U.S. Office of Personnel Management Office of Merit Systems Oversight and Effectiveness Report of a Special Study POOR PERFORMERS IN GOVERNMENT: A Quest for the True Story January 1999 TABLE OF CONTENTS I. Executive summary . . . . . . . . . . . . . . . . . . . .1 II. Introduction. . . . . . . . . . . . . . . . . . . . . . .3 III. Employee sample . . . . . . . . . . . . . . . . . . . . .5 IV. Personnel action sample. . . . . . . . . . . . . . . . .19 V. The non-federal environment. . . . . . . . . . . . . . .27 VI. Conclusions. . . . . . . . . . . . . . . . . . . . . . .31 VII. Recommendations. . . . . . . . . . . . . . . . . . . . .35 I. EXECUTIVE SUMMARY As part of its Fiscal Year 1998 oversight program, the Office of Personnel Management (OPM) examined the pervasive suspicion that there are too many poor performers in the Federal government. To do this, we identified a random sample of employees and interviewed their supervisors about the performance of all employees in the work unit, about the causes of poor performance, what was done to address the problem, and what results had been achieved. In a separate interview sample, we contacted supervisors who had successfully taken a formal personnel action against a poor performer, to record their experiences and the lessons they learned. Finally, we looked at the private sector and the non-Federal public sector for points of comparison. Our key findings and conclusions are summarized as follows. * Our best estimate of the proportion of poor performers in the Federal workforce is 3.7 percent. While there are no good benchmarks in the private or non-Federal public sectors for comparing this finding, it is lower than the "common wisdom" would lead us to believe. We do not believe it reflects a serious performance problem in the Federal government. * As a whole, the supervisors of poor performers have not surrendered to cynicism or despair. Many report that they are actively pursuing a solution through formal and informal means. * Supervisors report success with rehabilitating a substantial number of workers who were poor performers at one time. * The legal protections available to employees in non-Federal public and private organizations are often similar to the Federal system, and the trend seems to be toward increasing these protections. Federal supervisors and managers may be yearning in vain for a dramatic easing of their burdens and responsibilities in this regard. * The workforce's misperception of the facts revealed by this study is substantial and could be diverting our attention from other important issues and from developing a realistic approach to addressing poor performance. * Supervisors who have pursued formal performance-based personnel actions describe the experience in intensely emotional terms. The effort they put forth to overcome real and perceived obstacles may be characterized as "heroic." Of particular concern is their frequent perception that top management did not welcome their efforts. * Supervisors typically receive a general orientation to poor performance regulations and procedures during their initial supervisory training. Supervisors are unanimous in saying this approach is nearly useless. They want and need "just in time" training and counseling, including ready resource materials. OPM's CD-ROM guide, Addressing and Resolving Poor Performance, is just what the doctor ordered. II. INTRODUCTION The truth is out there. - Slogan from the TV series, "The X-Files" There is a prevailing perception about public service employment: poor performance is a big problem. Moreover, civil service employees are among the first to speak up about the situation. In questionnaire after questionnaire, civil service employees express disdain for a management team that they say cannot or will not remove from their midst co-workers who are not carrying their share of the load. In a 1997 report entitled Adherence to the Merit Principles in the Workplace, the Merit System Protection Board reported that, among the 9,700 Federal employees it surveyed, the issue of handling poor performance was the deepest area of concern. Nearly half of the respondents said that agencies had a major problem correcting poor performance, and even more said the same thing about the firing of poor performers. {... the full text of the following paragraph includes a GRAPHIC which cannot be rendered in this ASCII format....} OPM's own Merit System Principles Questionnaire (MSPQ) consistently finds that most employees do not believe corrective action is taken when employees do not meet performance expectations. An even smaller percentage believe that poor performance is successfully corrected or that employees will be removed or leave the organization for inadequate performance. Civil service supervisors join this chorus of complaint. In a 1995 Merit Systems Protection Board report entitled Removing Poor Performers in the Federal Service, supervisors indicated that: they frequently receive inadequate or confusing advice and assistance on performance matters; the procedures for dealing with poor performance are too complicated, time consuming, or onerous; they do not get higher level management support; and they perceive their decisions will be reversed or that they will be falsely accused of discrimination in their actions. Many supervisors also report there is little negative cost to them for taking no action. The complaints are not new. In a 1978 report entitled A Management Concern: How to Deal with the Nonproductive Federal Employee, the General Accounting Office found that: Supervisors and managers perceive firing as a difficult chore which often lacks top-level management support. People at all levels fear reprisals from employees who may file adverse action appeals, discrimination complaints, and lawsuits...Removal procedures are complex, especially the detail and specificity required in stating reasons for removal; the process is also lengthy and time consuming. Supervisors and managers instead tend to use an informal system of working around, isolating, reassigning, sending to long-term training, or even promoting unsatisfactory employees. The Congress responded to these concerns by passing the Civil Service Reform Act of 1978, which instituted a new set of processes for setting performance expectations and for dealing with those who fail to meet those expectations. Twenty years later, the National Performance Review (now the National Partnership for Reinventing Government) surveyed the results of the Government's reformed processes and found them still sorely lacking. In the 1993 report Creating Government that Works Better and Costs Less, Vice President Gore called for initiatives to "reduce by half the time required to terminate Federal managers and employees for cause," to "develop a culture that supports supervisors' efforts to deal with poor performers," and to "improve supervisors' knowledge and skills in dealing with poor performers." While there can be no doubt that dealing with poor performance is an issue in the Federal civil service, there is an insidious corollary to this idea, an implication that is not often spoken openly but which is inescapable: there are too many poor performers in the Government. If you don't floss your teeth for twenty years, the tartar will build up. This unsettling notion that the Government is harboring an ever-increasing number of poor performers tarnishes the whole institution, fueling the negative attitudes of a public predisposed to criticize it. As part of its Fiscal Year 1998 oversight program, OPM examined the foundation for the suspicion that the Government has too many poor performers. To do this, we identified a random sample of employees and interviewed their supervisors. For those employees identified as "poor performers," we asked the supervisor what caused the employee's poor performance, what was done to address the problem, and what had been achieved as a result. In a separate interview sample, we contacted supervisors who had successfully taken a formal action to deal with a poor performer, to obtain descriptions of their experiences, and to record the lessons they learned. Finally, we looked at the private sector and the non-Federal public sector for points of comparison. The stated criticism that it is "too difficult" and the implied criticism there are "too many" suggest that these problems are less prevalent outside the Federal government. We wanted to test this hypothesis and find out if it might be true. III. EMPLOYEE SAMPLE By a small sample, we may judge of the whole piece. - Miguel de Cervantes METHODOLOGY In the past, OPM successfully conducted Governmentwide classification surveys using random sampling and telephone interviews. We decided that a similar random sample/ telephone interview methodology could be used to answer our question about "how many" poor performers there