A COMPARATIVE ANALYSIS OF BLACK AND
WHITE LEADERSHIP IN A NATURALISTIC SETTING
WILLIAM ROBERT ALLEN
A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA IN PARTIAL
FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
To Marlies, my wife,
with love and appreciation
The author wishes to express his appreciation to his Supervisory
Committee for their support and help in completing this dissertation.
The efforts of Jack M. Feldman, H. Joseph Reitz and James F. Burns
have proven invaluable in many ways, particularly in light of the
untimely passing of Walter A. Hill, the original committee chairman.
Walter A. Hill was both mentor and friend, and his support and
influence on this work and others was appreciated and will long be
remembered. Jack M. Feldman, first a committee member, assumed the
chairmanship and helped greatly toward placing this in completed
form. H. Joseph Reitz agreed to serve during the latter stages of
completion and has been very helpful. James F. Burns has been
steadfast in his support and helpful in his comments. The contribu-
tions and friendship of these gentlemen are gratefully acknowledged.
Several others have also been helpful in many different ways.
Marvin E. Shaw and Robert C. Ziller offered many thoughtful comments
during the formative stages of the present work. Ira Horowitz,
William F. Fox, John H. James and William V. Wilmot have provided
considerable support and encouragement during the doctoral program.
Robert R. Bell and Fred J. Nutt have helped in their comments and
A special word of appreciation is in order for John A. Ruhe and
Jerome F. Dederick. John A. Ruhe's generous suggestions, support
and friendship proved extremely helpful and was greatly appreciated.
SMCS Jerome F. Dederick USN, acting in his capacity as liaison with
the Navy during data collection, did much more than required and
contributed considerably with his enthusiasm and friendship. In
addition, the fidelity and friendship of Frankie Hammond, who pre-
pared the final copy, is sincerely appreciated.
Finally, the author must recognize the many sacrifices made by
his wife, Marlies, and children, William, Robert and Elizabeth.
Their support, love and understanding made the doctoral program
and this study possible.
TABLE OF CONTENTS
ACKNOWLEDGMENTS. . .
LIST OF TABLES . .
LIST OF FIGURES. .... ..
ABSTRACT . ....
I INTRODUCTION, PURPOSE OF THE STUDY, LITERATURE
REVIEW AND HYPOTHESES .
Introduction ..... .....
Purpose of the Study .
Literature Review. . .
Hypotheses . .
II METHODOLOGY .. .
Subjects . .
Physical Environment . .
Design . .. .. .
Observer Training and Reliability .
Procedure. . .
II ANALYSIS AND RESULTS. ...
Methods of Analysis .... .. ..
Results .. .
IV SUMMARY, DISCUSSION AND IMPLICATIONS FOR FUTURE
RESEARCH. . .. ....
Summary. . .....
Discussion .. .......
Implications for Future Research ...
TABLE OF CONTENTS (CONTINUED)
REFERENCES . . .
A QUESTIONNAIRE FOR SQUAD LEADERS ...
B PERSONAL QUESTIONNAIRE. .
C IEC . .. .
D SOCIAL ORIENTATION TASKS. .
E RSE . ..
F LOQ . .
G SBD .
H JDI-GT. . .
I JDI-SL. .. . ...
J JDI-S QUESTIONNAIRE ABOUT SUBORDINATES.
K JDI-FS QUESTIONNAIRE ABOUT FELLOW SUBORDI
L LAST PAGE . ..
M INSTRUCTIONS. . .
N INSTRUCTIONS. . .
0 INSTRUCTIONS . .
P SHIP-ROUTING INFORMATION. ..
Q AVAILABLE ROUTE COMBINATIONS. .
. .E .
BIOGRAPHICAL SKETCH. . .. .
LIST OF TABLES
1 PERSONAL PROFILE OF SQUAD LEADERS AND SQUAD MEMBERS. 27
2 RELIABILITY BETWEEN BLACK AND WHITE OBSERVERS. 48
3 MEAN IPA SCORES FOR LEADER RACE MAIN EFFECTS -
GROUP I HYPOTHESES . . 64
4 MEAN IPA SCORES FOR LEADER RACE X GROUP TYPE
INTERACTIONS GROUP I HYPOTHESES. . 66
5 MEAN IPA SCORES FOR GROUP TYPE MAIN EFFECTS -
GROUP I HYPOTHESES . . 68
6 MEAN PERFORMANCE SCORES FOR LEADER ATTRIBUTE X
GROUP TYPE INTERACTIONS FOR GROUPS OF EQUAL RACIAL
COMPOSITION GROUP II HYPOTHESES. . 71
7 MEAN PERFORMANCE SCORES FOR LEADER ATTRIBUTE X GROUP
TYPE INTERACTIONS FOR GROUPS OF EQUAL SIZE -
GROUP II HYPOTHESES. . . 73
8 MEAN GROUP IPA SCORES FOR LEADER RACE MAIN EFFECTS 80
9 MEAN GROUP IPA SCORES FOR GROUP TYPE MAIN EFFECTS. 81
10 MEAN GROUP IPA SCORES FOR LEADER RACE X GROUP TYPE
INTERACTIONS .. . 82
11 MEAN LEADER IPA SCORES FOR GROUP TYPE MAIN EFFECTS IN
SOCIAL-EMOTIONAL BEHAVIOR. ... . 86
12 MEAN LEADER IPA SCORES FOR LEADER RACE X GROUP TYPE
INTERACTION IN LEADERSHIP STYLE. . 87
13 MEAN LEADER SCORES FOR GROUP TYPE MAIN EFFECTS FOR
THE GENERAL SATISFACTION SURVEY. . 89
14 RESULTS OF TESTING HYPOTHESES.. . 93
LIST OF TABLES (CONTINUED)
15 SUMMARY OF OBSERVED IPA DIFFERENCES BETWEEN
LEADERS. . 98
16 FREQUENCY COMPARISON OF DIFFERENCES IN OBSERVED IPA
BEHAVIOR BETWEEN LEADERS, SUBORDINATES AND GROUPS
BY TASK TYPE AND BY GROUP TYPE. . 99
LIST OF FIGURES
I Recruit Company Organization . ... 29
2 Physical Plan of Experiment Rooms and Observation
Stations . .. .. 32
3 Schema of the Basic Research Design. .. 35
4 Notation and Designs for Data Testing. . 62
Abstract of Dissertation Presented to the Graduate Council of the
University of Florida in Partial Fulfillment of the Requirements
for the Degree of Doctor of Philosophy
A COMPARATIVE ANALYSIS OF BLACK AND
WHITE LEADERSHIP IN A NATURALISTIC SETTING
William Robert Allen
Chairman: Jack M. Feldman
Major Department: Management
The objective of this study was to investigate differences in
attitudes and behavior between black and white leaders supervising
biracial groups of varying size and composition. In addition, differ-
ences between subordinates and groups were investigated.
A total of 288 male naval recruits participated, evenly balanced
between blacks and whites. Eight different black and white squad
leaders, 64 in all, supervised groups composed of members from their
own squads in performing a knot-tying task and a ship-routing task,
while a pair of racially mixed observers watched through a one-way
window and recorded group interaction using Bales' Interaction Process
Analysis (IPA). Each leader supervised one of four types of groups:
racially balanced dyads, 25 percent black tetrads, 50 percent black
tetrads and 75 percent black tetrads. The result was fundamentally
a two-by-four factorial design. An assistant in the experimental room
recorded the time to perform each task and each individual's speech
After performing the tasks each subject completed a series of
questionnaires. Leaders completed the Leadership Opinion Question-
naire, providing a self-rated measure of the leadership dimensions
initiating structure and consideration; subordinates completed the
Supervisory Behavior Description, providing a subordinate-rated
measure of the same dimensions. Individual satisfaction with
subordinates, tasks, leaders and/or subordinates was measured using
selected scales of the Job Descriptive Index. Individuals completed
Rotter's Internal-External Control Scale, Ziller's Self-Esteem
Instrument, Rosenberg's Self-Esteem Scale.and answered certain
As hypothesized, white supervised groups performed both tasks
faster than black supervised groups, and black leaders were less
expressive in their behavior and scored lower in internal control
than white leaders. Performance deficiency was explained in terms
of status incongruence, group heterogeneity and experimental stress;
the less expressive behavior, in terms of status characteristic
theory, interracial interaction disability and social stress.
Contrary to the hypotheses, white leaders were not higher in
self-esteem, and black leaders did not have a greater speech duration,
more satisfaction with subordinates, or more satisfaction with the
tasks. Also not supported were hypotheses stating that, for each
task, the performance of groups of equal size and of equal racial
composition would be faster when supervised by leaders high in
self- and subordinate-rated initiating structure and consideration,
self-esteem, internal control and intelligence (measured by the Navy
Basic Test Battery), compared to leaders of the same race low in
Comparing black and white leaders' use of the IPA categories
indicated race, group and interaction effects. White leaders were
clearly more active in interpersonal behavior. In every instance
of a significant difference between leader types for both tasks,
black leaders displayed less of the behavior in question than did
white leaders. Although white leaders gave and asked for more opinions
during the ship-routing task, the results were more pronounced
during the knot-tying task where white leaders exhibited more activity
in five of the twelve categories: showing solidarity, giving sugges-
tions, asking for information and opinions and disagreeing. These
results suggest inhibition on the part of the black leaders. Patterns
of behavior displayed through IPA differences suggest similar results
for black subordinates. During the knot-tying problem, black super-
vised groups displayed, on the average, less agreeing, less asking
for information, less asking for opinion.and more antagonism, suggest-
ing behavior that hindered performance. During the ship-routing
problem, black supervised groups displayed, on the average, less
asking for opinion but more asking for information, and were greater
in speech duration, which may have hindered performance by detracting
from task demands.
The data suggest that the leaders, regardless of race, experi-
enced supervision difficulties due to increasing the size of sub-
ordinate groups and due to increasing the relative number of blacks
in subordinate tetrads.
INTRODUCTION, PURPOSE OF THE STUDY, LITERATURE
REVIEW AND HYPOTHESES
The emergence of the "new Negro" or "black" is one of the most
striking. phenomena of the latter half of the 1960's (Miller and Dreger,
1973).1 As these minority-group Americans become assimilated into
social institutions In a manner, and at levels, never before achieved,
the appropriateness of current organizational practices may have to be
questioned. As Miller and Dreger (1973) have pointed out, the
behavioral setting can make a major difference In comparative behaviors.
One important modification in the behavioral setting will be the
emergence of more blacks into positions of leadership and influence.
Not only is this an expectation of society in general, it is, as King
and Bass (1970) point out, natural for blacks to desire managerial
positions when they have already achieved entry into the organization
and access to skilled jobs.2
When making racial designations, the terms black and white will be
favored over the terms Negro and Caucasian in the belief that they are
the currently preferred social terms.
2King and Bass (1970) have proposed a hierarchy of concerns about
integration in organizations that is analogous to Maslow's (1954) postu-
lation of a hierarchy of needs. Their hierarchy was conceptualized as
a three-tiered, truncated pyramid consisting of: entry into the organization,
access to skilled jobs, and access to managerial positions.
The research literature is presently equivocal as to how blacks
might compare to whites in leadership or supervisory roles. We know
very little about racial differences in these areas. The basic problem
is to determine what differences there are and to see If these
differences can be explained. Because blacks do form a significant
ethnic subgroup in our society and because our social institutions are
now dominated by whites, this problem should be considered as openly
and objectively as possible so that as organizational integration
proceeds difficulties can be properly anticipated and effectively
minimized. The need for social psychological research concerning
actual differences between blacks and whites and the nature of the
supervisor-subordinate interaction in job settings has been expressed
by many authors (see, e.g., Dreger and Miller, 1968; King and Bass, 1970;
Moskos, 1967; Triandis and Malpass, 1971).
Racial differences and the psychology of blacks have been discussed
in numerous works (e.g., Allen, 1970; Anastasi and Foley, 1949; Bendix
and Lipset, 1953; Benedict and Weltfish, 1943; Boyd, 1950; Deutsch,
Katz and Jensen, 1968; Dreger and Miller, 1960, 1968; Dunn and Dobzhansky,
1946; Frazier, 1939, 1957; Garth, 1925, 1931; Ginzberg, 1956; Harding
et al., 1969; Kardiner and Ovesey, 1951, 1962; Katz, 1970; Klineberg,
1935, 1944; Knox, 1945, 1949, 1952; Lindzey, 1954; Miller and Dreger,
1973; Montagu, 1952; North, 1957; Pettigrew, 1964; Sarason and Gladwin,
1958; Shuey, 1958; Tyler, 1956; Woodworth, 1916).3 Early comparative
3While the term race will be used for convenience, no meaning is
intended other than that of distinctiveness of appearance and commonality
of experience; the issue of whether there are consequential differences
in the genetic endowment of blacks and whites will not be considered
(cf., Katz, 1964).
research was concerned primarily with attempts to measure and describe
interracial differences within a normative framework where the behavior
of whites served as a norm against which black behavior was evaluated.
Social psychological research was concerned primarily with the response
of whites to blacks; thus, attempts to do such things as modify attitudes,
measure social distance, etc., were directed to whites rather than
blacks (cf., Miller and Dreger, 1973).
The relatively recent advances being made by minority-group Americans
into positions of leadership and supervision is a trend deserving further
study, particularly in field and naturalistic settings where blacks and
whites coexist in comparable organizational positions. The experimental
study of black-white relationships is a relatively unexplored field
offering unique research opportunities (Katz, 1970).
Purpose of the Study
The main concern of this study is the development of knowledJe and
information concerning differences In leadership qualities between
black and white leaders. This study from its Inception was conceived
to follow Ruhe (1972). It is not an exact replication, but several of
the same methodological techniques were employed in an effort to
generalize earlier results to natural work groups outside the laboratory,
to enhance the external validity of both studies. The focus is on
differential leadership qualities, though certain comparisons oP black
and white subordinate behavior were made.
At this point It is appropriate to briefly describe the Ru!ii study.
