Urban Institute Evaluation of EEOC’s Performance Measures Harry P. Hatry Katharine Mark Elizabeth Davies March 19, 2013 Prepared for the Office of Inspector General U.S. Equal Employment Opportunity Commission OIG Project Number # 2012-10-PMEV by The Urban Institute 2100 M Street NW Washington, DC 20037 Table of Contents Summary ......................................................................................................................................................... 3 I. Project Overview .......................................................................................................................................... 5 Purpose & Scope ..................................................................................................................................... 5 Methods .................................................................................................................................................. 5 II. Review of EEOC 2012-2016 Strategic Plan Performance Measures ........................................................... 6 Strategic Objective I: Combat employment discrimination through strategic law enforcement .......... 7 Strategic Objective II: Prevent employment discrimination through education and outreach ............. 8 Strategic Objective III: Deliver excellent and consistent service through a skilled and diverse workforce and effective systems ............................................................................................... 9 Further Observations on the Current Strategic Plan Measures ........................................................... 10 III. Recommended Revisions to EEOC’s Performance Measures .................................................................. 13 Strategic Objective I: Combat employment discrimination through strategic law enforcement ........ 15 Strategic Objective II: Prevent employment discrimination through education and outreach ........... 18 Strategic Objective III: Deliver excellent and consistent service through a skilled and diverse workforce and effective systems ............................................................................................. 20 Overall Objective: Reduce National Employment Discrimination Levels ............................................. 22 Measurements of “Efficiency” .............................................................................................................. 26 Overall Recommendations on Measures.............................................................................................. 27 Exhibit 1. Recommended Outcome Measures ..................................................................................... 28 IV. EEOC Use of Performance Measures: Findings & Recommendations .................................................... 30 V. Conclusion................................................................................................................................................. 33 Appendix A. List of Interviews....................................................................................................................... 35 Appendix B. References & Documents Reviewed ........................................................................................ 36 Appendix C. Performance Measures in the Strategic Plan for Fiscal Years 2012-2016 ................................ 38 Appendix D. Performance Measures in the Strategic Plan for Fiscal Years 2007-2012................................ 39 Appendix E. Notes on Existing National Surveys as Potential Source for Key EEOC Measures .................... 40 Appendix F. Using Paired-Comparison Testing to Measure Employment Discrimination ........................... 44 Appendix G. EEOC Reviews of January 30 Draft Version .............................................................................. 46 Urban Institute Evaluation of EEOC’s Performance Measures | Page 2 Summary ”Stop and Remedy Unlawful Employment Discrimination.” This is the mission of the U.S. Equal Employment Opportunity Commission (EEOC) as reported in its Strategic Plan for Fiscal Years 2012- 2016. Regularly providing data on progress towards this mission is the major purpose of EEOC’s performance measurement system. The performance measurement information is needed for use by EEOC in guiding its internal activities and in reporting to the President, Congress, and the public. This report seeks to provide recommendations to EEOC to help it meet this important need. The Strategic Plan for Fiscal Years 2012-2016 focuses on three strategic objectives: 1. Combat employment discrimination through strategic law enforcement; 2. Prevent employment discrimination through education and outreach; and 3. Deliver excellent and consistent service through a skilled and diverse workforce and effective systems. This report discusses outcome measures that can be used to track progress towards these three objectives. This report also suggests measures that can be used to track the nation’s progress towards EEOC’s overall mission of helping to stop unlawful employment discrimination. We examined EEOC’s current performance measures, focusing on those measures included in the latest strategic plan (FY 2012-2016) and their ability to track the progress of EEOC‘s mission and strategic objectives. We also sought to identify measures likely to be appropriate for future strategic plans and those measures likely to be important for EEOC commissioners and management for tracking the outcomes on a regular basis and encouraging continual service improvement. Most of the measures contained in the current strategic plan are primarily process and not outcome measures, with two important exceptions. The current measures do not cover the nation’s progress towards achieving the overarching goal: to reduce employment discrimination in the United States. This report provides recommendations for outcome measures for each of the three strategic objectives included in the current strategic plan. It also includes recommendations for tracking progress towards reducing employment discrimination in the United States. The latter measures, though particularly difficult to measure fully satisfactorily, nevertheless warrant major attention by EEOC. Suggestions as to data sources and data collection procedures are identified for each measure. Overall, we recommend these three major actions: 1. Include measures such as those proposed in Section III of this report as soon as possible in EEOC’s annual Performance and Accountability Report (PAR). These are the outcomes that citizens, Congress, and OMB are most likely to consider important. Progress on these measures should be a major focus of EEOC. Urban Institute Evaluation of EEOC’s Performance Measures | Page 3 2. Plan to include at least a subset of these measures as the principal outcome measures in EEOC’s next strategic plan. The data obtained from 2012-2016 on the measures will provide baseline data for establishing targets for the next strategic plan. 3. Begin introducing some of these measures into updates of the current strategic plan. This report concludes by providing a number of recommendations for improving the use of outcome information -- drawing on the new focus of the GPRA Modernization Act of 2010, such as the use of regular data-driven in-person reviews with agency staff. We recommend that EEOC: 1. Expand the new Strategic Enforcement Plan’s (SEP) requirement for quarterly reviews to include not only SEP progress but also progress reflected in the latest EEOC performance reports. While not currently required by OMB of small agencies such as EEOC, EEOC management would likely benefit considerably with the implementation of quarterly data-driven reviews, such as those required of large federal agencies. These regularly scheduled sessions with EEOC’s major units would focus on reviewing the latest performance information on both process and outcome measures. The sessions would be used to identify problem areas and to help identify ways to correct those problems. 2. Provide commissioners and managers with easy access to relevant disaggregations of the outcome measure values. Outcome data would be broken out by such characteristics as priority level, industry, and key characteristics of the charging parties. 3. Ask the appropriate office to provide explanations for unexpectedly poor, and very good, measurement data values shown in the latest performance report. Such explanations should be included as part of both external and internal performance reports. Such information would be part of the performance reports and would encourage management at all levels to focus on needed improvements. Urban Institute Evaluation of EEOC’s Performance Measures | Page 4 I. Project Overview Purpose & Scope The U.S. Equal Employment Opportunity Commission’s Office of the Inspector General (OIG) contracted with the Urban Institute to conduct an evaluation of the performance measures included in the EEOC’s Strategic Plan for Fiscal Years 2012-2016. The major objective of this study was to assess the quality, coverage/adequacy, alignment with agency goals and activities, and usefulness of EEOC outcome measures to key stakeholders within the Commission and to provide recommendations as to how the performance measures can be improved. This report provides our findings and recommendations. Methods The major part of the work took place between October 2012 and January 2013 but additional helpful input was obtained subsequently, especially from reviewers of our draft report. In order to complete this research, we collected information from a variety of sources, including: in-depth interviews with EEOC leadership and staff from EEOC district offices and Fair Employment Practices Agencies (FEPAs); a review of EEOC internal documents, including strategic plans for FY2007-2011 and FY2012-2016, EEOC Commission meeting transcripts, past evaluation studies, and the FY 2012 Performance and Accountability Report (PAR); and documents from other federal agencies (such as OPM, HUD, and DOJ). Lists of interviews and of documents consulted to date are included in Appendices A and B respectively. Urban Institute Evaluation of EEOC’s Performance Measures | Page 5 II. Review of EEOC 2012-2016 Strategic Plan Performance Measures The EEOC seeks to achieve its vision of “Justice and Equality in the Workplace” through an overall mission to “Stop and Remedy Unlawful Employment Discrimination.” In the Strategic Plan for Fiscal Years 2012-16, this mission is achieved through three strategic objectives: 1. Combat employment discrimination through strategic law enforcement 2. Prevent employment discrimination through education and outreach 3. Deliver excellent and consistent service through a skilled and diverse workforce and effective systems The first objective, combating employment discrimination through enforcement, encompasses two outcome goals: (1) to provide a platform for a fair consideration and resolution of employment discrimination complaints; and (2) to remedy discriminatory practices and secure meaningful relief for victims of discrimination. The second objective, preventing employment discrimination, seeks to ensure that (1) members of the public understand and know how to exercise their right to employment free of discrimination; and (2) employers, unions and employment agencies (covered entities) better address and resolve EEO issues, thereby creating more inclusive workplaces. The third objective, delivering excellent and consistent service, strives for all interactions with the public to be timely, of high quality, and informative. The Strategic Plan for Fiscal Years 2012-2016 articulates a set of thirteen measures (see Appendix C) intended to gauge the agency’s progress on these three objectives. The measures were the result of a consultative strategic planning process that sought to develop new performance measures that better aligned with the agency’s core mission and that represented a departure from previous measures that many in EEOC regarded as limited. The measures included in the previous Strategic Plan (for Fiscal Years 2007-2012) are presented in Appendix D. In particular, many of those interviewed for this report voiced discomfort in that plan’s use of case resolution time as a proxy for case “success,” as resolution times failed to reflect the different complexity among charges or EEOC priorities that might influence disposition time. A number of stakeholders also cited ongoing concerns about the “unintended consequences” of this measure (such as programs focusing deliberately on easier charges to meet the targets set for the agency). However, one stakeholder added that there was no actual evidence to suggest manipulation. In addition to worries about the validity of data, stakeholders also felt that the measures did not reflect core EEOC activities. Previous performance measures were described by one stakeholder as “quantifiable but not good.” The Strategic Plan for Fiscal Years 2012-2016 reflects the desire among EEOC leadership for a different process that moves away from past measures where success was quantified simply by counting activities. The Commission chose to switch the majority of measures in the current plan to process goals that would enable the Commission to develop more substantive performance measures in the future. Our review of the measures in the Strategic Plan for Fiscal Years 2012-2016 is summarized below. Urban Institute Evaluation of EEOC’s Performance Measures | Page 6 Strategic Objective I: Combat employment discrimination through strategic law enforcement Appendix C lists the seven measures included in the Strategic Plan for Fiscal Years 2012-2016 that address this objective. Most of these measures focus on achievement of internal processes believed to be important steps towards achieving EEOC’s strategic goals. EEOC has considerably more ability to measure the values of these measures than it does more outcome-focused measures. These measures can be very helpful in tracking internal progress. However, they do not measure important external outcomes. Measures 1, 3, 4, and 5 are process, not outcome, measures. They relate to the completion of EEOC activities (such as implementing the Strategic Enforcement Plan, categorizing cases ,1 increasing the percentage of systemic cases, and issuing compliance plans). These measures do not address what results from these activities. Measure 2, “the percentage of investigations and conciliations that meet the quality criteria established in the new quality control plan,” is primarily a measure for internal EEOC purposes. However, it is likely to be of interest to some external parties, such as OMB. This measure, thus, is included in the recommended list of measures for Strategic Objective 1 provided later in this report. Measure 4, “the percent of cases that are systemic cases,” does not appear to be an outcome (or an output) measure. The mix of systemic cases and non-systemic cases would be a policy call for EEOC and one depending to a significant extent on the resources available. (EEOC develops systemic cases from the various charges filed or otherwise obtained.) Such a measure would more appropriately be considered a “process” measure. Also, the use of a percentage means that this metric depends heavily on the number and characteristics of its cases (which affects the denominator of the percentage independently of what is done about systemic cases). If many more non-systemic–related cases are identified, this percentage will go down. What would this mean with respect to performance? This measure would not do justice to the work of the EEOC. This metric is certainly useful for EEOC to track. The FY 2012-2016 strategic plan defines systemic cases as “pattern or practice, policy, and/or class cases where the alleged discrimination has a broad impact on an industry, occupation, business, or geographic area.” How EEOC staffs interpret the definition, will affect how many cases are identified as “systemic.” This definition of systemic charges is subject to interpretation and judgment. An additional concern we heard was that if any targets are established for the measure, those targets could be considered by field staff to be quotas. This could encourage excessive litigation 1 The process of categorizing charges is fundamental to case management. All local, state, and private sector charges are currently divided into A, B, and C priority categories immediately after intake. This determines which charges go immediately into investigation (A), which are dismissed (C), and the remainder (B) most of which are channeled directly into Alternative Dispute Resolution. The Office of Federal Operations (OFO) used to prioritize federal cases (e.g., into A, B, and C categories) but does not any longer. Urban Institute Evaluation of EEOC’s Performance Measures | Page 7 on charges that might not otherwise be considered systemic or, or it could lead to failure to pursue sufficient “non-systemic” charges. Recognizing the importance of systemic cases and EEOC’s desire to include systemic cases in its performance measures, it appears considerably better to include in its internal monitoring the number of systemic cases on which it is working (a process measure) and the number of systemic cases resolved (an output measure). For an outcome measure, an appropriate one is the percentage of those charges that EEOC was able to pursue with its limited resources and were resolved satisfactorily. Some reviewers were concerned, legitimately so, that such a measure can have the unintended effect of encouraging selection by staff of the easiest cases. This is a standard problem with outcome measurements expressed as percentages (including EEOC’s new Strategic Plan Measures #6 and #7.) Despite the potential perverse incentive, achieving successful resolutions in systemic, as well as in non-systemic cases, has to be a major concern for EEOC. Ways to alleviate this problem are discussed in the next section. Measures 6 and 7, percentages of resolutions containing targeted, equitable relief relate, respectively, to charges processed by EEOC and by state and local Fair Employment Practices Agencies (FEPAs).2 These are outcome measures and are included in the suggested list of measures for Strategic Objective 1 provided later in this report. They seek to measure success in remedying discriminatory practices and securing meaningful relief for victims of discrimination. There seems to be wide agreement that “targeted, equitable relief” represents a key component of success. Such a measure is essential, according to one interviewee, as the role of the EEOC should go beyond benefiting just one complainant. The director of a district office commented that this measure was, “as close as I’ve seen to a good measure of success.” One stakeholder did not expect the measure to vary much from 100% for all reporting units, but others felt that some units would struggle to meet the yet-to-be-defined quota. This difference is likely related to different interpretations of what “targeted, equitable relief” means and how those definition influence calculations. Ensuring that a clear, specific definition of “targeted, equitable relief” is available and well- known throughout EEOC will be vital for obtaining valid, reliable outcome data. Although some interviewees felt that the definitions were unambiguous and well understood, several others felt that the definitions were not clear and were hoping for future guidance. Cautions were raised by one interviewee that pushing for certain kinds of relief might result in requiring changes when none are really needed and perhaps unnecessarily delaying settlement. This interviewee felt that the term “appropriate relief” might be better to use in this measure. Strategic Objective II: Prevent employment discrimination through education and outreach Appendix C lists the four measures included in the Strategic Plan for Fiscal Years2012-2016 that address this objective (Measures 8, 9 10, and 11). This objective seeks to prevent employment discrimination by 2 Congress has authorized EEOC to enter into agreements with FEPAs to carry out investigations and conciliations of charges that would otherwise come to the EEOC. Urban Institute Evaluation of EEOC’s Performance Measures | Page 8 proving information to the public and to businesses as to their rights and responsibilities relating to discrimination-free employment. The current wording of these four measures does not provide significant information on to the extent to which EEOC efforts to prevent employment discrimination through education and outreach have been successful or the extent to which employees and businesses understand their rights and responsibilities. Measures 10 and 11 on social media planning and clear guidance materials are process measures. They are very likely to be useful to EEOC for internal progress reporting. However, neither measure tracks any outcomes of those activities. Measures 8 and 9, which consider the number of “significant partnerships” with, respectively, worker and business organizations, would likely to be considered process measures by persons in OMB, though a case can be made that they are “intermediate” outcomes since external organizations are involved. However, the number of significant partnerships, while an internally useful number for tracking progress in developing partnerships, says little as to what those partnerships accomplished (the outcomes of partnerships). It has the additional problem, that EEOC offices might be encouraged to push for quantity at the expense of quality of the partnerships. The two partnership measures could be altered to transform them into measures that provide meaningful outcome information. This could be done by defining “significant” partnerships to be partnerships that had produced meaningful accomplishments/products, such as the number of partnerships in which partner organizations had developed appropriate and substantive guidelines and/or training sessions for their memberships. Partnerships can require considerable time to develop and sustain. They need to produce benefits and, for measurement purposes, provide tangible, measureable benefits. Also, the definition of partnership may need more definition. What minimum size, frequency, and form of interaction, are needed to qualify as a partnership? Strategic Objective III: Deliver excellent and consistent service through a skilled and diverse workforce and effective systems Appendix C contains the two measures included in the Strategic Plan for Fiscal Years 2012-2016 that address this objective, Measures 12 and 13. These two measures seek to measure quite different elements. Both these measures are expressed in general terms. Measure 12, strengthening the skills and improving the diversity of EEOC’s own workforce, is obviously important to the agency (“do as we do, not only do as we say”). No specific numerical measures are provided in the strategic plan. Because it is an “internal” issue, we have not considered it within our scope and have not included this topic in the set of recommendation measures in the next section. However, we later, under the fourth “overall” Strategic Objective, address the problem of measuring diversity as a way to track the level of potential national discrimination. Urban Institute Evaluation of EEOC’s Performance Measures | Page 9 Measure 13, “streamline services and increase responsiveness to customers,” addresses the strategic plan’s sub-objective: “All interactions with the public are timely, of high quality, and informative.” No specific numerical measures are provided. However, we suggest specific ways to measure various dimensions of this objective