A Path Goal Theory of Leader Effectiveness

Author(s): Robert J. House

Source: Administrative Science Quarterly, Vol. 16, No. 3 (Sep., 1971), pp. 321-339
Published by: Sage Publications, Inc. on behalf of the Johnson Graduate School of Management,
Cornell University
Stable URL: http://www.jstor.org/stable/2391905 .
Accessed: 19/05/2014 08:09

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Sage Publications, Inc. and Johnson Graduate School of Management, Cornell University are collaborating
with JSTOR to digitize, preserve and extend access to Administrative Science Quarterly.

http://www.jstor.org

This content downloaded from 129.233.1.252 on Mon, 19 May 2014 08:09:20 AM

All use subject to JSTOR Terms and Conditions
Robert J. House

A Path Goal Theory of Leader Effectiveness

An explanationof the effects of leader behavior on subordinate satisfaction,

motivation, and performance is presented. The explanation is derived from a
path-goal theory of motivation. Dimensions of leader behavior such as leader
initiating structure, consideration,authoritarianism,hierarchical influence, and
closeness of supervision are analyzed in terms of path-goal variables such as
valence and instrumentality.The theory specifies some of the situational mod-
eratorson which the effects of specific leader behaviorsare contingent.A set of
general propositionsare advanced which integrate and explain earlier fragmen-
tary researchfindings.Several specific predictionsare made to illustratehow the
general propositionscan be operationalized.The usefulness of the theory is dem-
onstrated by showing how several seemingly unrelated prior research findings
could have been deduced from its general propositionsand by applying it to
reconcile what appear to be contradictoryfindings from prior studies. Results
of two empirical studies are reported that provide support for seven of eight
hypotheses derived directly from the general propositionsof the theory. A third
study designed to test three of the originaleight hypothesesis also reported.Two
of these three hypothesesare successfullyreplicated. In the light of these results
and the integrativepower of the theory, it is argued that the theory shows prom-
ise and should be furthertested with experimentallaswell as correlationalmeth-
ods.

INTRODUCTION logical support, warmth, friendliness, and

Two major behavioral dimensions that helpfulness by doing such things as being
have emerged from leadership research are friendly and approachable, looking out for
those which sociologists have termed instru- the personal welfare of the group, doing little
mental and social-emotional, or expressive, things for subordinates, and giving advance
leadership behavior. Psychologists who have notice of change.
studied leadership have independently dis- Research has indicated that leaders who
covered these dimensions (Korman, 1966). initiate structure for subordinates are gen-
The terms most frequently used to describe erally rated highly by superiors and have
these behaviors are initiating structure and higher producing work groups than leaders
consideration. Leader initiating structure is who are low on initiating structure; and that
used to describe the degree to which the leaders who are considerate of subordinates
leader initiates psychological structure for have more satisfied employees (Filley and
subordinates by doing such things as assign- House, 1969); however the evidence with
ing particular tasks, specifying procedures to respect to the relationship between initiating
be followed, clarifying his expectations of structure and satisfaction of subordinates is
subordinates, and scheduling work to be very mixed. Several studies have shown that
done. This dimension of leader behavior de- initiating structure is frequently resented by
scribes leaders who are similar to those pre- unskilled and semiskilled employees and is a
scribed by classical management theorists, source of dissatisfaction (Filley and House,
that is, leaders who plan, organize, direct, 1969), grievances, and turnover (Fleishman
and control. Leader consideration is used to and Harris, 1962). However, employees in
describe the degree to which the leader large groups have been found either to pre-
creates a supportive environment of psycho- fer initiating structure more or to dislike
321

This content downloaded from 129.233.1.252 on Mon, 19 May 2014 08:09:20 AM

All use subject to JSTOR Terms and Conditions
322 ADMINISTRATIVE SCIENCE QUARTERLY

it less than employees in smaller groups estimate of the probability that his behavior
(Hemphill, 1950; Mass, 19,50; Vroom and will indeed result in the outcomes. Vroom
Mann, 1960). And, Oaklander and Fleish- (1964) formalized one perspective of expec-
man (1964) found initiating structure to be tancy motivation theory mathematically, and
negatively correlated with intergroup con- Galbraith and Cummings (1967) extended
flict. Recent studies have shown that among his formulation by pointing out that some
high-level employees, initiating structure is of the valences associated with a specific be-
positively related to satisfaction (House et havior are intrinsic to the behavior itself and
al., 1971a, 1971b), performance (House et some are the extrinsic consequences of that
al., 1971b), and perceptions of organizational behavior. To the extent that behavior is
effectiveness, but negatively related to role intrinsically valent it is also intrinsically
conflict and ambiguity (Rizzo et al., 1970). motivational because the behavior is highly
This paper presents a theory of leader instrumental to the outcome of satisfaction.
behavior that attempts to reconcile and in- A person will be motivated to engage in such
tegrate the conflicting results of previous behavior because his expectancy that satis-
studies under a set of general propositions faction will follow is nearly unity. That is, if
from which they could have been deduced, the outcomes were contingent on an external
and reports three studies designed to test rewarder-any significant other-the expec-
eight hypotheses derived from the theory.' tancy would be less that unity because the
behavior might not be observed or recog-
THEORETICAL BACKGROUND nized by the rewarder. However, when the
The theory advanced here is derived from reward is essentially self-administering, ex-
the path-goal hypothesis advanced by Geor- pectancy approaches unity.
gopoulos et al. (1957), and from previous The theory may be further extended and
research supporting the broad class of ex- broken down into parts that have specific
pectancy theory of motivation (Atkinson, relevance for leadership using the concept
1958; Vroom, 1964; Porter and Lawler, of path instrumentality advanced by Evans
1967; Galbraith and Cummings, 1967; Graen, (1968: 14): "This is the cognition of the
1969; Lawler, 1968). The central concept of degree to which following a particular path
expectancy theories is that the force on an (behavior) will lead to a particular outcome,
individual to engage in a specific behavior is it is akin to (but not identical to) the con-
a function of (1) his expectations that the cept of 'expectancy' introduced by Vroom."
behavior will result in a specific outcome; Evans (1968) has also advanced an extension
and (2) the sum of the valences, that is, of Vroom's (1964) theory and a path-goal
personal utilities or satisfactions, that he de- theory of leadership. His theory is different
rives from the outcome. The research find- from the one presented here in that its pre-
ings indicate that the function is a nonlinear, dictions are not contingent on situational
monotonically increasing product of expec- variables, and it is not an attempt to account
tations and valences. Thus, according to this for the conflicting findings just reviewed.
theory of motivation, an individual chooses According to the formulation advanced
the behaviors he engages in on the basis of here, the individual makes probability esti-
(1) the valences he perceives to be asso- mates with respect to two linking points
ciated with the outcomes of the behavior connecting behavior with its outcomes, and
under consideration; and (2) his subjective subjectively places values on the outcomes.
The magnitude of these probability estimates
1 This paper is based on a presentationmade at indicates the degree of path instrumentality
the SouthernIllinoisUniversitySymposiumon Con- of his behavior for work-goal accomplish-
temporaryDevelopmentin the Study of Leader- ment and valence. This formulation can be
ship in Carbondaleon April 30, 1971. The author expressed in the following formula:
is indebtedto LarryL. Cummingsfor his critical
commentsand suggestionswhich resultedin signifi-
cantimprovement of an earlierversionof this paper,
andto ThomasH. Stonefor his editorialsuggestions M = IVb + P[IVa + (P2iEVj)
whichhelped clarifythe discussionof the probabili-
ties associatedwith intrinsicallymotivatedbehavior. i= 1,..., n,