are. As a first step, we drew a random list of 200 employees from the Central Personnel Data File (CPDF). Data for September 1997 was used. We then contacted the servicing personnel office to locate the name and phone number of the listed employee's supervisor. However, when we finally contacted that supervisor, we completely disregarded the named employee and interviewed the supervisor about the performance of all the employees under his or her immediate supervision. It was the aggregate of all the employees in all these supervisors' work groups that became our random sample. This technique is called "cluster sampling," a generally accepted statistical sampling method that is used when a list of the entire population is not readily available or for administrative convenience. In cluster sampling, members of the population are not chosen directly. Rather, population members are sampled by randomly choosing easily identifiable groups (i.e., "clusters") of these members. In our methodology, the employees in a particular work unit were randomly identified by selecting the name of one unit member at random from CPDF. Estimates derived from a cluster sample are computed in special ways to reflect the indirect method of sampling. In general, cluster sampling introduces a greater degree of sampling variability than simple random sampling. In most cases, this increased variability, which would be reflected in wider confidence intervals, is not enough to prevent the use of cluster sampling. For example, in this study, the effect of cluster sampling was to add 0.2 percent to each end of our confidence interval for the percentage of poor performers. This method appealed to us because it allowed us to mitigate some of the obvious sensitivity and privacy issues (the employees in the "clusters" supervised were never identified by name) and because it produced a fairly large sample (a total of 3,114 employees were included in the units supervised by interviewees). We considered drawing a sample of civil service supervisors from CPDF and constructing our clusters more directly, but this method would not have allowed us to reach the military supervisors of civilian employees, a common situation in the Department of Defense. One challenge we faced in planning our interviews was to define the term "poor performer." Accepting the conventional wisdom that too few are identified through the official performance appraisal system, we felt it would be self-defeating to ask supervisors to use that system's definitions: to do so might be perceived as a challenge to a supervisor's integrity, or it might simply duplicate the results of the allegedly-flawed appraisal system. We wanted a broad, common-sense definition of poor performance, and settled on a three-category scheme: "good performer," "okay performer," and "poor performer." With the "good" and "okay" categories, we described a range extending from "outstanding" and "solid" to "meeting minimum performance objectives." The "poor performer" was described to the interviewee this way: These are employees with whom you are seriously disappointed. You have little confidence that they will do their jobs right. You often have to redo their work, or you may have had to severely modify their assignments to give them only work that they can do, which is much less than you would otherwise want them to do. They are just not pulling their weight. By orienting our interviewees toward these new definitions and descriptions of behavior, we hoped to focus the discussion on practical, "real world" concepts of worker performance. Our sole purpose was to identify as clearly as possible those employees whose performance falls below the line between the "okay" and "poor" categories. In developing our performance definitions and interview procedures, it became clear that, while employee performance was the subject of our interviews, the subject of our study was supervisory performance. If supervisors' perceptions or behaviors were a cause for the Government's alleged inability to deal with poor workers, we could not always rely on supervisors to accurately rate performance, even in our three common sense categories. To solve this problem, we developed a "bottom up" interview strategy. In our interview script, the interviewer reads the three category descriptions to the supervisor. Then, the interviewer says: "Now, put our discussion of the three categories aside for a minute. Let's think about your immediate workforce--just the ones who report directly to you. In your mind's eye, rank them from top to bottom in terms of their performance. OK, let's talk about the person whom you would rank at the bottom. Describe his/her performance in your own words." Our interviewers encouraged the supervisor to talk expansively about the employee's performance. Their goal was to get enough information to make an independent determination as to whether the employee at the bottom of the rank order was actually a "poor performer." The interviewer asked for and recorded examples of the employee's behavior. While the interviewer was to avoid prompting at first, the script suggested questions derived from the category definitions. "How much confidence do you have in this employee? How often do you have to re-do this employee's work? Have you severely restructured his/her assignments down to a much lower level than you have a right to expect? Have you ever considered trying to get rid of this employee? If you think about it, would you want to get rid of this employee if you could?" If the employee was deemed by the interviewer to be a "poor performer" (or if the interviewer was unable to make this determination on the spot), additional questions were asked about the employee's position and type of work, tenure with the Government and the supervisor, how the poor performance came about, efforts made to deal with the performance problem, and reasons why the efforts were unsuccessful. After completing the interview about the employee on the bottom of the supervisor's list, we then inquired about the next employee in the ranking and repeated this process until satisfied that there were no more poor performers to discuss. In theory, this approach enabled us to independently judge the performance of the employees in our sample. In practice, the process was collaborative. Most supervisors appeared quite capable and willing to distinguish the performance of their employees along the dimensions set out in our category definitions. In all cases, our interviewers recorded work examples and situation descriptions in sufficient detail to allow us to support and review a determination that an employee in the sample met our definition of a "poor performer." In all of the interviews, we went at least one step beyond the discussion of current poor performers and asked if the work unit included any "rehabilitated" poor performers. "Rehabilitated poor performer" is our term; we actually asked: "Let's talk for a minute about your Okay-Plus' performers. Are there any who used to be poor performers?" We then proceeded to ask questions similar to those that we asked about current poor performers: the circumstances causing the employee's poor performance; the corrective action attempted; and the results. However, in these cases, we were talking about success, and we were pleased to find a substantial occurrence of success in the work units sampled. If a supervisor had no or only a few current employees to discuss, we asked about "performers in days gone by," those who were either poor performers or those who had been "rehabilitated." Here, we departed from the rigors of predictive sampling in an effort to find out something about the dynamics of reality and perception. Do poor performers stay forever? Supervisors lament the effort required to deal with poor performers, but do they nevertheless put forth that effort? If so, what form did those efforts take, and what degree of success did they have? Finally, we asked the supervisors in our sample for their opinion on the usefulness of the regulations, tools, and systems available for dealing with poor performers and for any suggestions they might have on improvements. RESULTS General We conducted our interviews in March and April of 1998. A total of 200 supervisors were interviewed, who directly supervised 3,114 employees. Of those supervised, 429 were supervisors themselves (14 percent of the total). Twelve of the supervisors interviewed were in the military (6 percent). The supervisors interviewed had an average of 22 years of Federal civilian or military service and an average of 12 years experience in supervising Federal civilian employees. Most of the supervisors reported that they had received training in improving employee performance and in taking corrective personnel actions. However, many noted that the training was not timely they did not get it when they really needed it and that it was usually only a part of a broader supervisory training package. Several said that the training was too abstract and did not deal with such realities as health, reasonable accommodation, discipline, leave, and family-friendly issues. Further, it did not provide the step-by-step tools needed to monitor and counsel employees on their performance, make appropriate documentation, and effect the proper formal personnel actions. As one supervisor put it, "Management courses do not adequately address these technical aspects; instead, they deal with the warm and fuzzy' topics like leadership styles, and seem to think that everything else will take care of itself." Table 1. The Supervisors Interviewed Number interviewed 200 Civilian 188 Military 12 Average years of Federal civilian/military service 22 Average years supervising Federal civilian employees 12 Number of employees directly supervised 3,114 Average number of employees supervised 16 Employees directly supervised who are supervisors 429 Received training on how to improve performance 9.5% Received training on taking corrective personnel actions 6.5% Current Poor Performers How Many Are There? {... the full text of the following paragraph includes a GRAPHIC which cannot be rendered in this ASCII format....