Group Title: impact of the frequency of accounting-based performance reports on capital budgeting decisions
Title: The impact of the frequency of accounting-based performance reports on capital budgeting decisions
CITATION THUMBNAILS PAGE IMAGE ZOOMABLE
Full Citation
STANDARD VIEW MARC VIEW
Permanent Link: http://ufdc.ufl.edu/UF00102726/00001
 Material Information
Title: The impact of the frequency of accounting-based performance reports on capital budgeting decisions
Physical Description: Book
Language: English
Creator: Kite, Devaun Marie, 1959-
Copyright Date: 1992
 Record Information
Bibliographic ID: UF00102726
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 - 27779177
ltuf - AJM7682

Full Text










THE IMPACT OF THE FREQUENCY OF ACCOUNTING-BASED
PERFORMANCE REPORTS ON
CAPITAL BUDGETING DECISIONS













BY


DEVAUN MARIE KITE


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY


UNIVERSITY OF FLORIDA


1992














I dedicate this dissertation to my parents, Hayman and Joan Kite.














ACKNOWLEDGEMENTS


Having now experienced firsthand the competing demands upon my time in an

academic environment, I deeply appreciate the time my committee spent working with me.

I am grateful to my chairman. Dr. Doug Snowball, for his leadership, guidance,

and calming influence. His optimism and willingness to see me anytime I dropped by were

both instrumental in my completing the dissertation.

I am also grateful to Dr. Bipin Ajinkya, Dr. William Messier, and Dr. Stephan

Motowidlo for their direction in both the development of the case materials and the writing

of the dissertation. Their comments and suggestions taught me lessons that will continue to

help me through my research career.

I would also like to acknowledge the individuals who made the experiment

possible. First. I am grateful to the professors of the Fisher School of Accounting who

allowed me to interrupt their classes to recruit subjects. Most especially, my thanks go to

Ron Rasch for donating his classes for the experiment and to Susan Crosson for offering

extra credit to her class in order to boost the power in the nick of time. Thanks go also to

the Fisher School of Accounting administrative staff for preparing the experimental

materials. Also, the administrative assistance of Tammy Bowley, Tom Bristow, Sherry

Ropp, and Delta Sigma Pi was invaluable. Finally, Delta Sigma Pi's participation in the

many pilot studies proved extremely useful in developing the experimental materials.














TABLE OF CONTENTS


ACKNOWLEDGEMENTS..................................................................... iii
A BSTRA CT ................................................................................... ... vii

CHAPTERS

1 INTRODUCTION AND BACKGROUND........................................ 1

Introduction ............................................................................... 1
Purpose of the Research............................................................... 2
Motivation for the Research............................................................ 2
Overview of Research Method..................................... ............. 5
Organization of Remaining Chapters............................... ............. 6

2 REVIEW OF LITERATURE............................................................ 7

Introduction ............................................................................. 7
Performance Evaluation Literature ................................................... 7
Contingency Theory: Implications for the Design of ABPM............ 7
Behavioral Consequences of ABPM....................................... 8
Dysfunctional Consequences of ABPM ................................... 10
Self-presentation Literature............................................................. 13
Sunk Cost Literature ................................................................. .. 15
Project Determinants ......................................................... 16
Psychological Determinants................................................... 17
Social Determinants ............................................................ 17
Organizational Determinants ................................................ 18
The Present Research ................................................................... 19

3 MODEL DEVELOPMENT AND RESEARCH HYPOTHESES ................. 20

Introduction ........................................................................................ 20
The M odel............................................. ............. ............... 20
Frequency of Performance Evaluation.................................... .. 21
Perceived Benefits of an Action ............................................... 24
Possible Interaction Effects ................................................... 27
Action: Switching or Escalating ............................................... 30
Summary .......................................................... .......... 30








4 RESEARCH DESIGN AND DATA ANALYSIS METHODS .................... 32

Introduction .................................... ........................ ........... ...32
Experimental Method ................................................................... 32
G general Setting ................................................................. 32
Experimental Materials.......................................................... 33
Experimental Procedure ...................................................... 37
Subjects.................................................................... ... 39
Variables.................................................................... 41
Operationalization of Independent Variable ................................. 41
Measurement of Dependent Variables..................................... 42
Measurement of Moderating Variables......................................... 44
Data Analysis Methods .................................................................45
Data Screening.................................................................45
Missing Values ................................................................. 45
Manipulation Checks.......................................................... 45
Nuisance Variables............................................................ 46
Scales ..................................................... ............. ....... 46
Test of Hypotheses ....................................................................46
Personal Benefits from Escalating H 1....................................... 46
Ability to Manipulate One's Image H2....................................... 47
Doing the Job Right H3...................................................... 48
Periods Until Switching H4.................................. ........... 48
Summary ..................................................................... .....50

5 RESEARCH RESULTS.................................................................51

Introduction ...................................................................... ....51
Data Screening................................................................. .....51
Manipulation Checks............................................ ................. 52
Job Insecurity................................................................. 52
Policy Resistance............................................................... 52
Experimental Administration................................................... 54
Frequency of Performance Evaluation...................................... 55
Incentive system................................................................ 57
Nuisance Variables...................................................................... 58
Missing Values ........................... .................... ............... ........ 58
Scales ............................................................ .. ...... 58
Between-scale Convergent Validity ........................................ 60
Reliability-fixed-list Scales.................................................... 60
Reliability-aggression Questionnaire......................................... 65
Reliability-efficacy Questionnaire ............................................ 65
Tests of Hypotheses .................................................................... 65
Hypothesis 1 .................................................................. 67
H ypothesis 2................................................................. 78
Hypothesis 3 ......................... .................... ............. ........ .. 79
Hypothesis 4a................................................................. 88
H ypothesis 4b ................................. ................................. 90
Exploratory Analysis of Hypothesis 1....................................... 90
Sum m ary .................................... ....... ....... ........... ... .... .. ... 92








6 SUMMARY AND CONCLUSIONS ...................................................93

Summary and Discussion of the Results ........................................... 93
Frequency of Performance Evaluation................................. 93
Ability to Manipulate One's Image .......................................... 94
Doing the Job Right.......................................................... 94
Periods Until Switching and Benefits from Escalating .................... 95
Periods Until Switching and Benefits from Switching .................... 95
Limitations ............................................................................. 95
Implications and Conclusions.................................... ............... 96
Conclusions .................................................................... 97
Implications for the Model .................................................... 98
Implications for Future Research.............................................. 98
Implications for Policy Makers................................................ 99

APPENDIX EXPERIMENTAL MATERIALS ...................................... 101

Scenario .............................................. ................ ................ .. 102
Instructions ............................ ................................................... 103
Rules for Earning Tickets .............................................................. 104
Period 1 Experimental Materials........................................................ 105
Period 2 Experimental Materials....................................................... 111
Period 3 Experimental Materials...................................................... 123
Period 4 Experimental Materials...................................................... 126
Post-experimental Questionnaire ...................................................... 127
Aggression Questionnaire .............................................................. 130
General Self-efficacy Questionnaire.................................... ............ 131

REFERENCE LIST.......................................................................... 133

BIOGRAPHICAL SKETCH ................................................................... 140























vi














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

THE IMPACT OF THE FREQUENCY OF ACCOUNTING-BASED PERFORMANCE
REPORTS ON
CAPITAL BUDGETING DECISIONS

By

Devaun Marie Kite

August 1992

Chairperson: Doug Snowball
Major Department: Accounting

Performance evaluation research has identified dysfunctional consequences of using

accounting-based performance measures. Some critics have attributed American

corporations' lack of competitiveness to the short-term emphasis in financial statements.

The purpose of this study was to examine the effect of the frequency of performance

evaluation, using accounting-based performance measures, on the capital investment

decision process.

Based on self-presentation theory and the sunk cost phenomenon, hypotheses were

developed concerning the effect of the frequency of performance evaluation upon an

individual's self-presentation motivation. The major hypothesis was that the frequency of

performance evaluation would influence the perceived personal benefits associated with

escalating an investment commitment. Furthermore, the perception of these benefits, as

well as the perception of benefits to the organization from escalating and benefits from

switching to the alternative investment, were predicted to affect the capital investment

decision positively.

An experimental study was conducted to investigate the differential effects of long-

term and short-term performance evaluation. To disguise the true purpose of the study, the









capital budgeting decisions were made in the context of a business simulation, using an in-

basket scenario. The study was conducted over four periods during which subjects made

capital budgeting choices for a hypothetical company. The treatments differed only as to

when their performance was evaluated (short-term or long-term). Using self-reports as

data, the perceptions underlying the capital budgeting decisions of each subject were

determined. The length of commitment to the original investment decision was also

measured.

The results provided weak support for the effect of the frequency of performance

evaluation on the individuals' perceptions of the personal benefits from escalating

commitment to the original investment. These results were stronger when the extraneous

variability associated with risk aversion or ethical considerations was removed.


viii















CHAPTER 1
INTRODUCTION AND BACKGROUND


Introduction


Throughout its development, what accounting has measured has channeled the

efforts of the participants whose behavior was being measured. For example, early cost

accounting focused on conversion costs, and this caused factory owners to concentrate on

cost reduction through new forms of power and mechanization. Since accounting

measures can direct organizational activities, accounting research should consider the way

"in which a particular account can shape, mould, and even play a role in constructing the

setting of which it forms a part" [Hopwood 1983, p. 288].

Research on the use of accounting-based measures in performance evaluation has

examined contextual, economic, and psychological variables that are important in

explaining attitudes and performance. One stream of research has uncovered some of the

dysfunctional consequences of using accounting-based performance measures (ABPM).

These consequences, such as rigid bureaucratic behavior and invalid data reporting, occur

when managers act to improve their performance scores for specific accounting indices.

For example, in a field study observing the behavior of officials in two government

agencies, Blau [1955] found that employees acted in ways to affect the performance

measure and not the performance itself. Magee and Dickhaut [1978] reported that the

choice of compensation plans affected behavior. It appears that ABPM have powerful

repercussions within an organization; therefore, managers should be careful to select the

accounting-based performance measures that motivate behavior consistent with

organizational objectives.









Purpose of the Research


The purpose of this research is to examine the effect of the frequency of

performance evaluation, using accounting-based performance measures, on the capital

investment decision process. The model presented in the study posits a direct effect of the

frequency of ABPM on the cognitive processes underlying the capital budgeting decision.

In addition, an indirect effect of personality variables on these processes is predicted.

Environmental, organizational, individual and group variables influence both the

method of performance measurement and the effectiveness of the measurement. For

example, environmental variables such as task environment and societal values affect the

use and usefulness of accounting-based performance measures. In Japan, where managers

often are committed to their companies for life, the ABPM used affects behavior differently

than in the United States, where employees' tenure with firms may be relatively shorter.

The current research holds the effect of environmental, organizational, individual and group

variables constant, and investigates the effect of different types of performance evaluation

on behavior.

The study poses three major research questions. Does the frequency of

performance evaluation affect an individual's perception of the benefits of a decision? Is

the relationship between the frequency of performance evaluation and the perceived benefits

moderated by variables such as the perceived ability to manipulate one's image to others?

Do the perceived benefits weighted most heavily by subjects explain the actions taken?

Motivation for the Research


Performance evaluation research has identified dysfunctional consequences of using

ABPM. However, the influence of some fundamental aspects of ABPM has not received

the express attention of those interested in explaining individual motivation and behavior in

an organizational context. According to the framework developed by Smith [1976],










performance measures may vary with respect to their specificity, to their closeness to

organizational goals, and to the time span covered. Accounting research has indirectly

investigated both the specificity and the closeness to organizational goals through research

using the contingency paradigm and through research on the dysfunctional effects of

specific ABPM. Research investigating the effects of time span has only been performed at

the organizational level. Time span effects on individual behavior have been given little

research attention.

The lack of research on the effect of the time span of ABPM on individual behavior

is puzzling, since the ability of ABPM to influence behavior has been established by prior

research [Magee and Dickhaut 1978; Chow 1983]. Performance evaluation is used for

purposes such as compensation, counseling, training and development, promotion,

manpower planning, retention or discharge, and validation of a selection technique [Eichel

and Bender 1984]. Regardless of the intended purpose, however, it is inevitable that the

method chosen will have motivational implications. The existence of a relationship between

performance evaluation and behavior should cause superiors to consider more carefully the

ABPM chosen to evaluate subordinates.

This study attempts to expand the limited body of literature in this area by focusing

on one basic dimension of evaluation: the frequency of evaluation. It investigates whether

the frequency of performance evaluation mitigates one possible dysfunctional consequence

of short-term performance evaluation known as management myopia. Management myopia

occurs when managers who are responsible for short-term earnings become excessively

short-term oriented due to the lack of balancing long-term incentives [Merchant 1989].1

11t is possible that other short-term measures within an organization, such as analyst
forecasts or external reports, will also cause investment myopia. This study will ignore
these influences in order to determine whether one component of an organization's
information system, performance evaluation, does influence management myopia. If this
can be established, it will provide a basis upon which it may be determined whether other
information system characteristics also influence management myopia.










Investment myopia is a type of management myopia which occurs when a business unit

must invest for the future but chooses short-term investment strategies because it is

evaluated using short-term measures of performance. Therefore, the manager must create

and maintain the impression that will result in the best short-term image. These

consequences presumably diverge from the behavior desired by the owners.2 The sunk

cost effect, whereby an individual escalates commitment to an inferior course of action,

may be an example of investment myopia. The sunk cost effect stems from what Staw

[1980] called "retrospective rationality," i.e., considering previous decisions so as to

appear rational. He called the normative, economically rational approach "prospective

rationality." The prospectively rational approach ignores sunk costs, processes information

regarding future outcomes and chooses the alternative which will attain the highest

outcome.

Since the sunk cost effect may be an example of investment myopia and because the

act of accounting can influence behavior, it is possible that the choice of ABPM could be

used to mitigate a manager's tendency to escalate commitment to a faulty course of action.

Therefore, the effect of varying the frequency of performance evaluation on the tendency to

escalate commitment is examined. As Ijiri et al. [1970] argued, unless one can

demonstrate that different accounting methods in a given context result in different


2Major companies attempt to alleviate this goal incongruence by compensating executives
with a combination of compensation packages. For example, at IBM the components of an
executive's compensation consist of salary, variable compensation (incentive plans and
award plans) and long-term performance programs (e.g., stock option programs) [Edman
1990]. Some companies [e.g., Edman 1990] have attempted to align managers' interests
with those of the business by offering long-term incentives such as stock options. This
approach is also used by corporations on a broad level for senior managers, thereby tying
rewards to long-term performance accounting measures such as a five-year trend in EPS
[Merchant 1989]. However, there are two problems with the present approaches. First,
there is a question as to the point at which this type of ownership outweighs other personal
motivations to maximize utility. Second, the performance evaluation/rewards of lower
levels of management are usually not tied directly to project outcomes. Rather, Merchant
pointed out, long-term written promises are tied to multi-year profit of the profit center,
corporation or organization. Therefore, restructuring the application of ABPM (e.g., time
frame of evaluation) is a more direct approach than offering long-term incentives.










decisions, then there is no point in arguing the relative merits of accounting methods.

Therefore, in a common setting, the cognitive processes and decisions associated with

long-term performance evaluation will be compared to the the cognitive processes and

decisions associated with short-term performance evaluation.

This study is the first to offer explicit hypotheses concerning the cognitive

processes underlying an individual's reaction to accounting-based performance measures.

Furthermore, this study will expand the current body of accounting literature by

investigating a previously unexplored relationship between the frequency of evaluation and

individual behavior. The availability of a descriptive model for the effects of long-

term/short-term performance evaluation could enable managerial accounting systems and

reports to be designed to minimize escalation errors and other dysfunctional behaviors

resulting from the use of accounting-based performance measures.

Overview of Research Method


The effect of the timing of ABPM on capital budgeting decisions was examined in

an experimental study that used 109 student subjects. The experimental task encompassed

four periods during which subjects made capital budgeting decisions for a hypothetical

company. To disguise the true purpose of the study, the capital budgeting decisions were

made in the context of a business simulation, using an in-basket scenario. One half of the

subjects received a performance evaluation every period (ST) while the other half received

one performance evaluation at the end of the experiment (LT). Written self-reports

captured the perceptions underlying the capital budgeting decisions of each subject. The

length of commitment to the original investment decision was also measured. Subjects

earned tickets based on their performance in the study. The tickets were entered into a

lottery which awarded seven cash prizes. This reward scheme was used to motivate

student performance during the experiment.