This content downloaded from 129.233.1.252 on Mon, 19 May 2014 08:09:20 AM

All use subject to JSTOR Terms and Conditions
 House: LEADER EFFECTIVENESS 323

where: mient, P2. If he is consistent in his decision

M = motivation to work; making with respect to recognizing and
!Vb = intrinsic valence associated with goal- rewarding work-goal achievement, he will
directed behavior; clarify the linkage between work-goal
IVa= intrinsic valence associated with work- achievement and rewards. Thus, if he con-
goal accomplishment; sistently rewards achievement, this will most
EVi = extrinsic valences associated with probably increase the subordinate's path in-
work-goal accomplishment; strumentality, P2, for valent personal out-
P1 = path instrumentality of behavior for comes. Third, through his own behavior he
work-goal attainment; can provide support for the subordinate's
P2i = path instrumentalities of work goal for efforts and thereby influence the probability
extrinsic valences. that this effort will result in work-goal
achievement, that is P1. Fourth, the leader
In wAork situations the individual estimates influences the intrinsic valences associated
the path instrumentality, P1, of his behavior with goal accomplishment, IVa, by the way
for the accomplishment of some work goal. he delegates and assigns tasks to subordi-
Here he considers such factors as his ability nates, which determines the amount of in-
to behave in an appropriate and effective fluence the subordinate has in goal setting
manner as well as the barriers to work-goal and the amount of control he is allowed in
accomplishment in the environment, and the the task-directed effort. The greater the sub-
support he will receive from others to ac- ordinate's opportunity to influence the goal
complish the work goal. In addition, he esti- and exercise control, the more intrinsically
maates the path instrumentality, P2, of the valent the work-goal accomplishment. Fi-
work goal for attaining personal outcomes nally, the leader can increase the net intrinsic
that have valence for him. For example, he valence associated with goal-directed behav-
estimates the probability that his superiors ior, IVb, by reducing frustrating barriers, be-
will recognize his goal accomplishment and ing supportive in times of stress, permitting
reward him accordingly. He also considers, involvement in a wide variety of tasks, and
and places subjective values on the intrinsic being considerate of subordinate's needs.
valence associated with the behavior re-
quired to achieve the work goal, IVb, the PROPOSITIONS
intrinsic valence associated with the achieve-
ment of the work goal, IVa, and the extrinsic The above interpretation of motivation
valences associated with the personal out- theory as applied to leadership suggests the
comes that he accrues as a result of achieve- following general propositions:
ing the work goal, EVA. 1. The motivation functions of a leader
The behavior of the leader is clearly rele- are to increase the net positive valences asso-
vant to all of the independent variables in ciated with work-goal attainment, increase
this formulation. The leader, at least in part, the net positive valences associated with the
determines what extrinsic rewards should be path-behavior-to work-goal attainment,
associated with work-goal accomplishment, and increase the subordinate's path instru-
EVi. For example, he has some influence over mentality with respect to work-goal attain-
the extent to which work-goal accomplish- ment for personal outcomes and the behavior
ment will be recognized as a contribution required for work-goal attainment. This state-
and whether it will be rewarded with finan- ment assumes that when the subordinate is
cial increases, promotion, assignment of more working under ambiguous path-goal relation-
interesting tasks or opportunities for personal ships, his subjective probability estimates
growth and development. Consequently, he that his behavior will affect the valences he
influences the magnitude of the sum of the receives are less than the objective probabili-
personal outcomes available. Second, the ties that his behavior will affect the valences
leader, through his interaction with the sub- he receives. When this assumption does not
ordinate, can increase the subordinate's path hold, that is, when under conditions of role
instrumentality concerning the rewards forth- ambiguity, his subjective probability esti-
coming as a result of work-goal accomplish- mates exceed the objective probabilities, then

This content downloaded from 129.233.1.252 on Mon, 19 May 2014 08:09:20 AM

All use subject to JSTOR Terms and Conditions
 324 ADMINISTRATIVE SCIENCE QUARTERLY

clarification of path-goal relationships will faction of subordinates will result in in-

result in reduced motivation. creased performance to the extent that such
Stated less formally, the motivational func- satisfaction increases the net positive valence
tions of the leader consist of increasing per- associated with goal-directed effort.
sonal pay-offs to subordinates for work-goal
attainment, and making the path to these HYPOTHESES
pay-offs easier to travel by clarifying it, re-
ducing road blocks and pitfalls, and increas- From the above general propositions, sev-
ing the opportunities for personal satisfaction eral specific hypotheses concerning leader
en route. The function of making the path consideration, initiating structure, closeness
of supervision, hierarchical influence, and
easier and more satisfying to follow has been
dealt with only implicitly in the leadership authoritarianism can be derived. These hy-
literature and, as will be shown, has signifi- potheses are consistent with empirical find-
cant implications for leader's behavior. ings and thus illustrate how these prior find-
2. In increasing path instrumentality by ings could have been deduced from the
clarifying path-goal relationships, the lead- general propositions of the theory. These hy-
er's behavior will have positive motivational potheses do not constitute an exhaustive list
effects to the extent that it reduces role am- of relationships between the variables, but
biguity or makes possible the exercise of ex- rather, serve to illustrate how the general
ternally imposed controls. Reduction of role propositions can be operationalized.
ambiguity results in increased motivation 1. Leader initiating structure increases
because role ambiguity is both negatively the path instrumentality for subordinates
valent to subordinates (Rizzo et al., 1970), whose roles have nonroutine task demands
and because it is usually associated with low by decreasing role ambiguity (Rizzo et al.,
path instrumentality. Externally imposed 1970).
controls are motivational because they make 2. Informal leaders high in structure in-
possible the allocation of valences con- fluence positively the subjective probabilities
tingent on desirable behavior. Externally im- other group members assign to positively
posed control results in improved perfor- valent outcomes (Rim, 1965).
mance only to the extent that the rewards 3. Leader initiating structure and consid-
that are under the control of the leader are eration will have differential effects, depend-
positively valent to the subordinates; punish- ing on whether the task is satisfying or un-
ments that are under the control of the leader satisfying to the subordinate, and whether
are negatively valent to the subordinates; the task-role demands are clear or ambigu-
rewards and punishments are contingent on ous.

performance; and the contingency is clearly The more satisfying the task, the less posi-
perceived by the subordinates. Whether per- tive the relationship between consideration
formance motivated by external controls is and subordinate satisfaction and perfor-
satisfying to the subordinate, depends on his mance (Fleishman, 1971: 13-14). These cor-
unconscious needs, conscious values, and relations will vary from insignificant to
perceptions of equity in the exchange of ef- positive, depending on task satisfaction. For
fort for rewards. unsatisfying tasks, consideration will tend to
3. Where leader attempts to clarify path- offset dissatisfaction associated with the task;
goal relationships are redundant with existing for satisfying tasks, consideration will be less
conditions, that is, where path-goal relation- important.
ships are apparent because of the routine of The less satisfying the task the more nega-
the tasks or objective system-fixed controls, tive will be the relationship between struc-
attempts by the leader to clarify path-goal ture and satisfaction and the more positive
relationships will result in increased exter- will be the relationship between structure
nally imposed control and will be seen by and performance (Fleishman, 1971: 19-20).
subordinates as redundant. Although such For unsatisfying tasks, structure will be
control may increase performance, it will also viewed as an imposition of external control
result in decreased satisfaction. and, therefore dissatisfying, but will also be
4. Leader behavior directed at need satis- required to motivate subordinate effort to-

