Structure.
If we remember to reverse the sign of the correlation when necessary
in order to correct for the fact that high participation equals low De-
cision-Centralization, then it is obvious that the results of these studies
are remarkedly consistent. Decision-Centralization and Initiating Struc~

~Signifieanee levels for tlle correlations were not given, but judging from the
sample sizes, they should all be significantat the .05 level or better.
 BEHAVIORAL THEORY OF LEADERSHIP 419

ture appear to be independent dimensions. Decision-Centralization and

Consideration should probably be regarded as oblique rather than
orthogonal dimensions. That is, there will tend to be a low to moderate
negative correlation between them, but some leaders will have high
scores on both dimensions ("benevolent autocrat") and some leaders
will have low scores on both dimensions ("malevolent democrat").

A DISCREPANCY MODEL OF SUBORDINATE SATISFACTION

WITH THE LEADER
In this section, a discrepancy model of satisfaction will be used to
explain the relation of the three leadership dimensions to subordinate
satisfaction with the leader. Discrepancy or subtraction models of job
satisfaction have been proposed by a number of psychologists (Morse,
1953; Sehaffer, 1953; Rosen & Rosen, 1955; Ross & Zander, 1957;
Porter, 1962; Katzell, 1964; Locke, 1969). In a discrepancy model, satis-
faction is a function of the difference between a person's preferences
and his actual experience. The less the discrepancy between preferences
and experience, the greater the satisfaction. This hypothesis has received
some support in the studies cited above, but the evidence is by no
means conclusive. In some versions of the discrepancy model there is
a second hypothesis which states that the amount of dissatisfaction
with a given discrepancy also depends upon the importance of the needs
affecting the preference level. If importance varies from person to person,
the discrepancy scores cannot be compared unless first adjusted for
importance. Whether such a correction is necessary, and if so, how it
should be made appears to be a matter of growing controversy.
Although the discrepancy model appears to be applicable to the anal-
ysis of subordinates' satisfaction with their leader, only a few studies
have used it for this purpose. In two of these studies (Foa, 1957; Greet,
1961), leadership variables other than Consideration, Initiating Struc-
ture, and Decision-Centralization were used. No studies were found
which included subordinate preferences for Consideration and Initiating
Structure as a moderating variable. The results from studies which
have included subordinate preferences for participation in decision-mak-
ing tend to be consistent with the discrepancy model.
According to the proposed discrepancy model, the shape of the curve
relating leader behavior to subordinate satisfaction will vary somewhat
depending upon a subordinate's preference level. A preference level will
be defined tentatively as a range of leader behavior acceptable to subor-
dinates rather than as a single point on a behavior continuum. Figure 1
shows the theoretical curves for a low, medium, and high preference
level. The curves represent the relation for a single subordinate. When
420 GARY Y U K L

Z high A,
0

3
~o
q

to} J o w PL
low high
BEHAVIOR

z high B,
0

w
P,iX~
l o w

low high
BEHAVIOR

=h~gh
0
~ i
U
<

'<
vi low PL [
low high
BE H A V I O R

Fie. 1. The relation between leader behavior and subordinate satisfaction for a
low, medium, and high preference level (PL).

the preference levels of group members are relatively homogeneous, the

relation between leader behavior and average group satisfaction with
their leader will yield a curve similar to that for an individual. How-
ever, the more variable the preferences are in a group, the less likely it
is t h a t any significant relation wili be found between leader behavior
and average group satisfaction.
Subordinate preference levels are determined both by subordinate
personality and by situational variables (see Fig. 2). Preferences can be
expected to v a r y more for Initiating Structure and Decision-Centraliza-
tion than for Consideration. Except for a few masochists, is is probably
safe to assume that subordinates will desire a high degree of consid-
erate behavior by their leaders. As a result, the function relating Con-
sideration and subordinate satisfaction should resemble curve C in Fig. 1.
Preference levels for Decision-Centralization, i.e., the subordinate's
 B E H A V I O R A L T H E O R Y OF L E A D E R S H I P 421

SUBORDINATE~ I
PERSONALITY
SUBORDINATE
PREFERENCES
SITUATIONAL
VARIABLES SUBORDINATE
SATISFACTION
LEADER
I BEHAVIOR
Fro. 2. A discrepancy model of subordinate satisfaction with the leader.

desire for participation in decision-making, may be partially determined

by two personality traits: Authoritarianism (Vroom, 1959) and "need
for independence" (Trow, 1957; Ross & Zander, 1957; Vroom, 1959;
Beer, 1966, p. 51; French, Kay, & Meyer, 1966). Although none of these
investigators assessed the relation between a personality measure and
expressed behavior preferences, they did find that personality had the
expected moderating effect upon the relation between Decision-Centraliza-
tion and subordinate satisfaction. However, it should be noted that Tosi
(1970) was not able to replicate the results of the study by Vroom (1959).
The measurement of subordinate preferences in future replications may
aid in clearing up the contradiction between these two studies.
The major situational determinant of the preference level for partici-
pation in making a decision is probably the importance of that decision
for' the subordinate (Maier, 19'65, p. 165). When a decision is very im-
portant to subordinates, they are likely to prefer as much influence as
possible (e.g., joint decision-making or delegation). When decisions do
not involve matters of importance, consultation or even autocratic de-
cision-making is more likely to be preferred. Of course, the more that
subordinates trust their leader to make a decision favorable to them, the
less need they will feel to participate in order to protect their interests.
Also, when th~ subordinates are committed to group goal attainment or
survival and the task or environment favors centralized decision-making
(e.g., a crisis), then they are likely to expect the leader to make most
of the decisions (Mulder & Stemerding, 1963).
Preference levels for Initiating Structure are partially determined by
the subordinates' commitment to group goals and their perception of
the amount of structuring that is necessary to help the group attain these
goals. Subordinates who are indifferent about or hostile toward t h e goal
of maximum productivity are likely to prefer a leader who is not very
task oriented in his behavior.

Summary of the Discrepancy Model

The major features of the proposed discrepancy model can be sum-
marized in terms of the following hypotheses:
422 GAnY YUKL

Hyp 1: Subordinate satisfaction with the leader is a function of the

discrepancy between actual leader behavior and the behavior preferences
of subordinates.
tIyp 2: Subordinate preferences are determined by the combined
effect of subordinate personality and situational variables.
Hyp 3: Subordinates usually prefer a high degree of leader Consid-
eration, and this preference level results in a positive relation between
Consideration and subordinate satisfaction.
The discrepancy model in its present form is only a static model repre-
senting one-way causality at one point, in time. No attempt has been
made to include additional complexities such as the effects of leader be-
havior on subordinate preferences. For example, a leader who gradually
allows greater subordinate participation may find that the subordinates'
preference for decision-making increases over time. Nor does the model
explicitly deal with such other determinants of subordinate satisfaction
with the leader as his intelligence or the feedback effects from success-
ful or unsuccessful group performance. Finally, the influence of various
components of the model on leader behavior has also been ignored. For
example, subordinate preferences represent one of several sources of
role expectations for the leader, and these role expectations interact
with other situational variables and leader personality to determine his
behavior.

