Group Title: study of contemporary leadership models in CPA firm audit teams
Title: A study of contemporary leadership models in CPA firm audit teams
Full Citation
Permanent Link:
 Material Information
Title: A study of contemporary leadership models in CPA firm audit teams
Physical Description: Book
Language: English
Creator: Dillaway, Manson Peter, 1937-
Copyright Date: 1980
 Record Information
Bibliographic ID: UF00102826
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: oclc - 07295221
ltuf - AAL5576

Full Text



















The Audit . . . . . . .
The Audit Team. . . . . .
The Phases of an Audit . . ..
Audit Teams Leadership Research


MEASUREMENTS. . . . . . .






Ideal Leader Behavior Correlational
Study . . . . . . .
Vroom and Yetton Normative Model of
Leadership Styles . . . .
Item Analysis of the LBDQ Results
Factor Analysis of the LBDQ Results


Summary . ..
Conclusion. .
















. . 102
. . 104

S. 106













. 112

. 116

. 119

. 123

. 127

. 129

. 130

. 131

. 132


Abstract of Dissertation Presented to the Graduate Council
oftheUniversity of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy





Chairman: Gary L. Holstrum

Major Department: Accounting

The applicability of certain behavioral and situational

leadership studies models was tested using audit teams from

a national CPA firm engaged in field work.

Leader behavior predictor variables were measured using

an adaptation of the Fox Modified Leader Behavior Description

Questionnaire. They included Consideration, Initiating

Structure, Consultative-Participative Decision-Making, Work

Facilitation, Decisiveness, Goal Emphasis, Leader Support,

and Management of Rewards and Punishment. Situational char-

acteristics measured among team subordinate members included

Pressure, Task Certainty, Need for Information, Organiza-

tional Independence, and Intrinsic Job Satisfaction. Global

assessments were obtained for team performance from

supervisors and for job satisfaction from members. Addition-

ally, a measure of subordinate dissatisfaction was obtained

by using the shortfall between the expected leader behavior

and actual leader behavior item scores on each individual

team member's Fox Modified Leader Behavior Description Ques-

tionnaire. Each team leader completed the 32-item Fox Least

Preferred Coworker (LPC) instrument.

Consideration, Leader Support, Work Facilitation, and

Decisiveness showed a significant negative correlation with

Dissatisfaction. WorkFacilitation correlated positively with

Performance. Significant negative correlations were found

between LPC and Performance when leader member relations were

moderately poor (Octant VII). A factor analysis of the Fox

Modified Leader Behavior Description Questionnaire results

yielded a three-factor solution. None of the derived factors

approximated any of the predictor variables previously


Based upon the results of this study, none of the models

tested appeared to be notably applicable for predicting out-

comes in terms of audit team satisfaction and performance.

It was also concluded that situational factors may exert a

stronger influence on audit teams than on more homogeneous

groups engaged in less complex tasks.



The Audit

Such diverse services are performed by certified public

accountants in both the private and public economic sectors

of the United States that it is necessary to place a bound-

ary around those areas of public accounting which are to be

considered in this study. The primary service performed by

CPAs is discussed in the following:

The objective of the ordinary examination of financial
statements by the independent auditor is the expression
of an opinion on the fairness with which they present
financial position, results of operations, and changes
in financial position in conformity with generally ac-
cepted accounting principles. [AICPA SAS No. 1, 1973,
p. 1].

This examination of financial statements will subsequently

be referred to as an audit.

The primary role of auditing has been described as "to

improve the usefulness of decision-making information" [PM

& M, 1976, p. 2], and the fundamental type of decision-mak-

ing which auditing facilitates is the allocation of scarce

resources in a society. This role is not restricted by type

of economic system or organizational structure, since infor-

mation that is improved in terms of accuracy, timeliness,

understandability, or specificity should enable better

choices to be made both in terms of attracting capital from

investors and distributing internally generated funds among

alternative uses.

The Audit Team

For the purpose of this study, it was determined that

there exists a standard organizational structure used by the

tested CPA firm on all annual audits despite the fact that

each annual audit is a unique case. Notwithstanding this

uniqueness, there are common elements of organization which

may be observed in the firm tested and in comparable CPA


One member of a firm will normally retain overall re-

sponsibility for an audit. This is generally that individual

who assumes the primary role in maintaining communications

with the client and negotiates the particular services which

the firm will provide as well as the billings for those serv-

ices. This individual is normally a partner in the CPA firm.

The key role element is the entrusted authority and responsi-

bility for providing services to particular clients. The

partner will not normally take an active part in the routine

audit work (nor should he if he is to be an objective

reviewer of the audit evidence). Consequently, a manager is

appointed to organize specific client services and an "ac-

countant-in-charge" (AIC) is appointed for the field work

phase of the audit. The AIC generally prepares and admin-

isters an audit program which, when effectively carried out,

will provide sufficient evidence for the CPA firm's

publication of an opinion regarding the client's financial


The AIC should be an audit staff member with sufficient

experience and ability to effectively plan the nature and the

extent of the audit tests and supervise the work of the sub-

ordinate audit staff members while they are occupied with

this audit engagement--both in the field and in the CPA

firm's office.

It is the AIC position which is of primary interest in

this study. An AIC will customarily be delegated extensive

discretionary authority in an engagement as long as it seems

apparent that satisfactory results are being achieved. The

competent AIC will, quite literally, handle an entire audit


The number of assistants assigned to an audit engage-

ment is not necessarily stable and may fluctuate during the

course of the engagement depending upon the audit procedure

being conducted. For instance, a large group may be tempor-

arily required to observe and review a client's count of

year-end inventories. There is, however, a tendency to as-

sign audit assistants to engagements for the duration where

feasible. This tendency endures in practice because of the

learning period required for an assistant to master a client's

employees and the other members of the audit team.

Many audit assistants are quite limited in their work

experience and may have little more than their university

education and the CPA firm's indoctrination course to serve

as a basis from which to perform audit tests.

Since formal educational programs are, of necessity,

general in the type of knowledge that they convey, the audit

assistant finds that during the early stages of his career

there is a continual need for counsel as new situations are

encountered. As staff assistants become experienced, they

may be expected to perform work requiring the exercise of

professional judgment. During periods when a CPA practice

experiences rapid expansion, staff assistants will advance

rapidly in terms of the degree of responsibility demanded

by work assignments and may find themselves working at the

limit of their acquired skills.

On the typical audit engagement, staff assistants must

work with the AIC and their colleagues in order to make a

useful contribution to the audit. The degree of this inter-

dependency should be evident in the character of the rela-

tionship between a staff assistant and the various other

members of an audit team. Whereas most work assignments on

an audit do not require assistants to work in unison with

each other, the overall coordination of an audit team is

vital to effective performance if individual assistants are

not to interfere with the work of others and avoid needless


During the course of an audit there may be instances

where the members of the team are unable to perform certain

procedures or provide required services because they lack

... .... ... .

sufficient technical knowledge. To fit this need, most CPA

firms employ "specialists" for aid when these situations

arise. These employees specialize in computer technology,

personnel evaluation and placement, tax planning, SEC report-

ing, or some other area of concern to the audit team. Gen-

erally specialists serve on an engagement to the extent that

their expertise is needed. In his relationship with the AIC,

the specialist resembles an internal consultant, not a sub-

ordinate. Since this study will concentrate on the inter-

play between group leader and subordinates, the specialist

will not be included as a team member.

The Phases Of An Audit

There are three readily identifiable phases in an an-

nual audit. Each has implications with regard to supervis-

ion to staff assistants. First, there is a planning phase

in which a CPA firm manager and the accountant-in-charge de-

cide the manpower needs of the audit and establish deadlines

for the completion of certain portions of the audit work.

The second phase is where the AIC and his assistants

perform field tests to gather evidence which will enable them

to form an opinion as to the appropriateness of the client's

financial report. In this phase the audit "team" is func-

tioning and, it may be assumed, perceptions of leader behav-

ior become measurable. This is the phase which provides the

basis for the measurements in this study.

The third and final phase of an audit engagement norm-

ally takes place in the CPA firm's office after the field

work has been completed. The working papers are assembled,

checked, and reviewed. Normally, a smaller staff is required

in this phase than during the preceding field work phase.

If no gaps remain in the network of evidence that has been

gathered in accordance with the audit program, the CPA firm

will issue an auditor's report and the audit team will be


Audit Team Leadership Research

The relative importance of leadership in the audit team

context has not been documented. In this study an attempt

is made to eliminate this void. Accordingly, this study con-

centrates on groups engaged in annual audits during the field

work phase. It seeks to determine the applicability of cer-

tain theories which purport to explain the effects of leader

behavior on group performance and on group member satisfac-


It is expected that practitioners may benefit from this

research in several ways. It could enable them to discover

possible sources of employee dissatisfaction which results

in unnecessary staff turnover. It could point to leader be-

havior which tends to be associated with high quality work.

Certainly, improvements in audit team performance in an ab-

solute sense eventually reflect favorably upon the CPA

profession as a whole by reducing the possibility of inap-

propriate auditor's reports.

Perhaps the most positive aspect of this research is in

its potential for improving team efficiency. Efficiency of

operation is as important in a service-oriented CPA firm as

it is in a manufacturing firm. It is more difficult to

measure efficiency when output is not in the form of units

of standard quality and specification. A layman may assume

that if one AIC required five hundren man-hours for an annual

audit and another completed the annual audit in a subsequent

year in eight hundred man-hours, the AIC using the fewer

hours was the more efficient. An experienced reviewer of

annual audits would be quick to point out that the man-hours

expended in an audit was seldom a measure of efficiency even

when a standard audit program is used by and the experience

level and competence of the audit teams are comparable.

Each annual audit is unique. Efficiency in field work

is a long-range necessity for CPA firms. There are con-

straints in terms of the fees a client will pay. There is

also a time constraint in field work created by the neces-

sity to simultaneously divide the firm's professional staff

among numerous clients. Further need for efficiency is gen-

erated by the deadlines which exist for the publication of

financial statements and other documents which may rely on

the same financial data (SEC reports and tax returns).

The limited time span of an audit team's existence ob-

viates the need for a longitudinal measurement of the possible

effects of alternate leader behavior on group performance

and subordinate satisfaction. Although leadership research-

ers have pointed to the necessity for longitudinal studies

[Fox, 1976a; Fox, 1976b, Franklin, 1975;Likert, 1967] due to an

apparent time-lagbetween changes in supervisory behavior and

lasting changes in subordinate performance, the temporary

nature of an audit team rules out this consideration.

Clearly, in this respect, an audit team should have more sim-

ilarity with the laboratory study groups and temporary mili-

tary groups which have provided much of the support for lead-

ership models. It is the temporary change in group behavior

which is the appropriate criterion in studying audit teams.

An audit team is a group in which the leader (AIC) norm-

ally wields extensive authority relative to his assistants

and the extent to which the task of the group is structured

is quite low. Very little leadership research has been pub-

lished on groups of this type. Two studies were reported by

Fielder 1967, p. 129 and "the data obtained in these studies

are quite weak." This study, then, concentrates in an area

that has been neglected.

The foregoing provides the background information nec-

essary to consider the effects of leader behavior on per-

formance and satisfaction measures as applied to groups en-

gaged in annual audit work in the field (audit teams).

Subsequent chapters will briefly review the research to date

in leadership and show how it was applied to audit teams in

this instance.



The primary objective of this study is to determine the

applicability of certain contemporary research models and

techniques used for understanding leadership in formal organ-

izations for groups engaged in the field work phase of annual

audits. It is expected that the results should indicate

whether certain preoccupations in leader behavior and atti-

tude have noticeable effects on assessments of group effec-

tiveness and various individual measures of job satisfaction.

The major approaches to the study of leadership have been

classified by Filley et. al. [1976] into three categories--

trait, behavioral, and situational.

The trait approach is based upon the assumption that an

individual's behavior can be attributed to certain distinc-

tive psychological or physical characteristics. Certain

identifiable traits differentiate successful from unsuccess-

ful leaders to proponents of this approach. Traits cannot

be directly observed, but conceptualizations can be measured

which may approximate traits. For example, "I.Q." tests

are available which some presume to measure an approximation

of an individual's intelligence.

Filley et al. [1976, p. 219] state that "the weight of

evidence pertinent to the trait approach does not lend

strong support to it." They conclude that the proposition

that there are a finite number of identifiable characteristics

or traits which differentiate successful from unsuccessful

leaders is insufficiently supported by the research conducted

to date.

The behavioral approach to the study of leadership as

defined by Filley et al. [1976] is based upon the assumption

that leaders may be characterized by readily observable be-

havior patterns. These patterns, in turn, may be categorized

into four types: supportive, participative, instrumental,

and "great man."

Examples of supportive leaders would be individuals who

rate high in Consideration (Ohio State Leadership Studies

Model) or Leader Support (Fox Normative Model). These leaders

treat their subordinates in ways which increase their posi-

tive feelings toward their jobs. That is, they praise work-

ers for good performance, avoid authoritarian behavior, and

adopt a posture of thoughtfulness and sincerity in their

interactions with subordinates.

In their summary of the empirical evidence as to the

effectiveness of leaders characterized by considerate-sup-

portive behavior, Filley et al. [1976, p. 222] conclude that

"supportive leadership is often effective, resulting in im-

proved subordinate attitudes, satisfaction, and performance."