Using 96 male undergraduate students as subjects (48 black and 48 white),
Ruhe randomly assigned each of 12 black and 12 white older students (who
had volunteered to be supervisors) to supervise three different types of
subordinate dyads (one all-black, one all-white, and one racially-mixed).
The remaining 72 students, 36 black and 36 white, were randomly assigned
to one of the three subordinate dyads that participated once with a black
supervisor and once with a white supervisor. Each group performed
three tasks: knot-tying, ship-routing and letter writing. During the
performance of the tasks two trained observers (one black and one white)
coded the group interaction using Bales' (1950) Interaction Process
Analysis (IPA). Other output measures recorded were: duration of speech
for each group member, amount of time needed to complete each task and
cohesiveness. After performing the last of the tasks, both supervisor
and subordinates completed satisfaction rating forms (from Smith, Kendall
and Hulin, 1969) for each other and for the work in the tasks. At the
end of their participation each subject completed a self-esteem form
(from Ziller, 1971).
In designing the present study the desire was to draw from the
Ruhe study and extend the scope of research in this area, using a different
subject population in a more naturalistic setting. In keeping with this
desire, several methodological changes were made: (1) group size and
racial composition were varied; (2) instruments for measuring leader
behavior (the Leadership Opinion Questionnaire and Supervisory Behavior
Description, which provide leader dimensions known as initiating structure
and consideration, from Stogdill and Coons, 1957) were included to tie
in with recognized leadership literature; (3) the Internal-External
Control Scale (Rotter, 1966) assessing locus of control was included,
since this quality has been shown to be related to effectiveness in
attempts to influence others (Lao, 1970); (4) a second self-esteem
instrument (the Self-Esteem Scale, from Rosenberg, 1965) was included
because it is based on verbal self-report; (5) cohesiveness (defined
to be the proportionate usage of "we" or "1") was eliminated as an
output variable due to the apparent lack of a conceptual base in the
biracial situation (Ruhe, personal communication); and (6) the letter
writing task was eliminated because it was felt that the manner In which
this was handled by Ruhe was not appropriate in the field setting, since.
in the university setting each group was asked to develop a recruiting
letter urging college students to join an all-volunteer Navy.
From the above, therefore, the principal purpose of the present
study can be stated as an investigation of some of the possible differences
between black and white leaders supervising racially heterogeneous
groups, of varying size and racial composition, in the performance of
Synthesizing the literature addressed by this study Is a difficult
task due to its scope and diversity. Further, because of the relatively
recent dramatic changes in black-white relations, there is a very real
possibility that much of the research in existence may be partially,
if not totally, irrelevant to the present day black, assuming that
enduring changes in attitudes and behavior have occurred. This should
be borne in mind when the literature is examined. Notwithstanding the
difficulty of synthesis and the relevancy Issue, several highly plausible
general conclusions do emerge. Upon these hypotheses can be advanced arid
tested within the limits established by this study.
The review of the literature that follows focuses mostly on those
attributes which have been found to differentiate between blacks and
whites and which were addressed by this study, with the exception of the
research involving the leader dimensions initiating structure and
consideration. The review is grouped under the following sub-headings:
Biracial Work Groups, Leadership Studies, Leader Behavior and Task
Performance, Self-Concept, Locus of Control, Intell ience and Aptitude,
Job Satisfaction, Miscellaneous Related Research and Conclusion.
Biracial Work Groups
The behavior displayed in biracial work groups and problem-solving
situations may suggest how blacks and whites compare in leadership
positions. Blacks have displayed marked social inhibition and subordination
to white partners In cooperative problem-solving situations, even when
both races were matched on intelligence and made to display equal
ability at the task; blacks mostly ignored one another, made fewer
proposals, were less willing to argue for their point of view, were
more susceptible to group influence, ranked whites higher on mental
ability, favored whites when talking, favored one another as future
work companions, and expressed less satisfaction with group experiences
than did whites (Cohen, 1972; Katz and Benjamin, 1960; Katz, Goldston
and Benjamin, 1958).
Although some studies suggest that blacks alter their behavior in
an unfavorable direction or decrease in performance efficiency when
faced with a white rather than a black frame of reference (Hatton, 1967;
Katz and Cohen, 1962; Katz, Epps and Axelson, 1964; Katz, Roberts and
Robinson, 1965; Preston and Bayton, 1941), others suggest improved
performance in the white norm condition (Epps, Katz and Runyon, 1970;
Katz, Epps and Perry, 1970; Katz et al., 1972).
Some of the work done by Katz and his colleagues uses the race of
the experimenter as one of the major variables. These provide results
relevant to this study. Under a low stress conditions blacks worked
better in an all-white environment; however, under a high stress
condition black efficiency improved in the all-black environment and
went down in an all-white environment (Katz and Greenbaum, 1963).
These results were interpreted in terms of the hypothesis of an Inverted
U-shaped relationship between arousal and performance (cf., Malmo, 1957;
Duffy, 1957), with the low stress, all-black environment considered
insufficiently arousing for optimal performance and the high stress,
all-white environment too arousing, since the blacks worked best in a
low stress, all-white environment. In another study (Katz, Roberts and
Robinson, 1965), relating the most difficult task level to intelligence
reduced black performance under a white experimenter, but when the same
task level was described as a test of eye-hand coordination (i.e., an
ability which blacks are not stereotyped as lacking), black performance
was better with a white experimenter than with a black experimenter.
These results were related to the concept of task motivation as a
joint function of the subjective probability and incentive value of
success (cf., Atkinson, 1964); thus, when the likelihood of winning
approval was equally high, blacks would work harder for a white person
(i.e., a higher status individual) than a black person.
Actual white behavior in the biracial situation has not been studied
to the same degree as black behavior. Burnstein and McRae (1962)
investigated the attitudinal effect of participation in biracial
problem-solving groups and found that under conditions of shared threat
a reduction in the expression of prejudice occurred, although communication
to the black participants was significantly less by the high prejudiced
white subjects regardless of the presence or absence of shared threat.
Another study (Katz, Goldston and Benjamin, 1958) found that group
reward produced a greater amount of cooperative behavior in both black
and white subjects. Although no more likely to have correct solutions
than blacks in group problem-solving, whites still exercised more social
influence (Katz and Benjamin, 1960; Katz, Goldston and Benjamin, 1958).
When talking, whites have favored whites as the target of their inter-
action (Cohen, 1972; Katz and Benjamin, 1960; Katz, Goldston and Benjamin,
1958). While in one study (Katz, Goldston and Benjamin, 1958) whites
chose blacks as often as they chose one another as future working
companions and ranked blacks as high as themselves on task competence,
in another experiment (Katz and Cohen, 1962), whites downgraded the
problem-solving ability of black partners and expressed less willingness
to continue working with the blacks as compared to a control group of
From these studies we can reach several plausible conclusions.
Black performance probably depends upon the racial-environmental
conditions under which they must work and anticipation of a cross-racial
comparison may have a favorable motivational effect. Members of both
races may favor white recipients when talking, ostensibly because whites
are perceived as having higher social status. Whites are probably more
influential in biracial group decisions and blacks are probably more
susceptible to group Influence. The reaction of whites to blacks in the
biracial problem-solving situation remains equivocal.
If leadership is viewed as a social exchange process a satisfactory
conceptual base can be provided for predicting group performance on the
basis of the research concerning biracial work groups. The essence of
social exchange is the development of relationships with other persons
such that benefits of mutual value can be "traded" between participants
of both equal and unequal status (Jacobs, 1970). Thus, participants
accrue "assets" which determine their status within the group. Leaders
are afforded or possess the greater assets, allowing them to function
more effectively in their leadership role. Implicit in this exchange or
transactional approach is the understanding that Individual, situational
and cultural differences have the potential of impinging upon the
leadership process and influencing or moderating relevant variables.
Therefore, the research concern racial work groups suggests that
groups supervised by white leaders will outperform groups supervised by
black leaders because white leaders will have more transactional assets
than will black leaders.
An important consideration is the functioning of blacks in leadership
positions and their acceptance by whites as leaders. Support for a
proposal that whites in the job setting would be willing to have blacks
over them in a position of leadership and supervision can be found in
Campbell (1971) where 86 percent of the white respondents to a 15-city
survey said they would not mind at all having a qualified black as a
supervisor on their job and Cavanagh (1971), who found that whites
accepted a black supervisor more readily than did many of the blacks.
Using the Leadership Behavior Description Questionnaire (Fleishman,
1953), Glasgow (1970) found no significant differences between the
initiating structure and consideration leader behavior scores of a
sample of black public school principals and those of a sample of white
principals taken from a previous study. Glasgow points out that black
principals see themselves as being competent to assume the leadership
role in professionally staffed organizations.
Leadership assumption in a simulated clerical task was investigated
by Fenelon and Megaree (1971). The rate of leadership assumption by
high-dominance white women paired with low-dominance black women was
significantly lower than the rate of leadership assumption by both black
and white high-dominance women in the other groups. Analysis of the
process for deciding who should assume leadership suggested that the
above finding stemmed from the reluctance of the high-dominance white
women to assume leadership over the low-dominance black women, coupled
with the increased assertiveness of low-dominance black women when
paired with a white partner.
Using a business game to investigate interaction difficulties of
black and white coworkers with blacks in a supervisory position, Richards
and Jaffee (1972) found that the performance ratings by white observers
of black supervisors were significantly poorer than those of white
supervisors; that subordinates supervised by blacks gave more suggestions
and opinions and disagreed more than subordinates supervised by whites
whereas subordinates supervised by whites showed more solidarity and were
more accepting than subordinates supervised by blacks and some of these
behaviors (e.g., being less accepting) appeared to hinder the
effectiveness of black supervisors; and that subordinates with negative
racial bias gave poorer ratings to black supervisors than subordinates
with more positive racial attitudes.
Investigating the effects of varying racial compositions upon the
attitudes and behavior of black and white supervisors and subordinates,
Ruhe (1972) found no significant differences between black and white
supervisors in style of leadership (defined to be directive or nondirective),
duration of speech, self-esteem and satisfaction with subordinates and
work in three tasks.
Studies relating behavior scored by the IPA to leadership or
interracial leadership behavior (Richards and Cuffee, 1971; Richards and
Jaffee, 1972; Ruhe, 1972; Zdep, 1969) are generally Inconsistent. The
only notable commonality is the lack of an indicated relation for
categories 2 (showing tension release), 3 (agreeing), 7 (asking for
information), 10 (disagreeing) and 12 (showing antagonism). Richards
and Cuffee (1971) found that leaders (race unspecified) of Interacting
groups emitted more behavior in catagorles 4 (gives suggestion), 6
(gives information) and 9 (asks for information), and, in general,
emitted more behaviors; in addition, they found in interacting groups
that the effectiveness of leaders correlated with categories 1 (shows
solidarity), 4 (gives suggestion), 5 (gives opinion) and II (shows
tension). Richards and Jaffee (1972) found white leaders emitting more
behavior than black leaders in categories 1 (shows solidarity), 4 (gives
suggestion)and 6 (gives information). Ruhe (1972) found that white
leaders emitted more behavior than black leaders in categories 8 (asks
for opinion) and 9 (asks for suggestion). Zdep (1969) relates
leadership to categories 1 (shows solidarity), 4 (gives suggestion), 5
(gives opinion), 6 (gives information), 8 (asks for opinion) and 9 (asks
Overall, perhaps the best conclusion that can be reached Is that
the evidence fails to demonstrate great deficiencies for blacks in
positions of leadership. More specifically, based on the results of the
two studies relating IPA-scored behavior to leadership (Richards and
Jaffee, 1972; Ruhe, 1972) the most plausible general conclusion for
differences in IPA-scored behavior between black and white leaders
appears to be that when differences occur white leaders will emit more
behavior than black leaders.
Leader Behavior and Task Performance
Initiating structure and consideration as dimensions of leader
behavior emerged from studies initiated by the Personnel Research Board
of the Ohio State University (cf., Stogdill and Coons, 1957). These
dimensions resulted from a factor analysis of hypothesized dimensions
of leader behavior (Halpin and Winer, 1957) and were identified as the
smallest number of dimensions which would adequately describe leader
behavior as perceived by the leader's subordinates and as the leader
himself reported his own attitudes toward his role (Korman, 1966). They
are frequently used to account for a leader's behavior and its effects.
They may be defined as follows (cf., Harris and Fleishman, 1955;
Fleishman and Harris, 1962; Fleishman and Peters, 1962):
Initiating Structure: Includes behavior indicative
of the extent to which an individual is likely to
organize and structure his role and those of his
subordinates toward goal attainment. A high score
characterizes individuals who assume a more active
role in defining and facilitating group activity
through planning, scheduling, task assignment,
establishing ways of getting thigs done, and
establishing desired subordinate roles.
Consideration: Includes behavior indicative of the
extent to which an individual is likely to emphasize
mutual trust, respect for subordinate's ideas,
consideration for their feelings, and maintenance
of a certain warmth between himself and his group.
A high score characterizes individuals who emphasize
a deeper concern for subordinate needs and indicates
a climate of good rapport and two-way communication.