in Section III under Strategic Objective 3. We recognize the purposes of the measures included in the current strategic plan. Many of those involved in the strategic planning process felt that improvements to the framework for measuring performance and ensuring quality were needed. (For example, identification of adequate criteria was needed for assessing quality, and improvement of the case management system was needed. Making these needs part of the strategic plan would ensure their completion.) As a result, EEOC incorporated the accomplishment of two planning processes – the Quality Control Plan and the Strategic Enforcement Plan – as performance measures in the Strategic Plan. The Strategic Enforcement Plan was completed late in 2012, and the Quality Control Plan is slated for completion in spring 2013.3 In addition, most stakeholders felt it was premature to name measures for which robust data was not yet available, or could not be easily collected. Further Observations on the Current Strategic Plan Measures Need for more outcome measures As indicated above, many of the measures in the Strategic Plan for Fiscal Years 2012-2016 are process measures and do not appear appropriate as major indicators of performance over the long run. EEOC consciously decided to take major steps to develop a more robust performance measurement system. Many of the measures used in the plan are intended to motivate and measure progress in development of improved performance measurement. However, Strategic Plans are intended to address performance in achieving its strategic objectives and may not be the appropriate place for measurement of internal steps. Nevertheless, there is value in the EEOC continuing to track, and report on, these measures. These measures represent important milestones and steps towards achieving the outcomes associated with the plan’s primary objectives relating to reducing employment discrimination. However, EEOC should also begin tracking a number of key outcomes relating to its strategic goals, such as those recommended in the next section, even if the data are not “perfect,” even if they have not been included in the strategic plan. Past difficulties, especially over “resolution timeliness” Interviews with stakeholders suggest that measuring these outcome goals has been challenging in the past. Timeliness, an important intermediate outcome measure, was used as the focus of three of the seven performance measures included in EEOC’s 2007-2011 Strategic Plan. However, timeliness was not considered satisfactory because it did not reflect the different complexity (and priority) among charges that can influence the amount of time spent on a case. There is agreement that the complexity of cases differs greatly; overall average resolution times could, thus, be misleading. The initial categorization (into groups A, B, and C) that all charges go through is one indication of the level of complexity. One interviewee noted that “A” charges are likely to be of higher complexity (thus requiring more time), 3 The current performance measures in the strategic plan call for investigations, conciliations, hearings, and appeals, to meet criteria that will be developed in the Quality Control Plan. The Work Group for the Quality Control Plan was launched on January 11 2012, and the Plan is expected to be ready in April 2013.The Strategic Enforcement Plan was approved on December 17, 2012. Urban Institute Evaluation of EEOC’s Performance Measures | Page 10 while B charges tend to be delayed because of the limited resources assigned to them, and therefore also may take longer to do. Reporting resolution times separately by category (A, B, and C charges4) could thus help at least somewhat. Looking at case resolution times also does not reflect variation in the complexity of investigations – and may actually encourage lapses in quality. As one stakeholder pointed out, one way to resolve a case quickly is to conduct a superficial investigation. This problem could be alleviated if reasonable estimates of the complexity can be made for each charge (such as whether the charge is of high, low, or medium complexity)—and resolution times are calculated for each such category of charges. Need for quality control Most of those interviewed stressed the importance of outcome information quality control. It appears that several quality control processes exist. For example, we understand that the Office of Field Programs (OFP) annually reviews 50-125 charges from each of the 15 districts (plus DC) looking at: (a) the handling of each of these charges at intake (e.g., was the correct priority rating made and did the office get enough information); and (b) case closure decisions (does the information adequately support the closure resolution). The reports are used to gauge individual field office performance. At least ten percent of FEPA resolutions are examined annually. In the past, EEOC has tried to develop a peer quality-review process. The Federal Consulting Group developed and piloted a peer review process several years ago. One stakeholder described a desire for close systematic examination of case quality but noted pushback from attorneys who consider “every case to be different.” In addition to variation in cases, a lack of consistent policies across offices makes standardization difficult. An evaluation conducted in 2006 found that processes across agencies were not consistent, particularly around investigations and conciliations. OFP confirmed that all 15 district offices work differently and that it would be helpful to improve consistency. 5 Need for estimates of the level of national discrimination and how it is changing over time Although the EEOC has developed measures for its three strategic objectives, it has not developed measures that indicate national success at fulfilling its vision of justice and equality in the workplace through an overall mission to “stop and remedy unlawful employment discrimination.” A very basic measure of success would be a measure of employment discrimination nationwide. The size of the EEOC caseload for each discrimination category, while relevant, does not provide an adequate proxy for the national level of discrimination. (For example, this information does not identify unreported cases.) There appears to be hesitation within EEOC (as with many other federal agencies) about including indicators over which the commission does not have a major degree of control. This problem is due in part to the widespread use of the term “performance” measures as distinct from the term “outcome” measures. The term “performance” implies that the measured values are directly due to the work of the agency--and therefore, responsibility for successes and failures can be assigned. Of greater interest to citizens, and probably Congress, are the outcomes that result regardless of the extent to which the federal agency caused them. This report takes the position that EEOC should track and report on the level of employment discrimination that exists in the United States, even if these are rough estimates. EEOC should make the 4 The categorization of charges into A, B, and C categories is currently based on a peer review process. Because of its importance, correct categorization should be subject to quality control as well. 5 Development Services Group, 2006. Urban Institute Evaluation of EEOC’s Performance Measures | Page 11 point to the outside world that for any important outcome, the results are almost always based on partnerships among not only the federal agency reporting them but many other organizations and individuals – and therefore no one should expect EEOC to be solely responsible for ensuring lower levels of discrimination. In the next section, we suggest a number of ways to obtain data that estimate the amount of national employment discrimination. Diversity of the federal workforce While lack of diversity does not by itself mean that discrimination has occurred, such lack suggests the likelihood that discrimination exits. As articulated by the Office of Federal Operations (OFO), the extent of federal employee diversity can be considered an important indicator of equality in the workplace and of potential discrimination. For example, the EEO Program Compliance Assessment (EPCA) instrument scored federal agencies on the degree of diversity among employees. The EPCA was withdrawn because it was controversial and there was disagreement over what the scores meant. The Office of Personnel Management is working on a diversity index for the federal workforce. Although the details and methodology could not yet be shared, it will include: an objective measure of the diversity of federal hiring decisions and employee attrition (based on demographic breakdowns and GS- level); an inclusivity score based on employees’ ratings of their agency’s fairness, openness, cooperation, supportiveness, and staff empowerment; and a summation of agency efforts to promote and increase diversity (such as staff training and education). These findings are expected to be included in an annual report that helps managers to see how well their agency or unit compares to other agencies or units and better understand the composition of their labor force. The OPM representative indicated that the office cannot set targets for diversity in the federal workforce. Customer service EEOC officials interviewed for this assessment articulated the importance of customer service for any government agency. EEOC has some, but limited, experience with customer surveys. In December 2008, but not since repeated, it sponsored a Customer Satisfaction and Knowledge of Law Survey to identify the extent to which citizens understood their rights under the law. The questionnaire was administered to a national sample of citizens. OFP has also conducted surveys of both charging parties and respondents who participated in a mediation process to gauge their satisfaction. Completed surveys are sent to OFP and become part of an informal rating of this aspect of the program. However, the results are used only internally. One District Office is planning to survey all visitors to its offices to assess the courteousness and helpfulness of staff. Other offices have used comment cards, but the process has not been standardized. Two officials believed that client satisfaction with the services provided by an agency is determined almost entirely by whether the case was settled favorably or unfavorably. Urban Institute Evaluation of EEOC’s Performance Measures | Page 12 III. Recommended Revisions to EEOC’s Performance Measures Below we identify a list of the outcome measures that we recommend EEOC consider. The measures are grouped into the following four outcome dimensions, three of which map directly with the three strategic goals in the 2012-2016 Strategic Plan: 1. Combat employment discrimination through strategic law enforcement (Strategic Goal 1) 2. Prevent employment discrimination through education and outreach (Strategic Goal 2) 3. Deliver excellent and consistent service through a skilled and diverse workforce and effective systems (Strategic Goal 3) 4. Achieving justice and equality in the workplace by stopping and remedying unlawful employment discrimination (New Overarching Goal) The first three dimensions align with the three strategic objectives outlined in the Strategic Plan for Fiscal Years 2012-2016, with the addition of an overarching set of measures that address national levels of employment discrimination and equal opportunity employment. A complete list of the measures recommended is presented in Exhibit 1 at the end of this section. Several issues are important to note: The following set of outcome measures are not intended to be considered to be “final.” At least, however, we hope they will provoke further review, discussion, and modification. These measures, as with all federal “performance” measures, are intended only to indicate the outcomes that resulted. They do not identify the extent to which EEOC caused the results. For the latter information, additional explanatory, in-depth evaluation, information is needed. Each measure addresses a different outcome dimension and perspective on EEOC performance. Seldom, if ever, will a single measure provide a reasonably comprehensive picture as to the desired outcomes. There is no such thing as a perfect measure. Including more than one measure for an objective can, in some instances, also guard against the problem of overemphasizing one measure so that it has perverse effects on other important outcomes, such as overemphasizing closure times at the expense of obtaining correct charge resolutions. For each measure, brief comments are provided on potential data sources, data collection, and limitation issues. We also consider the cost (dollars and burden) on EEOC staff and overall feasibility of implementing measures that would require new data collection procedures. Included in this list are 4 of the 13 measures included in the current strategic plan (Measures 6 and 7 and modified versions of Measures 8 and 9). Not included are the measures that relate to progress on various processes internal to EEOC. While important for EEOC to track, they are not measures that OMB and other persons outside EEOC are likely to consider to be important outcome measures. Urban Institute Evaluation of EEOC’s Performance Measures | Page 13 Many of the candidate measures will require time to develop (perhaps a year or more). Thus, they can be considered EEOC’s next generation of strategic plan measures. However, we recommend that development and implementation of the procedures begin as soon as possible. These measures should not be considered as being solely for strategic planning purposes but should also be tracked at least annually, and preferably more frequently such as quarterly or at least semi-annually, and used for managing and continuous learning by EEOC personnel. For each measure, it will make the information on outcomes considerably more useful to EEOC decision makers if the information is also disaggregated by important interpretive characteristics (such as by type of discrimination, by priority level, by industry, by characteristics of the charging party, and by field office/department). EEOC’s IMS system already provides considerable opportunity for making such disaggregations quite feasible. Decision makers should be able to readily access these disaggregations when the information is wanted. We identify likely useful disaggregations in the discussion of the individual measures. Separate targets should be set for each such disaggregated outcome measure to provide a more reasonable set of targets and to avoid excessive focus on easy-to-resolve charges or complaints. Urban Institute Evaluation of EEOC’s Performance Measures | Page 14 Strategic Objective I: Combat employment discrimination through strategic law enforcement 1. Percent of EEOC’s administrative and legal resolutions that contain targeted non-monetary, equitable relief. This is Measure 6 in the Strategic Plan for Fiscal Years 2012-2016. The measure would be based on definitions developed to determine whether resolutions met the targeted, equitable relief criteria as articulated in the new Quality Control Plan. Breakouts should be provided for systemic and non-systemic cases.6 Because of the importance of systemic cases, EEOC might consider including “percent of systemic cases resolved successfully” as a separate, distinct, measure and not only as a breakout of this measure. (Some EEOC staff and officials were concerned, legitimately so, that this measure can have the unintended effect of encouraging selection of the easiest cases. This is a standard problem with outcome measurements expressed as percentages, including EEOC’s new Strategic Plan Measures #6 and #7.) Despite the potential perverse incentive, achieving successful resolutions in systemic, as well as in non-systemic cases, has to be a major concern for EEOC. To alleviate this problem, outcomes can be grouped by case difficulty. Our interviews with EEOC officials indicated that EEOC management routinely considers case difficulty (as well as available resources) in case selection. EEOC management already has the responsibility of achieving a balance between ”easy” and “difficult” cases. In addition, management can play an important role (and already seems to do so) in encouraging staff to see performance measures as management tools rather than as ways to criticize staff. Providing clear, well-anchored definitions for each of the terms “targeted,” “equitable,” and “relief” will be important for assuring that the data are reasonably reliable. It is also important to make this clear throughout the agency and to the outside world. For example, the current strategic plan states clearly that this relief must include non-monetary relief in order to be considered “equitable.” However, our interviews indicated that this exclusion of monetary-only relief is not universally known by EEOC staffs—and not likely to be understood by persons outside EEOC who do not have a legal background. Thus, and as recommended by the General Counsel, the word non-monetary is added above, and in Measures 2 and 3, to assure that this meaning is clear. 2. Percent of resolutions by FEPAs that contain targeted non-monetary, equitable relief. This is Measure 7 in the Strategic Plan for Fiscal Years 2012-2016. This measure would be based on definitions developed to determine whether resolutions met the targeted, equitable relief criteria. As noted above, providing well-anchored definitions for each of the terms “targeted,” “equitable,” and “relief” will be vital for assuring that the data are reasonably reliable. 3. Percent of federal sector hearing and appeal resolutions in which there has been a finding of discrimination or a settlement that contains targeted non-monetary, equitable relief. Like the two previous measures, this measure needs to be based on clear and detailed definitions to determine whether resolutions met the targeted, equitable relief criteria. 6 The current strategic plan defines systemic cases as “pattern or practice, policy, and/or class cases where the alleged discrimination has a broad impact on an industry, occupation, business, or geographic area.” Urban Institute Evaluation of EEOC’s Performance Measures | Page 15 4. Number of discrimination victims awarded monetary benefits. The information would come from the EEOC database. This measure does not include the many other cases where discrimination has occurred but has not been reported or has not led to monetary relief. Its advantage is that the data are already available to EEOC. 5. Amount of monetary benefits (financial relief) awarded to discrimination victims. The information should be available in the EEOC database (and was reported on the EEOC website for fiscal 2011). This measure has been regularly reported by EEOC. Its drawbacks are that this measure does not include non-monetary relief and can be greatly affected by a very small number of very large awards. 6. Number of direct recipients of monetary and non-monetary (equitable) relief, by type of relief. This metric helps the EEOC quantify the number of victims of employment discrimination who have been compensated by work conducted by the EEOC. These data are available from the EEOC database. This measure does not capture other employees who indirectly benefited. 7. Number and percent of charges that resulted in either: (a) a settlement (through ADR/mediation); or, among those classified as meriting relief, (b) a satisfactory settlement through conciliation (after a determination by investigators that discrimination had occurred), or (c) a litigated award. The denominator would include all charges other than those charges that were neither sent to mediation nor classified as meriting relief. This measure would use the relief definitions identified in the new Quality Control Plan called for in the Strategic Plan for Fiscal Years 2012-2016 and EEOC’s Integrated Mission System (IMS). Breakouts should be provided for systemic and non-systemic cases. 8. Percent of litigated cases that ended favorably to the EEOC position. This key measure addresses the important work of the office of the General Counsel and its litigation attorneys. OGC has regularly tracked the “success rate” for all litigation. 9. Number of employers found to have violated employee discrimination laws that have a charge filed against them within, three years of the resolution of the first charge and that resulted in a cause finding. This measure would help EEOC assess “recidivism” of employers for whom charges have been litigated successfully in the past. In theory, a lower recidivism rate over time should indicate success in preventing new incidents of discrimination. 10. Number and percent of charges reviewed by expert reviewers that meet EEOC quality standards and that have been properly assigned to EEOC level categories (e.g., A, B, and C). This measure is similar to Measure 2 in the current strategic plan. The measure would use the criteria and rating procedures identified by the new Quality Control Plan. A key to the credibility of this measure is the extent to which quality control procedures are applied to the measurement procedure so that “peer review” ratings are reliable. Notes 1. Much of the information needed for the above measures would come from existing sources and is available through the EEOC’s Integrated Mission System (IMS). The measures should be collected and reported using clearly stated and easy-to-locate definitions. Changes in defining Urban Institute Evaluation of EEOC’s Performance Measures | Page 16 terms such as “targeted, equitable relief” may later preclude comparisons over time with future- year values based on definitions derived from the emerging Quality Control Plan. 2. Not included in this list are measures such as “number of charges resolved.” EEOC uses this indicator and should continue to do so. However, the information from this measure does not indicate how many of these charges were resolved successfully.7 If significant numbers of resolved charges were resolved in ways not desired by EEOC, this would flag a potentially serious problem that EEOC would likely want to address. 7 For example, resolutions include settlements, withdrawals with benefits, no cause findings, and both successful and unsuccessful conciliations. In the case of mediation, this number would only indicate that the case was settled between the parties, not anything about the nature of the settlement. Urban Institute Evaluation of EEOC’s Performance Measures | Page 17 Strategic Objective II: Prevent employment discrimination through education and outreach 1. Percent of employable members of the public reporting they are reasonably clear about people’s rights relating to employment discrimination. Those interviewed might also be asked to rate their certainty that they have the correct information. The data can be collected in a national survey. To make the annual survey cost more manageable, the survey might be funded with Department of Labor or Department of Justice support or be added as a module to an existing annual national survey.8 This same survey could include questions for Measure 2 in the later discussion of measures for the “Overall Strategic Objective.” Existing national surveys often consider the addition of supplemental modules. Modules proposed to the survey administrator generally have to go through a thorough review process to ensure the questions won’t impact the primary survey responses. Many times the value of the added questions is also considered in screening an additional module. Alternatively, omnibus surveys are specifically designed for the addition of modules by multiple organizations, and there is no oversight or special consideration needed by the organization administering the survey. Appendix E provides a description of existing national surveys and omnibus surveys that specifically offer provisions and guidelines for the additions of supplemental questions. The footnotes in the Appendix provide links to documents describing the procedures and guidelines for the addition of supplemental modules. Such self-reports by no means guarantee they are adequately informed as to their rights. Such data, particularly when tracked over time, probably provides a reasonable, though rough, perspective on the public’s knowledge of employment discrimination. An option is to use a considerably more extensive, and expensive, survey such as that reported in EEOC’s study of customer satisfaction and knowledge of existing employment discrimination law.9 2. Percent of federal employees reporting they are reasonably clear about employees’ rights relating to employment discrimination. Employees might also be asked to rate their certainty that they have the correct information. The data could be collected through the Office of Personnel Management’s survey of federal employees. OPM might be amenable to adding, or revising, a few questions. 