However, the studies reviewed could not be construed as to

substantiate the proposition that effective leaders are char-

acterized by supportive behavior because of the effects of

individual, task, and environmental variables. The authors

also point out that "much remains to be learned" with regard

to the question of causality--studies have shown that sup-

portive leader behavior positively affects subordinate sat-

isfaction and performance and other evidence suggests that

supportive behavior may be the result of subordinate satis-

faction and performance.

Participative leaders are those who use participative

decision making and supervision techniques. Such leaders

would rate high in Consultative-Participative Decision Mak-

ing (Fox Normative Model). In their review of the studies

dealing with participative leadership, Filley et al. [1976,

p. 229] state that although findings suggest that participa-

tive leadership is often effective, the proposition that ef-

fective leaders are characterized by participative decision

making and supervision "cannot be said to be confirmed, be-

cause there is ample evidence that the effects of participa-

tive leadership are not omnipotent." They cite the necessity

for the leader to be otherwise skilled in the performance of

his task, for subordinates to be in favor of participation,

and for the task to be a complex one requiring a high quality

decision or subordinate acceptance of the decision--or a

combination of both. As was the case with supportive lead-

ership, situational variables must be considered in the

evaluation of participative leadership findings.

Instrumental leadership is based upon the management

activities of planning, organizing, controlling, and coordin-

ating the work of subordinates. These activities are presumed

to be "instrumental" in achieving group accomplishment of or-

ganizational activities. Leaders rating high in Initiating

Structure (Ohio State Leadership Studies Model) or Work Facil-

itation (Fox Normative Model) represent instrumental leaders.

Empirical evidence with regard to the effectiveness of

instrumental leadership in promoting group performance and

satisfaction criteria does not fully support the proposition

that instrumental leaders are effective leaders. Kerr et al.

[1974] reviewed the results of studies where Initiating Struc-

ture was measured and concluded that it was only effective

in situations where there was outside pressure, where the

task was satisfying in itself, where subordinates perceive

high labor mobility for their skills, and in other specific


"Great man" leaders are characterized by both supportive

and instrumental leadership behavior. Presumably the leader

faces conflicting demands from the two levels that he works

directly with--subordinates preferring supportive behavior

and superiors preferring instrumental behavior. Filley et al.

[1976, p. 233] question whether it is possible for a leader

to exhibit both behavioral styles. They feel that the prop-

osition that "great man" leaders are effective is only par-

tially supported by the empirical literature and that situa-

tional factors come into play on the two components of "great

man" leadership (supportiveness and instrumentality) in the

same fashion as they might for supportive of instrumental


Situational leadership approaches have been used to

eliminate the major weakness of the behavioral approaches

just mentioned--oversimplification due to the failure to

consider the moderating effects of situational factors. Con-

tingency theories have been proposed by Fiedler [1967), House

[1971], and Vroom and Yetton [1973], among others. These as

well as certain specific behavioral models will now be


The Ohio State Leadership Studies were undertaken by an

interdisciplinary group working within the Bureau of Business

Research of the Ohio State University. An eight step research

plan was adopted and reported in the monograph, Methods in

the Study of Administrative Leadership [Stogdill and Shartle,

1955]. Two of the eight steps--the steps of greatest inter-

est to this study--called for descriptions of actual leader

behavior by both group leader and group members and measures

of individual leader and group effectiveness. Two basic hy-

potheses were proposed with regard to leader behavior: (1)

the patterns of behavior in a given leadership position will

be determined in part by the performance demands made upon

that position and (2) the leader's behavior is related to (a)

status in the organizational hierarchy, (b) structure of

interactions among organization members, (c)

responsibility-authority structure of the organization, and

(d) performances of members of the organization.

Hemphill and Coons [1957, p. 7] report that the problem

of obtaining leader behavior descriptions was met by develop-

ing a methodology designed to answer a primary question,

"How does this leader operate?" The instrument that evolved

consisted of a set of questions which would enable researchers

to isolate and classify the relevant forms of behavior ex-

hibited by an individual when directing the activities of a

group toward a shared goal. Nine dimensions of behavior were

tentatively selected: integration, communication, production

emphasis, representation, fraternization, organization, eval-

uation initiation, and domination. Using these dimensions,

a large number of items were constructed and refined [Hemphill

and Coons, 1957, p. 9]. Among the several criteria used for

item selection were insistence that only specific behavior

as opposed to general traits or characteristics be described,

that an item be limited to one unit of behavior, that an item

not be emotionally or evaluatively toned, and that the times

be applicable to many different organizational structures,

groups, and situations. An initial factor analysis of the

intercorrelations of the dimensions of the Leader Behavior

Description Questionnaire (LBDQ) conducted by Hemphill and

Coons [1957, p. 24] yielded factors labelled Maintenance

of Membership Character, Objective Attainment Behavior, and

Group Interaction Facilitation Behavior. The first is based

on leader behavior that is socially agreeable to group mem-

bers while the other two are based on leader behavior which

structures group activities and communications.

Another factorial study utilizing the LBDQ was carried

out with air crew commanders by Halpin and Winer [1957].

They accounted for 100% of the common variance with four fac-

tors which were designated Consideration (49.6%), Initiat-

ing Structure (33.6%), Production Emphasis (9.8%), and Sens-

itivity (7.0%). The subsequent working model of leader be-

havior which evolved from these Ohio State studies concen-

trated on the first two factors. Consideration (C) is des-

cribed as leader behavior reflecting friendship, mutual

trust, respect, and warmth. Behavior on the part of a leader

which organizes and defines relationships, defines roles, and

clarifies how to do jobs is classified as Initiating Struc-

ture (I.S.).

The bulk of the early studies utilizing the LBDQ were

based on hypotheses such as supervisors rate favorably the

performance of leaders who show high I.S. behavior; subord-

inates prefer as leaders those who show high C behavior; and

leaders getting highest effectiveness ratings score above

the mean in both leader behavior dimensions while leaders

getting lowest ratings score beneath the mean in both dimen-

stions [Halpin, 1957]. The results of these studies did not

lend themselves readily to generalization due to the organi-

zationally specific nature of most leadership positions.

Jacobs [1970, p. 33] concluded: "These studies demonstrate

that situational variables influence the balance of leader

behavior that will be desired (or best)."

Stogdill [1974, p. 427], one of the developers of the

LBDQ, has cautioned researchers about placing too much empha-

sis on C and I.S. and other frequently used measures:

A small number of variables (authoritarianism,
democratic leadership, consideration, initiating
structure, and LPC, for example) has been overworked
at the expense of the other variables that are
equally important and about which little is known.

However, Stogdill [1974, p. 428] does reaffirm the key be-

lief that there is a functional dependency between leader be-

havior and group performance and emphasizes the situational

nature of such behavior:

The leader is required to (1) maintain role
structure and goal direction, (2) provide for role
freedom and group drive, and (3) maintain group
cohesiveness and norm conformity. The behaviors
that produce these outcomes differ from group to
group depending upon variations in member charac-
teristics, task characteristics, and group charac-

In an effort to encounter criticism that the Ohio State

Leadership Studies ignore situational variables, Kerr et al.

[1974], by reviewing published findings utilizing the Leader

Behavior Description Questionnaire, were able to develop ten

situational propositions and two general postulates of lead-

ership effectiveness. The variables which were hypothesized

to have a moderating influence between C and I.S. and satis-

faction and performance criteria were subordinate need for

information, job level, subordinate expectations of leader

behavior, perceived organizational independence, leader's

similarity of attitudes and behavior to managerial style of

higher management, leader upward influence, and the task

characteristics of pressure and intrinsic satisfaction. The

adaptation of these situational variables into the Ohio State

Leadership Studies tends to reconcile that massive body of

work with the other contemporary leadership studies.

House and Dessler [1974] have revised and extended

House's [1971] Path-Goal Theory of Leader Effectiveness which

attempts to explain the effects of intermediating (situational)

variables in the relationship between leader behavior and

performance and satisfaction measures. The leader is viewed

as one whose function is motivational--to increase personal

payoffs to subordinates for work goal attainment and to fa-

cilitatethe ease with which subordinates may travel the path

to goal attainment. They contend that the specific leader

behavior necessary to perform this motivational function is

determined by (contingent upon) the situation in which the

leader operates.

House and Dessler [1974, p. 31-2] mention two classes of

situational variables: (1) the characteristics of subordin-

ates and (2) the environmental pressures and demands the sub-

ordinates must cope with to -accomplish work goals and satisfy

their own needs. One subordinate characteristic is the de-

gree to which leader behavior is perceived to be either a

source of immediate satisfaction or instrumental to future

satisfaction. The other characteristic is the subordinates'

perception of their own ability with respect to task demands.

Environmental moderators proposed by House and Dessler were

the subordinates' task, the formal authority system of the

organization, and the primary work group. They suggest that

the effect of the leader on the subordinate's motivation will

be a function of how deficient the environment is with respect

to motivational stimuli, constraints, or rewards.

The leadership studies undertaken at the Survey Research

Center of the University of Michigan share certain similari-

ties with the Ohio State Leadership Studies. The Survey Re-

search Center methodology also relies upon the position that

leadership amounts to a large aggregation of separate behav-

iors [Bowers and Seashore, 1966, p. 240]. The basic differ-

ences between the two research methods are in how aspects of

leader behavior should be grouped and classified. There are

four dimensions emphasized in the survey Research Center

Studies: (1) support, (2) interaction facilitation, (3) goal

emphasis, and (4) work facilitation. These are based on an

integration of research results and writings by Katz and

Kahn, Cartwright and Zander, Mann, and Likert [Bowers and

Seashore, 1966, 247]. The Survey Research Center methodology

measures aspects of behavior which represent supervisory and

peer leadership. The Ohio State Leadership Studies concen-

trate on the leadership provided by one leader in a group.

A standardized questionnaire has been developed based

on the four factors proposed by Bowers and Seashore [Taylor

and Bowers, 1972]. It is a machine-scored instrument which

measures support, goal emphasis, work facilitation, and

interaction facilitation for both supervisory and peer lead-

ers. Also measured are various assessments of individual

satisfaction (with peers, with supervisor, with job, with the

organization in relation to others, with pay, with current

progress, and with future prospects). The entire question-

naire contains twenty-four short questions, thirteen of which

measure the four leadership factors.

Taylor and Bowers [1972, p. 55] initially administered

the Survey of Organizations to the employees of an oil refin-

ery (N = 325), and, after excluding three questions from the

set, obtained Spearman-Brown reliability coefficients of from

.85 to .91 for the four leadership factors. An analysis was

made to determine if there existed a single general factor

for supervisory leadership of sufficient magnitude to inval-

idate the four factor construct. It was concluded that the

general factor content was not large enough to regard any of

the four proposed factors as interchangeable measures of a

single leadership factor and that each of the four contained

sufficient unique variance to be considered a measure of some

distinguishable aspect of leadership.

Studies initially undertaken by the Group Effectiveness

Research Laboratory at the University of Illinois are based

on Fiedler's Contingency Model which suggests that changes

in group effectiveness from one situation to another, where

leadership style is held constant, depend upon the favorabil-

ity of the situation [Fiedler, 1967, Fiedler and Chemers,

1974]. The binary factors which determine situation

favorability are (1) leader-member relations (good, poor),

(2) task structure (high, weak), and (3) position power

(strong, weak). The eight combinations of these factors

form eight cells in Fiedler's classification of group situ-

ations. Four criteria are used to determine the degree of

task structure: decision verifiability, goal clarity, goal

path multiplicity, and solution specificity. Leader-member

relations are based upon the degree of acceptance the leader

obtains from his group. Position power is related to the ex-

tent to which the leader can reward and punish his subordin-


The measure used to classify leadership style in Fied-

ler's Contingency Model studies is the Least-preferred Co-

worker (LPC) score. This is obtained from a sixteen scale

semantic differential instrument [Fiedler, 1967.,p. 269].

The bipolar scales have eight positions that may be selected

by the respondent from least favorable to most favorable.

The usual scoring is one point for the least favorable posi-

tion up to eight points for the most favorable position.

The total score for a given leader is the summation of the

scores for every item. A high score on the LPC instrument,

in relation to a given norm, reveals a leader who holds his

least-preferred coworker in higher regard than low scorers.

Fiedler [1967, p. 46] hypothesizes that:

High-LPC leaders are concerned with having good
interpersonal relations and with gaining prominence
and self-esteem through these interpersonal relations.

Low-LPC leaders are concerned with achieving success
on assigned tasks, even at the risk of having poor
interpersonal relations with fellow workers.

When the situation is favorable for the leader in terms

of control and influence, Fiedler and Chemers [1974, p. 77-8]

argue that a leader will pursue his secondary goals:

the high LPC leader will concern himself
with such status-enhancing activities as ordering
people around, assigning tasks, and assuming re-
sponsibility. The low LPC leader, given this high
degree of control, will be relaxed, friendly, and
considerate, in the knowledge that the task pre-
sents no problem.

In situations where the leader control is low (unfavor-

able situations), Fiedler and Chemers [1974, p. 78] hypothe-

size that each leader will revert to his primary goal: the

high LPC leader will seek to enhance relationships with group

members; the low LPC leader will concentrate on getting the

task accomplished at the expense of good interpersonal rela-

tions, if necessary.