A low score is indicative of a more authoritarian and
impersonal relationship between the leader and his
From the literature there is little doubt that the leadership
dimensions initiating structure and consideration and similar behavior
categories describe important leader behaviors, but the lack of
consistent empirical evidence as to how these behaviors predict group
performance poses a major theoretical problem (Anderson, 1966; Campbell
et al., 1970; Korman, 1966; House, Filley and Kerr, 1971; Lowin,
Hrapchak and Kavanagh, 1969). With regard to structured leadership
behavior, the typical case seems to be that leaders high in Initiating
structure or similar measures of instrumental or structuring behavior
have higher performing subordinate groups (Bales, 1953; Bass and Dunteman,
1963; Bass et al., 1963; Dunteman and Bass, 1963; Halpin and Winer,
1957; Katz and Kahn, 1953; Katz, Maccoby and Morse, 1950; Moore, 1953;
Moore and Smith, 1952; Stouffer et al., 1949); however, Korman (1966)
revealed several studies showing no relationship between initiating
structure and performance. Two studies (House, Filley, Gujarati, 1971;
House, Filley and Kerr, 1971) have found a positive relationship between
initiating structure and satisfaction. With regard to supportive
leadership behavior, the typical case seems to be that leaders high in
consideration or expressive behavior have subordinate groups who are high
in measures of satisfaction (Argyle, Gardner and Cioffi, 1958; Baumgartel,
1956, 1957; Comrey, Pfiffner and Wallace, 1954; Danielson and Maler, 1957;
Fleishman and Harris, 1962; Halpin, 1954; Halpin and Winer, 1957;
Hemphill, 1957; Indik, Seashore and Georgopoulos, 1960; Moore, 1953;
Moore and Smith, 1952; Oaklander and Fleishman, 1964; Patchen, 1960;
Seeman, 1960; Spector, Clark and Glickman, 1960); however, this type
of leadership behavior has also been found to relate positively to
departmental and individual performance (Argyle, Gardner and Cioffi, 1958;
Blbu and Scott, 1962; Indik, Seashore and Georgopoulos, 1960; Katz and
Kahn, 1953; Katz, Maccoby and Morse, 1950; Llkert, 1961). Leaders rated
high on both initiating structure and consideration are more likely to
be judged effective by their superiors and to have desirable effects
on productivity and group morale (Fleishman and Harris, 1962; Fleishman
and Simmons, 1970; Halpin, 1954; Halpin and Winer, 1957; Misumi and
Tosaki, 1965; Oaklander and Fleishman, 1964; Shartle, 1956); one study
(Halpin, 1955) has shown that while the effective leaders of air crews
were both structuring and considerate, the same condition was not true
of leaders of educational institutions. Korman (1966) concludes that,
despite the fact that initiating structure and consideration have
become almost bywords in American industrial psychology, it seems
apparent that very little is now known as to how these variables
predict work group performance and the conditions which affect such
predictions. Indeed, Lowin, Hrapchak and Kavanagh (1969), after a
thorough reading of the relevant literature, suggest that there appears
to be much evidence that initiating structure and consideration can each
correlate positively, negatively, both positively and negatively (depending
on other variables) and only weakly if at all with effectiveness and
morale indices. Korman's (1966) review shows a predominance of low to
moderate correlations, almost all of a concurrent nature. There is as
yet almost no evidence on the predictive validity of initiating structure
and consideration nor on the kinds of situational moderators which might
affect such validity (Korman, 1966).
Given the above results, one cannot reliably predict the relationship
between initiating structure or consideration and task performance.
However, if one must predict the relationship the safer prediction appears
to be that task performance will be better for those leaders who are high
in these dimensions rather than low.
As Jacobs (1970) has noted, it is important to the individual that
his self-concept be as favorable as possible; the more favorable it is,
the greater will be his assets in social exchange relationships. Although
Coleman et al. (1966) found no difference in the self-concept of blacks
and whites in the educational setting, many studies of blacks suggest
the presence of an unfavorable self-image (Hodgkins and Strakenas, 1969).
Society has fostered a negative self-concept among blacks by nurturing
and rewarding feelings of inferiority and unworthiness (Poussaint and
Atkinson, 1968), and this sort of self-image may account for such behavior
as black parents favoring a light-skinned child (Coles, 1967; Grambs,
1964), dark-skinned men trying to marry wives of a lighter skin color
(Kardiner and Ovesey, 1962), and the tendency of the black child to
identify with the white majority (Clark and Clark, 1947; Goodman, 1952;
Landreth and Johnson, 1953; Morland, 1962; Radke and Trager, 1950;
Stevenson and Stewart, 1958). Considerable research supports the
contention that blacks almost inevitably develop feelings of low self-
esteem (Ausubel and Ausubel, 1963; Bernard, 1958; Bridgette, 1970; Clark,
1967; Guggenheim, 1969; Jefferson, 1957).
However the "black is beautiful" theme that emerged during the 1960's
raises serious doubts as to the enduring validity of many of these earlier
studies. Indeed, recent research does indicate an improved self-image
among blacks, often equalling or exceeding that of whites (Back and
Parmesh, 1969; Carpenter and Busse, 1969; Dennis, 1968; Douglas, 1970;
Greenwald and Oppenheim, 1968; Hodgkins and Strakenas, 1969; Hraba and
Grant, 1970; McElroy, 1971; Ogletree, 1969; Ruhe, 1972; White, 1971).
A careful examination of the literature leads one to the conclusion
that variables such as social setting (Ausubel, 1958; Coleman et al.,
1966; Hodgkins and Strakenas, 1969; Ruhe, 1972), sex (Carpenter and
Busse, 1969) and age (Clark and Clark, 1947; Morland, 1962; Proshansky
and Newton, 1968; Radke and Trager, 1950) are moderating variables In
the determination of self-concepts. Therefore, it is an oversimplication
to generalize and say that black self-concepts are necessarily less
favorable than white self-concepts.
In a recent review concerning self-concept and attitudes, covering
generally the late 1960's, Christmas (1973) found that the research
surveyed fell into three major categories which corroborate the
importance of moderating variables on self-concept:
The first was concerned with the influence of antecedent
factors upon current aspects of self-concept. In these
studies, self-concept was the inferred consequent of familial,
parent-child, or other social interaction; of counseling, guid-
ance or role modeling; or of variations in learning
experience or educational practice (Allen, 1969; Crovetto,
1968; Crovetto, Fischer and Boudreaux, 1967; Henderson,
1967; Talley, 1968). A number of Investigations explored
the possible effects of Integrated, desegregated and
segregated schooling on self-attitudes (Bass, 1969;
Blenvenu, 1968; McWhirt, 1967; Strauss, 1967; Taylor, 1968).
The findings were generally inconclusive.
The second major group included studies in which antecedent
self-concept was presumably related to behavior. They
included investigations of self-evaluation and self-esteem
(Guggenheim, 1969; Soares and Soares, 1969; Williams and
Byars, 1969, T970); differences between blacks and whites
in regard to ethnocentrism and self-acceptance (Freeman et al.,
1966; Getter and Satow, 1969; Greenwald and Oppenheim,
1968; Hraba and Grant, 1970); and studies of level-of-
aspiration performance in learning tasks and academic
achievement (Blair, 1972; Caplin, 1966, 1969; Curtis,. 1967;
Freyberg and Shapiro, 1966; Gay, 1966; Greenberg et a.,
1965; Lourenso, Greenberg and Davidson, 1965). Here, too,
results were inconclusive and failed to show a definite
relationship between self-concept, broadly defined, and
Finally, a number of research efforts were directed
toward determining possible correlations between self-
esteem and variables such as occupational level, socio-
economic status, residence, age, and sex, in blacks and
whites (McDonald and Gynther, 1965; Wendland, 1969; Wylie
and Hutchins, 1967; Yeatts, 1968). Here, at least,
recognition was given to the possibility, for example,
that socioeconomic status might outweigh the factor of
race. Yet, several investigators still compared blacks
of lower socioeconomic status with middle class whites,
while acknowledging this to be a limitation of the study
(Long and Henderson, 1968). The weakness of the design and
the methods of data analysis in these studies contributed
to the resultant lack of solidity In the findings.
Considering these findings on self-concept as a whole and even though
it is evident that black self-concepts are not necessarily less favorable
than white self-concepts, the evidence seems to favor the prediction
that whites will have greater self-esteem than will blacks. Since
leaders high in self-esteem possess a greater transactional asset than
leaders low in self-esteem it is reasonable to expect better task
performance from groups they supervise.
Locus of Control
It is often stated that locus of control Is an important determinant
of individual behavior. For example, two review articles (Lefcourt,
1966; Rotter, 1966) have shown that when a person believes that
reinforcements are controlled by internal rather than external forces,
he is likely to make greater attempts at mastering the environment; to
be more resistant to influence attempts by others, yet more effective in
attempts to influence others; to prefer high-probability choices in
risk-taking behavior; to be lower in anxiety and higher in achievement
orientation; to act more responsively to probability changes in the
situation; to place higher value on skill determined rewards; and to be
involved in social action (Lao, 1970). Internal-external control is
consistently related to a variety of scales, with internal scorers
describing themselves as more active, striving, achievingpowerful,
independent and effective (Hersch and Scheibe, 1967). Also, Internals
are more active in intellectual pursuits and show greater interest in
intellectual activities (Crandall, Katkovsky and Crandall, 1965;
Crandall, Katkovsky and Preston, 1962).
The research relating to this dimension in groups of blacks and
whites has consistently demonstrated class and racial differences (see,
e.g., Lao, 1970; Lefcourt, 1965; Lefcourt and Ladwig, 1965b), with a
sense of internal control being stronger in the middle class than in
the working class and stronger in whites than in blacks (Battle and
Rotter, 1963; Coleman et al., 1966; Crandall, Katovsky and Crandall,
1965; Gurin, 1970; Lefcourt and Ladwig, 1965a). The interaction of
class and race is such that lower class blacks stand out as having
a particularly external orientation (Battle and Rotter, 1963; Coleman
et al., 1966; Lefcourt and Ladwig, 1965a).
While one study found no racial differences between blacks and
whites in locus of control (Hall, 1969), and even though blacks in one
situation behaved in a more cautious, Internal fashion than did
whites (Lefcourt, 1965), blacks as a group seem to feel and behave as
if they have limited control of reinforcement. However, as Dreger and
Miller (1968) suggest, this reaction may be a function of the type of
situation being studied. Higher achievement scores and grades, greater
academic confidence and higher expectations and aspirations have been
reported on the part of black students with a strong feeling of personal
control (Gurinet al., 1969; Lao, 1970). In contrast to studies using
general locus of control measures, those using the Intellectual Achieve-
ment Responsibility scale (Crandall, Katkovsky and Crandall, 1965)
report no race effects (Katz, 1967; Solomon, Houlihan and Parelius, 1969)
and only very slight social class effects (Crandall, Katkovsky and Crandall,
1965; Solomon, Houlihan and Parelius, 1969).
Based on the above, it is reasonable to expect that white leaders
will be higher in internal control than black leaders. Further,
regardless of leader race, the performance of groups supervised by
leaders high in internal control should be better than that of groups
supervised by leaders low in this dimension because they have a
greater transactional asset and because they should be more effective
in influencing others.
Intell ience and Aptltude
The contribution of intelligence to leadership appears to be small
(cf., Shaw, 1971), but the review by Stogdill (1948) overwhelmingly
indicated that leaders are, on the average, more intelligent than
nonleaders. A low positive correlation between tested intelligence and
leadership behavior is indicated (Bass and Wurster, 1953a, 1953b;
Hollander, 1954). If we can view aptitude for learning as being a
subcategory of intelligence, similar results can be expected.
That blacks, as a group, show up as deficient in abstract abilities
and score below whites on most measures of academic achievement and
aptitude is well documented (APA, 1969; Baughman and Dahlstrom, 1968;
Coleman et al., 1966; Dreger and Miller, 1968; Epps, 1969; McFann, 1970;
Pettigrew, 1964);4 however, the significance of this to their performance
as leaders has not been investigated.
With the notable exception above, the ability of tests in general
to distinguish between blacks and whites appears to be subject to
numerous moderating variables; further, their ability to predict per-
formances for either race appears to be somewhat unreliable. These
conclusions are suggested by evidence such as the following:
(1) On personality-type tests, differences tend to disappear
when black and white subjects are matched on such variables
as intelligence (Thumin and Goldman, 1968) or socioeconomic
status (Flanagan and Lewis, 1969).
(2) Aptitude differences between blacks and whites have remained
even when the subjects were matched on socioeconomic status
(Flanagan and Lewis, 1969).
(3) While some studies strongly confirm that verbal and mathematics
tests predict scholastic performance equally well for blacks and
whites (Boney, 1966; Cleary, 1966; Cleary and Hilton, 1966;
Sunday, 1965; Stanley and Porter, 1967), other studies have
yielded no clear trends in predicting job performance (Gordon,
1955; Kirkpatrick et al., 1967; Tenopyr, 1967), and thus no
firm conclusions are possible (APA, 1969).
(4) Blacks have scored lower than whites on qualification tests
and then have performed equally well on job sample and job
knowledge tests (McFann, 1970).
(5) Blacks have been hired for certain positions for which their
measured intelligence disqualified them and then have gone
ahead and performed satisfactorily ("Merit for Hire," 1958).
(6) The contention that blacks are handicapped on tests of spatial
ability because of an environmental disadvantage has not
always been supported (Osborne and Gregor, 1968).
4The APA report (1969) offers five possible factors as to why blacks
typically score below whites: cultural deprivation, anxiety induced by
the testing situation, unfairness of test content, improper interpretation
of test scores, and lack of content relevance.
The positive correlation between tested intelligence and leader
behavior combined with the consideration that a more intelligent person
will have a greater transactional asset than will a less intelligent
person suggests that it is reasonable to expect groups supervised by
leaders high in intelligence to outperform groups supervised by leaders
low in intelligence. This is also suggested by the consideration that
leaders with more intelligence will be able to comprehend and execute
task demands with greater facility.
Depending on the demographic stratum to which they belong, blacks
may have higher, the same or lower job satisfaction that whites
(Feldman, 1972; Kahl and Goering, 1971; Katzell, Ewen and Korman, 1970;
Milutinovlch, 1971). Although black job attitudes less favorable than
those of whites have been found (Katzell, Ewen and Korman, 1970), the
evidence heavily favors the conclusion that black workers have job
attitudes as favorable or even more favorable than those of whites
(Feldman, 1972; Katzell, Ewen and Korman, 1970; Nelson, Achabel and
Investigating the relationship of race and job satisfaction and
the influence of participative and authoritative supervisory styles on
the job satisfaction of blacks and whites, Milutinovich (1971) found
that race has only limited influence on job satisfaction and that both
blacks and whites had higher job satisfaction under participative than
under authoritative supervision. Black blue- and white-collar workers
had similar job satisfaction, while white blue-collar workers were lower
than white-collar workers. Blacks, regardless of job type, had higher
job satisfaction than white blue-collar workers, but had lower satisfaction
than white white-collar workers.
