Research & Practice: Commentary:

>> Pinter-friendly version

Jeffrey Pfeffer Testifies to Congress About 
Evidence-Based Practices

For
United States House of Representatives
Committee on Oversight and Government Reform
Subcommittee on
Federal Workforce, Postal Service, and the
District of Columbia

Hearing on the Status of Federal Personnel Reform
Washington, D.C.
March 8, 2007

My name is Jeffrey Pfeffer, Ph.D. I am a professor of organizational behavior and human resource management in the Graduate School of Business at Stanford University, and have served on the faculties of the University of Illinois and the University of California, Berkeley, as well as being a visiting professor at Harvard Business School, Singapore Management University, London Business School, and IESE in Spain. I have written 12 books and more than 100 articles and book chapters, and write a monthly management column for Business 2.0.

I am pleased to be able to offer my thoughts and evidence as the Federal Government thinks about how to manage its substantial civilian workforce to ensure even higher levels of performance and service. There is no doubt that people and how they are managed matter tremendously for organizational success -- as literally scores of studies show.¹ However, much of the conventional wisdom about and current practices in managing people are inconsistent with both theory and evidence about how to attain the best from a workforce.²

In this short statement, I want to make five points as succinctly as possible, providing references for background and documentation for my arguments. First, organizations in both the public and private sector ought to base policies not on casual benchmarking, on ideology or belief, on what they have done in the past or what they are comfortable with doing, but instead should implement evidence-based management. Second, the mere prevalence or persistence of some management practice is not evidence that it works -- there are numerous examples of widely diffused and quite persistent management practices, strongly advocated by practicing executives and consultants, where the systematic empirical evidence for their ineffectiveness is just overwhelming. Third, the idea that individual pay for performance will enhance organizational operations rests on a set of assumptions. Once those assumptions are spelled out and confronted with the evidence, it is clear that many -- maybe all -- do not hold in most organizations. Fourth, the evidence for the effectiveness of individual pay for performance is mixed, at best -- not because pay systems don't motivate behavior, but more frequently, because such systems effectively motivate the wrong behavior. And finally, the best way to encourage performance is to build a high performance culture. We know the components of such a system, and we ought to pay attention to this research and implement its findings.

Evidence-Based Management Practice

There is a small, but growing, movement to transition management practice from being premised on some mixture of ideology, casual benchmarking -- mindlessly copying what other, apparently successful organizations seem to be doing, what people have done in the past and think has worked, what people are skilled and comfortable at doing, and what they hope will work -- to management practice based on evidence, including qualitative and quantitative evidence, and evidence that is based on the best, most current, social science theories.³ Such a movement follows from and is inspired by the evidence-based medicine movement, something that has been demonstrated to lead to better practice, keeping up to date with the relevant advances in medical science, and to better patient outcomes.⁴ There are also evidence-based movements in the policy sciences, particularly in the United Kingdom where there has been a substantial commitment to basing policy on facts, as well as elsewhere. For instance, we recently learned of work in the New Zealand Ministry of Education on an Outcomes Evidence and Monitoring Project. And, there are important evidence-based movements in criminology and education, among other fields of practice. With the help of two Stanford business school librarians, Daphne Chang and Paul Reist, we have developed a website, www.evidence-basedmanagement.com, that collects a number of these resources as well as commentaries and case studies from individuals from a variety of fields trying to push evidence-based practice forward. The Academy of Management, the leading association of management professors, is embarking on creating a collaborative of scholars and practitioners interested in developing and diffusing evidence-based management and the association and its leadership appears committed to stimulating evidence-based practice in management. In short, there are many nascent movements to get evidence into practice. Much as in the case of evidence-based medicine, such efforts will undoubtedly take some time to bear fruit.

It is neither necessary nor useful to reprise all of the reasons why so little management practice is based on the evidence.⁵ Here I want to comment mostly on two substantial barriers that need to be acknowledged and surmounted. The first is the idea that somehow acting on the facts is more time-consuming and costly than relying on experience, intuition, and other non-evidence based approaches. Although I hear this argument all the time from my business colleagues, I find it completely unpersuasive. Many of the people who tell me that collecting data and running experiments take too long are the same people who spend millions of dollars on studies by consulting firms that are at least as expensive and time-consuming. Moreover, as we should have learned from the quality movement and from medical practice, preventing errors and difficulties, to the extent that is possible by acting on the best theory and evidence, is the most cost- and time-effective way of going about our business. It is almost always easier and less expensive to prevent problems than to fix them once they have occurred.

