Few issues in the information system profession have generated as much controversy as computerized performance monitoring (CPM). Employers under increasing pressure to boost productivity have adopted these systems with the hope that the increased feedback and control they provide will lead to increased productivity. Despite the limited evidence concerning the impact of CPM on work quantity and quality (Grant and Higgins, 1991; Marx and Sherizen, 1986), many employers appear to be convinced of the benefits of this technology. According to one estimate, approximately 6 million workers in the United States have some or all of their work evaluated through computerized performance reports (Office of Technology Assessment [OTA], 1987). A more recent survey found that 73 out of 186 firms representing a variety of industries relied on some form of CPM (Rule and Brantley, 1992). Although there is need to better understand the effects of CPM on productivity, concerned scholars appear to be most interested in the harmful effects of CPM on monitored workers. Because of these presumed effects, many knowledgeable observers have raised questions about the ethics of CPM usage.

ETHICS AND COMPUTERIZED PERFORMANCE MONITORING

Even CPM's critics agree that this technology could be used humanely (Nussbaum and duRivage, 1986; Westin, 1986). This suggests that the choices employers make in how they use CPM ultimately determine the degree to which a particular system represents a humane or inhumane use of this technology. Many decisions must be made beyond the basic choice of whether or not to use a computerized performance monitor. What will the CPM measure? What information is to be reported and how will it be presented? Who will receive CPM output? What means are provided to employees to allow their input into the use of CPM data? These choices, or design alternatives, will largely determine the effects of a CPM on monitored workers (Marx and Sherizen, 1986). It would therefore appear to be more useful to examine the ethical considerations surrounding these more detailed design decisions rather than CPM in general. Before examining the potential problems surrounding various design alternatives, we first need to clarify the relationship between CPM and ethics. Ethical theory provides the basis for evaluating the morality of a given CPM. Two dominant ethical theories, Utilitarianism and Kantianism (Beauchamp and Bowie, 1993), both provide some insight into this issue.

Utilitarianism suggests that the morality of acts can be judged on the basis of their consequences. An action would be judged ethical to the extent that it results in the greatest good for all concerned parties. Business owners, employees, customers, and society in genera may all benefit from CPM. The profits returned to business owners could be increased if CPM improved the efficiency and quality of work performed by monitored employees. Employees could potentially benefit if such profitability resulted in increased wages or employment stability. Customers could benefit if monitoring brought about improvements in service/product quality, or lower prices. Finally, society in general could benefit from an increased efficiency and effectiveness of its workforce. These benefits, of course, need to be weighed against the potential harms resulting from CPM usage. The literature clearly presumes that monitored employees bear the brunt of CPM's harms. As noted earlier, research indicates that CPM causes stress and health problems, that it creates the potential for unfair performance evaluations, and that it is an invasion of privacy (OTA, 1987). Utilitarianism suggests that decisions surrounding CPM need to evaluate both the benefits and harms and choose the option that results in the greatest net good.

Kantian ethical theory suggests that the inherent features of an action mate it right or wrong. This perspective places importance on the intrinsic worth of individuals, emphasizing the obligation of business to respect the right of privacy, the dignity and the autonomy of their employees. For this reason, under the Kantian approach, monitoring systems would be judged to be unethical to the extent that implementing them breached these obligations. Consider, for example, a CPM system used to report on minute to minute actions of an employee. Whether or not the benefits to the employer of such a system outweighed the costs to the employee, a Kantian would reject the monitoring as morally objectionable if she concluded that it violates the employee's right of privacy, offends his dignity, or undermines his autonomy.

Reliance on CPM data for employee evaluation was measured using a procedure outlined by Hoffman (1960). Subjects were asked to estimate the weight their supervisors placed on CPM data during their performance evaluation relative to weights given other data sources (e.g. qualitative, non-computerized data). Scores on this measure can vary from 0 to 100 indicating a range of use from none to complete reliance on CPM data. Task coverage was measured using Grant and Higgins' (1991) instrument. This instrument asks subjects to indicate whether or not the computer keeps records on each of seven performance dimensions. Example dimensions are counting transactions, counting mistakes and recording terminal idle time. The task coverage score was determined by summing those dimensions for which the computer kept records. Three 5-point items were used to measure discussion of monitoring output (e.g. "to what extent do you have the ability to challenge the correctness of the monitoring information?", very little to very much). Frequency with which CPM output was received by supervisors and employees were each measured by a single item asking how often computer output which summarized the employee's performance was received by the supervisor and employee respectively. Each was measured on a five-point scale (never to daily).

Health problems was measured by presenting subjects with eight health complaints and asking them to place a check mark next to those problems they had experienced routinely during the past year. These items represent those complaints found by Smith et al. (1981) to be significantly different for monitored and non-monitored workers (e.g. headaches, stiff or sore wrists, and severe fatigue and exhaustion). A scale was constructed by adding up the number of checked health problems. Stress was measured using four 5-point items developed by Chalykoff and Kochan (1989). (e.g. "I experience tension on the job"--strongly disagree to strongly agree). Evaluation satisfaction was measured by using four 5-point items developed for this study. An example item is "the extent to which work quantity is only one of a well-rounded set of performance criteria"--very dissatisfied to very satisfied.

Measures of the following three non-CPM job characteristics were also included in the questionnaire; hours of computer use, task repetitiveness and task analyzability. Task analyzability refers to the degree to which there are well-know straightforward procedures that govern the performance of a task. Hours of computer use, task repetitiveness and task analyzability tend to increase the likelihood of monitoring (OTA, 1986). Because these job characteristics are also likely to affect stress and health problems, (Gardner et al., 1988). we need to control for them when examining the effects of CPM. Unless they are accounted for, the results will tend overstate the effects of CPM. Hours of computer use measured by asking subjects to estimate how many hours per day (on average) they worked on a computer terminal. Task repetitiveness and task analyzability were measured using an instrument developed by Withey et al. (1983). Three 5-point items measured task repetitiveness (e.g. "people in this unit do about the same job in the same way most of the time", strongly disagree or strongly agree). Three items measured task analyzability (e.g. "there is a clearly known way to do the major types of work I normally encounter", strongly disagree to strongly agree).