} Based upon the work done in this study, we estimate the number of poor performers in the Federal government is 3.7 percent. From a statistical standpoint, we are 95 percent confident that the percentage of poor performers is no lower than 2.8 percent and no larger than 4.6 percent. The average length of service in the poor performer group was 14 years. Less than 1 percent were reported to be serving either a probationary or trial period; 10 percent were supervisors. What Was Their Problem? In describing the poor performance of these employees, the supervisors gave these examples: low productivity; poor quality/frequent errors; incomplete work products; failure to meet deadlines; backlogged work; customer complaints of inadequate or discourteous service; and failure to keep the supervisor informed of work progress and problems. However, when the supervisors discussed the background to these situations, several common trends emerged concerning other issues associated with poor performance. The most prevalent were health, attendance and leave, and behavioral or conduct issues. One supervisor illustrated these mixed-issue situations very well by describing an employee who was reassigned to his work unit through a reorganization. The employee had a chronic illness and was often absent from work. This made learning the new job difficult and resulted in numerous errors and incomplete work. This also placed strain on the supervisor to monitor the work more closely and on other employees to make the corrections or handle additional workload. Whenever the supervisor counseled the employee on her performance, she would then be absent for several days and claiming that it was due to illness. There were also instances where the employee's actual work was acceptable, but there were overriding conduct issues that caused the supervisor to consider this employee a poor performer. These situations involved such things as being absent from the work site without authorization, substance abuse, disruptive or confrontational behavior with co-workers, or various illegal activities. The supervisors said that dealing effectively with these mixed situations was often challenging because the rules and procedures did not fit neatly. Table 2 shows the distribution of problems related to poor performance. Table 2. Poor Performance Estimate Current Employees Governmentwide 95% Confidence Issue Estimate Interval --------------------------- ------------ ------------ All poor performers 3.7% 2.8%--4.6% Performance only 1.8% 1.2%--2.5% Performance and attendance 0.8% * Performance & conduct 0.6% * Performance and health 0.4% * Conduct only 0.1% * * Due to the nature of the variance estimation in cluster samples, accurate confidence interval estimates are not available for very small percentages. How Did the Performance Problem Come About? Supervisors were asked how the employee's poor performance came about. Most supervisors told us that the employee came under his/her supervision with performance problems. One supervisor who was new to an organization described three of her employees as not performing work of the quality and quantity expected at their grade level. Her efforts to address these failings were hampered when she found out the employees had received excellent performance ratings over the past ten years under previous supervisors. Table 3 lists the factors to which supervisors attributed the poor performance. Table 3. Origin of Poor Performance Current Employees Percent in Reason Reported Group ---------------------------------------- ----------- Employee came as a poor performer 60.8% Employee developed a poor attitude or work habits 53.3% Wrong match of the person and the job 25.0% Employee has health problems 18.3% Employee's position or job assignment changed 13.3% Other 9.2% Note: More than one option may apply to an individual employee. What Have You Tried To Do About It, and How Did It Work? The performance management regulations state that employees should be provided assistance whenever performance is below the fully successful level and that appropriate action should be taken when performance is unacceptable (5 CFR 430.207). Several options are available to supervisors seeking to improve employee performance or in proceeding with a performance-based personnel action. Our interviews revealed that the more formal options, such as rating the employee below fully successful, developing a performance improvement plan, or initiating a performance-based action were rarely used. While supervisors reported using personal counseling much more than other options, it was rated as only "somewhat useful" in just over one-third of the cases in our sample. When supervisors reported that they had taken no formal action, they cited as reasons: lack of higher management support, dislike for a tedious and time-consuming process, and fear of grievances or other actions. One supervisor said that, within his agency, the culture is not to "rock the boat" by giving a less-than-fully- successful performance rating, let alone proposing to demote or terminate an employee for poor performance. Others stated that the unwritten policy was to avoid any situation that could lead to an appeal or law suit. Table 4 provides a summary of the interventions supervisors used and how they viewed the usefulness of these options when they were used. Table 4. Usefulness of Performance Interventions Current Poor Performers Some- No what Very Performance Intervention Used Effect Useful Useful ---------------------------------- ---- ----- ------ ----- Provided personal guidance 85.0% 61.8% 38.2% 0 Provided additional training or education 40.8% 69.4% 30.6% 0 Changed work assignments 37.5% 71.1% 26.7% 2.2% Assigned mentors or advisors 29.2% 57.1% 42.9% 0 Developed a performance improvement plan 15.8% 68.4% 26.3% 5.3% Rated the employee less than fully successful 14.2% 88.2% 11.8% 0 Initiated action to reassign, demote, or remove 7.5% 77.8% 11.1% 11.1% What Kind of Help Is Available to You? Supervisors were also asked about the usefulness of certain resources when dealing with poor performers. Only half of the supervisors felt that the current performance management regulations, policy, or guidance were helpful to them, and only 5 percent rated these resources as "very helpful." The same pattern was found with the reported usefulness of performance standards. Examples were given where performance standards were written so poorly that it was almost impossible to use them effectively. Some of these standards were imposed nationwide or had been negotiated with the union and thus could not be altered by the immediate supervisor. Several supervisors indicated their agency had recently gone to a pass/fail rating system. Many of them felt that the criteria for receiving a passing rating under these new systems were too broad to support a formal action. The personnel office staff were rated as "very helpful" by a little over a third of the supervisors. Another third rated this resource as "somewhat helpful." Consultation with other supervisors was used by more than half of these supervisors, and this was seen as useful by most of them. Table 5. Value of Resources to Supervisors in Dealing with Current Poor Performers Some- Not what Very Resource Used Helpful Helpful Helpful ------------------------------- ---- ----- ----- ------- Servicing personnel office staff 83.3% 23.0% 38.0% 39.0% Employee's performance standards/plan 81.7% 36.7% 57.1% 6.1% Regulations, policy, or guidance 73.3% 33.0% 60.2% 6.8% Other supervisors 62.5% 13.3% 70.7% 16.0% Other resources 18.3% 