3. Percent of employers reporting they are reasonably clear about their responsibilities relating to employment discrimination. Employers might also be asked to rate their certainty that they have the correct information. They might also be asked if they would like more information on their responsibilities. The data can be collected in a national survey of employers. To make the annual data gathering more feasible, the survey might be funded with the Department of Labor or added as a module to an existing annual national survey of employers. EEOC already has a complete list of employers from which it could draw a sample, from its annual “EEO-1” reports (see Appendix E for more information). This measure is focused on private employers; however, a version of this measure might also be appropriate for public sector employers. 8 It is not uncommon for some major surveys to allow different agencies to “piggy-back” by adding special questions. 9 Federal Consulting Group, 2008. Urban Institute Evaluation of EEOC’s Performance Measures | Page 18 These data will likely provide a reasonable, though rough, perspective on the extent to which employers feel they have adequate knowledge of their responsibilities relating to discrimination. An option is to use a considerably more extensive, and expensive, survey of employers, perhaps based on a variation of the above mentioned survey of customer satisfaction and knowledge of EEO law. 4. Number of significant partnerships with organizations that represent vulnerable workers and/or underserved communities and that produced meaningful accomplishments/products. Accomplishments might include, for example, development of appropriate and substantive guidelines and/or training sessions for their memberships. To provide reliable measurements, specific definitions of minimum requirements are needed for being a “significant” partnership and for providing “meaningful accomplishments/products.” 5. Number of significant partnerships with organizations that represent small or new business (or with businesses directly) and that produced meaningful accomplishments/products. Accomplishments might include, for example, development of appropriate and substantive guidelines and/or training sessions for their memberships. To provide reliable measurements, specific definitions of minimum requirements are needed for being a “significant” partnership and for providing “meaningful accomplishments/products.” Urban Institute Evaluation of EEOC’s Performance Measures | Page 19 Strategic Objective III: Deliver excellent and consistent service through a skilled and diverse workforce and effective systems All stakeholders interviewed for this evaluation agreed that customer service, including timeliness, is important to the work of EEOC. Though these measures are more accurately considered to address intermediate outcomes rather than end outcomes, they are sufficiently important to many EEOC stakeholders that they should be reported, preferably at least once per year. Measures of service timeliness 1. Number of pending charges and complaints at a specified point of time broken out by priority level. This is the size of the pending-charge inventory and represents the size of the backlog. It is important that this measure be disaggregated by charge category as well as by priority level. The data would come from the EEOC IMS data base. It would also be useful to provide separate, disaggregated, data for each sector (e.g., the state and local government sectors, the private sector, and the federal sector). 2. Average length of time for resolution of those charges resolved during the reporting period (between entry of charge and its “resolution”) over: (a) all charges; and (b) all complaints from federal employees. To address concerns by EEOC officials that time to case resolution is meaningless without knowing the complexity of the case, this measure should be disaggregated by priority category, by level of difficulty as rated by a supervisor at entry, and by discrimination category. The data would come from the EEOC database. Separate targets should be set for each such disaggregated group. A good option is to measure the median, rather than the average, length of time for resolution, disaggregated by discrimination category and level of difficulty. This measurement form avoids the effects of extreme resolution time values. 3. Percent of (a) charging parties and (b) federal employees who have requested hearings or filed appeals with EEOC who perceived that there were substantial unneeded delays in resolving their case. The data would come from surveys of (a) a random sample of charging parties and (b) a random sample of federal complainants whose charges had been resolved in the period for which data are being reported (probably surveyed electronically). 4. Percent of (a) respondents and (b) federal agencies whose cases were handled by the EEOC who perceived that there were substantial unneeded delays in resolving their case. The data would come from random samples of respondents – both (a) private, state, and local, and (b) federal agencies – whose charges had been resolved in the period for which data are being reported (probably surveyed electronically). Measure of access to information on charges/cases 5. Percent of (a) charging parties and (b) federal employees who have requested hearings or filed appeals with EEOC who reported having had significant problems in obtaining information on their charge or complaint. This information can be obtained from the same surveys as those used to obtain ratings of the timeliness of charge processing (that is, by surveying random samples of charging parties and of federal employee complainants whose charges had been resolved in the period for which data are being reported). Urban Institute Evaluation of EEOC’s Performance Measures | Page 20 6. Percent of respondents who reported having had significant problems in obtaining information on their charge or complaint. This information can be obtained from the same surveys as those used to obtain ratings of the timeliness of charge processing (that is, by surveying random samples of respondents and federal agencies whose charges had been resolved in the period for which data are being reported). Measures of courteousness 7. Percent of charging parties who reported satisfaction with the courteousness of EEOC personnel with whom they had communicated. This information can be obtained in the same survey as that used to obtain ratings of the timeliness of charge processing and adequacy of information on the status of the charge (that is, by surveying a random sample of charging parties whose charges had been resolved in the period for which data are being reported). 8. Percent of respondents who reported satisfaction with the courteousness of EEOC personnel with whom they had communicated. This information can be obtained from the same survey as that used to obtain ratings of the timeliness of charge processing and the adequacy of information on the status of the charge (that is, by surveying a random sample of respondents whose charges had been resolved in the period for which data are being reported). The above two measures may not be felt to be of sufficient importance to EEOC to warrant inclusion in its external reporting. However, if the surveys of charging parties or respondents are being conducted to obtain other information, the added cost for covering this customer service attribute would likely be quite small and would at least be useful for internal management. The cost of such surveys does not need to be large, especially with the potential for conducting them through the internet, at least for surveying businesses. Some reviewers of a draft of this report expressed concern about the meaningfulness and accuracy/validity of measurements obtained from surveys of customers, as required for a number of the above measures. Such measures have been widely used, and reported, by many federal agencies (such as in their Performance Accountability Reports). The premise is that businesses and charging parties are important “customers” of EEOC, and their perceptions are very important to EEOC. The views of customers, even those that are unhappy with the results of the services, are well worth counting. At noted earlier, at least one of the field offices, we found from our interviews has conducted such surveys. As noted earlier, EEOC already has been surveying both charging parties and respondents who participated in the mediation process, and at least one district office is planning a surveys of those using its services. To make customer surveys more useful we recommend that respondents who give poor ratings be asked to explain why and to provide suggestions to EEOC as to how to improve its work. Many government agencies have found that often much useful information can be obtained for improving services. This can be a major side benefit of surveys of customers. Urban Institute Evaluation of EEOC’s Performance Measures | Page 21 Overall Objective: Reduce National Employment Discrimination Levels The mission of EEOC, as quoted from its Strategic Plan for Fiscal Years 2012-2016, p. 12 is to “Stop and Remedy Unlawful Employment Discrimination.” These words indicate that progress in reducing national employment discrimination should be a major concern for EEOC. Yet, no measures are presented in the current strategic plan, or elsewhere, that reflect progress towards reducing employment discrimination in the United States. This is a badly neglected area of EEOC reporting, perhaps in part because of the limited control that EEOC has over these national levels—and the presumed difficulty and cost of collecting data on progress. However, the levels are likely to be of primary concern to citizens as well as to the President and Congress. Should not EEOC be the primary agency reporting on this major national goal? We emphasize, as EEOC should continue to do, that when reporting on national employment discrimination levels, many factors affect discrimination. EEOC by itself, and with its highly limited budget, can only do so much. However, seeking national data on the incidence of employment discrimination would seem to be an important task for EEOC because of its central role in helping reduce such discrimination. We note that many federal agencies annually report major national outcomes over which they have only limited control (such as DOJ-FBI’s annual crime reports; DOED’s national high school graduation reports; and DOT-NHTSA’s annual traffic mortality reports). Obtaining exact data on discrimination levels is very difficult, if not impossible, and expensive. However, reasonably practical ways to approximate the levels, or at least track progress in reducing discrimination, appear possible. Below we suggest a number of options for obtaining reasonable data on the levels of employment discrimination in the United States. We do not suggest that only one of these should be selected but rather that a number of these considered together would each provide useful national information on at least major population groups. Each measure provides a different, though incomplete, perspective. Some of these measures use data already available to EEOC. 1. Number of charges that led to a resolution that discrimination had occurred or for which mediation led to some form of relief being provided. The information would come from the EEOC database. This information is a quite weak indicator of the amount of actual existing employment discrimination. It does not include the number of charges that are closed at the request of charging parties, nor whether discrimination actually occurred even though EEOC did not find reasonable cause. However, its advantage is that at least it provides a count of the number of “proven” discrimination cases—and the data are readily available. Tracked over time, the data would become a somewhat stronger indicator of changes in the amount of employment discrimination in the nation. The measure should be disaggregated by such characteristics as the priority level, whether systemic or not, and by category of employment discrimination. To cover federal employment, the “number of cases that led to a resolution that discrimination had occurred” would be measured separately. These data should be available from the OFO’s data system. The “number of charges filed,” while useful to EEOC as a workload indicator, could also be considered an indicator of national discrimination (and the data are already available). However, this measure would be a much weaker indicator of national discrimination. Its drawback is that not all charges have a legal basis. It is not recommended as an outcome measure. Urban Institute Evaluation of EEOC’s Performance Measures | Page 22 2. Percent of members of the public reporting that they or someone they know had experienced workplace discrimination in the last 12 months, by category of employment discrimination. This information would be obtained from a national survey of a sample of individuals (perhaps funded with Department of Labor, DOJ, and/or HUD, which could include housing discrimination in the survey). The survey with carefully constructed question wording could potentially provide the most comprehensive picture of nationwide employment discrimination. Employment discrimination questions might also be added as a module to another existing annual national survey (see Appendix E and earlier discussion under Strategic Objective II). Somewhat complicating the sample design is the desire to over-sample vulnerable populations, such as particular racial/ethnicity groups. Respondents might also be asked to rate their certainty that discrimination occurred. While this limited information may not be very satisfying, such data, particularly when tracked over time, seems likely to provide a rough perspective on national discrimination levels. And this information is considerably more likely to be practical for EEOC to obtain on an annual basis than if the agency sought considerably more extensive information. This measure has important limitations. The public’s understanding of their legal employment discrimination rights undoubtedly is limited. In some instances they will believe they have been discriminated against when under the law they have not been. In other cases, they will not know that they have been the victim of discrimination. There is no question that this will yield imperfect data. Nevertheless, this may well provide as good a perspective on extent of employment discrimination in the United States as can be obtained. Any reports from this survey should make clear this limitation. The other major problem is the potential cost of the survey to obtain this information. The earlier discussion of Measure 1, under Strategic Objective 2, suggested ways to substantially reduce the cost. We repeat that information here. To make the annual survey cost more manageable, the survey might be funded with Department of Labor and/or Department of Justice support or be added as a module to an existing annual national survey, such as the American Community Survey (suggested as a possibility by one of the reviewers of our draft report).10 (This same survey could include questions for Strategic Objective 2, Measure 1.) Some national surveys consider the addition of supplemental modules. Modules proposed to the survey administrator generally have to go through a review process to ensure the questions won’t impact the primary survey responses. Many times the value of the added questions is also considered in screening an additional module. Alternatively, omnibus surveys are specifically designed for the addition of modules by multiple organizations, and there is no oversight or special consideration needed by the organization administering the survey. Appendix E provides a description of existing national surveys and omnibus surveys that specifically offer provisions and guidelines for the additions of supplemental questions. The footnotes in the Appendix provide links to documents describing the procedures and guidelines for the addition of supplemental modules. 3. Percent of federal employees reporting that they have been discriminated against in the last 12 months by category of discrimination. The data could be broken out by selected employee 10 It is not uncommon for some major surveys to allow different agencies to “piggy-back” by adding special questions. Urban Institute Evaluation of EEOC’s Performance Measures | Page 23 demographics (such as age, sex, and race/ethnicity categories). Survey questions could be added to OPM’s Annual Employee Survey or could replace some of the existing questions that only peripherally relate to discrimination, such as the following from the latest OPM survey: - Question 17. “I can disclose a suspected violation of any law, rule or regulation without fear of reprisal.” - Question 34. “Policies and programs promote diversity in the workplace (for example, recruiting minorities and women, training in awareness of diversity issues, mentoring).” - Question 38. “Prohibited Personnel Practices (for example, illegally discriminating for or against any employee/applicant, obstructing a person’s right to compete for employment, knowingly violating veterans’ preference requirements) are not tolerated.” - Question 45. “My supervisor/team leader is committed to a workforce representative of all segments of society.” This measure considers only one employment sector, federal employment, but this is a large and important group for EEOC and the country. 4. The extent to which demographic groups of concern to EEOC, in the (a) national and (b) federal workforce, reflect the actual composition of the US working population. This might be calculated as the percentage point difference between the percent of persons within a specific demographic group who are employed in particular positions (level/rank, salary, occupational and industry category) and the national estimate of that same group within the working age population as a whole. Calculations for the federal workforce are likely to be readily available. For the national workforce, data are available through the EEO dataset collected by the Census as part of the American Community Survey. The data provides detailed breakdowns of the workforce by race, sex, occupation, educational attainment and other factors that could be useful in comparison to the working population as a whole. Differentiation in the distribution amongst occupations or industries when compared to the distribution of the entire workforce would provide partial evidence of discrimination. OFO has indicated strong interest and has worked on this issue in the past. This measure has the drawback that it does not directly measure employment discrimination. It only indicates the potential that discrimination does exist. Other factors, such as education, might explain the difference. 5. Percent of paired-comparison testing cases in which discrimination occurred, for particular types of employment discrimination for which paired-comparison testing is feasible, especially those relating to hiring practices. Two procedures have been used: (1) In-person paired testing in which pairs of individuals apply for employment presenting credentials and capacities, differing solely on their race/ethnicity, gender, or other characteristic protected by law; and (2) Resume paired testing where equivalent resumes are randomly assigned, for example, as “white-sounding” or “black-sounding” names and submitted to employers. The differences between employer responses to the applications provide evidence of discrimination. Pairs of Urban Institute Evaluation of EEOC’s Performance Measures | Page 24 individuals do not need to be used and trained, thus, resume-testing significantly reduces the costs associated with conducting a national test for discrimination. Some limited role-play tests of discrimination have been carried out some years ago, though not to our knowledge by EEOC. (See Appendix F for more details.) We discussed with most of our interviewees surveys and role-play (“testing”) as potential sources for such measures of employment discrimination nationwide. Most interviewees liked the concept of such a process for proving strong evidence of discrimination but recognize the difficulties, and cost, of its use. As described in Appendix F, in-person paired testing requires highly specialized and expensive procedures undertaken by contractors. However, the process provides strong evidence about discrimination levels, especially those relating to hiring practices. Joint sponsorship with Department of Labor and/or Department of Justice would alleviate the cost problem. Resume paired-testing places the emphasis on the preparation of questionnaires that differ only by clues as to the applicant’s demographic group, such as name, where the applicant went to school, and the organizations in which the applicant reports having participated. This practice is more recent. It has the considerable advantage that it is much less expensive. The evidence obtained is weaker than that from in-person testing but still is likely to produce strong measures of discrimination. HUD has demonstrated that in-person paired testing can be an effective instrument to estimate housing discrimination nationally (Simonson & Wienk 1984; Cross, Kenney, Mell, and Zimmermann 1990; Turner, Struyk & Yinger 1991; Turner et al. 2002). No national testing estimates exist for discrimination in employment. Two paired testing pilot studies conducted by the Urban Institute in the early 1990’s provide evidence that paired testing methodology can be adapted to help test employment discrimination (Cross et al. 1990; Turner et. al. 1991). Paired testing for employment discrimination up to this point has been infrequent, limited in scale, directed at individual cities or regions. Alternatively, Resume-paired testing is a relatively new strategy that is unique to employment testing and has been less reported thus far (Newman 1980; Bertrand and Mullainathan 2003). The pilot studies and resume tests are discussed in Appendix F. A national paired testing study could provide insight into fruitful testing strategies that the EEOC could use to set an “industry standard” for testing employment discrimination. Proven standard practices would allow local practitioners the ability to test and litigate employment discrimination claims more effectively locally and help decrease the burden of litigation placed on the EEOC. In-person-paired is likely to be impractical for EEOC. EEOC, however, might want to encourage ad hoc testing in locations where discrimination is believed likely to be large. Funding might be provided by private sector organizations such as national or community foundations. Resume-paired testing shows promise and appears considerably more feasible. However, further analysis is needed to find out what such testing can do and what it cannot do, as well as estimation of its cost. Urban Institute Evaluation of EEOC’s Performance Measures | Page 25 Measurements of “Efficiency” Program costs have become a critical issue throughout the federal government. This, of course, includes EEOC. A greater focus on efficiency in future years can be expected by the federal government.11 However, neither the EEOC strategic plan nor the PAR includes any measures of efficiency. While measurement of efficiency is not the subject of this report, we would be remiss if we did not provide initial thoughts on the subject. The word “efficiency” has many different meanings. For this report, and for performance measurement purposes, efficiency is measured, as is typically done in federal government performance reports, as the “cost per unit of output or outcome.” (Cost can be expressed as either the amount of financial cost or in units of employee time, such as FTEs.) The danger is that performance measurements expressed as cost per unit of output, such as cost per charge, or cost per charge closed, may tempt employees to press for closures at the expense of quality. However, it is likely that EEOC managers will want to track such indicator values as “number of employees per case” to estimate workload, but this is not a measure performance. If, and when, EEOC introduces efficiency measures, EEOC should consider including cost per unit of outcomes rather than only cost per unit of output. Such a measure would be in the form: “Cost per charge, or case, closed for which the outcome was favorable (e.g., provided targeted, equitable relief).” 