A normative decision-making model based upon organiza-

tional psychology has been proposed by Vroom and Yetton [1973],

who had found significant variation in the amount of partici-

pation that was perceived as necessary by managers in given

problem situations. Their model is based on the assumption

that the most appropriate unit for the analysis of a situa-

tion requiring the exercise of leadership is the particular

problem to be solved and the context in which the problem

occurs. The discrete social processes by which organizational

problems are translated into decisions form the basis for

prescribed leadership styles which vary in the amount of

subordinate participation. Vroom and Yetton describe five

decision methods for group problems varying from a completely

autocratic decision style to a group generated decision

style. To make the model operative, a list of questions are

provided for the leader to consider in diagnosing a partic-

ular problem before choosing a decision style. Heller [1973]

independently developed a similar model and reached similar


Bass and Valenzie [1974] have proposed a Systems Model

of Leadership in which the system outputs (effectiveness and

employee satisfaction)are to be explainedby systems inputs

(organizational, work group, task, and subordinate personal-

ity variables), system relations (power distribution in-

formation distribution, and work group structure and objec-

tives), and managerial styles (directive, manipulative, con-

sultative, participative, and delegative). They have hypoth-

esized that the directive style should be most effective

when the leader has high power and information distributions;

the delegative style, when low. Participation is hypothesized

to be most effective when the leader and group members have

equal information. Consultation was proposed for situations

where the leader has high power but low information and man-

ipulation for low power-high information situations.

One of the recurring difficulties in leadership studies

is the isolation of significant moderator variables and the

statements of their effect. In addition to the variables

isolated in the contingency models of Kerr et al. [1974],

Fiedler [1967], Vroom and Yetton [1973], House and Dessler

[1974], and Bass and Valenzie [1974], as already discussed,

there have been numerous studies purporting to show the me-

diating effects of additional variables on measures of satis-

faction and performance.

Bowers and Seashore [1966, p. 257] found that the best

"predictor" of three out of four peer leadership variables

was the same variable as measured for the group leader, in-

dicating a relationship between the two. Certain writers

cited by Greene [1976, p. 65] have argued that subordinate

behavior may cause variance in leader behavior in opposition

to the usual pattern of causuality assumed in leadership

studies. Fox [1976a, p. 6] has cited performance and satis-

faction of his subordinates are moderated by the behavior of

higher level management,

The moderating effects of group member personality dif-

ferences have been discussed by Korman [1974, p. 192] and

Fox [1976a, p. 6]. Korman believes that personality differ-

ences may prove to have significant moderator effect if the

personality constructs used are situationally defined and the

measures construct validated. Korman believes that care in

these areas will eliminate criticism based on evidence per-

sonality test unreliability over time and over different


Kerr et al. [1974, p. 66] discuss a number of findings

with regard to the moderating effect of pressure from

external forces (interunit stress and physical danger) and

from internal forces (urgency of time demands). Significant

correlations between proficiency ratings and structure have

been found in studies where time pressure was involved.

In this review of the research literature, Filley et al.

[1976, p. 233] conclude that the effects of supportive lead-

ership are most positive with regard to performance when (1)

the subordinate has been previously denied some source of

satisfaction or (2) the subordinate operates under conditions

of frustration or stress. The effects of supportive leader-

ship on subordinate satisfaction were found to be positive

except in situations where work is intrinsically satisfying

[Filley et al. 1976, p. 220].

With regard to instrumental leadership, Filley et al.

[1976, p. 230] reportthat initiating structure has its most

positive effects when

(1) There is a high degree of pressure for output due to
demands imposed by sources other than the leaders
(2) The task is satisfying to subordinates
(3) The subordinates are dependent on the leaders for
information and direction
(4) The subordinates perceive that they have labor
(5) The subordinates are psychologically predisposed
toward being told what to do and how to do it
(6) Subordinates' tasks are nonroutine
(7) Subordinates have high occupational-level jobs
(8) Subordinates are under stress due to threat from
sources other than the leaders
(9) The number of people working together for the same
leader is high (12 or more)
(10) The leader is considerate

Other important moderators mentioned by Fox in his lit-

erature review [1976a, p. 7] include subordinate-leader

knowledge and expertise differences, supervisor's upward

influence, inter-group competition, and complementarity of

leader-subordinate behavior styles.

It should be apparent from the foregoing, that research

results in the leadership area are not only specific to sit-

uational characteristics, but also subject to the potential

mediating effects of many other variables. The difficulty

in drawing well grounded generalizations from these endeavors

has by no means diminished the volume of research being done;

nor, for that matter, have basic theoretical criticisms such

as the relative importance of leadership to other organiza-

tional variables [Hunt and Larson, 1976, p. 235] or the mer-

its ofacognitive versus operation approaches in leadership

studies [Fox, 1976b]. Many still consider an acognitive

study of leader-subordinate dyads as an effective research


This brief discussion of contemporary research methods

in the field of leadership studies is not intended to be

complete. It is introductory to the chapters which follow.

The purpose of this dissertation is to borrow and adapt to

an auditing environment that which has been successfully ap-

plied elsewhere. The general approach has been to use the

methods of many contemporary schools rather than select one

methodology. In this way, questions regarding the criteria

used for selection of particular leadership measurement

methods may be minimized and the study may be limited to the

usefulness of the methods toward predicting satisfaction and
performance in auditing situations

Chapter III


The foundation upon which this study is built is Ident-

ifying and Developing Leadership Aspects of Effective Manage-

ment in Team-Oriented Task Groups [Fox 1976a]. In this work,

Fox presents a normative model for leadership based upon cur-

rent research findings -- a model specifically developed to

assist managers of team-oriented (interacting) task groups

in adjusting their behavior toward greater managerial effec-


The adaptation of the Fox model to the field of auditing

follows due to the fact that an audit team engaged in the

field work phase of an annual audit represents an interacting

task group. Fiedler [1967] states that in such groups, there

exists a high degree of interdependence among group members.

The leader coordinates the various task functions of the

group's activities so that the work flows smoothly and harm-

oniously. The various segments of an audit program must be

closely coordinated if the work of the audit team is to be

successful. Certain audit procedures may be completed by an

individual, yet many require a coordinated effort by two or


Effective performances from audit teams are imperative

if professional standards are to be maintained and if the


required efficiency for the auditing firm's profitability is

to be met. Certain situational aspects of annual audits are

relatively comparable. The behavior of the audit in gather-

ing evidence is partially restricted in the interest of re-

maining an independent observer. Certain audit procedures

are demanded by generally accepted auditing standards. Min-

imum standards of record-keeping and reporting are forced

upon the client by the necessity to provide documentary sup-

port for items reported on tax returns and financial reports.

The manufacturers of business machines, business forms, and

computer software have tended to standardize the ways that

most large organizations produce, record, and store documents

and reports. Organizational structures often reflect simi-

larities within specific industries. The auditors, them-

selves, tend to be similar in education, interest, aptitudes,

and even standards of dress. This may be the result of a

certain amount of standardization in selection policies used

by major CPA firms. These conditions would seem to indicate

that the study of audit teams in the field represents a po-

tentially rewarding area for research in leadership. This

is especially appropriate at a time when auditing effective-

ness is continually questioned both in court cases and by

government and when signs of staff member dissatisfaction

are prevalent.

Fox [1976a, p. 2] regards the dominant emphasis in def-

initions of leadership as influence via personal interaction

and considers only face-to face aspects of planning and

controlling as falling within leadership. This leads to a

definition of leadership style as the behavioral prediposi-

tion or orientation on the part of a leader as he deals with

the performance of management or leadership functions that

result more from the leader's personality or values than

from the demands of the situation at hand [Fox, 1976a, p.9].

Leadership style factors suggested by Fox [1976a, p. 10]

based on his review of published factor analysis of leader

behavior descriptions [Halpin and Winer, 1957; Bowers and

Seashore, 1966; Wofford, 1970; Miller, 1973; House and Des-

sler, 1974] are (A) task relevant structuring emphasis, (B)

support emphasis, and (C) consultation, participation


The normative model hypothesizes the actual behavior,

as opposed to style or predisposition, that appears to be

optimal under various circumstances. The model categorizes

four major behavioral areas and related them to the three

leadership styles previously mentioned. A listing of each

behavioral area with some specific examples of positive

leader behavior follows [Fox 1976a, pp. 16-27]:

Leader Support. Leader is friendly and easy to approach,
is willing to listen and attentive to problems and other
matters people wish to discuss. He freely gives de-
served credit to others, does things to make it pleas-
ant to be a group member and to strengthen the self-
esteem of his subordinates.

Consultative-Participative, Decision-Making. Leader
encourages subordinates to exchange information and
ideas. When possible, he invites group member sugges-
tions and gives serious consideration to them before
finalizing decisions which affect either the individual
or the group. If in a position which requires his

his retention of veto power, he uses it as sparingly as
possible. He appropriately delegates decision-making
to individuals, the group, or subgroups with a level
of collaboration desired by them. When possible, he
gives advance notice of changes and is candid and open
to questions. Whenever possible, he strives for deci-
sions relative to the pursuit of given organizational
objectives which accommodate the needs and values of his

Work Facilitation, Leader trains subordinate, consults
with him on job related problems and ways to improve
performance, helps him to plan and schedule work ahead
of time anticipating detailed needs and problems, pro-
vides appropriate equipment, materials, and sees that
decisions are made and implemented in time.

Goal Emphasis. Leader uses appropriate process to de-
velop realistic goals and plans, and gain commitment to
them, he stresses high standards of performance for him-
self and his subordinates, and establishes appropriate
contingencies between rewards and individual and group

Table 1 lists the major constraints to leader behavior in

each of the behavior areas. Two of these constraints (lack

of decisiveness and management of contingencies) were singled

out by Fox to be of particular importance and are found as

separate predictor variables in this study (Decisiveness and

Management of Rewards and Punishment).

In conjunction with research carried out among ROTC

cadets while testing his normative model, Fox [1976c] devel-

oped a leader behavior questionnaire which measured both the

four behavioral areas in his normative model as well as the

two behavioral areas in the Ohio State Leadership Studies

model (C and I.S.). Each variable score for a leader's be-

havior is computed by combining the individual scores of a

subset of questionnaire items. A group score is obtained by

averaging the ratings for any one variable for all group


Table 1



Leader Support


Work Facilitation

Goal Emphasis


Required Social Distance
Therapist Role Pitfall
Goal Conflict

Real Time Pressure
Absence of Subordinate Desire
Selective Interest
Leader Can't Deliver
Differential Expertise
Need to Impose Decisions

Lack of Influence
Lack of Expertise
Lack of Decisiveness

Low Consultative-Participative
Low Technical-Conceptual Abil-
Lack of Commitment
Management of Contingencies

*[Fox, 1976a, p. 35]

members.. The first 54 items on Fox's 60 item questionnaire

were selected from Form XII of the Ohio State Leader Behavior

Description Questionnaire, slightly modified to fit the spe-

cific research situation --ROTC cadets on summer camp ex-


As used in this study, the Leader Behavior Description

Questionnaire for an Independent Auditor in Charge in Field

Work (Appendix B) was developed from Fox's leader behavior

questionnaire, although some items were not used and eleven

new items were introduced. These new items are designed to

cover those facets of Fox's four leader behavior area defin-

itions which were not included among specific questions on

his original instrument. This work was carried out at the

suggestion of Dr. Fox and under his guidance. The resulting

52 item questionnaire yields predictor variable scores based

on the question subsets as shown in Table 2. A measure of

the situational variable Leader-Member Relations is also

available from a question subset (Nos. 5, 6, 27, 44, 48 and

52). It will be noted that the entire set of questions con-

tains many that are used for more than one variable's meas-

urement. This stems from the fact that C and I.S. purport

to cover all of the behavioral classes that are significant

in leadership as do L.S., C-P D-M, W.F., and G.E. There

are simply two different models incorporated into the ques-

tionnaire design and they both cover the same kinds of behav-


Table 2




Initiating Structure

Leader Support

pative Decision-

Work Facilitation





C-P, D-M


Item Subset (see note below)

1, 6, 9, 10, 2, 25, 27, 44

18, 21, 22, 30, 32, 41, 43, 45

1, 4, 6, 19, 27, 33, 44, 48

5, 9, 10, 25, 26, 28, 32, 40

2, 8, 10, 17, 22, 24, 30,

38, 45, 46, 50


Goal Emphasis

Management of Rewards
and Punishment



3, 11, 20, 24, 34, 36, 39

7, 14, 16, 18, 23, 35, 42, 47,

13, 15, 29, 31, 37

NOTE: The score for each variable is a simple average of the
individual item scores from the subset. If the score
is from the Leader Behavior Description Questionnaire
for an Independent Auditor in Charge of Field Work
(Appendix B) it represents a measure of actual behav-
ior for the particular variable. If the score is
from the Ideal Leader Behavior questionnaire it repre-
sents a measure of optimal leader behavior as expected
by the subordinate (Appendix D) or the leader (Append-
ix E). Group (team) scores are produced by averaging
the subordinate scores for each variable in a group.


Certain minor details that are expected to rule out

potential sources of error are worth mentioning. The order

of items on the instrument was determined by a random dis-

tribution so as to avoid any biases that might result from

having each variable's subset intact. Three experienced

CPAs reviewed the appropriateness of the questions in the

Leader Behavior Description Questionnaire (Appendix B) as

well as subsequent research instruments in the study. Partic-

ular emphasis was placed on using language which would be

neither unfamiliar nor ambiguous inthe auditing environment.

The possibility of reading a sexist bias into the Ohio State

LBDQ question forms (every item begins with "He") was elim-

inated by replacing "He" with "This Auditor" and rewording

the questions where possible to eliminate masculine posses-

sive pronouns. In certain cases the range of possible an-

swers (Always, Often, Occassionally, Seldom, Never) was ex-

panded to include a "No Basis" selection. It was believed

that this would strengthen a respondent's confidence in the

instrument for those cases where his knowledge of how the

auditor in charge acted was vague or nonexistent.