27.3% 50.0% 22.8% What Percent are Being "Worked On?" During the interviews, supervisors indicated that they were still seeking to improve the performance of 40 percent of those employees identified as current poor performers. They said they were not actively working on the remainder for a variety of reasons: some poor performers were expected to retire soon; some had health problems that precluded immediate corrective action; and some posed problems that (in the supervisor's eyes) seemed impervious to corrective action (i.e., the supervisor had given up). The latter reports usually included stories of failed attempts and frustrated efforts to work with "the system." Rehabilitated Poor Performers How Many Are There? Based upon the number of "rehabilitated" poor performers identified in our interviews, the estimate for this category Governmentwide is 1.5 percent. None of the rehabilitated poor performers was reported to be serving a probationary or trial period at the time of their poor performance. Of this group, 12.5 percent were supervisors. The average length of service for rehabilitated poor performers was 14 years. As with current poor performers, the supervisors' descriptions of rehabilitated poor performers were often mixed with issues related to their health, attendance and leave, and conduct (see Table 6). For example, an employee, whose work quality was otherwise acceptable, procrastinated. He was frequently away from the work area, spent a lot of time gossiping, and distracted other employees from completing their work. Table 6. Issues Associated with Poor Performance Rehabilitated Poor Performers (Percent) Governmentwide 95% Issue Estimate Confidence Interval ------------------------ -------------- ----------------- All poor performers 1.5% 0.7%--2.2% Performance only 0.9% 0.3%--1.6% Performance & conduct 0.3% * Performance and health 0.1% * Performance and attendance 0.1% * Conduct only 0 * * Due to the nature of the variance estimation in cluster samples, accurate confidence interval estimates are not available for very small ercentages. Most of the rehabilitated poor performers reported in this survey appeared to have developed problems while on their current job (See Table 7). One supervisor described the situation as a performance "slump." However, a high percentage arrived in the current supervisor's organization carrying substantial "baggage" from a previous work situation. For some, this baggage consisted of poor skills and/or work habits that had not been addressed by their previous supervisors. For others, it was anger, resentment, confusion and/or bruised self-confidence from having been affected by a reorganization or reduction in force. Table 7. Origin of Poor Performance Rehabilitated Poor Performers Percent in Reason Reported Group --------------------------------------------- ----------- Employee developed a poor attitude or work habits 62.5% Employee came as a poor performer 43.8% Other 18.8% Employee has health problems 12.5% Employee's position or job assignment changed 9.4% Wrong match of the person and the job 3.1% Note: More than one option may apply to an individual employee. A comparison of the origins of poor performance for current employees (Table 3) and rehabilitated employees (Table 7) shows some interesting differences. The percentage of rehabilitated poor performers who came into their current organizations with performance problems is noticeably less than the current poor performer group. Conversely, a higher percentage of rehabilitated performers' problems originated on the current supervisors' watch than was the case with the current poor performers. A much lower percentage of rehabilitated performers were identified as a "wrong match" for their positions, and a lower percentage of rehabilitated performers were identified as having health problems. These differences may indicate that the performance or behavioral problems among the rehabilitated group were not as great or as deeply rooted as those among the current poor performers and, therefore, may have been more susceptible to improvement. It is also possible that the supervisors themselves contributed to the problems, recognized this fact, and were able to correct their own behavior. What Have You Tried To Do About It, and How Did It Work? Whatever the reasons may have been, the supervisors of rehabilitated poor performers tried to resolve their employees' performance problems and succeeded. Finding the solutions, however, meant getting to the causes of the problem performance. Primarily through discussion with the employees, these supervisors were usually successful in getting to the causes. Common causes were: personal/family problems; untreated substance abuse; health/medication problems; inadequate technical training, and poor time management skills. As with the current poor performers, these supervisors used personal guidance most often in seeking to improve the employees' performance. A noted difference in this group, however, is the reported usefulness of this intervention as well as the results. Of the supervisors using personal guidance, 65 percent indicated it was "very useful," and another 27 percent reported it as "somewhat useful" in correcting performance problems. Although these supervisors relied somewhat less on providing additional training, changing work assignments, and providing mentors, they found these interventions to be much more useful than did those supervising current poor performers. The more formal options were rarely used by these supervisors, but, when they were used, there was greater success than with the current poor performer group. Whereas the supervisors of current poor performers never considered performance ratings or performance improvement plans to be "very useful," this group of supervisors rated these options as useful more often than not. Table 8 provides a summary of the interventions supervisors used and how they viewed the usefulness of these interventions when they were used. Table 8. Usefulness of Performance Interventions Rehabilitated Poor Performers Some- No what Very Performance Intervention Used Effect Useful Useful ------------------------------------- ----- ------ ------ ------ Provided personal guidance 81.3% 7.7% 26.9% 65.4% Provided additional training or education 31.3% 20.0% 30.0% 50.0% Assigned mentors or advisors 31.3% 0 40.0% 60.0% Changed work assignments 28.1% 0 22.2% 77.8% Rated the employee less than fully successful 21.9% 42.9% 14.3% 42.9% Developed a performance improvement plan 18.8% 33.3% 0 66.7% Initiated action to reassign, demote, or remove 0 0 0 0 The relative usefulness of the various performance interventions suggests once again that the performance deviations of the rehabilitated group overall were not as extensive or as deeply ingrained as those of the current poor performers. Since a large majority developed their problems in the presence of the current supervisor, the problems may have been more quickly detected and easier to address. Several supervisors of rehabilitated poor performers gave credit to the Employee Assistance Programs (EAP) for playing a major role in helping employees to resolve the problems which impacted on performance. This was especially the case when substance abuse or family problems were known or suspected as contributing factors. A few supervisors mentioned that "last chance" agreements produced the desired changes in poor performers who had managed to resist other corrective tactics. These formal agreements got the employees' attention because they provided the visual confirmation that only two options remained, and one of them was removal. Sometimes those agreements included requirements to participate in EAP programs or undergo substance abuse treatment. For a few others, only a less-than-satisfactory performance rating was able to get the employee's attention. Confronted with the reality of the official performance rating, these employees finally improved their performance. One supervisor described an employee who arrived in his organization as a result of a downgrade through reduction in force. The employee was quite distraught over his demotion from a long-time position in program management to one involving program operations. He ignored his new responsibilities and persisted in functioning as though nothing had changed. For example, without informing management, he intercepted a speaking invitation intended for the program director, responded to it, and actually gave the speech himself as though officially representing the director. It took some very candid counseling and a less-than- fully-successful performance rating before the employee finally accepted the responsibilities of his new position and began functioning appropriately. Two supervisors reported that, in their team-based organization, peer pressure helped their efforts to convince problem employees