This content downloaded from 129.233.1.252 on Mon, 19 May 2014 08:09:20 AM

All use subject to JSTOR Terms and Conditions
 House: LEADER EFFECTIVENESS 325

ward goal achievement (Fleishman, 1971: fluence, consideration will be positively re-
19-20). lated to satisfaction and performance of
The more ambiguous the task the more subordinates (Pelz, 1952). Among indepen-
positive the relationship between leader ini- dent employees, or under leaders with low
tiating structure and subordinate satisfaction upward influence, consideration will have a
and performance. Structure serves to reduce lower positive relationship to subordinate
role ambiguity and clarify path-goal relation- satisfaction and performance (Wager, 1965;
ships for ambiguous tasks but is viewed as House et al., 1971a; 1971b). Leader influence
unnecessary and redundant for nonambigu- permits the leader to have more control over
ous tasks. rewards for subordinates and thereby per-
When task demands are self-evident due mits the leader to make subordinate valences
to a high degree of routinization, or where contingent on performance and to make the
roles are clearly defined by such factors as outcomes of work-goal attainment more val-
mechanization, legal constraints, contracts, ent or less valent.
professional ethics, or group norms, initiating 9. Under conditions of authoritarian or
structure will not result in role clarification punitive leadership, both leader initiating
and will be unsatisfying to subordinates. structure and leader hierarchical influence
4. Where the follower's tasks are varied will be negatively related to subordinate
and interdependent and where teamwork satisfaction. Under such conditions, both
norms are not developed within the group, structure and influence will be seen as bases
initiating structure and close supervision will of authoritarian power by subordinates.
regulate and clarify path-goal relationships.
RECONCILIATION OF PRIOR
Therefore, structure and close supervision
will result in increased coordination, satis- FINDINGS
faction, and performance (Patchen, 1962; The usefulness of the theory can be illus-
Mass, 1950; Hemphill, 1950; Fleishman, trated by showing how it can be applied to
1971: 26). reconcile what appear to be conflicting re-
5. Where tasks are interdependent, sults of prior research cited earlier.
varied, and ambiguous, consideration will Fleishman and Harris (1962) found that
result in social support, friendliness among when the leader's consideration toward sub-
group members, increased cohesiveness, and ordinates' needs was used as a moderating
team effort. These social outcomes will be variable-that is, when separate correlations
positively valent to the members and thus are computed for high-considerate, medium-
increase the net sum of the positive valences considerate, and low-considerate leaders-
associated with interdependent jobs requir- the relationship between leader structure and
ing cooperation and team spirit (Oaklander grievances and turnover varied. Specifically
and Fleishman, 1964; Fleishman, 1971: 13- they found that there was no relationship
14). between structure and grievances or turn-
6. Where tasks and/or environment are over for highly considerate leaders; but for
frustrating and stress inducing, considera- leaders with low consideration, the relation-
tion wlll result in increased social support ship was positive and significant.
for followers and thus reduce negative val- House et al. (1971b) attempted to conduct
ence associated with task-oriented behavior a replication of the findings of Fleishman
(Fiedler, 1967; Rim, 1965; Fleishman, 1971; and Harris (1962), using the same scales
12-13. for the predictor and moderator variables
7. Where stress is from sources external and eight measures of satisfaction with sub-
to the work unit and tasks are ambiguous, ordinate role expectations as criterion vari-
structure will result in increased ego pro- ables. The populations studied were salaried
tection, security, and satisfaction (Oaklander engineers, scientists, and technicians in three
and Fleishman, 1964). In this instance, struc- large research, design and development or-
ture serves as an umbrella which protects ganizations. Unexpectedly, the data not only
followers from externally imposed stress. failed to replicate Fleishman and Harris'
8. Among hierarchically dependent em- (1962) findings but suggested their opposite.
ployees under leaders with high upward in- Specifically it was found that leader initiating

This content downloaded from 129.233.1.252 on Mon, 19 May 2014 08:09:20 AM

All use subject to JSTOR Terms and Conditions
 326 ADMINISTRATIVE SCIENCE QUARTERLY
structure had a significant linear-unmoder- ductivity by preventing soldiering, work re-
ated-positive relationship to half the satis- striction, or slowdowns, it is also a source of
faction measures in two of the three com- dissatisfaction to employees.
panies, and significant positive correlations, Another hypothesis derived from path-goal
> .36, p < .05, with satisfaction with the theory explains the findings concerning the
company management in all three com- moderating effect of consideration in some
panies studied. Furthermore, in one com- studies and not in others. Where the path is
pany under conditions of high leader not viewed as satisfying, that is, for lower
consideration, leader initiating structure level jobs, it can be hypothesized that con-
had significant positive curvilinear rela- sideration serves as a source of extrinsic so-
tionships to measures of satisfaction with cial satisfaction and support to the employee,
job freedom, advancement, security, and thus making the path easier to travel. Conse-
family attitudes toward the respondent's job, quently, for Fleishman and Harris' (1962)
with the eta correlation ranging from .36 to blue-collar workers, leader consideration
.50. moderated the unsatisfying effects of leader
In a subsequent study of 192 nontechnical structure; whereas, for higher level jobs,
corporate office employees of a chemical where the path was intrinsically satisfying,
manufacturing company, House et al. (1970) the need for such support was lower and
failed to replicate the findings of Fleishman consequently consideration would be ex-
and Harris (1962) and showed significant pected to have little or no moderating effect
positive relationships between leader initiat- on the relationship between initiating struc-
ing structure and six measures of satisfaction. ture and satisfaction.
The following three hypotheses derived Similarly in the International Harvester
from the theory serve to reconcile these study reviewed by Fleishman (1971: 19-20),
findings. high leader initiating structure was found to
Leader initiating structure can be hypoth- be related to foremen ratings of proficiency,
esized to clarify path-goal relationships for but also higher grievances; high leader con-
higher occupational level jobs which are sideration was found to be related to lower
frequently ambiguously defined. Such clari- proficiency ratings, a tendency more pro-
fication reduces role ambiguity and increases nounced in production than in other de-
the employee's perceived instrumentality of partments. The specific variable that was
effort toward goal attainment. That is, it in- subsequently discovered to account for the
creases his subjective probability estimates differential relations across departments was
that his efforts will result in goal attainment. pressure for output. If it can be assumed
Thus, the path-goal theory advanced here that the tasks in the production departments
offers an explanation for the positive corre- were less satisfying, then it follows that under
lations between leader initiating structure conditions of high pressure for output, leader
and satisfaction among the high occupational initiating structure would be viewed as an
level groups studied by House et al. (1970, externally imposed form of control. Such
1971a, 1971b). The theory also explains the control would be more acceptable to higher
negative relationships found at lower occu- managers, but resented by the subordinates
pational levels by Fleishman and Harris on whom it was imposed. Leader considera-
(1962). If it can be assumed that lower level tion is more likely to serve as a stress reducer
jobs are generally more routine, that their as tasks become more unsatisfying and
path-goal relationships are usually self-evi- pressure for output increases. Thus the dif-
dent, and that the job itself is frequently not ferential relationship found across types of
intrinsically satisfying, then it can be hy- departments can be explained in terms of
pothesized that leader initiating structure differences in task satisfaction, that is, path
would be viewed by subordinates as an im- valence and pressure for production differ-
position of external control that does little to ences. This explanation is directly deducible
clarify path-goal relationships and is viewed from the path-goal theory presented here and
by subordinates as being directed at keeping again illustrates the ability of the theory to
them working at unsatisfying activities. Al- accommodate and explain otherwise con
though such control is likely to increase pro- fusing empiric findings.