Organization of Remaining Chapters


Chapter 2 discusses prior research relevant to the research questions. The model of

the differential effect of the frequency of performance evaluation on the cognitive processes

underlying capital budgeting decisions is presented in Chapter 3, along with the research

hypotheses. Chapter 4 describes the research design used to test the hypotheses. The

results of the experiment are described in Chapter 5. A summary of the study and

conclusions are presented in Chapter 6.














CHAPTER 2
REVIEW OF LITERATURE


Introduction


Three bodies of literature are relevant to this study: performance evaluation [e.g.,

Hopwood 1972, 1974; Hayes 1977], self-presentation [Schlenker 1980], and sunk costs

[e.g., Staw 1980]. The performance evaluation literature is reviewed to provide evidence

of the contingent nature of the effects of performance evaluation, the possible dysfunctional

consequences of using ABPM, and the possible effects of varying the time dimension of

ABPM. Since it is posited that individuals' self-presentation motivation will be affected by

the frequency of accounting-based performance measures, the self-presentation literature is

reviewed next. Finally, since individuals may consider sunk costs in their decisions

because they are concerned about the impression they present to others, the sunk cost

literature is reviewed in the final part of this section.

Performance Evaluation Literature


Performance evaluation may focus either on the process of performing or on the

results of the performance. Most evaluation instruments that measure the process of

performing utilize character traits. On the other hand, those evaluation instruments that

measure performance outcomes generally employ cost-related variables [see Latham and

Wexley 1981]. The studies reviewed below are concerned with the latter approach.


Contingency Theory: Implications for the Design of ABPM


Contingency theories have been applied to many organizational phenomena.

Performance evaluation is no exception. Hayes [1977] provided a contingency framework









for analyzing the effects of performance evaluation. He hypothesized that the effectiveness

of various measures of organizational performance is dependent upon internal,

interdependency, and environmental factors (i.e., that there was no one "best" method of

performance evaluation). Internal factors are those defined within the organizational

subunit such as nature of the tasks, types of people and the ability to measure and quantify

functions. Interdependency factors describe interactions with other subunits such as

reciprocal or sequential relationships. Finally, environmental factors refer to such factors

as market share, environmental stability, and environmental diversity. Hayes's results

supported the contingency framework: ABPM were not effective for all organizational

subunits.

A significant body of research on variables that might temper the outcome of

performance evaluation has used the contingency paradigm [Brownell 1982; Kida 1984;

McNamee 1988]. Contingency variables have been examined for their effects on

organizational performance, individual performance, and attitudes. The studies have

explored the effects of such contingency variables as evaluation style [Hopwood 1972;

Otley 1978], leadership style [Hopwood 1974], task uncertainty [Hirst 1981],

environmental uncertainty [Govindarajan 1984], business strategy, incentive bonus

systems, and strategic business unit effectiveness [Govindarajan and Gupta 1985].


Behavioral Consequences of ABPM


Research using the contingency paradigm has typically correlated behavior with

performance evaluation system characteristics. Rarely has the performance evaluation

system been treated as an independent variable with behavior as the dependent variable.

The limited research that did treat performance evaluation as the independent variable was

based upon the expectation that the choice of ABPM would affect behavior on the

dimension measured. These behavioral consequences have been explored at both

organizational and individual levels and are discussed next.









Behavioral consequences at the organizational level. Larcker [1983] evaluated the

changes in decisions associated with changes in performance evaluation. He posited that

the adoption of performance plans would be associated with increased corporate capital

investment.1 His prediction was based upon the idea that since performance plan

compensation is based upon a longer time period than the typical short-term plan, managers

would lengthen their decision-making horizon. This increased horizon should heighten the

attractiveness to managers of projects that exhibit later payoffs. Larcker's predictions were

supported. In his study of firms that adopted performance plans and had changes in

corporate capital investment, he found that corporate capital investment increased with the

adoption of performance plans.

Behavioral consequences at the individual level. In a field study of five industrial

companies, Hofstede [1968] found that 'tightness' of job standards affected motivation.

Motivation increased as standards became more difficult to achieve, but this levelled off.

When standards became too difficult to achieve, motivation began to decrease. Rockness

[1977] found similar results for performance and satisfaction. He used a laboratory setting

to study the effects of alternative budget levels, reward structures, and performance

feedback on subject performance and satisfaction. Higher budget levels resulted in higher

performance. Performance and satisfaction were increasing with the number of periods for

which the subjects were participating. Magee and Dickhaut [1978] also found evidence that

the type of performance evaluation affects behavior. They investigated the effect of

performance evaluation on the variance investigation behavior of individuals. They

assigned individuals to one of two compensation plans. Compensation plan one (CP1)

reduced individuals' compensation for the cost of investigation; compensation plan two


1Performance plans have the following characteristics: 1) they are long-term (3-6 years);
2) compensation is deferred until a future date and is forfeited if the employee leaves the
organization during the compensation period; 3) goals are explicitly stated in growth terms
(accounting-based measures) over the award period: and 4) the payoff is bounded by zero
from below and increases as performance exceeds the target [Larcker 1983].









(CP2) paid a fixed amount or zero, depending on whether the costs of investigation and of

operation were less than a predetermined standard. Magee and Dickhaut found that the

groups used different decision rules. Specifically, individuals in CP1 set the upper bound

for investigation higher than the mean departmental costs (which were distributed according

to a conditional distribution contingent upon the state of nature). On the other hand, those

in CP2 showed no preference for setting the upper bound equal to or greater than the mean.

Chow [1983] explored the effect of job standard tightness and compensation

scheme type on job performance. He found that tight standards resulted in higher

performance than average standards and that piece rate compensation resulted in higher

performance than fixed pay or budget-based pay.


Dysfunctional Consequences of ABPM

Agency theory recognizes that the utilities of managers and owners may be

divergent. Therefore, contracts must be created to motivate managers to perform as desired

by the owners. If the contracts do not provide this motivation, the managers' performance

may result in dysfunctional consequences. Ridgway [1956] pointed out that "even where

performance measures are instituted purely for purposes of information, they are probably

interpreted as definitions of the important aspects of that job or activity and hence have

important implications for the motivation of behavior" (p. 247). In other words, even if the

performance measures are not tied to compensation, the act of measurement itself can

motivate dysfunctional behavior. Thompson and Dalton [1970] stated that comprehensive

performance appraisal systems often have results directly opposite from those intended and

that managers need to consider the human consequences of any systems which they initiate.

In fact, Mintzberg [1975] pointed out that "organizational objectives are often rigid and

dysfunctional and encourage the manager to use inappropriate information" (p. 2).

A stream of literature has investigated the dysfunctional consequences of using

ABPM. This body of literature recognized that consequences inevitably accompany the use









of ABPM and that these may lead to behavior contrary to the desires of the owners of the

corporation. Babchuck and Goode [1951] described a situation where a company changed

sales performance measures in order to eliminate such dysfunctional consequences. The

old measures had encouraged individual antagonism and a parochial focus on sales. The

new measures increased group cooperation and motivated employees to perform on

nonsales duties which were essential to company success.

Other researchers also provided evidence of dysfunctional consequences of

performance evaluation. Granick [1954] studied Soviet management and reported that they

delayed repairs and maintenance to improve their individual performance records. These

employees behaved in order to improve their performance on production indices at the

expense of overall plant production. Later Blau [1955] found that employees' behavior

diverged from that desired by the business because of the way employees were evaluated.

Finally, Dearden [1960] discussed instances where the method of evaluating the investment

performance of decentralized divisions caused the division managers to make asset

replacement and retirement choices which were not in the best interests of the overall

company. The decisions improved the performance indices for the divisions but decreased

the overall profit of the organization.

The dysfunctional consequences of budgeting have been investigated by a number

of researchers. Studies have shown, for example, that individuals responded to budgeting

with data manipulation [Lowe and Shaw 1968; Hopwood 1972; Yetton 1976], game

playing [Collins, Munter and Finn 1987]. and slack building [Young 1985; Merchant 1985;

and Waller 1988].

Dysfunctional consequences of short-term performance measures. Thurow [1981]

blamed short-term performance measures for the difficulties encountered by American

corporations in international competition. He argued that the problem arises not because

American managers are stupid or impatient, but because business has created an

environment where it is rational to have a short time horizon. The choice of information









systems emphasizing short-term performance may be a major contributor to the inability of

American corporations to compete with foreign competitors. Consistent with this theory,

Merchant [1989] pointed out that managers follow "Gresham's Law of Planning," that is,

they have a tendency to let short-term concerns dominate. This creates a need to redirect

attention to long-term concerns (for example, through ABPM that emphasize long-term

performance).

Hayes and Garvin 1982] reported that choices of ABPM may bear some

responsibility for the changes in capital and R&D expenditures between 1948 and 1973.

When adjusted for inflation, changes in GNP, or work force size, these expenditures had

actually declined over the 25 years. Hayes and Garvin posited that this decrease may have

been due to a move toward multidivisional organizations, which primarily evaluate

managerial performance using short-term financial measures.

Indeed, it may be that the source of this myopic emphasis on short-term profitability

(quarterly and yearly results) is traceable to the accounting system. Accounting systems

have been based on traditional accounting measurements that require the life of the business

to be separated into artificial accounting "periods." And as critics of accounting have

remarked, managers have been rewarded for profit performance in the current period

[Merchant and Bruns. 19861, rather than over the long term.2 Elliot [1990] referred to this

as the U.S. financial accounting paradigm.

The above "accusations" that short-term performance measures have affected

decision making have some limited empirical support. The little empirical work that has

been done to determine how time influences the decision process has been in psychology.

2This is in contrast to Japan where long-term growth is pursued. Tsurumi and Tsurumi
[1985] point out that, "Few Japanese firms use quarterly or even annual earnings per share
or other similar financial indices as their performance guides" (p. 29). However, it is not
possible to infer that long-term measures caused Japan to be internationally competitive.
Their use of long-term measures may only reflect their long-term commitment to the
company or their focus on long-term goals. However, given that the United States uses
short-term performance measures, regardless of the reason why, it is interesting to
investigate whether these measures do indeed lead to dysfunctional behavior.









Nisan and Minkowich [1973] explored how variation of the distance between action

and outcome affects the salience of decision variables. They found that if the outcome was

temporally distant, subjects considered probabilities more salient than values. The reverse

held true for temporally close outcomes. Further, subjects behaved differently if they were

dealing with gains or losses that were temporally distant. Gains induced lower risk-taking

than losses.

In an attempt to develop a theory to explain patients' tenure under a treatment

program, Christensen-Szalanski and Northcraft [1985] found that the temporal distribution

of costs and benefits have a direct impact on behavior. They explored the use of contracts

used to bridge the gap between the short-term costs of medical treatment and the long-term

gains from the outcome of the treatment. The success of these contracts has direct

implications for the possibility of varying the temporal aspects of ABPM to influence

behavior.

The results of Stevenson's [1986] study also had design implications for ABPM.

She found that outcomes with positive expected value were more desirable the closer they

were to realization and those with negative expected value were preferred the further they

were from realization. This and other studies of time's impact on the decision process

[Loewenstein 1988; Benzion. Rapoport, and Yagil 1989] imply that it may be possible to

use the time span of ABPM to channel individuals' decision processes toward that desired

by owners.

In this study it is proposed that the frequency of accounting reports will impact the

"self-presentation" behavior of individuals. Accordingly, the self-presentation literature is

reviewed in the next section.

Self-presentation Literature


The effect of ABPM on behavior may be moderated by various psychological

determinants of behavior. The act of evaluating a manager's performance may cause the









manager to have evaluation apprehension [Rosenberg 1969] and to behave in ways that he

perceives will improve his evaluation. For example, Tan [1991] found that auditors who

anticipated a review behaved differently from those who did not. When anticipating a

review, auditors who had generated prior-year work papers paid more attention to audit

evidence inconsistent with their expectations and were less extreme in their judgments than

the auditors in a new audit situation. When no review was anticipated, there was no

difference between the groups. Prakash and Rappaport [1982] referred to this process,

where an information sender is influenced by the information required for communication,

as information inductance.

One potentially important theory for explaining the psychological processes that

may moderate the effect of ABPM on decision making is self-presentation theory (hereafter

referred to impression management). The essence of impression management, according to

Schlenker [1980], is that people attempt to establish, monitor and control their identities in

front of others so that they may control the outcomes that stem from interactions with

others.3 In organizations, desired outcomes such as bonuses are often contingent upon the

results of performance evaluations. Because the outcome of the performance evaluation is

important to the manager, the motivation exists to manage the image presented to the

evaluator. It is possible that the manager's perception as to which image will result in a

better outcome is contingent upon the type of evaluation used.

Schlenker [1980] posited that people behave in order to maximize expected rewards

and minimize expected punishments. If people expect that a certain impression will


3People may feel accountable to themselves, to people with whom they interact, and to
others who are prominent in their life [Schlenker, 1986]. This study is designed to focus
the subjects on accountability to others. According to Schlenker, "When people's attention
is focused on how the self appears to others, the expectations of those others become
salient. People are then more likely to conform in the face of real or imagined social
pressures and to play socially expected roles" (p. 73). In this study there is no attempt to
measure or increase accountability to the self. Because subjects are randomly assigned to
treatments (discussed in Chapter 4). it is assumed that individual differences in personality
variables that are likely to produce effects where the private self is most salient will be
randomly dispersed.









increase rewards, they will go to great lengths to ensure this particular impression has been

conveyed to others. This is not surprising since most desired outcomes are dependent not

only upon the person performing the behavior, but also upon others. Therefore, people

may influence the attainment of desired outcomes by influencing the impressions others

form of them.

As evidence of impression management, Jones et al. [1965] found that subjects

behaved differently, depending upon the personal values they ascribed to the person

responsible for their rewards. Subjects behaved in an independent manner when they

believed the superior valued productivity but in a conforming manner when they believed

that the superior valued solidarity and "going along."

More recent research provides evidence of impression management Schlenker et

al. [1983] conducted a laboratory experiment with undergraduate psychology students.

They found that the students adjusted the image they presented to others depending upon

what they had been told about the beliefs of the interviewers. The laboratory experiment by

Kardes and Kimble [1984] reported that subjects who anticipated future interaction with an

individual were more informative in good news than in bad news conditions. The news

described the other individual's performance on the "Social Sensitivity Test." It was good

news if the subject was told the individual scored high. The converse was true in the bad

news condition. Subjects who did not anticipate future interactions were more informative

in bad news conditions than in good news conditions.

Sunk Cost Literature


An individual's tendency to commit escalation errors, i.e.. to consider sunk costs in

decision making, may be driven by impression management. An individual who commits

these errors may be attempting to present a positive image to those to whom he or she is

responsible. This study considers impression management in a scenario where there is









pressure to commit such an error, and therefore the sunk cost literature is especially

relevant.

Much of the research investigating the sunk cost effect has been based on the

premise that individuals need to justify their actions [Brockner 1992]. This justification

represents an attempt to manage a positive image regarding past decisions. Staw [1980]

suggested that the more an individual is motivated to predict and control the environment,

the more likely that the individual will be susceptible to forces that are counter to rationality.

He proposed that the stronger these forces, the more the individual will feel the need to

justify his or her decisions. The justification may be internal, to protect one's self-concept,

or external, to protect one's image. Staw predicted that "The need to justify one's actions

increases as the irrationality of one's actions is exposed both to self and others" (p. 57).

Some of the determinants of this need to justify past decisions, according to Staw, are

personal responsibility for negative consequences (prior choice and foreseeability of

outcomes) and the organizational norms for rationality.

Staw and Ross [1987] proposed a model of commitment to a course of action.

They identified four categories of escalation determinants: project, psychological, social,

and structural (hereafter referred to as organizational). The following sections will review

some of the relevant research on these determinants of individuals' tendencies to escalate

commitment to a course of action.


Project Determinants


The characteristics of a project may have an impact on the tendency toward

escalation. These are the objective characteristics of a project (e.g., project goals). Staw

and Ross [1978] investigated whether an individual's investment allocation decision would

differ according to whether the causes of success or failure were endogenous (directly

related to the project itself, most likely to recur if the project continued, and relatively

foreseeable) or exogenous (not directly related to the project, least likely to recur if the









project continued, and not foreseeable). Staw and Ross found that student subjects who

received positive feedback following an original investment did not vary their second

commitment of resources, regardless of the cause of success (exogenous or endogenous).

