COMPUTER-AIDED PERFORMANCE MONITORING OF EMPLOYEES IN LARGE-VOLUME OFFICE OPERATIONS
by
John Chalykoff
AB, Boston College, 1976
MBA, University of Western Ontario, 1979
Submitted to the Alfred P. Sloan School of
Management in partial fulfillment of the
requirements for the degree of
DOCTOR OF PHILOSOPHY
at the
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
December 1987
@John Chalykoff, 1987
The author hereby grants to MIT permission to reproduce and
to distribute copies of this document in whole or in part
Signature of author
Signature redacted
Alfred P. Sloan School of Management,
December 20, 1987
Signature redacted
Certified by
Thomas A. Kochan, Thesis Supervisor
Certified by
Signature redacted
Robert B. McKersie, Thesis Supervisor
Signature redacted
Accepted by
Arnold Barnett
Chairman, Departmental Ph.D. Committee
Abstract
Computer-aided monitoring of employees in office settings has become
the subject of intense debate. The critics on the one hand, view it as
rife with abuse -- denigrating to employees and an invasion of privacy. Proponents on the other, view its use as essential to conducting business and largely beneficial to employees in terms of its capability to instruct
them in their work. Other than anecdotal evidence and a handful of
case studies, there is little data to support either claim.
This dissertation reports results of a study based on data collected in twenty computerized tax-collection offices of the Internal Revenue
Service (IRS). These tax-collection offices are collectively referred to
by the IRS as their Automated Collection System (ACS). The data were
collected in two stages. First, a total of ninety-one interviews were
conducted in five of these offices with employees, supervisors and
managers. Second, based largely on the interviews three survey
instruments were developed and administered in all twenty of these offices, one survey instrument for each of the above three groups.
The broad goals of the study are to assess the contributions and risks that information technology brings to office work and speak to the
larger policy issue at stake in debates over the use of the technology, as well as inform managerial practice in these settings.
The analysis suggests first, that computerization of office settings provides new mechanisms for control that serve to rationalize job
activities and intensify and expand performance assessment. The
immediate consequence is a perception of "close supervision." This is not only felt by non-managerial employees, but also by first-line
supervisors. Second, the extent to which the technology is viewed as a
useful control tool differs by level in the organization. Managers as a
group have a stronger orientation toward the technology as a means for control than either first-line supervisors or non-managerial employees, and supervisors perceive more control value in the technology than
employees do. Third, employees affective response to computer-aided
monitoring is largely predicted by supervisory approaches to the use of
performance information. Specifically, many of the factors previously
found to influence responses to performance appraisal sessions in
general, most notably the nature of feedback, are also predictive in this
context. Fourth, employee responses to computer-aided monitoring
exerts an influence on their general work satisfaction and turnover propensity.
Committee:
John Van Maanen, Professor of Organizational Psychology and Management N. Venkatraman, Assistant Professor of Management
Robert B. McKersie, Professor of Industrial Relations, Co-Chair Thomas A. Kochan, Professor of Industrial Relations, Co-Chair
DEDICATION
This dissertation is dedicated to my wife Gerry and daughter Leigh
without whose love this would not have been possible.
ACKNOWLEDGEMENTS
There are several people who contributed to what is good in this
dissertation.
The members of my committee -- Tom Kochan, Bob
McKersie, N. Venkatraman, and John Van Maanen -- are at the top of
the list. Tom and Bob saw me through four and a half years at Sloan.
What I believe to be of value as a teacher and researcher is in large
part a reflection of them.
If example is the best teacher, I have been
taught well.
Venkat, a comrade throughout the dissertation, was a
constant source of encouragement, and particularly helpful with the
statistical methodology.
John Van Maanen challenged every word I
wrote and some I didn't write.
Our frequent discussions broadened my
view of research.
Together, they provided an intellectual and personal
environment that I will miss.
Fellow doctoral students Joel Cutcher-Gershenfeld and Nitin Nohria
deserve special mention.
Joel, my office mate (the first half of the IR
team) was a constant source of inspiration. My late-night and
early-morning discussions with Nitin kept me sane.
There are others who I
have benefited from during the course of the dissertation -- Sumantra
Ghoshal, Tom Allen, John Carroll, Don Lessard, Bob Thomas, John-Paul MacDuffie, Paul Osterman, Lisa Lynch, and Aks Zaheer.
My thanks to the "Management In The 1990s" program which sponsored the research and to the people in the Internal Revenue Service who gave of their time and energy in making the research a reality. Those that were most instrumental were, Judith Tomaso, Ralph Reeder, John Wedick,
Bill Wauben, Henry Reches, and John Johnson.
The University of New Brunswick, as well as colleagues in the Division of Administration made every effort to help me complete this work. For this I am most grateful.
A special thanks to both my wife's family and my own who offered constant encouragement and support.