A MULTIPLE LINKAGE MODEL OF LEADER EFFECTIVENESS

When a leader is dependent upon his subordinates to do the work,
subordinate performance is unlikely to improve unless the leader can
increase one or more of the following three intermediate variables: (1)
Subordinate task motivation (i.e., effort devoted to their tasks), (2)
subordinate task skills, and (3) Task-Role Organization (i.e., the tech-
nical quality of task decisions).4 Consideration, Initiating Structure, De-
cision-Centralization, and various situational variables interact in their
effects on these intermediate variables. The intermediate variables in-
teract in turn to determine group performance (see Fig. 3).

Consideration, Initiating Structure, and Subordinate Motivation

Consideration and Initiating Structure interact in their effect upon
subordinate task motivation. Subordinate task motivation will be highest
when the leader is high on both Consideration and Initiating Structure.

~The leader can also improve productivity by obtaining necessary information,

resources, and cooperation from other organization members and outside agencies,
but this involves leader behavior outside the context of the work group.
 ]3EHAVIOI~AL THEOIgY 0F LEADERSI-IIP 42~

LEADER ~--~ SUBORDINATE k

[ eONSIDE~TION , . ] / ~ MOTIVATION

I DEClSION- ~ >
L CENTRALIZATION! ///~' ORGANIZATION RMANCB

INITIATING ~ _ _ _ > ~I SUBORDINATE

I STRUCTURE SKILL LEVEL

1 SITUATIONAL
L VARIABLES
FIG. 3. A multiple linkage model of leader effectiveness.

The ordering of the other combinations is less certain, because the in-
teraction appears to be highly complex and irregular. If leaders were
subgrouped according to their Initiating Structure scores, for high struc-
turing leaders there would probably be a positive relation between Con-
sideration and subordinate task motivation. For low structuring leaders,
there is some reason to suspect that the relation between Consideration
and subordinate motivation is described by an inverted U-shaped cm've.
In other words, subordinate task motivation can be adversely affected
when the low structuring leader is either very supportive and friendly
or very hostile and punitive.
There are at least two hypotheses for explaining the interaction be-
tween Consideration and Initiating Structure, and it is not yet clear if
either or both are correct. From instrumentality theory (Vroom, 1964,
p. 220; Galbraith & Cummings, 1967), comes the hypothesis that a
leader can improve subordinate performance by being highly considerate
to subordinates who make an effort to perform well, while withholding
Consideration from subordinates who show little task motivation. In
effect, considerate behavior is a reward which is contingent upon the
display of certain task-motivated behavior by subordinates.
The "identification" hypothesis proposes that subordinate motivation
is a response to previous leader Consideration rather than an attempt to
obtain future Consideration. As Consideration increases, subordinate at-
titudes toward the leader become more favorable and his influence over
the subordinates increases correspondingly. In effect, the considerate
leader has greater "referent power" (French & Raven, 19,59). However,
in order for subordinate loyalty to be translated into task motivation,
it is necessary for the leader to communicate a concern for productivity.
If the leader is highly considerate but does not stress productivity, the
subordinates are likely to feel that they can safely neglect their tasks.
424 QAaY YUKL

If a leader actually becomes hostile and punitive, it is likely that

subordinate task motivation will be adversely affected, regardless of the
level of Initiating Structure. Punitive leadership can lead to counter-
aggression by subordinates in the form of slowdowns and subtle sabotage
(Day & Hamblin, 1964).

Decision-Centralization and Subordinate Motivation

Although there is some direct evidence that subordinate participation
can result in increased task motivation (Baumgartel, 1956), the nature
and relative importance of the psychological processes accounting for
the relation and the prerequisite conditions for their occurrence are not
yet clear. A number of explanations for the effect of participation on
subordinate motivation have been proposed during the last two decades.
Probably the most important of the proposed processes is the pos-
sibility that subordinates become "ego-involved" with a decision which
they have helped to make. When subordinates identify with a decision,
they become motivated to help make the decision successful, if only to
maintain a favorable self-concept. However, there may be several limit-
ing conditions for this causal sequenee (Strauss, 1964; Vroom, 19'64;
Lowin, 1968). It is possible that there is some minimal amount of in-
dividual influence, actual or perceived, which is necessary before iden-
tification will occur. As a group gets larger, the influence of each member
over a decision will necessarily decline; thus the size of the group may
be one limiting factor. Also, it is not clear whether a person who supports
a proposal that is rejected will become committed to the proposal finally
selected by the group. Another prerequisite may be the subordinate's per-
ception that the decision process is a test of his decision ability and those
skills of his which are used in implementing the decision. In the ease where
subordinates participate in making decisions unrelated to their tasks,
there is no reason to assume that any increased commitment to these de-
cisions will generalize to task decisions. Finally, if responsibility for
making decisions is thrust upon subordinates who do not want it or who
see it as the legitimate role of the leader, then these subordinates may
fail to identify with the decisions (French, Israel, & As, 1960).
Another explanation of the relation between Decision-Centralization
and task motivation is that participation facilitates reduction of subor-
dinate resistance to change (Coch & French, 1948). One way this could
occur is through direct persuasion. Since the leader is usually not aware
of all the subordinates' fears an.d doubts regarding a proposed change,
consultation provides him with an opportunity to uncover these fears
and to persuade subordinates that the change will be beneficial rather
 BEHAVIORAL THEORY OF LF~gDERSHIP 4:25

than harmful. When a leader's proposal involves features which clearly

are detrimental to subordinates, mere persuasion is not likely to win their
support. However, consultation or joint decision-making provides the
opportunity for bargaining and agreement on a compromise proposal
which the subordinates can support (Strauss, 1964).
When the leader allows his subordinates to make a group decision,
the interaction dynamics of the group are yet another possible source
of increased task motivation. If the work group is cohesive, its members
are subject to direct social pressure to conform to group norms (Schachter,
Willerman, Festinger, & Hyman, 1961; Berkowitz, 1954; Seashore, 1954).
In addition, the work group may function as a "reference group" for its
members (Neweomb, 1965, p. 109). Subordinates who have positive atti-
tudes toward their work group will tend to support group norms, includ-
ing group decisions made in a legitimate manner. This tendency for mem-
ber attitudes and behavior to be consistent with reference group norms
will occur even in the absence of direct social pressure.
Of course, increased commitment to carry out decisions is not con-
ceptually equivalent to increased task motivation. Subordinates can
make task decisions which in effect restrict output or resist change.
Subordinate attitudes toward the leader and the organization constitute
an important situational variable which moderates the effect of partici-
pation upon task motivation. If relations between the leader and the
subordinates are very poor, or the subordinates are in opposition to the
goal of maximum group performance, then participation in decisions in-
volving production goals, standards, quotas, etc., is not likely to result in
increased subordinate task motivation (Strauss, 1964:). Since Considera-
tion is an important determinant of subordinate attitudes toward the
leader, participation is more likely to be effective if combined with high
Consideration than if combined with low Consideration.

