You are on page 1of 5

238 BRIEF ARTICLES

(Ed.), Advances in experimental social psychology. PAVAGEORGIS, D., & McGuiRE, W. J. The generality
New York: Academic Press, 1964. Pp. 191-229. of immunity to persuasion produced by pre-ex-
McGuiRE, W. J., & PAPAGF.ORGIS, D. The relative posure to weakened counterarguments. Journal oj
efficacy of various types of prior belief-defense in Abnormal and Social Psychology, 1961, 62, 475-
producing immunity against persuasion. Journal of 481.
Abnormal and Social Psychology, 1961, 62, 327- ROSENBERG, M. J., & ABELSON, R. P. An analysis of
337. cognitive balancing. In M. J. Rosenberg et al.,
OSGOOD, C. E. Cognitive dynamics in the conduct of
Altitude organization and change. New Haven:
human affairs. Piiblic Opinion Quarterly, 1960, 24,
341-361 Yale Univer. Press, 1960. Pp. 112-163.
OSGOOD, C. E., Suci, G. J., & TANNENBAUM, P. H. The TANNHNBAUM, P. H. Initial attitude toward source
measurement of meaning. Urbana: Univer. Illinois and concept as factors in attitude change through
Press, 1957. communication. Public Opinion Quarterly, 1956,
OSGOOD, C. E., & TANNKNBAUM, P. H. The principle 20, 413-425.
of congruity in the prediction of altitude change.
Psychological Review, 19;5, 62, 42-55. (Received August 4, 1964)

Journal of Personality and Social Psychology


1966, Vol. 3, No. 2, 238-242

EFFECTS OF LEADERSHIP STYLE UPON GROUP PERFORMANCE


AS A FUNCTION OF TASK STRUCTURE 1
2
MARVIN E. SHAW AND J. MICHAEL BLUM
University oj Florida

Fiedler's contingency model holds that directive leadership is more effective


when the group-task situation is either highly favorable or highly unfavorable
for the leader, whereas nondireclive leadership is more effective in the inter-
mediate ranges of favorabilily. An experiment was conducted to test the
generality of this hypothesis. 5-person groups attempted 3 tasks under either
directive or nondirective leadership. Leadership behavior was manipulated by
instructions. The 3 tasks were selected to vary along the solution multiplicity
dimension, hence presumed to reflect different levels of favorability for the
leader. The results indicated that the directive leader was more effective than the
nondirective leader only when the group-task situation was highly favorable
for the leader, thus only partially supporting the hypothesis.