Second, because ideology and belief loom large, and because self-fulfilling prophecies are potent, care must be taken in gathering and analyzing data to ensure the integrity of the process. In medicine, the double-blind, placebo-controlled study is the gold standard for a good reason -- even in the case of physical symptoms and disease processes, we know all too well that we see what we expect, or hope, to see. Therefore, in medicine and other domains as well, to the extent possible, data gathering must be done in a way to ensure against simply confirming our biases, recognizing that in the world of the policy sciences, blinded studies are seldom possible.

One of my observations from years of experience is that consulting firms know the answers that their clients want, and will almost invariably give it to them. In that regard, care should be taken in doing evaluation studies to ensure the greatest level of objectivity and rigor or else such activities become simply exercises to provide the expected answers that sponsors have paid to hear.

Diffusion and Persistence Don't Provide Proof of Effectiveness

Many people will correctly note that individual pay for performance has increased in popularity, at least as measured by the extent of its implementation, not only in the United States but also in other countries as well. The adoption and persistence of some management practice, particularly by profit-seeking entities such as business firms, is often taken as prima facie proof that the practice must be effective. The argument is that if some management approach weren't effective, it would not be adopted in the first place or would be abandoned if it proved not to be effective.

However, this line of reasoning is completely fallacious. The study of management fads and fashion repeatedly illustrate the widespread adoption of techniques that have little or no value as well as the abandonment of management approaches such as total quality management that demonstrate effectiveness when interest wanes.⁶ The difficulty in learning from experience, when that experience has a systematic sampling bias -- for instance, oversampling on success and undersampling on failure -- has been explored in depth by my colleague Jerker Denrell in a series of insightful papers.⁷

Two concrete examples briefly illustrate the argument that diffusion and persistence say nothing about the effectiveness of some managerial intervention. Stock option plans and other schemes that encourage equity ownership on the part of top management have been widely adopted. Many companies have, as part of "good corporate governance," implemented required stock ownership guidelines for their senior managers. As this Congress certainly knows, any attempt to make such plans less financially attractive -- for instance, by requiring that stock options be expensed -- has been met with fierce, passionate opposition by executives who plead that such ownership incentives are crucial for creating a culture that encourages enhanced retention of key employees and improved organizational performance. Nonetheless, a meta-analysis of more than 200 studies of the effects of senior management equity ownership on company financial performance finds no effect.⁸ Thus, this is one example of a widely-adopted and vigorously defended management practice that has no evidence to support it.

Almost as near and dear to the hearts of senior private sector leaders is the idea of downsizing and the belief that U.S. companies are comparatively more competitive because they enjoy much greater flexibility than their counterparts in other countries in being able to adjust their workforce levels in response to changing economic and competitive conditions. As documented by Peter Cappelli⁹ and others, the rate of corporate restructurings has increased over the past several decades. However, once again, systematic reviews of the evidence find little positive effect from downsizings and, instead, much evidence of problems and various adverse consequences.¹⁰ There is, as reviewed by Wayne Cascio, no evidence that corporate downsizings increase productivity or stock price, reduce costs, stimulate innovation, or make organizations more successful. Nonetheless, such activities persist, providing yet another example of a widespread management practice that is apparently growing in prominence in spite of, not because of, the evidence for its effectiveness.

Adoption and persistence of some management practice should be taken as nothing other than evidence of what is currently in fashion or believed. Recall that at one time blood-letting was a widely adopted and advocated method of medical treatment, and a treatment that resulted in the death of George Washington.¹¹ In fact, the first examples of what has become evidence-based medicine were directed at evaluating this practice.

The Assumptions that Underlie Individual Pay for Performance

Every managerial intervention is premised on a set of assumptions about people and organizations. These assumptions are seldom specified or confronted with what we know, which is how so much bad management practice gets done. For individual pay for performance systems to produce the desired positive results, the following must be true:

Money is an important motivator. The reason why monetary incentives are recommended for solving almost every organizational problem is because we have come to believe that financial incentives are potent ways to drive behavior. There are two issues with this assumption. First, there are at least two effects of instituting some incentive program, be it based on financial rewards or some other type. The reward may increase or redirect motivation. But instituting a reward also provides information about what the organization's priorities are. People may respond to this informational effect in a desire to please their bosses and to do good work, regardless of whether or not they are motivated by the reward.