This content downloaded from 129.233.1.252 on Mon, 19 May 2014 08:09:20 AM

All use subject to JSTOR Terms and Conditions
 House: LEADER EFFECTIVENESS 327

Three studies were conducted in which the larger research project from which these
eight hypotheses specifically derived from the data were taken.
theory were tested, and three of these hy- In the satisfaction and role-ambiguity
potheses were retested. scales, the respondents were asked to indi-
cate the degree to which descriptive state-
STUDY NUMBER 1 ments were true or false on a 7-point scale
Method ranging from 1, very false, to 7, very true.
One sample item for each of the satisfaction
The first study was based on an analysis of scales follows:
data collected as part of a larger ongoing
research project. The scale means, standard Satisfaction category
deviations, and intercorrelations were re- Advancement opportunity
ported by Rizzo et al. (1970). Job autonomy
Sample. The sample for the study consisted Intrinsic job rewards
of 199 office employees of a heavy equipment Job security
manufacturing company. All respondents Pay
were salaried and performed either profes- Recognition
Social environment
sional, administrative, or somewhat admin-
istrative-clerical tasks. The sample was Sample item
randomly selected and constituted 35 percent
Chances for advancing at a reasonable rate
of the total population of salaried employees The opportunity for independent thought
at the corporate offices and main plant of The feeling of worthwhile accomplishment at
the company, excluding first-level produc- work
tion foremen. Chances of keeping this job as long as I want it
Hypotheses. The following hypotheses The pay I receive for my work
were tested: The awareness that others have of my perfor-
Hypothesis 1. Leader initiating structure mance when I perform well
will be positively related to subordinate sat- The dignity with which I am treated
isfaction. The items in the role ambiguity scales
Hypothesis 2. Leader initiating structure were:
will be negatively related to subordinate role
ambiguity. I feel certain about how much authorityI have.
Hypothesis 3. The variance in role am- Clear, planned goals and objectives for my job.
biguity will account for the relationship I know that I have divided my time properly.
I know what my responsibilitiesare.
between leader initiating structure and sub- I know exactly what is expected of me.
ordinate satisfaction. Explanationis clear of what has to be done.
The rationale for these hypotheses is as
follows. Since the sample consisted primarily Scale validities. The leader initiating struc-
of employees doing quasi-professional or ture scale was shown by Halpin (1957) to
quasi-administrative work, it seemed likely have concurrent criteria validity and by
that much of their time would be devoted to Stogdill (1969) to have experimental criteria
ambiguously defined tasks, for which the validity. The role ambiguity measure has
path to goal achievement was not clear; been construct validated (Rizzo et al., 1970)
consequently, initiating structure by the against measures of organization practices,
leader would be expected to reduce am- leader behavior, and satisfaction. Although
biguity and therefore increase the clarity of not validated, the reliabilities of the satisfac-
path-goal relationships and increase satis- tion measure are moderately high and ade-
faction. quate for research purposes. The reliabilities
Measures. The scales used were the leader- for all measures are reported in Table 1.
initiating-structure scale taken from the Ohio Statistical procedure. Hypotheses 1 and 2
State Leader Behavior Description Ques- were tested by simple product-moment cor-
tionnaire (Stogdill, 1965a) and measures of relations; hypothesis 3 by comparing the
satisfaction and role ambiguity derived by zero-order correlations between structure
factor analysis and developed specifically for and satisfaction with partial correlations be-

This content downloaded from 129.233.1.252 on Mon, 19 May 2014 08:09:20 AM

All use subject to JSTOR Terms and Conditions
328 ADMINISTRATIVE SCIENCE QUARTERLY
TABLE 1. ZERO-ORDER AND PARTIAL CORRE- interpretation that the theory is not discon-
LATIONS BETWEEN LEADER INITIATING STRUC- firmed, but only weakly tested and sup-
TVRE AND SUBORDINATE SATISFACTION WITH ported. A more sensitive test of the same hy-
AND WITHOUT AMBIGUITY HELD potheses would require careful selection of
CONSTANT a sample of employees known to be per-
forming primarily ambiguous but also satis-
Correla- fying tasks.
tions
Zero- with role Re-
order ambiguity lia- STUDY NUMBER 2
correla- held bili-
Variables tions constant ties Method
Satisfactionwith The second study tested hypotheses about
Advancement the moderating effect of task characteristics
opportunity .20** .09 .83 on the relationships between leader initiat-
Job autonomy .10 -.05 .82 ing structure and consideration, and subordi-
Intrinsic
job rewards .17* .03 .92 nate satisfaction and performance.
Job security .03 -.06 .70 Sample. The sample was that used by
Pay .05 -.01 .86 House et al. (1970); however, two additional
Recognition .16* -.06 .85 measures were used to stratify the popula-
Social tion according to task autonomy and job
environment .21** .04 .78
Role ambiguity -.41** .78
scope, thus permitting tests of specific hy-
Initiating
potheses derived from the theory.
structure .65 Hypotheses. The following hypotheses
N = 199 were tested:
Hypothesis 1. Job autonomy will have a
* p < .05.
positive moderating effect on the relation-
p < .01. ship between leader initiating structure and
tween the same variables, with role ambigu- subordinate job satisfaction; that is the re-
ity held constant. lationship will be stronger under high job
autonomy than under low job autonomy.
Findings This hypothesis was based on the expecta-
The results of the analysis, reported in Ta- tion that for autonomous tasks the role de-
ble 1, indicate support of all the hypotheses. mands are likely to be more ambiguous than
Four of the relationships between initiating for nonautonomous tasks. Thus, under auto-
structure and satisfaction are significant, as nomous tasks, leader structure will serve to
is the relationship between structure and role reduce role ambiguity, clarify path-goal re-
ambiguity. It can also be seen that when the lationships, and thereby increase satisfac-
effect of role ambiguity is partialled out, the tion.
four significant correlations are no longer Hypothesis 2. Job autonomy will have a
significant. negative moderating effect on the relation-
The zero-order correlations, while signifi- ship between leader initiating structure and
cant and in the hypothesized direction, are subordinate performance; that is, for non-
very low, but not inconsistent with the theo- autonomous jobs the relationship will be
retical interpretation, since a significant pro- stronger than for autonomous jobs.
portion of the sample were probably doing This hypothesis is based on the expecta-
relatively routine work and for which the tion that for nonautonomous tasks, role de-
path-goal relationships were relatively clear mands are more likely to be dissatisfying, so
regardless of the amount of structure initi- that initiating structure will control behavior
ated by the leader. and ensure higher performance. For sub-
ordinates with autonomous jobs, initiating
Conclusion structure will be irrelevant to performance,
Although the above findings support all since the subordinate is less dependent on
three hypotheses, inability to control for the the leader for resources, assistance, or guid-
characteristics of the sample studied and the ance.
magnitude of the correlations require the Hypothesis 3. Job autonomy will have a