TABLE OF CONTENTS
Page
1. Theoretical Framework
7
1.1 The Controversy
8
1.2 Alternative Approaches to Conceptualizing
13
The Effects of Monitoring
1.3 Performance Appraisal/Feedback Literature
22
1.4 Chapter Overview
29
2. Research Design And Implementation
33
2.1 Research Sites
34
2.2 Developing the Questionnaires
37
2.3 Implementing the Questionnaires
39
2.4 Focus of Dissertation
42
3. Organizational Scope Of Technological Change
43
3.1 Introduction
44
3.2 ACS Technology
46
3.3 Work Structure 48
3.4 Control of Work Process
55
3.5 Office Climate
57
3.6 Performance Reviews 59
3.7 Supervisor and Employee Reactions to the Change 64
4. Computer-Aided Monitoring In ACS 74
4.1 Introduction
75
4.2 Perceptions and Practice 76
4.3 Orientations Toward the Use of Computer Monitoring 82
4.4 Knowledge of Computer-Aided Monitoring
87
5. Determinants Of Employee Satisfaction With
Computer-Aided Monitoring
103
5.1 Introduction 104
5.2 Hypotheses 106
5.3 Model Estimation 114
5.4 Support For Hypotheses 121
5.5 Discussion
126
6. Centrality Of Computer-Aided Monitoring To The Work Situation, And Influence On General Work Satisfaction
And Turnover 132
6.1 Introduction 133
6.2 Model and Method of Analysis 135
6.3 Development of Hypotheses 142
6.4 Model Estimation 148
6.5 Support for Hypotheses Relating to Centrality
of Computer-Aided Monitoring 155
6.6 A Test Of Discriminant Validity of the
Endogenous Constructs 161
6.7 Support for Hypotheses Relating to Turnover 162
7. Findings, Implications, And Conclusions 169
7.1 Principal Findings 170
7.2 Implications for Theory and Research 172
7.3 Managerial Implications 174
7.4 Implications for Public Policy 177
7.5 Conclusions 180 BIBLIOGRAPHY 182 APPENDIX 189 Interview Guides 190 Employee Questionnaire 194 Supervisor Questionnaire 205 Manager Questionnaire 216
CHAPTER 1
The Controversy
The introduction of information technology in office settings is said to provide managers with increased control over employees work behaviors and output (Nussbaum and DuRivage, 1986; Marx and Sherizen, 1986; 9 to 5, 1986; Manoocheheri, 1985; Westin et. al., 1985; Jacobs, 1984; Zuboff,
1982;). Indeed, Harley Shaiken (1987) warns that today's office technology provides a forum for unprecedented control of labor. He
notes that what is profoundly different about the new office technology, is the ability to monitor work as it takes place.
Integral to new office technology is the vast amount of information on employees' behaviors including their productivity that is provided to management. The flexibility of the software allows for "real time"
reports on a vast array of previously opaque aspects of work. Alan Westin (1985) lists some of the more general measures that are utilized
in these type of operations.
Computer technology can automatically record when operators sign on and off their terminals each work period; how long their transactions take (whether these are telephone calls, orders processed, sales made or claims adjusted); when operators leave their machines; how long the downtime is between transactions; how many transactions are completed each day; and similar aspects of operator performance
(ch 6, p 2).
Thus, the ability to monitor performance is considerable, limited only by management's control needs, desires, and intentions. The information
available can be as general or fine grained as desired.
For example,
software is available that can count individual operators' keystrokes per
second.
Moreover, in large-scale office operations the integration of telephone
technology with video display terminal (VDT) computer-based systems,
provides management with a means to monitor an employee's telephone
calls while simultaneously viewing a screen display of the account the
employee is working on.
This control mechanism facilitates
managements' monitoring of whether or not the operator was courteous
to the client, adhered to the company script or procedures, and
controlled the conversation.
The capability to simultaneously view the
case the employee is working on makes this form of computer-aided
monitoring
qualitatively
different than previous forms of telephone
monitoring where only the employee's interaction with the client was
assessed.
Currently, as many as 28 million Americans use computer terminals in
their jobs.
Although no precise data are available, it is estimated that
one third of them are computer monitored (Hickey, 1987).
Moreover, the
number of terminals in use is expected to substantially increase by 1990.
Given this prediction, the impact of computer monitoring on the
workplace should not be underestimated.
Recent writings have focused on the potential exploitation of clerical
employees occasioned by managerial use of the technology; specifically,
the setting of relentless standards of performance.
The ability to count keystrokes and time phone calls down to the second enables managers to increase production standards and establish piece rates, chillingly reminiscent of management practices in 19th century garment industry sweatshops (9 to 5,
1986).
Statements such as these, conjuring up images of a managerial
ideology associated with an early period in our industrialization, are put forth as descriptive of present day automated offices in America.
Questions concerning working conditions, that were recently thought to apply to only a small segments of American workers, if any, are now raised concerning increasing numbers of office workers in large and reputable companies.
An equally important issue associated with managerial use of new
office technology pertains to employee privacy. Under traditional
methods of supervision employees were likely to know when they were
being observed. Today, supervision can be carried out electronically,
from remote locations, and often without the employee's knowledge. In this regard, one of the more visib'e control mechanisms in electronic
offices with the potential to directly invade employee's privacy is
telephone monitoring. Not only is it possible for management to listen
in on both work related and personal calls, but also to conversations between co-workers (Nussbaum and duRivage, 1986). Thus, supervision can be present all of the time, blurring in these settings the distinction between what is work related and therefore might be reasonably
observed, and what is of a personal nature, where surveillance constitutes an invasion of privacy (Marx and Sherizen, 1986).
Are we indeed in the midst of a control revolution with the potential to permanently alter the very nature of work relations in office
settings? Has, as is claimed, managerial control of the labor process become ubiquitous, as it is now accomplished by tireless machines? Whether or not this potential describes its actual consequences is still
largely unknown due to a lack of research studies in actual office settings (Attewell, 1987; Time, 1986; Westin, 1985). However, the potential for abuse in the use of the technology has sparked a public policy debate over the issue of computer-aided monitoring.
Policy-level Initiatives
There are currently bills before both houses of congress and several
state legislatures pertaining to telephone monitoring (Daily Labor
Reporter, 1987).
One of these bills, sponsored by Representative Don
Edwards (D - California), would require employers to use an audible
"beep tone" during monitoring.