Second, as nicely documented in a series of studies by Chip Heath, many individuals believe that others are primarily motivated by money, even as they know that they themselves are motivated by other factors. So, for instance, a survey conducted by Kaplan Educational Centers of almost 500 people preparing to take the Law School Admissions Test (LSAT) found that 64% reported that they were thinking of pursuing a legal education because it was intellectually appealing or because they were interested in the law, but only 12% thought their peers were similarly motivated. Instead, 62% thought that others were interested in a legal career primarily for the financial rewards.¹² Heath has referred to this as an extrinsic incentives bias. Numerous surveys report that financial rewards are seldom listed as the most important to respondents. One might particularly question the relevance of financial remuneration compared to a public service motive in a work force that has chosen to work for the government instead, for instance, of going into investment banking.

Motivation is the issue for enhancing performance. Incentives affect motivation. Incentives have only a delayed effect, if any, through differential attraction of job candidates on the skills of the people doing the work, and no effect at all on the systems through which work gets done. That's why Deming and other management writers in the quality movement eschewed the use of individual financial incentives.¹³ Skill and training matter for performance, and increasing financial incentives can not substitute for providing development so people can do their work more effectively. Nor can financial incentives solve problems of how work is organized and led.

In his famous "red ball, white ball" demonstration, Deming would randomly select a participant from one of his lectures and have the person sit in front of a covered urn filled with red and white balls. Deming would tell the individual to pull only white balls out of the urn, since the reds were considered defects. When the person pulled a red ball, Deming would note that perhaps the incentives weren't sufficient, and would put money on the table and tell the person he or she could have it if they continued to draw white balls without drawing a red. When a red ball would invariably be drawn, Deming would then comment that perhaps punishment was necessary, and would smack the individual with a ruler the next time a red ball was drawn. The simple, but frequently, overlooked point is that systems produce performance, not individuals. The recent publicity about Toyota, which, by the way, has somehow succeeded in spite of not having individual pay for performance in its factories or large financial rewards for its executives, makes the point, yet again, about the wisdom of the quality movement with its emphasis on systems not people that is repeatedly forgotten in management practice, particularly in the U.S.

Individual performance can be reliably and unambiguously assessed. One of the reasons why pay for performance plans frequently fail is that they are seen as arbitrary and capricious. That's because few organizational work settings have the requirements argued my colleague, Edward Lazear, as being necessary to have individual financial rewards work effectively. In his classic study of piecework at Safelite Glass, Lazear noted that three conditions were present to make instituting a piecework incentive scheme so effective in the case of this installer of automobile windshields: 1) installing windshields was a job with almost no interdependence -- there was little need to learn from colleagues because the task was relatively straightforward and unchanging, and it was a job done by one person so that performance could be unambiguously attributed to that single individual; 2) quality could be measured -- did the windshield stay in or not? -- and again could be attributed to an individual; and 3) the company had already installed a measurement system that reliably tracked individual productivity, the use of materials, and defects.¹⁴

In most work situations, there is substantial interdependence, making the assignment of relative contributions necessarily complex and subjective. In most work situations, there are multidimensional aspects to performance, again making quantitatively assessing performance a difficult task. In most work situations, pay for performance involves scoring high on subjective ratings of work performance. But the evidence is persuasive that such performance evaluation ratings are flawed and imperfect measures of what employees actually do.

Consider the following simple illustration of the subjectivity of supervisory evaluations. Supervisors frequently provide evaluations for two types of employees: people who they have hired themselves since assuming their position as the head of some unit, and people who were already in that unit when the supervisor arrived or was promoted to the position of supervisor. In a study of a public sector workplace, research showed that, not surprisingly, supervisors gave higher performance ratings to people they had hired -- because, of course, they were more psychologically committed and attached to those individuals, having actively participated in choosing them. That study found that whether or not the supervisor had hired the individual being rated actually had a larger effect than objective performance on the rating that was awarded.¹⁵ Other forms of possible bias in subjective performance appraisals include the tendency to prefer people who are similar to us along dimensions of demography (race and gender) as well as attitudes, liking people who agree with us and don't challenge our decisions, and preferring people who provide "positive" news -- even if such good news turns out not to be accurate.

Although there are obviously exceptions, such as the installation of automobile windshields, in most jobs performance measurement is fraught with error and subjectivity. That's why pay-for-performance systems often devolve into pay for ingratiation with the boss systems -- and these are not the same thing.