11 EEOC’s FY2012 “Performance and Accountability Report (PAR),” p.25, addresses the concern under the heading “Serving the Public More Efficiently.” Urban Institute Evaluation of EEOC’s Performance Measures | Page 26 Overall Recommendations on Measures Each of the measures listed above, and listed together in Exhibit 1, measure a somewhat different EEOC outcome. Tracking each measure will provide important information on EEOC’s progress in reducing some important aspect of employment discrimination. Only two of these are the same as measures included in the currently strategic plan (Measures 6 and 7 on targeted, equitable relief – see Appendix C). However, at least close versions of some of the other measures recommended are already being reported regularly in EEOC reports, especially the measures for Strategic Objective I. The primary new data collection needed for these measures are: (a) surveys of citizens (for Strategic Objective II and IV); and (b) periodic “paired comparison testing” of employers in hiring employees (for Strategic Objective IV). Overall, we recommend that EEOC: 1. Review the measures; improve them; determine the priorities for testing and data collection; and begin tracking them. We understand that the multi-year research and data collection plan called for by the Strategic Enforcement Plan is an appropriate venue for discussing what is possible with EEOC’s resources. 2. Include such measures as these as soon as possible in its data collection systems and then in annual Performance and Accountability Reports (PAR). These are the outcomes that citizens, Congress, and OMB are most likely to consider most important. Progress on these measures should be a major focus of EEOC. 3. Plan to include at least a subset of such outcome measures as the principal outcome measures to be included in EEOC’s next strategic plan. The data obtained during the 2012-2016 years will provide baseline data for establishing targets for the next strategic plan. 4. Some of the candidate measures will require time to develop (perhaps a year or more). Thus, they can be considered EEOC’s next generation of strategic plan measures. However, we recommend that development and implementation of the procedures begin as soon as possible. 5. Begin introducing some of these measures, as appropriate, into updates of the current strategic plan. 6. These measures should not be considered as being solely for strategic planning purposes but should also be tracked at least annually, and preferably more frequently such as quarterly or at least semi- annually, and used for managing and continuous learning by EEOC personnel. Urban Institute Evaluation of EEOC’s Performance Measures | Page 27 Exhibit 1 Recommended Outcome Measures Strategic Objective 1: Combat employment discrimination through strategic law enforcement 1. Percent of EEOC’s administrative and legal resolutions that contain targeted, non-monetary equitable relief. 2. Percent of resolutions by FEPAs that contain targeted, non-monetary equitable relief. 3. Percent of federal sector hearing and appeal resolutions in which there has been a finding of discrimination or a settlement that contains targeted, non-monetary equitable relief. 4. Number of discrimination victims awarded monetary benefits. 5. Amount of monetary benefits (financial relief) awarded to discrimination victims. 6. Number of direct recipients of monetary and non-monetary (equitable) relief by type of relief. 7. Number and percent of charges that resulted in either: (a) a settlement (through ADR/mediation); or among those classified as meriting relief, (b) a satisfactory settlement through conciliation (after a determination by investigators that discrimination had occurred), or (c) a litigated award. 8. Percent of litigated cases that ended favorably to the EEOC position. 9. Number of employers found to have violated employee discrimination laws that have a charge filed against them within three years of the resolution of the first charge and that resulted in a cause finding. 10. Number and percent of charges reviewed by expert reviewers that meet EEOC quality standards and that have been properly assigned to EEOC level categories (e.g., A, B, and C). Strategic Objective 2: Prevent employment discrimination through education and outreach 1. Percent of employable members of the public reporting they are reasonably clear about people’s rights relating to employment discrimination. 2. Percent of federal employees reporting they are reasonably clear about employees’ rights relating to employment discrimination. 3. Percent of employers reporting they are reasonably clear about their responsibilities relating to employment discrimination. 4. Number of significant partnerships with organizations that represent vulnerable workers and/or underserved communities and that produced meaningful accomplishments/products. 5. Number of significant partnerships with organizations that represent small or new business (or with businesses directly) and that produced meaningful accomplishments/products. Urban Institute Evaluation of EEOC’s Performance Measures | Page 28 Exhibit 1 (continued) Strategic Objective 3: Deliver excellent and consistent service through a skilled and diverse workforce and effective systems Measures of service timeliness 1. Number of pending charges and complaints at a specified point of time broken out by priority level. 2. Average length of time for resolution of those charges resolved during the reporting period (between entry of charge and its “resolution”) over: (a) all charges; and (b) all complaints from federal employees. 3. Percent of (a) charging parties and (b) federal employees who have requested hearings or filed appeals with EEOC who perceived that there were substantial unneeded delays in resolving their case. 4. Percent of (a) respondents and (b) federal agencies whose cases were handled by the EEOC who perceived that there were substantial unneeded delays in resolving their case. Measure of access to information on charges/cases 5. Percent of (a) charging parties and (b) federal employees who have requested hearings or filed appeals with EEOC who reported having had significant problems in obtaining information on their charge or complaint. 6. Percent of respondents who reported having had significant problems in obtaining information on their charge or complaint. Measures of courteousness 7. Percent of charging parties who reported satisfaction with the courteousness of EEOC personnel with whom they had communicated. 8. Percent of respondents who reported satisfaction with the courteousness of EEOC personnel with whom they had communicated. Overall Strategic Objective: Reduce National Employment Discrimination Levels 1. Number of charges that led to a resolution that discrimination had occurred or for which mediation led to some form of relief being provided. 2. Percent of members of the public reporting that they or someone they know had experienced workplace discrimination in the last 12 months, by category of employment discrimination. 3. Percent of federal employees reporting that they have been discriminated against in the last 12 months by category of discrimination. 4. The extent to which demographic groups of concern to EEOC, in the (a) national and (b) federal workforce, reflect the actual composition of the US working population. 5. Percent of paired-comparison testing cases in which discrimination occurred, for particular types of employment discrimination for which paired-comparison testing is feasible, especially those relating to hiring practices. Urban Institute Evaluation of EEOC’s Performance Measures | Page 29 IV. EEOC Use of Performance Measures: Findings & Recommendations Critical to the value of performance measures is their usefulness, not only for accountability purposes but also as a management tool to allow the EEOC and its managers to better allocate resources and optimize effectiveness. Therefore it is important to assure that measurement data are not only potentially useful but that they are actually used. Our interviews indicated that periodic managerial meetings are very much part of the regular functioning of the EEOC and its managers. However, it does not appear that these meetings currently contain systematic reviews of the progress reflected by the latest performance measurement values, such as do the quarterly data-driven reviews required by OMB for large agencies’ priority measures. Stakeholders identified a variety of reports, produced typically on an annual or quarterly basis, that include data on EEOC performance.12 Some of these reports include extensive data, usually from the IMS or the OFO’s IMS equivalent. The internal reports are often broken down by field office and include the number of charges and cases, the number and percent of resolutions, the number of people who benefit both monetarily (including money amount) and non-monetarily, the types of resolution by charge categories, and similar data on mediations carried out under the ADR program. Internal data such as caseloads, number of staff, and transfers among offices are also provided. However, several interviews demonstrated considerable uncertainty about precise definitions of some of the measures. Our review of reports confirmed that not all terms are clearly defined and easy to find. It is unclear how the new measures in the Strategic Plan for Fiscal Years 2012-2016, many of which are insufficiently defined or difficult to quantify, will be integrated into the reports and tracked on a regular basis. According to one stakeholder, the Government Performance and Results Act Modernization Act (GRPA MA) requires that the Office of Research, Information, and Planning (ORIP) review all documentation of performance measures to ensure quality control. The stakeholder noted that this process has become much more difficult and time consuming with the new strategic plan measures, adding, “as much as everyone hated the old measures, we still could run numbers and scenarios – there were milestones for each performance metric.” Another official affirmed that the new strategic plan measures would be incorporated into regular reports “once conceptualized,” but also added that the official’s office would retain many of the old measures (such as the measure of case completion times) because they are useful to the office. There is clear desire among EEOC leadership to integrate performance information into routine processes and to use it to “identify weak performers and focus efforts” more strategically. As the set of measures increasingly addresses the core objectives of the agency (such as: the number of resolutions that provided targeted or meaningful relief; the level of public knowledge about discrimination; the levels of discrimination; and customer satisfaction), high level meetings to discuss performance will become increasingly valuable. 12 These documents include: the Federal EEO Complaints Processing Reports; the large statistical reports issued quarterly by the Office of Research, Information, and Planning (ORIP); performance agreements between district offices and central Office of Field Programs, such as Performance Management Agreement for Senior Executive Service staff; and Performance and Accountability reports (PARs). Urban Institute Evaluation of EEOC’s Performance Measures | Page 30 There is movement in that direction. The Strategic Enforcement Plan (SEP), which was approved in late 2012, asks office leaders to meet quarterly to discuss “implementation of the SEP” (page 22 of the SEP). Those meetings could be expanded to incorporate review of the performance measures identified in the Strategic Plan for Fiscal Years 2012-2016. Each measure is likely to be assigned to an office that will serve as its formal “goal leader.” For example, the Office of Field Programs is responsible for measures 6, 7, 8, 9, and 13, and co-responsible for measure 3 with OFO, the office that handles appeals by federal workers. These efforts to establish a regular process are particularly laudable in light of the fact that the EEOC, as a smaller federal agency, is not required by OMB to set high priority goals or to convene quarterly meetings to review them.13 Recommendations The following recommendations will likely be considerably more useful to EEOC management and Commissioners if EEOC includes such outcome measures as those identified in Section III. We believe these recommendations are about good management. They should require little added work. These recommendations pertain to both the Commissioners’ Office and to internal EEOC management. 1. While not currently required by OMB, EEOC management would likely benefit considerably if it uses the type of quarterly data-driven reviews that OMB requires for the large federal agencies. These would be regularly scheduled sessions with EEOC’s major units to review systematically the latest performance information on both the process and outcome measures. This is a variation of the “PerformanceStat” process, used to track and improve the effectiveness and efficiency of an agency’s programs already being used by a number of federal agencies, including the Department of Housing and Urban Development, the Department of Veterans Affairs, the Environmental Protection Agency, the Federal Aviation Administration, the Food and Drug Administration, the Internal Revenue Service, and the National Aeronautics and Space Administration. The EEOC version need not be as extensive or elaborate as are some of the federal department efforts. The process involves the use of regularly scheduled, structured, data-driven (and typically quantitative) meetings to review the outcome and process measures with department or program personnel.14 It is adaptable to several levels within agencies, from the Executive leaders to individual offices and programs, and occurs in several forms depending on the goals of the agency. These need not be additional meetings: this might just mean using performance data as a way to focus discussion at regular EEOC review meetings already scheduled. The classic structure, such as used by the FDA, involves separate executive meetings with each individual reporting unit. This organization can be useful in ensuring that offices and programs focus on their core mission and encourages participants to be more forthcoming about problems than meetings held with other offices within the agency. Depending on the organization of their agency, some leaders have used a model that includes joint participation with managers from sister offices.15 This arrangement may be particularly applicable to OFP in designing a review of district offices and FEPAs. 13 However, the Director of ORIP attends the Performance Improvement Council (PIC). 14 Specific suggestions on these procedures are provided in Hatry and Davies, 2011. 15 See Patusky, Botwink and Shelley, 2007 Urban Institute Evaluation of EEOC’s Performance Measures | Page 31 Another way of organizing sessions, such as used by HUD, is by thematic areas (such as alternative dispute resolutions, conciliation, and litigation). This thematic organization can help: (a) encourage joint responsibility for performance between reporting units; (b) foster communication among agencies jointly responsible for meeting preparation; (c) make finished products more accessible and easier to understand for citizens; and (d) ensure that blame is not shifted to other reporting units not present during a meeting. Unfortunately, this approach also makes it difficult to define who is responsible for performance.16 The EEOC already has a number of regularly scheduled meetings that would be well suited to incorporating this data-driven review process. For instance, the Chief Operating Officer holds regular meetings with the head of each office. Similar meetings are held at the District Office level as well. At these meetings a portion of the time could be allocated to review the latest performance reports. Alternatively, at regular intervals, one of these meetings could be fully dedicated to a review of the recent data on the performance measures most relevant to that office. 2. EEOC should request that relevant disaggregations of the outcome measure values be provided to its managers so that findings for each measure are readily available, not only in aggregate but also by key breakout categories. Such information will very likely provide considerably more informative and more actionable information to EEOC managers than can be obtained by only an examination of the aggregate data. Such disaggregations might, for example, include the form of discrimination (e.g., religion, sex, age, race/ethnicity, etc., as identified in the statutes); sector (federal, state or local government, or private sector); coverage (systemic, class action, or single incident); geographical region; priority level; employer NAICS code; and/or method (mediation, formal charging). Such categorizations for most measures are available through the information systems of OFP and OFO. Each EEOC district office should continue to be provided regular performance reports covering its own workload (such as the charges that the office has received).The performance reports might use hyperlinks so that a manager can readily obtain disaggregated data by the breakout categories selected for each measure. 3. EEOC offices should be asked to provide explanations for unexpectedly poor, or good, measurement data values as part of performance reports. The major purpose of performance measurement is to help suggest ways to improve the quality of services. Obtaining explanations is a key step towards such improvements. Obtaining explanations often is not easy but should be attempted, even if only qualitative judgments are possible. Such explanations should be made available in both internal and external reports. 4. The performance reports should be provided in a clear, concise and understandable way to the staff and Commissioners. Obvious as this may seem, too often performance reports across the nation have been overly difficult to read and interpret. Charts and graphics can help display data clearly and show trends and comparisons. 16 More detail on these examples and other aspects of the Performance Stat process can be found in Hatry and Davies 2011. Urban Institute Evaluation of EEOC’s Performance Measures | Page 32 V. Conclusion Our evaluation of EEOC’s measures focused on: the measures in the latest strategic plan; those measures likely necessary for future strategic plans; and those measures likely to be important for EEOC management and Commissioners for tracking the outcomes on a regular basis for encouraging continual service improvement and learning. These are inter-related sets of measures. The current strategic plan measures are primarily process and not outcome measures with two important exceptions. They currently neglect coverage of progress towards the overarching purpose of EEOC: to help reduce employment discrimination in the United States. This report provides recommendations for a number of outcome measures for each of the three strategic objectives included in the EEOC strategic plan plus recommendations for also tracking progress towards the over-arching objective of reducing employment discrimination in the United States. The latter measures, though particularly difficult to measure fully satisfactorily, nevertheless warrant major attention by EEOC. Suggestions as to data sources and data collection procedure are identified for the measures. Finally, the report provides a number of basic recommendations for improving the usefulness and use of the outcome information, drawing on the new focus of the GPRA Modernization Act of 2010. Examples include: the use of regular data-driven reviews with agency staff; the development of reports that provide agency officials with ready access to considerably more targeted disaggregated data, such as outcomes by priority level, by demographic characteristics of complainants, by industry, and by district office; and including as part of each performance report, explanations for unexpected results. Urban Institute Evaluation of EEOC’s Performance Measures | Page 33 APPENDICES Urban Institute Evaluation of EEOC’s Performance Measures | Page 34 APPENDIX A List of Interviews EEOC Commissioners & Staff Director, Compliance and Control Division Constance Barker Dexter Brooks Commissioner Director, Federal Sector Programs Kimberly Essary EEOC Office of Research, Information and Planning Chief of Staff to Commissioner Constance S. Barker Deidre Flippen Director, Office of Research, Information Chai R. Feldblum and Planning, EEOC Commissioner EEOC Office of Field Programs Sharon Masling Chief of Staff for Commissioner Feldblum Nick Inzeo Director Steven P. Zanowic Special Assistant, Office of Commissioner Katharine Kores Feldblum Director, Memphis District Office Claudia Withers Spencer Lewis Chief Operating Officer, EEOC Director, Philadelphia District Office EEOC General Counsel Beverly L. Watts P. David Lopez Executive Director, Tennessee Human General Counsel Rights Commission James Lee The Urban Institute Deputy General Counsel Margery Turner Vice President for Research Gwendolyn Reams (a national expert on paired-comparison Associate General Counsel testing) Leslie Annexstein United States Office of Personnel Management Senior Attorney, Advisor to the General Veronica Villalobos Counsel Jennifer Goldstein Senior Attorney, Advisor to the General Counsel EEOC Office of Federal Operations Carlton Hadden Director Robert Barnhart Urban Institute Evaluation of EEOC’s Performance Measures | Page 35 APPENDIX B References & Documents Reviewed Abravanel, Martin D and Mary Cunningham. 