Subjects in the negative feedback condition, however, did react differentially to the cause.

If it was exogenous endogenouss), they invested more (less) than in any other case. The

subjects were more aware of causal data in the case where they had a failure.


Psychological Determinants


Psychological determinants are factors that "influence one's goals and beliefs about

the consequences of an action" [Staw and Ross 1986, p. 55]. Arkes and Blumer [1985]

conducted a number of experiments using psychological determinants (such as the need for

self-justification) and found consistency with the sunk cost effect. For example, in a field

study, theater patrons who spent more for their season tickets attended significantly more

performances during the first half of the season than those who spent less. The decision

whether or not to attend a performance appeared to be influenced by the size of the season

ticket investment Arkes and Blumer did not find that personal involvement increased the

sunk cost effect or that learning economic principles lessened the effect. They concluded

that the sunk cost effect is a robust judgment error.


Social Determinants


Social determinants such as the need for external justification also affect

commitment to a course of action. Staw [1976] pioneered research regarding the effects of

social determinants on escalation behavior. He investigated the effects of personal

responsibility and decision consequences on the amount allocated to a previously chosen

investment alternative. The task was the allocation of R&D funds to one of two divisions

over two periods. Following the original allocation, the subjects received feedback (either

negative or positive) regarding the success of the first-period allocation. Subjects in the









high-responsibility condition made the allocation decision in both periods; those in the low-

responsibility condition were told the first-period allocation had been made by another

party. A significant interaction effect was found for personal responsibility and feedback.

The amount invested in the high-personal-responsibility, negative-feedback condition was

higher than in any other condition. Under low responsibility, the feedback did not have

any effect.

Staw and Fox [1977] replicated Staw's study and extended his work to consider the

the effects of the probability of investment of resources resulting in positive outcomes and

the effects of intertemporal decisions (three periods). All subjects received feedback that

their previous decision had resulted in negative consequences. Staw and Fox found that

over time the low probability and low-personal-responsibility subjects did not vary their

commitment. The high-responsibility and high-probability subjects, however, were

unstable. They invested most at time one, least at time two and an intermediate amount at

time three. Due to this instability and the lack of variability in commitment over time by

both low-responsibility and low-probability subjects, the researchers concluded that

escalation did not diminish over time with negative feedback.


Organizational Determinants


The structural features of an organization may influence escalation tendencies of

individuals. Fox and Staw [1979] investigated the effects of organizational determinants.

They manipulated job security and resistance to policy choices to determine their effects on

the amount of money committed to policy choices. They hypothesized that if a subject

received negative feedback regarding a policy choice, faced high policy resistance, and had

low job security, the subject would be most likely to escalate commitment to the policy

choice. This hypothesis was based on the idea that if the subject had implemented an

unpopular policy, there would be more pressure to avoid failure. Similarly, if an

individual's job security were low, there would be more motivation to save a failing policy









rather than admit failure. The hypothesis was supported. Additionally, commitment of

resources was lowest in the low-job-insecurity and low-resistance condition. There were

main effects for both job insecurity and policy resistance on the amount of resources

committed.

The Present Research


The literature reviewed above has examined both differences in performance

evaluation effectiveness due to organizational factors and differences in behavior due to

individual reactions to performance evaluation. However, the effect of the time span of

performance evaluation on individual behavior has not been the central focus of research to

date. The question that this study poses is whether varying the frequency of performance

evaluation using accounting-based performance measures will affect an individual's

psychological processes and resulting decisions. This query is especially relevant in light

of evidence of potential dysfunctional consequences of ABPM and the nature of concerns

expressed about the short-term emphasis in financial statements. The answer to this

question will provide an extension to the present accounting-based performance evaluation

literature and may have implications for motivating managers behavior to behave in a

manner more consistent with organizational objectives.














CHAPTER 3
MODEL DEVELOPMENT AND RESEARCH HYPOTHESES


Introduction


This chapter describes the influence that the frequency of accounting-based

performance reports may have on an individual's tendency to escalate commitment to a

course of action. Kanodia, Bushman, and Dickhaut [1989] showed analytically that

escalation errors are driven by the lack of commitment by the employer to ignore switching

decision information. Furthermore, they found that when a manager's reputation is

considered, escalation behavior can be economically rational. This study incorporates their

findings by examining how varying the frequency of performance evaluation may change a

manager's perception of possible negative repercussions from switching to a superior

investment. The purpose of the current model is to present the effect of performance

evaluation on behavior as a function of cognitive processes. Hypotheses are developed for

the effects of performance evaluation.

The Model


The conceptual model of the effect of performance evaluation on psychological

processes underlying the investment choice is developed from the relevant psychological

literature.

When an individual chooses between two alternatives, it is posited that he or she

considers the results (hereafter called benefits) of each alternative.1 The benefits can affect



1Because the costs of a decision can be reframed as the benefits of the alternative, only the
benefits are considered. This eliminates the possibility of double-counting.









either the individual (called personal) or the organization (called organizational). How an

individual will be evaluated should affect the perception of these benefits.

In this model it is posited that the Frequency of Performance Evaluation (short-term

or long-term) will affect the perceived benefits of a choice between remaining committed to

an investment (escalating) or changing to an alternative (switching).2 Specifically, because

the act of evaluating a manager's performance represents a form of public scrutiny, it will

affect an individual's perception of the image which he or she can claim and the expected

"value" of this claim.3 As explained in the following sections, through its effect on the

image, performance evaluation influences the perception of Personal Benefits from

Escalating.

The effect of the Frequency of Performance Evaluation on Personal Benefits from

Escalating may be moderated by the individual's perception of the ability to manipulate his

or her image to others. The effect of the Frequency of Performance Evaluation on Personal

Benefits from Escalating may also be moderated by whether an individual places a higher

priority on personal welfare than on doing the job "right." Finally, all types of perceived

benefits (personal or organizational, resulting from switching or escalating) will affect the

decision to escalate or switch. The variables in this model are shown in Figure 3-1,

described in the following section, and summarized in Table 3-1.


Frequency of Performance Evaluation


Since the purpose of this study is to examine the influence of the Frequency of

Performance Evaluation on decision making, the focus is on two discrete levels of time:

short-term (ST) and long-term (LT). This independent variable is higher if an individual is


2Hereafter all variable names will be capitalized.

31n this study, value and expected value refer to the general expectations of the worth of a
particular action. These terms do not refer to the traditional mathematical representations
where probabilities and values are combined to determine expected value.









Table 3-1
Summary of Variables


DEFINITION


TVPF T 1;VFT~ M ThAF; IP1 Th


Frequency of
Performance
Evaluation


Hl-IV
H2-IV
H3-IV


Perceived Weighted number of
Benefits perceived benefits


OBS


OBE


PBE




PBS


Ability to
Manipulate
One's Image


Organizational Benefits
from Switching

Organizational Benefits
from Escalating

Personal Benefits from
Escalating



Personal Benefits from
Switching

Perceived ability to
manipulate image


Discrete
Two Levels:
ST, LT


Continuous


H4b-IV


H4a-IV


H1-DV
H2-DV
H3-DV
H4a-IV

H4b-IV


H2-MV Continuous


JOBRITE Degree to which a subject H3-MV Continuous
considers doing the job
right instead of
personal consequences


Randomly
Assigned


Measured:
Open-list and
Fixed-list


Measured:
General Self-Efficacy
Scale, Aggression
Scale

Measured:
Post-Experimental
Question


Periods
Until
Switching


Number of periods until H4a-DV
the subject switches to H4b-DV
the alternative investment
or does not switch


Discrete Measured
Three Levels:
Switch Period 2,
Switch Period 3,
Never Switch


Key: DV-Dependent Variable
IV-Independent Variable
MV-Moderating Variable


M A Ar?


Frequency of
Performance
Evaluation


I AV1.. L.. IIIIi 111 Y. 1 I r i. .-, l-- l.R yJT-
















PERIODS
UNTIL
SWITCHING


Figure 3-1
Conceptual Model of the Effect of Performance
Evaluation on Escalation Behavior

evaluated frequently over the course of a project (ST) and lower if an individual is
evaluated at the end of a project (LT).


KEY:

I Information Cue

C) Cognitive Process

0 Management Behavior
7 Personality Variable









Perceived Benefits of an Action


When the individual is presented with a superior alternative investment following

acceptable feedback regarding the original investment choice, there are two possible

actions: escalate commitment or switch to the alternative investment.4 During this

decision, the individual considers the benefits of both options. The benefits which result

from the decision are personal or organizational.

The following discussion describes the predicted perceived benefits of each action:

Organizational Benefits from Escalating, Organizational Benefits from Switching, Personal

Benefits from Switching and Personal Benefits from Escalating. These benefits are

summarized in Table 3-2. Benefits resulting from the original investment's performance

exist regardless of whether the individual decides to escalate or to switch. Therefore, it is

predicted that the individual will eliminate this information from the evaluation (see

Kahneman and Tversky [1979] for a discussion of editing). After this elimination, the

benefits associated with each action diverge.

Organizational Benefits from Escalating. If an individual escalates, the benefit to

the organization for remaining with the current project is a continuance of the neutral

economic performance which occurred in the previous period.

Organizational Benefits from Switching. If an individual switches, it is assumed

that the new project will be successful, or at least more successful than the current project.5

Therefore, the company will receive improved cash flows in future periods.

Personal Benefits from Switching. The individual who switches will have a

positive image associated with the future positive cash flows.


4In this study subjects received feedback that their investment choice met the budget.
Hereafter, this feedback is referred to as neutral feedback.

5The case developed for this study was intended to present the alternative project as one
with a high probability of success. However, some subjects perceived it as a risky project.
The implications of this are discussed in Chapter 5.








Table 3-2
Predicted Perceived Benefits of an Investment Choice


Benefits from Switching


-- I I


ST organizational


Personal


1. Future Positive
Economic Outcome





2. Future Positive Image
for project success


4 4 1*


LT Organizational


Personal


1. Positive Overall
Economic Outcomes



2. Positive Overall
Image for project
success


Benefits from Escalating


1. Future Neutral
Economic Outcome





2. Future Neutral Image
for project meeting
budget

3. Positive consistent
image

4. Avoid negative image of
mistake, bearer of bad news
associated with switching


1. Neutral Overall
Economic Outcomes



2. Neutral Overall
Image for project
meeting budget


II I I


The Frequency of Performance Evaluation may affect an individual's perception of

these three types of benefits. Consistent with economic theory and the research on time

discussed in Chapter 2, it is possible that temporally distant outcomes will be discounted

relative to those that are closer. With ST performance evaluation, future benefits for

escalating or switching are not explicitly recognized at the time of the action. With LT

evaluation, these benefits are recognized because the project has been completed. This

could cause individuals to weight LT benefits greater than ST benefits. However, this









relationship is complicated by the fact that the LT performance evaluation is also temporally

distant. The effect of Frequency of Performance Evaluation may be minimal in this

situation. Therefore, it is not predicted that Frequency of Performance Evaluation will

affect the perception of these three types of benefits.

Personal Benefits from Escalating. It was predicted above that the Frequency of

Performance Evaluation would not affect the perception of the Organizational Benefits from

Escalating, Organizational Benefits from Switching or Personal Benefits from Switching.

Using the same logic, the Frequency of Performance Evaluation is unlikely to affect one of

the Personal Benefits from Escalating. Specifically, the Frequency of Performance

Evaluation does not affect the neutral image associated with neutral economic performance

because this benefit occurs with either type of evaluation.

However, two additional Personal Benefits from Escalating may occur only if an

individual is evaluated in the short run. First, the individual may present an image of

consistency and confidence. Norms of society tend to encourage consistency and

discourage inconsistency, and therefore, people manage impressions to establish the image

of consistency [Schlenker 1980]. Second, escalating makes it possible to avoid the

admission of having made a mistake. Research has demonstrated how distasteful it can be

to admit to having made a mistake. It has been shown that subjects will sacrifice monetary

payoffs to minimize public embarrassment [Brown 1968; Brown 1970; Brown and

Garland 1971; Garland and Brown 1972]. In fact, this tendency has been labelled the

"MUM effect" whereby individuals tend to keep mum about negative messages [Kardes

and Kimble 1984]. Individuals seem to perceive that the image of the messenger will be

directly related to the message.

Based upon these expected perceived benefits, it is predicted that the frequency of

accounting-based performance reports will influence the individual's perceived Personal

Benefits from Escalating. The Frequency of Performance Evaluation is expected to have a

negative effect on the continuous dependent variable Personal Benefits from Escalating.









Therefore, it is predicted that Personal Benefits from Escalating will be greater in the ST

than in the LT condition.

H1: Where there exists pressure to escalate commitment, the perceived
Personal Benefits from Escalating will be greater when an individual is
evaluated in the short run than when an individual is evaluated over the
long run.


Possible Interaction Effects


Hypotheses 2 and 3 explore the possibility that the effect of the Frequency of

Performance Evaluation on Personal Benefits from Escalating will be moderated by other

variables. Two possible interactions are proposed. The first is that an individual's

perceived ability to manipulate his or her image to others may influence the relationship

between the Frequency of Performance Evaluation and Personal Benefits from Escalating.

The second is that the same relationship is also moderated by an individual's attitude

toward doing the job "right."

Ability to Manipulate One's Image. An individual's perceived ability to manipulate

his or her image to others has been described as self-presentational efficacy expectancies

[Leary and Atherton 1986; Maddux et al. 1988]. These expectancies consist of self-

efficacy and outcome expectancies. This study is concerned with the self-presentational

outcome expectancy, defined as the subjective probability that a person can perform a

specific behavior intended to convey a particular impression.

In the short-term evaluation condition, there are more benefits from escalating than

from switching because of the need to protect one's image. It is possible, however, that if

an individual has a low self-presentational outcome expectancy, he or she will not perceive

an ability to control the image presented to others. In this case, the individual probably

would not consider those benefits to the image in the decision (or would consider them as

less important); and the benefits from switching would outweigh the benefits from

escalating. Lack of consideration of the benefits to the image would cause the benefits of









short-term evaluation to be the same as the benefits of long-term evaluation. In both cases

the individual would not consider benefits to the image. This is true in the long-term

condition because there is no need to protect one's image since switching behavior is no

longer salient (see Table 3-2). Therefore, the Frequency of Performance Evaluation may

only affect those who believe they can control their image to others. It is predicted that the

perceived Ability to Manipulate One's Image will moderate the relationship between the

Frequency of Performance Evaluation and Personal Benefits from Escalating. The effect of

the Frequency of Performance Evaluation will be greater the higher the perceived Ability to

Manipulate One's Image.

This prediction is illustrated in Figure 3-2.



Personal High MANIPULATE IMAGE
Benefits
from
Escalating
Low MANIPULATE IMAGE


LT ST


Figure 3-2
Predicted Interaction of the Perceived Ability to Manipulate One's Image to Others and the
Frequency of Performance Evaluation


The following hypothesis is based upon this prediction.

H2: Where there exists pressure to escalate commitment, there is an
interaction between the Frequency of Performance Evaluation and the
perceived Ability to Manipulate One's Image affecting Personal Benefits
from Escalating.

The interaction is such that the effect of the Frequency of Performance
Evaluation on Personal Benefits from Escalating becomes stronger as
the perceived Ability to Manipulate One's Image becomes greater.

Doing the Job Right. The degree to which a person considers doing the job "right"

affects the salience of performance evaluation. If a person considers personal welfare to be









more important than doing the job right, the type of performance evaluation is significant

because it differentially affects personal benefits. There are more implications for personal

benefits from escalating with short-term performance evaluation than with long-term. On

the other hand, if doing the job right has a higher priority than personal welfare, the type of

performance evaluation is irrelevant because the correct action is identical in either

condition. The variable, "Doing the Job Right" has a higher value if a person considers

doing a job "right" before considering how he or she is affected personally. Because the

level of Doing the Job Right affects the importance of performance evaluation, it will

moderate the relationship between the Frequency of Performance Evaluation and Personal

Benefits from Escalating. It is predicted that the Frequency of Performance Evaluation will

not affect Personal Benefits from Escalating if the individual has a high level of Doing the

Job Right, because there is no distinction between long-term and short-term performance

evaluation. The lower the level of Doing the Job Right, the greater the effect the Frequency

of Performance Evaluation, because the distinction between long-term and short-term

performance evaluation is more salient.