Leader Behavior and Subordinate Taslc Skill

The second way in which leaders can increase group performance is
to increase the ability of subordinates to perform their individual tasks.
A number of studies (reviewed in Vroom, 1964, p. 197) support Maier's
(1965) hypothesis that performance is a function of a person's Motiva-
tion X Ability. According to this hypothesis, even highly motivated sub-
ordinates will not perform well if they lack the necessary knowledge or
skills to carry out their assignments. Therefore, one way for a leader to
improve group performance is to correct deficiencies in subordinate task
skills and knowledge by means of on-the-job instruction and improved
downward communication of task-relevant information. Instruction and
426 GAnY YV~_~

communication of this nature are, by definition, elements of Initiating

Structure. A more complex analysis of the relation between Initiating
Structure and subordinate task skill was beyond the seope of this article.

The Nature of Task-Role Organization

Task-Role Organization refers to how efficiently the skill resources of
subordinates are utilized to perform the group's formal tasks. Adequacy
of Task-Role Organization depends upon how well job assignment de-
cisions and work method decisions are made. The making of job assign-
ment decisions is usually referred to in industrial psychology as "place-
ment" or "classification." When the jobs of each subordinate are identical
and subordinates work independently of each other, it doesn't matter
what subordinates are assigned to what jobs. However, when jobs are
highly specialized, each job has different skill requirements, and skill
differences among subordinates are substantial, then job assignments
are an important type of task decision. If work assignments are not made
carefully, the skills of some workers will not be fully utilized, while other
workers will be placed in jobs which they cannot perform adequately.
Furthermore, if the jobs are interdependent, bottlenecks will occur at
various points in the flow of work.
Work method decisions are important whenever a task can be per-
formed in many different ways, and some ways are better than others.
Work methods and procedures can be designed with the available skills
of a particular work group in mind, but it is common practice in indus-
trial engineering to ignore individual differences and develop methods
which maximize the efficiency of the typical worker. Decisions about
work procedures are not always the responsibility of the leader. In some
organizations, work methods are designed by staff specialists or are
rigidly prescribed by company or union regulations.
Task-Role Organization was included in the multiple linkage model to
account for any variability in group productivity which is not attribut-
able to subordinate motiva{ion, subordinate ability, or to extraneous
events such as an improvement in the flow of material inputs, a break-
down in equipment, etc. The identification of Task-Role Organization as
a separate variable is analogous to Maier's (1965) distinction between
the quality of a decision and group acceptance of the decision. Although
Task-Role Organization is an important conceptual component of the
multiple linkage model; measurement of this variable is likely to prove
troublesome. Any measure of Task-Role Organization will be highly
specific to a given set of tasks and subordinates. Within a specific situa-
tion, one could attempt to scale the adequacy of job assignment de-
cisions by evaluating the match between iob requirements and subor-
 BEHAVIORAL THEORY OF L]~ADERSHIP 427

dinate skills for all possible combinations of iob assignments. Adequacy

of work method decisions could be evaluated in several ways. In some
situations, the accumulated knowledge of industrial engineering spe-
cialists may permit the subjective ranking of various possible work pro-
cedures according to their relative efficiency. When objective measures of
group performance (e.g., quantity or quality of output, labor time, errors)
are available to use as a criterion of efficiency, then alternative work
methods may be experimentally compared. However, it may be difficult
to hold task motivation constant, even within a single work group, be-
cause job design can affect the intrinsic motivation of workers as well
as their efficiency.

Initiating Structure, Decision-Centralization, and

Task-Role Organization
Both Initiating Structure and Decision-Centralization appear to be
related to Task-Role Organization. By definition, a leader who is high in
Initiating Structure will attempt to improve the efficiency of his group.
However, simply engaging in structuring behavior does not guarantee
t:hat Task-Role Organization will improve. The leader's success depends
upon his organizing skills, technical knowledge, and the extent to which
he taps the knowledge of his subordinates by allowing them some degree
of participation in making task decisions. The relation between Decision-
Centralization and Task-Role Organization is moderated by the relative
amount of leader and subordinate organizing skills and task knowledge.
When the leader is very capable in this respect but the subordinates
lack the appropriate talents, then there will be a negative relation be-
tween participation and Task-tl.ole Organization. When the subordinates
have more relevant knowledge and organizing talent than the leader, we
would expect a positive relation between participation and Task-Role
Organization. We have already seen that Decision-Centralization can
affect the task motivation of subordinates as well as the quality of task
decisions. This means that in the situation where there is a negative
relation between participation and Task-Role Organization, there may
also be a positive relation between participation and subordinate mo-
tivation. When such a trade-off dilemma occurs, some intermediate de-
gree of Decision-Centralization will probably be optimal with respect
to group performance.
In some situations, the quality of task decisions involves a time di-
mension. That is, the effectiveness of decisions depends in part upon
how quickly they are made (Strauss, 1964; Lowin, 1968). Autocratic
decision-making is faster than other decision procedures because little
communication with subordinates is necessary. Therefore, participation
428 GARY YUICL

is likely to be negatively related to group performance when rapid

decision-making is required. The magnitude of this negative relation will
be greatest when the leader already has the necessary knowledge and
ability to make good decisions, the subordinates are motivated by the
urgency of the situation, and the task group is very large.