Representative Edwards argues that
telephone monitoring without the employee's or customer's knowledge
''erodes the employee's dignity and invades the privacy of the person on
the other end of the line."
Others have argued that its use is potentially much more sinister than
suggested above, and that what employers are really doing is monitoring
workers and not their work.
For example, the technology, they say, can
be programmed to listen in on employees and "detect a worker's religious
or political involvement, union activity, or personal problems" (Nussbaum
and duRivage, 1986).
On the other hand, employers argue that legislation that would, for example, require them to continuously signal that monitoring is being done would distract and confuse customers and increase stress for employees rather than decrease it. More specifically, they point out that monitoring is essential to doing business. In service industries where competitive advantage is achieved primarily by providing a superior level of service, employers view monitoring as part of a competitive edge. Monitoring employees' interactions with the
organization's clients, it is argued, helps to ensure customer satisfaction in their dealings with the organization.
Moreover, other virtues, beyond customer satisfaction, are alluded to.
It can be used to protect employees who are accused of
misrepresenting a product they are selling over the phone, for example. It can flag managers to particular problems that their telemarketers are having in their selling approach ....
And mainly it is used to instruct employees (Bradley, 1987).
Thus, there is a heated debate surrounding the use of computer-aided monitoring. The critics on the one hand, view it as rife with abuse --denigrating to employees and an invasion of privacy. Proponents on the other, view its use as essential to conducting business, and largely beneficial to employees in terms of its capability to instruct them in their work.
While the public policy debate has served to sensitize people to the potential of information technology to radically alter office work, our current inability to adequately inform this debate places in sharp relief our lack of theory and empirical knowledge pertaining to the impact of information technology in office settings.
Alternative Approaches To Conceptualizing The Effects Of Monitoring
Although monitoring has been the subject of intense debate, little
effort has been made to construct a theoretical framework for either assessing its actual role at the workplace or for grounding public policy. One conceptual approach favored by some organizational theorists has focused on the concept of control (Blau and Scott, 1962; Edwards, 1979;
Ouchi, 1977; Eisenhartd, 1986). However, the conceptual literature on
control has been developed based on previous generations of control
mechanisms embedded in physical technology. Information technology in
office settings adds new dimensions to past conceptualizations of
technological control. This technology is not only capable of initiating
and pacing work (Blau and Scott, 1962; Edwards, 1979), but also of monitoring and evaluating work.
Others have attempted to draw a distinction between an organization's output and/or behavioral control strategy (Ouchi and McGuire 1975;
Ouchi, 1977, Eisenhardt, 1986). With information technology it is
possible to intensify and expand behavioral control. Not only is it
possible to assess employees general work behaviors, but also behaviors
that in the past were assessed informally, if at all.
None of the previous efforts to conceptualize control strategies have developed testable propositions relating their actual use to organizational
or individual outcomes. Instead previous models have treated control
strategies as dependent variables to be explained by the nature of
technology or the organization's task environment. In contrast,
examining the effects of a control strategy requires a model that is capable of predicting how individuals respond to various control
strategies.
Thus, a new approach is needed that goes beyond existing literature on control to assess the actual impacts of computer-aided monitoring. One promising approach is to cast the study of computer-aided
monitoring within the framework of performance appraisal since this literature, a) has long debated the issue of assessing employee's
performance on the basis of outputs and/or behaviors, and b) is focused on the individual-level of analysis.
Moreover, research on performance appraisal offers a sharper analytical model because it suggests that a monitoring and evaluation process has both potential control and feedback characteristics, and that the response of employees will differ depending on whether control or
feedback dominate. That is, the more the process takes on feedback
characteristics, the more positive the employee's response is expected to be.
This dissertation will develop and test the power of performance
appraisal research to provide a theoretical framework in which to assess
the consequences of computer-aided monitoring. The essence of the
argument developed is that by adopting this approach we can: a) produce a more accurate conceptual or theoretical view of the potential
contributions and risks or abuses that this technology offers; b) develop an analytical strategy for examining the actual use of monitoring and its response from employees, supervisors, and managers; and c) provide data that can speak to the larger policy issues at stake in debates over the use of this technology.
To do this, data were collected through interviews and questionnaires from employees first-line supervisors and managers in computerized
offices. Chapter two describes the sample and the methodologies in
more detail. Chapters three and four examine employee supervisor and
manager responses to computer-aided monitoring, whi!e chapters five and six, grounded in the interviews and rooted in the performance appraisal
literature, test theoretical models focused on explaining the factors that influence employee responses to monitoring, as well as the influence of monitoring on other employee-level outcomes.
The remainder of this chapter is divided into three sections. Section
one explores the link between information technology and control. Section two examines completeness of control strategies occasioned by
the technology. Section three, rooted in the performance
appraisal/feedback literature, examines factors that are likely to influence employees' affective response to computer-aided monitoring, and presents in a preliminary fashion the theoretical model.
Information Technology and Control
Information technology provides different possibilities for control than
those presented by previous technology. As such, it does not conform
to the historical view of technological control. As is argued below, the
result of information technology is to make control more visible to employees rather than hidden.
There is universal agreement that an essential component of an organization's control system consists of the monitoring and evaluation of work; either work in process, or completed work (Merchant and
Simon, 1986; Horngren, 1982; Flanholtz and Das, 1985; Ouchi, 1977; Edwards, 1979). Moreover, it can be stated with little controversy, that
a) organization implies control (McMahon and Ivancevich, 1976), and b) that historically organizations have strived to maximize control over their operations while minimizing worker resistance to control (Yates,
1987).