Individual, rather than collective, rewards are important because of the need to overcome free-riding problems. Pay for performance, of course, need not be individual pay for performance. There are many group-based systems that provide more collective rewards when people work together to increase organizational capacity and outcomes. In the private sector, such systems include profit sharing and stock ownership. In the public sector and the private sector, there is gain-sharing, in which enhancements in efficiency get shared with those responsible for producing such gains. Because of the interdependence inherent in most complex organizations, where outcomes depend on the actions of many people acting in concert, collective rewards may be preferred because they don't encourage the internal competition of individual pay-for-performance and reinforce cooperative effort to make things better.

The common objection to such systems is that they are prone to free-riding problems -- that individuals can benefit from the efforts of their colleagues even if they don't work hard or effectively. There are two responses to this objection. The first is that systems that organize people into self-managed teams activate peer pressure, and that peer pressure is much more potent than hierarchical supervision. People will, and do, let down their bosses, but will seldom disappoint their peers. Second, although free-riding is a big deal in the economics literature, empirically, there is much less free-riding than theory predicts, except, of course, among people who have studied economics.¹⁶

Differentiation in individual rewards, a necessary and frequently explicit consequence of individual pay-for-performance systems, leads to higher unit performance. Yet another assumption of individual pay-for-performance systems is that by differentiating rewards among individuals, affirming the best and ignoring the rest, to use the phrase from the McKinsey War for Talent book, system performance is encouraged. That's because presumably those with the most skill and motivation receive outsize rewards and therefore are encouraged to continue their efforts, while those who don't perform at the same level don't get rewarded and, as a consequence, either leave or are motivated to try harder and do better.

Once again, the existing evidence is largely inconsistent with this idea. In even individualistic setting such as baseball teams¹⁷ and university faculties,¹⁸ the evidence is that the greater the dispersion in individual pay, the lower the performance. In less individualistic settings with even higher levels of task-related interdependence, such as companies, the evidence is that greater pay dispersion is also associated with diminished quality and financial performance.¹⁹

There is evidence that in sports that involve almost no interdependence, such as golf²⁰ and automobile racing (where there is interdependence within the crew but not between crews), higher differentiation in the prizes awarded to winners compared to the others does spur greater levels of performance. However, in the automobile racing case, the evidence seems to suggest that speeds go up but so do accidents.²¹

In short, individual pay-for-performance is premised on a set of assumptions which rest on very shaky logical and empirical foundations. It is, as a consequence, not surprising that such systems are fraught with problems, an issue to which I next turn.

The Evidence on the Effects of Individual Pay-for-Performance
is Mixed, at Best

Numerous surveys, including surveys conducted by the same compensation consulting firms that frequently advise on and advocate pay-for-performance systems, provide evidence of widespread dissatisfaction. For instance, a 2004 Watson Wyatt study of employee attitudes and opinions found that only 30% of U.S. workers believed that their company's performance management program did what it was intended to do -- improve performance. That's because fewer than 40% of people felt that the systems generated clear goals or provided honest feedback, while almost 40% believed their performance was inaccurately evaluated and about the same number said they did not understand the measures used to assess their performance.²² A 2004 Hewitt survey of some 350 companies reported that more than 80 percent of the organizations believed their pay-for-performance programs were at best partly successful or were not successful at all at accomplishing their goals.²³

Case examples provide detail behind these overall assessments. When TriQuint Semiconductor implemented pay-for-performance, the company confronted substantial pushback from managers and employees.²⁴ That's because people prefer less differentiated reward and recognition systems and understand the often arbitrary and capricious way in which pay-for-performance gets implemented. Moreover, the data suggests that few organizations -- and this is almost going to certainly be true of the Federal government -- adequately fund their plans to provide the resources necessary to award meaningful financial incentives. Thus, organizations first upset people through pay-for-performance programs and then don't fund the programs adequately to even sufficiently reward the people who have presumably been recognized as outstanding performers through this contentious process.

There is little evidence that things are getting better as more learning and experience with pay plans accumulates over time. I see nothing in the recent survey data to contradict the conclusion so aptly stated by the William Mercer company more than a decade ago: "most plans share two attributes: They absorb vast amounts of management time and resources, and they make everybody unhappy."²⁵

And when various financial incentive plans do "work," they may affect behavior, but not in ways that would generally be perceived as desirable. Steve Levitt and Brian Jacob have done research showing that the implementation of merit pay for teachers -- in which teachers in Chicago were rewarded when their students achieved higher test scores -- led to more cheating on the part of teachers, by giving students the answers or the test questions in advance. Jacob and Levitt concluded that such cheating behavior was quite sensitive to the size of the rewards offered.²⁶ Implementation of an incentive plan in Albuquerque, New Mexico, that rewarded garbage truck drivers for finishing their routes faster found that they did finish their routes faster -- by failing to pick up all the garbage, speeding (which caused accidents), and driving their trucks filled to over the legal weight limit.²⁷

The idea that financial incentives are the way to solve organizational performance and service problems is one of the most dangerous "half-truths" of management. The evidence for widespread dissatisfaction with such pay plans is pervasive. Both employee and company survey data suggest that the likelihood of success is low and the odds of problems and dissatisfaction are high.