This prediction is illustrated in Figure 3-4 and is the basis of the following

hypothesis.


Personal Low JOBRITE
Benefits
for
Escalating
High JOBRITE


LT ST


Figure 3-4
Predicted Interaction of the Consideration of Doing the Job Right (JOBRITE)
and the Frequency of Performance Evaluation








H3: Where there exists pressure to escalate commitment, there is an
interaction between the Frequency of Performance Evaluation and an
individual's consideration of Doing the Job Right, affecting Personal
Benefits from Escalating.

The interaction is such that the effect of the Frequency of Performance
Evaluation on Personal Benefits from Escalating becomes stronger the
less the individual considers Doing the Job Right.


Action: Switching or Escalating


After receiving information about a superior alternative investment, if an individual

perceives more benefits from switching than from escalating, it is predicted that the

individual will change to the alternative investment project. Alternatively if an individual

perceives more benefits from escalating than from switching, it is predicted that the

individual will escalate commitment to the original investment project. The ordinal

dependent variable (three levels), Periods Until Switching, has a low value if the individual

makes the economically rational choice and switches to a superior alternative investment

and has a high value if the individual does not switch. These predictions are presented in

the following hypotheses.


H4a: Where there exists pressure to escalate commitment, there will be
two positive correlations: 1) between Personal Benefits from Escalating
and Periods Until Switching, and 2) between Organizational Benefits
from Escalating and Periods Until Switching.

H4b: Where there exists pressure to escalate commitment, there will be
two negative correlations: 1) between Personal Benefits from Switching
and Periods Until Switching, and 2) between Organizational Benefits
from Switching and Periods Until Switching.


Summary


In this chapter, a model of the differential effect of the frequency of accounting-

based performance reports on the capital budgeting decision in an escalation context was

presented. Dichotomizing the perceived benefits into personal and organizational

categories, the model described the effect of the frequency of accounting-based





31


performance reports on a specific subset of benefits: Personal Benefits from Escalating. In

addition, variables that moderate the relationship between the Frequency of Performance

Evaluation and Personal Benefits from Escalating were included in the model. Chapter 4

details the experiment that was conducted to test the predictions of this model.














CHAPTER 4
RESEARCH DESIGN AND DATA ANALYSIS METHODS


Introduction


This chapter describes the research method used to test the hypotheses discussed in

Chapter 3. First, the design of the experiment conducted to investigate the effects of the

frequency of accounting-based performance reports on capital budgeting decisions is

presented. Next, the chapter describes the data analysis methods and the predicted results.


Experimental Method


The study was conducted in a laboratory setting, and used students as subjects.

The following sections present, in turn, the general setting for the experiment, the

experimental materials, the experimental procedures (including the incentive structure), and

the subjects.


General Setting


The original and subsequent investment choices were made in the context of a

business simulation using an in-basket scenario. Subjects were asked to assume the role of

Acting Capital Procurement Manager (a newly-created and possibly temporary position) of

the Paging Division of Mototronics, an electronics corporation. As the Capital

Procurement Manager they were asked to respond to corporate memos and narratives with

comments and decisions concerning capital projects.

Subjects were randomly assigned to long-term or short-term performance

evaluation conditions so that comparisons could be made between their cognitive processes

and between their decisions. It was critical to the examination of the effects of the









Frequency of Performance Evaluation that the type of performance evaluation was made

salient to the subjects. Therefore, the study necessarily encompassed several periods. In

the experimental setting, each period represented a fiscal year. Subjects were told at the

outset that the study would continue for multiple periods but were not told the actual

number of periods.1

Based on the results and predictions of the sunk cost literature, information

designed to encourage escalation was contained within each case (see the Appendix for an

example of these materials).


Experimental Materials


The case materials were developed over a series of pilot studies. Background

information, instructions for completing the materials and the rules for earning payment

were presented at the outset of the experiment. The subjects were asked to assume the role

of Lee Chambers who had been temporarily assigned to fill the position of Acting Capital

Procurement Manager. They were told that the position was not secure:

Unfortunately, you have only been temporarily assigned to fill the
position (your official title is Acting Capital Procurement Manager). A
complete evaluation of the reorganization which resulted in your
promotion will occur within the next five years and your job will
become permanent or you will be demoted back to Production Manager,
depending on your performance. There is some uncertainty as to
whether your former position will still be available because of the
current recession and rising unemployment. In the meantime, you have
to deal with other executives who are well-qualified and envious of your
position. Indeed, they were unhappy that you were chosen over them to
fill the temporary position. Thus, you cannot expect support and
assistance from your peers, especially if you do not perform well
immediately.



1Subjects were not told the number of periods in order to avoid the "finite-period paradox."
This paradox occurs in multiple-stage experiments when the actual end of the experiment is
known. If the timing of the end of the experiment is known, instead of treating the
experiment as a multiple-stage game, subjects treat it as a series of single-stage games. If
the timing of the last interaction is unknown, this problem should not occur [Luce and
Raiffa 1967; Kreps and Wilson 1982; Axelrod 1984].









They were also informed that would receive a performance evaluation on the degree of

success or failure of their decisions. This evaluation would be based on their "reactions to

the situations described in the case and would account for a wide variety of information:

cost attainment and relationships between the principal actors in the case, to name a few."

After being informed of the experimental procedure, they began the experiment The

experimental materials, listed in order of presentation, are described below.

Period 1. Subjects received the following materials:

1. A memo from the Personnel Manager congratulating them on their new position

and providing details about the position, including an organizational chart and information

regarding the timing of their performance evaluation (annually or at the end of the project's

life).

2. The planned production of pagers through 1995.

3. Information regarding the choice between two machines which would be used to

manufacture outside housings for a new pager. The choice was between machine 0036 and

machine 0073. The scenario was designed so that the only major difference between the

two machines was that 0036 was produced in Arizona and 0073 was produced in New

York.

Period 2. Subjects received these materials:

1. A memo from their superior discussing the negative reaction of the Board of

Directors to their decision made in period 1 and their superior's defense of their decision.

This memo was included to encourage escalation by creating high policy resistance to the

original investment choice. An example of this policy resistance is:

... several Board members were very dissatisfied and critical of your
recommendation and were firmly prepared to vote against it. Although
they were highly skeptical and critical of your recommendation and were
firmly convinced you had recommended the wrong course of action, in
the final analysis, the Board reluctantly deferred to your judgment
(Lee, I really have to point out that I believe the Board finally agreed to








support your recommendation because I went out on a limb for you and
defended your decision, not because they were pleased). 2

2. A Production Cost Report indicating that the costs of their choice, as well as

other current investments under their management, had remained within budget (neutral

feedback).3

3. A performance evaluation (ST condition only). The performance evaluation was

based upon the fact that expenses related to the investments did not exceed budget and

"upon overall performance." Performance for the period was rated "good."

4. A memo from their assistant with private information about an alternative

machine which would cut costs by 75%.4


2The Job Insecurity and Policy Resistance manipulations were modeled after Staw's
[1979] Adams and Smith Decision Case. Staw's study showed that escalation of
commitment was strongest when these variables were present. In the present study these
manipulations are used verbatim with slight changes to reflect the different decision,
company, etc. These are reprinted from "The Trapped Administrator: Effects of Job
Insecurity and Policy Resistance upon Commitment to a Course of Action" by Frederick V.
Fox and Barry M. Staw published in Administrative Science Quarterly (volume 24, number
3) by permission of Administrative Science Quarterly. Copyright 1979 Cornell University.

3The instrument used in the first pilot studies contained negative feedback regarding the
original investment choice. Subjects who participated in the pilot studies which tested this
instrument complained that they had no control over the outcome and felt hopeless.
Therefore, the feedback was changed to neutral in order to alleviate these problems and to
direct the subjects' attention to the questions of interest.

4This feedback was structured so that the subsequent correct decision was to switch to the
superior alternative investment. Subjects continued to receive this feedback until they
switched to the alternative investment. For example, a subject deciding to invest in
machine 0036 received feedback that the alternative investment was superior until he or she
switched (if ever) to Machine 0073. This decision is illustrated in Figure 4-1. Once a
subject switched, he or she received feedback which indicated that it was best to remain
with the new investment. If the subject received feedback that it was best to switch back, it
would have been possible to infer lack of causality, inducing the subjects to make
meaningless future decisions (as may have applied in Staw and Fox
[1977]).


Figure 4-1
The Decision Context









5. A list of facts regarding the case designed to remind the subjects of pressures to

escalate, followed by a request for their choice of machine (the original machine or the

alternative machine with 75% lower costs).

6. A request to complete four open-ended lists: "Specific benefits for me

personally if I remained with xxxx, Specific benefits for Mototronics if I remained with

xxxx, Specific benefits for me personally if I changed to xxxx, and Specific benefits for

Mototronics if I changed to xxxx."5

7. A request to weight those benefits by allocating a total of 100 points among all

the items in the four lists.

8. One list of 14 "Specific benefits if I remained with xxxx" (7 Personal and 7

Organizational) and one list of 14 "Specific benefits if I changed to xxxx" (7 Personal and 7

Organizational). The subjects were asked to indicate the importance of each benefit to their

decision on a scale from 1 (not at all important) to 5 (extremely important).6

Period 3. Subjects received the following:

1. Production Cost Report indicating that the cost of their choice, as well as the

other current investments, had remained in budget.

2. Performance Evaluation (ST condition only). If the subject had escalated, the

performance evaluation was "good," if the subject had switched, the performance

evaluation was "excellent."



5Machine numbers are designated "xxxx" because the materials contained either 0036 or
0073 depending upon the original choice of the subject.

6The items on the lists were developed from the results of open-ended responses from the
pilot studies and from the theory discussed in Chapter 2. They consisted of two 14-item
lists of benefits for switching and benefits for escalating. Within each list there were seven
organizational benefits and seven personal benefits which were randomly mixed. The
results of the final pilot study tested the reliability of the four subscales (Personal Benefits
from Escalating, Personal Benefits from Switching, Organizational Benefits from
Escalating and Organizational Benefits from Switching). All had alphas .77 or greater,
therefore the scales were used in the actual experiment.









3. Information regarding the choice between their machine and the alternative. If

they had escalated they had a choice between further escalation or switching. If they had

switched they could remain with the new project or change back to the original investment.

Period 4. Subjects received the following:

1. Performance Evaluation: "good" if they escalated during the entire experiment,

"excellent" if they switched at any point during the experiment.

2. Post-experiment questionnaire.

3. Aggression and General Self-efficacy questionnaires.


Experimental Procedure


The experiment began with a briefing by the researcher. Next, the subjects were

given the experimental materials in a set of envelopes and were required to work with one

envelope of materials at a time. When cued, subjects chose the next envelope which

reflected their previous choice. For example, the first decision required subjects to choose

either Machine 0036 or Machine 0073. If the subject chose Machine 0036, he or she then

completed the next envelope of materials which was labelled, "Yr 1 0036" and similarly

"Yr 1 0073" if he or she had chosen Machine 0073.

Because the students participated in the experiment during one class period it was

not possible to conduct the experiment over multiple periods. Therefore, in order to create

the illusion of "years" passing, participants were given a set amount of time for each "year"

(year one-20 minutes, year two-35 minutes, year three-5 minutes, year 4 and post-

experimental questionnaire-10 minutes). Participants were required to work only on the

applicable year during the time allotted and were not permitted to advance to the next year's

set of materials until a timer rang, signifying the completion of one year.

All subjects completed a post-experimental questionnaire which elicited data on the

strength of the manipulations. This questionnaire measured the face validity of some of the

manipulations by asking the subjects to rate on a five point scale (coded l=strongly









disagree to 5=strongly agree) the degree to which they agreed with certain statements. The

subjects were also asked to provide general comments regarding the experiment and to

answer questionnaires measuring their the perceived ability to manipulate their image to

others.

Incentive structure. To provide the subjects with motivation both to participate in

the experiment and to behave in a self-interested manner, lottery tickets were used as

rewards. Subjects were informed that the tickets would be awarded based on three

measures. The first measure was how well the subjects responded (as Lee) to the

situations provided in the experiment. The second measure was based upon Lee's final

position with Mototronics. Finally, subjects earned tickets based upon Lee's performance

evaluation.

It was expected that providing monetary compensation based upon these variables

would heighten the salience of the time, job security, and policy resistance manipulations,

and accentuate the fact that compensation depended on the superiors' overall assessment of

their behavior.

The actual allocation of tickets was based upon the following rules. For the first

category (how well they responded) each subject began with 10 points. Points were then

subtracted for missing answers and for incorrect answers. For the ending position

category, all subjects were retained as the permanent Capital Procurement Manager (5

tickets) unless they had done an extremely poor job on completing the materials in which

case they were demoted to Production Manager (2 tickets) or fired (0 tickets). Finally,

tickets were awarded based upon on the final performance evaluation rating: 6 tickets for a

"good" evaluation, 10 tickets for an "excellent" evaluation. The average number of tickets

earned for overall performance was 8.18, for ending position 4.49. and for the final

performance evaluation 8.48. Each lottery ticket represented a chance to win one of seven

cash prizes. The prizes offered were one award of $200, two awards of $100 and four









awards of $50. The drawing was held the Saturday following the completion of the

experiments.

Whether a performance-contingent incentive results in better performance than

straight pay or no financial incentive, seems to depend on the situation [Ashton 1990].

Libby and Lipe [1991] showed that performance-based payments increased effort when

effort-sensitive processing was required for the task. Their study provides support for

using the lottery system in this experiment since effort is required to process the

information provided to the subjects (i.e., it is not a highly-structured decision).

Furthermore, Bolle [1990] found that, in experiments where decision costs were small and

choices were anonymous, subjects' behavior did not differ between a randomized reward

structure and a deterministic reward structure.


Subjects


Subjects were recruited from a first-year Masters of Business Administration cost

accounting class and from undergraduate and graduate accounting classes. Sixty-five MBA

students participated during their regular class. The remainder of the subjects (43)

participated outside of regular class time. Participating undergraduate Cost 1 students

received extra-credit toward their class grade.7

Table 4-1 contains a summary of demographic information for the subjects. A

majority of the subjects were male (71%), graduate students (63.3%), had held full-time

jobs (68.8%), and had GPA's 3.0 or above (91.7%). Accounting majors made up 39% of

the sample, while the remainder were finance (25.7%), management (3.7%), marketing



7Subjects drawn from corporate management represent the group to which researchers
wish to extend their experimental results. However, as the theory upon which this study is
based relies on basic psychological principles which apply to all individuals, the use of
student subjects is considered acceptable as a starting point for examining the research in
this area.









(13.7%), and other (14.7%). In addition, 33% of the subjects were between the ages of 18

and 22, 45% were between 23 and 27 years of age, and the remainder were over 28 (22%).

Table 4-1
Summary of Subject's Demographic Information

ST LT
Grouping Variable Number Percentage # (%) # (%)

Gender
Female 38 34.9% 19 (50) 19 (50)
Male 71 65.1 37(52) 34(48)

Major
Accounting 42 38.5% 21(50) 21(50)
Economics 0 0 0 0
Finance 28 25.7 11(39) 17 (61)
Management 4 3.7 2 (50) 2 (50)
Marketing 15 13.7 9 (60) 6 (40)
Other 16 14.7 11(69) 5(31)
Not Reported 4 3.7 2 (50) 2 (50)

Classification
Undergraduate 38 34.9% 20 (53) 18(47)
Graduate 69 63.3 35 (51) 34 (49)
Not Reported 2 1.8 1 (50) 1 (50)


Age
18-22 36 33.0% 20 (56) 16(44)
23-27 49 45.0 26 (53) 23 (47)
28-32 19 17.4 8 (42) 11(58)
Over 32 5 4.6 2 (40) 3 (60)


Have held a full-time job
Yes 75 68.8% 42 (56) 33 (44)
No 34 31.2 14(41) 20(59)

GPA
2.5-2.9 9 8.3% 3 (33) 6 (67)
3.0-3.4 52 47.7 28 (54) 24 (46)
3.5-4.0 45 41.2 24 (53) 21(47)
Not Reported 3 2.8 1 (33) 2 (67)








Variables


Due to concerns about the dysfunctional effects of short-term emphasis in

performance evaluations, the investment decision modeled here is the choice of switching

or escalating after the receipt of feedback regarding a superior alternative choice. Because

the focus is on the effects of the Frequency of Performance Evaluation on the psychological

processes underlying the investment decision, all other variables (e.g., personal

responsibility, job security) were held or assumed to be constant across the Frequency of

Performance Evaluation conditions.