Studies of the effects of leadership style tion for the leader, where favorableness refers
(autocratic versus democratic, directive versus to the degree to which the group environment
nondirective, etc.) upon group effectiveness have makes it easy or difficult for the leader to in-
yielded ambiguous and often contradictory re- fluence group members. When the group-task
sults (Anderson & Fiedler, 1964; Fiedler, 1958; situation is either highly favorable or highly
Lewin, Lippitt, & White, 1939; Preston & Heintz, unfavorable for the leader, controlling, managing,
1949; Shaw, 1955). A theory proposed recently directive leadership behavior is most effective,
by Fiedler (1964) suggests a possible explanation whereas permissive, considerate, nondirective
of these inconsistent findings. His basic thesis leadership is needed for moderately unfavorable
is that the type of leadership behavior required group-task situations.
for effective group performance is contingent According to this theory, the favorableness
upon the favorableness of the group-task situa- of the group-task situation is determined by three
1 dimensions: the affective relation between the
This research was supported by the Office of leader and his members, the power inherent in
Naval Research, Contract NR 170-266, Nonr-
the leadership position, and the degree to which
580(11). Reproduction in whole or in part is per-
mitted for any purpose of the United Slates the task is structured. Although it is recognized
Government. that the interaction of these dimensions is com-
2 plicated, Fiedler suggests that the leaders rela-
Now with the American Institute for Research,
Pittsburgh, Pennsylvania. tions with his members is the most important,
BRIEF ARTICLES 239
the structure of the task is next most important, nondirective) i and three degrees of group-task
and inherent power of the leadership position is favorability (high, moderate, and low).
least important for the favorableness continuum.
Once measures of the three dimensions are avail- Subjects
able, it is possible to order group-task situations The subjects for this experiment were 90 male
along the favorableness continuum, by first order- undergraduates at the University of Florida. They
ing the group-task situation on the basis of the were randomly assigned to 18 five-member groups.
leader's relation with his group, then on the basis Nine groups worked under directive leadership and
of task structure, and lastly on the basis of 9 under nondirective leadership.
position power. This ordering may be considered
to be an operational definition of the favorability Tasks
continuum. The most favorable group-task situa- Each group attempted the same three tasks, which
tion is one in which leader-member relations were chosen to vary the favorability dimension, as
are good, the task is highly unstructured and the described below. Task A was a discussion task
position power is strong; the most unfavorable adapted from Cleveland and Fisher (1957). It re-
quired the group to list the five most important
group-task situation is one in which leader- traits that a person needs for success in our culture.
member relations are very poor, the task is Task B was a discusison task taken from a report
unstructured, and the position power is weak. by Bass (1960) involving the case of a young
In support of his theory, Fiedler reexamined politician who is burdened with an alcoholic wife.
the findings of several studies of leadership, in- The group was asked to decide which of five pos-
volving 21 different kinds of groups. In these sible courses of action would be the best one. Task C
studies leader-member relations had been mea- called for the group to identify five objects
sured by a variety of self-report procedures ("wrench," "ruby," etc.) by asking questions (Smith,
(sociometric choice, leader ratings of group 1957). The subjects were given a clue that the object
was either vegetable, animal, or mineral, and then
atmosphere, etc.). In Fiedler's reexamination, permitted 40 questions to identify it. These tasks
task structure was operationally denned in terms have been described fully in an earlier report (Shaw,
of four task dimensions (decision verifiability, 1963).3
goal clarity, goal path multiplicity, and solution
specificity) suggested by Shaw (1963), and in- Experimental Manipulations
herent power of the leadership position was Leadership style. Type of leadership behavior was
tentatively defined by a check-list rating of the manipulated by instructions to the assigned leader.
leader's position. Using these measures, a favor- One member of each group was briefed in private by
ability continuum was determined by the order- the experimenter's assistant. He was then directed to
ing procedure described above. Median correla- report to the experimental room in the same manner
tions between least preferred co-worker (LPC) as the other subjects. Half of the leaders were
scores and group effectiveness scores were then instructed to behave in a controlling, directive man-
related to this favorableness continuum. This ner and the other half were instructed to be permis-
yielded a U shaped curve showing that leaders sive and nondirective in their behavior toward other
group members.
with low LPC scores (directive leaders) were Group-task favorability. As Fiedler (1964) noted,
more effective at both extremes of the con- it is difficult to obtain a laboratory situation in
tinuum, whereas leaders with high LPC scores which affective leader-member relations are poor,
(nondirective leaders) were more effective in the and when a leader is assigned to the group his power
middle range of the favorableness continuum. position is strong. Hence, the group-task favorabil-
This support for the theory is impressive, but ity variable was manipulated by means of task
is open to the objection that since the interpre- structure only. The three tasks were selected to vary
tation is ex post facto it may not stand up under on the solution multiplicity dimension, based upon
cross-validation. Furthermore, leadership behav- the scale values determined by Shaw (1963) ; scale
values on other task dimensions were essentially
ior was inferred from personality scores, a the same.
hazardous procedure at best. The purpose of the
3
present study is to test the generality of Fiedler's Copies of tasks, general instructions, and special
theory by experimentally manipulating both the instructions to leaders have been deposited with the
group-task favorability dimension and the be- American Documentation Institute. Order Document
havior of the leader. No. 860S from ADI Auxiliary Publications Project,
Photoduplication Service, Library of Congress,
METHOD Washington, D. C. 20540. Remit in advance $1.25
for microfilm or $1,25 for photocopies and make
The experimental design was a mixed factorial, checks payable to: Chief, Photoduplication Service,
involving two styles of leadership (directive and Library of Congress.
240 BRTKF ARTICLES