This content downloaded from 129.233.1.252 on Mon, 19 May 2014 08:09:20 AM

All use subject to JSTOR Terms and Conditions
 House: LEADER EFFECTIVENESS 329
negative moderating effect on the relation- This hypothesis is based on the assump-
ship between leader consideration and sub- tion that a wide variety of tasks is more
ordinate satisfaction; that is, the relationship likely to be satisfying and, therefore, subor-
will be weaker for autonomous jobs than for dinates with varied tasks have less need for
nonautonomous jobs. social support, that is, consideration from
Hypothesis 4. Job autonomy will have a their leaders; whereas for highly routine
negative moderating effect on the relation- jobs, leader consideration is a source of
ship between leader consideration and sub- support to the employee, thus making the
ordinate performance. path easier to travel.
Hypotheses 3 and 4 are based on the as- Measures. The measures used were the
sumption that under high job autonomy, the leader-consideration and initiating-structure
role demands are more likely to be satisfy-
scales taken from the Ohio State question-
ing and, consequently, leader behavior will
be less relevant to the needs or performance naire, the Ohio State job-description scale
of subordinates than under low job auton- and job-expectation scales (Stogdill, 1965b),
omy. multimethod-multitrait ratings of individual
Hypothesis 5. Job scope will have a neg- performance, a job-scope and a task-auton-
ative moderating effect on the relationship omy scale.
between leader consideration and subordi- The job-scope scale developed by Wigdor
nate satisfaction and performance; that is, (1969), consisted of five questions designed
the wider the variety of tasks performed by to measure the extent to which the subject
subordinates, the weaker the correlations performs various tasks, sees projects through
between leader consideration and subordi- to completion, and determines job objectives
nate satisfaction and performance. and methods. The questions were:
How often are you required to perform tasks which previously had not been part of your job
responsibility?
5. very often 4. often 3. sometimes 2. occasionally 1. rarely
How often do you see projects through to completion?
1. rarely 2. occasionally 3. sometimes 4. often 5. very often
To what extent do you set objectives, goals and procedures for your job rather than following di-
rections or established procedures?
5. very large 4. large 3. somewhat 2. little 1. almost never
To what extent do you participate in decisions concerning the methods to be used in performing
your job?
1. almost never 2. occasionally 3. frequently 4. usually 5. almost always
To what extent are you able to allocate a portion of your time to tasks related to corporateobjec-
tives but not specificallyassigned to you?
S. very large 4. large 3. sometimes 2. little 1. almost never
The task-autonomy scale also developed for financial resources, nonfinancial resources,
by Wigdor (1969) consisted of 16 items in- and directions; independently scheduled and
tended to reflect the degree to which the planned his activities; and innovated inde-
respondent was able to perform his job with- pendently of others. Three sample questions
out depending upon his superior or others and response choices follow:
To what extent are you able to act independently of your superiorin performingyour job function?
1. hardly ever 2. seldom 3. occasionally 4. frequently 5. almost always
When someone else in the company requests you to perform a task for them, how frequently do
you seek advice from your superiors?
1. almost always 2. very often 3. often 4. occasionally 5. rarely
To what extent are you able to schedule and plan your task requirementsindependent of others in
the organization?
1. hardly ever 2. seldom 3. occasionally 4. frequently 5. almost always

This content downloaded from 129.233.1.252 on Mon, 19 May 2014 08:09:20 AM

All use subject to JSTOR Terms and Conditions
330 ADMINISTRATIVE SCIENCE QUARTERLY
The job-description questionnaire essen- sideration and leader initiating structure
tially measured employees' attitudes toward scales, corrected with the Spearman-Brown
the company and its management. Factor prophesy formula, were .86 and .90 respec-
analysis by Stogdill (1965b) of a previous tively.
satisfaction scale yielded a single factor with The job-autonomy scale was validated by
high loadings for the 12 items of the job-de- Wigdor (1969) on two samples assumed to
scription questionnaire, which were related be significantly different on job autonomy
to satisfaction with the company, manage- because of their occupational level. Wigdor
ment, and with recognition. (1969) found that the scale discriminated,
The job-expectation questionnaire mea- as predicted, between a sample of 33 cor-
sured employee satisfaction with role expec- porate vice presidents and a sample of white-
tation with respect to work, advancement, collar trainees and secretaries. The difference
the prestige of the respondent's job as com- in the distribution of the two samples were
pared with the jobs of others, pay, freedom, in the predicted direction and significant at
family attitudes toward the respondent's job, the .0001 level, Mann-Whitney U = 801,
and job security. Respondents were asked Z = 6.10. A similar result was found when
to indicate the degree to which each charac- 18 college professors with tenure were used
teristic of the job met their expectations. in place of the vice presidents, p < .0001;
Each class of expectation was measured by U = 421; Z = 4.97. This scale had a .90
averaging the scores of the questions. reliability for the present sample.
In the present study the items pertaining The job-scope scale was also validated by
to advancement, pay, and prestige expecta- Wigdor (1969) who compared 30 employees
tions were combined to provide a single mea- at a high organizational level making a wide
sure-extrinsic job satisfaction. variety of decisions and performing a large
The performance scale used for this study number of functions, with 30 employees who
was a multitrait-multirater scale. Lawler were low on these characteristics, and found
(1968: 370), in a review of methods for a difference in responses significant at the
measuring performance, indicated that the .0001 level, Mann-Whitney U = 861, Z =
multitrait-multirater method provides many 6.08, in the hypothesized direction. This scale
of the advantages of both the more objective has a .70 reliability for the present sample.
and more subjective measures, and added: The satisfaction scales were based on fac-
"Moreover, with this approach it is possible tor analyses reported by Stogdill (1965b).
to assess the criterion by determining its con- Although no validity data are available, the
vergent and discriminant validity, and it is reliabilities of these scales are moderately
not necessary to depend on an objective in- high and adequate for research purposes.
dicator such as sales or profits that may miss Reliabilities of all multiple-item scales are
the essence of the job." given in Table 2.
The index of performance was the average The performance measures were tested for
of the ratings given the subject when rated convergent and discriminant validity accord-
by one peer and one superior. The obvious ing to the criteria suggested by Campbell
constraint in the choice of the raters was that and Fiske (1959). Of the following criteria,
they had to be familiar with the aspects of the first shows convergent validity and the
the individual's performance that they were rest discriminant validity.
to rate, otherwise, the ratings tend to be af-
fected by the halo tendency (Bescoe and 1. There should be high and significant
Lawshe, 1959). To ensure the choice of interrater agreement, that is, correlation,
knowledgeable raters, the subject was asked among measures of the same traits.
to indicate the names of his superiors and 2. The correlations among two ratings
two peers who knew his performance best. of a single trait should be greater than the
The choice of which superior and which peer correlation between any one rating of that
ratings was used was made randomly by the trait and any other trait rated by the other
investigator. rater.
Scale validities and reliabilities. Kuder 3. The correlation between two ratings
Richardson reliabilities for the leader con- of a single trait should be greater than the