Summary of the Multiple Linkage Model

The major features of the multiple linkage model of leadership effec-
tiveness can be summarized by means of the following hypotheses:

Hyp 1: Group productivity is a function of the interaction among

subordinate task motivation, subordinate task skills, and Task-Role
Organization for the group.
ttyp 2: Initiating Structure and Consideration interact in the deter-
ruination of subordinate task motivation. Task motivation is highest when
the leader is high on both behavior variables.
Hyp 3: Decision-Centralization is negatively correlated with subor-
dinate task motivation (i.e., high participation causes high motivation)
when subordinate relations with the leader are favorable, the decisiofis
are relevant to subordinate tasks, and subordinates perceive their par-
tieipation to be a test of valued abilities.
Hyp 4: Initiating Structure interacts with Decision-Centralization in
the determination of Task-Role Organization. The relationship is mod-
erated by the level and distribution of task knowledge and planning
ability in the group.
Hyp 5: Initiating Structure is positively related to the level of subor-
dinate task skill.

REVIEW OF RELATED RESEARCH

Most studies of the relation between leader behavior and subordinate
satisfaction with the leader have not measured subordinate preferences
or the personality and situational variables which determine these
preferences. Most studies of the relation between leader behavior and
group productivity have not included measures of the intermediate and
situational variables in the proposed linkage model. The approach typical
of mosi~ leadership research has been to look for a linear relation between
leader behavior and one of the criterion variables. Nevertheless, previous
research does provide some direct and some indirect evidence for eval-
uating the proposed models.
In the following sections of this article, relevant leadership research
will be reviewed. The review will include studies dealing with variables
which are reasonably similar to those in the proposed discrepancy and
linkage models. However, it should be emphasized that in many of these
 BEHAVIORAL THEORY OF LEADERSHIP 429

studies, the operational measurement of a variable only approximates

the conceptual definition presented in this article. Studies using scales
which can be regarded as a measure of leader attitudes (e.g., LPC
scale, P scale, Leadership Opinion Questionnaire) rather than leader
behavior were not included. Also exeluded were studies of general vs
close Supervision. This leadership dimension, as usually defined, con-
founds Decision-Centralization with Initiating Structure. Finally, the
review does not include studies of emergent leaders in informal groups,
studies using children, studies involving an entire organization rather
than individual work groups or departments, and studies in which leader
behavior is obviously confounded with organizational variables such as
the incentive system.

Consideration and Satisfaction

In seven studies of the relation between Consideration and subordinate
satisfaction with their leader, Consideration was measured by means of
subordinate responses on leader behavior description questionnaires. In
five of these studies (Halpin, 1957; Halpin & Winer, 1957; Nealey &
Blood, 1968; Yukl, 1969a; Anderson, 19'66) there was a strong positive
relation between Consideration and subordinate satisfaction. In the re-
maining two studies (Fleishman & Harris, 1962; Skinner, 1969) there
was a significant curvilinear relation between Consideration and two ob-
jective measures which reflect subordinate satisfaction, namely turnover
and grievances. The curve describing the relation corresponded roughly
to curve C in Fig. 1. If subordinate preferences were homogeneous, this
curve would represent supporting evidence for the concept of a zone of
indifference within which leader behavior does not affect subordinate
satisfaction. Below this indifference zone, the relation between Consid-
eration and satisfaction was positive.
In research reported in Likert (196.1, p. 17), aspects of Consideration
such as "supervisor takes an interest in 'me and understands my prob-
lems" and "supervisor thinks of employees as human beings rather than
as persons to get the work done," were related to favorable attitudes on
job-related matters. In two laboratory experiments (Day & Hamblin,
1964; Misumi & Shirakashi, 1966) punitive leader behavior (i.e., low
Consideration) was associated with low subordinate satisfaction. In an-
other laboratory experiment, Lowin (19'69) found a significant positive
relation between subordinates' satisfaction and their ratings of leader
Consideration, but the difference in satisfaction between high and low
Consideration conditions, although in the right direction, was not signifi-
cant.
Only two studies were found in which a positive relation between Con-
430 (~AR~" YL-~:L

sideration and subordinate satisfaction with the leader did not occur. In
a study by Argyle, Gardner, and Cioffi (1958), leader self-reports of
punitive behavior did not correlate significantly with subordinate turn-
over and absences. Pelz (1952) found an interaction between the degree
to which a leader acts as a representative of his subordinates when deal-
ing with higher management (one form of Consideration) and the
leader's upward influence in the organization. For leaders with little up-
ward influence, subordinates were less satisfied when the leader "went
to bat" for them than when he did not go to bat. Presumably the leader
representation raised expectations which he could not fulfill, thereby
frustrating subordinates. In terms of the discrepancy model, the subor-
dinates' preferences for leader representation are probably lower when
it repeatedly causes frustration. Whether the negative effects of unsuc-
cessful representation can completely cancel out the positive effects of
other considerate behavior by the leader is not clear. It does not seem
likely.
In summary, the research literature indicates that in most situations,
considerate leaders will have more satisfied subordinates. Although none
of the investigators included subordinate preferences in their analysis,
the results are consistent, with the discrepancy model if we can make
the relatively safe assumption that most subordinates prefer considerate
leaders.

Initiating Structure and Satisfaction

A consistent linear relation between Initiating Structure and subor-
dinate satisfaction was not found, even within sets of studies using
comparable measures. Unfortunately, none of the studies reviewed in-
eluded subordinate preferences. Baumgartel (1956), ttalpin and Winer
(1957), Argyle et at. (1958), 3/iisumi and Shirakashi (1966), Lowin
(1969), Anderson (1966), and Likert (1961, pp. 16-18) failed to find a
significant relation. IIalpin (1957) and Yukl (1969a) found positive cor-
relations. Vroom and Mann (1960) found a significant negative correlation
between pressure for production and job. satisfaction for delivery truck
drivers but not for loaders. Nealey and Blood (196.8) found a negative
correlation between Initiating Structure and subordinate satisfaction
for second-level supervisors and a positive correlation for first-level
supervisors.
Only three studies were found which examined the possibility of a
curvilinear relation between Initiating Structure and subordinate satis-
faction. Liker~ (1955) found that the relation between pressure for
productivity and subordinate satisfaction took the form of an inverted
U-shaped curve which is similar to curve B in Fig. 1. Fleishman and
 BEttAVIORAIJ T H E O R Y OF D E A D E R S H I P 431

Harris (1962) and Skinner (1969) found a curvilincar relation between

Initiating Structure and both turnover and grievances. Although subor-
dinate preferences were not measured, the relationships in these studies
were roughly comparable to curve A in Fig. 1.
Yleishman and Harris also tested for an interaction between Initiating
Structure and Consideration. The results of their analysis suggest that
Consideration has a greater effect upon subordinate satisfaction than
does Initiating Structure. High Consideration leaders could increase
Initiating Structure with little accompanying increase in turnover or
grievances. Fleishman and Harris provide two possible explanations for
this interaction. One explanation is that considerate leaders are more
likely to deal with any dissatisfaction caused by high structuring be-
havior before the dissatisfaction results in official grievances or with-
drawal (i.e., turnover). Another explanation is that Consideration affects
the way subordinates perceive structuring behavior. In terms of the dis-
crepancy model, subordinates of highly considerate leaders are more
likely to have a higher preference level for Initiating Structure because
they do not perceive leader structuring as threatening and restrictive.