In an extensive exploratory study of hardcore unemployed and
working-class blacks and whites, Feldman (1972) found that black subjects
evaluated several material and social job outcomes more highly than
whites and black working-class subjects rated a variety of outcomes,
including some higher order ones (Maslow, 1954), more highly than any
other group. He summarized his general pattern of results as suggesting
that the black subjects, particularly the working-class, perceive work
as a source of valued rewards, while the white working-class does not.
If differences in satisfaction between black and white leaders are
predicted the more plausible prediction seems to be that black leaders
will indicate greater satisfaction with work and subordinates than
will white leaders. This is suggested by the fact that in this study
the subjects (i.e., Navy enlisted personnel) should typically have
their origins in predominately working-class/blue-collar environments.
The data suggests that blacks from these origins probably have greater
work-related satisfaction than whites with similar backgrounds.
Miscellaneous Related Research
Much of the available research, while not falling into any of
the previous categories, suggests important differences between blacks
and whites. Although In one study (Veroff et al., 1960) blacks and
whites were shown to be similar, it appears that blacks are less
achievement oriented than whites (Lott and Lott, 1963; Merbaum, 1962;
Mingione, 1965; Rosen, 1959). Consistent with recent studies of black
consciousness and black politics where power is the central theme
(Dizard, 1970; Kelman, 1970; Tomlinson, 1970), blacks have been shown to
have a higher need for power (Greene and Winter, 1971; Veroff at al., 1960).
Further, blacks, relative to whites, (a) generally score lower on need
for dominance and autonomy (Brazziel, 1964) and self-liking (Clark,
1967), (b) are similar on need for affiliation (Veroff et al., 1960) or
high in affiliation (Ferman, Kornbuk, and Miller, 1968), (c) have a
higher degree of self-doubt (Deutsch, 1960) and social conformity
(Sistrunk, 1971), (d) exhibit essentially the same creative talent
(Doyle, 1970) and (e) place a greater emphasis on verbal interaction
Results presented by two studies (Broom and Glenn, 1966; Cameron,
1971) weaken the notion of a separate subculture insofar as such a
culture is indicated by opinion and personality differences. Indeed,
based on these studies, blacks and whites appear far more similar
psychologically than different. Analyzing responses to questions asked
on national public opinion surveys, Broom and Glenn (1966) found, in
general, that black and white differences in attitudes were smaller
than the differences between Southern and non-Southern whites and
between low- and high-education whites. In the other study, Cameron
(1971), using various instruments, found that blacks tested the same as
whites on masculinity-femininity, extraversion, ego-strength and liking
of the generalized other; higher than whites on religiosity and "claimed
judged liking of the generalized other"; lower than whites on neuroticism
and hostility; and, on a lie scale, there was a slight tendency for
blacks to be more candid. Parallel sex and developmental differences
suggested to Cameron that the differential social influences are much
the same and are handled psychologically fhe same for either race.
From these results we can conclude that blacks have stronger power
orientations, are less achievement oriented, emphasize verbal interaction
more and are probably psychologically similar to whites.
Relating the above now to the present study, black leaders can be
predicted to have a longer speech duration than white leaders when
directing groups in task accomplishment because of the greater emphasis
blacks place on verbal interaction.
The studies reviewed, particularly those concerning black-white
differences, represent many research designs, populations, situations
and methodologies. Although their results can be considered as suggestive
of enduring differences and thus helpful in understanding black-white
interaction In the work environment, additional research is necessary
to better establish these relationships, particularly in view of the
apparent change in the attitudes and behaviors of blacks since the
mid-1960's. The relative scarcity of blacks in positions of leadership
and supervision has seriously limited the knowledge in this area.
Based on the literature review, hypotheses relevant to this study
can now be formally advanced. For ease of reference, they have been
categorized into two groups: group I deals with differences between
black and white leaders; group II with differences in task performance
depending upon specific leader attributes.
Group I: BETWEEN BLACK AND WHITE LEADERS
A. Groups supervised by white leaders will outperform groups
supervised by black leaders in accomplishing assigned tasks
(where the shortest time for task accomplishment is the
B. White leaders, as compared to black leaders, will: (a) exhibit
more interpersonal behavior, (b) score higher in self-esteem,
and (c) score higher in internal control.
C. Black leaders, as compared to white leaders, will: (a) have
a greater duration of speech, (b) indicate more satisfaction
with subordinates, and (c) indicate more satisfaction with work
in the tasks.
Group II: BETWEEN LEADERS OF THE SAME RACE
For each task, the performance of groups of equal size and of
equal racial composition will be faster when supervised by leaders
who: (a) perceive themselves high in consideration, (b) perceive
themselves high in initiating structure, (c) are rated by their
subordinates as high in consideration, (d) are rated by their subordinates
as high in initiating structure, (e) are high in self-esteem, (f) are
high in internal control, (g) are high in intelligence, compared to
other leaders of the same race who, respectively, are low In these
Considering the findings and evidence presented in Chapter I as a
whole, it is evident that research and intuition are not clear as to
just what enduring differences exist between blacks and whites in
instances of leadership. This problem has been exacerbated by the
paucity of systematic, empirical information on the subject of black-
white leadership differences. Thus, this area is rather fertile for
study and an investigation focused here is most timely and relevant.
In Chapter II the methodology is presented in detail. Chapter III
presents the analysis and results. Chapter IV contains a summary,
discussion and implications for future research.
The methodology used in this study is presented in this chapter
in the following order: (a) subjects, (b) physical environment, (c)
design, (d) tasks, (e) observer training and reliability, (f) procedure
and (g) questionnaire. The study was carried out on the following
dates: August 28-31; September I, 4-8, 11, 14, 1972. Training for
investigator assistants was conducted on August 28 and 29; data
collection was accomplished during the other days.
The cooperation of the Commanding Officer of a large U. S. Naval
Training Center located in the Southeastern United States was obtained
so that recruits attached to the Recruit Training Command could serve
as subjects. A total of 288 male recruits, 144 black and 144 white,
participated in the study. Sixty-four of these were squad leaders, 32
black and 32 white. A profile of the black and white squad leaders
and members obtained from a biographical questionnaire is shown in
All of the subjects were engaged in the seven week basic training
program prescribed by the Curriculum for Recruit Training (NAVPERS 92353B,
PERSONAL PROFILE OF SQUAD LEADERS AND SQUAD MEMB.ERS.
Squad Leaders .. Squad Members
Black White Black White
(N=3) (N ) Nr2) (NT
Marital Status (Percent)
Average number of Dependents
Region of Birth (Percent)
Region where Raised (Percent)
Type of Community where Raised Percer
Perception of Family Income (Percent)
Higher than Average
Less than Average
Very Much Below Average
Mostly on Welfare
Raised by (Percent)
Mother and Father
Average Number of Siblings
Perception of Social Class (Percent)
Father's Occupation (Percent)
...... .2 .
aAll percentages are rounded to the nearest whole percentage point.
In some classifications the percentages do not equal 100 due to rounding.
1968). Recruit training companies were organized as shown in Figure 1.
Squad leaders and occupants of other company positions are selected by
the company commander (who is usually a chief or first-class petty
officer) during the first few days of recruit training, usually on the
basis of the recruits' vita and initial interviews. The position of
squad leader was selected as the level of leadership comparison in order
to obtain a large enough sample for statistical reliability. According
to the Company Commander's Guide (CRUITRACOMORLINST 5400.1C, 1972),
squad leaders during recruit training have the following as their formal
duties and responsibilities:
(a) Be in charge of and responsible for the men in their squads.
(b) Assist and supervise the men in their squads in attaining high
standards of personal cleanliness, military appearance, conduct
and academic achievement.
(c) Assume responsibility, under the duty platoon leader, for the
performance of their squad on cleaning details.
(d) Insure that the yeoman is kept informed of personnel changes
within their squads.
Random sampling for the selection of participants was not possible
due to the rigor of their training schedule and the uneven distribution
of blacks in some companies. Altogether, thirty companies representing
approximately 2,150 men were canvassed to arrive at the final set of
subjects. To reduce variability in the length of time the squad leaders
and members had been in training as much as possible, selection of the
sample was further restricted to those companies engaged in the fifth day,
fourth week and In the first, second and fourth days, fifth week of
training. These days corresponded to the latest availability possible
for the recruits before entering the latter portion of their training,
during which no periods of availability could be scheduled.
Recruit Company Organization
Recruit Chief Petty Officer
First Platoon Leader
Second Platoon Leader
Squad Squad Squad
Leader Leader Leader
*Squad *Squad '*Squad
Members. Members Members
Squad Squad Squad
Leader Leader Leader
"Squad *Squad .'Squad
Members Members Members
*Note: Approximately 8 squad members made up the typical squad.
The Navy provided a coordinator as liaison between the company
commanders and the researchers. The coordinator, SMCS Jerome F.
Dederick USN, scheduled groups from the appropriate companies on a
day-by-day and as-available basis. He had recently been a company
commander himself and was held in high esteem by his colleagues. Because
of this, all 30 company commanders providing squads for the study were
cooperative and harmonious relations prevailed throughout the duration
of the study.
In selecting groups for study on a particular day, the coordinator
would, during the previous afternoon or evening, ascertain (by
consulting the daily master training schedule) which companies would be
available for possible participation (i.e., to find out those engaged
In the fifth day, fourth week and in the first, second and fourth days,
fifth week of their training and not scheduled for any special activity).
The coordinator, after checking with the investigator to find out what
additional types of groups were needed at that point, would then
contact the commanders of the available companies to solicit their
cooperation. Then the company commander was asked for the racial
distribution of squad leaders and members, and if or when those squads
meeting racial composition needs were available. The company commanders
were told only that the squads were needed for a study concerning the
problem-solving effectiveness of small groups; no mention was made of
leadership or race (although the interracial aspects were undoubtedly
guessed). The company commanders in turn decided which squad leaders
and squad members were to participate. Selection depended upon the
personal availability of the individual subject, not in the sense of
their volunteering to participate but more in the sense of whether or
not they were Individually scheduled for some other activity. No known
consistent biasing occurred because the company commanders were not
told how to select the participants.
This particular setting for the experimental study offered several
(1) the population from which the subjects were drawn represented
a fairly wide cross-section of young males, both black and
white (although nearly all of the blacks turned out to be born
and raised in the South; see Table ).
(2) the leaders of the groups studied were, in fact, squad leaders,
who would have been appointed to their position in the same
manner whether or not a study was being conducted;
(3) similarly, the subordinates were, in fact, members of the
squad led by the squad leader who was the leader during the
(4) the phenomena under investigation were studied as they existed,
providing a "natural experiment" as described by Shaw (1971).
The study was conducted in the Conference Facility at the U.S.
Naval Training Center. Two rooms, each with a one-way window, were
utilized (see Figure 2). Trained observers were able to observe the
participants without disturbing or distracting them. Each room was
(a) a rectangular table and five chairs for the participants,
(b) a desk and chair for the investigator's inside assistant,
o m -
0 .U) 4J
0o 0 -
V0 L L. t .
41 I !
0 0- = 0)
to V 0U -X 0I
S. > O Cl O 0
C- 0.- 0.IDC
N O L
m \ VI M
L.Iin in L -L
I I I EE
a 1U0 10 "
m l a m o v mQ
\ a f o U U U
_o__________ o -*c-J pf
(c) lengths of rope for knot-tying,
(d) a microphone, on the rectangular table, connected to a
speaker/amplifier operated by the observers on the other
side of the one-way mirror,
(e) stop watches mounted In a specially fabricated rack for
manipulation by the investigator assistant,
(f) a suitable supply of necessary forms and instruction cards.
The observation station outside for each room was provided with:
(a) suitable seating for the observers near the one-way window,
(b) a speaker/amplifier located at the observation station,
(c) recording forms and two large clipboards for the observers.
There were 64 subordinate groups in the study:
(1) 16 two member subordinate groups, one black and one
white (designated a Type 0 group);
(2) 16 four member subordinate groups, one black and three
white (designated a Type 1 group);
(3) 16 four member subordinate groups, two black and two
white (designated a Type 2 group);
(4) 16 four member subordinate groups, three black and one
white (designated a Type 3 group).
Eight different black and eight different white squad leaders supervised
one of the four types of subordinate groups shown above, allowing 64
leaders In all to participate. In way of clarification, the term
group denotes a group that includes the leader plus the subordinates;
the term subordinate group denotes a group consisting of subordinates
The study addressed itself to the effects of three factors: (1)
leader race, (2) group size and (3) group racial composition. Leader
race was varied by having equal numbers of black and white squad leaders
in charge of groups of equal size and groups of equal racial composition.
Group size was varied by holding racial composition constant at 50
percent and varying the number in the subordinate group from two to four.
Group racial composition was varied by holding subordinate group size
constant at four and varying the number of members of each race from
one to three (and thus racial composition was 25 percent, 50 percent and
75 percent black or white). The four-member size and 50 percent racial
composition subordinate group is common to both the analysis of data
with subordinate group size varying and the analysis of data with
subordinate group racial composition varying. Otherwise, size and racial
composition would be confounding variables. Data analysis between
subordinate groups of varying size but equal racial composition was
accomplished by utilizing Type 0 and Type 2 groups, and between
subordinate groups of varying racial composition but equal size,
Types 1, 2 and 3. Basically, the result was a 2-X-4 factorial design
as shown in Figure 3.
Within the framework of the 2-X-4 design, the following were
(1) interpersonal behavior (defined to be the group interaction
(2) leader's style of leadership (directive versus non-directive),
(3) leader's social-emotional behavior (positive versus negative),
(4) group performance for two tasks (i.e., the time for completing
(5) leader's and subordinate's duration of speech during each of the
(6) leader's and subordinate's self-esteem,
(7) leader's satisfaction with the subordinates and the tasks,
Schema of the Basic Research Design
Number of Groups
Number of Groups
B = black subject
W = white subject
(8) subordinate's satisfaction with the leader, the tasks and
(9) leader's perception of his initiating structure and consideration,
(10) subordinate's perception of the leader's initiating structure
(11) leader's and subordinate's generalized expectancies for
internal versus external control,
(12) leader's general intelligence.