Edwards (1979) providing an evolutionary control framework, argues that worker resistance has over time been minimized by depersonalizing
control, by making control a function of the organization's "technical"
1or "social" structure. Previous technology (such as that found in traditional manufacturing establishments) automated the production
process. It served a control function in as much as the initiating and
pacing of work was no longer the province of individual supervisors, but was subsumed in the structural properties of production -- in the speed of the "line". Thus, according to Edwards (1979), the exercise of control became less "visible" since it was tied to the logic of production and not a function of the whims of individual supervisors.
Supporting this argument, Blau and Scott (1962) reported that ". .. the
assembly line reduces the need for close supervision because the production process exerts some of the constraints on worker behavior
that otherwise would have to be exerted by foreman" (p 181). They
noted that employee-supervisor relations improved because the supervisor job consisted less of initiating and pacing work and more as a "helper" to employees on the line.
Information technology on the other hand is not only capable of
1. "Technical control" as used by Edwards resides or is embedded in the physical technology of the firm, and as such is distinct from the concept of "work technology" which is defined in terms of "work flow uncertainty, task routiness, task predictability, and task complexity" (Ovalle, 1984). The latter, "work technology" is taken in the managerialist literature to be largely determinant of the characteristics of a control system, while the former, "technical control" defines a control system.
initiating and pacing the work
2,
but also of monitoring and evaluating
employee work behaviors
3and output.
Moreover, there are two distinct
sets of employee behaviors that can be monitored.
First, in
computerized offices the technology is capable of monitoring behaviors
that pertain to compliance with the organization's rules, such as
attendance and punctuality (employee starting time, time taken for lunch
and breaks).
Second, and more importantly, the technology mediates the monitoring
and evaluation of employee behaviors considered to be crucial to
achieving "organizational goals" -- the quality and efficiency of employee
interactions with clients (following the prescribed procedures, controlling
the conversation, and accomplishing the interaction in a relatively short
period of time).
It is not only that the technology can monitor "work as it takes
place" (Shaiken, 1986), but it makes visible to management the
2
For example, the work to be done is stored within
the computer and the cases that employees work on
is determined by their priority order in the
computer queue.
Thus, the work is initiated by the
computer.
The computer can also be programmed to
pace the work by allowing a short period of time,
for example, eight seconds between cases.
3
The term behaviors, as used here refers to specific
employee actions that have a direct bearing on their
work, and not a more general set of behaviors or
behavioral characteristics that have at best a
tangential relation to the work process.
An
example of the former is adherence to work
procedures, while an example of the latter is dress
or general demeanor.
conversion process. That is, employee behaviors that were previously part of a "black box" inferred primarily from finished output if at all, are now assessed through computer-aided monitoring, and become integral to evaluation of what constitutes good or poor performance in these settings.
Control could remain less visible to the extent that it was subsumed in
the structural properties of the production process -- in the speed of
the line. However, information technology does not merely initiate and
pace work (which would allow it to be characterized as a purely structural element in the workplace), but it serves to monitor and evaluate employee work behaviors -- a characteristic that sets it apart from structure.
Indeed, this distinguishing characteristic of information technology --the capability to continually monitor behavior -- can make control in office operations somewhat analogous to a form of control that relies on
relentless monitoring by individual supervisors. (This form of control is
often associated with an earlier period in American industrialization, or thought to apply to only a small segment of present day work
establishments). Largely accomplished by machines rather than human
supervision, the raw exercise of power that often characterized this form
of control (Jacoby, 1985) is minimized. Nevertheless, from the
employee's perspective, the feeling can be one of being "closely
watched", under "constant surveillance." Rather than control becoming more "hidden", it has far more potential to become "visible" to employees
in computerized offices. Thus, individual consequences are likely to be far different than those spawned by earlier technology.
At its core the technology provides increased information on various
aspects of employee performance. In the past, some of these were not
directly assessed (for example, the quality of employees' interactions
with clients). A point that is implicit in the above discussion is that
the technology provides the means to shift the relative control emphasis in these offices from one that primarily monitored employees' completed work (output control), to one that also monitors employee behaviors
(behavioral control).
Output or Behavioral Control Strategies
Another control-theoretic approach attempts to draw a sharp
distinction between output and behavioral control strategies (Ouchi and
Mcguire,1975, Ouchi, 1977; Eisenhardt, 1986). According to this
perspective, behavioral control is the monitoring and evaluating of employees behaving in ways that managers agree they should behave in
order to produce the desired outcome. Output control is the monitoring
of whether or not the desired outcome was achieved.
Ouchi (1977) notes that behavioral control consists of observations of
employees by supervisors. Thus, in organizations that relied on
behavioral control there was a reliance on supervisors to gather and
first-line supervisors systematically observing and recording employee work behaviors is the exception rather than the rule (Baird, Beatty, and
Schneier, 1982). Thus, in a majority of work situations the actual
practice of behavioral control could be properly characterized as general and sporadic observations of employees by supervisors.
In computerized office settings the practice of behavioral control
becomes qualitatively different. Information technology makes it possible
to raise the pitch and refine the focus of behavioral control. For
example, general work behaviors monitored by supervisors in the past such as time taken for lunch and breaks are constantly monitored by the
computer. Moreover, as noted earlier, computer-aided monitoring makes
it possible to assess a range of behaviors that were previously hidden --behaviors inherent in the employee-client interaction.
It is arguable that at least some of these behaviors could be assessed
without the technology. The supervisor needed only to sit next to the employee and listen to his or her part of the conversation. However, how the interaction was handled in the supervisors presence may or may
not reflect the employee's general interactions with clients -- that is
the range of behaviors or actions taken on cases. This can now be
assessed.