Decision-Centralization and Satisfaction

Six studies were found which examined the correlation between subor-
dinate satisfaction and participation as perceived either by the leaders
or by the subordinates (Baumgartel, 1956; Argyle et al., 1958; Vroom,
1959; Bachman, Smith, & Slesinger, 1966; Yukl, 1969a; Tosi, 1970). In
each of these studies, evidence was found to support a positive relation
between participation and subordinate satisfaction, although within some
of the studies, a significant relation was not obtained for every subsample
or for every alternative measure of the variables. A significant positive
relation was also found in each of five studies in which participation was
experimentally manipulated (Coch & French, t948; Shaw, 1955, Morse
& :Reimer, 1956; Solem, 1958; Maier & Hoffrnan, 1962). The results of
these studies are generally consistent with the discrepancy model if one
can assume that the subordinates preferred a substantiaI degree of par-
ticipation.
In those eases where a significant relation between participation and
subordinate satisfaction was not found, there was usually some reason
to expect that the subordinates preferred a moderate or low amount of
participation. In the study by Vroom (1959), a positive correlation oc-
curred for subordinates with a high need for independence but not for sub-
ordinates with a low need for independence. Bass (1965, pp. 169-170) and
French et al. (1960) found that subordinate participation did not result
in more favorable attitudes toward a leader unless the subordinates per-
432 GARY YUKL

ceived the decision-making as a legitimate part of their role. Further

evidence for the moderating effect of subordinate preferences can be found
in a study by Baumgartel (1956) and in two unpublished studies (Jacob-
son, 1951; Tannenbaum, 1954) which were reported in Likert (1961, pp.
92-93). In the Tennenbaum study, some subordinates reacted adversely
to a sudden substantial increase in participation. Finally, Morse (1953,
p. 64) found that, regardless of whether workers made some decisions
or none, they reported more intrinsic job satisfaction when the amount
of decision-making equalled the amount desired than when they were
not allowed to make as many decisions as they desired. Although intrinsic
job satisfaction is conceptually distinct from satisfaction with the leader,
these two variables are probably highly correlated when the leader de-
termines how much responsibility a subordinate has for making task
decisions.
Consideration, Initiating Structure, and Productivity
Considering the complexity of tl~e interaction between Consideration
and Initiating Structure, it is not surprising that research on the relation
between Consideration and productivity does no~ yield consistent results.
In the large majority of studies there was either a significant positive
relation (Katz, Maccoby, Gurin, & Floor, 1951; Argyle et al., 1958;
Beseo & Lawshe, 1959; Schachter e~ al., 1961; Kay, Meyer, & French,
1965) or there was no significant linear relation (Bass, 1957; Halpin,
1957; Rambo, 1958; Day & Hamblin, 1964; Anderson, 1966; Nealey &
Blood, 1968; Rowland & Scott, 1968). Lowin (19'69) found a positive
relation for objectively manipulated Consideration in an experiment but
not for subordinate ratings of Consideration. A significant negative rela-
tion was found by EIalpin and Winer (1957) for aircraft commanders and
by Fleishman, Harris, and Burtt (1955, p. 80) for foremen of production
departments but not for nonproduction departments. In both of these
studies, productivity was measured by superior ratings, and the highest
ratings went to leaders low on Consideration but high on Initiating
Structure. It is possible that the ratings were influenced more by the
raters' task-oriented stereotype of the ideal leader than by actual group
performance.
Turning to research on the relation between Initiating Structure and
productivity, we again find mixed results. In a number of studies a
significant positive relation was reported (Fleishman et al., 1955; Likert,
1955; tIalpin &Winer, 1957; Maier & Maier, 1957; Beseo & Lawshe,
1959; Anderson, 1966; Nealey & Blood, 1968). For some subsamples in
three of these studies, and for leaders studied by Argyle et al. (1958),
Bass (1957), Halpin (1957), Rambo (1958), and Lowin (1969), a signifi-
 BEHAVIORAL THEORY OF LEADERSHIP 433

cant relation was not found. In no case was a significant negative rela-
tion reported.
It is unfortunate that so few investigators measured intermediate
variables or tested for an interaction between Consideration and Initiating
Structure. However, the few studies which are directly relevant to the
proposed linkage model do provide supporting evidence. In a laboratory
experiment in Japan, Misumi and Shirakashi (1966) found that leaders
who were both task oriented and considerate in their behavior had the
most productive groups. Halpin (1957) found that aircraft commanders
were rated highest in effectiveness when they were above the mean on
both Consideration and Initiating Structure. Hemphill (1957) obtained
the same results for the relation between the behavior of department
chairmen in a Liberal Arts College and faculty ratings of how well the
department was administered. Fleishman and Simmons (1970) translated
the Supervisory Behavior Description into Hebrew and administered this
questionnaire to the superiors of Israeli foremen. Proficiency ratings for
the foremen were also obtained from their superiors. Once again, the fore-
men with the best ratings tended to be high on both Consideration and
Initiating Structure. P atehen (1962) found that personal production
nerms (i.e., task motivation) of workers were highest when the leader
encouraged proficiency as well as "going to bat" for them. These pro-
duetion norms were related in turn to actual group production. Finally,
although he didn't measure Consideration, Baumgartel (1956) found a
significant positive relation between subordinate motivation and the
concern of research laboratory directors for goal attainment (i.e., Initi-
ating Structure).

Decision-Centralization and Productivity

Seventeen studies were found which examined the relation between
Decision-Centralization and group productivity. A significant positive
relation between participation and productivity was found by Baebman
et al. (1966), Coch and French (1948), Fleishman (1965), French (1950),
French, Kay, and Meyer (1966), Lawrence and Smith (1955), Likert
(1961 , p. 20), Mann and Dent (195zi), McCurdy and Eber (1953),
Meltzer (1956), and Vroom (1959). Argyle et al. (1958) found a positive.
relation only for departments without piece rates, suggesting that the
organizational incentive system, a situational variable, interacts with
Decision-Centralization in determining the subordinates' task motiva-
rich. Tosi (1970), French et al. (1960), and McCurdy and Lambert
(1952) failed to find a significant relation between participation and
productivity. In two other studies (Shaw, 1955; Morse & Reimer, 1956)
a significant negative relation was found. Several of these sbldies demon-
434 ~AI~Y Y~1o~

strafe that various situational variables can moderate the effects of leader
decision behavior on group performance. Nevertheless, the high percentage
of studies reporting a positive relation is an indication that some degree
of participation leads to an increase in group performance in most situa-
tions. However, this generalization is not equivalent to concluding "the
more participation there is, the greater will be group productivity." For
a particular group, there is probably some optimal pattern of decision-
making which will consist of various amounts of delegation, joint de-
cision-making, consultation, and autocratic decision-making (Heller &
Yukl, 1969). The optimal pattern is likely to involve some intermediate
amount of subordinate influence, rather than the greatest possible
amount.