Since the Frequency of Performance Evaluation was predicted to influence Personal

Benefits from Escalating directly and to influence the number of Periods Until Switching

indirectly through Personal Benefits from Escalating, it was necessary to measure Personal

Benefits from Escalating and the number of Periods Until Switching as dependent

variables. Perceived Benefits were also used as independent variables in the switching

behavior model.

Following is a discussion of the independent, dependent and moderating variables.


Operationalization of Independent Variable


The theoretical construct of interest is the temporal relationship between the

investment choice and the performance evaluation. Therefore, those in the long-term

condition were exposed to a longer time period between the investment choice and

evaluation than those in the short-term group. To operationalize this, the timing of the

subjects' performance evaluation was varied between the short term and the long term. The

Frequency of Performance Evaluation was manipulated between-subjects. Each subject

was randomly assigned to one of two experimental treatment conditions: short-term or

long-term performance evaluation. Those in the short-term condition (56 subjects) received

the performance evaluation following the end of each period, prior to their subsequent









investment decision. Those in the long-term condition (53 subjects) received their

performance evaluation at the end of the project life (at the end of four periods).


Measurement of Dependent Variables


The hypotheses discussed in the Chapter 3 concern five response variables: four of

these concern perceived benefits (Organizational Benefits from Switching. Organizational

Benefits from Escalating, Personal Benefits from Switching, and Personal Benefits from

Escalating) and the other is the number of Periods Until Switching. The measurement of

these response variable will be discussed next.

Perceived benefits. The benefits considered prior to making investment choices

were measured, following the first feedback, directly after the subjects reported their

second investment choice. The subjects were asked to list what specific benefits would

accrue to themselves and to Mototronics as a result of their investment decision.

These representations are self-reports. The use of self-reports as data is a

controversial issue. First, there is an unresolved question as to whether subjects are

capable of self-reporting internal states [Nisbett and Wilson 1977; Smith and Miller 1978;

Wright 1987; Wright 1988]. Ericsson and Simon [1980] posit that certain conditions

make it possible for accurate self-reporting of internal states. For example, because of the

limited capacity of short-term memory the experimenter can only expect to access the most

recently heeded information. Moreover, Gordon and Braun [1986] stated "yet it is the self-

report (verbal) data that provides some of the clearest insights into the confounding effects

of common aspects of the instructional treatments" (p. 299).

Second, there are no recognized solutions to the problems with methodology

[Eriksen 1962; Wright 1980; Ericsson and Simon 1980; Perone 1988]. Therefore, once an

experimenter has decided to use self-reports as data the method of data collection must be

tailored to the specific experiment. The relevant methodological issues in this experiment









were: the timing of data collection, the number of collections, and the types of probes.

Each is addressed in the following paragraphs.

The first methodological issue concerns the timing of data collection. Subjects'

thoughts may be collected during the actual decision (concurrently) or following the

decision (retrospectively). Wright [1980] stated that it is difficult to judge which method

more seriously affects completeness. Ericsson and Simon [1980, 1984] recommended

retrospective collection if the processing episode is brief and if it is the most recent

cognitive process that is measured. Because the processing episode is brief in this study,

and because concurrent collection might cause subjects to alter their decisions, the data was

collected immediately following the decision.

The second methodological issue is the proper number of data collections. The

process of probing and self-reporting may itself affect subsequent behavior. For example,

subjects may be alerted as to the experimenter's goals by the original probes and may then

act in ways to meet those perceived goals. To avoid this problem, the data was collected

once in the current study.

Finally, there is a question whether to use a directive or general probe. Each type

of probe has relative advantages and disadvantages [Payne, Braunstein and Carroll 1978;

Wright 1980; Ericsson and Simon 1980]. For example, a general probe may fail to

encourage subjects to elicit the information desired by the experimenter. On the other hand,

directive probes may encourage subjects to provide information they would not otherwise

have provided. In order to overcome these problems both types of probes were used in

this experiment. First, a general probe was used where subjects were asked to list the

specific benefits for them personally and for Mototronics for both alternatives (open-list).

This was followed by a fixed-alternative probe where subjects were asked to rate the

importance of each specific benefit provided on a list (fixed-list).

Periods until Switching. This dependent variable is the number of periods until the

subject switched. If the subject switched after the first feedback he or she received a score









of one. If the subject switched after the second feedback the score was two. Finally, if the

subject did not switch the score was three.


Measurement of Moderating Variables


Ability to Manipulate One's Image. Leary and Atherton [1986] posited that self-

efficacy (perceived ability to produce desired results) and outcome expectancies would be

affected by dispositional factors such as General Self-efficacy, self-esteem, social skills or

situational factors such as novel or ambiguous encounters.

In this study the construct of the perceived Ability to Manipulate One's Image is

proposed to be represented by two individual variables: General Self-efficacy and

Aggression. The perceived Ability to Manipulate One's Image should be greater the more

that person has personal expectations of mastery over situations (General Self-efficacy).

Furthermore, the perceived Ability to Manipulate One's Image of an aggressive individual

should be greater than that of a less aggressive individual.

General Self-efficacy was measured using the Self-efficacy Scale [Sherer et al.

1982]. The Cronbach alpha reliability coefficient for this scale was .86 (p. 665).

Aggression was measured using the Personal Assertion Analysis [Hedlund and Lindquist

1984]. Cronbach's alpha reliability coefficient was not provided by this study. However,

the test-retest reliability coefficient was .70 (p. 381).

Doing the Job Right. The concern with Doing the Job Right was measured with a

question on the post-experimental questionnaire. The question asked subjects to respond

on a five-point scale (coded l=strongly disagree to 5=strongly agree) to the statement, "In

general, when I do a job, I am more concerned about doing the job "right" than with how it

will affect me, my bank account, my family, etc."

Table 4-2 contains a summary of information on subjects' levels of Aggression,

General Self-efficacy and consideration of Doing the Job Right. This information was

collected after the experiment so that the subjects would not be sensitive to the researcher's









interest in their personality variables. The levels of Aggression, General Self-efficacy and

Doing the Job Right were equal across long-term and short-term conditions, ensuring that

this distribution would not confound the results of the hypotheses tests.

Table 4-2
Summary of Subject's Levels of Aggression, General Self-efficacy and Doing
the Job Right (JOBRITE)


Overall ST LT
Variable Mean Mean Mean I- RE


Aggression 31.51 31.40 31.62 .33 .744

Efficacy 69.26 69.36 69.15 -.14 .890

JOBRITE 3.53 3.64 3.41 .97 .333


Data Analysis Methods


Data Screening


First, the computer output of raw data was compared to the original documents.

This was followed by an analysis of frequencies for reasonableness.

Missing Values


Analyses were conducted with and without subjects whose output included missing

values. Furthermore, subjects with missing values were compared to subjects without

missing values to determine if there were commonalities among the groups.


Manipulation Checks


Whether subjects responded to the Frequency of Performance Evaluation

manipulation was determined by comparing the means of the responses on the post-

experimental questionnaire between the short-term subjects and the long-term subjects. If

there was a significant difference between the two groups as to whether their performance









was evaluated annually it was assumed that they were cognizant of this manipulation.

Significant differences on other post-experimental questions were not expected.


Nuisance Variables


If inspection of the data revealed the necessity to remove the effect of variables not

hypothesized to affect the dependent variables, it was necessary to conduct partial

correlation analysis, ANCOVA, or to include an additional variable in the multiple

regression analysis.8


Scales


Two scales were used to measure subjects' perceived benefits: open-list and fixed-

list. The open-list asked subjects to list their perceived benefits and weight each according

to its importance in their decision. Subjects were then provided with a prepared list of

benefits (fixed-list) and asked to rate the importance of each benefit listed in their decision.

The fixed-list was tested for reliability.

The convergent validity of the open-list and the fixed-list was tested using

correlations.


Test of Hypotheses


Personal Benefits from Escalating H1


The following equation was developed for the purpose of testing H1:
Y = BI + B2X9



8This was important to ascertain due to the results of the original pilot study. This study
had 31 subjects and was administered to an undergraduate cost accounting class. The
effect of the Frequency of Performance Evaluation on Periods Until Switching was
moderated by gender. Females tended to switch immediately while ST males escalated
more than LT males.









where Y is Personal Benefits from Escalating

X is the Frequency of Performance Evaluation, (a dichotomous variable)

X = 1 for LT performance evaluation

2 for ST performance evaluation

H1 was supported if the mean of Perceived Benefits from Escalating in the LT was

significantly less than the mean of Perceived Benefits from Escalating in the ST. The

manner in which H was tested was dependent upon the treatment of the dependent

variables. Personal Benefits from Escalating, Personal Benefits from Switching,

Organizational Benefits from Escalating, and Organizational Benefits from Switching could

either be tested separately or simultaneously. If tested separately, a t-test would be

appropriate. If they were treated simultaneously a MANOVA (multivariate analysis of

variance) would be appropriate. Because it is the univariate relationship between the

Frequency of Performance Evaluation and Personal Benefits from Escalating that is of

interest it was considered appropriate to conduct univariate tests (t-tests).


Ability to Manipulate One's Image H2


The following equation was developed for the purpose of testing Hypothesis 2:
Y = BI + B2X + B3M + B4XM

where Y is Personal Benefits from Escalating

X is the Frequency of Performance Evaluation

M is the perceived Ability to Manipulate One's Image

XM is the interaction between the Frequency of Performance Evaluation and
the perceived Ability to Manipulate One's Image


9In order to simplify the presentation of the equations, basic regression models are used.
Technically, when conducting mean comparisons, the ANOVA model is more appropriate.
For example, for H1 the ANOVA model would be: Xij = M + Pi + eij, where Xij is the
observed response of subject number j (j =1,2,...,ni) to treatment i (i =1,2), M is the
superpopulation grand mean, Pi is the variance associated with treatment i. and eij is
random error.









Multiple hierarchical regression was used to test H2.10 The interaction's significance was

tested by determining whether the addition of the interaction term to the equation resulted in

a significant increase in R2.


Doing the Job Right H3


The following equation was developed for the purpose of testing Hypotheses 3, 3a

and 3b:
Y = BI + B2X + B3J + B4XJ

where Y is Personal Benefits from Escalating

X is the Frequency of Performance Evaluation

J is consideration of Doing the Job Right

XJ is the interaction between the Frequency of Performance Evaluation and
consideration of Doing the Job Right

Multiple hierarchical regression was used to test H3. The statistical significance of the

interaction was tested by determining whether the addition of the interaction term to the

equation resulted in a significant increase in R2 (See Cohen and Cohen 1975 for a

discussion of testing the significance of interactions).


Periods Until Switching H4

The following equations were developed for the purpose of testing H4a:
S= B1 + B2PBE

S = B3 + B40BE

where S is the number of Periods Until Switching (three levels)

S = 1 if subject switched in Period 2

= 2 if subject switched in Period 3


10The question of univariate vs. multivariate analysis is also applicable in the
supplementary analysis. If multivariate analyses were deemed necessary, the appropriate
analyses would be canonical correlation or path analyses.









= 3 if subject never switched

PBE is Personal Benefits from Escalating

OBE is Organizational Benefits from Escalating

H4a is supported if the correlations between the number of Periods Until Switching and

Personal Benefits from Escalating, and between the number of Periods Until Switching and

Organizational Benefits from Escalating were positive and significantly different from

zero. 1

The following equations were developed for the purpose of testing H4b:
S= B1 +B2PBS

S = B3 + B4OBS

where S is the number of Periods Until Switching (three levels)

S = 1 if subject switched in Period 2

= 2 if subject switched in Period 3

= 3 if subject never switched

PBS is Personal Benefits from Switching

OBS is Organizational Benefits from Switching

H4b was supported if the correlation between Periods Until Switching and Personal

Benefits from Switching and between Periods Until Switching and Organizational Benefits

from Switching were negative and significantly different from zero.

The appropriate analysis for hypotheses 4a and 4b was dependent upon the

treatment of the independent variables. As discussed above univariate analysis was used

(correlations).







11Although this variable is not continuous, because there are three ordered levels,
correlation analysis is considered appropriate for the analyses of H4a and H4b.





50

Summary


This chapter described the experiment designed to test the hypotheses developed in

Chapter 3. An instrument was developed to measure cognitive processes underlying the

investment decision. Subjects were given incentives to participate in the experiment

The chapter also proposed data analysis methods to test the hypotheses. The results

of these analyses are presented in Chapter 5.














CHAPTER 5
RESEARCH RESULTS


Introduction


This chapter reports the results of the analyses proposed in Chapter 4. First, data-

screening methods, results of manipulation checks and analyses of nuisance variables are

discussed. This is followed by a description of the methods used to develop scales for the

measurement of Benefits, Aggression, and General Self-efficacy. Finally, the results of

tests of the research hypotheses are presented.


Data Screening


In order to ensure the data conformed to basic standards, the frequencies of the data

were analyzed for reasonableness. All variables were within acceptable ranges. Data

screening revealed response errors by subjects. The types of errors and their resolutions

are listed below:

1. Due to the complexity of the measurement of perceived benefits using the open-

list, some subjects did not understand how to assign points among the four open-ended

lists of perceived benefits. Six subjects applied 100 points to each of the lists; two

distributed 25 points to each. Because the instructions were to distribute 100 points among

all four lists as a reflection of the importance of the perceived benefits to their decision, it

was apparent that these subjects did not understand the instructions. These measures

therefore were treated as missing values.

2. Five subjects assigned weights which did not add to 100. The proportions of

the assigned weights were multiplied by 100 with the resulting weights used in analysis.









3. Four subjects circled two responses for one statement on the fixed-list. The

answers were coded as missing.

4. If a subject assigned fractional weights, anything equal to or less than x.5 was

rounded down to x.

5. In two cases subjects said they "saw few benefits" or "I really don't see any

benefits" in a specific category, yet still assigned weights to the overall category. These

weights were retained in the analysis.

Manipulation Checks


Job Insecurity

Three statements on the post-experimental questionnaire addressed the success of

the job insecurity manipulation. The results presented in Table 5-1 show that the

manipulation was somewhat successful. Although only 31.2% disagreed or strongly

disagreed with statement 2, "I felt secure in my job as Capital Procurement Manager," the

manipulation appeared to be more successful when measured by statements 5 and 7. A

majority (77%) of the subjects agreed or strongly agreed with statement 5, "Poor outcomes

from my recommendations clearly would have meant losing the job of Acting Capital

Procurement Manager." Furthermore, 66% agreed or strongly agreed with statement 7, "I

needed to protect my position as Acting Capital Procurement Manager in the company."


Policy Resistance


Manipulation checks reported in Table 5-1 revealed that the policy resistance

manipulation was moderately successful. First, a majority (54.1%) of subjects disagreed

or strongly disagreed with statement 3, "The Board of Directors was supportive of my

recommendations." Second, 49.6% of subjects agreed or strongly agreed with statement

6, "There was much resistance to my recommendations." Finally, a majority (67%) agreed









or strongly agreed with statement number 8, "The Board of Directors was reluctant to

accept my recommendations."

Table 5-1
Responses to Manipulation Checks
Full Sample


1. I felt pressure to "stick with" my
original decision.

2. I felt secure in my job as Capital
Procurement Manager.

3. The Board of Directors was
supportive of my recommendations.

4. I cared what "grade" I was to receive
on my performance evaluation.

5. Poor outcomes from my
recommendations clearly would have
meant losing the job of Acting
Capital Procurement Manager.

6. There was much resistance to my
recommendations.

7. I needed to protect my position as
Acting Capital Procurement Manager
in the company.

8. The Board of Directors was reluctant
to accept my recommendations.

9. During the experiment I acted as I
thought best, not as I thought "Lee"
or or other managers might act.

10. I wanted to perform well and make
the best decisions I possibly could
in this experiment.

11. My performance as Acting Capital
Procurement Manager was evaluated
annually.


Mean SA (%) A


N D SD


3.06 11.0 33.0 20.2 22.0 13.8


3.13 8.3


34.9 24.8 23.9 7.3


15.6 29.4 35.8 18.3


3.94 32.1 45.0 11.9 7.3 3.7



3.95 28.4 48.6 12.8 10.1 0


3.44 8.3 41.3 36.7 13.8 0



3.63 12.8 53.2 18.3 12.8 1.8


3.64 10.1 56.9 20.2 12.8 0



4.32 57.8 25.7 10.1 3.7 2.8



4.54 60.6 34.9 3.7 0 .9


3.45 22.9 37.6 13.8 12.8 12.8








Table 5-1--continued

Mean SA(%) A N


12. I looked back at my original list of
consequences even though the
instructions asked me not to do so
(Please answer honestly, there is no
penalty or reward for any answer). 1.37

13. I found the format of the experiment
(e.g., using the envelopes)
understandable. 4.17

14. In general, when I do a job, I am
more concerned about doing the job
"right" than with how it will affect
me, my bank account, my family,
etc. 3.53

15. I wanted to earn lottery tickets. 3.25

16. I believed that the better my decisions,
the more lottery tickets I would earn. 3.41



Note Sample size was 109 for all questions except
108 subjects due to missing values.