The Bales (1950) Interaction Process Analysis was used to measure
the group interaction processes (cf., Zdep, 1969; Richards and Jaffee,
1972; Katzell et al., 1970; Ruhe, 1972). Initiations of interaction
in each of the 12 IPA categories by all leaders and subordinates were
recorded by trained observers while the tasks were being performed.
Two observers were used in scoring the IPA to establish reliability for
the observation process. In this situation one black and one white male
observer were used to observe each group session In an effort to
compensate for possible perceptual differences. No female observers were
Certain categories of the IPA have been interpreted as reflecting
directive and non-directive leadership (Katzell et al., 1970). IPA
categories 4 (giving suggestions), 5 (giving opinion) and 6 (giving
information) were interpreted as directive, while categories 7 (asking
for information), 8 (asking for opinion) and 9 (asking for suggestions)
were interpreted as non-directive. Style of leadership was assumed to
be measured by these categories.
In a similar way IPA categories 1 (shows solidarity), 2 (shows
tension release) and 3 (agrees) were interpreted as positive social-
emotional behavior in the same sense as presented by Bales (1950), while
categories 10 (disagrees), 11 (shows tension) and 12 (shows antagonism)
were interpreted as negative social-emotional behavior.
The time required to complete each of the two tasks was used as a
measure of the group performance under the direction of the squad leader.
The time required was recorded, with the assumption that the less the
time needed for task accomplishment, the better the performance. No
quality considerations were given to group performance.
Speech duration for each group member during both tasks was
recorded on time-accumulating stop watches activated by the investigator's
inside assistant. This method has been shown to be a reliable technique
for recording absolute duration of speech (Bass, 1960; Jaffee, Cohen
and Cherry, 1972). This measure was assumed to reflect the degree of
participation by the individual group members.
Self-esteem of each group member was measured by two instruments:
(1) the self-esteem portion (adult and student form) of the Self-Other
Orientation Tasks of Ziller (1971) (see Appendix D) and (2) the Self-
Esteem Scale of Rosenberg (1965) (see Appendix E).
Self-esteem, according to Ziller (1968), is the component of the self
system concerning the Individual's perception of his worth with:in.a
social context. The stem presents a horizontal array of six circles and
a list of significant other people such as a friend, a selfish person,
grandmother, someone you hope to be like, a principal and yourself.
The task requires the subject to assign each person to a circle. The
score is the weighted position of the self. In accordance with the
cultural norm, positions to the left are associated with higher self-
esteem (Ziller, 1971). This assumption has been examined and validated
in a series of separate studies (cf., Ziller, Megas and DeCencio, 1964;
Henderson, Long and Ziller, 1965; Mossman and Ziller, 1968; Ziller et al.,
In contrast to Ziller's instrument, Rosenberg's Onmitem
Guttman scale is designed to measure attitude toward the self along
a favorable-to-unfavorable dimension. Rosenberg designed the scale
with several criteria in mind, one being his conception of self-
esteem (Rosenberg, 1965):
When we speak of high self-esteem, then, we
shall simply mean that the individual respects himself,
considers himself worthy; he does.not necessarily con-
sider himself better than others, but he definitely does
not consider himself worse; he does not feel that he is
the ultimate in perfection but, on the contrary, recog-
nizes his limitations and expects to grow and improve.
Low self-esteem, on the other hand, implies self-
rejection, self-dissatisfaction, self-contempt. The
individual lacks respect for the self he observes. The
self-picture is disagreeable, and he wishes it were
The 10 Items, all of the Likert type, are scored through the use of
contrived items (Stouffer et al]., 1952) to yield a seven point scale.
Rosenberg (1965), before the larger study reported therein, pretested
the instrument and found a significant association between the
individual's self-esteem and the likelihood that he would appear
depressed to others. Not only were people with low self-esteem
scores more likely to appear depressed to others but they were also
more likely to express feelings of unhappiness and discouragement .
A pilot sociometric investigation of 272 high school seniors (also
reported In Rosenberg, 1965) found that individuals rated low in
self-esteem by the instrument: (1) were less likely to be selected
as a leader by two or more classmates; (2) were judged as less
likely to be chosen by others if an election were held; (3) were
less likely to be described as active participants in classroom
discussions; and (4) were more likely to be described as relatively
subdued and inactive in classroom discussions. In the larger study
of slightly over 5,000 high school juniors and seniors (Rosenberg,
1965), he found that students with low self-esteem were less likely
than those with high self-esteem to be active participants in
formal groups, to be active and frequent participants in informal
discussions, to be informal opinion leaders, and to be formal group
leaders. He also mentions in a footnote (p. 30) that a study by
E. Silber and J. S. Tippett showed a test-retest reliability of .85
for a group of college students retested after two weeks. The Rosen-
berg instrument was included in addition to the Ziller instrument
because it is based upon verbal self-report rather than spatial
Individual satisfaction with subordinates, tasks, leader,
and/or fellow subordinates was measured using selected scales of
the Job Descriptive Index (JDI) described by Smith (1967) and
Smith, Kendall. and Hulin (1969). The JDI in its complete form
measures satisfaction in five areas: type of work, pay, oppor-
tunities for promotion, supervision. and co-workers. For each area
there is a list of adjectives or short phrases, and the respondent
is instructedd to indicate whether or not each word or phrase applies
to the facet in question (e.g., the supervision) by circling Y
for yes or N for no. If he cannot decide if a word or phrase
applies, he is asked to circle a question mark. All group members
completed the work scale (see Appendix H) as a measure of satisfaction
with the tasks. Only the squad members completed the supervision
scale (see Appendix I) as a measure of satisfaction with the squad
leader. Both the squad leaders and the squad members completed
the co-workers scale; the leaders rated their subordinates indi-
vidually (see Appendix J), while the squad members rated their
fellow subordinates (see Appendix K). Smith, Kendall and Hulin feel
that the JDI has several advantages as a measure of satisfaction:
it is directed toward specific areas of satisfaction rather than
global or general satisfaction, the verbal level required to answer
the JDI is quite low, and the responses are not self-referent.
There is a deliberate attempt to avoid the use of a self-referent,
since the basic needs or drives of the organism or their relevance
to satisfaction are not clearly established.
Assessment of the squad leader's perception of his own initi-
ating structure and consideration was accomplished through the use
of the Leadership Opinion Questionnaire (see Appendix F), described
by Fleishman (1957b). This 40-item instrument was designed to
provide independent measures of initiating structure and considera-
tion in leadership-group situations where primary concern is in the
assessment of leadership attitudes. Internal consistency as well
as test-retest reliability has been assessed (DiVesta, 1954; Fleish-
man, Harris. and Burtt, 1955; Harris and Fleishman, 1955); reliability
for initiating structure was .67-.74, and for consideration, .77-
.80. Validity has been evaluated through correlations with inde-
pendent leadership measures, such as merit rating by supervisors,
peer ratings, forced-choice performance reports by management and
leaderless group situation tests (Fleishman, 1957b).
Assessment of the leader's initiating structure and consideration
by subordinates was accomplished through the use of the Supervisory
Behavior Description (see Appendix G), described by Fleishman
(1957a). This 48-item instrument is meant to parallel the Leadership
Opinion Questionnaire in scope and purpose. Reliability has been
assessed in terms of internal consistency, inter-rater agreement,
and stability of repeated measures over time (Fleishman, Harris, and
Burtt, 1955; Harris and Fleishman, 1955); reliability for initiating
structure was .46-.53, and for consideration, .55-.58. Its validity
has also been assessed through correlations with independent leadership
measures, such as objective group indices (absenteeism, turnover),
productivity ratings, peer ratings. and leadership group situation.
tests (Fleishman, 1957a).
Both the Leadership Opinion Questionnaire and Supervisory
Behavior Description are considered independent of the general
intelligence of respondents (Fleishman, 1957a, 1957b). While both
initiating structure and consideration were found to be orthogonal
in factor analyses by Fleishman (1953) and Halpin and Winer (1957),
which has been interpreted as suggesting that the two are Independent
(Fleishman, 1953, 1957a; Fleishman and Harris, 1962; Stanton, 1960),
recent evidence (Lowin, Hrapchak. and Kavanagh, 1969) suggests a
relationship between them.
Expectancy for internal versus external control was measured
by Rotter's (1966) Internal-External Control Scale (see Appendix C).
The instrument and its developmental history is documented in
Rotter's (1966) monograph. In Rotter's theory, the control construct
is considered a generalized expectancy, operating across a large
number of situations, which relates to whether or not the indi-
vidual possesses or lacks power over what happens to him.
Individuals are conceived to vary along a "locus of control"
dimension, with the end points labeled internal and external. The
29-item, forced-choice scale has a test-retest reliability that is
consistent and acceptable. Little relationship has been found
between intelligence and control measures (Hersch and Scheibe,
1967; Lefcourt, 1966; Rotter, 1966). There is, however, some evi-
dence that intelligence is positively related to perceived internal
control (Bialer, 1961; Crandall, Katkovsky, and Preston, 1962).
The general Intelligence of leaders was measured by the U.S.
Navy Basic Test Battery (BTB). given to all recruits upon entering
the recruit training program. This battery consists of four
pencil-and-paper examinations: (1) the general classification test
(GCT), (2) an arithmetic test (ARI), (3) a mechanical ability test
(MECH), (4) a clerical ability test (CLER). The classification
tests are primarily designed to measure an individual's aptitude
or capability for learning, rather than his achievement. For the
BTB, the raw score Is converted to a Navy Standard Score (NSS).
According to the Manual of Enlisted Classification Procedures
(NAVPERS 15812B, 1970), the NSS is designed to provide standardiza-
tion in test score recording and Is meant to ease the function of
score comprehension, comparison, and interpretation; it is distributed
normally, with a mean of 50 and a standard deviation of 10. The
BTB is designed primarily to determine minimum eligibility for
These scores were extracted from recruit records. The sum of
the four individual scores on the GCT, ARI, MECH. and CLER tests
was considered a measure of general intellectual ability.
Two structured, verbal tasks were employed In this study: a
knot-tying and a ship-routing task. Both were selected because of
their applicability to a situation involving Navy recruits, and to
assure interaction between the leaders and subordinates in accom-
plishing the tasks. They differ largely in the degree to which
verbal skill is required. The knot-tying task was instructional and
emphasized a certain degree of spatial orientation and physical
dexterity. The ship-routing task required the coordination of
information provided only to the subordinates and emphasized a certain
degree of mental agility. Both types of tasks were used by Ruhe
(1972); the ship-routing task is similar to that used by Fiedler
The knot-tying task required that group members each perform
satisfactorily the physical task of tying a Double Carrick Bend.
This unusual knot, used in the maritime world for joining two hawsers
together (a hawser is a large rope by which a ship is towed), was
selected in an effort to eliminate the possibility of prior knowledge
by any of the subjects, due perhaps to membership in various youth
groups or an early nautical interest.
Prior to commencement of the task, the leader was given an
instruction sheet (see Appendix M), face down, displaying how to tie
the knot, and each group member was given two lengths of rope. After
receiving directions for task accomplishment (see Appendix N) from
the investigator's assistant, the signal to begin was given. The
leader instructed the subordinates how to tie the knot and they were
asked to duplicate it. Group members were permitted to give verbal
but not physical help so that group effort and verbal interaction
would be enhanced. After all subordinates had tied the knot, and
upon a signal from the leader, the assistant determined whether or
not the knot had been tied correctly by each subordinate. The per-
formance of the group was measured by the total elapsed time for the
instruction by the leader and the duplication of the knot by the
subordinates. If the task was not completed within 25 minutes,
the assistant was instructed to take the leader outside the room
and show him how to properly tie the knot. The leader then would
return and continue instructions.
The ship-routing task has been rated high on cooperation re-
quirements, decision verifiability, and intellectual-manipulative
requirements (Shaw, 1963). In this problem each group was asked to
find, in the least amount of time, the shortest route for a ship to
leave one port and touch at each of four other ports.
Each member was given a copy of the instruction sheet (see
Appendix 0) showing a diagram designed to assist in arriving at the
shortest route combination. Pencils were also provided. With the
same intent as Fiedler (1967), i.e., assuring that group members
will have to interact to solve the problem, each subordinate received
only partial information concerning the distances and availability
of the routes between ports. For a five-port problem, there are 10
elements of information (e.g., the distance from A to B is 150 miles
by direct route). Each element was listed on a small index card
(see Appendix P). In addition, two cards with no information were
included where the total number of cards would be divisible by two
and by four. For a three-man group, each squad member was given six
information cards, randomly chosen; for a five-man group, three
cards, also randomly chosen. Cards could not be exchanged; all
information had to be transmitted verbally. A second ship-routing
problem, having information and a solution exactly double that shown
in Appendix P, was used whenever a second group or groups from the
same company performed the task subsequent to a first group or groups
without an .intervening time period. This was done in case the
solution to the first problem was compromised when the first group
or groups returned to the barracks. Between company communication
was virtually zero under regular circumstances. However, within
company communication was greater because each company occupied one
large room in the barracks.
After receiving directions for task accomplishment (see
Appendix N) from the assistant, the signal to begin was given.
When the group members were satisfied they had reached a correct
solution, the leader signaled the assistant. The assistant then
determined if the solution was correct. If the correct solution was
not reached in three attempts, the task instructions were repeated
and the group was allowed to continue. If the correct solution was
not reached after two more attempts, the squad leader was given a
copy of the available route combinations (see Appendix Q) and the
group was allowed to continue. The performance of the group was
measured by the total elapsed time required to solve the problem.
Observer Training and Reliability
Eight men, four black and four white, were hired to assist the
investigator. All were trained as IPA observers and investigator
assistants. Of the four blacks, two (both 24 years old) held
Bachelor's degrees and were seeking admission to :graduate school,
while the other two (both 21 years old) were upper division under-
graduate students. All of the four whites were graduate students;
they were 22, 23, 23, and 26 years of age. One of the white ob-
servers was familiar with the IPA and provided valuable assistance
during the training period and during the subsequent data collection.
An additional graduate student was hired only to assist in the
training of observers. Chapter 3 and the Appendix of Bales (1950)
were used as the fundamental training materials. Each observer was
provided a personal copy of this material for study and consultation.