DISCUSSION
The Multiple Linkage Model and FiedIer's Contingency Model
A considerable number of leadership studies have been conducted by
Fred Fiedler and his associates at the University of Illinois (Fiedler,
1967). Fiedler has developed a theory of leadership effectiveness to ex-
plain the results of this research. According to Fiedler's theory, group
performance is a function of the interaction between the leader's "esteem
for his least preferred co-worker" (LPC) and three situational variables:
task structure, leader-member relations, and the position power of the
leader. Leaders with low LPC scores have the most productive groups
when the leadership situation, in terms of the three situational variables,
is either very favorable or very unfavorable. Leaders with high LPC
scores are more effective when the situation is intermediate in favorable-
ness. Although Fiedler provides a behavioral explanation for these hy-
pothesized relations, most of his studies did not measure leader behavior.
The few studies which have attempted to identify the behavioral cor-
relates of LPC scores have not yielded consistent results (Sample &
Wilson, 1965; Fiedler, 1967, p. 53; Nealey & Blood, 1968; Yukl, 1970;
Gruenfeld, Ranee, & Weissenberg, 1969; Reilly, 1969). Thus, it is not
possible at this time to determine whether Fiedler's model is compatible
with the proposed linkage model. Both theories are generally supported
by their own separate bodies of empirical research. Reconciliation of the
two approaches will probably require additional research which includes
variables from both theories.
Direction for Future Research
The theoretical framework and ~he literature review presented earlier
point out some empirical gaps which badly need filling. The central
feature of the linkage model is the set of intermediate variables. A
 BEHAVIORAL THEORY OF LEADERSHIP 435

leader can do little to improve group productivity unless he can alter one
or more of these variables. Yet the mediating role of these variables, their
relation to each other, and their interaction in the determination of pro-
ductivity have seldom been investigated in leadership studies. Future
research should be more comprehensive in scope. Leader behavior vari-
ables, intermediate variables, situational variables, subordinate prefer-
enees, criterion variables (i.e., satisfaction and productivity), and relevant
leader traits should all be included. Situational variables other than those
discussed in this article also need to be investigated. Likely candidates
are the organizational limiting conditions for participation suggested by
Lowin (1968) and Strauss (1964), the structural variables found to be
associated with leader decision behavior by Heller and Yukt (1969),
the situational variables in Fiedler's model, the situational variables
cluster-analyzed by Yukl (1969b), and Woodward's (1965) system for
classifying production technology. Finally, the way in which the three
behavior dimensions interact in determining the intermediate variables
should be investigated. If possible, the leader behavior variables should
be experimentally manipulated in order to avoid the measurement prob-
lems associated with leader behavior descriptions by subordinates.
The analysis of leader effectiveness has utilized leader behavior vari-
ables which maintain a basic continuity with traditional conceptualization
and research. However, in speculating about future research, it is ap-
propriate to evaluate the continued usefulness of these broadly defined
behavior dimensions. It is obvious that Consideration and Initiating Struc-
ture are composed of relatively diverse elements, while Decision-Cen-
tralization is an average based on many different types of decisions. In
order to improve the predictive power of the model, it may be necessary
to identify which components of the behavior variables are the most
important determinants of each intermediate variable.
The discrepancy model and the multiple linkage model provide only
the skeleton of a static leadership theory which purposely ignores the
additional complexities of feedback loops and circular causality. Much
additional research and revision will be necessary to transform the skele-
ton into a full-fledged dynamic model which permits accurate predictions
about leader effectiveness in formal task groups.

REFERENCES
A:~DERS05r, L. R. Leader behavior, member attitudes, and task performance of inter-
cultural discussion groups. Journal of Sociat Psychology, 1966, 69, 305-319.
ARaYLE, M., GARDXER, G., & CIOFFI, F. The measurement of supervisory methods.
Human Relations, 1957, 10, 295-313.
AEa~E, M., GARDXEmG., & CmFFI, F. Supervisory methods related to productivity,
absenteeism, and labor tmmover. Human Relations, 1958, 11, 2340.
BACH~AN, J. G., SMIT~, C. G., & SLESI~GER,J. A. Control, performance, and saris-
436 GARY YUKL

faction: An analysis of structural and individual effects. Journal o] Pe~'sonaIity