This content downloaded from 129.233.1.252 on Mon, 19 May 2014 08:09:20 AM

All use subject to JSTOR Terms and Conditions
 House: LEADER EFFECTIVENESS 331

TABLE 2. CORRELATIONS BETWEEN LEADER INITIATING STRUCTURE AND SUBORDINATE SATISFAC-

TION AND PERFORMANCE MODERATED BY JOB AUTONOMY

Study Number 2
Job autonomy Differences in
correlations-
Low Medium High low versus
Variables (N = 61) (N = 68) (N = 62) high group Reliabilities
Satisfactionwith
Company and management 3544 4504 4744 .90
Intrinsic job rewards 30 16 20 .64
Extrinsic job rewards 11 26* 5144 .01 .72
Family attitude
toward job -9 14 15* .72
Security 14 17 25* .61
Job freedom 3744 7 3644 .70
Average satisfaction 19 21 33
Performance ratings
Quality 33** 22 6
Ability 34** 8 12
Effort 43** 20 20
Initiative 4644 13 27*
Ability without
guidance 36** 27* 23
Quantity 55** 17 18 .05
Average performance 47 18 18
p .05.
*P
pV .01.
V

correlation between any one rating of that vergent validity and the third test of dis-
trait and any other trait rated by the same criminant validity. Four of the six ratings
rater. meet the first requirement of discriminant
4. There should be the same pattern of validity and two were marginally significant
trait interrelationships in all the multitrait (p = .055), but none of the ratings meet the
triangles of the multirater-multitrait correla- second. Sign tests were used for the first two
tion matrix. tests of discriminant validity, and the W
The validity tests, using Spearman rank- coefficient of concordance for the third.
order correlations, are summarized in Ta- These tests were first used for this purpose
ble 3. by Evans (1969) and are explained by Siegel
The sample used to test the validity of the (1956). There is also indicated a high level
performance ratings consisted of a subsam- of halo. The W test of concordance indicated
ple of 172 subjects of the 192 used to test that this halo was common among all four
the hypotheses. The subsample of 172 con- multitrait triangles of the multitrait-multi-
sists of all subjects for whom self-, superior, rater matrix, thus indicating not only inter-
and peer ratings were obtained. As can be rater agreement on individual scales but
seen from Table 3, for study number 2, the also a similarity of rater response sets, that
superior-peer correlations indicated substan- is, halo. These findings suggest that any tests
tially higher convergent and discriminant of the hypotheses based on these perfor-
validity than the self-superior or the self- mance scales are susceptible to a type one
peer correlations. For this reason the self- error; that is, an error in which data might
ratings were discarded and the superior and result in rejecting a true hypothesis because
peer ratings for each subject were averaged the measure of the dependent variable, per-
to arrive at a measure of performance for formance, included a large halo and had only
this study. moderate-to-low discriminant validity. Thus,
All ratings meet the requirement of con- use of the average of peer-superior ratings

This content downloaded from 129.233.1.252 on Mon, 19 May 2014 08:09:20 AM

All use subject to JSTOR Terms and Conditions
332 ADMINISTRATIVE SCIENCE QUARTERLY

TABLE 3. SUMMARY OF TESTS OF CONVERGENT AND DISCRIMINANT VALIDITY FOR

MEASURES OF PERFORMANCE

Study Number 2, N = 172

Ratings 1 2 3 4 5 6
Peer-superiorratings
Convergentvalidity, p = .01 .01 .01 .01 .01 .01
Discriminant validity
Test 1, p = .001 .055 .001 .001 .055 .001
Test 2, p = N.S. N.S. N.S. N.S. N.S. N.S.
Test 3, overall test, w = .8Q
X2 =44.8
p .001
Peer-self ratings
Convergent validity, p .025 N.S. .01 .05 N.S. .05
Discriminant validity
Test 1, p = .001 N.S. .055 N.S. N.S. N.S.
Test 2, p = N.S. N.S. N.S. N.S. N.S. N.S.
Test 3, overall test, w = .55
X2= 30.8
p ? .01
Self-superior ratings
Convergent validity, p = .Q1 .05 .01 .025 .10 .025
Discriminant validity
Test 1, p = .001 N.S. .055 N.S. N.S. N.S.
Test 2, p = N.S. N.S. N.S. N.S. N.S. N.S.
Test 3, overall test, w = .56
X2= 31.4
p .01

Study Number 3, N = 122

Ratings 1 2 3 4 5 6
Peer-superiorratings
Convergent validity, p= .01 .01 .01 .01 .01 .01
Discriminant validity,
Test 1, p = N.S. N.S. N.S. N.S. N.S. N.S.
Test 2, p = .055 .828 .001 .01 .001 .172
Test 3, overall test, w .61
X2= 34.1
p .01

constitute conservative tests of the hypoth- Findings

eses. Table 2, showing the findings relevant to
Statistical procedure. The sample was di- hypothesis 1, indicates that the average cor-
vided into respondents with low-, medium-, relation and four of the individual correla-
and high-task autonomy. The medium group tions between initiating structure and satis-
consisted of the third closest to the mean of faction increase monotonically with increases
the total population and included all tied in job autonomy, and that the correlation
subjects at the borderlines. Correlations were between extrinsic job satisfaction and initiat-
computed for the relationship between leader ing structure is significantly higher for groups
consideration and subordinate satisfaction with high autonomy than for groups with
and performance. The same procedure was low or medium autonomy. These findings,
repeated using job scope as the moderating while in the predicted direction for four of
variable. the classes of satisfaction and for the average

This content downloaded from 129.233.1.252 on Mon, 19 May 2014 08:09:20 AM

All use subject to JSTOR Terms and Conditions
 House: LEADER EFFECTIVENESS 333

correlations, are only clearly and strongly between leader consideration and perfor-
supportive, that is, differences in correlations mance is rather profound. The average cor-
are large and significant for hypothesis 1, relation decreases from .52 for the low job-
with respect to extrinsic satisfaction. scope group to .02 and .09 for the medium
Table 2 also presents the findings relevant and high job-scope groups respectively. Five
to hypothesis 2. Again, the average of the of the six correlations between leader con-
correlations between leader initiating struc- sideration and performance in the low job-
ture and subordinate performance decrease scope group are significantly higher than
in the predicted direction from .47 for the their respective correlations in the medium
low-autonomy group to .18 for the medium- or high job-scope groups. Finally, all of the
and high-autonomy groups. Three of the correlations in the low job-scope group are
correlations decrease monotonically in the significant while none in either the medium
predicted direction and all 6 of the correla- or high job-scope groups are.
tions in the low-autonomy group are signifi-
cant at the .01 level, whereas only one in the Conclusion
high-autonomy group is significant. Finally, Although the tests were conservative be-
the correlation between initiating structure cause of measurement error in the perfor-
and quantity of subordinate performance is mance ratings, that is halo, the findings from
significantly higher for the low-autonomy this study provide rather strong support for
group than for either the medium- or high- the hypotheses derived from the theory.
autonomy group.
Table 4 presents the findings relevant to STUDY NUMBER 3
hypotheses 3 and 4. For hypothesis 3, while
Method
none of the correlations between leader con-
sideration and subordinate satisfaction in the This study consisted of replications of hy-
group with low job autonomy are signifi- potheses 3, 4 and 5 of study number 2. The
cantly higher than those in the groups with data for this study were collected as part of
medium or high autonomy, both the average a larger study. Unfortunately, the leader-
correlations and three of the individual cor- initiating-structure scale was not included
relations decrease monotonically in the pre- in the larger study, thus precluding the pos-
dicted direction. sibility of replicating hypotheses 6 and 7.
For hypothesis 4, four of the correlations Sample. The sample for this study con-
between leader consideration and subordi- sisted of 122 employees, that is, 97 percent
nate performance are significantly higher for of the employees of a chemical manufactur-
the low-autonomy group than for the med- ing plant. Those not included were absent on
ium- or the high-autonomy group. All six the day the questionnaire was administered,
correlations in the low-autonomy group are so that the final sample consisted of 13 man-
significant at the .01 level, while none are agers, 8 technicians, 2 secretaries, and 99
significant in the medium- or high-autonomy hourly production workers. The sample was
groups, and the average of the correlations not divided by occupational level because
decreases monotonically from .42 to .11 to this would have restricted the vaiiance in
.08 with increases in job autonomy. task autonomy and job scope and would have
Table 4 also presents the findings relevant precluded a test of the hypotheses.
to hypothesis 5. The differences in the corre- Scale validity and reliability. The scales
lations between leader consideration and used in this study are the same as those used
subordinate satisfaction are neither signifi- in study number 2, with the exclusion of the
cant nor consistently in the hypothesized initiating-structure scale.
direction. Evidently, consideration has a per- The performance measures were based on
vasive effect and when not controlled for multitrait-multirater scales completed by the
subordinate autonomy the relationship be- subject's peers and immediate superiors, with
tween leader consideration and subordinate tests of discriminant and convergent validity
satisfaction remains constant, regardless of based on those subjects for whom peer and
subordinate job scope. However, the mod- superior ratings were available. The same
erating effect of job scope on the relationship statistical tests were applied to these data