0 4.6 3.7 15.6 75.2



38.5 48.6 4.6 7.3 .9


24.8

20.2


34.9

26.6


14.7

27.5


20.2

9.2


5.5

16.5


19.3 32.1 27.5 12.8 8.3


numbers 2, 7, and 12 which each had


Experimental Administration


Two statements in the post-experimental questionnaire presented in Table 5-1

measured procedural concerns with the experiment and revealed no apparent problems with

the administration. Most (90.8%) of the subjects disagreed or strongly disagreed with

statement 12, "I looked back at my original list of consequences even though the

instructions asked me not to do so (Please answer honestly, there is no penalty or reward

for any answer)." Furthermore, 87.1% of the subjects agreed or strongly agreed with

statement 13, "I found the format of the experiment (e.g., using the envelopes)

understandable."


D SD








Frequency of Performance Evaluation


Table 5-2 details responses to the post-experimental questionnaire for LT and ST

subjects. This analysis revealed four significant differences between the LT and ST


Table 5-2
Responses to Manipulation Checks
ST vs LT


ST Mean LT Mean it


1. I felt pressure to "stick with" my
original decision.

2. I felt secure in my job as Capital
Procurement Manager.


3.30


3.16


3. The Board of Directors was
supportive of my recommendations. 2.46


4. I cared what "grade" I was to receive
on my performance evaluation.

5. Poor outcomes from my
recommendations clearly would have
meant losing the job of Acting
Capital Procurement Manager.

6. There was much resistance to my
recommendations.

7. I needed to protect my position as
Acting Capital Procurement Manager
in the company.

8. The Board of Directors was reluctant
to accept my recommendations.

9. During the experiment I acted as I
thought best, not as I thought "Lee"
or or other managers might act.


4.25




3.93


3.43



3.73


3.66



4.55


2.79


3.10


2.43


3.62




3.96


3.42



3.52


3.62



4.08


2.18 .032


.30 .763


.16 .875


3.31 .001



-.30 .764


.31 .759



1.19 .238


.24 .813



2.59 .011


1Additional analyses were conducted excluding all subjects who gave neutral answers to
this question. This did not change the significance, or lack of significance, for any of the
analyses. However, in the ANOVA analysis with Personal Benefits from Escalating
reduced fixed-list as the dependent variable and the Frequency of Performance Evaluation
as the independent variable, the marginal significance found in the original analysis (p =
.089) was no longer apparent (p = .238).








Table 5-2--continued

ST Mean LT Mean


10. I wanted to perform well and make
the best decisions I possibly could
in this experiment. 4.57

11. My performance as Acting Capital
Procurement Manager was evaluated
annually. 4.30

12. I looked back at my original list of
consequences even though the
instructions asked me not to do so
(Please answer honestly, there is no
penalty or reward for any answer). 1.32

13. I found the format of the experiment
(e.g., using the envelopes)
understandable. 4.32

14. In general, when I do a job, I am
more concerned about doing the job
"right" than with how it will affect
me, my bank account, my family,
etc. 3.64

15. I wanted to earn lottery tickets. 3.28

16. I believed that the better my decisions,
the more lottery tickets I would earn. 3.46


4.51



2.55




1.42



4.00




3.42

3.21


3.36


.49 .626



9.25 .000




-.69 .495



1.91 .058




.97 .333

.30 .761


.47 .642


subjects. First, statement 1 showed that short-term subjects felt more pressure to "stick

with" their original decision than did long-term subjects (t=2.18, p=.032). This provides

support for the theory presented in Chapter 3 which was based on the idea that ST

performance evaluation encourages escalation. Second, statement 4 revealed that short-

term subjects cared more about what "grade" they would receive on their performance

evaluations than did long-term subjects (t=3.31, p=.001). This difference indicates that the

more often subjects were evaluated the greater their concern with the outcome of that

evaluation. It is possible that since ST subjects received more evaluations that the outcome

of these evaluations was more salient. Furthermore, statement 9 indicated that more short-

term subjects than long-term subjects acted as they thought best, as opposed to how "Lee"


i









or other managers might act (t=2.59, p=.011). There is no apparent reason for this

difference. Finally, statement 11 revealed that the subjects were aware of the timing of

their performance evaluation. In response to the statement "My performance as Acting

Capital Procurement Manager was evaluated annually" the mean for short-term subjects

was significantly higher than the mean for long-term subjects (t=9.25, p=.000). These

responses provided evidence that the subjects were at least aware of when their

performance was evaluated. Although the reason for some differences between the two

groups isn't entirely explainable, it is encouraging that the two groups responded

differently to some measures. This provides evidence that in this study performance

evaluation differentially affected cognitive processes.


Incentive system


Subjects were not required to attend the drawing to win and fewer than ten students

attended. Furthermore, two students who won cash prizes did not claim the prizes.

Because subjects had remained anonymous it was not possible to contact the winners who

did not claim their prizes. The lack of interest in tickets and prizes was also evident in the

response to measures which had been taken to ensure that subjects internalized the reward

system. The post-experimental questionnaire included two statements, 15, "I wanted to

earn lottery tickets" and 16, "I believed that the better my decisions, the more lottery tickets

I would earn." The mean responses were 3.25 and 3.41 respectively on a scale of 1

(strongly disagree) to 5 (strongly agree). Only 46.8% agreed or strongly agreed with the

first statement and only 51.4% agreed or strongly agreed with the second. Finally, the

post-experimental questionnaire asked the subjects to estimate the tickets they expected to

earn. Only 38 (35%) of the subjects understood how the tickets were awarded. The

responses to the post-experimental questionnaire and the lack of participation in the final

drawing indicate a general failure of the reward system to motivate self-interested behavior.

This is addressed in the discussion of tests of Hypothesis 1.








Nuisance Variables


An examination of the correlations between possible nuisance variables (gender,

GPA, age, major, experience, experimental session, and educational level) and the

dependent and independent variables, did not reveal any significant correlations greater than

.30. Supplemental analysis including nuisance variables was not considered necessary.


Missing Values


Additional analyses were performed that excluded subjects who had weighted

benefits incorrectly (13 subjects), who had answered "very few benefits" (2 subjects), or

who had failed to respond on the fixed-list or the open-list (8 subjects). This exclusion

caused the main effect of the Frequency of Performance Evaluation on Personal Benefits

from Escalating open-list to become marginally significant (p=.077). It also changed the

statistical significance of the main effect of the Frequency of Performance Evaluation on

Personal Benefits from Escalating reduced fixed-list, to p=.109. The results of the

remainder of the analyses did not change from those found with the full sample.

A sample containing the twenty-three subjects whose values were excluded under

the missing values analyses was compared to a sample of all other subjects. The means of

each group for the Frequency of Performance Evaluation, GPA, age, major, experience,

experimental session, and academic level were compared in order to determine whether

there were any underlying differences between the two groups. The comparisons did not

reveal any such differences.


Scales


As discussed in Chapter 4, the perceived Benefits from Escalating and Switching

were elicited in two ways. First, subjects were asked to provide open-ended lists of

benefits within the four categories presented by the personal-organizational and escalating-









switching dimensions. They were also asked to assign 100 points among all four lists to

represent the extent to which they actually considered the benefits in making their choice.

Second, subjects were presented with two predetermined lists of benefits: benefits

from switching and benefits from escalating. Seven personal and seven organizational

benefits were mixed randomly within each list. Subjects were asked to rate the extent to

which they considered each of the benefits in their decision by circling the number on a

five-point scale (coded l=not at all important, to 5=extremely important). Scale

correlations for the two measures of perceived benefits are presented in Table 5-3.

Table 5-3
Scale Correlations

OBEO OBSO PBEO PBSO OBEF OBSF PBEF PBSF

OBEO 1.000 -.4913^ -.0717 -.5027* .4071* -.1490 -.0361 -.2206**

OBSO 1.000 -.6147* -.0574 -.3974* .4708* -.4716* .0357

PBEO 1.000 -.2246AA .2053** -.5455* .4421* -.2289**

PBSO 1.000 -.1835** .1975** .1736 .4735*

OBEF 1.000 .0530 .3514* .0181

OBSF 1.000 -.0672 .6155*

PBEF 1.000 .4006A

PBSF 1.000


* p <.01,1 -tailed significance
** p<.05, 1-tailed significance
A p<.01, 2-tailed significance
A^ p<.05, 2-tailed significance

Note:

PBEO = Personal Benefits from Escalating Open-List
PBSO = Personal Benefits from Switching Open-List
OBEO = Organizational Benefits from Escalating Open-List
OBSO = Organizational Benefits from Switching Open-List
PBEF = Personal Benefits from Escalating Fixed-List
PBSF = Personal Benefits from Switching Fixed-List
OBEF = Organizational Benefits from Escalating Fixed-List
OBSF = Organizational Benefits from Switching Fixed-List








Between-scale Convergent Validity


The between-scale correlations provide some information as to whether the two

scales measure the same constructs. All between-scale correlations of different measures of

the same constructs should be positive. For example, the open-list measure of Personal

Benefits from Escalating should be positively correlated with the fixed-list measure of

Personal Benefits from Escalating. The results of correlation analyses support these

predictions. The convergent validity correlations for the two measures of perceived

benefits (open and fixed lists) ranged from .4071 to .4735. All were significant at p < .01

indicating some convergent validity.


Reliability-fixed-list Scales


Each of the fixed-list scales is presented in Tables 5-4 through 5-7 with its

respective coefficient of reliability, measured by Cronbach's alpha [Cronbach 1951]. The

elimination of statements from scales measuring Organizational Benefits from Escalating,

Organizational Benefits from Switching, or Personal Benefits from Switching did not

improve the reported reliability. However, for the scale measuring Personal Benefits from

Escalating, elimination of all statements except 4 and 5, improved the reliability to an alpha

of .8812.2

The increase in reliability with the deletion of items 2, 6, 11, 12 and 14 points to a

problem with the scale's measurement. An examination of the Personal Benefits from

Escalating reveals that the scale, although measuring personal benefits, captures three

different constructs. Items 2, 4, 5 and 14 relate specifically to an individual's image:

2. I would appear to have confidence in my decisions
and to be standing up for what I believe in.



2Cronbach's alpha provides an estimate of how well the measures of a trait in a scale agree.
It represents "the average of all the possible split-half coefficients for a given test"
[Cronbach 1951, p. 300].








Table 5-4
Fixed-List Measuring
Organizational Benefits from Escalating (alpha=.7617)

1. Mototronics would avoid the risk of incurring
unexpected costs, unexpected production problems,
etc.

3. Mototronics would not have to retrain any workers.

7. Mototronics would not have to stop production.

8. Mototronics would continue to produce XD3 at or
under budget.

9. Mototronics would have no conversion costs.

10. Mototronics would not have an unbudgeted, extra
cash outlay for the conversion in year two.

13. Mototronics would not have to lay off any workers.


Table 5-5
Fixed-List Measuring
Personal Benefits from Escalating
(Alpha all items=.8250 Alpha items 4 and 5=.8812)


2. I would appear to have confidence in my decisions
and to be standing up for what I believe in.

4. I would not have to admit to the Board of Directors or
my boss that my original choice was not the best,
that I had made a "mistake."

5. I would not look like I couldn't make up my mind or
that I was uncertain of my decisions.

6. I would receive acceptable performance evaluations
and might become permanent Capital Procurement
Manager.

11. I would avoid possible questions about my ethics for
acting on "private" information.

12. I would avoid receiving an unacceptable performance
evaluation and possible demotion.

14. I would appear to know what I was doing and as
competent in my original decision process.








Table 5-6
Fixed-List Measuring
Organizational Benefits from Switching (alpha=.8311)

1. Mototronics' would save a lot of money and have
higher profit.

2. Mototronics would produce XD3 significantly below
budget.

3. Mototronics would get a jump on advanced
technology, it would be on the "cutting edge."

7. Mototronics might be able to use the cost savings for
other profitable ventures.

9. Mototronics would get a jump on competitors with
the new production techniques.

12. Mototronics would decrease overall cash outflows.


13. Mototronics would improve its ability to attract
outside investors due to the advanced technology and
higher profits.

Table 5-7
Fixed-List Measuring
Personal Benefits from Switching (alpha=.8311)

4. I would please the Board of Directors since they were
resistant to my original choice.

5. I would receive a better performance evaluation and
might become permanent Capital Procurement
Manager.

6. I would be perceived as having the confidence and
flexibility to adapt to new information and to change
to a better option.

8. I would look like a genius if it panned out and would
be considered a hero.

10. I would appear to be doing the best thing for the
company, and to be willing to put the company ahead
of myself.

11. My reputation would not be harmed if the savings
weren't recognized, because the beepers would still
be produced efficiently.

14. I would please the Production Managers due to the
decreased costs.








4. I would not have to admit to the Board of Directors or
my boss that my original choice was not the best,
that I had made a "mistake."

5. I would not look like I couldn't make up my mind or
that I was uncertain of my decisions.

14. I would appear to know what I was doing and as
competent in my original decision process.

Items 6 and 12 represent extrinsic benefits relating to performance:

6. I would receive acceptable performance evaluations
and might become permanent Capital Procurement
Manager.

12. I would avoid receiving an unacceptable performance
evaluation and possible demotion.

Item 11 reflects the ethical consideration underlying the decision:3

11. I would avoid possible questions about my ethics for
acting on "private" information.

In order to determine whether the scale does indeed represent these underlying

dimensions, factor analysis was performed.4 As three dimensions were predicted, three

factors were specified (Affifi and Clark 1984, p.334).5 The results of the factor analysis

are presented in Table 5-8. After five iterations, items 2, 4, 5, and 14 loaded highly on

3The concern with ethics is included as a separate dimension rather than with the image
dimension because it represents a different concept. The image dimension symbolizes how
a person appears to others. The ethical dimension reflects an innate moral standard. This
moral standard, or the inclination to make a moral choice, probably exists within some
subjects regardless of any experimental treatment. The propensity to consider one's image,
on the other hand, may be affected by external circumstances (in fact, it is this contention
upon which this study is based). Another possible distinction between the two dimensions
is that there is a difference between an image to others and an image to self. Statements 2,
4, 5, and 14 are associated with the external image while statement 11 is associated with the
internal. It is the image to others with which this study is concerned.

4The same analyses were not performed for the remaining three scales (Organizational
Benefits for Escalating, Personal Benefits for Switching, and Organizational Benefits for
Switching) because these scales were only used to test Hypothesis 4. Tests of Hypothesis
4 were not sensitive to the underlying dimensions of the scales. Instead they were sensitive
as to whether the scales measured benefits from switching or benefits from escalating.

5The alternative is to use the Kaiser criterion, choosing factors with eigenvalues greater
than 1. Using eigenvalues greater than one resulted in 2 factors. Items 2, 4, 5, and 14
loaded heavily on Factor 1. Items 6 and 12 loaded heavily on Factor 2. Item 11 loaded
heavier on Factor 2 (.51797) than Factor 1 (.11366).









Factor 1, 6 and 12 on Factor 2, and 11 on Factor 3. These loadings support the contention

that the fixed-list scale measuring Perceived Benefits from Escalating captures three

underlying constructs: image, economic and ethical.


Final Statistics:

Variable

PBEF2
PBEF4
PBEF5
PBEF6
PBEF11
PBEF12
PBEF14

Factor

1
2
3


Table 5-8
Factor Analysis for
Benefits from Escalating
Items 2,4,5,6,11,12,14


Communality

.68221
.76364
.82808
.85685
.94057
.89930
.77601

Eigenvalue

3.61349
1.20948
.92369


Pct of Var

51.6
17.3
13.2


Varimax Converged in 5 iterations.