and Social Psychology, 1966, 4, 127-136.
BAss, B. M. Leadership opinions and related characteristics of salesmen and sales
managers. In 1{. M. StogdilI and A. E. Coons (Eds.), Leader behavior: Its de-
scription and measurement. Columbus: Bureau of Business Research, Ohio State
University, 1957.
Bhss, B. M. Organizational psychology. Boston: Allyn & Bacon, 1965.
BAU~ARTEL, H. Leadership, motivations, and attitudes in research laboratories.
Journal o/Social Issues, 1956, 12, (2), 24-31.
BEER, M. Leadership, employee needs, and motivation. Columbus: Bureau of Busi-
ness Research, Ohio State University, Monograph No. 129, 1966.
BERKOWITZ, L. Group Standards, cohesiveness, and productivity. Human Relations,
1954, 7, 509-519.
BEsco, R. O., • LAWSHE, C. I-I. Foreman leadership as perceived by supervisor and
subordinate. Personnel Psychology, 1959, 12, 573-582.
Cocg, L., & FaENCE, J. R. P. Overcoming resistance to change. Human Relations,
1948, 1, 512-532.
DAY, R. C., & HAMBLI~~, R. L. Some effects of close and punitive styles of supervision.
American Journal o] Sociology, 1964, 16, 499-510.
FIEDLER, F. ]~. A theory o] leadership ef]ectiveness. New York: McGraw-ttill, 1967.
]~LEISHlVIAN,E. A. A leader behavior description for industry. In R. M. Stogdill and
A. E. Coons (Eds.), Leader behavior: Its description and measurement. Co-
lumbus: Bureau of Business Research, Ohio State University, 1957 (a).
FLEISHlV~A~, E. A. The Leadership Opinion Questionnaire. In R. M. Stogdill and
A. E. Coons (Eds.), Leader behavior: Its description and measurement. Ca-
lumbus: Bureau of Business Research, Ohio State University, 1957 (b).
FLEISHMAN,E. A. Attitude versus skill factors in work group productivity. Personnel
Psychology, 1965, 18, 253-266.
FLEISHMA~, ]~. A., c~ SImMOnS, J. Relationship between leadership patterns and
effectiveness ratings among Israeli foremen. Personnel Psychology, 1970, 23,
169~172.
FLEISHMA~, ]~. A., & HARRIS,]~. F. Patterns of leadership behavior related to em-
ployee grievances and turnover. Personnel Psychology, 1962, 15, 43-56.
FLEISH~AZ% E. A., HARalS, E. F., & BURTT, It. E. Leadership and supervision, in in-
dustry. Columbus: Bureau of Educational Research, Ohio State University, Re-
search monograph No. 33, 1955.
FoA, U. G. Relation of worker's expectations to satisfaction with his supervisor. Per-
sonnel Psychology, 1957, 10, 161-168.
FaENCE, g. R. P. Field experiments: Changing group productivity. In J. G. Miller
(Ed:), Experiments in social process: A symposium on social psychology. New
York: McGraw-Hill, 1950.
FRENCh, J. R. P., ISRAEL,J., & As, D. An experiment on participation in a Norwegian
factory. Human Relations, 1960, 13, 3-19.
FRENCE, J. R. P., KAY, E., & MEYEa, I-I. Participation and ~he appraisal system.
Human Relatiows, 1966, 19, 3-20.
F~ENC~, J. R. P., & RAVEN, B. The bases of social power. In D. Cartwright (Ed.),
Studies in social power. Ann Arbor: Institute for Social Research, University of
Michigan, 1959.
GnLBRAIT~, J., & CVM~INGS, L. L. An empirical investigation of the motivational
determinants of task performance: Interactive effects between instrumentMity-
 BEHAVIORAL THEORY OF LEADERSHIP 4~7

valence and motivation-ability. Organizational Behavior and Human Per]orm-

ance, 1967, 2, 237-257.
GO~BERa, W. The trouble with democratic management. Transaction, 1966, 3, (5),
30-35.
GRE~R, F. L. Leader indulgence and group performance. Psychological Monographs,
1961, 75 (12, Whole No. 516).
GRCENFELD, L. W., RANCh, D. E., & WEISS~NBER0, P. The behavior of task-oriented
(low LPC) and socially-oriented (high LPC) leaders under several conditions
of social support. Journal o] Social Psychology, 1969, 79, 99-107.
thLPI~, A. W. The leader behavior and effectiveness of aircraft commanders. In
1~. M. Stogdill and A. E. Coons (Eds.), Leader behavior: Its description and
measurement. Columbus: Bureau of Business Research, Ohio State University,
1957.
HALPIN, A. W., & WI~ER, B. J: A factorial study of the leader behavior descriptions.
In 1~. M. Stogdill and A. E. Coons (Eds.), Leader behavior: Its description and
measurement. Columbus: Bureau of Business Research, Ohio State University,
1957.
HELLER,F., ~g YUKL, G. Participation, managerial decision-making, and situational
variables. Organizational Behavior and Human Per]ormance, 1969, 4, 227-241.
H]~MPttILL, J. K. Leader behavior associated with the administrative reputations of
college departments. In R. M. Stogdill and A. E. Coons (Eds.), Leader behavior:
Its description and measurement. Columbus: Bureau of Business Research, Ohio
State University, 1957.
HE~PmLL, J. K., & COONS, A. E. Development of the leader behavior description
questionnaire. In R. M. Stogdill and A. E. Coons (Eds.), Leader behavior: Its
description and measurement. Columbus: Bureau of Business Research, Ohio
State University, 1957.
JAC0BSOX, J. M. Analysis of interpersonal relations in a formal organization. Unpub-
lished doctoral dissertation, University of Michigan, 1953.
KA~:Z, D., MACCOBY,N., GURIN, G., & FLOOR,L. Productivity, supervision, and morale
among railroad workers. Ann Arbor: Survey Research Center, University of
Michigan, 1951.
I~A'rZELL, P~. A. Personal values, job satisfaction, and job behavior. In H. Borrow
(Ed.), Man in a World o] Work. Boston: Houghton-Mifflin, 1964.
KAy, E., ME,R, H. It., & FRE~'eH, J. R. P. Effects of threat in a performance ap-
praisal interview. Journal o] Applied Psychology, 1965, 49, 311-317.
NORMAN, A. IK. Consideration, initiating structure, and organizational criteria--A
review. Personal Psychology, 1966, 19, 349-362.
LAWRENCE,L. C., ~5 S~¢IITI-I, P. C. Group decision and employee participation. Journal
o] Applied Psychology, 1955, 39, 334-337.
LEWlN, I(. The dynamics of group action. Educational Leadership, 1944, 1, 195-200.
LIKERT, R. Developing patterns in management. Strengthening management ]or the
new technology. New York: American Management Association, 1955.
LIKERT, R. New patterns oJ management. New York: McGraw-Hill, 1961.
LOCKE, E. A. What is job satisfaction? Organizational Behavior and Human Per]orm-
ance, 1969, 4, 309-336.
Lowi~, A. Participative decision-making: A model, literature critique, and prescrip-
tions for research. Organizational Behavior and Human PerJormance, 1968, 3,
68-106.
LowIN, A., ~-~RAPCHA1K,W. J., (~5 ~VANAGI-I, M. J. Consideration and Initiating
438 ~A~Y V~KL