2002. How Much Do We Know?: Public Awareness of the Nation’s Fair Housing Law. Washington, D.C.: The Urban Institute. Abravanel, Martin D. 2006. Do We Know More Now?: Trends in Public Knowledge, Support and Use of Fair Housing Law. Washington, D.C.: U.S. Department of Housing and Urban Development. American Association for Public Opinion Research. 2009-2010 Blue Book of Agencies and Organizations that Support AAPOR and WAPOR. Available online at http://www.aapor.org/Blue_Book1.htm. Benard, Stephen, In Paik, and Shelley J. Correll. 2008. “Cognitive Bias and the Motherhood Penalty.” Hastings Law Journal 59:101-129. Bertrand, M. and S. Mullainathan. 2003. Are Emily and Greg More Employable than Lakisha and Jamal?: A Field Experiment on Labor Market Discrimination. NBER Working paper 9873. Cross, H., G. Kenney, J. Mell, and W. Zimmermann. 1990. Employer Hiring Practices: Differential Treatment of Hispanic and Anglo Job Seekers. Washington D.C.: Urban Institute Press. Development Services Group (DSG), Inc. 2006. Evaluation of Intake and End of Fiscal Year Closure of the Equal Employment Opportunity Commission Private Sector Charge Process. Development Services Group (DSG), Inc. Epstein, P., J. Fountain, W. Campbell, T. Patton, and K. Keaton. 2005. Government Service Efforts and Accomplishments Performance Reports: A Guide to Understanding. Norwalk, CT: Governmental Accounting Standards Board. Federal Consulting Group (FCG). 2008. Equal Employment Opportunity Commission Customer Satisfaction and Knowledge of Law Study Final Report. Federal Consulting Group (FCG). ---. 2010. U.S. Equal Employment Opportunity Commission Evaluation of the Priority Charge Handling Procedures Report. Federal Consulting Group (FCG). GPRA Modernization Act of 2010. Public Law 111–352—JAN. 4, 2011. Available online at http://www.gpo.gov/fdsys/pkg/PLAW-111publ352/pdf/PLAW-111publ352.pdf Hatry, H., and Elizabeth Davies. 2011. A guide to data driven performance reviews. Washington, DC: IBM Center for the Business of Government. Hatry, H., P. S. Schaenman, D. M. Fisk, J. R. Hall Jr., and L. Snyder. 2006. The Urban Institute and International City/County report “How Effective Are Your Community Services? Procedures for Performance Measurement,” third edition. Washington, DC: Urban Institute Press. Matos, K. and E. Galinsky. 2012. 2012 National Study of Employers. Families and Work Institute. http://familiesandwork.org/site/research/reports/NSE_2012.pdf Mincy, R. The Urban Institute Audit Studies: Their Research and Policy Context. Clear and Convincing Evidence: Measurement of Discrimination in America. M. Fix and R. Struyk. Washington D.C.: The Urban Institute Press, 1992. 165-186. Print. Newman J. M. 1980. “Discrimination in Recruitment: An Empirical Analysis,” Industrial and Labor Relations Review, 32 (1): 15-23. NORC at the University of Chicago. General Social Survey. http://www3.norc.org/GSS+Website/. ---. Frequently Asked Questions. http://publicdata.norc.org:41000/gssbeta/faqs.html. ---. Guidelines For Paid Supplements To The GSS. http://publicdata.norc.org:41000/gss/documents//OTHR/Paid%20Supplements2014.pdf. ---. Call For Proposals To Add Questions To The 2014 GSS. http://publicdata.norc.org:41000/gss/documents/OTHR/Module%20Competition.pdf Patusky, C., L. Botwink, and M . Shelley. 2007. The Philadelphia School Stat Model. Washington, DC: IBM Center for the Business of Government. Urban Institute Evaluation of EEOC’s Performance Measures | Page 36 Princeton Survey Research Associates International. PSRAI Omnbius Fact Sheet. http://psrai.com/pdf/psrai_telephone_omnibus.pdf Simonson, J. C. and R. Wienk. 1984. Racial Discrimination in Housing Sales: An Empirical Test of Alternative Models of Broker Behavior, unpublished manuscript. Social Science Research Solutions. http://www.ssrs.com/ Survey Technology & Research Center (STR). http://strcenter.com Turner, M., S. L. Ross, G. Galster and J. Yinger. 2002. Discrimination in Metropolitan Housing Markets: National Results from Phase I HDS 2000. Washington, D.C.: Urban Institute Press. Turner, M. and C. Herbig. Measuring the Extent and Forms of Discrimination in the Marketplace: Lessons from Paired- Testing Research. All Things Being Equal. Ed. B. Smedley and A. Jenkins. New York, NY: The New Press, 2007. 148-189. Print. Turner, M., M. Fix and R. Struyk. 1991. Opportunities Denied, Opportunities Diminished: Racial Discrimination in Hiring. Washington D.C.: Urban Institute Press. Turner, Margery A., Raymond J. Struyk and J. Yinger. 1991. Housing Discrimination Study Synthesis. Washington DC: Urban Institute Press. U.S. Census Bureau. 2006. Current Population Survey. Design and Methodology. Technical Paper 66. Also see http://www.census.gov/cps/methodology/ ---. U.S. Economic Census. http://www.census.gov/econ/census/ U.S. Equal Employment Opportunity Commission. 2011. Annual Report on the Federal Work Force Part I EEO Complaints Processing. Fiscal Year 2011. Available online at http://www.eeoc.gov/federal/reports/fsp2011/index.cfm ---. 2011. Fiscal Year 2012 Congressional Budget Justifications. Available online at http://www.eeoc.gov/eeoc/plan/2012budget.cfm ---. 2012. Fiscal Year 2013 Congressional Budget Justifications” Available online at http://www.eeoc.gov/eeoc/plan/2013budget.cfm ---. 2012. Meeting of February 22, 2012 - EEOC Strategic Plan for Fiscal Years 2012-2016. Available online at http://www.eeoc.gov/eeoc/meetings/2-22-12/transcript.cfm ---. 2012. FY 2012 Performance and Accountability Report. Available online at http://www.eeoc.gov/eeoc/plan/upload/2012par.pdf. ---. December 2012.United States Equal Employment Opportunity Commission Strategic Enforcement Plan. ---. Federal EEO Complaint Processing Procedures. Available online at http://www.eeoc.gov/eeoc/publications/fedprocess.cfm ---. Job Patterns For Minorities And Women In Private Industry (EEO-1). http://www.eeoc.gov/eeoc/statistics/employment/jobpat-eeo1/ ---. United States Equal Employment Opportunity Commission Strategic Plan for Fiscal Years 2012 – 2016. Available online at http://www.eeoc.gov/eeoc/plan/strategic_plan_12to16.cfm ---. United States Equal Employment Opportunity Commission Strategic Plan for Fiscal Years 2007 – 2012. Available online at http://www.eeoc.gov/eeoc/plan/strategic_plan_07to12_mod.cfm ---. 2011. Integrated Mission System (IMS): Private Business Review Manual. ---. 2012. Strategic Enforcement Plan. U.S. Office of Management and Budget. 2012. Circular A-11, e.g., Section 240.8, based on the 2010 GPRA Modernization Act. U.S. Office of Personnel Management. 2012. 2012 Federal Employee Viewpoint Survey Results. Governmentwide Management Report. Williams, Adley & Company-DC, LLP. 2011. Evaluation of the Management of the EEOC's State and Local Programs. Williams, Adley & Company-DC, LLP. Urban Institute Evaluation of EEOC’s Performance Measures | Page 37 APPENDIX C Performance Measures in the EEOC Strategic Plan for Fiscal Years 2012-2016 Objective I 1. By FY 2016, the EEOC develops, issues, implements, evaluates, and revises, as necessary, a Strategic Enforcement Plan. 2. By FY 2016, TBD% of investigations and conciliations meet the criteria established in the new Quality Control Plan. 3. By FY 2016, 100% of federal sector case inventory is categorized according to a new case management system and TBD% of hearings and appeals meet the criteria established in the new federal sector Quality Control Plan. 4. By FY 2016, TBD% of the cases in the agency’s litigation docket are systemic cases. 5. By FY 2016, the EEOC uses an integrated data system to identify potentially discriminatory policies or practices in federal agencies and has issued and evaluated TBD number of compliance plans to address areas of concern. 6. By FY 2016, a TBD% of the EEOC’s administrative and legal resolutions contain targeted, equitable relief. 7. By FY 2016, a TBD% of resolutions by FEPAs contain targeted, equitable relief. Objective II 8. By FY 2016, the EEOC is maintaining TBD significant partnerships with organizations that represent vulnerable workers and/or underserved communities. 9. By FY 2016, the EEOC is maintaining TBD significant partnerships with organizations that represent small or new business (or with businesses directly). 10. By FY 2016, the EEOC implements a social media plan. 11. The EEOC reviews, updates, and/or augments with plain language materials its sub-regulatory guidance, as necessary. Objective III 12. The EEOC strengthens the skills and improves the diversity of its workforce. 13. The EEOC improves the private sector charge process to streamline services and increase responsiveness to customers throughout the process. Urban Institute Evaluation of EEOC’s Performance Measures | Page 38 APPENDIX D Performance Measures in the EEOC Strategic Plan for Fiscal Years 2007-2012 1. Percent increase in the number of individuals benefiting from improvements to organizations’ policies, practices and procedures because of the EEOC’s enforcement programs. 2. Percent of the public confident in EEOC’s enforcement of Federal equal employment laws. 2.1 Percent private sector charges resolved in 180 days. 2.2 Percent federal sector hearings resolved in 180 days. 2.3 Percent federal sector appeals resolved in 180 days. 2.4 Percent investment files meeting quality criteria. 2.5 Percent parties confident in EEOC’s mediation program. 2.6 Percent lawsuits successfully resolved. Urban Institute Evaluation of EEOC’s Performance Measures | Page 39 APPENDIX E Note on Existing National Surveys as Potential Source for Key EEOC Measures17 This note looks at potential national surveys to which the EEOC could add questions about employment discrimination. Adding questions to existing surveys represents a cost-effective way to conduct a national survey. NATIONAL HOUSEHOLD SURVEYS Surveys directed at members of the public can provide insight into the knowledge and perceptions employees hold about discriminatory actions. It could also provide a measure of perceived and actual incidence of discrimination nationally. The surveys reported below are national in scope and be conducted regularly, at least once a year. Current Population Survey The Current Population Survey (CPS)18 is a joint effort conducted by the Bureau of Labor Statistics and the Census Bureau and is the primary source for labor force statistics in the U.S. Surveys from the CPS are conducted monthly using a sampling of 60,000 households. Respondents are surveyed for the first four months then they are removed from the sample for eight months and then surveyed for four more months before leaving the sample permanently. Also included in the CPS are supplemental questions on other topics related to the labor market. There is no information on the costs that are associated with supplementing questions on the CPS. However, the U.S. Census Bureau (2006) offers guidelines for the addition of supplemental questions to the CPS. The ability to supplement questions, the large sample size and the interconnection between the purpose of the CPS, to survey the labor force, and the interests of EEOC in conducting a survey, to examine the role of discrimination in the labor force, makes the CPS a strong survey option. General Social Survey Another ongoing national survey that might be a partner in conducting a national discrimination survey is the General Social Survey (GSS).19 The GSS is a national biennial survey of American opinions that has been conducted since 1972 by the National Opinion Research Center (NORC). It measures societal change within the U.S. over time and asks standard demographic, behavioral and attitudinal questions. The target sample size for the GSS is approximately 3000 using two samples of 1500 respondents each. The GSS has traditionally been open to the addition of topics of special interest which are considered in July of the year preceding the survey. Questions submitted for consideration in the GSS must go through an extensive vetting process if the researcher wants it added at no additional cost: Pending the availability of sufficient funds, the General Social Survey (GSS) project expects to include some items or topical modules designed by users in its 2014 survey, and invites users to submit proposals recommending such items or modules. Proposals submitted in response to this call need not be accompanied by funding that would support costs of data collection and data processing. 20 They will be judged with their scientific merit as a primary consideration. Additionally questions can be added as a paid supplemental module to the GSS. These also need to go through a review process but because the researcher is funding the question the criteria are less stringent. 17 This Note was prepared by Brent Howell, Urban Institute. 18 U.S. Census Bureau, Current Population Survey. See references for more information. 19 For more information about the General Social Survey, see references. 20 NORC’s call for 2014 proposals can be found at http://publicdata.norc.org:41000/gss/documents/OTHR/Module%20Competition.pdf. Urban Institute Evaluation of EEOC’s Performance Measures | Page 40 There is still a significant amount of coordination with the GSS staff to determine the structure of the questions to be added. There was not cost information readily available. The cost and consideration of adding questions is dependent on the number, structure and nature of the questions.21 Omnibus Surveys There are a number of organizations that conduct weekly or monthly omnibus surveys. Omnibus surveys provide the opportunity for multiple organizations and companies to attach questions onto one survey in the interest of reducing the expense to conducting a survey. The surveys generally consist of primary demographic questions and subsequent modules from other parties. The following is a list of organizations that have been identified22 and conduct omnibus surveys: Princeton Survey Research Associates International23 Frequency: Weekly Sample Size: 1000 adults Cost: Questions 1-2: $1,000 per closed end question Questions 3-4: $950 per question Questions 5+: $900 per question Standard Questions: gender, age, education, race/ethnicity, income, employment status, parent status, party identification and voter status Survey Technology and Research24 Frequency: Weekly or Monthly Sample Size: 1000 Cost: Unknown Standard Questions: Gender, age, household income, education, race/ethnicity, region, metro status, Custom Banner Options: employment status, marital status, party identification, Head of Household, Parental status, voter registration, own/rent, household composition, Age/gender of children, State, MSA, DMA Social Science Research Solutions (SSRS)25 Frequency: Weekly Sample Size: 1000 adults Cost: Unknown Standard Questions: gender, age, education, race/ethnicity, geographic information, Income, homeownership, household composition, party identification, political ideology and religion. NATIONAL EMPLOYER SURVEYS A search for potential resources that would be useful for a survey of employers related to employment discrimination was carried out along the following parameters: 1) it should be a national survey 2) that it should be conducted annually or at least regularly 3) Sample employers or have the ability to capture employers’ responses to employment discrimination. There were no surveys that met every parameter above. It was found that the surveys were either conducted at intermittent intervals or that their sample was not nationally representative. Unlike the 21 Guidelines for paid supplements to the GSS can be found at http://publicdata.norc.org:41000/gss/documents//OTHR/Paid%20Supplements2014.pdf. 22 Organizations were identified through their association with American Association for Public Opinion Research (AAPOR) and World Association for Public Opinion Research (WAPOR) in their 2009-2010 Blue Book of Agencies and Organizations that Support AAPOR and WAPOR. 23 Princeton Survey Research Associates International, PSRAI Omnibus Factsheet. See references. 24 Survey Technology & Research Center. See references. 25 Social Science Research Solutions. See references. Urban Institute Evaluation of EEOC’s Performance Measures | Page 41 employee-based surveys it is important that the survey capture the specific sampling of employers only. This narrows the available surveys and prevents the use of omnibus surveys that capture the general population which is representative of employees. Below is a list of employer based surveys that could be expanded on or stabilized to capture data annually related to employer perceptions of discrimination. There were three options that were turned up in the research process. The National Study of Employers (NSE) is conducted by the Families and Work Institute (FWI) looks at practices, policies, programs and benefits provided to U.S. employers. It has been conducted 3 times (2005, 2008 & 2012) and succeeds FWI’s Business Work Life Study conducted in 1998. The 2012 NSE sampled 1,126 employers with 50 or more employees and weighted samples to for-profit and non-profit industries.26 Already included in the survey are questions on the demographics of the workplace including the percentages of women, racial and ethnic minorities, and women and racial and ethnic minorities in top positions or reporting to top positions. With increased regularity and stability to this survey it is feasible that the EEOC could use this survey to establish a national annual employer survey. Another survey conducted regularly and targeted to employers is the Economic Census.27 The Economic Census is conducted every 5 years and would represent a more formal forum to gauge employer discrimination. It is required by law and 4 million businesses respond to operational questions and also provide performance data for their companies to the Census Bureau. As a result of its rigid structure and the requirement that employers respond to requests for information there is a much smaller chance of adding modules related to discrimination. Additionally there is a greater chance for insincere responses to any question related to employment discrimination because there may be a perception that a truthful response would result in legal repercussions. The limitations of the Economic Census make it a sub-optimal option to survey employers. Finally the EEOC could potentially partner with other interested parties such as the Department of Labor/Bureau of Labor Statistics to attempt to generate their own national survey of employers. The EEOC is particularly well positioned to engage on this front due to their access to employment data from EEO-1.28 EEO-1 data are collected from employers with federal contracts as well as employers exceeding 100 employees. This captures a significant numbers of employers and could be utilized to create a sample of organizations and companies to draw from in a national survey related or unrelated to data collection from EEO-1. EEO-1 data suffers from its rigid structure and legal implications surrounding compliance with employment law and so concerns about sincerity persist. This is why sampling from EEO-1 rather than adding employment discrimination questions could be useful. With BLS’s capacity for data collection and DOL’s interest in civil rights compliance, a partnership could be mutually beneficial for all parties involved. SURVEY METHODOLOGY When asking questions of employers or employees it is important to consider how to structure the survey questions. Knowledge of discrimination can be just as important as perceptions of discrimination. Both interact with one another and can affect the responses to survey questions. For example, if I perceive I am the victim of discrimination, it is important to consider why I believe that. It is entirely possible that I am not familiar with the law and think I have been treated unfairly when in actuality no legal wrongdoing has occurred. For both employers and employees it is important to gauge awareness of the law as well as perceived incidence of discrimination in the workplace. Abravanel and Cunningham (2002) accomplish this in their survey on housing discrimination. Respondents were told a scenario and then asked if they thought that it was against the law for the housing provider to 26 See Matos and Galinsky 2012 (p. 2) 27 U.S. Economic Census. See references. 28 U.S. Equal Employment Opportunity Commission, Job Patterns For Minorities And Women In Private Industry (EEO- 1). Urban Institute Evaluation of EEOC’s Performance Measures | Page 42 act in that way. Ten scenarios were given some in violation of fair housing law and others not. Abravanel and Cunningham compiled the answers to gauge specifically the respondent’s, and generally the overall awareness of fair housing law. In addition to questions on awareness the survey also included questions on respondent’s perceived encounters with discrimination, including the actions that they took if they had encountered discrimination. When these answers were compared to the questions on awareness of the law a rich picture of discrimination was provided. The picture included knowledge of the law, perceived incidence of discrimination and whether the perception was actually illegal. The survey was repeated in 2006 and questions were added to include the steps taken if the respondent perceived they had been discriminated against. It is conceivable that this format could also be used by EEOC to serve the purpose of gauging public knowledge and awareness of equal employment law and also gauging perceived incidence or encounters with employment discrimination and also whether that led to filing a complaint with EEOC. Conducted regularly this could prove to be an invaluable measure of discrimination and public awareness to determine the success of EEOC’s goals. The versatile format of this survey also means it could easily be adapted to translate employer responses that could be attached to any of the employer based surveys listed above. Urban Institute Evaluation of EEOC’s Performance Measures | Page 43 Appendix F Using Paired-Comparison Testing to Measure Employment Discrimination29 BACKGROUND30 Paired-comparison testing originated as an enforcement mechanism to detect discrimination. The initial form of paired test, calls for two equally qualified individuals to pose as applicants. The testers present themselves in the same way so that their identification within a protected class is the only difference between testers. Therefore, different treatment among testers indicates that discrimination may have played a role in that preference. In this way, paired tests offer useful and easy to understand descriptive evidence of the existence of discrimination. While, paired testing has traditionally concentrated on racial discrimination, evidence suggests it is possible to measure discrimination against other classes protected under the civil rights act.31 While using paired testing to measure discrimination in the housing market has been around since the 1970’s, paired testing in hiring is relatively new. Two Urban Institute pilot studies conducted in the early 1990’s used paired testing to measure discrimination in the hiring process between white/Anglo job seekers and Hispanic job seekers (funded by the Government Accountability Office) and white job seekers and black job seekers (funded by the Rockefeller Foundation).32 The pilot studies showed there was in fact discrimination in hiring practices directed at both African American and Hispanic job seekers. The studies demonstrated that paired testing can measure discrimination in hiring practices. The results provided evidence that systemic discrimination in hiring exists. However, unlike HUD’s decennial housing market discrimination studies, the pilot studies do not measure the full extent of discrimination in hiring nationally. Much of the research on paired testing in hiring focuses on in-person paired tests. However, recent work done by Bertrand and Mullainathan (2003) suggests that discrimination in hiring practices can also be measured using resume testing during the application process of the job search. Using a similar methodology to in-person paired tests, resumes submitted to employers are assigned pre-determined “black sounding” and “white sounding” names to indicate the race of the applicant. Differential treatment of the applicants is measured by the rate at which minority applicants get called back compared to white applicants. Both methods have various advantages and disadvantages. Below we focus on the comparative aspects of: cost, strength of evidence, and level and scope of coverage for both in-person and resume testing methodologies. Some of the comparisons are difficult to engage in beyond a cursory analysis because there have not been efforts to measure hiring discrimination nationally using paired testing or resume testing. IN-PERSON PAIRED TESTING (ROLE PLAYING) Cost 29 This Note was prepared by Brent Howell, Urban Institute. 