Certain basic assumptions underly the research enumer-

ated in this chapter and stipulate the measurements selected.

In part, this study is a test of the applicability of the

Fox Normative Model and the Ohio State Leadership Studies

Model in the public accounting context. Some of the working

hypotheses which apply to these models and this study follow:

Hypothesis 1
There will be significant variance among individual
teams on measured dimensions of tear leader behavior
(predictor variables) in terms of perceptions of ac-
tual behavior.

Hypothesis 2
There will be significant variance among individual
team members on measured dimensions of ideal leader

These two hypotheses reflect an assumption common to

most leader behavior research -- that there are individual

differences with regard to perceived leader behavior and

preferences of normative leader behavior. Aggregations of

particular individual member's scores on leader behavior di-

mensions into group average scores for a leader should also

result in a range of scores for the leaders whose groups are

averaged, since leaders may be expected to behave differently

in response to similar situations.

Hypothesis 3
There will be significant correlation between average
team measurements on dimensions of leader behavior
(predictor variables) in terms of perceptions of ac-
tual leader behavior and average team measurements
of satisfaction and performance (criterion measures).

This hypothesis assumes that the models have predictive ability

for the effects of alternate forms of leader behavior on job

performance and satisfaction.

In view of the current consensus regarding the relation-

ship between leadership effectiveness and situational vari-

ables, a decision was made to measure situational character-

istics. The most comprehensive summary of the moderating

effect of situational variables between criterion and leader

behavior measures was found in the Kerr et al. [1974] situa-

tional propositions. Proposition 1 of this study follows:

(1) The greater the amount of pressure, the greater
will be subordinate tolerance of leader Initiat-
ing Structure, and the greater will be the (pos-
itive) relationships between Structure and sat-
isfaction and performance criteria. Pressure
may stem from the nature of the task (degree of
time urgency, uncertainty, permissible error
rate) or from some threatening source external
to the task.

Based upon this proposition, the following items were in-

cluded in the Audit Staff Member Opinion Survey (Appendix

C): measures of degree of time pressure (questions 9 and 10),

permissible error rate (questions 11 and 12), and threats

from exterior sources (questions 13 and 14). Uncertainty

was dealt with separately. (See the discussion following

Proposition 6).

Propositions 2, 3 and 4 consider the effects of intrin-

sic satisfaction on subordinates [Kerr et al. 1974]:

(2) The greater the intrinsic satisfaction provided by
the task, the less positive will be relationships
between Consideration and satisfaction and perform-
ance criteria. Intrinsic satisfaction may be de-
rived from high job autonomy, broad job scope, or
the opportunity to do interesting and meaningful

(3) The greater the intrinsic satisfaction provided by
the task, the less negative will be relationships
between Structure and subordinate satisfaction.

(4) The greater the intrinsic satisfaction provided by
the task, the less positive will be relationships
between Structure and performance.

In accordance with the sources listed in Proposition 2, three

facets of intrinsic satisfaction have been incorporated in

the Audit Staff MemberOpinion Survey: job autonomy

(questions 15 and 16), breadth of job scope (questions 17 and

18), and meaningful work (question 19).

Proposition 5 considers the effects of job-related in-

formation on subordinate criteria [Kerr et al. 1974]:

(6) The greater the amount of task certainty, the
greater will be the (positive) relationships
between leader Consideration and subordinate

Proposition 9 is particularly appropriate to the CPA firm

environment [Kerr et al. 1974]:

(9) The greater the perceived organizational independ-
ence of subordinates, the greater will be the
(positive) relationships between leader behavior
variables and satisfaction and performance criter-
ia. Particularly when perceived independence is
low. Consideration-satisfaction and Structure-
satisfaction relationships will be relatively in-

Traditionally, the organizational independence of CPA firm

staff members has been high. As with other professionals,

a certain degree of job autonomy must be accorded CPA firm

staff members to provide the necessary atmosphere of profes-

sionalism. However, the primary generator of independence

has probably been the rapid growth of the profession in re-

cent decades which has tended to create a shortage in public

accounting, as well as in other areas, of qualified profes-

sionals. This has resulted in a prevailing 'situation where

any competent staff member could expect to find employment

with other CPA firms or among a wide choice of accounting

positions outside of CPA firm practice. Questions 7 and 8

of the Audit Staff Member Opinion Survey (Appendix C) are

included to measure how the staff member perceives his in-

dependence with regard to alternate employment in these two


No items are included in the Audit Staff Member Opinion

Survey to specifically test Propositions 7, 8, or 10 [Kerr

et al. 1974]:

(7) The less the agreement between subordinate expec-
tations of leader Consideration and Structure and
their observations of these behaviors, the lower
will be the levels of satisfaction and performance
of subordinates. Such expectations typically re-
sult from a host of cultural, experiential, and in-
formational sources.

(8) The less higher management is perceived to exhibit
Consideration, the lower will be the (positive)
relationships between lower-level supervisors'
consideration and subordinate satisfaction.

(10) The greater the perceived upward influence of the
supervisor, the greater will be the (positive)
relationships between consideration and subordinate
satisfaction. This will be especially true for
subordinates who are highly dependent on their loss
for such things as recognition, freedom, and phys-
ical and financial resources.

Proposition 7, a postulation of the relationship between

the degree to which subordinate expectations of leaderbehav-

ior are met and performance and satisfaction measures, may be

tested using the difference between Leader Behavior Descrip-

tion Questionnaire and Ideal Leader Behavior (Appendix D)

results as a measure of the degree to which expectations are


Propositions 8 and 10 [Kerr et al. 1974] deal with (1)

the contrasting degrees of Consideration between higher and

lower-level management as perceived by subordinates and (2)

the perceived upward influence of the supervisor. It was de-

cided that these moderator variables are not appropriate in

the CPA firm environment. All staff members are a part of

"management," Even relatively inexperienced staff members

find themselves in supervisory roles on some engagements.

There is an opportunity for exchange between junior staff

members and the managers and partners of a CPA firm that pre-

cludes the usual sorts of barriers which isolate workers

from management. The flexible, informal type of organization

in CPA firms also tends to reduce the importance of a super-

visor's upward influence, It is also worth noting that a

superior-subordinate relationship in a CPA firm is a tempor-

ary condition. As soon as an audit is completed, the staff

members take positions on new audit teams. The hierarchy in

aCPA firm tends to be flatter than in most organizations,

with usually about four levels of professional staff and

numerous members at each level. Consequently, the bulk of

the power in personnel and operational matters is disbursed

among a group of partners and managers making upward influence

a highly complex item for measurements.

The Kerr et al. [1974] study provides the basis for

the first nineteen items in the Audit Staff Member Opinion

Survey (Appendix C) --which are designed to measure moder-

ator variables. The remaining six items (Nos. 20 to 25) in-

clude four measures of satisfaction (with colleagues, with

compensation, with the leader, and with the present job

situation), a measure of leader-member relations, and a

measure of the current leader's decision-making style. The

necessity for the latter two will be discussed later.

The working hypothesis which is to be tested with the

Kerr et al. [1974] moderator variables follows:

Hypothesis 4 (Moderated Ohio State Leadership Studies
Model) The correlation between average team predictor
variable measurements and criterion measures will be
enhanced by classifying teams into groups based upon
average team scores in certain follower characteristics
as proposed by Kerr et al. [1974].

In other words, the relationship between predictor variable

and satisfaction or performance will be stronger when the

influences of certain follower characteristics are screened


The four subordinate satisfaction items included in the

Audit Staff Member Opinion Survey (see questions 20, 21, 23

25, Appendix C) are derived from the Survey of Organizations.

Taylor and Bowers [1972, p. 76] assumed that satisfaction

was multidimensional and they used seven questions to cover

satisfaction with peers, supervisor, job, organization,

compensation, progress, and opportunity. For the purpose of

this study, job and organizational satisfaction were combined

into one question. It was concluded that satisfaction with

a job in a CPA firm would be tantamount to satisfaction with

the organization since every staff member moves throughout

the organization, working with a variety of colleagues and

supervisors. Subordinate audit staff members also perform

varied tasks, depending upon a client's needs. To a certain

extent, then, the concept of a job with fixed, familiar tasks

is foreign to audit staff members.

Questions regarding satisfaction with progress and op-

portunity were considered unnecessary in this study which is

geared to large CPA firm staff members. Those who do not make

satisfactory progress are soon separated from the firm--

often, placed with a client. Promotions, for those staff

members who remain, are practically automatic. Consequently,

lack of progress should not be a problem in large CPA firms.

Satisfaction related to opportunity for advancement in the

organization should not be a problem, either. Generally,

subordinate staff members hired are believed to have "part-

nership potential." The partner's position in a large CPA

firm is both responsible and financially rewarding--a suf-

ficiently advanced opportunity to satisfy most.

The Ideal Leader Behavior questionnaire (Appendix D) is

identical in most respects with the Leader Behavior Descrip-

tion Questionnaire (Appendix B). The major difference is that

instead of asking for the actual behavior of a specific lead-

er, the Ideal Leader Behavior questionnaire asks "what the

IDEAL auditor should do" in similar situations. The order

of questions is identical and the same subsets are used to

arrive at the eight leader behavior predictor variables. Dis-

crepancies between the ideal version and the actual version

may be used as a measure of subordinate satisfaction. This

is in accordance with the Kerr et al. [1974] Proposition 7

and Porter and Lawler [1968, p. 31].

The three questionnaires discussed up to this point rep-

resent the entire set administered to audit team subordinates.

What is obtained, then, is the subordinate staff member's

perception of his team leader's actual behavior with regard

to 52 items or situations. Subsets of these provide measures

of the leader's behavior in terms of Consideration, Leader

Support, and certain other predictor variables (see Table 3)

as observed on a specific audit engagement. Measures in terms

of the subordinate staff member's ideal for behavior from a

leader in the same situations are also gathered. And, fin-

ally, measures of a number of potentially significant moder-

ator variables are obtained (see Table 4) as well as several

individual satisfaction measures.

The criterion measure for this study is a global assess-

ment instrument (see Appendix G). It asks the manager or

partner to evaluate the auditor in charge (supervisor) of the

field work phase in a particular audit in terms of other per-

sons in similar positions within the same firm. In part 1 of

this questionnaire the auditor in charge is evaluated for

deadline promptness, efficiency of staff utilization (two

aspects), training ability, and crises avoidance. Parts 2

and 3 require the manager or partner to evaluate subordinate

staff satisfaction (three aspects) and client satisfaction

(four aspects). The purpose of these three preliminary parts

of the "Manager's Rating of Field Work Performance" is to

force the manager to think of the audit from the client's and

subordinates' standpoint as well as from his own. When this




Measuring Instrument


Initiating Structure
Leader Support
Decision Making
Work Facilitation
Goal Emphasis
Management of Rewards
and Punishment

LPC (Self Description)
Social LPC
Task LPC
LPC Difference Score


Behavior Description

Leader Behavior


Fox LPC Scales


Task Certainty
Information Needs
Organizational Independence
Intrinsic Work Satisfaction
Leader-Member Relations
Leader's Decision-Making

Audit Staff Member Opinion


Team Performance
Team Leader Performance

Satisfaction with Colleagues
Satisfaction with Compensation
Satisfaction with Current
Overall Job Satisfaction

Satisfaction with Leader

Supervisor's Rating of Field
Work Performance

Audit Staff Member Opinion

Difference (ILB LBDQ)







LPC Self

Task LPC

Social LPC

LPC Difference

Version of
LPC Instrument

Least Preferred

Least Preferred

Least Preferred

Least Preferred

Total Score
Item Subset (See Appendix F)





(Task LPC Social LPC)

is done, the possibility of a standardized overall evaluation

of leader effectiveness and team effectiveness may be improved

over a point blank, one-item rating. Question 4a (How would

you evaluate the overall effectiveness of this team based

upon your review of the field work?) is the criterion measure

for team performance. Question 4b (How effective was the

auditor in charge of field work as a leader on this particu-

lar audit?) is a performance evaluation for the leader, alone,

separated from the success or failure of the team. These will

be analyzed against the results of the other questionnaires

as two distinct performance criterion measures.

Each audit team leader (supervisor) completed the Ideal

Leader Behavior questionnaire (Appendix E) as did his sub-

ordinates on the audit. Any differences between a supervis-

or's behavior expectations and the average responses of his

subordinates could then be analyzed to determine whether ex-

pectations are role dependent. The cover sheet on the Ideal

Leader Behavior questionnaire administered to team leaders

contains an additional question which illicits leader-member

relations as judged by the team leader on a five point scale

ranging from very good to very poor. This question is similar

to question 22 of the Audit Staff Member Opinion Survey which

asks for the subordinate's judgment of leader-member relations.

Each team leader must complete the Fox LPC Self Descrip-

tion instrument (Appendix F) in duplicate -- one version for

a least preferred coworker and the second version describing

himself. The instrument is expanded over the standard

sixteen scale version used by Fiedler to include scales that

fully cover the five recurring dimensions found by Tupes and

Christal 11961]: urgency, agreeableness, dependability, emo-

tional stability, and culture. A complete review of the de-

velopment of this instrument is included in Fox [1976d].