Structure: An experimental investigation of leadership traits. Administralive

Science Quarterly, 1969, 14, 238-253.
MAIER, N. R. F. Psychology in industry. Third ed. Boston: Houghton-Mifflin Co.,
1965.
MAmR, N. R. F., & HOFFMAN, L. R. Group decision in England and the United
States, Personnel Psychology, 1962, 15, 75-87.
MAmR, N. R. F., & MAIER, R. A. An experimental test of the effects of "develop-
mental" vs. "free" discussions on the quality of group decisions. Journal o] Ap-
plied Psychology, 1957, 41, 320-323.
MAN~, F. C., & DENT, J. The supervisor: Member of ~wo organizational families.
Harvard Business Review, 1954, 32, (6), 103-112.
MeCuRDY, I-I. O., • EBE~, I.i.W. Democratic vs. authoritarian: A further investiga-
tion of group problem-solving. Journal o] Personality, 1953, 22, 258-269.
McCuRD¥, tI. G., & LAMBERT,W. E. The efficiency of small groups in the solution
of problems requiring genuine cooperation. Journal o/ Personality, 1952, 20,
478-494.
McMuRaAY, R. N. The case for benevolent autocracy. Harvard Business Review,
19'58, 36, (1), 82-90.
MELTZEa, L. Scientific productivity in organizational settings. Journal o] Social Issues,
1956, 12, (2), 32-40.
MIsu~I, J., & SHIRAKASEI,S. An experimental study of the effects of supervisory
behavior on productivity and morale in a hierarchical organization. Human Re-
Iations, 19'66 19, 297-307.
Mo~sE, N. Saris]action in the white-collar job. Ann Arbor: Institute for Social Re-
search, University of Michigan, 1953.
MoRsE, N. C., & REI~ER, E. The experimental change of a major organizational
variable. Journal o] Abnormal and Social Psychology, 1956, ~2, t20-129.
M~-LDER, M., & STE~ERDING,A. Threat, attraction to group and strong leadership: A
laboratory experiment in a natural setting. Human Relations, 1963, 16, 317-334.
NEALEY, S. M., & BLOOD, M. R. Leadership performance of nm'sing supervisors at
two organizational levels. Journal o] Applied Psychology, 1968, 52, 414-422.
NEWEOMB, T. I-I., TS~NEI~, R. H., & CON-WRSE,P. E. Social psychology. New York:
Holt, l~inehart, & Winston, 1965.
NEwPoaT, G. A study of attitudes and leadership behavior. Personnel Administration,
19'62, t~ (5), 42-46.
PATCHE~r, M. Supervisory methods and group performance norms. Administrative
Science Quarterly, 1962, 7, 275-294.
PELZ, D. C. Influence: A key to effective leadership in the first-line supervisor. Per-
sonnel, 1952, ~9, 209-217.
PORTER, L. W. Job attitudes in management: I. Perceived deficiencies in need ful-
fillment as a function of job level. Journal o] Applied Psychology, 1962, 46, 375-
384.
RAMSO, W. W. The construction and analysis of a leadership behavior rating form.
Journal o] Applied Psychology, 1958, 42, 409-415.
I%EILLY,A. J. The effects of different leadership styles on group performance: A field
experiment. Paper presented at the American Psychological Association Conven-
tion, Washington, D. C., Sept. 1, 1969.
RosEN, R. A. I-I., & 1RosE~, R. A. A. Suggested modification in job satisfaction cur-
veys. Personnel Psychology, 1955, 8, 303-314.
 BEHAVIORAL THEORY OF LEADERSHIP 439

Ross, I. C., & Z,~'DER, A. Need satisfactions and employee turnover. Personnel Psy-
chology, 1957, 10, 327-338.
ROWLA~TD,K. M., & SCOTT,W. E. PSychological attributes of effective leadership in a
formal organization. Personnel Psychology, 1968, 21, 365-378.
SALES, S. M. Supervisory style arid productivity: Review and theory. Personnel Psy-
chology, 1966, 19, 275-286.
SAMPLE, J. A., & WILson-, T. R. Leader behavior, group productivity, and rating of
least preferred eoworker. Journal o] Personality and Social Psychology, 1965, 1,
266-270.
SCHACHTER, S., WILLERMA~T,B., FESTI~,TGER,L., & I-IYMAZV,R. Emotional disruption
and industrial productivity. Journal o] Applied Psychology, 1961, 45, 201-213.
SCHAEFER,R. H. Job satisfaction as related to need satisfaction in work. Psychological
Monograph, 1953, 67, (14, Whole No. 364).
SCHOE~FELD, E. Authoritarian management: A reviving concept. Personnel, 1959, 36,
21-24.
SEASHORE, S. Group cohesiveness in the industrial work group. Ann Arbor: Institute
for Social Research, University of Michigan, 1954.
SHAW, M. E. A comparison of two types of leadership in various communication nets.
Journal of Abnormal and Social Psychology, 19'55, 50, 127-134.
S~;~h'NZR, E. W. Relationships between leadership behavior patterns and organiza-
tional-situational variables. Personnel Psychology, 1969, 22, 489~I94.
SOLE~, A. R. An evaluation of two attitudinal approaches to delegation. Journal of
Applied Psychology, 1958, 42, 36-39.
STANTO~¢,E. S. Which approach to management--democratic, authoritarian, or . . .?
Personnel Administration, 1962, 25 (2), 44-47.
STO~DILL, R. M., GOODE, O. S., &' DAY, D. R. New leader behavior description sub-
scales. Journal o] Psychology, 1962, 54, 259-269.
SrOaDILL, R. M., GOODE, O. S., & DAY, D. R. The leader behavior of corporation
presidents. Personnel Psychology, 196,3, 16, 127-132.
STOaDILL, R. M., G00DE, O. S., & DAY, D. R. The leader behavior of presidents of
labor unions. Perso~nel Psychology, 1964, 17, 49~57.
STRAUSS, G. Some notes on power equalization. In H. J. Leavitt (Ed.), The social
science o] organizations: Pour perspectives. Englewood Cliffs, New Jersey:
Prentice-Hail, 1964.
TosI, H. A re-examination of personality as a determinant of the effects of partici-
pation. Personnel Psychology, 1970, 23, 9,1-99.
TAN~CE>-BAUX,A. S. The relationship between personality and group structure. Un-
published doctoral dissertation. Syracuse University, 1954.
Taow, D. B. Autonomy and job satisfaction in task-oriented groups. Journal o] Ab-
normal and Social Psycholo.gy, 1957, 54, 204-209.
VRoo~, V. tI. Some personality determinants of the effects of participation. Jou~,'nal
o] Abnormal and Social Psychology, 1959, 59, 322-327.
VROOM, V. It. Work and Motivation. New York: John Wiley and Sons, 1964.
VR00~, V. tI., & M.4N>~. F. C. Leader authoritariallism and employee attitudes. Pe~'-
sonnet Psychology, 1960, 13, 125-140.
WOODWAaD, J. Industrial orga)zizatio.~a: Theory and practice. London: Oxford Uni-
versity Press, 1965.
YtrKL, G. A. Conceptions and consequences of leader behavior. Paper presented at
440 GARY Y U K L

the annual convention of the California State Psychological Association, Newport

Beach, January, 1969 (a).
YrzKb, G. A. A situation description questionnaire for leaders. Educational and
Psychological Measurement, 1969, 29, 515-518 (b).
YCKL, G. A. Leader LPC scores: Attitude dimensions and behavioral correlates.
Journal o] Social Psychology, 1970, 80, 207-212.

RECEIVED: A p r i l 29, 1970