From this manipulation, it is clear that Ilic group-


task situation was most favorable lor Task C M KAN T l M K SCORES ( M l N U T U S ) I''"R LEAPEKS1J
(leadcr-mcmber relations good, position power AND TASK CONDITIONS
strong, task structured), most unfavorable for Task
A (leader-member relations good, position power Task
strong, task unstructured), and intermediate for
Task B (leader-member relations good, position
power strong, task moderately structured). Thus, A B c
directive leadership should be more effective for Directive 23.42 13.36 24.67
Tasks A and C, and nondirective leadership for (4.80) (3.71) (4.97)
Task B. Nondirective 16.76 5.29 34.73
(4.03) (2.36) (5.85)
Procedure
The five subjects in each group were seated around Note.—Mean transformed scores are given in parentheses.
an oval work table. Each person was given a copy
of the instructions so he could follow the experi- were transformed by the square-root transforma-
menter's verbal presentation. The instructions indi- tion prior to analysis. Table 1 gives the means
cated that the group would be asked to solve a of raw scores and of transformed scores for
number of tasks working together as a group, and
leadership and task conditions.
that Person A (the instructed subject) was appointed
leader to facilitate interaction. The instructions fur-
Analysis of variance yielded significant F
ther stated that the group must follow the leader's values for tasks (F = 31.96, df = 2/32, p < .01)
directions. and for the Leadership Style X Tasks interaction
The order in which the three tasks were presented (F=7.54, df-2/32, p < .01). For purposes
to the groups varied according to a systematic Latin of this study, the significant interaction term
square, such that each task was attempted first, is of greatest interest. Inspection of the data in
second, and third an equal number of limes. After Table 1 reveals that directive leadership was
each task was explained to the group, 5 minutes were more effective than nondirective leadership only
given for planning the method of attack. Records on Task C; on both Tasks A and B nondirective
were kept of time required for completion and of
final solutions for each of the tasks. After all three
leadership was more effective. Separate t tests
tasks had been completed, each subject responded revealed significant differences between leader-
to a questionnaire which called for ratings of satis- ship styles on Tasks B and C (p < .01 in each
faction with the group, group cooperation, group case), but the difference on Task A was not sig-
performance, leadership performance, and directive- nificant (p < .20). This finding only partially
ness of the leader's behavior. supports the prediction that directive leadership
should be more effective on both Task A and
RESULTS Task C.
Questionnaire data are given in Table 2,
Before examining the results, it is necessary broken down for leaders and followers taken
to inquire whether the leader's behavior was ma- separately. Although ratings of satisfaction, co-
nipulated in the intended direction. Evidence on operation, group performance, and leader per-
(his was obtained from the ratings of the leader. formance were all higher for nondirective than
Ratings on a S-point scale ranging from "very for directive leadership, differences were signifi-
directive" to "very nondirective" were asked for cant only for ratings of cooperation (p < .05).
in response to the question, "How would you None of the differences between leaders and
classify your leader?" Rating scores could range
from 1 for a "very nondirective" response to
5 for a "very directive" response. The mean TABLE 2
rating given by followers of leaders instructed MEAN RATINGS OF LEADERSHIP DIRECHVENESS,
SATISFACTION, COOPERATION, GROUP PERFOR-
to behave in a directive manner was 4.12 as MANCE, AND LEADER PERFORMANCE
compared with 3.52 for followers of nondirective
leaders (p < .02). The manipulation thus appears Diicclivc leader Nomlireetivt' leader
to have produced the intended differences, al-
though both types of leaders apparently were Leader Follower Leader Follower
seen as more directive than nondirective.
The time scores are the only measures of group Leader directiveness 3.78 4.12 2.94 3.52
performance that are comparable across tasks. Satisfaction 4.28 4.01 4.34 4.29
Cooperation 3.83 3.87 4.50 4.41
These scores were approximately normal in dis- Group performance 4.17 3.84 4.33 4.22
tribution, but the means and standard deviations Leader performance 2.84 3.20 3.67 3.74
tended to be correlated. Therefore, raw scores
BRIEF ARTICLES 241