For this study, two LPC instruments are employed. The

supervisor's rating of his least preferred coworker will be

obtained by combining the scores for fourteen scales from the

32 item instrument-- least preferred coworker version. This

measure will be referred to as LPC [Fox, 1976a, p.4]. The

supervisor's rating of himself is termed LPC self-description

and is obtained by combining scores for a different subset

of fourteen items on the appropriate version of the instru-

ment. It should be noted that the respondent checks either

of two boxes at the top of the instrument to identify the in-

strument as a description of "My Least Preferred Co-worker"

or "Me." Additional LPC measures are possible from these

instruments (Task LPC, Social LPC, and a Difference Score)

by combining different subsets of scales (see Table 4).

The various LPC measures obtained may then be used as

predictor variables to test the following working hypothesis:

Hypothesis 5 (Fiedler Contingency Model)
There will be a correlation between leader LPC score
and measures of leader and group effectiveness as
proposed by Fiedler [1967].

One item has been included in the study in an effort to

single out each team leader's decision method (item 24, Ap-

.pendix C). This is based upon the Normative Model of

Leadership proposed by Vroom and Yetton [1973]. Five deci-

sion methods open to a leader are proposed in the case of

individual problems -- decision problems encountered by one

subordinate member of a group. The methods and their assigned

symbolic notations are IVroom and Yetton, 1973, p. 13]:

A 1 where the leader makes the decision for
himself, using only information available
to him at that time.

A 11 : where the leader makes the decision by
himself after obtaining information from
the subordinate.

C 1 : where the leader makes the decision after
sharing the problem with his subordinate
and hearing his suggestions and ideas.

G 1 :where the leader and subordinate arrive
at a mutually agreeable decision after
analyzing the problem together.

D 1 : where the subordinate makes the decision--
the leader having delegated that respons-
ibility to him and provided him with any
information that he held.

Vroom and Yetton categorize eighteen problem types based

upon the presence or absence of eight problem attributes.

Their Normative Model of Leadership included one or more of

the above decision methods as the preferred solution for each

of the problem types. An analysis of the eighteen problem

types revealed that only four were applicable to the situa-

tions encountered by subordinate audit staff members. In

each of these problem types, style G-l was appropriate. In

some cases D-l and C-1 were also appropriate depending upon

the sufficiency of subordinate information and the importance

of the subordinate's acceptance of the decision.

In the particular question selected for use, it is

stated that the subordinate has sufficient knowledge and in-

formation to make a quality decision. Consequently, for the

case included on the questionnaire the preferred solution may

be any one of the latter three. However, the objective of

this study is not to test the Vroom and Yetton model, but to

determine if there is a significant relationship between any

one decision-making style and the criterion measures.

The research design for this study is present in Appendix



The field study was performed with employees of one of

the "big-eight" accounting firms. Audit teams were selected

which met the following criteria:

(a) Four or more regular members (excluding specialists)
(b) One individual in charge of field work during that
entire phase of the audit.
(c) Ninety days or less from the time the team completed
the field work.
(d) Each team member must have been on the audit for at
least ten days to be considered a regular member.

Five offices in this accounting firm (all located in the

southeastern U.S.) participated in the study. Thirty-four

teams were found in these offices which met the above criter-

ia. An effort was made to include all possible teams, since

in this way a representative cross-section of large audit

teams would be tested. The firm could not have the option

of submitting only its best audit teams (in terms of overall

performance and member satisfaction) for testing.

The firm did not consider assembling all of the employ-

ees for testing at one time as a feasible way to conduct the

study. Their contention was that this would be too expensive.

They also preferred to designate one of their own represent-

atives in each office to administer the questionnaires. In

each case, it was a partner or personnel specialist who had

been involved in the scheduling of audit field work.


Meetings were held with each of the five individuals who

were to administer the questionnaires. The purpose of these

meetings was to provide these individuals with all of the

background information that they would need in order to ad-

minister the instruments. The models utilized in the study

were explained and then each questionnaire was reveiwed to

show its contribution to the study. During these meetings,

the teams to be tested were selected.

In order to provide anonymity to the testing process

special cover sheets were designed for the sets of question-

naires provided for staff members (team leaders' supervisors).

A code number for each team was assigned by the administra-

tor and entered on the cover sheets (see Appendices A, B, and

C). The audit engagement and team leader were entered fol-

lowing the code number. Each individual was instructed to

enter the team code number on all questionnaires in his set.

Staff members were further required to coin a six digit per-

sonal code number and enter it on each questionnaire. The

result was that each questionnaire could be sorted by team

and team member, yet no name remained on any of the documents.

The name was entered on the cover sheet, which was to be re-

moved and destroyed by the individual completing the question-


Although it would perhaps have been preferable to admin-

istor all of the questionnaires at the same time of year,

this was not possible. Most of the annual work is scheduled

at various times of the year --not necessarily following

the fiscal year-end. A further complication is that many

corporate clients use fiscal years which end throughout the

calendar year.

Twenty-two correctly completed sets of questionnaires

were eventually obtained. Of the twelve teams that were not

usable, there were several reasons. In one case, the audit

manager did not feel that he should evaluate the team's per-

formance for an outsider, even with the anonymity provided.

The most frequent problem was the loss of one team member due

to resignation, transfer, sickness, or insufficient time in

the field.

Frequent inquiries were made to the test administrators,

urging them to follow up on those team members that had not

been tested. A few of the discrepancies were eventually

filled; however, once an audit team had been disbanded in ex-

cess of ninety days, the team was excluded from the sample.

The result of this strict enforcement of the criteria

for audit team selection was to limit the statistical base

for the study. This is the case, particularly, where the

analysis of results for the moderator variable effects is


One of the reasons why there were so few teams large

enough to be suitable for the study is that auditing tech-

niques in recent years have tended to become more scientific

and less laborious. Greater emphasis is placed on an evalu-

ation of internal control, statistical sample selections, and

upon analytical techniques which provide assurance without

extensive verification. Also, since virtually all clients

of a large CPA firm use a computerized information system,

a major concern for the auditor now is to evaluate the system

to determine the probability for material breakdowns in con-

trol. Audits of this genre do not require large teams to

test and evaluate account balances in search of human error.

The fact that all of the offices were located in the

southeastern U.S. is perhaps a positive factor in the study.

To whatever extent there remain regional differences in atti-

tudes toward authority, work habits, or job expectations,

these would appear to have been screened out by using offices

only in one region.

The use of firm representatives to administer the ques-

tionnaires has both positive and negative aspects; the chief

advantage being that the individuals tested realized that the

study was sanctioned by the firm and might be expected to

adopt an attitude of sincerity in completing the question-

naires. The negative aspects of having the firm's represent-

atives administer the questionnaires are that they lack the

commitment toward the project and the intimate knowledge of

the study that the originator would bring.

Chapter V


That leaders were perceived to behave differently by

their teams may be demonstrated by noting the statistics on

Table 5.. Perceived leader behavior on the eight dimensions

measured showed a certain amount of variability as assumed in

working hypothesis number one, Chapter III. The greatest

variability was measured in the two Ohio State Leadership

Studies variables [Stogdill and Shartle, 1955], Considera-

tion and Initiating Structure, with standard deviations of

.35 and .29, respectively. It should be pointed out that

scores in any dimension could range from a minimum of one to

a maximum of five. The six Fox Normative Model variables

[Fox, 1976a] showed somewhat less variability, standard de-

viations ranging from .27 down to .15. Means for all eight

predictor variables fell within a range between 3.57 and 4.12.

No particular conclusions should be drawn from these mean

values, due to the lack of comparative studies using the same

instrument with other groups.

There was some variability in expectation of ideal be-

havior from leaders as seen by individual team members, which

was assumed in working hypothesis number 2 (see Table 5).

As a matter of fact, the variability in the ideal behavior

dimensions was generally greater than in the perceived


Table 5


A. Perceived Leader Behavior (n = 22 items)

Variable Mean S.D. Min.

Consideration 3.94 .35 3.38

Initiating Structure 3.72 .29 2.92

Leader Support 4.12 .26 3.75


Work Facilitation


Goal Emphasis

Management of Rewards and

B. Ideal Leader Behavior (n



Initiating Structure

Leader Support


Work Facilitation


Goal Emphasis

Management of Rewards and






= 69 respon

Mean S.D.

4.32 .34

4.05 .41

4.46 .32

























.19 3.44 4.20



































4.01 .25 3.40 5.00 .03

behavior dimensions for the same variable. This was a sur-

prising outcome--an indication that the perceived behavior

of the 22 leaders was somewhat less variable than the con-

sensus of how the leaders should act. Evidently, the team

leaders tended to behave similarly-- possibly this is a re-

sult of the firm's stated personnel policies and informal


The mean scores in seven out of the eight ideal behavior

variables were higher than the comparably perceived leader

behavior scores in the same variable. This was the expected

result. The one variable where leaders perceived behavior

scored higher than the ideal was on the dimension of Decisive-

ness. Apparently, CPA firm audit staff members prefer less

decisiveness. They expect to be involved in the decisions

which affect them.

A comparison of leaders' and subordinates' expectations

with regard to leader behavior may be seen in Table 6. The

greatest discrepancy between the mean predictor variable

scores for the two groups was .29 (Consultative-Participative

Decision-Making). All but two other variables showed mean

differences of less than .20. Perhaps the most notable find-

ing is that if the mean scores are rounded to the nearest

whole number, all are equal. In six of the eight variables,

the subordinates' ideal scores are higher. However, this

difference is always less than one standard deviation. Ap-

parently, there is little role dependency with regard to

Table 6


Comparison of Leaders' and Subordinates' Expectations



Initiating Structure

Leader Support

tive Decision-Making

Work Facilitation


Goal Emphasis

Management of Rewards
and Punishment


Mean S.D.

4.07 .21

4.14 .31

4.41 .22










Mean S.D.

4.32 .32

4.05 .41

4.46 .32









4.01 .25

4.00 .23

expectations of leader behavior among the 69 subordinates

and 22 leaders tested.

In accordance withworking hypothesis number three, Chap-

ter III, Pearsonian correlation coefficients were computed

between predictor variables and assessments of satisfaction

and performance. (See Tables 7, 8, and 9.) Of the six Fox

Normative Model variables, Consultative-Participative Deci-

sion-Making, Goal Emphasis, and Management of Rewards and

Punishment did not show any significant correlation with the

criterion scores. Leader Support correlated .41 with Leader-

Member Relations and -.47 with the difference between the

group averages for the 52 leader behavior items. This dif-

ference is labelled Discrepancy Measure, and is expected to

be a measure, as opposed to an assessment, of group dissatis-


This result raises a question of to what (negative) de-

gree the Discrepancy Measure indicates overall job satisfac-

tion. Porter and Lawler [1968, p. 31] define satisfaction

as "the extent to which rewards actually received meet or

exceed the perceived equitable level of rewards." The Dis-

crepancy Measure used in this study is an average of all of

the differences (ideal minus actual score) for each of the

52 items selected for this study from the LBDQ (see

both Appendices B and D). A review of the items on these

questionnaires demonstrates an element of positive or nega-

tive reward in most: for example; item (1), "treated all

subordinates on the audit as professional colleagues; item

Table 7


Correlation between Predictors and Criterion Scores

Predictor Variable

L.S. D-M

W.F. D G.E. M.R.P.

0 -.06

p -.13

Q .41*

T -.04





Y .00 -.01

Z .17





.08 -.10

.21 -.16

.20 .19

.42** .29

.46** .36*

Q31 -.47** -.28




-61** -.56** -.27

* denotes probability of randomness less than .10

** denotes probability of randomness less than .05

Symbol Variable


Leader Support
Consultative-Participative Decision-Making
Work Facilitation
Goal Emphasis
Management of Rewards and Punishment
Satisfaction with Colleagues
Satisfaction with Compensation
Leader-Member Relations
Overall Job Satisfaction
Team Performance
Leader Effectiveness
Discrepancy Measure (Ideal Behavior Minus Actual









Table 8


Correlation between Predictors and Criterion Scores

Predictor Variable

C I.S.

0 -.07 -.05

P -.12 -.10

Q .33 .07

Criterion Measure

T -.08

Y -.07 -.05

Z .07 .17

Q31 -.49** -.26

** denotes probability of randomness less than .05


Initiating Structure
Satisfaction with Colleagues
Satisfaction with Compensation
Leader-Member Relations
Overall Job Satisfaction
Leader Effectiveness
Discrepancy Measure (Ideal Behavior Minus Actual




Table 9


Correlation between LPC Measures and Criterion Scores





O .11

P -.15

Q -.03

T -.06



.02 -.09 -.08

.14 -.04 -.15 -.02

.17 -.03

Y -.42** .19

Z -.24

Q31 -.06




-.46** -.53**

-.29 -.22




.39 -.34

* denotes probability of randomness less than .10

** denotes probability of randomness less than .05

Symbol Variable

Task LCP
Social LPC
LPC Difference Score (Social LPC-Task LPC)
LPC Self Description
Satisfaction with Colleagues
Satisfaction with Compensation
Leader-Member Relations
Overall Job Satisfaction
Team Performance
Leader Effectiveness
Discrepancy Measure (Ideal Behavior Minus Actual




(4), "gave deserved credit to subordinates when dealing with

senior colleagues," item (6), "did little things to make it

pleasant to be on the audit," and item (9), "was willing to

make adjustments in the audit program to suit individual


Those items that do not represent a reward in the per-

sonal sense (for example; item (2), "anticipated problems

and planned for them" and item (7), "pushed to get the field

work done with a minimum of wasted time and effort") may

still be regarded as related to overall job satisfaction.