Building High Performance Cultures

Tinkering with pay appears to be easier than fixing organizational cultures and leadership capabilities. It is apparently "fashionable" because it does not seem to require the systemic intervention along multiple dimensions implied in the idea of building high performance work arrangements. But there is no free lunch. Isolated, disconnected interventions often work at cross purposes with other aspects of management practice. And there is little evidence that isolated interventions can profoundly affect organizational performance. The large literature on high performance work systems tends to speak to the systemic, complementary nature of the various management practices required to provide an environment that produces innovation, discretionary effort, and high levels of performance and service. So, for instance, a prominent human resources text book has a chapter on consistency in management practices, noting that the whole can, and often is, more than the sum of its parts.²⁸

Although the list of high commitment or high performance work practices differs slightly among authors and studies, most such lists include: a) sustained investment in training and development, including job rotation, both formal and on-the-job training, and a tendency to promote from within as a consequence of the successful internal development of skill and people; b) an egalitarian culture in which formal status distinctions are downplayed, salary differences across levels are less than in the general economy, and in which people feel as if their contributions are important and valued; c) delegation of decision making responsibility so that skilled and developed people can actually use their gifts and skills to make real decisions; d) high pay to reduce turnover and attract the best people, coupled with rewards that share organizational success with its members; and e) employment security and a policy of mutual commitment, so that the workforce does not fear for the outcomes of events over which it has no control and instead, feels reciprocally committed to the employer.

It is important to note, however, that these management practices are not "techniques" to be imported. As David Russo, the former head of human resources at SAS Institute once commented, what matters is not what the company does -- its various programs and policies -- but the philosophy that provides foundation for all of its choices. That philosophy is one of valuing the workforce, treating it with respect and dignity, and making people as if their jobs are meaningful and their contributions important.

In that regard, government employees at all levels face a paradox. Even as there is emphasis on reinventing government and doing things to enhance employee performance, the message, delivered in numerous ways, is that the government work force is a "bureaucracy" and that, the smaller the better. The often disdainful way in which various agencies and their employees are spoken about makes instilling a culture of pride in performance difficult. The continuing efforts to outsource, downsize, and cut governmental agencies send not-so-subtle messages that the work, and the workforce, are not valued.

It is difficult to believe that a pay-for-performance system is going to be able to overcome the message, delivered both explicitly in what is said and implicitly in what is done, that the government work force and, by extension, their activities, are not really valued. It is particularly difficult to believe that a high performance culture will be produced by an intervention that has such a questionable set of assumptions and difficult history as individual pay-for-performance has. There is a large research literature on what it takes to build a high performance workplace. We should implement what we know, rather than what we hope, or wish, might be true.

NOTES

1. There are a number of summaries of the research linking people management practices to performance. Examples include Brian E. Becker and Mark A. Huselid, "High Performance Work Systems and Firm Performance: A Synthesis of Research and Managerial Implications," in G. R. Ferris (ed.), Research in Personnel and Human Resources Management, Vol. 16, Greenwich, CT: JAI Press, 1998, 53-101; Eileen Appelbaum, Thomas Bailey, Peter Berg, and Arne L. Kalleberg, Manufacturing Advantage: Why High Performance Work Systems Pay Off, Ithaca, NY: Cornell University Press, 2000; Brian E. Becker, Mark A. Huselid, and David Ulrich, The HR Scorecard: Linking People, Strategy, and Performance, Boston: Harvard Business School Press, 2001.

2. See, for instance, Jeffrey Pfeffer, What Were They Thinking? Unconventional Wisdom About Management, Boston: Harvard Business School Press, 2007 (July).

3. For a description of evidence-based management, see Jeffrey Pfeffer and Robert I. Sutton, Hard Facts, Dangerous Half-Truths, and Total Nonsense: Profiting from Evidence-Based Management, Boston: Harvard Business School Press, 2006.