of their performance or behavior deficiencies. Just knowing that their co-workers agreed with the supervisor's assessment was enough to bring about a change. The supervisors were also asked how useful they found certain official resources in rehabilitating poor performers. The most notable difference between these successful supervisors and the supervisors of current poor performers was that the successful supervisors named the personnel office as their most valuable resource (84 percent rated personnel office consultation as "very helpful"). In addition, the successful supervisors very often volunteered high compliments about the assistance provided by the personnel office staff. Table 9 shows how the supervisors of rehabilitated poor performers evaluated the various resources for dealing with poor performance. Table 9. Value of Resources to Supervisors in Dealing with Rehabilitated Poor Performers Some- Not what Very Resource Used Helpful Helpful Helpful ------------------------------- ---- ------- ------- ------- Employee's performance standards/plan 84.4% 22.2% 55.6% 22.2% Servicing personnel office staff 81.3% 3.8% 11.5% 84.6% Regulations, policy, or guidance 68.8% 22.7% 63.6% 13.6% Other supervisors 62.5% 5.0% 60.0% 35.0% Other resources 12.5% 0 25.0% 75.0% Former Employees Who Were Poor Performers Supervisors were asked about their experience in dealing with poor performers who were no longer under their direct supervision. In the 200 interviews conducted, 108 supervisors reported at least one experience. At the time of the poor performance, 12 percent of these former employees were reported to have been serving either a probationary or trial period, and almost 15 percent were supervisors. Nearly half of these employees were considered poor performers based solely on performance, with no other issues present. In the remaining half, conduct was cited as the most prevalent related issue, followed by attendance, and then health. Supervisors were asked how the former employee's poor performance came about. As was found with the current and rehabilitated poor performers, the supervisors indicated that the employees typically came into the organization or under this supervisor as a poor performer. The next most prevalent origin was the development of poor attitudes or work habits. Of the interventions available for improving employee performance, supervisors once again indicated they most often used personal guidance. However, over half indicated this effort had "no effect" and only 8 percent found it "very useful." We inquired into what happened with these former employees. In almost a quarter of the cases, management actually took action to remove or reassign the employee, and, in 16 percent of the cases, the poor performer was rehabilitated. A large share (39 percent) left after "prompting" from the supervisor (one supervisor said that the employee "felt the footsteps getting closer"). The remaining employees (20 percent) left the organization totally on their own volition or are, in fact, still working for the organization under a different supervisor. The results of this part of our survey confirm a commonly-held opinion that many poor performers move themselves out of a poor performance situation without formal action. IV. PERSONNEL ACTION SAMPLE "Ask the man who owns one!" - Ad from the Packard Motor Company PURPOSE This portion of the study involved telephone interviews with supervisors and managers who have personally initiated performance-based actions. We wanted first-hand information about these individuals and their experiences with these much-maligned and infrequently-used processes. The specific questions we asked were: What led you to initiate the action? What prior knowledge or concerns did you have about the process? What kind of support did you expect, and what kind did you get? What impediments did you encounter? What consequences have you, your subordinates, and your organization experienced? What would you do if the situation arose again, and why? What suggestions do you have for improving the process? METHODOLOGY For Fiscal Year 1997, we found very few actions in the CPDF which were recorded only as performance-based removals or reductions, and the number of actions with "mixed" codes (performance and conduct) was only slightly larger. Therefore, we did not attempt to select a random sample from this small population. Rather, we started at the top of the list of actions and contacted the associated personnel offices for names and telephone numbers of supervisors and managers we could interview. When we contacted the personnel offices, we also asked the person with whom we spoke if he or she personally knew of any supervisor or manager who had taken a personnel action based on poor performance and, if so, might we have that person's name and telephone number. This two- pronged approach turned out to be very effective. Using this approach, eight interviewers who were specially trained and dedicated to this part of the study, conducted a total of 42 interviews. Interviews took place by telephone in April and May of 1998. All interviewers used a structured, open-ended interview format to reduce the variance of responses due to interviewer variability. RESULTS Because of the nature of the issues and the relatively small number of interviews con-ducted, we have chosen to present the information in summary fashion, organized into what we believe are common themes and useful categories. General The information in this part of the study comes from the first- hand accounts of people who initiated and completed a personnel action to address poor performance. The information is not hearsay. It reflects, for the most part, intensely emotional experiences etched in the memories of the participants. Our interviewees were anything but reticent. They often went into considerable detail, and their responses indicated considerable thought devoted to the experience. The 42 supervisors interviewed were from 12 different agencies, located in 14 states and the District of Columbia. The grades of the employees against whom the supervisors took action ranged from GS-4 to GS-13. We did not learn the grades of ten employees due to their supervisors' wish to assure the anonymity of the employees. The interviewees work in widely divergent agencies and organizational cultures. However, their responses were remarkably consistent. Reasons for Taking the Action All supervisors stated that the trigger for the process which ultimately led to the action was that the employees did not perform at their current grade level. Very few interviewees were more specific than that. A few added that the employee was disruptive in some way. Where numerical performance standards were used, as in the Internal Revenue Service's Total Performance System or the Veterans Health Administration, more specific performance inadequacies were mentioned. Five interviewees said they took action because the employees lacked the knowledge, skills, or abilities required for their jobs, e.g., communication skills, current technical knowledge. Three stated that the employee did not even perform adequately at the lowest grade level in the career ladder. Another three stated that the employee could not do the work at the journeyman level and that they had no need for "permanent apprentices." Several comments indicated a link between poor performance and conduct issues, some with indications of possible emotional disturbance. Some examples: The employee gave incorrect responses to inquiries and did not return calls. She stated "If it's important, they'll call back." The employee did not give the briefings and presentations he was assigned to do. He shifted assignments without approval and claimed others' work as his own. The employee wandered around and annoyed co-workers. The employee was hostile to a new supervisor and manifested it by refusing to perform adequately, by attempting to get co- workers to resist the supervisor, and by denigrating the agency to its customers. The employee could not have a coherent conversation, would forget what she was doing in the midst of the task, came to work only partly dressed, and was unable to do simple tasks. It is interesting to note that the supervisors often considered conduct issues, such as insubordination, to be performance problems. This blending of behaviors in areas covered by different regulations and processes is similar to what was portrayed by supervisors in the employee sample (Part III of this report). Support for the Supervisor To determine the level of assistance provided to the supervisor who initiated a performance-based action, we asked the following questions: What kind of assistance or support did you get in the course of taking the personnel action? Was there any assistance you would have liked but did not get? Were there any impediments encountered in the