followers was significant, except for ratings of ership simply because his attempted directions are
leader directiveness. Leaders rated themselves ignored, whereas the attempted permissiveness
less directive than did their followers (p < .02). (encouraging contributions for all, etc.) of the
nondirective leader would only distract the group
DISCUSSION from its goal. This difference in leadership ef-
fectiveness would be unrelated to favorability of
The most useful results for evaluating Fied- the group-task situation for the leader, as opera-
ler's contingency model are those obtained from tionally defined.
the time scores. Under the conditions established In short, it is suggested that task structure is
in this experiment, it is clear that directive an important variable in the determination of
leadership was more effective than nondirective leadership effectiveness, but that this variable is
leadership only when the group-task situation was ambiguously related to the favorability continuum
highly favorable for the leader. Fiedler's model postulated by Fiedler. A more complete analysis
predicts that the directive leader should also have requires a consideration of the particular task
been more effective when the group-task situa- requirements in relation to leadership type. Fied-
tion was highly unfavorable, and to this extent ler's (1964) discussion of coacting versus inter-
the present findings appear to be at variance with acting groups is a step in this direction, but
the model. Obviously, this discrepancy may be greater consideration of task dimensions is called
due either to a failure to manipulate the group- for.
task dimension effectively or to inadequacies of Regardless of their implications for the con-
the model. tingency model, the results of this experiment
The most likely explanation is that the highly show clearly that directive leadership is more
unfavorable situation (Task A) is, at most, only effective than nondirective when the task is
moderately unfavorable. Favorability was manip- highly structured; that is, when there is only
ulated only by task structure as reflected by the one solution and one way (or only a few ways) of
solution multiplicity task dimension. The task is obtaining this solution. The requirements for
thus unstructured in the sense that there are leadership are quite limited, and nondirective
many possible solutions; however, the other leader behaviors may only interfere with the
variables postulated as determinants of favor- problem-solving process. However, on tasks that
ability supposedly contributed to high favorabil- require varied information and approaches, non-
ity. If this reasoning is correct, the conditions in directive leadership is clearly more effective. On
this experiment actually varied only from highly such tasks the requirements for leadership are
favorable to moderately unfavorable, and the great. Contributions from all members must be
findings agree with theoretical expectations. encouraged, and this requires motivating, advis-
Despite the apparent correctness of this in- ing, rewarding, giving support—in short, nondi-
terpretation, there is some reason to believe that rective leadership.
the theory can be improved. Fiedler noted that
the model is an oversimplification, and pointed REFERENCES
to the difficulties in weighting the dimensions re-
flected by the favorability continuum. It is the ANDERSON, L. R., & FIEDLER, F. E. The effect of
participatory and supervisory leadership on group
view of the writers that the oversimplification is
creativity. Journal of Applied Psychology, 1964,
most evident with respect to the task structure
48, 227-236.
dimension. Differences among tasks that are rele- BASS, B. M. An evaluation of the use of objective
vant to leadership requirements are almost cer- social data for training problem-solving discus-
tainly not limited to differences in task struc- sants. Technical Report No. 22, 1960, Louisiana
ture. For example, variations in difficulty and State University, Contract N7 ONR 35609, Office
cooperation requirements probably call for dif- of Naval Research.
ferent leader behaviors. Furthermore, it seems CLEVELAND, S. E., & FISHER, S. Prediction of small
unlikely that the task structure dimension is un- group behavior from a body image schema. Human
ambiguously related to the favorability contin- Relations, 1957, 10, 223-233.
uum. An unstructured task may in some instances FIEDLER, F. E. Leader attitudes and group effective-
be favorable for the leader in the sense that it ness. Urbana: Univer. Illinois Press, 1958.
encourages the members to accept his leadership. FIEDLER, F. E. A contingency model of leadership
effectiveness. In L. Berkowitz (Ed.), Advances in
On the other hand, a highly structured task may experimental social psychology. Vol. 1. New York:
be unfavorable to the leader in the sense that his Academic Press, 1964. Pp. 149-190.
leadership is superfluous. In this case, directive LEWIN, K., LIPPITT, R., & WHITE, R. K. Patterns of
leadership might be superior to nondirective lead- aggressive behavior in experimentally created "so-
242 BRIEF ARTICLES