After being familiarized with the 12 IPA categories and the
applicable definitions, the observers practiced observing and
recording the types of interactions that could be expected in each
category. Mock knot-tying and ship-routing task sessions and motion
pictures were used. Observer teams of one white and one black were
established and maintained throughout the study. Approximately
17-1/2 hours were devoted to establishing initial observer relia-
bility; after training, agreement within each observer team was 80
percent or better. After establishing initial observer reliability,
three pilot sessions using recruits as subjects served to familiarize
all concerned with the tasks, desired procedures, techniques, and
methods, and to seek out and solve operational problems. Originally,
the ship-routing task was planned as a six-port problem. During the
first three sessions it became obvious that this problem was too
difficult and time consuming to be suitable. Thus, the ship-routing
problem was changed to a five-port problem similar to that used by
When both study rooms were in use, one black/white observer
team observed the tasks in each study room; one black served as an
assistant inside one room, while one white performed the same duties
in the other room; one black and one white served as assistants
outside the study rooms in such things as administering question-
naires, coding data for subsequent analysis, and transporting
recruits. The inside assistant conducted the task sessions by means
of a prepared script. He recorded the task performance times and
the individual duration of speech by means of stop watches.
Interrater reliability between black and white observers
(Table 2) was calculated for each team, over both tasks and each
RELIABILITY BETWEEN BLACK AND WHITE OBSERVERS
1. Shows solidarity
2. Shows tension release
4. Gives suggestion
5. Gives opinion
6. Gives information
7. Asks for information
8. Asks for opinion
9. Asks for suggestion
11. Shows tension
12. Shows antagonism
No.. of subjects
1 2 4 1 2 12 4
95 74 84 35 95 74 84 35
Weighted observer reliability for knot-tying task 65
Weighted observer reliability for ship-routing task 76
Overall observer reliability 71
NOTE: All numbers are rounded to nearest whole percentage point.
IPA category, to obtain the agreement between the observers. For
each task, the average reliability over all IPA categories was
calculated for each team. Then, because the number of subjects
observed was not equal for each team, a weighted observer reliability
was calculated for each task. The overall interrater reliability of
.71 was obtained by averaging the weighted figure reliability for
each task. (All figures are based on the intraclass correlation;
[Snedecor and Cochran, 1967].)
Nearly all of the groups participating in the study had to be
transported by private automobile from the Recruit Training Command
section of the Naval Training Center to the Conference Facility, a
distance of approximately one mile. The number, size: and composi-
tion of the groups varied depending upon the availability of the
subjects. Two groups could be studied at one time; individual
sessions were conducted during the day and at night. If other groups
were waiting, arriving groups were escorted to an unused area of the
building and not permitted to come within earshot of the proceedings.
The study rooms had closed doors whenever groups were under study
and the observers were stationed at the one-way windows (which could
not be seen by waiting groups). Thus, none of the groups had any
prior knowledge of the study methods.
At the onset of the study, an effort was made to schedule the
same number of black and white leaders with black and white inside
investigator assistants and balance them between the two rooms to
minimize racial experimenter effects (Sattler, 1970). This proved
impossible when it became necessary to schedule groups at night,
since there were too many conflicts between the schedules of the
subjects and the observers. The presence of a visible black at
some time during each of the sessions was designed to mitigate any
racial experimenter effects.
Upon arrival at the Conference Facility, the participants were
escorted either to a study room or to a waiting area. As a group
entered the study room they were greeted by the assistant; the
squad leader was asked to sit at the middle position on one side of
a rectangular table, and the squad members were asked to sit alpha-
betically at positions 1 and 2 or 1, 2, 3, and 4, depending on group
size (see Figure 2). Alphabetical seating was done to randomize
the racial seating arrangement.
The purpose of the study was explained as an analysis of the
problem-solving effectiveness of groups (Katzell et al., 1970).
They were told that they would perform two tasks. The participants
were informed that two additional observers besides the assistant
giving the instructions were necessary for the analysis but had to
sit outside and listen by means of a microphone on the table because
of the physical arrangement of the room. The importance of com-
pleting the tasks as rapidly as possible was stressed by telling the
group they were in "friendly competition" with other squads. This
phrase has special meaning for the recruits. It means that they
are In competition and should perform as best they can, but no
special reward will be forthcoming. To further enhance their desire
to perform effectively, they were told that their company commander
would be informed as to how well they compared with all the squads
participating in the study. The group was told that the squad
leader was in charge due to the way they are organized during recruit
training. This was mentioned in an effort to enhance the leadership
position of the squad leader and stress the idea that the squad
leader was the one to be in charge of the group during the task
performance. The same introduction was made to each group from a
prepared script (see Appendix N).
The group then received instructions for the knot-tying task.
Individual duration of speech for the task for each group member
was recorded on time-accumulating stop watches. When a member of
the group began to talk the assistant started the watch and allowed
it to run until the member stopped talking. Thus, total verbal
participation time was measured for each member of the group.
Another stop watch recorded the total time for task completion.
After completing the first task, instructions were given for
accomplishing the ship-routing task. Again, individual duration of
speech and total task completion time were measured.
Upon completion of the second task, the group was ushered from
the study room to an unused area of the building. No conversing
with a waiting group was allowed. Each subject was given a ques-
tionnaire booklet, one type for squad leaders and another type for
squad members. After receiving brief instructions emphasizing
the need to complete every item in the booklet, the participants
were encouraged to relax and complete the questionnaire at their own
pace. Each subject was asked to complete his booklet without dis-
cussing his responses with fellow recruits. Coffee and donuts (when
available) were offered and smoking was permitted (this was a
privilege for the recruits).
After the questionnaire was completed by the group member,
it was checked by one of the assistants and then double-checked by
the investigator to assure that the individual responded to every
question and to look for response sets. When found (and this
occurred only about four or five times), the investigator would
first attempt to determine if the responses were genuine and then,
if the individual agreed that he had not been careful in responding,
he was asked to reconsider his answers. They were thanked for
participating in the study and cautioned not to discuss any details
of the study with other recruits back in the barracks. The idea of
the friendly competition was again raised in an effort to solicit
Two different questionnaire booklets were prepared: one for
squad leaders and one for squad members. Each booklet consisted
of several separate items arranged In the order Indicated by the
Internal versus External
Control Scale (IEC)
Ziller Self-Esteem Instru-
Rosenberg Self-Esteem Scale
Leadership Opinion Ques-
Job Descriptive Index for
with tasks (JDI-GT)
with squad leader (JDI-SL)
with subordinates (JDI-S)
with fellow subordinates
For example of
The various instruments were headed by the above indicated
letter designations to keep from disclosing what the particular
instrument measured, especially in the case of the two self-esteem
measures and the internal-external control measure, and for ease in
verbal identification whenever recruits were present.
The Personal Questionnaire contained a biographical data sheet
(see Table 1 for results), a short 10-item general satisfaction survey,
and asked for squad members' feelings about the squad leader, the
name of the best squad leader in the company, the rank desired
during first enlistment, and the perceived odds for making the rank.
The biographical data was requested to check on the similarity of
the subjects and to compare these subjects to those used In Ruhe's
(1972) study. The satisfaction survey was intended to assess a
general attitude toward several activities related to recruit train-
ing, home life, and life in general. The intent behind asking for a
sociometric rating of the leader by the subordinates was to see if
back and white squad members would differ in their personal liking
for their particular leader. Having all of the subjects nominate
the best squad leader in their company was intended to see if sub-
ordinates would choose their own leader more or less often than
another squad leader, and to see if leaders would choose themselves
more or less often than another squad leader. Each subject was
asked to indicate the rank he would like to reach during his first
enlistment in an effort to estimate his level of aspiration, and,
similarly, he was asked to indicate his feeling as to the odds of
making this rank in an effort to estimate his expectancy of
ANALYSIS AND RESULTS
This chapter is divided into two parts: (a) Methods of
Analysis and (b) Results.
Methods of Analysis
This part of the chapter presents the methods used in analyzing
the data and is subdivided into the following sections: (a) Inter-
action Process Analysis Scoring, (b) Between Black and White Leaders,
(c) Between Leaders of the Same Race, (d) Between Black and White
Subordinates, (e) Between Groups and (f) Additional Data.
Interaction Process Analysis Scoring
The behavioral data collected utilizing the IPA is essentially
count data possessing the characteristics of the Poisson distribution.
The basic conditions defining the Poisson distribution (Burford,
1968) are: (1) that the distribution is independent through time or
space (i.e., that the probability of occurrence of an event in any
given interval of time or space is independent of other such inter-
vals), (2) that increasing or decreasing the interval of time or
space increases or decreases proportionately the probability of
occurrence of an event, (3) that if the given interval of time or
space is small the probability of two or more occurrences of the
event within the interval is also very small and (4) that the
probability of X occurrences of the event in the interval increases
or decreases continuously as the interval increases or decreases
The raw count data observed by the black and white observers
for each subject in each IPA category was converted to frequency per
unit time for each task and then linearly transformed into whole
numbers by multiplication by a factor of 10,000 and rounding to the
nearest whole number. The transformed data was then given a square
root transformation in an effort to achieve uniform variance. The
square root of Poisson count data yields a response that will possess
approximately a constant variance independent of Its mean (Mendenhall,
The transformed count data for each of the two observers in
each IPA category were added together (by task) to arrive at a score.
Thus, for each task, every subject had 12 scores, one for each of the
IPA categories. To arrive at a score for positive social-emotional
behavior for each leader, the scores in IPA categories 1 (shows
solidarity), 2 (shows tension release) and 3 (agrees) were summed;
similarly, a score for negative social-emotional behavior was calcu-
lated by adding the scores in categories 10 (disagrees), 11 (shows
tension), and 12 (shows antagonism). To arrive at a score for
directive leadership for each leader, the scores in IPA categories
4 (giving suggestions), 5 (giving opinion),and 6 (giving information)
were summed; similarly, a score for non-directive leadership was
calculated by adding the scores in categories 7 (asking for informa-
tion),8 (asking for opinion) and 9 (asking for suggestions).
Between Black and White Leaders
Differences in the various measures between leaders for the
four group types were analyzed using a 2-X-4 analysis of variance
(ANOVA) design with eight observations per cell. An ANOVA was run
for each of the following: (a) group performance for both tasks;
(b) each IPA category for both tasks to analyze interpersonal be-
havior; (c) self-esteem (ZSE); (d) self-esteem (RSE); (e) internal
control; (f) duration of speech for both tasks; (g) satisfaction with
subordinates; (h) satisfaction with the work during both tasks;
(i) positive social-emotional behavior for both tasks; (j) negative
social-emotional behavior for both tasks; (k) directive leadership
behavior for both tasks and (1) non-directive leadership behavior
for both tasks. While the RSE score is essentially an ordinal
measure and would normally be analyzed using non-parametric tech-
niques, the number of data points on the scale (i.e., seven) is
sufficient to allow the data to be analyzed by an ANOVA due to the
robustness of the technique.
Between Leaders of the Same Race
Comparisons between leaders of the same race were analyzed
using the ANOVA method. To understand how these analyses were
accomplished the example of black leader, equal racial composition
group, is presented. The median score for each attribute
(e.g., self-rated consideration) of the black leaders for Type 0 and
Type 2 groups was established. A high attribute classification was
given to scores above the median; low, to those below the median.
Ties were broken by random selection. The performance time (for one
task) associated with the black leaders, within a group type and
classified as high versus low in the attribute, became the cell
entries for a 2-X-2 ANOVA, where the two rows represented a high
versus low classification and the two columns represented a Type 0
group versus a Type 2 group. Therefore, for black leaders, equal
racial composition situation, two 2-X-2 ANOVAs were performed:
(a) one for the knot-tying task and (b) one for the ship-routing
task. For the black leader, equal size group situation, high and low
classifications were again established and two 2-X-3 ANOVAs were run
for each attribute, where the rows represented the high versus low
classification and the columns represented Types 1, 2 and 3 groups.
Thus, four ANOVAs (i.e., two 2-X-2 for the equal racial composition
situation and two 2-X-3 for the equal size situation) were performed
for each attribute over each leader type. Specifically, a multi-
variance general linear hypothesis program, BMDX63 (Dixon, 1969),
was run for each ANOVA to Investigate the performance of subordinate
groups of equal size and equal racial composition supervised by black
and white leaders, using high and low classifications of: (a)
consideration (self-rated); (b) initiating structure (self-rated);
(c) consideration (subordinate-rated); (d) initiating structure
(subordinate-rated); (e) self-esteem, using both the ZSE and RSE
scores; (f) internal control and (g) intelligence. The subordinate-
rated scores for consideration and initiating structure were means
for the ratings given by all subordinates (black and white) for the
particular group type.
Between Black and White Subordinates
Within each subordinate group type, differences in the various
measures between subordinates were analyzed for each leader type
using the t-statistic. For every individual subordinate measure,
the average black subordinate score was subtracted from the average
white subordinate score to arrive at a "difference between" score
for each group type and leader type. A 2-X-4 ANOVA with eight
"difference between" scores per cell was then run. A t-statistic for
each cell of the 2-X-4 matrix was then computed by dividing the
individual cell mean by the square root of the quantity: mean square
error divided by the number of observations in the cell. This was
accomplished for each of the following: (a) each IPA category for
both tasks to analyze interpersonal behavior; (b) duration of speech
for both tasks; (c) self-esteem, using both ZSE and RSE scores; (d)
internal control; (e) satisfaction with fellow subordinates;
(f) satisfaction with the leader and (g) satisfaction with work in
Differences in the various measures between groups were analyzed
using the ANOVA technique. Excepting performance of the group for
both tasks (which was analyzed as a between leader measure), a
group score for each measure in question was calculated by taking
the average of the measure for the particular group type. The
average Included both leader and subordinates. A 2-X-4 ANOVA with
eight observations per cell was then run for each of the following:
(a) each IPA category for both tasks; (b) duration of speech for
both tasks; and (c) satisfaction with work in the tasks.