This content downloaded from 129.233.1.252 on Mon, 19 May 2014 08:09:20 AM

All use subject to JSTOR Terms and Conditions
334 ADMINISTRATIVE SCIENCE QUARTERLY

~~~~i1~ ~ ~ 1

C~~~~~~~~~~~-
z E-iq
IIo cD oo

u m t 11 cO t ~ rcscq O cs O s s O t dq

o~~~~~~~t

C S jC 1 U t _ E
cjd

~cJ

gDo
%i1~ : t vo z t .co o
~i0

r-.CIr\%t
OsCO N~ 0 t
E~~~~~~~~~~~~~~~~~~r-

~?-~ S bJJ ~ ~ E~ Uo~ ~ -cio~ i~ o~~~~b

ci'o G E
?~~~~~~~~~~C E Q0 Io
1 It oo Q0 ,,,
O I- (x D
0 CIO r--

H ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~;
O~~~~~~~~~~~~~~1 11
CI c Io Ir c
ol co10 o Itc
O
10

E
0
t
~~~~~~~
C
|; E . Co
| E 2 - C mvcK
1> 0 Q W t X 3 X E;

This content downloaded from 129.233.1.252 on Mon, 19 May 2014 08:09:20 AM

All use subject to JSTOR Terms and Conditions
 House: LEADER EFFECTIVENESS 335

as were applied to the performance data in significantly higher than their correlations
the second study. As shown in Table 3, the with the groups with high job scope.
results of these tests are very similar to those
Conclusion
of Lawler (1966) and the results of the sec-
ond study. Specifically, all of the peer-supe- The findings about the moderating effect
rior correlations met the requirement of con- of job scope replicated the findings of the
vergent validity. Four of the correlations met second study and thus provide rather strong
the second test of discriminant validity and support for the hypothesis. However, the
the overall pattern of concordance was sig- findings about the moderating effect of job
nificant at the .01 level. As in the second autonomy only partially replicated those of
study, the ratings had a large amount of halo, the second study, possibly because the popu-
which accounted for the failure to pass the lation for the third study consisted primarily
first test of discriminant validity. The relia- of blue-collar workers who were probably
bilities for the job-scope, job-autonomy, and quite dependent on their superiors, even
consideration scales were .64, .88, and .87 when they scored in the upper third of the
respectively. job-autonomy scale. The job-autonomy scale
Statistical procedure. The sample was di- means for the sample in the second study
vided into approximately equal thirds on the were 2.9, 3.4, and 4.0 for the low, medium,
basis of scores on job scope and job auton- and high groups and 2.6, 3.3, and 3.8, respec-
omy, using the same procedure as described tively, for the third study. The finding that
in study number 2. The medium group was consideration has a significant and positive
slightly larger because boarderline ties were relationship to satisfaction for groups with
placed in this group. For each group, cor- both low and high job autonomy is consis-
relations were computed for the relation- tent with the interpretation that all respon-
ships between leader consideration and sub- dents were probably dependent on their su-
ordinate satisfaction and performance. periors for satisfaction, regardless of their
response to the job-autonomy scale. Thus,
Findings this study provides moderate support for the
hypothesis with respect to the influence of
From Table 5 it can be seen that the cor- job scope and ambiguous results with respect
relations between consideration and satis- to the influence of job autonomy.
faction with intrinsic job rewards decreases
from .61 to .24 to .10 for those in jobs with CONCLUSION
low, medium, and high autonomy, and that The theory advanced here has been shown
the correlation for the low-autonomy group to reconcile apparently conflicting findings
is significantly less than for the high-auton- from previous research. It has also been
omy group. Other differences between shown to provide an integrated explanation
groups appear to be negligible. Thus, hy- of the results of findings about authoritarian-
potheses 3 is replicated only with respect to ism in leader hierarchical influence, closeness
intrinsic job satisfaction, and hypothesis 4 of supervision, initiating structure, and con-
is not supported. sideration.
Table 5 also presents the data relevant to The theory was tested by correlational tests
hypothesis 5. The average correlations be- of 8 hypotheses derived from general propo-
tween consideration and satisfaction de- sitions. The tests are somewhat weak in that
crease from .38 to .24 to .18 for the groups the theoretical constructs, such as intrinsic
with low, medium and high job scope. Two task satisfaction and ambiguity of task-role
of the correlations for low job scope are demands, were inferred from situational mea-
significantly higher than their respective cor- sures of task autonomy and job scope and
relations for the groups with medium or high from occupational characteristics of the pop-
job scope. ulations studied. These inferences make the
The average correlations between leader tests susceptible to the error of rejecting a
consideration and performance decrease valid hypothesis, so that the tests are con-
from .33 to .18 to -.04 as predicted. Two servative ones. A further limitation is inher-
of the correlations with low job scope are ent in cross-sectional survey research, which

This content downloaded from 129.233.1.252 on Mon, 19 May 2014 08:09:20 AM

All use subject to JSTOR Terms and Conditions
336 ADMINISTRATIVE SCIENCE QUARTERLY

-4 It tt m O

C~~~~~~~~~~~~C

0~~~~~~~~
CR r
?~~~~~~~~~~~~o Qq ll

>1 O~ZO 0

"rCo

1 0 00 tC m 00 m co r-- CO 00C,-rCO
?n coCt0 0t co co ) C 0
D10CO.