course of taking the action? The responses fell into three categories: Lack of support from upper management. Most of the interviewees indicated that upper management "support" ranged from indifference to active opposition. In some organizations, upper management has let it be known that it will not support any actions based on poor performance, or will only do so if the employee does not file a formal grievance. Some supervisors believe that upper management will not support them against any charges the employee chooses to make. Varying quality in technical guidance. The interviewees generally reported more support from personnelists than from upper management, and several praised the quality of the support they received. However, some said the assistance they received was untimely, inaccurate, or incomplete. Commitment of resources. Interviewees found the investment of time and energy required over an extremely long period to be daunting. This was compounded by the stress resulting from the employee's counter-charges, grievances, accusations, appeals, general hostility and attempts to subvert the supervisor. One described the documentation required as "horrendous." Consequences of the Action To determine the consequences experienced by our interviewees, we asked the following questions. What effect did taking the action have on you personally? What effect did taking the action have on your subordinates? Were there any effects on your organization as a whole as a result of the action? The supervisors interviewed mentioned negative consequences far more frequently than the positive. Although we did not gather information about the relative intensity and duration of those consequences, the sheer preponderance of negatives may explain the rarity of the action. Negative Consequences The negative consequences described fell in three main areas: effects on the supervisor's performance and relations with other employees; effects on the supervisor's emotional and physical condition; and effects on the group's performance and relations with the supervisor. The following comments are typical. Supervisor's Performance and Relations with Other Employees Because of the amount of time I had to spend on the action, my performance suffered and I received a rating of needs improvement.' Because the action was kept confidential and I had to spend a lot of time on it, my other employees thought I was not doing my job. After removing the poor performer, I was told I couldn't fill the job. So I exchanged a poor performer for no performer at all. The employee managed to intentionally damage the reputation of the organization with our customers in the period between initiating the action and removal of the employee. The employee maligned me to everyone she could. You open yourself to charges of discrimination and harassment. If there is an opportunity, someone is going to use it against you. The accuser becomes the accused. Supervisor's Emotional/Physical Condition I felt guilty about depriving him of a job. I faced constant hostility and resistance from the employee and his friends. This persisted over the lengthy period it took to complete the action. I felt that I was not being fair to the other employees because I was not able to give them enough attention or direction. It affected my blood pressure. I felt I failed because I was unable to train and motivate the employee up to standard. The employee was very threatening. I feared for my safety. Group Performance and Relations with Supervisor Everyone got very legalistic. For example, they were asking things like "Exactly what time do you want the report?" It stifled creativity of the entire group because we had to institute the same formal monitoring done on the poor performer on everybody else in the group. The removal action was not successful. This led to a general feeling in the group that it doesn't matter if you perform or not. The employee became disruptive at staff meetings and stirred up arguments. Other employees became cynical as the process dragged on and nothing happened to the poor performer. When the employee did even less, they had to pick up the slack. Positive Consequences Positive consequences were fewer and mentioned less frequently. It increased my credibility with the other employees. The better employees supported my taking the action; they were tired of supporting her, but would not "rat" on a fellow employee. It sent a message to other employees that poor performance was not acceptable. It restored our faith in the system; poor performers are "not getting away with it." I felt personal satisfaction that I had done my job and done the right thing. The last of these positive consequences, the feeling of having done the right thing, was mentioned by virtually all of the supervisors, either as a consequence of taking the action or as the motivation for taking it again, if necessary. Supervisory Resolve A sense of moral imperative appears to be a primary force impelling supervisors to take a performance-based action despite the negative consequences. When we asked if they would do it again, a number of interviewees simply said "yes," "definitely," or "absolutely." Only one of the 42 interviewees had reservations, saying: "Not at my age. I might if I were younger." When asked if they would do anything differently, most said "no," but about one-quarter said they would start the process earlier and not keep hoping the employee would improve spontaneously. Five stated they would be more organized on their documentation, one reported being a little too cautious about taking the action for fear of a discrimination complaint, one would have maintained closer contact with the legal office, and one regretted not reviewing personnel records before accepting a transferee. Suggestions by Interviewees for Improving the System Virtually all supervisors had suggestions for improving the system or complaints about specific aspects of it. Some of the complaints imply suggestions. A number of suggestions or complaints may derive from misunderstandings of or confusion about laws, regulations, agency policies, negotiated agreements, and local management policies. They fell into several common areas that we have seen before. Performance Standards. Several supervisors suggested developing more objective and defensible performance standards. As with the supervisors in the employee sample, these supervisors were particularly critical of "standardized" performance plans developed outside their control. They said that better standards would increase the confidence of the supervisor in both the correctness of the action and its legal defensibility. Probationary Period. There were recommendations to increase the length of the initial probationary period to two or three years, provide a probationary period following each promotion, and, for positions in which there is no use for a permanent employee below the full-performance level, establish the "up or out" rule. Lesser Penalties. There were a number of suggestions related to promoting more use of alternative penalties for poor performance, such as withholding of a within-grade increase. These penalties can provide early warning to the employee that poor performance is not acceptable and that the supervisor intends to do something about it. Support for Supervisors. The interviewees suggested improved support in a number of areas: supportive attitudes and actions from upper management; more knowledgeable personnelists to guide them through the complexities of the process; printed and electronic information, including model letters and forms; and training in the process for the supervisors themselves. The Process. Recommendations included: simplify it; shorten it; reduce the number of avenues for appeal or consolidate them so that the same lengthy appeal cannot be made consecutively in different venues; and shorten the time periods for performance improvement plans and responses to charges. V. THE NON-FEDERAL ENVIRONMENT "Bumstead, you're fired!" - Mr. Dithers, in the comic strip "Blondie" PURPOSE In this phase of the study, we set out to develop a frame of reference. Based on our random sample, we estimate that 3.7 percent of Federal workers are poor performers. While zero would be ideal, what is a reasonable expectation? Is 3.7 percent too many? To best answer this question, we set out to learn what is known about poor performance norms in the non-Federal workforce. POOR PERFORMERS IN THE NON-FEDERAL WORKFORCE As of this writing, we do not know how many poor performers there are in the non-Federal workforce, and we do not think anybody else does, either. After searching extensively for reports of studies and surveys, we have concluded that nobody has attempted a study like ours in the public or private sectors. We suspect the reason is simply that non-Federal employees have not fallen prey to the