One must assume that a staff member serving on an audit team

in the field would be less satisfied with his job, regardless

of the level of social deference directed toward him, if the

main goal--to gather evidence--was thwarted by poor organ-

ization and planning.

Work Facilitation correlated .42 with Team Performance

and .46 with the Leader Effectiveness assessment. Decisive-

ness correlated .36 with Leader Effectiveness. Work Facilita-

tion and Decisiveness correlated most strongly with the Dis-

crepancy Measure, -.61 and -.56, respectively.

The only significant correlation between criterion meas-

ures and the two Ohio State Leadership Studies predictor vari-

ables was between Consideration and the Discrepancy Measure,

-.49. Initiating Structure did not show any predictive abil-

ity in this sample.

The Ohio State studies are not in disagreement with these

findings. Initiating Structure and Satisfaction seem to be

positively related in situations where group members perceive

a need for structuring activities on the part of the leader

and negatively related where the need does not exist. It

would seem apparent that an audit team could fit in either

category. The situation is unique in that most of the audit

tests and procedures may never have been performed before in

the context of the current client organization and accounting

system. The audit team, itself, would be a recently assembled

group where the members may have had no previous working re-


Although the LPC Model does not presume any predictive

ability for the LPC measures, Pearsonian correlations were

computed between the five LPC measures and the satisfaction

and performance assessments and also with the Discrepancy

Measure. Figure 5 shows the results. LPC, Social LPC, and

the LPC Difference Score correlated strongly with Team Per-

formance. It should be noted that these were negative cor-

relations, indicating that high performance was found in

teams led by low LPC leaders--persons holding their least-

preferred coworker in low regard. LPC Self Description

showed some correlation with the assessment of Leader Effec-

tiveness and the LPC Difference Score correlated with the

Discrepancy Measure .39.

In this study, variability was measured both in per-

ceived leader behavior and in expected ideal behavior. There

was little variation between leaders' and subordinates' no-

tions of what ideal behavior ought to be. Three variables

from the Fox Normative Model correlated with the Discrepancy

Measure--Leader Support, Work Facilitation, and Decisive-

ness. Work Facilitation correlated with both performance

assessments--Team Performance and Leader Effectiveness.

Small correlations also showed up between Leader Support and

Leader-Member Relations and between Decisiveness and Leader

Effectiveness. Both of these corrleations appear logical.

While a discussion of the shortcomings of the study will

be reserved for later, it would seem appropriate at this time

to deal with the performance measures (Team Performance and

Team Leader Effectiveness). Of the 22 teams, 15 were rated

as being "slightly above average for our firm" and four were

given the top rating, "significantly above average for our

firm. Only one team was rated as having performed "average

for our firm." Two were rated "slightly below average for

our firm."

It is possible that the performance measure is biased.

The individual making the rating does have final responsibil-

ity for the audit and by implication is deeply concerned with

its adequacy. Each partner or manager may believe that the

audits performed by his subordinates are better than the

firm's average because he sees to it that those areas which

he considers most important are performed to his satisfaction.

Partners and managers may also have a tendency to believe

that their office is above average for the firm--a preference

based on familiarity rather than experience.

Consideration was the variable from the Ohio State

Leadership Studies Model which had a significant correlation

in this study. That the Discrepancy Measure, a measure of

dissatisfaction, was the criterion measure that correlated

(negatively) was not unexpected.

Three of the LPC variables correlated with Team Perfor-

ance. As a matter of fact, no other variable was more

strongly correlated with Team Performance than LPC, Social

LPC, or the LPC Difference Score. These are interesting ob-

servations, but not tests of the Contingency Model because

the groups are left unclassified as to Leader-Member Relations

in this part of the study. Chapter VII will evaluate the

study results in accordance with the Contingency Model.)

These results discussed to this point demonstrate the

applicability of working hypotheses 1, 2, and 3. Neither the

Fox Normative Model nor the Ohio State Leadership Studies

Model variables appeared to indicate relationships with the

assessments of satisfaction. The situation improves when

the criterion measure of satisfaction is the Discrepancy

Measure. The Ohio State Leadership Studies variables did not

correlate with performance in this study, whereas, certain of

the other predictor variables measured showed some relation-




As discussed in Chapter III, eight of the ten situational

propositions suggested by Kerr et al. [1974] were selected for

testing in this study. The expectation was that a sharper re-

lationship between Consideration and Initiating Structure and

assessments of satisfaction and performance could be found by

using the proposed moderator variables as screening devices.

The usefulness of these results is limited by the performance

measurement bias, discussed in Chapter V, as well as by the

limited sample size.

A quadrant analysis was used to determine whether the

classification by predictor variable (HI or LO ranking in

Consideration or Initiating Structure) was independent of a

similar classification by moderator variable (Pressure, Task

Certainty, Need for Information, Organizational Independence,

and Intrinsic Job Satisfaction). Table 10 shows the results

of the quadrant analyses for the two Ohio State predictor

variables as paired against the five moderator variables

selected as feasible for this study. Although the Kerr et al.

11974] propositions were stated only in terms of the relation-

ship of the Ohio State predictor variables to satisfaction

and performance, it seemed plausible to search out any effects

that the moderator variables might have on the Fox Normative

Table 10

Phi Coefficient
Probability of Randomness

Moderator Variable





Structure .000 1.473 .733 .733 .733
.000 .259 .183 .183 -.183
1.000 .22 .39 .39 .39

Leader .182 .000 .182 1.636 .182
Support -.091 .000 -.091 .237 .091
.67 1.000 .67 .20 .67

Consultative- 4.545 .786 .182 1.636 .182
Participative -.455 .189 .091 .273 .091
D.M. .03Y* .38 .67 .20 .67

Work 1.636 .786 .182 1.636 .182
Facilitation -.273 -.189 .091 -.273 -.091
.20 .38 .67 .20 .67

Decisiveness .182 .786 4.545 .182 .189
.091 -.189 .455 -.091 .091
.67 .38 .03* .67 .67

Goal .182 .000 1.636 .182 1.636
Emphasis -.091 .000 .273 .091 -.273
.67 1.00 .20 .67 .20

Rewards and




Need for Information
Task Certainty
Organizational Independence
Intrinsic Satisfaction





Items from App. C, CI III

** Denotes probability of randomness less than .05




Model predictor variables as well. For each combination of

predictor variable (8) and moderator variable (5), the 22

teams were assigned to one of four quadrants: quadrant 1,

HI classification in both predictor and moderator; quadrant

2, LO classification in both predictor and moderator; quad-

rant 3, LO predictor classification and HI moderator classif-

ication; and quadrant 4, HI predictor classification and LO

moderator classification.

Table 10 shows the results of the test for covariation

between the predictor and moderator variables based upon the

number of cases falling in each category represented by the

four quadrants. The chi-square statistic was computed and

the strength of association was then found by measuring the

phi coefficient. None of the Ohio State predictor variables

showed a strong covariation for these tests--the relation-

ship between Initiating Structure and Task Certainty appeared

to be the greatest (phi coefficient of .259 and a probability

of randomness of .22). There were two strong relationships

between the Fox predictor variables and the moderator vari-

ables; and, in retrospect, both appear to be quite logical.

Consultative-Participative Decision-Making and Need for In-

formation had a phi coefficient of -.455. The indication

here being that where the subordinate's need for information

is high, he does not experience a high degree of participa-

tion in the decision making process. Decisiveness and Organ-

izational Independence has a phi coefficient of .455. Those

subordinates who felt independent of their firm, or highly

mobile in the job market, perceived their leaders as highly

decisive; whereas those subordinates who did not consider

themselves as mobile at the present time perceived their

leaders as indecisive. It is likely that the mobile employ-

ees are the more experienced staff members--ones that would

force their leaders into decisiveness by making decisions on

their own as their level of expertise warranted. Consequent-

ly, these employees would tend to present relevant problems

to the leader for solution in situations where the leader

must act. Inexperienced subordinates might ask for decisions

in situations where none is required or where information is

too scant for a thoughtful decision. In these cases they

could be left with a perception of their leader as indecisive.

Table 11 also shows the results of chi-square tests.

However, in this instance, the predictor variable scores used

are the ideal behavior measures. The moderator variables ap-

pear to exert a greater influence on expected (ideal) behav-

ior than on actual behavior as perceived by subordinates.

A strong relationship was noted between Initiating Struc-

ture and Intrinsic Satisfaction. The phi coefficient was

negative, -.455, which may be an indication that where there

is a high degree of satisfaction inherent in the job, subord-

inates do not want a leader to engage in structuring activi-

ties to a high degree.

The Fox Normative Model predictor variables (ideal be-

havior) yielded four strong relationships with the moderator

variables. Leader Support and Task Certainty, Consultative-


Table 11


Phi Coefficient
Probability of Randomness

Moderator Variable



Consideration .000

Initiating ,182
Structure -.091

Leader ,188
Support -.092

Consultative- 2.933
Participative .365
D.M. .08

Work .182
Facilitation .091



Management of
Rewards and





Items from Ap. C, Ch. III

Need for Information 1,2
Task Certainty 6
Organizational Independence 7,8
Pressure 9,1
Intrinsic Satisfaction 15,



** Denotes probability of randomness less than .05


































Participative Decision-Making and Need for Information, and

Decisiveness and Pressure all appear to be strongly related

where the ideal leader behavior and actual situational char-

acteristics are concerned. The phi coefficients for these

three relationships were .524, .365, and -.365, respectively.

The strongest relationship of all was between Decisiveness

and Organizational Independence, phi coefficient.548. The

relationship between Decisiveness and Organizational Inde-

pendence appears logical for the same reason when the pre-

dictor variable is ideal as for the previously discussed per-

ceived case. The relationship between Leader Support and

Task Certainty also follows a logical pattern--where there

is a high degree of task certainty, leader support is desir-

able, The negative relationship between Decisiveness and

Pressure could indicate that in high pressure situations,

subordinates do not want a decisive leader. Perhaps they

would prefer to operate in an atmosphere free of the con-

straint imposed by a highly decisive leader. The relation-

ship between Consultative-Participative Decision-Making and

Need for Information is not parallel in the ideal behavior

situation to the findings for the perceived behavior situa-

tion. In the latter, it was surmised that leaders did not

consult with subordinates in those cases where the subordin-

ates reflected a high need for information. However there

appears to be a positive relationship between the two vari-

ables when ideal behavior is substituted for perceived Con-

sultative-Participative Decision-Making. This may indicate

that even though relatively uninformed, subordinates still

prefer to be consulted,

The chi-square test does not take into consideration the

criterion measure effects. Consequently, a further step was

added to incorporate these into the quadrant analysis so that

the Kerr et al. [1974] situational proDositions could be ex-

amined in this environment. Table 12 contains the results.

The test was basically a simple observation of the difference

between quadrants. Each group was ranked HI or LO among the

twenty-two groups for each predictor variable (Consideration

and Initiating Structure as perceived by subordinates) and

moderator variable in this phase of the study. The appropri-

ate pairings were made to test each of the eight Kerr et al.

[1974] propositions that were feasible for this study. When

each group had been classified into a quadrant for a specific

pairing, the average Overall Satisfaction or Team Performance

was computed for the set teams in the quadrant. These aver-

age criterion measure scores were then compared to determine

if they followed the direction of the proposed results.

Table 12 shows that of the sixteen tests made of the

propositions studied, seven followed the expected direction

of change in criterion measure. Table 13 shows the quadrants

and criterion scores for propositions 1, 2, 3, 4, 5, 6 and 9.

Table 14 shows the results of the tests for proposition 7.

Among the propositions where two or more tests were con-

ducted (1, 2, 7, and 9), the agreement of observations with

expected results was always mixed. Of the propositions

Table 12

Congruity of Prop-sed Effect and seed :srved~ fct ri-=tmality

0-.i State
Pre Cr

pressure s.S.
Intrinsic Sat. C
Tr-.t-.sic at. C
Intrinsic Sat. .S.

Infcrr.s.t c. Need .S.
Task Certainty C
Discrecancy C
Dis C
(-A oerc r varea:e r.n:t

Org. i nc .. ...e c
Org. In dc enence r

Or(.oderzcr variale n.t
(coderatcr variab-e nct

SFtisfs" ic

Sa i fact icr


S 'a--c-- ct


SPropositions are reproduced in Chapter II1
*" See FYcure 2 for listing of i !e:ns used in -r--:ue-en

TESTING P'DCEDURE: Each of the 22 groups was placed into a H: cr LO
classfication for each of the moceratcr variat-es r-nasurei us.-.-
the mean sccre for the variable for each croup as a score fr rac :.
in an array. The 11 highest scoring grous in tne array were c
nated H:. The other groups were desigated L. A s lar rocedur
was used tc cas:Irfy the 2 f zre s 7 I l ; 'fr
thE Chic State rrecdctcr var:abCes. A arat ana,
crCpsltiton testec aE follow:

rredictor Tr 1 e

Quacrarnt I ,u a.. i .ra
(insert aver-i
criteric. sEcr e c.
of group fal!
inc Ir. __ _.-_


Kerr, e- al,

Nr- ra-
\a ZaI

D- T'-e" -t o .



Cr terinC.

Table 13


(See Figure

HI cases: 7

3.86 3.50
M 7--- y ---- o

LO 3.80 14.50

Frcpos-itir. 2
HI cases: 5 6


LO .35 3. 3

Proposition 5
I cases: 65


L 3.85 3.673

LO 3.41 3.52

HI cases: 6 5



for coding cf mierator varabt~l s:
on rcsl o- '-eL'
ProposiOn tn

HI cases: 7 HI cases: 5

sat. perf.
-. 3.0L 3-7 .