4. E.g., David Sackett et al., Evidence-Based Medicine: How to Teach and Practice EBM, London: Wolfe, 2002.

5. This material is reviewed in Chapters 1 and 2 of Hard Facts, op. cit.

6. David Strang and Michael Macy, "In Search of Excellence: Fads, Success Stories, and Adaptive Emulation," American Journal of Sociology, 107 (2001), 147-182; Eric Abrahamson,"Management Fashion," Academy of Management Review, 21 (1996), 254-285; Eric Abrahamson and Gregory Fairchild, "Management Fashion: Life Cycle, Triggers, and Collective Learning Processes," Administrative Science Quarterly, 44 (1999), 708-740.

7. See, for example, Jerker Denrell, "Vicarious Learning, Undersampling of Failure, and the Myths of Management," Organization Science, 14 (2003), 227-243.

8. Dan R. Dalton, et al., "Meta-analyses of Financial Performance and Equity: Fusion and Confusion?" Academy of Management Journal, 46 (2003), 13-27.

9. Peter Cappelli, The New Deal at Work: Managing the Market-Driven Workforce, Boston: Harvard Business School Press, 1999.

10. Wayne F. Cascio, Responsible Restructuring: Creative and Profitable Alternatives to Layoffs, San Francisco: Berrett-Koehler, 2002.

11. Kevin Patterson, "What Doctors Don't Know (Almost Everything)," New York Times Magazine, May 5, 2002, 77.

12. Chip Heath, "On the Social Psychology of Agency Relationships: Lay Theories of Motivation Overemphasize Extrinsic Incentives," Organizational Behavior and Human Decision Processes, 78 (1999), 25-62.

13. Demings fourteen points, which begin with driving fear out of the organization and go on to critique the use of individual financial incentives to solve performance problems, are described in numerous books. One of the best is Andrea Gabor, The Man Who Discovered Quality, New York: Times Books, 1990.

14. Edward P. Lazear, "Performance Pay and Productivity," American Economic Review, 90 (2000).

15. F. D. Schoorman, "Escalation Bias in Performance Appraisals: An Unintended Consequence of Supervisor Participation in Hiring Decisions," Journal of Applied Psychology, 73 (1988), 58-62.

16. Gerald Marwell and Ruth E. Ames, "Economists Free Ride, Does Any Else? Experiments on the Provision of Public Goods," Journal of Public Economics, 15 (1981), 295-310.

17. Matt Bloom, "The Performance Effects of Pay Dispersion on Individuals and Organizations," Academy of Management Journal, 42 (1999), 25-40.

18. Jeffrey Pfeffer and Nancy Langton, "The Effect of Wage Dispersion on Satisfaction, Productivity, and Working Collaboratively: Evidence from College and University Faculty," Administrative Science Quarterly, 38 (1993), 382-407.

19. E.g., Phyllis A. Siegel and Donald C. Hambrick, "Pay Disparities within Top Management Groups: Evidence of Harmful Effects on Performance of High Technology Firms," Organization Science, 16 (2005), 259-274. D. M. Cowherd and D. I. Levine, "Product Quality and Pay Equity Between Lower Level Employees and Top Management: An Investigation of Distributive Justice Theory," Administrative Science Quarterly, 37 (1992), 302-320.

20. R. G. Ehrenberg and M. L. Bognanno, "The Incentive Effects of Tournaments Revisited: Evidence from the European PGA Tour," Industrial and Labor Relations Review, 43 (1990), 74S-88S.

21. Brian E. Becker and Mark A. Huselid, "The Incentive Effects of Tournament Compensation Systems," Administrative Science Quarterly, 17 (1992), 336-350.

22. "WorkUSA 2004: An Ongoing Study of Employee Attitudes and Opinions," www.watsonwyatt.com.

23. "Many Companies Fail to Achieve Success with Pay-for-Performance Programs," Hewitt & Associates News and Information, June 9, 2004.

24. "Communicating Beyond Ratings Can Be Difficult," Workforce Management, April 24, 2006, p. 35.

25. William Mercer, Leader to Leader, 1 (Winter, 1997), 61.

26. Brian Jacob and Steven D. Levitt, "Rotten Apples: An Investigation of the Prevalence and Predictors of Teaching Cheating," working Paper 9413, National Bureau of Economic Research, New York, 2002.

27. "Garbage Truck Drivers Rushing to Finish Work Are Safety Risk," Associated Press, January 30, 2004.

28. James N. Baron and David M. Kreps, Strategic Human Resources: Frameworks for General Managers. New York: Wiley, 1999, Ch. 3.