The data obtained by means of the Personal Questionnaire was
analyzed by various methods. The general satisfaction survey score
was analyzed in exactly the same manner as the other individual
attributes were analyzed. Analysis of the leader sociometric rating by
group members was accomplished either through the use of the t-statistic
with a pooled variance estimator (Mendenhall, 1971) or with the z-
statistic, as appropriate. Comparisons were made by group type, by
leader type.and by leader for all subordinate types. Analysis of
the nomination made by individual subjects for the best squad
leader in their company was accomplished through the use of the X2
test statistic for the chi-square distribution. Two-by-two contin-
gency tables were used to provide the test statistic for comparing:
(a) both subordinate types under black leaders, (b) both sub-
ordinate types under white leaders, (c) black subordinates nominating
their own leader as the best, (d) white subordinates nominating their
own leader as the best and (e) both leader types nominating them-
selves as the best. The X2 test statistic included Yates' correction
for continuity (Champion, 1970). The question concerning the rank
which each subject would like to reach and the question concerning
the odds to make that rank were analyzed through the use of the z-
statistic, with the following comparisons being made: (a) between
leaders, (b) between subordinates, (c) black leader to both sub-
ordinate types, (d) white leader to both subordinate types and
(e) black subjects to white subjects.
All significance testing was standardized and performed at the
.05, .01 and .001 levels of significance. If an abbreviated form is
used in presenting results, the following shorthand notation will
be used: (1) black leaders and white leaders will be designated
BL and WL, respectively; (2) black subordinates and white subordi-
nates will be designated BS and WS, respectively; (3) group types
0, 1, 2 and 3 (i.e., 2-person, equal racial composition subordinate
groups, and 4-person subordinate groups with 1, 2 or 3 blacks,
respectively) will be designated TO, TI, T2 and T3, respectively;
(4) the high/low classification within leader types will be desig-
nated HI and LO, respectively; and (5) the cells of the various two-
way matrices will be designated by the row variable and group type,
separated by a slant (for example, BL/T3 means black leader,
Type 3 group). Figure 4 summarizes both the shorthand notation
and the designs for data testing.
This part of the chapter presents the tests of the various
hypotheses and incidental data. It is divided into the following
sections: (a) Differences Between Black and White Leaders,
Notation and Designs for Data Testing
Design for testing data between black and white leaders:
Leader T T) MlT (TL
Black (BL) BL/TO BL/TI BL/T2 BL/T3
White (W) L/T W/T WL/T WL/T2 WL/T3
Design for testing data between leaders of the same race:
(A) Equal Size Subordinate Groups
High (HI) HI/TI
Low (LO) LO/TI
(B) Equal Racial Composition Subordinate Groups
Attribute Group Type
Classification (TO) 1T2)
High (HI) HI/TO HI/T2
Low (LO) LO/TO LO/T2
Design for testing data between black and white subordinates:
Design for testing data between
(TO) (Ti) .(T2) .I
WS-BS WS-BS WS-BS WS-BS
WS-BS WS-BS WS-BS WS-BS
(b) Differences Between Leaders of the Same Race and (c) Incidental
Differences Between Black and White Leaders
This section reports the results of the tests run on the Group I
Group performance. Hypotheses I.A states that groups supervised
by white leaders will outperform groups supervised by black leaders
in accomplishing assigned tasks.
Groups supervised by black leaders took significantly longer to
perform both the knot-tying task (F,56=5.15, p< .05) and the ship-
routing task (F1 56=6.61, p<.05), as compared to those supervised by
white leaders. The mean time for groups supervised by black leaders
to perform the knot-tying task was 677.10 seconds; for groups super-
vised by white leaders, 489.32 seconds. For the ship-routing task,
the mean times were 1315.47 and 883.13 seconds, respectively.
Hypothesis I.A was supported.
Interpersonal behavior. Hypothesis I.B(a) states that white
leaders will exhibit more interpersonal behavior than black leaders.
During the knot-tying task, black leaders, as compared to
white leaders, exhibited: (a) less showing of solidarity (F1,56=
9.37, p<.O1, (b) less giving of suggestions (F1,56=19.01, p<.01),
(c) less asking for information (F1,56-16.65, p<.01), (d) less
asking for opinion (F1,56=5.72, p<.05) and (e) less disagreeing
(FI,56=6.72, p<.05). (See Table 3 for the mean leader scores for
these categories.) No main effects of group type were observed.
MEAN IRA SCORES FOR LEADER RACE MAIN EFFECTS -
GROUP I HYPOTHESES
Black Leaders White Leaders
Category 1: Shows solidarity 1.58 6.16
4: Gives suggestion 25.45 40.77
7: Asks for information 2.33 8.75
8: Asks for opinion 0.80 3.76
10: Disagrees 4.57 8.81
Category 5: Gives opinion 6.50 9.96
8: Asks for opinion 2.65 6.12
A significant interaction effect (F3,56=4.13, p<.05) was found for
category 6 (gives information) showing that: (a) black leaders of
racially balanced groups gave more information as the number of their
subordinates increased from two to four (p<.05); (b) when they super-
vised four-man subordinate groups, black leaders gave less informa-
tion when the number of black subordinates increased from two to three
(p<.01); (c) white leaders gave less information to their subordinates
in racially balanced dyads than they did in four-man groups whose
composition was 75 percent black (p<.05); (d) white leaders gave
more information in four-man groups when the racial composition
changed from 25 to 75 percent black (p<.05) and (e) black as compared
to white leaders gave less information when supervising four-man
subordinate groups whose composition was 75 percent black (p<.001).
(See Table 4 for the cell means.)
During the ship-routing task, black leaders, as compared to
white leaders, exhibited: (a) less giving of opinion (F1,56 6.46,
p<.05), and (b) less asking for opinion (F1,56=9.44, p<.0).
(See Table 3 for the mean leader scores for these categories.) Main
effects were found between group types in categories 2 (shows
tension release) (F3,56C3.76, p<.05), 11 (shows tension) (F3,56=
4.73, p<.01), and 12 (shows antagonism) (F3,56=3.61, p<.05).
Leaders, regardless of race, emitted fewer tension release comments
when supervising 25 percent black tetrads than either 50 percent
black tetrads (p<.0l) or 75 percent black tetrads (p<.05), displayed
less tension when supervising racially balanced dyads than either 50
percent black tetrads (p<.Ol) or 75 percent black tetrads (p<.0l),
displayed less tension when supervising 25 percent black tetrads rather.than
MEAN IPA SCORES FOR LEADER RACE X GROUP TYPE INTERACTIONS -
GROUP I HYPOTHESES
Leader (TO) (TI) (T2) (T3)
Category 6: Gives information
Category 1: Shows solidarity
50 percent black tetrads (p<.05), showed less antagonism when super-
vising racially balanced dyads than either 50 percent black
tetrads (p<.05) or 75 percent black tetrads (p<.05), and exhibited
less antagonism when supervising 25 percent black tetrads than either
50 percent black tetrads (p<.05) or 75 percent black tetrads (p<.05).
(See Table 5 for the mean group type scores for categories 2, 11 and
12.) Significant interaction effects between leader race and group
type were found in categories 1 (shows solidarity) (F3,56=3.04,
p<.05) and 10 (disagrees) (F3,56=3.32, p<.05). In category I the
following differences were found: (a) black leaders emitted fewer
solidarity-type comments when supervising racially balanced dyads
than when supervising 25 percent black tetrads (p<.01), and (b) black
leaders of 25 percent black tetrads emitted more solidarity-type
comments than did black leaders of 50 percent black tetrads (p<.0l)
and 75 percent black tetrads (p<.05) and white leaders of 25 percent
black tetrads (p<.01). In category 10 the following differences
were found: (a) black leaders of racially balanced dyads displayed
less disagreeing behavior than did white leaders of the same type
subordinate groups (p<.01) and black leaders of 75 percent black
tetrads (p<.01), (b) black leaders displayed less disagreeing be-
havior when supervising 25 percent black tetrads than when super-
vising 75 percent black tetrads (p<.05) and (c) white leaders of 25
percent black tetrads displayed less disagreeing behavior than did
white leaders of 50 percent black tetrads (p<.05). (See Table 4
for the cell means for categories 1 and 10.)
MEAN IPA SCORES FOR GROUP TYPE MAIN
GROUP I HYPOTHESES
Category 2: Shows tension release
11: Shows tension
12: Shows antagonism
Hypothesis I.B(a) was supported.
Self-esteem. Hypothesis I.B(b) states that white leaders will
score higher in self-esteem than black leaders.
No significant differences were found between black and white
leaders in either the ZSE or RSE measures.
Hypothesis I.B(b) was not supported.
Internal control. Hypothesis I.B(c) states that white leaders
will score higher in internal control than black leaders.
White leaders scored significantly higher (F1,56=9.91, pc.Ol)
in internal control than did black leaders. The mean score for
white leaders was 7.72; for black leaders, 10.51. It should be
noted that a low score signifies more internal control.
Hypothesis I.B(c) was supported.
Duration of speech. Hypothesis I.C(a) states that black leaders
will have a greater speech duration than will white leaders.
No significant differences were found between black and white
leaders in speech duration in either task.
Hypothesis I.C(a) was not supported.
Satisfaction with subordinates. Hypothesis I.C(b) states that
black leaders will indicate more satisfaction with subordinates
than will white leaders.
No significant differences were found between black and white
leaders in their satisfaction with subordinates.
Hypothesis I.C(b) was not supported.
Satisfaction with work in the tasks. Hypothesis I.C(c) states
that black leaders will indicate more satisfaction with work in the
tasks than will white leaders.
No significant differences were found between black and white
leaders in their satisfaction with work in the tasks.
Hypothesis I.C(c) was not supported.
Differerides Between Leaders of the Same Race
This section reports the results of the tests run on the
Group II hypotheses.
Consideration (self-rated). Hypothesis I (a) states that the
performance of groups of equal size and of equal racial composition
will be faster when supervised by leaders who perceive themselves
high in consideration compared to those who perceive themselves as
No significant main effects were found for either black or
white leaders; however, a significant interaction effect (F1,12
5.09, p<.05) occurred in the measure for white leaders supervising
groups of equal racial composition in the performance of the ship-
routing task. (See Table 6 for the cell means.) Investigation of
this effect revealed that under white leaders low in self-rated
consideration racially balanced dyads completed the problem faster
than racially balanced tetrads (p<.05). This may well be due to
Hypothesis I.(a) was not supported.
MEAN PERFORMANCE SCORES FOR LEADER ATTRIBUTE X GROUP TYPE
INTERACTIONS FOR GROUPS OF EQUAL RACIAL COMPOSITION -
GROUP II HYPOTHESES
Cons ideration (self-rated)
Initiating structure (self-rated)
Initiating structure (self-rated). Hypothesis 11(b) states
that the performance of groups of equal size and of equal racial
composition will be faster when supervised by leaders who perceive
themselves high in initiating structure compared to those who per-
ceive themselves as low.
No significant main effects were found for either black or
white leaders. A significant interaction effect (F2,18'4.49,
p<.05) occurred for black leaders supervising groups of equal size
in the performance of the ship-routing task: (a) under black
leaders high in self-rated initiating structure, 25 percent black
tetrads completed the problem faster than 75 percent black tetrads
(p<.05); (b) under black leaders high in self-rated initiating struc-
ture, racially balanced tetrads completed the problem faster than
75 percent black tetrads (p<.05), and also faster than racially
balanced tetrads supervised by black leaders low in self-rated
initiating structure (p<.01) and (c) under black leaders low in
self-rated initiating structure, 25 percent black tetrads completed
the problem faster than racially balanced tetrads (p<.05).
(See Table 7 for the cell means.)
Another significant interaction effect (F1,12=5.51, p<.05)
occurred only for white leaders supervising groups of equal racial
composition in the performance of the knot-tying task. Investigation
of this effect revealed that under white leaders high in self-rated
initiating structure racially balanced dyads completed the problem
faster than racially balanced tetrads (p<.05). (See Table 6 for
the cell means.)
MEAN PERFORMANCE SCORES FOR LEADER ATTRIBUTE X GROUP TYPE
INTERACTIONS FOR GROUPS OF EQUAL SIZE GROUP II HYPOTHESES
21 Wj. 300
I 44o. DO
Hypothesis 11(b) was not supported.
Consideration (subordinate-rated). Hypothesis 11(c) states
that the performance of groups of equal size and of equal racial
composition will be faster when supervised by leaders rated by their
subordinates as high in consideration compared to those rated as
Groups of equal size performing the ship-routing task under
white leaders high in consideration, as rated by their subordinates,
took significantly less time (F1,18=5.00, p<.05) than those under
white leaders low in this measure. Mean task performance time was
706.75 versus 1198.56 seconds. No group type main effects were found
between groups of equal size and groups of equal racial composition.
An interaction effect occurred (F ,12=7.45, p<.05) in this measure
for groups of equal racial composition performing the knot-tying
task under black leaders. Investigation of this effect revealed
that under black leaders high in subordinate-rated consideration
racially balanced dyads completed the problem faster than racially
balanced tetrads (p<.05), and also faster than racially balanced
dyads supervised by black leaders low in subordinate-rated
consideration (p<.05). (See Table 6 for the cell means.)
Hypothesis I1(c) was not strongly supported.
Initiating structure (subordinate-rated). Hypothesis 11(d)
states that the performance of groups of equal size and of equal
racial composition will be faster when supervised by leaders rated
by their subordinates as high in initiating structure compared to
those rated as low.
No significant differences were found for either black or white
Hypothesis 11(d) was not supported.
Self-esteem. Hypothesis 11(e) states that the performance of
groups of equal size and of equal racial composition will be faster
when supervised by leaders high in self-esteem compared to those low
No significant differences were found for either black or white
leaders in either self-esteem measure.
Hypothesis II(e) was not supported.
Internal control. Hypothesis 11(f) states that the performance
of groups of equal size and of equal racial composition will be
faster when supervised by leaders high in internal control compared
to those low in internal control.