Proposition 3
HI cases: 5 6


LO 3.6- 3.25

Proposition 6
HI cases: 7 7

3.a9 3.76

T 7
Lo 3-7. 3-

H cases: 71




HI esers: 5 6

LO _i

Propcstior. 9

T 3.83 '-

HI cases: 7

I^ -- ------- E ----------
sat .-

L.3 C


Table 14

Ccc-arsnr. of Fr-ccsed Effect tc

Average -E ult F.esull
A- Z
_____ rd

Perfc ance




t* The difference te.ween cservec (cerce'ved) Conrsderat..n ar. Ideal
(expected) Cons ierat ion was computed fcr eac tea.. The teams -ere
then ranked intc LC and HI classiflcat ons tasec u-cn tne amcur.t cf
discrepancy as represented by the com-pued difference. Eieven t-rms
were included in each clas3ificatlon. The average Overall Satisa-
tion and Team Performance were cooputed for each classlflcation. The
same procedures were also followed usinc g Ir.tlati Structure as
predictor variable. Propcsitfon 7 states that the higher the dis:r;-
ancy between expeczations of tenavlcr rad ofser-vaticns cf beha-
for both Consideratlcn and Initiatring :ruCt're activities, the lower
will be the levels of satisfaction arnd performance of suocrdinates.




I C W E r

where one test was made (3, 4, 5, and 6), two showed observ-

ations in accordance with the proposed result. Four of the

seven propositions relating to performance and three out of

nine propositions relating to satisfaction were in agreement

with the proposed results. Due to the absence of any pattern

in the results of the tests of the contingency theory propo-

sitions, it is not appropriate to make further conjecture as

to the applicability of these propositions in the auditing




The 32 item Fox LPC instrument (Appendix F, Chapter III)

was completed by each team leader in two versions--for

least preferred coworker and for self, Five predictor vari-

ables were obtained from these instruments; LPC, Task LPC,

Social LPC, and LPC Self Description. Table 4, Chapter III,

lists the item subsets which comprise these variables.

Assignment of teams into octants of Fiedler's classifi-

cation system acquired three measurements; position power,

task structure, and leader-member relations [Fiedler, 1967,

pp. 22-32]. Position power was measured using a set of 13

questions as utilized by J. G. Hunt [Fiedler, 1967, p. 281].

These questions were discussed with the partner or specialist

in charge of assigning staff to audit teams in three of the

offices visited. In every case, at least ten of the questions

were answered in the affirmative-- indicate an attribute

of high position power.

Group task structure was evaluated by having the same

three experts who answered the position power questions eval-

uate an audit team member's job in the manner suggested by

Fiedler and Chemers [1974, p. 68]. Four dimensions of task

characteristic (goal clarity, goal-path multiplicity, deci-

sion verifiability and decision specificity) were scored by

experts on a scale from 1 (low structure) to 8 (high struc-

ture). In each the mean score of the four dimensions was

below 5.0, indicating low task structure. The task structure

for a leader's job would be lower than the team member's job.

Consequently, it can be safely assumed that the task struc-

ture of any team member, to include the leader, or the task

structure of the team as a whole, is low for CPA firm audit


High position power combined with low task structure

indicates octants III and VII of Fiedler's classification

system. Fiedler [1967, p. 121] stated that there are rela-

tively few groups of this type in real life. Two studies

were undertaken by Fiedler and his associates using groups

created in a laboratory setting. The median correlations

reported for these groups were: -.33 where leader-member

relations was good octantt III) and .05 where leader-member

relations was moderately poor octantt VII), [Fiedler, 1967,

pp. 137-8].

Fiedler and Chemers [1974, p. 64-5] indicate two ways

to measure leader-member relations. One method was to ask

group members to indicate on a sociometric scale their ac-

ceptance of their leader as opposed to other supervisors.

The second method was to ask the leader to answer ten bipolar

(eight-point) questions, the sum of which yields a group at-

mosphere score. Fiedler and Chemers [1974, p. 65] admit

that the first method is awkward due to the fact that most

employees do not care to show low regard for their

supervisors. The second method may be criticized on the

grounds that it represents an evaluation by the leader, where

it would be potentially embarrassing for him to admit the

truth in certain cases, For instance, some of the scales are

Efficient-Inefficient, Cold-Warm, and Friendly-Unfriendly.

For this study, four measures of leader-member relations

have been taken. Six items from the Leader Behavior Descrip-

tion Questionnaire (items 5, 6, 27, 44, 48, and 52) when com-

bined and averaged provide a measure of leader-member rela-

tions [Fox, 1976c, Table 19C]. Each team member was asked to

give a global assessment of leader-member relations. Subord-

inates answered the question, "All in all, how would you

judge your relationship with the auditor in charge of field

work for the engagement listed on the cover sheet to this

set of questionnaires?" A group average of these responses

provides a second measure of leader-member relations. Lead-

ers were also asked to answer a similar question in an appro-

priately modified version, providing a third measure of

leader-member relations. A fourth measure was obtained by

taking a simple average of the second and third measures.

It was assumed that equal weight should be given to the

supervisor's and subordinates' respective assessments, since

the phenomena being measured represents an average of a

series of dyadic relationships. Therefore, in each dyad the

leader and the subordinate should be given equal weight.

Table 15 contains the results where the teams are split

into respective octants using the LBDQ subset as a measure


Table 15








Rank-Order Ccrrelation Coefficients between LPC and Criterion Scores

Octant III (Leader-Mlember Relations: Good) n=ll

LPC Task Social LPC LPC
Coworker LFC LPC Difference Self

am Perfcr.ance -.35 .09 -.56* -.70" .0

ader Effectiveness -.40 -.29 -.52' -.23 .2

screpancy Measure -.38 .09 .oE -.1 -.7

.lleaeue Sat. .00 .09 -.17 -.23 .7:

rmpensatlcr Sat. -.32 .18 -.21 -.-5 -.3

era J St. .08 .25 .0 -.3 -.




Octant VII (Leader-Member 'olatlons: .oderately Foor) n=32

LPC Task Social LPC LiC
Coworker LPC LPC Difference Self

Team Perforance -.57* -.02 -.50 -.29 .19

Leader Effectiveness -.39 -.21 -.33 -.0b .27

Discrepancy Measure .50 -.47 .55* .7i** -.0o

Colleague Sat. .12 .03 .4 .34 .15

Compensation Sat. .06 -.16 .19 .19 .2''

Overall Jct Ea:. -.15 -.27 -.C .12 -.2

denctc: prcbatl!ity of rra.-.ness oest that .]C

'' cenctes probat1llty of ranc-.cnes less trar .c

Leader-.'em-:er .elatcnrr scores are a cc-rposte of items 5,E,2,,,
and 52 frc~ the Leader Behavicr Descrlctlon Questionnaire (Apt. z,
Chapter III'. Oct.nt III Teamc: Nos. 3 ,,7,, 9,15 12,1 17,1 ,c
Octant : 2,2,4,,,13, -,:5,20,~, ... ..

of leader-member relations. The rho-value (Spearman rank-

order correlation) between LPC and Team Performance for Oc-

tant III closely agrees with Fiedler's predication of -.33,

however, Octant VII results are quite different. LPC and

Team Perfomrance show a rho-value of -.57, as opposed to

Fiedler's result of .05. Strong correlations appear between

Team Performance and Social LPC -.56 and LPC Difference -.70.

The LPC Self Description Correlates with two measures of

satisfaction; .71 with Satisfaction with Colleagues and -.72

with the Discrepancy Measure, a measure of dissatisfaction.

These last four correlations, all Octant III values, should

be compared with their Octant VII counterparts. In this way,

it can be determined to what extent the LPC variables dis-

criminate between the two octants. It is apparent that the

LPC Difference Score shows the greatest disparity in Team Per-

formance rank order correlations. The other LPC measures

seem to be in the same vicinity, regardless of octant. The

same holds true for all LPC measures in their correlations

with Leader Effectiveness.

The Discrepancy Measure appears to be the criterion

measure which contains the most diversity in correlations

when shifting from Octant III to Octant VII. The spread be-

tween the rank-order correlations is .88 LPC Coworker, .56

Task LPC, .47 Social LPC, .92 LPC Difference Score, and .64

LPC Self Description. The only other correlation spreads

that enter into this range of magnitude are: .41, Team Per-

formance and LPC Self Description; .56, Colleague Satisfaction

and LPC Self Description; .64, Satisfaction with Compensa-

tion andLPC Difference Score; .96, Satisfaction with Com-

pensation and LPC Self Description; .52, Overall Job Satis-

faction and Task LPC; and .42, Overall Job Satisfaction and

LPC Difference Score.

Table 16 contains the rank-order correlation values

where the measure of leader-member relations is a single-

item assessment given by team subordinates. In this partic-

ular analysis, there were nine median valued teams which were

eliminated in the process of dividing the teams into octants.

Perhaps as a consequence of this, the resulting rank-order

correlations are stronger than in the other analyses in which

far fewer teams were eliminated. Once, again, the results

do not coincide with Fiedler's predictions; LPC and Team

Performance correlated -.52 and -.80 for Octants III and VII,

respectively (-.33 and .05, predicated).

A large number of significant rho-values were computed

in the analysis where leader-member relations was measured

by using the team subordinate assessment. Eleven correla-

tions of .71 or greater may be noted in Table 16. It may be

of particular interest that the three strongest correlations

with Team Performance (LPC Coworker -.80, Social LPC -.90,

and LPC Difference Score -.78) are all negative values.

This is an indication that leaders with low LPC scores, pre-

sumably task-oriented, are more successful than high LPC

leaders [Fiedler and Chemers, 1974, p. 78]. Similar results

may also be noted on the other performance measure, Leader


Table 16


Rank-Order Correlation Coefficients between LFC and Criterion 5:ores

Octant III (Leader-Member Relations: Good) n=6

LPC Task Social LPC LF
Coworker LPC LPC Difference Self

Team Performance -.52 -.39 -.52 -.46 -.03

Leader Effectiveness -.56 -.56 -.56 -.37 .15

Discrepancy Measure -.49 .03 -.49 -.77' -.37

Colleague Sat. .10 .29 .10 -.10 .49

Compensation Sat. -.81" -.67 -.81** -.61 --32

Overall Job Sat. -.41 -.58 -.41 .03 -.70

Octant VII (Leader-Member Relations: Moderately Poor) n=-

LPC Task Social LPC LF:
Coworker LPC LPC Difference Self

Team Performance -.80'' .30 -.90* -.78** .50

Leader Effectiveness -.85" .23 -.81" -.58 -50

Discrepancy Measure .65 -.52 .58 .77** -.L5

Colleague Sat. .36 .24 .02 -.07 -.32

Compensation Sat. -.16 .22 .00 -.54 .71'

Overall Job Sat. -.61 .16 -.47 -.63 .77*

denotes probability of randomness less than .10

** denotes probability of randomness less than .05

Leader-Member Relations scores are a group average of a single iter,
global assessment from the Audit Staff Member Opinion Survey (App. C,
Chapter III, item 22). Octant III Teams: Nos. 6,12,16,17,19,22. c-
tant VII Teams: Nos. 7,8,11,13,14,18,21. Deleted Teams: 1,2,3,4,:,

Effectiveness, which also shows strong negative correlations

with LPC Coworker, Social LPC, and LPC Difference Score.

There were numerous large differences between the Octant

III and Octant VII rho-values. Once again, the Discrepancy

Measure appears to show the greatest amount of diversity when

shifting between octants. There were differences of 1.14,

LCP coworker;.55, Task LPC; 1.17, Social LPC; and 1.54, LPC

Difference Score. Team Performance, Leader Effectiveness,

and Overall Job Satisfaction showed differences when shift-

ing between octants of .69, .79, and .74 for correlations

with Task LPC. Team Performance, Colleague Satisfaction, and

Overall Job Satisfaction when correlated with LPC Self Des-

cription differed .53, .51, and 1.47 between octants. There

was a spread of .66 when comparing the rho-values for Overall

Job Satisfaction and the LPC Difference Score.

The above results were not unexpected. Since a large

number of median values were eliminated, differences should

have been more marked for a shift in a major element of sit-

uational favorableness.

Table 17 contains the rank-order correlations between

the LPC predictor variables and the criterion measures where

the measure of leader-member relations is the leader's global

assessment. The Team Performance correlation for Octant VII

comes closest to the Fiedler [1967, pp. 137-8] predication,

-.09 in lieu of .05. Team Performance and LPC Coworker cor-

relate -.64 in Octant III, compared to the predicted value

of -.33


Table 17


Rank-Order Correlation Coefficients betweer-lpCand Criterion Scores

Octant III (Leader-Member Relations: Good) n=ll

LPC Task Social LPC LPC
Coworker LPC LPC Difference Self

am Performance -.64" -.34 -.53' -.29

ader Effectiveness -.29 -.75** -.40 .29

screpancy Measure -.21 .02 .27 -.18 -.

league Sat. .01 .31 -.19 -.34 .

mpensation Sat. -.28 .13 -.12 -35 .

erall Job Sat. -.64** -.07 -.40 -.40 -.