criticism aimed at the Federal sector, and, therefore, there has been no need to study the situation. We thought we could look at statistics about those fired from employment for a general indicator of the scope of the performance problem and the level of effort devoted to addressing it. Most readers will expect, as we did, that the Department of Labor has such figures. They do not. The Bureau of Labor Statistics interviews the members of a sample of households each month to determine who is unemployed and the reason for unemployment. However, their definitions of "job losers" and "job leavers" do not distinguish between those fired for cause and those laid off for purely business reasons. The Bureau of National Affairs, a private organization, conducts quarterly surveys of human resources executives and reports monthly turnover rates in the U.S. that hover around one percent. We also found references in business literature to annual "dismissal" rates in the United States, Europe, and Japan of about four percent. While the figures are so general as to provide little concrete foundation for a conclusion, one could argue that, if the four percent figure includes layoffs and performance-based dismissals, the latter number does not reflect an effort much more vigorous than what occurs in the Federal sector. Comparable CPDF statistics for the Federal workforce for the last three full fiscal years hover around three percent. This line of speculation led us to look beyond the statistics. Perceptions of the Federal system based on comparisons with the private sector may not take into account the extent to which poor performance may also be a problem in the private sector, and the range of legal and administrative appeals available to non-Federal employees. While no one to our knowledge has systematically tested Federal workers on the accuracy of their perceptions of employment practices in the non-Federal sector, a Federal manager whose only exposure to private sector personnel practices is the cartoon image of an enraged Mr. Dithers shouting "Bumstead, you're fired!!" at a hapless Dagwood might well be expected to have a distorted view of the relative ease with which private sector managers deal with poor performers. LEGAL PROTECTIONS FOR PRIVATE SECTOR EMPLOYEES Our research disclosed that private sector managers have their own set of employee protections to keep in mind. Some of these protections flow from Federal laws that are not applicable to civil service employees, such as the Immigration Reform and Control Act of 1986, the Employment Retirement Income Security Act of 1974, and the Vietnam Era Veterans' Readjustment Assistance Act of 1974. Each of these laws provides at least one cause of court action for private sector employees that cannot be brought against a Federal agency. In addition to Federal laws, there are redress mechanisms available to private sector employees under state statutes and civil law that do not apply to Federal employees at all. An allegation of "wrongful discharge" is perhaps the most important avenue of redress available to private sector employees, and it is an expensive process for the employer. A national survey of 260 wrongful discharge cases between January 1986 and October 1988 found that, in the 166 cases won by employees, the average jury award was $602,303, and the median award was $158,800. A 1988 RAND study estimated that the average cost of defending a wrongful discharge case is $80,073, and the median cost is $65,000. Each of the 50 states has its own body of case law defining the grounds on which a former employee can sue an employer for wrongful discharge. These grounds vary widely. At one end of the spectrum, five states still hold to the doctrine of "employment- at-will" with minimal exceptions. This doctrine holds, in the words of a 1884 Tennessee decision, that "all may dismiss their employee(s) at will, be they many or few, for good cause, for no cause, or even for cause morally wrong without being thereby guilty of legal wrong." At the middle of the spectrum, case law puts some limits on the employment-at-will doctrine. Typical exceptions to this doctrine are: (1) public policy, (2) implied contract, (3) good faith and fair dealing, and (4) various damage claims, such as intentional infliction of emotional distress. However, the standards for a successful damage claim are infrequently met, and many states allow employers a defense to the implied contract and good faith and fair dealing exceptions by recognizing express disclaimers, such as an agreement to forego litigation as a condition of employment. The public policy exceptions offer the most numerous and widest variety of possible wrongful discharge suits. State case law definitions of "public policy" are typically open-ended. For example, in 1959, a California court declared that anything that contravenes "good morals or any established interests of society" constitutes action against public policy. In 1981, an Illinois court defined public policy as "that which is right and just and collectively affects the state's citizenry." And, in Oregon in 1975, a court declared that an employer can be held at fault for dismissing an employee "for a socially undesirable motive." At the far end of the spectrum of possible grounds for wrongful discharge suits is the 1987 statute adopted by Montana. This law establishes a compulsory arbitration system with statutory rights to job reinstatement or financial compensation for unjustified discharge. Over the past few years, legislation has been offered in 40 states proposing approaches similar to Montana's. An organization called the Uniform Law Commissioners has developed a Model Employment Termination Act (META) which includes a provision that employees cannot be terminated without good cause. Many state legislatures have introduced legislation influenced by the META, though none have yet adopted it. State laws also provide private sector employees protections unfamiliar to Federal supervisors and managers. Federal managers and supervisors might be surprised to know that smokers are a protected class in 26 states, and private sector employee whistle- blowers are protected by law in over 30 states. It can be safely said that private sector employers cannot separate poor performers without incurring some risk. Not surprisingly, a survey of human resource textbooks found that most advise private sector managers and supervisors to take steps fairly comparable to those outlined in OPM's recently-published guide, Addressing and Resolving Poor Performance, with the additional admonition to "document, document, and then document some more." PROTECTIONS FOR STATE AND LOCAL GOVERNMENT EMPLOYEES The avenues for challenging dismissal available to private sector employees are typically available to state and local employees as well. In addition, the U.S. Supreme Court gave state and local employees new grounds to challenge removal when it found dismissals for purposes of patronage to be unconstitutional in Elrod v. Burns (1976). This protection was expanded in 1990 when the Court held in Rutan v. Republican Party of Illinois that employment decisions in hiring, promotion, and rehiring after a lay-off that are based on patronage are also unconstitutional. Recognizing that the U.S. Constitution and Federal courts provide many protections, the State of Georgia, in 1996, eliminated administrative appeals processes for employees hired after a certain date. Georgia is the exception in this regard, however, and every other state provides internal appeal mechanisms, most of which are similar to the Federal process. For example, Pennsylvania has a general policy of "just cause" described in its Supervisor's Guide to Employee Counseling & Discipline: Supervisors and managers need a good reason for taking any disciplinary action and should be able to demonstrate that the decision to impose discipline is not arbitrary, capricious, or discriminatory. The burden of proof to show just cause' for disciplinary action rests with supervisors and managers. Appeals processes are described in the Internet-accessible state employment codes and manuals of Indiana, North Carolina, and Oklahoma. Even Georgia's civil service board recommends that, in lieu of the rescinded Statewide system, State agencies consider establishing their own appeals processes for recently-hired employees who do not have access to the older system. CONCLUSION In summary, legal protections for employees in the non-Federal sector are similar in varying degrees to those in the Federal sector. This is true in both public and private employment. Moreover, there are indications that the level of protection is slowly increasing. Finally, the administrative procedures for separating a Federal employee for poor performance are not radically different from those for separating an employee in many state governments. VI. CONCLUSIONS