Rotated Factor Matrix:

Factor 1 Factor 2

PBEF2 .70928 .18040
PBEF4 .85955 .15320
PBEF5 .88287 .22019
PBEF6 .37267 .84521
PBEF11 .09174 .12868
PBEF12 .08319 .93356
PBEF14 .85311 .14062


Kaiser Normalization.



Factor 3

.38286
-.03658
-.01174
.05990
.95687
.14439
.16861


The theory discussed in Chapter 3 explained that the impact of performance

evaluation occurs through its affect on an individual's need to present a positive image to

others. Accordingly, it is probable that the impact of performance evaluation will be most

apparent in the measurement of Personal Benefits from Escalating which reflects an


Cum Pct

51.6
68.9
82.1









individual's concern with his or her image. Because of this, and based on the results of the

reliability and factor analyses, hypotheses tests using Personal Benefits from Escalating

fixed-list, used the full list and a reduced measurement (called reduced fixed-list) that

included only items 2, 4, 5, and 14. The coefficient of reliability of this reduced scale is

alpha = .8756.


Reliability-aggression Ouestionnaire


The reliability coefficient of the Aggression Questionnaire was alpha=.6047. The

deletion of specific items did not result in improved reliability.


Reliability-efficacy Questionnaire


The reliability coefficient of the General Self-efficacy Questionnaire was

alpha=.7966. The deletion of items did not result in improved reliability.

Tests of Hypotheses


This section repeats each of the hypotheses developed in Chapter 3 and presents the

results of their tests. Since the model is based on the assumption that performance

evaluation will affect escalation behavior, tests were conducted to ascertain whether this

assumption held. Table 5-9 reports the switching behavior for long-term and short-term

subjects.
Table 5-9
Switching Behavior

ST LT

Periods 1 28 (50%) 33 (62%)
Until
Switching 2 6 (11%) 3 (6%)
3 22 (39%) 17 (32%)









Of the long-term subjects, 62% switched immediately compared to only 50% of the short-

term subjects. At the second opportunity to switch, 6% of the long-term subjects switched

and 11% of the short-term subjects switched. More short-term subjects never switched

(39%) than long-term subjects (32%). The results of ANOVA presented in Table 5-10

reveal that the Frequency of Evaluation is not significantly associated with Periods Until

Switching (F = 1.169, p = .141).6

Table 5-10
Examination of the Effect of the Frequency of Performance Evaluation
on Periods Until Switching



Source DF SS F p=

Main Effects

The Frequency
of Performance
Evaluation 1 1.033 1.169 .141*


Residual 107 94.527

Total 108 95.560

*1-tailed probability



However, eliminating the within cell variability by controlling for risk aversion

changes the results.7 As Table 5-11 reports, the results of ANCOVA with risk aversion as

61t is not clear that using ANOVA for this data is proper since the dependent variable is
ordinal with three levels: 1) switched in period 2; 2) switched in period 3; and 3) never
switched. However, since the first two levels are continuous ANOVA is appropriate for an
analysis using only these levels. If it can be established that the results of ANOVA using
levels 1 and 2 are similar to the results using all three levels, than the applicability of
ANOVA will be clearer. ANOVA using only levels 1 and 2 with Frequency of Evaluation
as the independent variable and Periods Until Switching as the dependent variable are
similar (F = 1.341, p =.126)

7The rationale underlying the use of risk aversion and concern with ethics as covariates is
described in the results of hypothesis tests.









a covariate are significant (F=8.263, p =.003). Similar results are illustrated in Table 5-12

using subjects' concerns with ethics as a covariate (F=7.809, p = .004).

Based upon these analyses it appears that the Frequency of Performance

Evaluation directly affects escalation behavior. This provides justification for exploring the

links between the Frequency of Performance Evaluation and escalation behavior through

the hypotheses developed in Chapter 3.

Table 5-11
Examination of the Effect of the Frequency of Performance Evaluation
on Periods Until Switching with Risk Aversion as a Covariate




Source D SS F

Main Effects

The Frequency
of Performance
Evaluation 1 .680 8.263 .003*


Covariate
Risk 1 1.028 12.492 .000

Residual 65 5.350

Total 67 7.059

* 1-tailed probability


Hypothesis 1


H1: Where there exists pressure to escalate commitment, the perceived
Personal Benefits from Escalating will be greater when an individual is
evaluated in the short run than when an individual is evaluated over the
long run.


Table 5-13 presents the results of the t-tests using the open-list

measure and the fixed-list measure of Personal Benefits from Escalating. They

do not support predictions.








Table 5-12
Examination of the Effect of the Frequency of Performance Evaluation
on Personal Benefits from Escalating with Ethical Concerns as a Covariate
Reduced Fixed-List



Source DF SS F p=

Main Effects

The Frequency
of Performance
Evaluation 1 .273 2.924 .046*


Covariate
Ethics 1 .728 7.809 .092

Residual 65 6.058

Total 67 7.059

*1-tailed probability


For the open-list measure, the long-term mean and the short-term mean are 22 and 26

respectively and the difference is not statistically significant (t = 1.02, p=.156). For the

fixed-list measures the long-term mean and the short-term means are 25 and 23

respectively, and the difference is not statistically significant (t = 1.02, p=.155). The

reduced fixed-list t-test results are marginally significant. The long-term mean is 12, and

the short-term mean is 13 (t = 1.036, p =.089).

Conceivably, the limited support for the first hypothesis could have been due to low

power of the test to reject the null hypothesis, lack of psychometric reliability, failure of the

experimental manipulations, a lack of subject motivation, difficulty with the experimental

procedure, subjects' focus on confounding variables, or the hypothesis.

Assuming a medium effect size, with samples ranging from 96 to 100 subjects and

a .05 significance level, the power for all tests was .79 or greater. Since this means that the









analyses had at least a 79% chance of rejecting the null hypothesis it is doubtful that lack of

power explains the insignificance of the tests.

The second possible explanation for the insignificant results was a lack of

psychometric reliability for the fixed-list scales. This is also a doubtful explanation since

the full fixed-list of seven Personal Benefits from Escalating was highly reliable (alpha =

.8250) Furthermore, reliability was higher when the scale was reduced to statements 2, 4,

5 and 14 (alpha = .8756). Factor analysis, specifying three factors, indicated that these

four statements loaded highly on one of the factors. This factor appeared to measure

benefits specifically related to one's image. Therefore, this reduced list of 4 statements,

along with the full fixed-list and the open-list, was used in the analyses. Because both lists

were highly reliable, it is doubtful that psychometric reliability was the cause of the

insignificant results.

A third possible reason for the lack of support for Hypothesis 1 could be a failure

of the manipulations. First, although statistically significant mean differences between the

post-experimental questionnaire responses for long-term and short-term subjects were

obtained for a number of questions, this may have indicated only that they were aware of

the timing of their performance evaluation. The subjects may not have internalized the

differences even though they were aware of them.

The lack of response to the incentive system is a fourth possible scapegoat.

Manipulation checks in Table 5-1 revealed that 77.1% of the subjects agreed or strongly

agreed with statement 4, "I cared what 'grade' I was to receive on my performance

evaluation." The manipulation checks also showed that 83.5% agreed or strongly agreed

with statement 9, "During the experiment I acted as I thought best, not as I thought "Lee"

or or other managers might act." Furthermore, 95.5% strongly agreed or agreed with

statement 10, "I wanted to perform well and make the best decisions I possibly could in

this experiment." These findings suggest that the subjects performed to the best of their

ability without a complete understanding of the incentive system. As subject 54 wrote, "I









Table 5-13
Results of t-tests
Hypothesis 1

t-test for: Personal Benefits from Escalating Open-list


Number
of
Cases


Short-term
Long-term


49
49*


Standard
Mean Deviation

26.266 17.552
22.469 19.410


Pooled Variance Estimate

t =1.02
df = 96
p =.156

T-test for: Personal Benefits from Escalating Fixed-list


Number
of
Cases


Short-term
Long-term


Standard
Mean Deviation

23.000 6.171
21.714 6.334


Pooled Variance Estimate


t =1.02
df = 97
p =.155

T-test for: Personal Benefits from Escalating Reduced Fixed-list


Number
of
Cases


Short-term
Long-term


Standard
Mean Deviation

13.255 4.088
12.082 4.532


Pooled Variance Estimate


t = 1.36
df = 98
p = .089

*In this and other hypotheses tests the total number of cases does not equal the
total sample size due to missing values.









wanted to make the best decisions possible. The lottery tickets are a nice incentive to

participate but they really didn't influence me during the experiment." Subjects seem to

have acted as though there was no external reward for their performance but nonetheless as

though they were motivated to perform during the experiment.

It is possible that the students may not have been equally motivated to participate.

MBA students, hereafter called Group 1, were "surprised" with the experiment during

regular class time. They were permitted to stop at any time, but since their professor

remained in the class during the experiment, some may have completed the experiment even

though they did not wish to do so. On the other hand, the remaining subjects (hereafter

called Group 2) were recruited during their classes and asked to participate in the

experiment at another location outside of regular class time. The majority of these students

received extra credit toward their grade for participation. Group 2 may have been more

motivated than Group 1 and this differential motivation may have influenced the lack of

results for the first hypothesis.

The results of t-tests for Hypothesis 1 revealed a difference between the groups.

The students who participated outside of class (Group 2) appeared to be sensitive to the

Frequency of Performance Evaluation for both the fixed-list and reduced-fixed-list

measures of Personal Benefits from Escalating. Those in the ST condition perceived more

Personal Benefits from Escalating (fixed-list 24.9, reduced fixed-list 14.6) than those in the

LT condition (fixed list-21.8, reduced fixed-list 11.9). The differences were significant

(fixed-list t = 1.62, p = .057, reduced fixed-list t = 2.04 p = .0245).8 On the other hand,

for Group 1, no statistically significant differences were observed between LT and ST

subjects with respect to the various measures of Personal Benefits from Escalating.


8The lack of significance using the open-list of Personal Benefits for Escalating may be due
to the relative insensitivity of this measure. For example, some subjects listed items in this
section which, if subjected to coding procedures, might be considered company benefits.
It is possible that this may have muddied the results for this measure of Personal Benefits
from Escalating.






72


The results of t-tests analyzing the relative motivation to participate in the

experiment are presented in Table 5-14. There were no statistically significant differences

Table 5-14
Mean Differences
Experiments During Class (Group 1) and Experiments Outside of Class (Group 2)

Mean Mean
Group 1 Group 2 t= D-
1. I felt pressure to "stick with" my
original decision. 2.97 3.16 -.79 .433

2. I felt secure in my job as Capital
Procurement Manager. 3.09 3.19 -.45 .657

3. The Board of Directors was
supportive of my recommendations. 2.48 2.40 .41 .680

4. I cared what "grade" I was to receive
on my performance evaluation. 3.92 4.00 -.38 .707

5. Poor outcomes from my
recommendations clearly would have
meant losing the job of Acting
Capital Procurement Manager. 4.00 3.88 .65 .519

6. There was much resistance to my
recommendations. 3.32 3.63 -1.88 .063

7. I needed to protect my position as
Acting Capital Procurement Manager
in the company. 3.60 3.69 -.49 .627

8. The Board of Directors was reluctant
to accept my recommendations. 3.55 3.79 -1.45 .150

9. During the experiment I acted as I
thought best, not as I thought "Lee"
or or other managers might act. 1.34 4.28 .30 .762

10. I wanted to perform well and make
the best decisions I possibly could
in this experiment. 4.55 4.53 .15 .443*

11. My performance as Acting Capital
Procurement Manager was evaluated
annually. 3.46 3.44 .08 .940








Table 5-14--continued

Mean Mean
Group 1 Group 2 t = LE

12. I looked back at my original list of
consequences even though the
instructions asked me not to do so
(Please answer honestly, there is no
penalty or reward for any answer). 1.42 1.30 .75 .457

13. I found the format of the experiment
(e.g., using the envelopes)
understandable. 4.08 4.28 -1.16 .249

14. In general, when I do a job, I am
more concerned about doing the job
"right" than with how it will affect me,
my bank account, my family, etc. 3.60 3.44 .65 .514

15. I wanted to earn lottery tickets. 3.11 3.49 -1.46 .074*

16. I believed that the better my decisions,
the more lottery tickets I would earn. 3.35 3.56 -.89 .187*

17. Gender 1.34 1.37 -.36 .723

18. GPA 3.50 3.18 4.95 .000

19. Experience 1.09 1.65 -7.50 .000

20. Fixed-list, Benefits from Escalating,
Statement 1: "Mototronics would
avoid the risk of incurring
unexpected costs, unexpected
production problems, etc." 3.48 3.83 -1.62 .108

21. Fixed-list, Benefits from Escalating,
Statement 11: "I would avoid
possible questions about my
ethics for acting on "private"
information." 2.57 3.37 -2.81 .006

*Because it was predicted that Group 2 was more motivated than Group 1, this p-value is
based upon 1-tailed probabilities.


between the groups' desires to perform well or to make the best decisions they possibly

could in the experiment. No differences were found with respect to caring what 'grade'

they received on their performance evaluation or with respect to beliefs that the better the

decisions, the more lottery tickets would be earned. A marginally significant difference (t =









-1.428, p = .074) was found in the desire to earn lottery tickets. Group 1 had less desire to

earn tickets than Group 2. Of Group 1, 41.5% agreed or strongly agreed that "I wanted to

earn lottery tickets" while 55.8% of Group 2 agreed or strongly agreed with the statement.

The different reactions of Group 1 and Group 2 to the Frequency of Performance

Evaluation could stem from other differences. The results of t-tests reported in Table 5-14

reveal that the groups differed on a number of measures. Group 2 perceived significantly

more problems with the use of "private" information (t = -2.81, p =.006) than Group 1.

Group 1 had higher GPA's (t = 8.53, p = .000) and more experience (t = -7.50, p = .000)

than Group 2.9

It is possible that Group 2's lack of experience may account for its differential

reaction to the Frequency of Performance Evaluation. On the other hand, this group also

appeared to be affected more by the manipulations, to be more sensitive to ethical issues,

and to be motivated more than Group 1. Additional research that equalizes the experimental

setting across all subjects is needed, especially in order to separate the effects of motivation

and experience.

A fifth possible explanation for the lack of support of Hypothesis 1 may have been

the complexity of the experimental procedure. However, 87.1% agreed or strongly agreed

that "I found the format of the experiment (e.g., using the envelopes) understandable."

An additional explanation for the lack of effect of the treatment is that subjects may

have been influenced by variables other than the Frequency of Performance Evaluation.

Two possible confounding variables were the perceived riskiness of the project and the

ethical considerations surrounding the project.

A review of subjects' open-list answers indicated that some subjects considered

the alternative project too risky. Because it was a new product which had not been


9The difference in GPA's is not necessarily informative since MBA students must maintain
a 3.0 GPA and undergraduate accounting students must maintain a 2.8 GPA. Therefore it
is not considered in the discussion.









introduced to the market, it appeared that some did not consider it a viable option.

Individuals who considered the project as a viable option may have been less risk-averse.

Therefore, it is possible that the variability associated with risk aversion may have affected

the results of tests of hypothesis 1. As this is a post-hoc consideration, measures of the

subjects' risk aversion were not available. However, the fixed-list did measure indirectly

an individual's consideration of the riskiness. The first statement on the Benefits from

Escalating fixed-list stated that "Mototronics would avoid the risk of incurring unexpected

costs, unexpected production problems, etc." Using this rough measure of risk-aversion

as a covariate and 1-tailed probabilities, ANCOVA (Table 5-15) revealed that



Table 5-15
Examination of the Effect of the Frequency of Performance Evaluation
on Personal Benefits from Escalating with Perceived Risk as a Covariate



Open-List



Source DF SS F p=

Main Effects

The Frequency
of Performance
Evaluation 1 652.391 1.989 .081*


Covariate
Risk 1 1388.861 4.235 .022

Residual 89 29186.574

Total 91 31227.826








Table 5-15--continued


Fixed-List


SS F


Source

Main Effects

The Frequency
of Performance
Evaluation


Covariate
Risk

Residual

Total


1 89.915 2.526


325.556

3417.438

3832.909


9.145


Reduced Fixed-List


Source

Main Effects

The Frequency
of Performance
Evaluation


Covariate
Risk

Residual

Total

* 1-tailed probability


1 62.093 3.614


SS F


.030*



.004


7.405


127.211

1666.457

1855.760


the Frequency of Performance Evaluation affected all three measures of Personal Benefits

from Escalating, at least marginally, (open-list F=1.989. p = .081. fixed-list F = 2.53, p =

.058. reduced fixed-list F = 3.614, p = .03). Apparently, reducing the within-cell


.058*



.003









variability by controlling for risk aversion allowed the affect of the Frequency of

Performance Evaluation to emerge.