Thus, information technology in large-volume office operations makes
behavioral control salient. It is qualitatively different than previous
attempts at behavioral control in both intensity and focus. It arguably
makes behavioral control more "complete." Therefore, we might expect dramatically different impacts and consequences than resulted from previous behavioral control strategies.
The impact of monitoring and evaluation strategies has been the
mainstay of the performance appraisal and feedback literature. That
literature has not only been concerned with situations in which either behavioral or output control dominate, but also their likely consequence
for employees. More importantly, this literature provides a theoretical
framework at the individual-level of analysis that allows for assessment of the effects of process (feedback characteristics) inherent in
performance evaluation. While it is recognized that computer-aided
monitoring is qualitatively different from past practice, this literature can nevertheless serve as a very useful starting point for examining the
individual-level consequences of information technology.
Performance Appraisal/Feedback Literature
In recent years there has been considerable activity centered on developing behaviorally based performance appraisal instruments.
Behaviorally Anchored Rating Scales (BARS) was first suggested in 1963 by Smith and Kendall. Other variations of this are: Behavioral
Observation Scales (BOS) by Latham and Wexley (1981); and Behavior
Summary Scales by Borman, Hough, and Dunnette (1976). While
based methods,
4the empirical evidence is far from conclusive (Dyer and
Schwab, 1982).
Moreover, the critics view behavioral-based assessments as less
desirable than assessments based on outputs because they require an
evaluator's judgement and are thus subject to personal bias and error
(Cummings and Schwab, 1973).
One the other hand, those who favor
behaviorally based measures argue, among other things, that they 1) hold
the employee accountable for what he actually does, 2) minimize
irrelevant factor outside of the individual's control, 3) facilitate
feedback and development, and 4) reduce role ambiguity by making
expected behaviors explicit (Latham and Wexley, 1981).
According to Wexley and Klimoski (1984), similar arguments
differentiate managerial preferences for one or the other of the above
methods.
In their own research, they found managers object to
behavioral measurement systems because they believe that it 'stifles
creativity,' forces them to 'supervise too closely,' and generally reminded
them of 'Theory X management'.
Those preferring behaviorally based
measures were likely to
"see
their managerial role as coach or helper"
that is, in training and developing employees.
4 Many performance appraisal instruments are based
on "traits". For example, "commitment, creativity,
loyalty, initiative" etc..
An advantage of trait
scales is that they can be developed easily, and can
be used across jobs.
Critics of this assessment
method argue that the words are too ambiguous, and
are therefore open to a variety of interpretations.
23
While managers have seemingly clear preferences in terms of the
assessment method, the literature strongly suggests that they also differ in terms of the use made of performance information (Deci, 1975;
McClelland, 1970; Hershey and Blanchard, 1977). The basic distinction
made by the above authors is between those mangers that seek to develop their employees and those that seek to control or subjugate
them. This distinction is particulary relevant to computerized office
settings where increased performance information in the hands of supervisors can potentially accentuate either predisposition.
Given the distinguishing characteristic of information technology --the possibility for a constant and specific form of behavioral
measurement, rather than the sporadic and largely general form of assessment that previously existed -- the direct implication is an increased focus on the measurement and evaluation of individual
performance. Therefore, performance assessment and feedback in these
settings is not a quarterly, or yearly occurrence, but rather, is
compressed into weekly or even daily time periods.
Viewed from this perspective, the functional or dysfunctional impacts of the use of information technology in monitoring and evaluating
employees is likely to be little different than the functional or
dysfunctional impacts of performance assessment and feedback in general. The main differences are that performance assessment and feedback
takes place much more frequently, and is focused on employee's behaviors -- their interactions with clients.
The technology is not benign. The potential to invade employee
privacy is real, as are potential approaches to its use that on the whole
disregard employee interests and welfare. However, the link between
potential and consequence is not predetermined. To date, the critics
have viewed its use as predetermined by its potential -- a potential
characterized as abusive. On the whole, the literature has failed to ask
whether or not its use gets transformed in such a way that is beneficial
for employees. This is largely a question of whether control or
feedback characteristics dominate the monitoring and evaluation process. It requires that we ask: what are the factors embedded in the use of the technology that determine its consequence (positive or negative) in
office settings?
While there is no empirical research in office settings that provides theoretical propositions to be tested, case studies that exist suggest that managerial orientations toward the use of the technology (Attewell,
1987), and supervisory practices embedded in its use (Westin, 1985) largely determine employee-level outcomes such as job satisfaction and
turnover (Nussbaum and duRivage, 1986).
Prescriptions for the effective conduct of Performance Appraisal
sessions rely heavily on the findings of the performance feedback
literature (Walther and Taylor, 1983, Reinhardt, 1985). This literature
has focused on the factors associated with recipients perception of
feedback, acceptance of feedback, and desire to respond to the feedback
(Ilgen et. al., 1979).
Summarizing the literature, Ilgen, et. al., suggest
three broad categories that influence the effectiveness of performance
feedback. These are, 1) message characteristics, 2)source characteristics,
and 3) recipient characteristics.
Message Characteristics
Factors that influence the perceived feedback message are 1) the
timing, 2) the sign, and 3) the frequency of feedback. In general,
positive reactions to performance feedback are correlated with feedback that is presented with little time delay, is positive in nature, and
frequent. There is a strong suspicion that the more frequent the
feedback the better, although this is expected to be moderated by the
recipient's task experience. People with relevant job knowledge are
likely to perceive frequent feedback as unnecessary and therefore be less
receptive to it. However, field studies suggest that this is not likely to
be an issue in most settings as performance feedback is seldom given
(Wexley, 1979; Taylor et. al., 1984).