30 For a comprehensive examination of paired testing methodology across multiple areas of study including paired testing for hiring see Turner and Herbig, 2007 In “All Things Being Equal.” 31 In employment, researchers are using paired testing to determine the incidence of discrimination against women and parental status (Benard, 2008). In housing, efforts are underway to determine if it is possible to use paired testing to measure discrimination against families with children and individuals with physical and mental disabilities. The versatility of paired testing can therefore help determine discrimination against various protected classes. 32 Cross et a. 1990; Turner et al. 1991. Urban Institute Evaluation of EEOC’s Performance Measures | Page 44 In-person paired testing tends to be very labor intensive and therefore also very costly. In-person tests require numerous tests over an extensive period of time in order to get a statistically significant sample size. For example Turner, Fix and Struyk (1991) conducted over 476 tests over more than four months using approximately 20 different testers in order to measure hiring discrimination in just two metropolitan areas. To extrapolate their efforts into a national test would require significantly more testers in multiple metro areas with varying costs in each metro area. Strength of Evidence In-person paired tests offer an effective and descriptive depiction of discrimination. In employment testing, in-person paired tests provide insight into discrimination at three various stages of the hiring process: the application stage, the interview stage, and offer stage. Differential treatment between the applicants at any point in the process can be measured and documented by the testers. This depth of information captures the point at which discrimination is occurring and how it is occurring in each test. Alternative methods for measuring discrimination have difficulty capturing the same depth of in their design. That is why in many instances the increased cost is justifiable. Level and Scope of Coverage One of the significant weaknesses of the in-person methodology is the limitation of testing beyond entry level positions. Out of necessity the design of In-person tests have exclusively focused on entry level positions because it becomes increasingly difficult to train testers to apply for advanced positions and then to have testers perform identically in the hiring process. Leaving out advanced positions reduces the likelihood of getting an accurate measure of discrimination in the labor force. RESUME TESTING Cost Since resume testing focuses only on the application process and does not require in-person resources, it is significantly less costly than in-person tests. The relatively low marginal cost to produce resumes results in a significant cost savings. Resumes can also be reused during the study and randomly reassigned different racial identities. The resulting cost savings enables more tests, generating a larger sample size and greater statistical significance of the application process. Bertrand and Mullainathan (2003) exhibit the advantages of the reduced costs by conducting 1500 tests during a 3 month period in two metropolitan areas. This is more than three times the number of tests conducted by the in-person methodology. Strength of Evidence The design significantly limits the strength of the evidence of the findings compared to in-person tests. Resume testing only allows analysis at the application stage of the interview process. As discussed above in- person testing is more effective at measuring discrimination and differential treatment across the entire hiring process. The increased statistical significance offered by the increased number of tests may prove illusory if discriminating employers fail to identify racial identity during the application stage of hiring and instead discriminate primarily based on (unobserved) in-person contact with applicants. This would underrepresent the incidence of discrimination and produce a poor national estimate. Level and Scope of Coverage The removal of human testers from the experiment results in the ability to test across various levels of job experience. Resume tests are not limited exclusively to the examination of entry level positions and this provides a tremendous extended benefit to the tests. Additionally, in resume tests multiple resumes of varying levels of experience could be sent to the same employer for both minority and majority applicants. Urban Institute Evaluation of EEOC’s Performance Measures | Page 45 APPENDIX G EEOC Reviews of January 30 Draft Version Urban Institute Evaluation of EEOC’s Performance Measures | Page 46 U.S. EQUAL EMPLOYMENT OPPORTUNITY COMMISSION Washington, D.C. 20507 Office of the Chair March 1, 2013 TO: Milton A. Mayo Inspector General FROM: Jacqueline A. Berrien, Chair RE: Response to Draft Report on the Evaluation of EEOC’s Outcome Measures (OIG Report Number 2012-10-PMEV) Thank you for the opportunity to respond to the draft report on the EEOC Strategic Plan for Fiscal Years 2012-2016. Like the Office of the Inspector General (OIG) and Urban Institute, we believe that a strategic plan provides an opportunity to re-assess our work, design a path forward for our agency, and establish measures to gauge and encourage our success. It is with that spirit that the Strategic Plan Workgroups and Commissioners endeavored to develop an ambitious yet realistic plan for the Commission’s work in upcoming years. To ensure we had the best thinking throughout our development process, we welcomed insights from a wide range of perspectives, internal and external to the agency, including OIG. Moreover, post adoption, as we have moved into the implementation stage, we still welcome input into the plan’s design, in the hopes that we can learn and improve as we proceed with implementation. We appreciate the efforts of the Urban Institute in drafting this report, and respect their expertise and the role of the IG in this assessment. Many recommendations in the draft report are consistent with our observations. There are, however, some matters that we want to bring to your attention before the report is finalized to ensure that the recommendations are of greatest utility to the agency. As we move forward, the Office of the Chair (OCH), Office of Research, Information, and Planning (ORIP), and other agency leaders charged with implementation will incorporate the report’s recommendations where appropriate. We also appreciate the opportunity to comment on the draft report. If OIG and/or the Urban Institute would like to follow-up with anyone in OCH or ORIP or with a goal leader concerning these comments, we would be pleased to facilitate that follow-up. General Areas of Agreement Regular and Consistent Communication with Agency Leadership: We agree with the recommendation of regular and consistent interactions with agency leadership throughout the plan monitoring and implementation process. Currently, Claudia Withers, Chief Operating Officer and Performance Improvement Office (PIO), Deidre Flippen, Director of ORIP and Deputy PIO, and their staff meet regularly with performance measure goal leaders to ensure understanding of the plan’s requirements, monitor their progress, and learn of any challenges they are facing in implementation. We also regularly update the Commissioners on the plan’s progress, directly and through their staff. Moreover, we recently held a public Commission meeting on the Strategic Plan’s implementation. As a result of these conversations, we have made – and will continue to make – appropriate adjustments to the plan’s implementation and our agency operations as necessary. Though generally we have been successful, your report revealed areas where additional communication is needed, especially with those outside of headquarters. We can begin immediately to incorporate more frequent communication with field offices during implementation of the plan. More Outcome Measures, Fewer Process Measures: We agree with the report’s finding that a number of our performance measures are process measures and not outcome measures. To the extent this poses concerns, the concerns are not new to us. As we discussed and cleared with the Office of Management and Budget, our strategic plan represented a major culture shift for our agency and included performance measures that are a complete departure from previous measures. As a result, in many cases we had to establish new baselines in the short term so that we could measure our success long-term. Similarly, we know that there is a desire to quantify discrimination and specify a measure to assess the impact of our work. Though there is not universal agreement that this can be achieved, we agree that it is an idea worthy of additional research and discussion. As we have discussed with OMB and our sister agencies that face similar challenges, the key is in finding the correct measurement. Based on our deep knowledge of employment discrimination and the nation’s workforce, we believe that some of the gauges proposed in your report would not result in an accurate measurement of the level of employment discrimination in the nation and, therefore, could not be used to assess our success or failure in reducing that level. Nevertheless, we believe that what gets measured gets done, and are looking for ways to better measure our success in addressing the nation’s employment discrimination challenges. General Areas of Concern Timing: As stated previously, we appreciate the effort invested in drafting this report. However, the recommendations come both somewhat late, and somewhat early, for practical and useful implementation. We believe the agency would have benefited from the Urban Institute’s expertise and assessment at an earlier stage in the development of the plan. The workgroup and the Commission’s efforts could have been greatly informed by your observations. Unfortunately, at this point – a year into implementation – it would be highly disruptive to change our performance measures to incorporate all of your recommendations, even if there were no other impediments, including fiscal constraints, to modifying them as you have recommended. An evaluation would be most helpful after baseline data for a number of the measures has been collected, which will begin after FY 2013. Thoroughness: In addition to our concerns about timing, we are also concerned that the report is based on a limited number of interviews. In contrast, the Strategic Plan was developed by a work group of more than 30 EEOC staff from across the agency, and was informed by extensive input 2 from the full Commission, additional EEOC staff, and dozens of stakeholders representing employers and employees, and many other interested members of the public. In addition, we consulted with other agencies concerning development and implementation of their Strategic Plans and measurements. Assumptions. Several assumptions underlying the recommendations reflect a misunderstanding of EEOC’s enforcement. For example, the report recommends that we measure discrimination by looking at the “Percent of citizens 1 reporting discrimination.” A measure on reported discrimination fails to acknowledge the underreporting of discrimination, the lack of awareness of legal protections, and the difficulties complying with procedural requirements that reduce reporting. Without controlling for these factors, we could not draw any credible conclusions about whether discrimination has increased or decreased. Moreover, conducting a survey of our stakeholders – in our case, most of the nation’s workforce – would have significant budgetary, legal, and staff implications. The report also suggests as a performance measure the “number of charges that led to a resolution that discrimination had occurred or for which mediation led to some form of relief being provided.” (Page 19) This suggested measure fails to recognize that many charges are dismissed without any finding or resolution. 2 Many of these charges may be meritorious but are closed at the request of charging parties represented by counsel who wish to litigate their claims. It also relies on the mistaken premise that if EEOC does not find reasonable cause or settle through mediation, there was no discrimination. 3 Few, if any, believe that the Commission’s cause or no cause rate can reliably measure whether discrimination actually occurred or whether an individual may prevail in a settlement or court litigation after the EEOC process Further, we are also concerned with the report’s recommendation to alter Performance Measure 4 regarding systemic litigation.(p.5) The report states that systemic cases are “not a product of EEOC’s work,” but rather depend “primarily on the characteristics of the cases [that] come to EEOC.” This assumption is incorrect. The EEOC carefully develops systemic cases from charges filed by individuals and Commissioner Charges and Directed Investigations, where EEOC becomes aware of a particular systemic practice yet no charge has been filed. 1 We note that the laws enforced by EEOC protect individuals without regard to their citizenship status. 2 See Laura Beth Nielsen, Robert L. Nelson, and Ryan Lancaster, “Individual Justice or Collective Legal Mobilization? Employment Discrimination Litigation in the Post Civil Rights United States,” 7 Journal of Empirical Legal Studies 175, 191 (June 2010). “In 80 percent of the lawsuits in our sample, the EEOC made no finding and provided the plaintiff with a right-to-sue letter. In the 20 percent of cases in which there was an EEOC finding, the EEOC supported the plaintiff’s charge 21 percent of the time and did not support on the merits 79 percent of the time. These EEOC administrative decisions also have relatively little effect on litigation outcomes.” 3 Id. 3 Additionally concerning is that the report states, “what seems to be most important is to track is the number and percent of systemic cases that were resolved satisfactorily” and recommends a measure of the “percent of litigated cases that ended favorably to the EEOC position.” (pp. 5, 25) The focus on “satisfactory resolution” would inadvertently encourage the selection of the easiest cases, i.e. the ones mostly likely to result in a win or quick resolution, which are more likely to be individual and not systemic cases or cases that have broad impact. We would posit that civil rights enforcement should focus on efforts to have the greatest impact. Indeed, it is that belief that drives our Strategic Plan and Strategic Enforcement Plan. Next Steps We recognize the value of an independent examination of processes that could benefit the agency and greatly appreciate and share OIG’s interest in ensuring the most efficient and effective performance of the agency’s work. However, for the reasons stated above, we do not believe that all of the recommendations in the draft report can be implemented by the Commission. Nevertheless, we can learn from many of its observations and will keep them in mind as we implement the strategic plan and review the success of its measures. 4 U.S. EQUAL EMPLOYMENT OPPORTUNITY COMMISSION Washington, D.C. 20507 Office of Commissioner Chai R. Feldblum TO: Milton A. Mayo Jr. Inspector General FROM: Chai R. Feldblum Commissioner DATE: March 1, 2013 RE: Response to Evaluation of EEOC’s Outcome Measures Draft Report (OIG Report Number 2012-10-PMEV) CC: Jacqueline A. Berrien Chair Claudia Withers Chief Operating Officer Constance S. Barker Commissioner Victoria A. Lipnic Commissioner P. David Lopez General Counsel Deidre Flippen, Director Office of Research, Information, and Planning Carlton Hadden, Director Office of Federal Operations Nicholas Inzeo, Director Office of Field Programs Thank you for providing us with the opportunity to offer comments on the draft report, “Evaluation of EEOC’s Outcome Measures,” that has been prepared by the Urban Institute (“draft Report”). Reports of this nature are a vital component of the Commission’s efforts to operate in an effective manner and I greatly appreciate the efforts that the Urban Institute and its researchers invested in this project. 131 M Street, N. E. Washington, D.C. 20507 Phone (202) 663-4090 TTY (202) 663-7172 FAX (202) 663-7101 CHAI.FELDBLUM@EEOC.GOV As Chair of the Performance Measurements Group – the working group of EEOC staff that developed the performance measures that were recommended to the Commission for adoption –I particularly welcome the opportunity to provide my reactions to the draft Report. In the pages that follow, I offer detailed comments on the draft Report’s findings and recommendations. My three main observations are as follows: 1) The draft Report criticizes the Commission’s decision to focus on process, rather than outcome, measures in the Strategic Plan. As I noted in my statement at the Commission Meeting held to approve the Strategic Plan,1 this was a very deliberate choice by the Commission. The Strategic Plan’s process measures create the infrastructure necessary to assess whether the Commission is achieving the strategic objectives outlined in the Plan. Once that infrastructure is in place, the Commission will be able to develop quantitative outcome measures. Indeed, the Strategic Plan specifically requires the Commission to develop such outcome measures once the infrastructure is in place. 2) The draft Report proposes a number of outcome measures to replace or supplement the process measures contained in the Strategic Plan. As I explain below, I believe that some of these measures are not particularly useful, either because they do not measure what they purport to measure, do not provide meaningful information about whether the Commission is meeting its strategic objectives, and/or are not feasible in the current budgetary climate. In a number of instances, the Commission considered, but then rejected, the measures now suggested for precisely these reasons. 3) Finally, a number of the new outcome measures recommended in the draft Report are not actually new. Some are already included in the Strategic Plan, while others are contemplated by the Strategic Plan once the process activities establishing the infrastructure have been completed. Since the first observation impacts a number of the recommendations made in the Report, I explain in the first section of this memo my thoughts on this matter. In the second part, I provide my assessment of each alternative measure proposed in the draft Report. A. Process, Rather than Outcome, Measures The draft Report begins with the important point that the Commission’s Strategic Plan for 2012-2016 contains measures that “are primarily process and not outcome measures,” and that the “current measures do not cover the nation’s progress towards 1 See Opening Statement of Commissioner Chai Feldblum, February 22, 2012 and Witness Statement of Commissioner Chai Feldblum, February 22, 2012 (Attachment A). 2 achieving the overarching goal: to reduce employment discrimination in the United States.” (Draft Report, Page 1) Unlike the draft Report, I consider this aspect of the Strategic Plan to be its singular strength, rather than its weakness – in that it reflects the Commission’s commitment to returning to first principles in developing appropriate outcome measures. When we voted on the Strategic Plan, I said the following: This Strategic Plan . . . is . . . a departure from previous plans. It focuses less on measuring numbers for numbers’ sake, and more on measuring what we need to do in order to achieve our long-term goals. This plan recognizes that some of our previous numbers-based performance measures may have had unintended adverse consequences. Thus, we force ourselves in this plan to take a step back, establish real and useful baselines in various areas and then think critically about how to move up from those baselines. There are various ramifications to stepping back and figuring out existing baselines and then setting targets to enhance our outcomes. One ramification is that, in the first few years of implementing the Strategic Plan, the Commission must necessarily focus on various process activities that help us figure out what our specific outcomes should be. Such a focus on process does not obviate the need for outcome measures. To the contrary, how well we carry out these process activities will directly affect how smart and strategic our final outcome measures will be. One area where I strongly disagree with the draft Report is its suggestion that the Commission may have elected to adopt process measures because they are easier to meet.2 None of these process measures are particularly easy to achieve. Moreover, although it is true they are more within the control of the Commission to affect, that was not the reason for their adoption. Obviously, the Commission could have undertaken all of these process activities absent the strategic planning framework called for by the Government Performance and Results Modernization Act (GPRAMA). Nevertheless, I am grateful that the requirements of GPRAMA forced us to critically evaluate the agency’s activities and strategically plan what we need to do to reach our objectives. Moreover, using the framework of the Strategic Plan meant that there was greater input from EEOC staff in the development of these target activities and greater transparency with EEOC stakeholders about the purpose of these activities. I believe those are positive side effects of using this process. Drafters of the agency’s previous Strategic Plan clearly spent a lot of time thinking about and developing its eight performance measures. Unfortunately, those measures 2 See Draft Report at Page 5 (because the Commission “has considerably more ability to affect directly the values of these measures than it does more outcome-focused measures,” that makes “these process measures attractive as performance measures.”) 3 resulted in a number of unintended adverse consequences that may have undermined our mission, rather than helped us achieve it. Rather than simply continuing the same numeric goals, or developing new numeric goals, the Commission took the much harder step of establishing significant process measures that would set us on the course of crafting appropriate and effective outcome measures for years three and four of this Plan and ultimately, for future strategic plans. B. Assessment of the Proposed Alternative Outcome Measures The draft Report proposes specific outcome measures for the Commission’s consideration. I list each measure suggested in the draft Report below and offer my assessment. 