Table 5-16
Examination of the Effect of the Frequency of Performance Evaluation
on Personal Benefits from Escalating with Ethical Concern as a Covariate


Open-List


SS F


Main Effects


The Frequency
of Performance
Evaluation


Covariate
Ethics

Residual

Total


1 486.996 1.855



1 7379.861 28.116


89 23360.969

91 31227.826


Reduced Fixed-List


Source


SS F


Main Effects


The Frequency
of Performance
Evaluation


Covariate
Risk

Residual


Total


1 38.469 2.176



1 102.276 5.785


97 1715.015

99 1855.760


* 1-tailed probability


.089*



.000


.072*



.009


Total









Many subjects considered that the private information about the alternative was

"inside" information, and questioned whether its use was legal or ethical. The fixed-list

measured an individual's consideration of the ethical nature of the choice. The eleventh

statement on the Benefits from Escalating fixed-list stated that "I would avoid possible

questions about my ethics for acting on 'private' information." The results of ANCOVA

using statement 11 as a covariate for the effects of the Frequency of Performance

Evaluation on the open-list and reduced fixed-list measures and 1-tailed probabilities are

presented in Table 5-16.10 These tests revealed that the Frequency of Performance

Evaluation marginally affected both measures of Personal Benefits from Escalating (open-

list F=1.855, p = .089, reduced fixed-list F = 2.176. p = .072). As with subjects' risk-

aversion, reducing the within-cell variability by controlling for concern with ethics allowed

the affect of the Frequency of Performance Evaluation to emerge.


Hypothesis 2


H2: Where there exists pressure to escalate commitment, there is an
interaction between the Frequency of Performance Evaluation and the
perceived Ability to Manipulate One's Image affecting Personal Benefits
from Escalating.

The interaction is such that the effect of the Frequency of Performance
Evaluation on Personal Benefits from Escalating becomes stronger as
the perceived Ability to Manipulate One's Image becomes greater.

Recall that the construct of the perceived Ability to Manipulate One's Image to others was

measured using two personality measures--Aggression and General Self-efficacy.

The results of the hierarchical regression with Aggression as the moderating

variable, and Personal Benefits from Escalating open-list as the independent variable, are




10Because statement 11 also was part of the fixed-list measure of Personal Benefits from
Escalating, it could not be used as a covariate for the effect of the Frequency of
Performance Evaluation on the full fixed-list.









presented in Table 5-17.11 As the significance test indicates, the addition of the interaction

between the Frequency of Performance Evaluation and Aggression does not add to the R2.

The results of the hierarchical regression analysis with Aggression as the moderating

variable and Personal Benefits from Escalating, fixed-list, as the independent variable are

presented in Table 5-18. As the significance test indicates, the addition of the interaction of

the Frequency of Performance Evaluation and Aggression does not add significantly to the

R2. Finally, the results for the same tests using Personal Benefits from Escalating reduced

fixed-list as a measure of Personal Benefits from Escalating also are insignificant. They are

presented in Table 5-19.

The results of the analyses using General Self-efficacy as a measure of the construct

of the perceived Ability to Manipulate One's Image to others resulted in similarly

insignificant results. These results are presented in Tables 5-20 through 5-22.


Hypothesis 3

H3: Where there exists pressure to escalate commitment, there is an
interaction between the Frequency of Performance Evaluation and an
individual's consideration of Doing the Job Right, affecting Personal
Benefits from Escalating.

The interaction is such that the effect of the Frequency of Performance
Evaluation on Personal Benefits from Escalating becomes stronger the
less the individual considers Doing the Job Right.

The results of the multiple regression analysis are presented in Tables 5-23 through 5-25.

As the significance tests indicate, the addition of the interaction between the Frequency of

Performance Evaluation and Doing the Job Right did not improve the equation's explained

variance.




11For H2 and H3, the results of the hierachical regression are presented first. This is
followed by significance tests which determine whether the addition of the interaction to the
regression equation which contained only the main effects, results in a significant increase
in R2.








Table 5-17
Results of Multiple Regression Analysis
Dependent Variable Personal Benefits from Escalating
Open-List


Variable Be Sig

Aggression .207222 2.076 .0406
Frequency -.100727 -1.009 .3155
(Constant) -.197 .8443

R2*=.03364




Variable Beta T SiT

AggXFreq 1.002741 1.081 .2826
Aggression -.103714 -.341 .7342
Frequency -1.051466 -1.188 .2380
(Constant) .945 .3470

R2=.03534


Significance Test

Independent Variables:

Aggression + Frequency
Aggression + Freq + AggXFreq


Incremental
R2 R2


.0336
.0353


.0017


n =98


0.05








Table 5-18
Results of Multiple Regression Analysis
Dependent Variable Personal Benefits from Escalating
Fixed-List


Variable Beta SigT

Aggression -.090335 -.893 .3738
Frequency -.102222 -1.011 .3145
(Constant) 5.065 .0000

R2=-.00161




Variable BetaT SiT

AggXFreq .952958 1.020 .3102
Aggression -.388959 -1.256 .2121
Frequency -.997695 -1.129 .2616
(Constant) 2.702 .0082

R2=-.00118


Significance Test

Independent Variables:

Aggression + Frequency
Aggression + Freq + AggXFreq


Incremental
R2 R2

-.0016
-.0012 .0004


n= 99


F


0.01








Table 5-19
Results of Multiple Regression Analysis
Dependent Variable Personal Benefits from Escalating
Reduced Fixed-List


Variable Beta I S

Aggression -.052495 -.523 .6024
Frequency -.136323 -1.357 .1779
(Constant) 4.142 .0001

R2=..00111




Variable Be S T

AggXFreq .454784 .492 .6241
Aggression -.195431 -.635 .5269
Frequency -.565672 -.643 .5215
(Constant) 1.892 .0615

R2=-.00676


Significance Test

Independent Variables:

Aggression + Frequency
Aggression + Freq + AggXFreq


Incremental
R2 R2

.0011
-.0068 -.0079


n =100


*The R2's used in the hierarchical regression analyses are adjusted R2's.


F


-0.24








Table 5-20
Results of Multiple Regression Analysis
Dependent Variable Personal Benefits from Escalating
Open-List


Variable Beta T Si T

Efficacy -.099247 -1.052 .2956
Frequency -.108487 -1.052 .3384
(Constant) 2.531 .0131

R2=.00086




Variable Beta T SigT

EffXFreq 1.934470 1.895 .0613
Efficacy -.669867 -2.107 .0378
Frequency -1.933575 -1.996 .0489
(Constant) 2.643 .0097

R2=.02823


Significance Test

Independent Variables:

Efficacy + Frequency
Efficacy + Freq + EffXFreq


Incremental
R2 R2

.0009
.0282 .0274


n= 95


F


0.83








Table 5-21
Results of Multiple Regression Analysis
Dependent Variable Personal Benefits from Escalating
Fixed-List


Variable BetSig

Efficacy -.105586 -1.029 .3064
Frequency -.110932 -1.081 .2827
(Constant) 4.912 .0000
R2=.00127




Variable Be T Si

EffXFreq -1.294459 -1.321 .1899
Efficacy .284813 .911 .3649
Frequency 1.135284 1.196 .2347
(Constant) .476 .6349
R2=.00920


Significance Test

Independent Variables:

Efficacy + Frequency
Efficacy + Freq + EffXFreq


Incremental
R2 R2


.0013
.0092


.0079


n =96


F


0.24








Table 5-22
Results of Multiple Regression Analysis
Dependent Variable Personal Benefits from Escalating
Reduced Fixed-List


Variable Beta T

Efficacy -.048664 -.476 .6353
Frequency -.132679 -1.297 .1976
(Constant) 3.750 .0003

R2=-.00152




Variable Beta T

EffXFreq -2.245310 -2.333 .0218
Efficacy .625217 2.046 .0436
Frequency 2.029316 2.177 .0320
(Constant) -.854 .3950

R2=.04368


Significance Test


Independent Variables:

Efficacy + Frequency
Efficacy + Freq + EffXFreq


Incremental
R2 R2

-.0015
.0437 .0452


n= 97


F


1.44








Table 5-23
Results of Multiple Regression Analysis
Dependent Variable Personal Benefits from Escalating
Open-List


Variable Beta T SigT

JOBRITE -.340264 -3.494 .0007
Frequency -.158776 -1.631 .1063
(Constant) 6.177 .0000

R2=. 10485


Variable Beta Sig

JOBRITExFreq -.196401 -.497 .6202
JOBRITE -.193719 -.624 .5342
Frequency -.010800 -.034 .9726
(Constant) 2.328 .0000

R2=.09770


Significance Test

Independent Variables:

JOBRITE + Frequency
JOBRITE + Freq + JOBRITExFreq


Incremental
R2 R2

.1049
.0977 -.0072


n =98


F


-0.22








Table 5-24
Results of Multiple Regression Analysis
Dependent Variable Personal Benefits from Escalating
Fixed-List


Variable Si T

JOBRITE -.437447 -4.756 .0000
Frequency -.155409 -1.690 .0944
(Constant) 12.800 .0000

R2=. 18264


variable Be T ig

JOBRITExFreq -.057104 -.153 .8784
JOBRITE -.395457 -1.369 .1743
Frequency -.114015 -.400 .6903
(Constant) 2.328 .0000

R2=. 17424


Sipgificance Test

Independent Variables:

JOBRITE + Frequency
JOBRITE + Freq + JOBRITExFreq


Incremental
R2 R2

.1826
.1742 -.0084


n= 99


F


-0.26








Table 5-25
Results of Multiple Regression Analysis
Dependent Variable Personal Benefits from Escalating
Reduced Fixed-List


Variable Beta T Si T

JOBRITE -.372245 -3.963 .0001
Frequency -.178496 -1.900 .0604
(Constant) 10.752 .0000

R2=.13789


Variable Beta T SigT

JOBRITExFreq -.181116 -.475 .6356
JOBRITE -.239634 -.814 .4178
Frequency -.047390 -.163 .8712
(Constant) 4.490 .0000

R2=.13096


Significance Test

Independent Variables:

JOBRITE + Frequency
JOBRITE + Freq + JOBRITExFreq

n= 100


Incremental
R2 R2

.1379
.1310 -.0069


Hypothesis 4a


H4a: Where there exists pressure to escalate commitment, there will be
two positive correlations: 1) between Personal Benefits from Escalating
and Periods Until Switching, and 2) between Organizational Benefits
from Escalating and Periods Until Switching.

The correlations between the number of Periods Until Switching and Benefits from

Escalating, and between the number of Periods Until Switching and Benefits from

Switching, are reported in Table 5-26. Personal and Organizational Benefits from

Escalating Open-list both are correlated significantly and positively with the number of

Periods Until Switching (Personal: r = .5827 p = .000, Organizational: r = .5160, p =


F


-0.22








Table 5-26
Hypothesis 4
Correlations Between Benefits and
Periods Until Switching


Panel A: Open-List


Personal
Benefits
From
Escalating


Periods
Until
Switching


.5827*


Panel B: Fixed List. all items

Personal
Benefits
From
Escalating


Periods
Until
Switching


.3372*


Organizational
Benefits
From
Escalating



.5160*


Organizational
Benefits
From
Escalating



.3898*


Organizational
Benefits
From
Switching



-.4108*


Organizational
Benefits
From
Switching



-.4148*


Personal
Benefits
From
Switching



-.6624*


Personal
Benefits
From
Switching



.2888*


Panel C: Fixed List. Reduced Items

Personal
Benefits
From
Escalating


Periods
Until
Switching

*p <.01


.2958*


.000). Results for the fixed-list also are significant and in the predicted direction (Personal:

r = .3372 p = .000, Organizational: r = .3898, p = .000). Finally, Personal Benefits from

Escalating, reduced fixed-list, and the number of Periods Until Switching are correlated

significantly and positively (r = .2958, p = .001).






90


Hypothesis 4b


H4b: Where there exists pressure to escalate commitment, there will be
two negative correlations: 1) between Personal Benefits from Switching
and Periods Until Switching, and 2) between Organizational Benefits
from Switching and Periods Until Switching.

These predictions were supported and their results are presented in Table 5-26. Personal

and Organizational Benefits from Switching open-list are correlated negatively and

significantly with the number of Periods Until Switching (Personal: r = -.6624 p = .000,

Organizational: r = -.4108, p = .000). The correlation between Personal Benefits from

Switching, fixed-list, and the number of Periods Until Switching is significant (r = -.2888,

p = .002) as is correlation between Organizational Benefits from Switching, fixed-list, and

the number of Periods Until Switching (r = -.4148, p = .000).


Exploratory Analysis of Hypothesis 1


A closer look at the four statements comprising the reduced fixed-list revealed that

although the statements are associated with the image of an individual, they actually

represent two types of benefits. Statements 4 and 5 represent avoiding embarrassment.

4. I would not have to admit to the Board of Directors or
my boss that my original choice was not the best,
that I had made a "mistake."

5. I would not look like I couldn't make up my mind or
that I was uncertain of my decisions.

Statements 2 and 14, on the other hand, represent receiving positive benefits affecting

one's image.

2. I would appear to have confidence in my decisions
and to be standing up for what I believe in.

14. I would appear to know what I was doing and as
competent in my original decision process.

The factor analysis reported in Table 5-28 is consistent with these propositions. The

addition of a fourth factor causes the "image" factor to divide into two factors: Factor 1

(statements 4 and 5) and Factor 3 (statements 2 and 14). Analyzing Hypothesis 1 with a















Final Statistics:

Variable

PBEF2
PBEF4
PBEF5
PBEF6
PBEF11
PBEF12
PBEF14

Factor

1
2
3
4


Table 5-28
Factor Analysis for
Benefits from Escalating
Items 2,4,5,6,11,12,14
with 4 Factors Specified


Communality

.90665
.91164
.87150
.89050
.99199
.91870
.80397

Eigenvalue

3.61349
1.20948
.92369
.54830


Pct of Var

51.6
17.3
13.2
7.8


Varimax Converged in 5 iterations.

Rotated Factor Matrix:

Factor 1 Factor 2

PBEF2 .27582 .17043
PBEF4 .92674 .14063
PBEF5 .84800 .20758
PBEF6 .21084 .84017
PBEF11 .05402 .12414
PBEF12 .13328 .93161
PBEF14 .59238 .12867


Kaiser Normalization.


Factor 3

.87618
.17192
.32969
.37413
.15040
-.01386
.65642


Factor 4

.18393
.05889
.02481
-.01361
.97521
.18124
.07495


scale of only statements 4 and 5 (alpha = .8812) resulted in a significant mean difference

between the mean Perceived Benefits from Escalating in the long-term (5.08) and the short-

term (6.21) conditions (t=2.36, p=.01). Tests of Hypothesis 1 with a scale of statements 2

and 14 revealed no mean differences (LT-7.00, ST-7.00, t = 0). This may explain why the

Frequency of Performance Evaluation did not affect the open-list, fixed-list or reduced

fixed-list measures of Personal Benefits from Escalating. These results are consistent with


Cum Pet

51.6
68.9
82.1
89.9









the view that the Frequency of Performance Evaluation affects a person's consideration of

the avoidance of negative consequences from switching.


Summary


The predictions for the effect of the Frequency of Performance Evaluation on

Personal Benefits from Escalating were not supported strongly. One measure of Personal

Benefits from Escalating, the reduced fixed-list, weakly supported this prediction. The

results were strengthened with supplementary analyses. Neither an individual's

consideration of doing the job "right" nor the perceived Ability to Manipulate One's Image

to others moderated the relationship between the Frequency of Performance Evaluation and

Personal Benefits from Escalating. The predicted relationships between Perceived

Benefits and the number of Periods Until Switching were supported strongly. Discussions

of these findings are presented in Chapter 6.




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 2.9.7 - mvs