Source Characteristics
When the primary source of performance feedback is the immediate
supervisor, the salient characteristics are the extent to which the employee perceives him/her as possessing the technical expertise to judge performance behaviors accurately, and the extent to which the
has been measured by asking employees direct questions pertaining to how much they trust their supervisor (Ilgen, et. al. 1981), or in
theoretical writings by some measure of the employees perception of the supervisor's leadership style, for example consideration behavior (Kerr
and Slocum, 1982).
Recipient Characteristics
Individuals' Locus of Control (Rotter, 1966), and age have been shown
to effect the recipients reception of feedback. Ilgen, suggests that the
finding with respect to age is most probably due to the positive
correlation between age and experience or job tenure. Thus, they
recommend that future studies should include some measure of experience or tenure.
In addition to the above factors, specific support for the feedback is
suggested as important. That is, specific critical incidents should be included in the feedback session so that the employee understands the basis for the assessment (Berger, 1983; Leskovec, 1967).
A schematic representation of these factors, as well as others
(discussed below) is presented in Figure 1. The primary outcomes of
interest are employees' satisfaction with computer-aided monitoring, their
general work satisfaction and their turnover. Following the performance
feedback literature these are presented as a function of message
characteristics, source characteristics, and recipient characteristics. As
FIGURE 1
FA CTORS INFLUENCING EMPLOYEE'S OUTCOME S
IN COMPUTERIZED OFFICES
INFLUENCING FACTORS
MESSAGE CHARACTERISTICS
-Immediacy of Feedback
-Frequency of Feedback
-Sign of Feedback
SOURCE CHARACTERISITCS
-Task Knowledge
-Consideration Behavior
RECIPIENT CHARACTERISTICS
-JobTenure
CLARITY OF RATING CRITERIA
BASIC ATTITUDES TOWARD
MONITORING
-Invasion of Privacy
-Shouldn't be Used
-Good if used properly
MONITORING METHOD
-Knowledge of when monitored
OTHER FACTORS
-Job Stress
-Pay
-Alternative Employment
EMPLOYEE-LEVEL
CONSEQUENCES
LII
SATISFACTION WITH
COMPUTER-AIDED
MONITORING
GENERAL WORK
SATISFACTION
TURNOVER
well, factors specific to computer-aided monitoring such as whether employees view it as an invasion of privacy, and the method of
monitoring (employees know or do not know when it is taking place) are
included. In addition, some factors that have been empirically found to
influence turnover in most work settings are included. These are pay,
job stress, and alternative employment opportunities.
The measurement, and motivation for all of these factors is presented in chapters five and six of this thesis where the formal models are
developed and tested. These chapters may be considered the heart of
the thesis. Their focus is on examining whether or not supervisory
approaches to computer-aided monitoring make a difference to
employee-level outcomes. However, the scope of the dissertation is slightly
broader. This can be seen from the following chapter outline.
Chapter Overview
In this chapter, I have presented a major focus of the thesis, which
is: to examine the extent to which supervisory approaches to the use of
computer-aided monitoring influence employee-level outcomes in office settings.
Chapter two is devoted to describing the research sites and the
methodologies employed in the research. Two separate methodologies
were used. To develop an understanding of the role and impact of
information technology in office settings I interviewed employees,
line supervisors, and managers in computerized offices. On the basis of these interviews, and prior literature, three questionnaire surveys (one for each of the groups mentioned above) were developed and
administered. Analyses throughout this thesis are rooted in both
methodologies.
In chapter three I examine, using both qualitative and quantitative data, employees' and supervisors' responses to these computerized offices in comparison to the offices that were replaced; that is the previous,
non-computerized work setting. The first section describes the physical
technology, and the changes in 1) work structure, 2) work process, 3) employees' interactions, and 4) performance reviews, that occurred in
conjunction with computerization. Section two, utilizing questionnaire
data, compares employees' and supervisors' reactions to the change. The basic result is that both groups report that they are more closely
supervised than they were when the technology was absent. However,
first-line supervisors are more inclined than employees to view the overall change as positive.
The analysis in chapter four begins to speaks to the policy-level
debate. Section one of that chapter explores the question of whether or
not it matters who controls the technology. Utilizing the interviews and
the questionnaires, employee, supervisory, and managerial control
orientations toward the technology are assessed. The results show that
managers as a group have stronger orientations toward technological control than do first-line supervisors, and that first-line supervisors
have a stronger control orientation than employees.
Section two of chapter four examines the following question: Do employees prefer to have knowledge of when they are actually being
monitored? The findings strongly indicate that in general employees
prefer not to know when they are being monitored. Reasons for this
are explored making use of written comments on the employee questionnaire.
The following two chapters are devoted to explicating the factors that
influence employee-level outcomes in computerized offices. In chapter
five, a model depicting the determinants of employee's satisfaction with computer-aided monitoring is developed and tested based on the
employee interviews and the previous literature pertaining to feedback
and performance appraisal. The results strongly support the proposition
that employee satisfaction with computer-aided monitoring is influenced by supervisory approaches to its use.
In chapter six two questions are examined. The first is: Does
employee satisfaction with computer-aided monitoring play a central role in explaining other outcomes such as employee general work satisfaction
and turnover propensity. The second is: What are the range of factors
that explain employee turnover in computerized offices. The results
indicate that employee satisfaction with computer-aided monitoring does influence these other outcomes, but that turnover is related to other factors not related to the use of the technology.