We have met the enemy, and he is us! - Pogo Human behavior is complex, and it should be no surprise that the results of nearly 250 interviews would produce findings that are complex and varied. For every question we answered, new questions appeared. Nevertheless, we are able to draw a number of important conclusions from our interviews and research. There is good news to report. Our estimate that 3.7 percent of Federal workers are poor performers is lower than the "common wisdom" about Government would suggest. It is unfortunate that we could find no good private or public sector benchmarks with which to compare our findings. However, the generally accepted performance norms for organizations and individuals suggest that the Government has "scored" a 96.3 on this "test." Managers do manage. Our surveys disclose that supervisors do take action to address poor performance. This often takes the form of informal measures which resolve the situation before the need to pursue formal action, e.g., a poor performer will voluntarily move on when they "feel the footsteps getting closer." Supervisors report that they are working in a variety of ways with almost half the poor performers in the current workforce, and many report having had success with rehabilitating workers who were poor performers at one time. Finally, the reports from supervisors who took performance-based adverse actions demonstrate an almost-heroic determination to address serious performance problems. In selecting a course of action, supervisors often report that they weigh the time and effort it would take to confront a poor performer with a formal personnel action against: (1) the prospects for success with that action; and (2) the cost of continuing the status quo in terms of accommodation and the impact on other workers. In this way, supervisors are making what seem to be informed judgments related to their accountability for the "bottom line," which is their organization's effectiveness and productivity. This behavior reflects pragmatic realism more than cowardice or cynicism. Things are not so easy in the non-Federal sector. While employment protections in some non-Federal settings may be such that employees can be dismissed virtually at will, the trends reported in this study seem to be in the opposite direction. Employment protections exist in both the Federal and non-Federal sectors. The challenge for all managers is to balance the risks, costs, and benefits, and to make responsible trade-offs. Our perceptions of reality are not accurate. Federal managers, supervisors, and employees do not have a clear and accurate perception of the scope of the poor performer "problem." Their responses to OPM and MSPB surveys indicate that Federal workers are major purveyors of the misperception that the Government has too many poor performers. For a number of reasons (one of them may be simply that they are asked so often), Federal workers seem almost obsessed with describing the poor performer problem. In point of fact, many employees may be reporting numerous observations of relatively few poor performers. We think there is also substantial misinformation about the difference between employment protections in the Federal and non- Federal sectors. Federal managers and supervisors who believe that "at will" employment is the norm outside the Federal government are ill-informed and may therefore suffer from a "grass is greener" syndrome. This study is consistent with and supports other survey findings. In conveying the positive findings of this study, we do not want to suggest that the concerns reported in previous surveys and questionnaires are without merit. The supervisors we interviewed were definitely not disposed to use formal procedures to "get rid" of poor performers, and those who finally did use those procedures found the experience distasteful, to put it mildly. This is consistent with many previous studies, and there are clearly opportunities for continuous improvement in how we handle poor performers. Organizational culture does not support taking action against poor performers. We were struck by supervisors' consistent perception that higher- level management was not supportive of was even resistant to direct action to deal with poor performers. Clearly, one of the factors the supervisors had to weigh in making a pragmatic decision was the degree of support or resistance expected from higher management. It is quite possible that what supervisors are perceiving and reporting is the accumulated effect of the legitimate concerns of higher level managers, attorneys, and human resource specialists. However, to the extent that this situation creates a perception that management is predisposed to resist or obstruct corrective action, it prevents dealing with poor performers. Supervisory training has not been useful. Most supervisors reported that, early in their supervisory careers, they received training or orientation related to dealing with poor performers. However, they also reported that, when faced with a poor performance situation, the training was too remote in time or too general in nature to be useful. This disparity may be partly due to the supervisors' infrequent need to draw upon the training; complex knowledge related to an infrequent problem is not likely to be retained to a degree that maintains its usefulness. VII. RECOMMENDATIONS The primary purpose for this study was to obtain new information about the scope of the "poor performer problem" in Government. We have accomplished that purpose. However, our findings and conclusions lead us to several suggestions for further action. 1. Get accurate information into the public discourse. Our findings indicate that the public, including the Federal workforce itself, tends to exaggerate the scope of the poor performer problem in Government. The proportion of poor performers is not as large as the rhetoric suggests, and there are indications that efforts to deal with poor performers have achieved and will continue to achieve a measure of success. However, the prevailing opinion that the situation is otherwise may in itself constitute a stumbling block to even more success. At the very least, it adds to the burden of supervisors and employees who share an uninformed perception of their situation and who believe that the only way to improve the situation is to escape or roll back public sector employment protections. Wide dissemination of the findings in this report should serve to move the public discussion toward a firmer foundation in reality. There is room for improvement in how we handle poor performance, but goals for improvement should be based on accurate information and realistic expectations. 2. Top agency management should do more to support managers and supervisors in taking action to deal with poor performers. The NPR has called for a change in management culture with regard to dealing with poor performance, and our study suggests that this call is right on the mark. The accounts of the supervisors in this report are eloquent and compelling, and no top manager should be comfortable seeing his or her actions or inactions reflected in those accounts. To the extent that these perceptions are fair and accurate, top managers must try harder to support justifiable actions against poor performers. The processes and interactions in this arena are complex, and the players are many and varied. It will take extra effort on the part of top managers to articulate and convey clear expectation amidst the confusion of fact and fancy. 3. Training and support for supervisors should emphasize "just in time" approaches. It is clear that supervisors do not absorb enough information about dealing with poor performers during routine supervisory orientation. It is also clear that, when they need such information, they need it badly and in a hurry. This suggests a phased learning and assistance model: (1) Orientation training that stresses how to recognize a performance problem at the early stages; (2) Self-study resources that will answer specific questions and provide general directions when a problem must be dealt with; (3) Expert coaches to guide, evaluate, and facilitate the process of dealing with a poor performer. Because the poor performer problem arises infrequently, the effort to provide the resources in phases (2) and (3) must be governed by considerations of efficiency. OPM's recently-published CD-ROM guide, Addressing and Resolving Poor Performance, exemplifies the kind of self-study resource called for by the model. Likewise, the model suggests that a relatively few strategically-placed and readily-available technical experts should be able to provide the level of coaching necessary to meet the demand. To the extent that computer-based libraries and "expert systems" might be developed, the bridge between phases (2) and (3) will be strengthened.