1. The draft Report’s proposed measures for Strategic Objective One (Report, Page 13) Strategic Objective 1: Combat employment discrimination through strategic law enforcement 1. Percent of EEOC’s administrative and legal resolutions that contain targeted, equitable relief. I am pleased that the draft Report agrees with the Commission’s determination that it is important to measure the extent to which the Commission is obtaining targeted, equitable relief in its administrative and legal resolutions of charges of discrimination. It was also helpful to learn from the Report that some EEOC staff do not understand the meaning of “targeted, equitable relief.” To the extent there is any confusion, however, I recommend that our Office of Field Programs and Office of General Counsel conduct outreach and education on the types of relief that investigators and lawyers should be seeking rather than recommending a change in the measure itself. 2. Percent of resolutions by FEPAs that contain targeted, equitable relief. I am pleased that the draft Report agrees with the Commission’s determination that it is important to measure the extent to which the FEPAs are obtaining targeted, equitable relief in their resolutions of charges of discrimination. As I noted above, it was helpful to learn from the Report that some FEPAs do not understand the meaning of “targeted, equitable relief.” Again though, to the extent there is any confusion, I recommend that our Office of Field Programs conduct outreach and education on the types of relief that the FEPAs should be seeking. 3. Percent of resolutions of the hearings and appeals carried out by Office of Federal Operations that contains targeted, equitable relief. I agree with the draft Report that this is a helpful outcome measure that would assess the extent to which the Commission is obtaining targeted, equitable relief in the 4 resolution of federal sector complaints of discrimination. However, to more accurately reflect what the Commission’s administrative judges and Office of Federal Operations do, I recommend that the measure be reworded along the following lines: “The percent of federal sector hearings and appeals resolutions, in which there has been a finding of discrimination or a settlement, that contain targeted, equitable relief.” 4. Number and percent of charges reviewed by expert reviewers that meet EEOC quality standards and that have been properly assigned to A, B, and C levels. The draft Report explains that “[t]his measure is similar to Measure 2 in the current strategic plan.” I believe that this measure is more appropriately described as dependent on the process activities called for in Measure 2 during the first two years of the Strategic Plan. In the Strategic Plan, the Commission decided that in order to adopt an outcome measure that assesses the “number and percent of charges reviewed by expert reviewers that meet EEOC quality standards,” the following process activities must first occur: 1) The Commission must develop a list of criteria that accurately measures quality in investigations and conciliations (the Quality Control Plan (“QCP”)); 2) The Commission must determine the baseline of the number and percent of charges that currently meet such quality standards; and 3) The Commission must develop realistic targets for increasing the quality of investigations and conciliations. The QCP outcome measures, once they have been developed with appropriate data, will assess the overall quality of our investigations and conciliations, as called for in this proposed measure. (Whether a charge has been appropriately categorized as an A, B, or C charge will be just one of a myriad of factors that will be considered in determining whether our investigations and conciliations meet quality standards.) In addition, many of the “customer service” issues that the draft Report recommends that the Commission measure, such as timeliness and communication, will be incorporated in these quality standards. 5. Number and percent of charges classified as meriting relief that resulted in either: (a) a settlement (through ADR/mediation); (b) a satisfactory settlement through conciliation (after a determination by investigators that discrimination had occurred), or (c) a litigated award. I must admit that this proposed measure confuses me a bit. Under our charge structure, if a charge is sent to mediation, it has not yet been classified as “meriting relief,” because the Commission has not yet conducted an investigation. The purpose of mediation is not to determine whether discrimination has occurred, but rather, whether there is a solution to the charging party’s complaint that might be agreeable to all parties. 5 The only stage at which the Commission classifies a charge as “meriting relief” is when it has found “reasonable cause.” The proposed measure appears to reference that stage in its proposed provisions (b) and (c). As I understand it, therefore, this proposal is suggesting that – from the group of charges in which the Commission finds reasonable cause (currently approximately 5% of our charges) – we should measure the percentage of those charges in which the Commission subsequently obtains a positive resolution through conciliation or litigation. While I understand the appeal of this sort of measure, I fear it could result in unintended adverse consequences – for example, encouraging the conciliation or settlement of charges on terms that are not warranted, simply to “count” the resolution as a positive one for purposes of this measure. 6. Percent of litigated cases that ended favorably to the EEOC position. A variation of this measure was included in the previous Strategic Plan. For that reason, the Performance Measurement Group discussed a measure along these lines at some length but ultimately chose not to recommend its inclusion in the Strategic Plan. Some members felt that the previous measure might have caused some Regional Attorneys not to bring cases that did not have a very high chance of success – even if it might have been appropriate for the agency to bring such cases in order to help develop the law. Other members felt this measure did not cause harm because they did not believe it affected Regional Attorney behavior in that way. Members also expressed concern that this statistic did not accurately represent the EEOC’s litigation success rate, in that it included all cases that had settled (including those that were settled because the litigators no longer believed that they would prevail in court). I continue to agree with the assessment of the Performance Measurement Group that, on balance, this is not a worthwhile measure. 2. The draft Report’s proposed measures for Strategic Objective Two (Report, Pp. 15-16) Strategic Objective 2: Prevent employment discrimination through education and outreach Before commenting on each of the draft Report’s proposed measures for Strategic Objective Two, let me note that the Strategic Plan for FY 2007-2012 did not include any performance measures for public outreach and education. In developing the Strategic Plan for FY2012-2016, therefore, it was a priority for the Commission to highlight the importance of preventing discrimination before it occurs through the effective use of 6 public outreach and education. The Commission did so by making these activities an overall strategic objective and by developing performance measures in this area. The Commission considered performance measures that would have set a target of an increasing number of events to be held or an increasing number of individuals to attend our outreach and education events. However, the Commission elected not to include such measures as we did not feel that they would necessarily tell us how effective our outreach was, nor did we feel the need to pressure our staff to focus on numbers in that fashion. The Commission also considered the use of surveys to determine whether the individuals who attended such events had changed their practices (if they were employers) or better understood their rights (if they were applicants or employees), as well as the idea of using the number of charges it received subsequent to outreach events as a measure of whether the outreach events had been successful. The Commission rejected the survey measures as we believed that there were too many variables that might affect the results of such surveys and that they were not worth the expenditure of resources. The Commission also opted not to include a measure on charges received after outreach events as we felt that there was an inherent causality problem in using such a measure. An increase in charges might indicate that the outreach is working – because more people understand their rights and bring charges. Conversely, a decrease in charges might indicate that the outreach is working – because more employers understand their responsibilities and are preventing discrimination. Thus, the Commission ultimately settled on Performance Measures 8 and 9 that, as the Strategic Plan explains, are focused on “rewarding and encouraging interactive and sustained partnerships with community organizations and businesses that are in the communities we are trying to reach.” The idea behind Performance Measures 8 and 9 is that an ongoing, interactive, and sustained relationship with a small number of community groups and business groups might be the best way to deepen the understanding by employees, applicants, and employers of the requirements of employment antidiscrimination laws. Moreover, the Commission believed that once such relationships were established, they might also serve as the venue for determining in a more sophisticated fashion whether our outreach and education efforts were working. I have a few additional comments with regard to the specific measures the draft Report offers for measuring the efficacy of our outreach and educational efforts. 1. Percent of employable citizens reporting they are reasonably clear about people’s rights relating to employment discrimination. 2. Percent of federal employees reporting they are reasonably clear about employees’ rights relating to employment discrimination. 7 3. Percent of employers reporting they are reasonably clear about their responsibilities relating to employment discrimination. Whether employable citizens, federal employees or employers believe they are reasonably clear about their rights or responsibilities regarding employment discrimination does not necessarily mean that their beliefs are correct. As such, this measure has some limitations. Given the current fiscal climate, I question whether the development of a new survey is the best use of our resources. However, with regard to federal employees, given that this measure proposes adding a question to a survey already conducted by OPM, it may be worth considering. 4. Number of employers found to have violated employee discrimination laws that have a charge filed against them within three years of the resolution of the first charge. The premise of this measure is that it will “help EEOC assess ‘recidivism’ of employers for whom charges have been litigated successfully in the past.” According to the draft Report, “[i]n theory, a lower recidivism rate over time should indicate success in preventing new incidents of discrimination.” I question the utility of this measure. Because many charges filed against an employer do not have merit, the number of charges filed against an employer cannot be used as a proxy for whether an employer has violated the law. Moreover, the fact that charges drop against an employer, or that charges increase against an employer, will not necessarily tell us very much about whether our outreach and education efforts have been successful. Finally, our success in preventing any additional cases of discrimination by a specific employer is more likely a measure of whether the Commission has imposed an appropriately strong remedy in the first charge. The measure of targeted, equitable relief is designed to capture that result. 3. The draft Report’s proposed measures for Strategic Objective Three (Report, Pp. 17-18) Strategic Objective 3: Deliver excellent and consistent service through a skilled and diverse workforce and effective systems The draft Report’s proposed measures for Strategic Objective Three focus on “customer service” measures, including measures of service timeliness, access to information on charges/cases, and courteousness. In developing the Strategic Plan, the Commission spent considerable time grappling with how to measure these issues. In the end, the Commission determined that it did not yet have the infrastructure in place to measure these issues, and thus decided it first needed to develop Quality Control Plans for the work being done in the private and 8 federal sectors. As mentioned above, once quality standards are established in those plans and a baseline of current quality is set, the Commission will be in the position to develop accurate outcome measures regarding its level of customer service. That is why, rather than provide explicit measures of customer service under Strategic Objective Three, the Strategic Plan simply cross references Performance Measure 2 (by FY 2016, TBD% of investigations and conciliations will meet the criteria established in the new Quality Control Plan) and Performance Measure 3 (by FY2016, TBD% of hearings and appeals will meet the criteria established in the new Federal Sector Quality Control Plan). With that explanation as a backdrop, I offer my assessment of the performance measures suggested by the draft Report for Strategic Objective Three. Measures of service timeliness 1. Number of pending charges and complaints at a specified point of time broken out by priority level. I agree that we should collect this data in the disaggregated form suggested. Indeed, I believe we already collect a fair amount of this data. But setting a performance measure that assesses the number of pending charges and complaints may have the inadvertent result of valuing the number of case closures over the number of quality investigations. 2. Average length of time for resolution of those charges resolved during the reporting period (between entry of charge and its “resolution”) over: (a) all charges; and (b) all complaints from federal employees. I agree that we should collect this data in the disaggregated form suggested. Indeed, I believe we already collect a fair amount of this data. But experience has shown that performance measures that establish timelines for resolutions may inadvertently incentivize premature closure or incomplete investigation of charges in the private sector and premature completion or poor analysis in appeals and hearings in the federal sector. 3. Percent of (a) charging parties and (b) federal employees who have requested hearings or filed appeals with EEOC who perceived that there were substantial unneeded delays in resolving their case. 4. Percent of (a) respondents and (b) federal agencies whose cases were handled by the EEOC who perceived that there were substantial unneeded delays in resolving their case. I do not believe that charging parties/respondents and federal employees/federal agencies are in a position to judge whether there has been an “unneeded delay” in the 9 resolution of their charges and cases, respectively. With regard to private sector charges, there are numerous factors that influence the time in which a charge is resolved. These include, among others: the investigator’s caseload, the complexity of the investigation, the amount of resources available to conduct the investigation, how quickly the respondent responds to the Commission’s requests for information, whether the Commission has to subpoena information from the respondent, and whether the respondent appeals the subpoena in court. Charging parties do not have access to such information and thus are not in the position to determine whether any delay is “unneeded.” Moreover, charging parties/respondents understandably want swift resolution of their charges and presumably will perceive anything other than a swift resolution as an unnecessary delay. Similarly, in the federal sector, numerous factors influence the time in which a hearing or appeal on a complaint is resolved. These include, among others: the amount of time it takes an agency to produce the investigative record, the quality of that record, the parties’ responsiveness to discovery requests, the caseload of the administrative judge, the availability of the parties and witnesses in order to schedule a hearing, the number of motions filed, the time it takes for an agency to issue a final decision, the time it takes an agency to produce the record on appeal, the caseload of the appellate attorney, and the need to supplement the record through additional investigation. Federal employees do not have access to such information and thus are not in the position to determine whether any delay is “unneeded.” Moreover, federal employees/agencies understandably want swift resolution of their cases and presumably will perceive anything other than a swift resolution as an unnecessary delay. With regard to all four of the above measures, I do believe we need to have timeliness targets for moving charges and complaints along at different priority levels. Indeed, I hope such targets will be part of the Quality Control Plan for the private charge sector that is currently being developed, and ultimately part of the Quality Control Plan for the federal sector that will be developed. Measure of access to information on charges/cases 5. Percent of (a) charging parties and (b) federal employees who have requested hearings or filed appeals with EEOC who reported having had significant problems in obtaining information on their charge or complaint. 6. Percent of respondents who reported having had significant problems in obtaining information on their charge or complaint. I agree that it is important to measure the quality of communication with charging parties/respondents and federal employees/federal agencies. I do not, however, believe this should be measured through a survey. Rather, I expect the Quality Control Plan to contain standards for communication, and I believe that compliance with those standards will be appropriately assessed through the peer review system called for in the Strategic Plan. 10 I also note that Performance Measure 13 of the Strategic Plan calls for the development of new processes and technologies that will provide charging parties on-line access to check the status of their charge. Agency officials recently confirmed that similar processes and technologies are being developed to provide respondents on-line access to check the status of a charge against them. In addition, Performance Measure 13 requires the development of a secure portal for electronic transmittal and receipt of charge-related documents. These new processes and technologies should significantly improve the ability of charging parties and respondents to obtain information about their charges and complaints. In addition, while Performance Measure 13 refers solely to the private sector charge process, the text accompanying that measure explains that the agency is also farther along in the development of an electronic portal for federal sector complaints. The existence of such a portal should ultimately make it easier for the EEOC to figure out mechanisms for allowing federal employees and agencies to check on the status of a complaint. Measures of courteousness 7. Percent of charging parties who reported satisfaction with the courteousness of EEOC personnel with whom they had communicated. 8. Percent of respondents who reported satisfaction with the courteousness of EEOC personnel with whom they had communicated. I agree that courteousness is an important element that should be measured. While I question the utility of surveys to obtain this information, I believe we should explore other means by which to obtain feedback about the level of courteousness our customers have received. 4. The draft Report’s proposed measures for a new Strategic Objective Four, Reduce National Employment Discrimination Levels (Report, Pp. 19- 22) Overall Objective: Reduce National Employment Discrimination Levels The draft Report recommends a new strategic objective for the Strategic Plan – Reduce National Employment Discrimination Levels – and then proposes eight new performance measures for that goal. I would argue that this is not actually a new objective, but rather a global performance measure for Strategic Objectives One (Combat Employment Discrimination Through Strategic Law Enforcement) and Two (Prevent Employment Discrimination Through Education and Outreach). That said, I believe a reduction in national employment discrimination levels would be an excellent measure of the success of the agency’s efforts. Indeed, the Performance 11 Measurement Group spent a significant amount of time trying to devise such a measure. In the end, we were unable to do so – particularly in light of our operating assumption of no increases in funding for the agency and potentially a decrease in funding. This is an inherently difficult endeavor and while I applaud the efforts of the Urban Institute in this section, I fear that the proposed measures will not accomplish the goals the draft Report set out. 1. Number of charges that led to a resolution that discrimination had occurred or for which mediation led to some form of relief being provided. There are a myriad of reasons why victims of employment discrimination do not file charges of discrimination with the EEOC, including lack of knowledge about employment rights, lack of knowledge about how or where to file a charge, inability to file a charge (e.g., no access to transportation or time off from work), fear of retaliation by an employer, lack of desire to participate in an investigation or court proceeding, etc. As such, and as the draft Report itself acknowledges, the number of charges filed is a weak indicator of national discrimination. Since this measure is based on the number of charges that have been filed, it incorporates all of these weaknesses. 2. Percent of citizens reporting that they or someone they know had experienced workplace discrimination in the last 12 months, by category of employment discrimination. This data point has the best chance of providing a good approximation of the national level of employment discrimination. That said, it would probably significantly over- represent the amount of employment discrimination that is actually occurring, given that many individuals have a poor understanding of what employment practices are actually illegal, rather than just “unfair.” Moreover, given the current financial climate, I do not believe development of a new survey is realistic at this time. However, I do believe it is worthwhile discussing with those responsible for the American Community Survey the possibility of including a question along these lines. 3. Percent of federal employees reporting that they have been discriminated against in the last 12 months by category of discrimination. The thoughts expressed above apply equally to this measure, including concerns about the cost of creating a survey. That said, given the fact that OPM’s Annual Employee Survey already exists as a survey vehicle, it might be worth adding this question to the survey. Trends in the data might serve over time as a useful barometer of, at least, perceptions of discrimination. 4. The extent to which demographic groups of concern to EEOC, in the (a) national and (b) federal workforce, reflect the actual composition of the US working population. 