The seventh and final chapter summarizes the findings and draws the major theoretical, practical and policy-level implications of the study.
CHAPTER 2
RESEARCH DESIGN AND IMPLEMENTATION
Research Sites
The results presented in this thesis are based on research in 20
tax-collection offices of the United States Internal Revenue Service (IRS).
In
1983-84, the IRS completely automated its Collection Office Function
(COF), reducing the number of offices from 63 to 20.
These 20 tax
collection offices comprise the Automated Collection System (ACS) within
the Collection Division of the IRS.
Central to the work of these 20 offices is the collection of over-due
taxes and the tracking of individuals who have not filed a return.
They
are in essence a telephone operation, making outgoing calls and taking
incoming calls.
A detailed description of the work carried out in these
automated offices is provided in the next chapter.
Choice of IRS
This research was funded by the "Management in the 1990s," a five year research project based in the Sloan School of Management at the
Massachusetts Institute of Technology. The Internal Revenue Service was
one of ten organizations sponsoring the project. Thus, I was able to gain access to research sites that have been noted to be difficult to obtain
(Time, 1986). They assert that "corporations tend to be hesitant about describing the inner workings of their monitoring programs."
The advantage of these research sites is that all of the twenty offices have the same technology and carry out the same work. Thus,
differences in employee reactions to monitoring are not attributable to
the work done or differences in technology. The power of the data rests
on the ability to assess a variety of views on telephone monitoring, both
across and within levels in a single large organization. Often, senior
managers views can be compared directly with those of first-line
supervisors and non-managerial employees on the same set of questions. Therefore, this line of analysis will allow for a complete account of monitoring as it is viewed in a single organization.
While in a strict sense, this limits the generalizability of the results, external validity is rarely achieved in any one study (Kidder & Judd, 1986; p. 35). The ultimate test of external validity is replication in other
settings and with other samples. For any one study, generalization must
proceed on a theoretical basis. For the present research, that implies a
discussion of possible differences with other settings where computer-aided monitoring takes place.
For this research, the population of interest is automated,
large-volume office operations. They have important similarities. First the
work is relatively routine in nature. That is, even though the work is
"case" based, procedures for handling cases are highly prescribed. Second,
the technology used in these offices is fairly standard. Third
computer-aided monitoring (such as telephone monitoring) is not likely to be too
different across organizations. The phenomenon is the same. Indeed, the
work setting and telephone monitoring in the New York Telephone
Company (as described by Anger, 1970) was very similar to that in ACS.
to not only listen in on the call, but to simultaneously view on a video display terminal the case the employee is working on.
Differences among organizations will exist in terms of approaches to
the use of the technology for performance monitoring. That is with
regard to managerial practices embedded in the monitoring. This is the
principal focus of this thesis. Thus, to the extent that this research adds
insight into effective approaches to computer-aided monitoring, the results should be generalizable to these similar settings.
Research Scope
At the outset, the research in ACS was focused on examining the impact of information technology on the job of first-line supervisors. The general questions proposed were:
1. What are the supervisory skills needed to operate effectively in this
new context? How has the role of the first-line supervisor
changed?
2. To what extent are supervisory career ladders altered? Do they
expand or become more limited?
3. What are the factors that effect turnover of supervisors in automated offices?
4. What are the implications of information technology for the selection and training of supervisors?
On the basis of interviews conducted in ACS this initial focus was
broadened to include the impact of information technology on employees,
and managers in these offices. As will be clear below a tremendous
Only a limited amount could be included in the dissertation. Thus, the questions above that originally started the investigation are not formally assessed in the thesis.
Developing The Questionnaires
Initially, twenty unstructured interviews were carried out in the
Boston district of the IRS in order to gain a preliminary understanding of the general impact of the technology on supervision and work in ACS.
Based on these interviews ( each of one and a half to two hours duration), interviews guides were developed for use in four additional office locations. These were St. Louis, Chicago, Seattle and Laguna Niguel, (which is just outside of Los Angeles). These interview guides are shown in Appendix one. Over a period of sixteen non-consecutive days, four in each of these locations, interviews were conducted with twenty-four non-managerial employees, thirty-two first-line supervisors, and eleven middle managers in ACS. In addition, an interview was conducted with the head of the Collection Division in each district.
The employees and supervisors interviewed were picked by managers on the basis of two criteria. The first was that they had worked in the
Collection Office Function (the office which ACS replaced). This was done in order to understand the impact of the technology on the overall work situation. Approximately one-half of the interviews were with
personnel that had worked in COF. The second criteria, was to choose a mix of personnel in terms of their overall satisfaction with the work
situation. Based on the comments in the interviews, this criteria was also met.
After writing up and examining the interview notes it became clear that monitoring was an issue for all levels in the organization, not just or primarily for supervisors. Thus, a fuller understanding of automated offices could be gained by conducting three questionnaire surveys rather than just the one with supervisors that was originally planned. A
separate questionnaire survey was proposed for non-managerial employees, first-line supervisors, and middle managers in ACS. The middle managers
survey was immediately agreed to by the IRS. The employee survey was sanctioned by both the IRS and the National Treasury Employees Union (NTEU) which represents non-managerial employees in ACS.
The three questionnaire surveys are shown in Appendix One. These
questionnaires were developed almost entirely on the basis of the
interviews. Many of the questionnaire items are direct quotes from the interviews and are stated verbatim. Most of the quotes concerning telephone monitoring were placed on all three questionnaires to compare views of telephone monitoring between levels in the organization.