C ~ coc10co
0~~~~~~~~~ L-- ?-qC )CO1 c
--qC cq cq q o--

s~~~~~oull~~~~~~~o I I

pqZ

0o

tw 11
- 1 10 r- rc
C- c ot t cq co ot ~ co co

c~n .t .> . ~. . . . . . .
pq
Z I
C, 0000C) -O
c I
x m I

C~~~~~~~C
Z 8q C)o r-- o
CY) CY) 10o r rC
co

0 CY) CY)~~~~~~~~~~~~~~~~~~~~~~~0
X W II "t "t tD t- oc 0x qc C)

i! ~ ~~~~~ lr :lrc
~~~~ :
0 c)
.s
Oc

: c s co c C oQ

This content downloaded from 129.233.1.252 on Mon, 19 May 2014 08:09:20 AM

All use subject to JSTOR Terms and Conditions
 House: LEADER EFFECTIVENESS 337
can rule out invalid hypotheses, but cannot Atkinson (ed.), Motives in Fantasy,
establish causal relationships among the vari- Action and Society. New York: Van
ables. Nostrand.
The findings, when viewed collectively, Bescoe, Robert O., and C. H. Lawshe
1959 "Foreman leadership as perceived by
generally support the theory. Among high-
superiors and subordinates." Personnel
occupational groups, leader initiating struc- Psychology, 12: 573-582.
ture was generally positively related to sub- Campbell, Donald T., and Donald W. Fiske
ordinate satisfaction and performance. This 1959 "Convergent and discriminant valida-
relationship was accounted for in terms of tion by the multitrait, multimethod
variance in subordinate role ambiguity, matrix." Psychological Bulletin, 56:
which was shown to have a negative cor- 81-105.
relation with initiating structure. The rela- Evans, Martin G.
tionships between leader structure and 1968 The Effects of Supervisory Behavior
subordinate role ambiguity, and satisfac- upon Worker Perception of Their
Path-Goal Relationships. Doctoral dis-
tion, although significant and in the theo-
sertation, Yale University.
retically predicted direction, were quite 1969 "Convergent and discriminant validi-
low, probably because it was not pos- ites of the Cornell job descriptive in-
sible to control for contaminating variables dex and a measure of goal attain-
that would be expected to suppress these ment." Journal of Applied Psychology,
relationships. The relationships between ini- 55: 102-106.
tiating structure and consideration, and sub- Fiedler, Fred E.
ordinate satisfaction and performance varied 1967 A Theory of Leadership Effectiveness.
significantly and widely in the directions pre- New York: McGraw-Hill.
dicted when moderated by job scope. When Filley, Alan C., and Robert J. House
1969 Managerial Process and Organizational
moderated by task autonomy the theoretical
Behavior. Glenview, Ill.: Scott Fores-
predictions were supported by one sample man.
and not supported by another which raised Fleishman, Edwin A.
a question about the appropriateness of task 1971 "Twenty years of consideration and
autonomy as an indicator of ambiguity of structure." In Symposium on Con-
task-role demands and satisfaction among temporary Development in the Study
blue-collar workers as well as a question of Leadership. Carbondale: Southern
about the validity of the general proposition Illinois University, in press.
from which the hypothesis was derived. Fleishman, Edwin A., and Edwin F. Harris
On balance, the ability of the theory to re- 1962 "Patterns of leadership behavior re-
concile and integrate earlier findings, to- lated to employee grievances and
turnover." Personnel Psychology, 15:
gether with moderate-to-strong support, for
43-56.
seven of the eight hypotheses tested, two of Galbraith, Jay, and Larry L. Cummings
which were replicated in a second study, 1967 "An empirical investigation of the
suggests that the theory shows promise and motivational determinants of past per-
warrants further testing with more direct formance: interactive effects between
measurement of the theoretical constructs instrumentality, valence, motivation
using experimental as well as correlational and ability." Organizational Behavior
methods. and Human Performance 2: 237-257.
Georgopoulous, Basil S., Gerald M. Mahoney,
and Nyle W. Jones, Jr.
Robert J. House is a professor of manage- 1957 "A path-goal approach to productiv-
ity." Journal of Applied Psychology,
ment at the Bernard M. Baruch College of
41: 345-353.
the City University of New York. Graen, George
1969 "Instrumental theory of work motiva-
REFERENCES
tion: some empirical results and sug-
Atkinson, J. W. gested modifications." Journal of Ap-
1958 "Towards experimental analysis of hu- plied Psychology, 53: 1-25.
man motivation in terms of motives, Halpin, Andrew W.
expectancies, and incentives." In J. W. 1957 "The leader behavior and effective-

This content downloaded from 129.233.1.252 on Mon, 19 May 2014 08:09:20 AM

All use subject to JSTOR Terms and Conditions
 338 ADMINISTRATIVE SCIENCE QUARTERLY
ness of aircraft commanders." In Ralph Oaklander, Harold, and Edwin A. Fleishman
H. Stogdill and A. E. Coons (eds.), 1964 "Patterns of leadership related to or-
Leader Behavior: Its Description and ganizational stress in hospital settings."
Measurement. Ohio State University, Administrative Science Quarterly, 8:
Bureau of Business Research. 520-532.
Hemphill, John K. Patchen, Marvin
1950 "Relations between the size of the 1962 "Supervisory methods and group per-
group and the behavior of superior formance norms." Administrative Sci-
leaders." Journal of Abnormal and ence Quarterly, 7: 275-294.
Social Psychology, 32: 11-12. Pelz, Donald C.
House, Robert J., Alan C. Filley, and Domo N. 1952 "Influence: a key to effective leader-
Gujarati ship in the first-line supervisor." Per-
1971a "Leadership style, hierarchical influ- sonnel 29: 209-221.
ence and the satisfaction of subordi- Porter, Lyman, and Edward E. Lawler, III
nate role expectations: a test of Li- 1967 Managerial Attitudes and Perfor-
kert's influence proposition." Journal mance. Homewood, Ill.: Irwin Dorsey.
of Applied Psychology, in press. Rim, Y.
House, Robert J., Alan C. Filley, and Steven 1965 "Leadership attitudes and decisions
Kerr involving risk." Personnel Psychology,
1971b "Relation of leader consideration and 18: 423-430.
initiating structure to R and D sub- Rizzo, John R., Robert J. House, and Sidney
ordinate satisfaction." Administrative E. Lirtzman
Science Quarterly, 16: 19-30. 1970 "Role conflict and ambiguity in com-
House, Robert J., Lawrence A. Wigdor, and plex organizations." Administrative
Kenneth Shulz Science Quarterly, 15: 150-153.
1970 "Leader behavior, psychological par- Siegel, Sidney
ticipation, employee satisfaction and 1956 Nonparametric Statistics. New York:
performance: an extension of prior in- McGraw-Hill.
vestigations and a motivation theory Stogdill, Ralph M.
interpretation." In W. M. Frey (ed.), 1965 Manual for the Leader Behavior De-
Proceedings Seventh Annual Confer- scription Questionnaire. Columbus:
ence of Eastern Academy of Manage- Ohio State University of Business Re-
ment: 179-195. Amherst: University search.
of Massachusetts. Vroom, Victor H.
Korman, Abraham K. 1964 Work and Motivation. New York:
1966 "Consideration, initiating structure Wiley.
and organizational criteria-a review." Vroom, Victor, and Floyd Mann
Personnel Psychology, 19: 349-361. 1960 "Leader authoritarianism and em-
Lawler, Edward E. III ployee attitudes." Personnel Psychol-
1966 "Ability as a moderator of the rela- ogy, 13: 125-139.
tionship between job attitudes and job Wager, Wesley L.
performance." Personnel Psychology, 1965 "Leadership style, influence and su-
19: 153-164. pervisory role obligations." Adminis-
1968 "A correlation-causal analysis of the trative Science Quarterly, 9: 391-420.
relationship between expectancy atti- Wigdor, Larry
tudes and job performance." Journal 1969 "Effectiveness of Various Management
of Applied Psychology, 52: 462-468. and Organization Characteristics on
Mass, H. S. Employee Satisfaction and Perfor-
1950 "Personal and group factors in leader's mance as a Function of Employees
social perception." Journal of Ab- Need for Independence. Doctoral dis-
normal and Social Psychology, 45: 54- sertation, Bernard M. Baruch College,
63. City University of New York.

This content downloaded from 129.233.1.252 on Mon, 19 May 2014 08:09:20 AM

All use subject to JSTOR Terms and Conditions
 I :I

|~~~ U;: ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

*~~~~~~~~~~~~~A A-

- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

-~~~~~~~~~~~~~i

The Pink Dismissal Slip by Philip Evergood

Courtesy of the Andrew Dickson White Museum of Art, Cornell University

339

This content downloaded from 129.233.1.252 on Mon, 19 May 2014 08:09:20 AM

All use subject to JSTOR Terms and Conditions