cial climates." Journal oj Social Psychology, 1939, SHAW, M. E. Scaling group tasks: A method for
10, 271-2')'). dimensional analysis. Technical Report No. 1,
PRESTON, M. G., & HEINT/, R. K. Effects of partici- 1963, University of Florida, ONK Contract NR
patory vs. supervisory leadership on group judg- 170-266, Nonr-580(ll), Office of Naval Research.
ment. Journal of Abnormal and Social Psychology, SMITH, E. E. The effect of clear and unclear role ex-
1949, 44, 345-355. pectations on group productivity and defensive-
ness. Journal of Abnormal and Social Psychology,
SHAW, M. E. A comparison of two types of leader- 1957, 55, 213-217.
ship in various communication nets. Journal of Ab-
normal and Social Psychology, 1955, 50, 127-134. (Received August 27, 1964)

Journal of Personalty and Social Psychology


1966, Vol. 3, No. 2, 242 244

CONFORMITY AS A FUNCTION OF BIRTH ORDER AND TYPE


OF GROUP PRESSURE:
A VERIFICATION

SELWYN W. BECKER MELVIN J. LERNER


Graduate School of Business, University of Chicago Department of Behavioral Science, University oj
Kentucky
AND JEAN CARROLL
Graduate School of Business, University of Chicago

1st born and later-born adolescents were tested in conditions resembling some
of those used by Deutsch and Gerard. In the condition most closely resembling
the Asch situation the Ist-born Ss yielded more than the later born. In a group-
reward condition the ] si-born Ss showed a significant increase in conforming
errors while the laler-born Ss were relatively unaffected. In a memory condition
the later-born Ss yielded more than in either of the other 2 conditions while
the Ist-born Ss were relatively unaffected. These findings were interpreted as
confirming the hypotheses that Ist-born persons arc more responsive to
normative influences while later-born persons will be more affected by in-
formational influences.

In a recent experiment (Becker, Lerner, & same country club. It is probably safe to sug-
Carroll, 1964) first-born and later-born adoles- gest that they were from the lower middle class
cents were tested in an Asch (1956) situation. and were somewhat special within that class in
By introducing the anticipation of a small or terms of the kind of employment they sought
large "payoff" for each correct judgment the and achieved.
amount of yielding was significantly affected. A The second, more serious problem derives from
small payoff greatly decreased conforming errors the line of reasoning concerning the effect of a
in the first-born group, and somewhat decreased small versus a large payoff as a technique for
conforming errors in later-born subjects. A large manipulating the normative and informational in-
payoff, however, led to increased yielding only fluence (Deutsch & Gerard, 19SS) operating in
for the later-born subjects. These findings were the Asch situation. A small payoff was assumed
interpreted as confirming the hypothesis that to provide motivation to resist the normative in-
first-born persons are more dependent on others fluence in the situation and a large payoff was
for social support while later-born persons rely supposed to enhance the informational value of
more on others for validation of their beliefs. the unanimous majority's judgments. Though
Confidence in the validity and generality of perhaps it was a logical manipulation, there was
this interpretation is somewhat limited because very little historical precedent for using payoffs
of two aspects in the experimental situation. The in this manner.
first is based on the characteristics of the sub- The study reported here was designed to vali-
jects, all of whom were golf caddies from the date the original hypotheses by testing them

You might also like