For this measure a reversal occurred. Groups of equal size
performing the knot-tying task under black leaders high in internal
control took significantly longer (F1 18=4.42, p<.05) than those under
black leaders low in internal control. Mean task performance time
was 850.25 versus 602.25. No group type main effects were found
for groups of equal size and groups of equal racial composition
supervised by either black or white leaders.
Hypothesis 11(f) was not supported.
Intelligence. Hypothesis 11(g) states that the performance
of groups of equal size and of equal racial composition will be
faster when supervised by leaders high in intelligence compared to
those low in intelligence.
No significant differences were found for either black or white
Hypothesis 11(g) was not supported.
This section reports the results of data analyses involving
differences between black and white subordinates, differences between
groups, differences between leaders In leadership style and social-
emotional behavior, and information derived from the Personal
Differences between black and white subordinates.
During the knot-tying task, black subordinates, as compared to
white subordinates, exhibited behavior as follows:
(a) less giving of suggestions in racially balanced dyads under
black leaders (t=3424 p<.01; x =27.33, xBS=9.63);
(b) less giving of opinion in racially balanced dyads under
white leaders (t=3.74, p<.001; xWS=7.67, xBS=1.51) ;
(c) more giving of opinion in 75 percent black tetrads under
white leaders (t=2.00, p<.05; xi =1.76, xBS=5.05);
(d) less giving of information in racially balanced dyads under
white leaders (t=2.93, p<.01; S=9.66, BS=6.7) ;
(e) less asking for information in racially balanced tetrads
under white leaders (t=2.46, p<.05; x~WS6.80, ;BS=1.58);
(f) more asking for opinion in racially balanced dyads under
white leaders (t=3.56, p<.001; WSO0.66, ZBS=3.80) ;
(g) less disagreeing in racially balanced dyads under white
leaders (t=3.04, p<.01; xWS=ll.57, XBS=4.01);
(h) less showing of tension in racially balanced dyads under
white leaders (t=2.07, p<.05; x =4.30, BS5=1.32); and
(i) more showing of antagonism in 75 percent black tetrads
under black leaders (t=2.18, p<.05; xs=1.59, XBS=4.76).
Every t-test for differences between subordinates was accomplished
using 56 degrees of freedom.
During the ship-routing task, black subordinates, as compared to
white subordinates, exhibited behavior as follows:
(a) more showing of solidarity in racially balanced dyads under
black leaders (t=2.42, p<.05; xWS=0.00, xBS=1.73);
(b) less showing of solidarity in racially balanced dyads under
white leaders (t=2.54, p<.05; XWS=1.82, XBS=0.00);
(c) less giving of suggestions in 25 percent black tetrads
under black leaders (t=2.90, p<.01 ; WS,=11.75, Xg=2.37);
(d) less giving of opinion in racially balanced tetrads under
white leaders (t=2.43, p<.05; x~S=6.64,, xBS=2.48);
(e) less giving of information in 25 percent black tetrads
under black leaders (t=2.13, p<.05; xWS=19.84; xBS=16.20);
(f) less giving of information in racially balanced tetrads
under white leaders (t=2.33, pc.05; xWS=16.90, xBS=12.91);
(g) less asking for information in racially balanced tetrads
under white leaders (t=2.58, p<.05; WS=8.80, XBS=2.97);
(h) more asking for suggestions in racially balanced tetrads
under black leaders (t=2.06, p<.05; WS=2.08, BS= 4.39);
(i) less asking for suggestions in racially balanced dyads
under white leaders (t=3.09, p<.01; xWS=3.47, B =0.00);
(j) less asking for suggestions in 25 percent black tetrads
under white leaders (t=2.04, p<.05; x=3.79, xBS=1.51);
(k) less disagreeing in 25 percent black tetrads under black
leaders (t=2.02, p<.05; xws2.36, S=.0.00) ; and
(1) less disagreeing in racially balanced tetrads under white
leaders (t=3.06, p<.O1; xWS=4.78, ;BS=1.21).
In their duration of speech, black subordinates, as compared to
white subordinates: (a) during the knot-tying task,spoke less in
racially balanced dyads under black leaders (t=2.40, p<.05; XWS=
131.13, xBS=62.13) and in racially balanced tetrads under white
leaders (t-2.42, p<.05; x S=101.94, xBS=32.38); and (b) during the
ship-routing task, spoke more in 75 percent black tetrads under black
leaders (t=2.22, p<.05; W=175.88, gBS=231.92), and less in racially
balanced dyads (t=3.07, p<.01; xWS=158.00, XBS=80.38) and tetrads
(t=2.56, p<.05; xW=107.94, xBS=43.25) under white leaders.
In the other measures, black subordinates, as compared to white
subordinates: (a) were higher in self-esteem in racially balanced
dyads under black leaders, as measured by Ziller's instrument
(t=2.21, p<.05; BWS=19.38, 3BS=26.38); (b) were not significantly
different in internal control; (c) were higher in satisfaction with
fellow subordinates in racially balanced dyads under black leaders
(t=2.12, p<.05; xW=38.88, x.s=46.50); (d) were higher in satisfac-
tion with the leader in racially balanced dyads under black leaders
(t=2.94, p<.01; 'WS=26.63, BS=36.25) and (e) were lower in satisfac-
tion with the work in the tasks in racially balanced dyads under
white leaders (t=2.30, p<.05; xWS=34.00, XBS=25.50).
Differences between groups. During the knot-tying task, the
behavioral differences exhibited in groups supervised by black leaders,
as compared to those supervised by white leaders, were as follows:
(a) less agreeing (F1,56=6.62, p<.05), (b) less asking for informa-
tion (F1,56=9.08, p<.01), (c) less asking for opinion (FI,56=
10.98, p<.01), and (d) more showing of antagonism (F1,56=4.13, p<.05).
(See Table 8 for the mean group scores for these IPA categories.)
Main effects of group type were found in category 4 (gives suggestion)
(F3,56=8.58, p<.01). Irrespective of leader race, three-person
groups, on the average, gave more suggestions when the subordinate
group consisted of racially balanced dyads than when the subordinate
group consisted of 25 percent black tetrads (pc.O01), racially
balanced tetrads (p<.00:), or 75 percent black tetrads (p<.001).
(See Table 9 for the mean group type scores for category 4.) A
significant interaction effect (F3,56=3.12, p<.05) was found in
category 6 (gives information): (a) three-person, white-supervised
groups with racially balanced subordinate dyads gave more information
on the average than five-person, white-supervised groups with both
25 percent black subordinate tetrads (p<.05) and racially balanced
subordinate tetrads (p<.01); (b) three-person groups with racially
balanced subordinate dyads gave less information on the average
under black leaders than under white leaders (p<.01) and (c)
five-person groups with 75 percent black subordinate tetrads gave
less information on the average under black leaders than under white
leaders (p<.05). (See Table 10 for the cell means.)
MEAN GROUP IPA SCORES FOR LEADER RACE MAIN EFFECTS
Groups Supervised Groups Supervised
by Black Leaders by White Leaders
Category 3: Agrees 8.52 13.46
7: Asks for information 2.36 4.52
8: Asks for opinion 0.65 2.39
12: Shows antagonism 3.35 1.89
Category 7: Asks for information 11.50 9.37
8: Asks for opinion 1.18 2.22
MEAN GROUP IPA SCORES FOR GROUP TYPE
Category 4: Gives suggestion
Category 2: Shows tension release
6: Gives information
11: Shows tension
24.87 15.83 14.62 15.89
MEAN GROUP IPA SCORES FOR LEADER RACE X GROUP TYPE INTERACTIONS
Leader (TO) (T) T2)
Category 6: Gives information
Category 1: Shows solidarity
0.94 1.88 3.33
7.81 2.66 2.07
0.51 2.41 0.78 2.09
0.87 0.23 0.60 0.78
During the ship-routing task, the average behavior exhibited in
groups supervised by black leaders, as compared to those supervised
by white leaders, was as follows: (a) more asking for information
(F1,56=5.06, p<.05) and (b) less asking for opinion (F1,56=7.33,
p<.01). (See Table 8 for the mean group scores for these IPA cate-
gories.) Main effects of group type were found in category 2 (shows
tension release) (F3,56=2.94, p<.05), 6 (gives information) (F3,56=
5.18, p<.01), 10 (disagrees) (F3,56=3.84, p<.05).and 11 (shows ten-
sion) (F3,56=5.52, p<.01). Irrespective of leader race, groups:
(a) showed less tension release on the average when the subordinate
group was a 25 percent black tetrad than when it was a racially
balanced tetrad (p<.05) or a 75 percent black tetrad (p<.05);
(b) three-person groups gave less information on the average than
five-person groups with 25 percent black subordinate tetrads (p<.001)
or 75 percent black subordinate tetrads (p<.05); (c) three-person
groups disagreed less than five-person groups having racially balanced
subordinate tetrads (p<.01) or 75 percent black subordinate tetrads
(p<.05), and five-person groups with 25 percent black subordinate
tetrads less than those with racially balanced subordinate tetrads
(p<.05) and (d) three-person groups showed less tension than five-
person groups with both racially balanced subordinate tetrads (p<.001)
and 75 percent black subordinate tetrads (p<.01). (See Table 9 for
the mean group type scores for categories 2, 6, 10 and 11.) A
significant interaction effect (F3,56=3.02, p<.05) was found in
category 1 (shows solidarity): (a) three-person, black-supervised
groups with racially balanced dyads showed less solidarity than
five-person, black-supervised groups with both 25 percent black
subordinate tetrads (p<.05) and 75 percent black subordinate tetrads
(p<.05); (b) five-person, black-supervised groups with 25 percent
black subordinate tetrads showed more solidarity than either five-
person, black-supervised groups with racially balanced subordinate
tetrads (p<05) or five-person, white-supervised groups with 25 percent
black subordinate tetrads (p<.0l); (c) five-person, black-supervised groups
with racially balanced subordinate tetrads showed less solidarity
than five-person, black-supervised groups with 75 percent black
subordinate tetrads (p<.05) and (d) five-person, black-supervised
groups showed more solidarity than five-person, white-supervised
groups with 75 percent black subordinate tetrads (p<.05). (See
Table 10 for the cell means.)
In average duration of speech, there was no significant differ-
ence between black- and white-supervised groups performing the knot-
tying task. However, black-supervised groups talked among themselves
more than white-supervised groups while performing the ship-routing
task (F1,56=6.89, p<.05). The mean speech duration for black-
supervised groups was 203.53 seconds; for white-supervised groups,
There were no significant differences between black- and white-
supervised groups in average measures of self-esteem, internal
control,and satisfaction with work in the tasks.
Differences between leaders in leadership style and social-
emotional behavior. During the knot-tying task, black leaders
exhibited less non-directive leadership behavior (F1 ,56=15.11,
p<.01) and less positive social-emotional behavior (F1,56=11.24,
p<.01). For non-directive leadership, the mean scores were 5.83
and 15.66; for positive social-emotional behavior, 13.61 and 23.46.
A main effect of group type was found for negative social-emotional
behavior (F3,56=3.04, p<.05). Leaders, regardless of race, displayed
more negative social-emotional behavior when supervising racially
balanced groups as the number of subordinates increased from two to
four (p<.0l). (See Table 11 for the mean scores for each group
type.) A significant interaction effect (F3,56=3.03, p<.05) was
found for the directive style of leadership: (a) black leaders of
racially balanced dyads used more directive comments than black
leaders of 75 percent black tetrads (p<.05) and (b) black leaders
showed less directiveness as compared to white leaders of 25 percent
black tetrads (p<.05), racially balanced tetrads (p<.05) and 75 percent
black tetrads (p<.00l). (See Table 12 for the cell means.)
During the ship-routing task, black leaders exhibited less
directive leadership behavior than did white leaders (F1,56 5.14,
p<.05). The mean score for black leaders was 37.76; for white
leaders, 51.02. A main effect was found among group types in
negative social-emotional behavior (F3,56=6.32, p<.01). Leaders,
regardless of race, displayed less negative social-emotional behavior
when supervising racially balanced dyads than when supervising
racially balanced tetrads (p<.0l) or 75 percent black tetrads (p<.01),
and when supervising 25 percent black tetrads as compared to racially
balanced tetrads (p<.01) or 75 percent black tetrads (p<.01). (See
Table 11 for the mean scores for each group type.) There were no
MEAN LEADER IPA SCORES FOR GROUP TYPE MAIN EFFECTS
IN SOCIAL-EMOTIONAL BEHAVIOR
(TO) (I) (T 2) (L
Negative social-emotional behavior 7.71
Negative social-emotional behavior 5.91
13.11 21.39 15.48
5.95 16.10 15.13
MEAN LEADER IPA SCORES FOR LEADER RACE X GROUP TYPE
INTERACTION IN LEADERSHIP STYLE
Leader (TO) (TI)
Directive style of leadership
General satisfaction survey. There were no significant differ-
ences found on this measure between leaders or between black and white
supervised groups. In the analysis of differences between leaders,
a significant difference was found between group types (F3,56=343,
p<.05); leaders, regardless of race, indicated more general satis-
faction if they had supervised racially balanced tetrads rather than
25 percent black tetrads (p<.01) or 75 percent black tetrads (p<.05).
(ee Table 13 for the mean scores for each group type.) Black subordi-
nates, as compared to white subordinates, were higher in general
satisfaction in 75 percent black tetrads under white leaders (t=
2.20, p<.05, df=56; xWS=26.38, xBS=30.21).
Sociometric rating of leader by subordinates. White leaders of
racially balanced dyads were rated higher by white subordinates than
by black subordinates (t=2.32, p<.05, df=14). The mean rating for
white leaders by white subordinates was 3.50 and by black subordinates,
2.50. No other significant differences were found in the comparisons
by group type, by leader type or by leader for both subordinate
Nomination for best squad leader. Black subordinates under
black leaders choose their leader as the best squad leader in their
company more often than did black subordinates under white leaders
(X =5.54, p<.05, df=l). Black leaders choose themselves as the best
squad leader in their company more often than did white leaders
(X =6.56, p<.05, df=l). No significant differences were found between
black and white subordinates under black leaders, black and white
subordinates under white leaders, and white subordinates nominating
their own leader regardless of leader race.