Octant VII (Leader-Ne-ber Relations: Moderately Poor) .1=

LPC Task Social LPC LPC
Coworker LPC LPC Difference Self

Team Performance -.09 .53* -.19 -.47 .29

Leader Effectiveness -.14 .45 -.09 -.36 .57'

-Discrepancy Measure .16 -.70** .33 .64*1 -.0

Colleague Sat. .51 -.08 .51 .35 .25

Compensation Sat. .38 -.01 .54' .17 .12

Overall Job Sat. .65** .32 .48 .13 .-1

denotes probability of randomness less than .10

** denotes probability of randomness less than .05

Leader-Member Relations scores are a single-item global assessment
given by the group leader (instruction page, Ape. E, Chapter III).
Octant III Teams: Nos. 1,2,5,12,15,16,17,18,20,21,22. Octant VII Ie-Z:
Nos. 3,4,6,7,8,9,10,11,13,14 ,15.







Significant correlations appear between LPC Coworker

and Overall Job Satisfaction; -.64, Octant III and .65, Oc-

tant VII. From this, it seems task-oriented leaders (low

LPC) have teams with higher satisfaction when leader-member

relations are perceived as good by the leader and high LPC

(relations-oriented) leaders have teams with higher satisfac-

tion when the leader perceives leader-member relations as

moderately poor. Fiedler and Chemers [1974, p. 78] hypoth-

esize that a low LPC leader in a favorable situation will be

relaxed, friendly, and considerate; whereas, the high LPC

leader in an unfavorable situation will seek a close relation-

ship with his group members. The negative correlation for

Octant III, a favorable situation, could result from the sat-

isfying effect of human-relations activities coming from the

low LPC (task-oriented) leader. The positive correlation in

Octant VII is the result of high LPC (relations-oriented)

leaders resorting to their primary style in this unfavorable


Table 17 reveals many examples of extreme shift in cor-

relation between octants for similar sets of variables. How-

ever, as in Tables 15 andl6, in most cases both correlations

are not statistically significant, limiting the amount which

one may conclude from the differences.

The division of teams into Octants III and VII for all

of the analyses up to this juncture has been based upon per-

ceptions or assessments by team subordinates (Tables 15 and

16) or team leaders (Table 17). The final analyses used an

average of team leader assessments and subordinate average

assessments. Table 18 reveals that some of the results are

similar. The negative to positive satisfaction shift for

LPC Coworker as conditions become less favorable continues.

Strong correlations continue between the Discrepancy Measure

and the LPC Difference Score in Octant VII. Octant III cor-

relations between LPC Self Description and Team Performance,

Leader Effectiveness, and Overall Job Satisfaction surfaced

where none had been statistically significant before.

When comparingTables 15, 16, 17, and 18 one might be

concerned that there are not more similarities. However, if

one carefully compares the assignment of teams to octants,

as noted at the bottom of each table it is apparent that

only Teams 16 and 17 are consistently classified in Octant

III and only Teams 11, 13, and 14 are consistently classified

in Octant VII. The difference measures of leader-member re-

lations give quite different rankings. None of the measures

matches the methods suggested by Fiedler and Chemers dis-

cussed earlier. Perhaps this accounts for the disparity of

the results in the study with their median rho-values. How-

ever, it should be noted that Fiedler's median results fell

within rather large ranges [Fiedler, 1967, pp. 137-8] and

encompassed only two studies.


Table 18

Rank-Order Correlation Coefficients between LPC and Criterion Scores

Octant III (Leader-Memoer Relations: Good) n=9

LPC Task Social LPC LPC
Coworker LPC LPC Difference Self
Team Performance -.64' -.46 -.55 -.27 -.75"

Leader Effectiveness -.17 -.87** -.35 .52 -.58'

Discrepancy Measure -.18 .17 .12 -.46 -33

Colleague Sat. -.06 .19 -.06 -.13 .23

Compensation Sat. -.28 .12 .17 -.19 -.20

Overall Job Sat. -.56 -.07 -.32 -.46 -.831*

Octant VII (Leader-.ember Relations: ".odertely Poor) n=

LPC Task Social LPC LPC
Coworker LPC LPC Difference Self
Team Performance -.23 .60* -.50 -.60' .14

Leader Effectiveness -.24 .52 -.37 -.45 .49

*Discrepancy Measure .22 -.75'* .40 .80** -.50

Colleague Sat. .58* -.09 .44 .36 .11

Compensation Sat. .34 -.02 .56 .21 -.03

Overall Job Sat. .65** .37 -55 .03 .47

denotes probability of rando-ness less than .13

** denies prcbfSbliZ t of randc ness less than .05

Leader-Member Relations scores are a simple average of the leader's
global assessmert (insnru-tion page, App. E, Chapter III) and the
average global assessment by tea m members (item 22, ApP. C, Chapte-
III). Octant III Teams: Nos. 1,2,5,12,15,16,170,0,22. Octant VII
Teams: Nos. 3,4,7,8,9,,1 ,11,13,4. Deleted Tean-s: 6,1 ,19,21.



Ideal Leader Behavior Correlational Study

Computations were made to determine whether any possible

relationships existed between ideal (expected) leader behav-

ior and the assessments of satisfaction and performance and

the Discrepancy Measure--a measure of job dissatisfaction.

Neither the Fox Normative Model nor the Ohio State Leader-

ship Studies Model suggest any correlation between ideal

leader behavior and criterion measures; however, the search

seemed appropriate in this study due to the availability of

the necessary data.

Table 19 reveals four possible significant relationships

between Fox Normative Model predictor variables and criter-

ion scores. Expectations of Leader Support correlate posi-

tively with the Discrepancy measure. This may be an indica-

tion that the greater the short-fall in leader behavior, the

greater the subordinate's need for supportive activities in

the future. The negative correlation between expected Man-

agement of Rewards and Punishment and Leader-Member Relations

may indicate that these activities, in the CPA firm environ-

ment, may detract from good supervisor-subordinate relations.

Difficult to explain is the set of correlations with expected

Consultative-Participative Decision-Making. The latter cor-

related positively with the Discrepancy Measure, a measure


Table 19


Relationship between Predictors and Criterion Scores

Predictor Variable (Ideal Behavior)
L.S. D-M W.F. D G.E. M.R.P.

O -.05

P .21



.12 -.11

.22 .22

Q -.16 .05 -.08

T -.08

45** .00

Y -.28 -.07 -.20

Z -.24 -.09 -.07

Q31 .39*

.41* .31

.25 -.02 -.16

.16 -.17 -.03,

.13 -.10 -.38*



.09 -.11 -.05



.03 -.06



* denotes probability of randomness less than .10

** denotes probability of randomness less than .05


Leader Support
Consultative-Participative Decision-Making
Work Facilitation
Goal Emphasis
Management of Rewards and Punishment
Satisfaction with Colleagues
Satisfaction with ComDensation
Leader-Member Relations (subordinate assessment)
Overall Job Satisfaction
Team Performance
Leader Effectiveness
Discrepancy Measure (Ideal Behavior Minus Actual



of job dissatisfaction, and positively with Overall Job Sat-

isfaction. Possibly this indicates that dissatisfaction as

measured in this study belongs on a continuum which is dis-

tinct from the overall satisfaction continuum.

The test for correlations of ideal behavior predictor

variables with the set of criterion measures using the Ohio

State Leadership Studies Model (Table 20) showed a strong

correlation between expected structuring behavior and dis-

satisfaction, as represented by the Discrepancy Measure.

This could be interpreted as an indication that groups having

higher expectations of emphasis on structuring activities on

the part of their leaders were more dissatisfied. The nega-

tive correlation between expected Initiating Structure and

Team Performance may indicate that groups that prefer higher

standards for structuring activities on the part of their

leaders tended to perform less effectively than others.

Vroom and Yetton Normative Model of Leadership Styles

Based upon item no. 24 in the Audit Staff Member Opin-

ion Survey, which was completed by each team subordinate

(App. C, Chapter III), it was possible to classify the deci-

sion method used by each team leader. The average response

for the team subordinates was computed. A score of 4.5 or

above was classified as style Al; 3.5 to 4.4, All; 2.5 to

3.4, Cl; 1.5 to 2.4; Gi; ans 1.4 or less, D1 (refer to p.

46, Chapter III, for descriptions of decision methods in

the Vroom and Yetton model).

Of the 22 team leaders, 8 used decision method C1 and

Table 20


Relationship between Predictors and Criterion Scores

Predictor Variable
(Ideal Behavior)

C I.S.

0 -.01


Q -.05

Criterion Measure




* denotes probability of randomness

** deontes probability of randomness

less than

less than .05


Initiating Structure
Satisfaction with Colleagues
Satisfaction with Compensation
Leader-Member Relations
Overall Job Satisfaction
Team Performance
Leader Effectiveness
Discrepancy Measure (Ideal Behavior Minus Actual














13 used decision method Dl. Groups using decision method Cl

showed a higher average for both Team Performance (4.125 to

3.750) and Overall Job Satisfaction (3.750 to 3.622). The

feasible solution for the typical auditing problem presented

in item no. 24 suggested by Vroom and Yetton [1973, p. 194]

would be decision method Cl.

The mean value reported by teams on the decision method

question was 2.33. The range of possible scores was from

1.00 (completely delegative style) to 5.00 (completely author-

itarian style). The median style for this range would be

3.00, which compares to the actual median of 2.33, indicating

a perceived bias in the direction of delegative styles. Cor-

relation coefficients (Pearsonian) were computed between

mean team response to the decision method item and the satis-

faction and performance criterion measures used in other parts

of this study. The only correlation of any size was with the

Discrepancy Measure, .37 (probability of randomness .09).

While the correlation is small, it does indicate a possible

line between delegation of decision making and dissatisfaction.

This result goes against the thrust of one of the major as-

sumptions inherent to the traditional human relations school

of thought in industrial relations. However, numerous stud-

ies have indicated that participatory leadership methods may

not be effective in terms of performance and satisfaction of

team members IJacobs, 1970, pp. 76-84].

Item Analyses of the LBDQ Results

Pearsonian correlation coefficients were computed

between each item of the Leader Behavior Description Ques-

tionnaire for an Independent Auditor in Charge of Field Work

(Appendix B) and five of the criterion measures (Satisfaction

with Colleagues, Satisfaction with Compensation, Leader-

Member Relations, Overall Job Satisfaction, and the Dis-

crepancy Measure). Correlations were computed on an indi-

vidual (n = 69) and group (n = 22) basis. Additionally, cor-

relation coefficients were computed for each of the 52 LBDQ

items and Team Performance and Leader Effectiveness on a group


Only six items when taken on an individual basis gave

correlation coefficients of greater than .50; and, all cor-

related with the Discrepancy Measure. Table 21 lists the

items and correlation coefficients. After reviewing the con-

tent of these items, it is apparent that they should corre-

late negatively with dissatisfaction. Leaders who perform all

of these functions well may be eliminating serious potential

sources of dissatisfaction. It is interesting to note that

no variable other than Work Facilitation includes more than

one of these items. Most of the predictor variables used in

this study are made up of at least eight items (Table 2,

Chapter III). Work Facilitation includes eleven items.

Consequently, it is safe to conclude that the subset of items

selected in the item analysis of individual responses does

not approximate any of the predictor variables used. This

Table 21
















Item (See Appendix B, Chapter III)

25. Gave advance notice of changes in the audit
where possible

35. Inspired enthusiasm among subordinates.

36. Took full charge when emergencies arose.

45. Scheduled the work to be done.

46. Skillfully coordinated the necessary inter-
action of subordinates with the client's
employees and with each other.

50. Gave the appropriate amount of direction to

Member of Variable Item Subset

Consideration, Consultative-Participative

Goal Emphasis


Initiating Structure, Work Facilitation

Work Facilitation

Work Facilitation

should not be taken to imply that the items correlating

highest on an individual basis should approximate one or more

of the predictor variables; because the analysis by item ig-

nores the multicollinearity inherent in a subset.

Group averages of the 52 item scores from the LBDQ were

also analyzed for correlation with the criterion measures

used in the study. Six correlation coefficients with values

over .50 were found and are listed on Table 22. Four of the

items which correlated strongly in the analysis of individual

responses (items 25, 35, 36, and 50). The criterion measure

which correlated in all cases was the Discrepancy Measure.

It is logical that where an item showed a high correlation

on an individual basis, it may also show a high correlation

when the individual responses in each group are averaged.

Table 22 shows the correlation coefficients for the

group average responses to the LBDQ items. Two items were

correlated with Team Performance. It seems appropriate that

Item 2 (Anticipated problems and planned for them) would be

related to Team Performance, but the positive correlation of

Item 24 (Failed to take necessary action) with Team Perform-

ance is puzzling. This item was supposed to be reverse-

scored. Failure to reverse the scoring.would account for a

correlation coefficient sign reversal; however, a review of

the computer instructions and independent tests of computa-

tions showed that all items that were to be reverse-scored

had, in fact, been so treated. It must be reported, then,

Table 22


lation Criterion*

.52 Y


















Item (See Appendix B, Chapter III)

2. Anticipated problems and planned for

24. Failed to take necessary action.

25. Gave advance notice of changes in
the audit where possible.

35. Inspired enthusiasm among subord-

36. Took full charge when emerengencies

50. Gave the appropriate amount of di-
rection to subordinates.

* Y = Team Performance; Q31 = Discrepancy Measure

Member of Variable Item Subset

Work Facilitation

Work Facilitation, Decisiveness

Consideration, Consultative-Participative Decision-

Goal Emphasis


Work Facilitation

University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - Version 2.9.9 - mvs