Pretests
Pretests of all three questionnaires were conducted in two separate offices. The objective of the pretesting was to make sure that all
questions were readily understandable and answerable. The pretests were
explained to the respondent. The respondent filled out the questionnaire and then each question was assessed in terms of the meaning of the
response given and the interpretation of the question. Respondents were
also asked if they felt the questionnaire covered the important aspects of
the work situation. Each of these pretest interviews took approximately
one and a half hours. Altogether twelve employees, twelve supervisors
and four middle managers participated in the pretests.
On the basis of these pretests several wording changes were made,
particularly to questions that asked for a rank ordering of items. In
addition a question was added to the first-line supervisor's questionnaire pertaining to the ability of supervisors and managers to work as a team. Also, it was revealed that newer employees, those that were still in
training, had difficulty answering questions that referred to the office as a whole and some that referred to the monitoring because they had not been there long enough to form clear opinions.
Implementation of the Questionnaire
To administer the questionnaire a contact person in each of the 20
offices was chosen by local management. The only requirement was that
the contact person be a non-managerial employee, i.e., part of the
bargaining unit. In four of the offices, managers were chosen as the
contact person. In telephone conversations with them, I insisted that
while they could administer the supervisory and managerial questionnaires, the employee questionnaire had to be administered by a bargaining unit employee.
Administering all three questionnaires consisted of distributing them to
the relevant personnel.
The questionnaires themselves were in postage
paid envelopes addressed to the Massachusetts Institute of Technology.
This was done in order to assure respondents complete confidentiality.
Upon completing the questionnaire the employee was to seal the envelope
and drop it in the mail.
There is every reason to believe that this
procedure was followed.
First, every envelope received was sealed.
Secondly, questionnaires from offices were received in a staggered
fashion, not in large bunches.
Receiving a large number of questionnaires
from any one office on the same day might indicate that the procedure
was not followed, i.e., that they were collected by the contact person and
then mailed.
Size Of The Samples
Altogether, 960 employees were surveyed.
At any given time, it is
difficult to know how many employees there are in ACS.
The number of
employees fluctuates with workload and the budgets of the individual
offices.
At the time of the survey there were anywhere from 1,576 to
1,970 employees in ACS.
This was arrived at in telephone conversations
to each office asking for the number of work groups that they had.
There were 197 work groups in total.
Each work group generally consists
of 8 to 10 employees.
Thus, as a rough estimate, anywhere from 48
percent to 61 percent of employees were surveyed.
there are three basic work functions 1) contact function, 2) research
function and, 3) investigation function.
Each of these work functions is
divided into work groups.
A typical office will have more contact units
than investigation or research units.
While these units do slightly
different tasks (described in chapter three) they are all subject to
monitoring.
Moreover, depending on workload in a particular function
some units shift their focus of work.
For example, in some of these
offices research units often do contact work for long periods of time.
The contact person in each office was instructed to give four
questionnaires to each research and investigation unit and six to each
contact unit in the office.
They were instructed to give them primarily
to employees who had finished their training.
It was decided to weight
the sample in favor of contact employees since they are generally more
heavily monitored.
Since there was approximately 197 first-line supervisors in ACS (one
for each work group) it was decided to survey all of them.
Similarly, all
60 middle managers were surveyed.
Response Rate
Of the 960 employee questionnaires, 740 were returned providing a
response rate of 77 percent.
This is fairly high for surveys of this kind.
While some people may have been cajoled into filling it out, there were
many comments on the questionnaires that showed employees found it
interesting.
Some employees stated that they looked forward to seeing
the results. Five of the 740 questionnaires were unusable in the analysis. Two of them had only two or three questions filled out, two employees commented that they were not able to fill out the questionnaire because they had not been in ACS long enough, and another employee, who had answered no questions, wrote that in his opinion the survey was
garbage".
One hundred and fifty-three, or 78 percent of the first-line
supervisor's questionnaires were returned. Similarly, a high percentage of
middle managers, 53 out of 60 or 88 percent, returned a completed questionnaire.
Focus Of The Dissertation
At the outset, this research was primarily focused on the impact of
information technology on supervisory jobs. The case studies and the
questionnaires allow us to examine not only the impact of information
technology on supervisors, but also non-managerial employees and
managers in automated offices. However, to make the thesis manageable
it was decided to concentrate primarily on the non-managerial employees. The supervisory and managerial surveys are used where appropriate to compare views on computer-aided monitoring between levels in the
organization. The other two surveys will be analyzed more formally in
CHAPTER 3
For example, it is well known that when computers are used to fragment and control clerical work, their arrival is almost always accompanied by reorganization that increases the pitch of
bureaucracy. Consequently, it is difficult to
separate the 'effect' of the machine from the 'effect' of organizational change. (Barley, 1987).
I ntroduction
In chapter one it was argued that information technology allows for a more focused form of behavioral control than exists in
non-automated offices or in most other work settings. Consequently there
can be a feeling of being "closely supervised," under "constant
surveillance." As the analysis in this chapter will show, this is
partly a result of the rationalization of work that can result from
information technology. Work organization becomes more highly
structured. What people are doing andwhen they should do it is not
only clearer, but also subject to direct and constant monitoring.
Moreover, the effect of this structure is not only felt by
employees, but also by first-line supervisors. What they have to do
and when they have to do it becomes more salient for this group.
Where in the past, a large part of their job consisted of initiating
and following the work (i.e., making sure that employees had enough cases to work on and that batches of cases were completed within a
reasonable period of time, this is now accomplished via the
technology. In automated offices first-line supervisor's jobs are