12 There are many factors that contribute to whether certain demographic groups reflect the actual composition of the U.S. working population. These include economic factors, education levels, access to health care for people with disabilities, family responsibilities, etc. Nevertheless, I think this is a useful measure to explore, in that trends may be indicative of whether the level of discrimination in the U.S. is increasing or decreasing. As the draft Report notes, much of the raw data is already readily available to the EEOC. The final Report might wish to note that the multi-year research and data collection plan called for by the Strategic Enforcement Plan is the appropriate venue for discussing what is possible for the agency to do with this data within our current resources. 5. Percent of paired•comparison testing cases in which discrimination occurred, for particular types of employment discrimination for which paired•comparison testing is feasible, especially those relating to hiring practices. I think paired-comparison testing is an excellent indicator of discrimination, at least by a certain employer or in a certain region or industry. However, given our very limited resources and constraints on our funding, I do not believe it is possible to pursue such testing at this time. 6. Number of discrimination victims awarded monetary benefits. As noted above, many victims of employment discrimination never file charges with the EEOC. Of those who do, many do not receive monetary benefits – because, for example, there is only an award of non-monetary relief; or conciliation fails but the EEOC chooses not to file suit and the employee cannot afford to hire an attorney; or because the attorney that the employee retains does not litigate the case well; or because the case law in a particular circuit is not favorable to the employee’s claim; etc. As a result, this measure would likely fail to provide a useful picture of national levels of employment discrimination. 7. Amount of monetary benefits (financial relief) awarded to discrimination victims. Damages are designed to compensate the harm experienced by particular individuals; they do not necessarily reflect the amount of discrimination that has occurred. Moreover, the amount of damages awarded to victims of discrimination is affected by numerous factors, including the desire of both parties to settle the matter, the financial resources of the employer, etc. As a result, I do not think this measure would provide an accurate picture of national levels of employment discrimination. 8. Number of direct recipients of monetary and non•monetary (equitable) relief, by type of relief. The concerns I note above apply equally with regard to this measure. In addition, I am 13 not sure what the draft Report means by “direct recipients of non-monetary (equitable) relief,” or how we would measure the number of such recipients. 5. The draft Report’s Final Recommendations (Report, Pp. 28-29) The draft Report concludes with some overarching recommendations, including recommendations that the EEOC: 1. Expand the new Strategic Enforcement Plan’s (SEP) requirement for quarterly reviews to include not only SEP progress but also progress reflected in the latest EEOC performance reports. 2. Provide commissioners and managers with easy access to relevant disaggregations of the outcome measure values, … broken out by such characteristics as priority level, industry, and key characteristics of the charging parties. 3. Ask the appropriate office to provide explanations for unexpectedly poor, and very good, measurement data values shown in the latest performance report. These are all excellent ideas. Indeed, these are precisely the types of actions the Commission can now take, once changes in infrastructure that the process requirements of the Strategic Plan were designed to achieve have taken hold. Thank you for providing me with the opportunity to comment on this draft Report. I hope that some of my observations and analyses will help inform the final version of the Report. 14 U.S. EQUAL EMPLOYMENT OPPORTUNITY COMMISSION Washington, D.C. 20507 Office of Field Programs March 1, 2013 MEMORANDUM TO : Milton A. Mayo, Jr. Inspector General FROM : Nicholas M. Inzeo, Director Office of Field Programs SUBJECT : Comments on Draft Report, “Evaluation of EEOC’s Outcome Measures” (OIG Report Number 2012-10-PMEV), prepared by the Urban Institute This is in response to your request for review of the draft report, “Evaluation of EEOC’s Outcome Measures.” We have completed our review and offer the following comments. The draft report presents some different approaches for building upon the Strategic Plan for FY 2012 – 2016. The report concludes that the EEOC’s strategic planning process could do more to cover and measure progress toward the overarching purpose of the EEOC to stop and remedy employment discrimination in the United States. The report recommends transforming the performance measures in the EEOC’s current and future strategic plans from process to outcome measures. It suggests several potential new or revised outcome measures and data sources for the three Strategic Objectives in the current plan and recommends the introduction of broader measures for a new overarching Strategic Objective of “reducing national employment discrimination levels.” It also recommends quarterly data-driven reviews of performance reporting by EEOC management and leadership. We believe these suggestions are worthy of discussion and consideration. We agree that the strategic planning process should continually strive toward ways to measure progress on the EEOC’s fundamental mission of achieving justice and equality in the workplace. We note, however, that the agency’s ability to implement some of the broader proposed measures may be limited by resource constraints or other external factors. Also, many of the current Strategic Plan’s performance measures are process-oriented because they are necessary first steps to build a foundation and set baselines for measures focused on external outcomes. The report will be a helpful resource as we move forward in the implementation and updating of the Strategic Plan. We have the following comments on specific parts of the draft report: The draft report recognizes the importance of Performance Measures 6 and 7 -- the percent of EEOC and FEPA administrative and legal resolutions containing Targeted Equitable Relief (TER) -- as outcome measures that track progress toward eliminating unlawful employment discrimination. We agree that Targeted Equitable Relief represents “a key component of success” and that “a clear, specific definition” should be “available and well-known throughout the EEOC” in order to obtain reliable outcome data (draft report, page 6). Based on interviews conducted in 2012, however, the report notes that there may be differing interpretations of what Targeted Equitable Relief means. We have taken significant steps in the first half of FY 2013 to ensure that the definition of Targeted Equitable Relief is known and applied consistently. Changes in IMS to capture TER took effect on January 7, 2013. We have provided training on data collection and the IMS changes in order to ensure that staff of both the EEOC and the FEPAs have a common understanding of what is to be measured. These i-Seminar training sessions were delivered to Field Enforcement Staff, including Investigators, Mediators, ISAs, Intake Staff, Supervisors and Managers, Field Hearings Staff and FEPA Staff. Over 1,100 individuals participated in the training and FEPA agencies were provided with a link to a recorded version of the training for their staff who could not attend one of the two live sessions designated for them. We believe that this training and the implementation of the IMS changes will lead to reliable, consistent measurement of Targeted Equitable Relief in all EEOC and FEPA resolutions. We will monitor this reporting as implementation continues. The report suggests future modifications to Performance Measures 8 and 9, maintaining significant partnerships with organizations that represent vulnerable workers or underserved communities (measure 8) and with organizations that represent small or new businesses, or with businesses directly (measure 9). The draft report suggests transforming these measures to examine outcomes such as the development of meaningful products such as guidelines or training for the members of the organization (draft report, page 7). We agree that these may be worthwhile areas for future performance measurement, but believe they are already encompassed to a large extent within these measures’ requirements that the partnerships be “significant.” To be significant, a partnership should be sustained and meaningful, and often this will mean that the partnership yields specific accomplishments such as those noted in the draft report. We also believe the current measure is more useful to the EEOC at this point because it encourages the establishment of more partnerships with employee and community organizations and small or new businesses, both areas where it is important to expand awareness and knowledge of the equal employment laws. We have already made progress toward establishing baselines for these measures. We believe we have the capacity to continue to increase the numbers of significant partnerships and that as we do so, we can then build upon this work and use these relationships to foster the types of outreach activities, products, and educational programs that could be the focus of future performance measures. We also think that the establishment of the partnerships in and of themselves should be recognized as a valid outcome measure. The number of partnerships does not represent merely a process measure of only internal usefulness. The existence of a partnership gives organizations representing workers or small or new businesses a valuable channel of communication with EEOC. The establishment of productive and trusting relationships to guide future interactions with the EEOC gives the workers represented by these organizations a better opportunity to vindicate their rights and gives more small businesses the resources they need to comply with the law and keep their workplaces free of discrimination. From the perspective of the outside organization, the establishment of a significant partnership with the EEOC is an outcome that directly aids in the prevention and elimination of discrimination. The draft report describes Performance Measure 4, the percent of cases in the litigation docket that are systemic cases, as having a number of conceptual issues. The report expresses concern that the metric depends primarily on the characteristics of the cases that come to the EEOC and is not a product of the EEOC’s work. It also describes the definition of “systemic” as subject to differing interpretation and judgment (draft report, page 7). The report recommends focusing on the number of systemic cases “received” rather than litigated. This portion of the report reflects some misunderstandings of the Systemic Program. After the Systemic Initiative began in 2006, we have implemented a clear and specific definition of "systemic case" and applied it consistently in enforcement and litigation. Systemic cases are not entirely dependent on the charges that come to the EEOC from individual members of the public but are primarily the product of the EEOC’s actions. The EEOC initiates systemic investigations and pursues systemic litigation in several ways. Under authority provided by our statutes, the EEOC may begin a systemic investigation on its own initiative through the filing of a Commissioner charge under Title VII or the ADA or a directed investigation under the ADEA or EPA. The EEOC also may decide to expand charges filed by members of the public or consolidate individual charges into national or regional systemic cases against particular respondents when the agency has reason to believe the alleged discrimination is more widespread. We have recently expanded our capacity to identify potential systemic discrimination cases through technology. Because we collect employer data nationally through the EEO-1 and maintain detailed charge data in IMS, the EEOC is uniquely capable of identifying systemic discrimination and perceiving potential systemic cases meriting investigation. We continue to leverage technology to facilitate communication and sharing of information and data among our district offices throughout the country. We use the Systemic Portal -- an internal EEOC systemic website -- and have recently developed and launched the Systemic Watch List Tool which draws data from IMS to help network all EEOC offices in looking at existing and potential systemic issues and cases. The Watch List can provide EEOC investigators and attorneys with notification when a charge or inquiry implicating specified priority bases, issues or practices is received into the inventory or when existing charges meet specified systemic criteria. Thus, with these and other tools, the EEOC has significant capacity to address systemic discrimination on its own initiative. The Performance Measure for the percentage of cases that are systemic thus represents an important intermediate outcome measure for the EEOC. It reflects that the EEOC has made systemic work a high priority because these cases have the potential to benefit large numbers of individuals and bring about more widespread improvements in employer policies and practices. We agree with the draft report’s recommendations that relevant disaggregations of data be collected and provided to managers to evaluate performance measurement data. The report provides some useful suggestions for categories to track, including form of discrimination and sector (private, state and local, and federal). With the increasing diversity of the population and the dynamics of the economy, we agree that it is more and more important to have detailed breakdowns of data in areas such as geographic region, specific national origin or other charge basis, and employer characteristics such as size of business. As indicated in the introduction of these comments, you have characterized EEOC’s overall mission in your “Summary” as “reducing national employment discrimination levels.” We believe you have mischaracterized the mission in a way that could produce perverse results. The Strategic Plan states our mission, “to stop and remedy unlawful employment discrimination.” The Commission’s statement of mission suggests the dynamic activity that must occur for EEOC to perform well. Your statement would encourage many to measure success by measuring fewer charges being filed or fewer findings of reasonable cause being made. These possible measures, which could be viewed as acceptable at a future time, would reward the agency for taking fewer charges or making fewer findings, regardless of the true merits of taking a charge or making a finding. We believe your statement of mission must be rejected. We understand the focus of the report on the desire to measure quality service by tracking the timeliness of the charge/complaint process, as noted in proposed measures #1 and #2 (SO 3). However, we believe that such a measure is not a true performance measure. Instead, it measures resources. The report appropriately discusses the differences in charges and the reasons why charges may not be resolved as quickly as the parties or EEOC would desire. Having tracked processing times and workloads for a number of years, it became clear that processing time correlated to workload. As EEOC was unable to hire front-line investigators from 2002 to 2008, average time to process rose. When resources permitted the hiring of more staff, thus bringing down average workloads, the average time to process fell. We believe the report should reject a timeliness measure when that measure is correlated to workload and not to the performance of the agency. We have concerns about the proposed measures that deal with the perceptions of charging parties, complainants, respondents and federal agencies. Two measures (SO 3 - #3 and #4) suggest that the parties could perceive “that there were substantial unneeded delays in resolving their case.” However, there are so many factors outside the control of the agency that can contribute to delays, but are still possible events that can occur in a charge/complaint process, such as a respondent’s delay in producing information or by the need to initiate subpoena enforcement proceedings to obtain the information. Also, two other measures (SO 3 - #5 and #6) focus on the parties’ perceived problems in obtaining access to information. These measures seem to be focused on similar areas to #3 and #4, and would likely reflect responses from those who are also frustrated with delays beyond the agency’s control that could be linked to expressing frustration with lack of information. In all four of these measures, we would expect that the responses would be influenced by the actual outcome of the charge/case. The agency has long attempted to avoid measuring a satisfaction level with the parties to our enforcement activity because of the subjective reactions associated with the final outcomes (e.g. parties who are advised that we find no reasonable cause in a case that they believe discrimination occurred are less likely to have a favorable reaction than are those who secured monetary or other relief in their charge). We do think that the measures of courteousness (SO 3 - #7 and #8) are good measures, as they focus on something important but easy to measure through a survey similar to the customer satisfaction survey. Since we are continually focusing on ways to better serve the public and ensure they receive quality customer service, this type of measure would allow us to measure the effects of our efforts. For the measure on employer recidivism (SO 2 – #4), we have concerns about the utility of a measure based on subsequent charge filings. This is a creative approach, but to more effectively capture the intent, it might be more appropriate to scale it to those employers who have subsequent charge filings that result in cause findings or repeat litigation. We recognize that charge filing activity may vary by the size of employer (largest employers may often have charges) or that charges filed may not have merit. Additionally, we would not want to measure something that might prompt employers to discourage charge filing. The report provides significant discussion about using paired-comparison testing and proposes a measure (SO 4 -#5) that would capture the results of the agency’s use of such tests. The EEOC has not pursued its own testing program for several reasons, including resource constraints and past objections to any initiation of an agency-conducted testing program expressed by our Congressional oversight and appropriations committees. However, the EEOC has always taken the position that charges may be filed by third- party organizations that have obtained evidence of discriminatory practices through matched-pair or other testing. Testing may provide useful evidence of discrimination, particularly in hiring, and we would accept and investigate any charges that are filed based on such evidence. However, we likely do not have a sufficient number of such charges that could be included in such a measure and reflect a meaningful assessment of the agency’s performance. We have also identified some clarifications to items in the draft report. For proposed measure #1 (Strategic Objective 2) and proposed measure #2 (Strategic Objective 4), we would recommend replacing the word “citizens” with “members of the public.” EEOC does not use citizenship as a qualifying condition for charges to be filed and using the limiting term of “citizen” may inadvertently express a limitation to our scope of coverage that is not correct. The discussion on p.23 regarding efficiency measures is incomplete. While the report notes that the EEOC does not currently have an efficiency measure, we believe that it would make for a more complete discussion for the narrative to reflect that EEOC did have an efficiency measure in its last Strategic Plan. Clearly, EEOC has considered such measures, but decided that such a measure in the current plan was not appropriate. Definitions would be useful for some terms that are used in proposed measures #1 and #2 under Strategic Objective 2. Measure #1 refers to “employable” citizens and it is unclear if this is attempting to capture individuals in the labor market, those seeking employment or some other government measure on employment in the U.S. Both measures refer to responders being “reasonably clear” about rights relating to employment discrimination, but there should be greater explanation provided about how to test a measurement of knowledge that would meet a ‘reasonably clear” standard. If you have any questions about our comments, please contact me or Sharon Shoemaker or Sue Murphy of my staff. cc: Claudia Withers, COO P. David Lopez, General Counsel U.S. EQUAL EMPLOYMENT OPPORTUNITY COMMISSION Washington, D.C. 20507 Office of General Counsel March 1, 2013 To: Milton A. Mayo Inspector General From: P. David Lopez General Counsel Re: Comments on draft report “Evaluation of EEOC’s Outcome Measure,” prepared by the Urban Institute This is in response to your request for comments on the Urban Institute’s draft report. We appreciate the time taken by the Urban Institute to learn about the work of our agency. We believe that many of the approaches suggested in the report should be seriously considered by the EEOC. However, the EEOC’s resources are extremely limited and we do not think the agency can bear the burden of additional measures that might require the creation of new tracking systems, significant time by a shrinking staff and/or contracts with outside experts. The following are specific comments focused on those recommendations in the report that address the litigation program: First, with respect to Performance Measure 4, “the percent of cases that are systemic cases,” the draft report suggests that the EEOC measure the number (not percentage) of systemic cases received. However, the EEOC’s systemic litigation docket does not come about simply from systemic charge filings. While a charge alleging systemic discrimination might be the basis for a systemic lawsuit, there are other sources of such suits. For example, systemic litigation may be the result of a Commissioner’s charge or an individual charge that did not necessarily allege systemic discrimination. Thus, counting the number of systemic charges received provides limited information about systemic litigation case filings. We do agree with the draft report’s recommendation that it is important to track the number and percent of systemic cases that are resolved satisfactorily. In our experience, this is information that our stakeholders are always interested in obtaining and OGC tracks this information for our own program assessment purposes. Of course, the systemic cases are only a part of the litigation docket and, as discussed below, we track the successful resolution of non- systemic cases, as well. Indeed, it is important to have measures that we have at least some ability to control and tracking successfully resolutions, while not perfect, is such a measure. Second, we agree that Performance Measure 6, percent of EEOC’s administrative and legal resolutions that contain targeted, equitable relief, is an important measure of success. There was some initial confusion in OGC about the definition of the term “equitable” because this term has a very specific meaning in our Title VII cases and that definition includes monetary relief, such as backpay. Therefore, we do believe it would have been more desirable – and less confusing to litigating lawyers – to have used the term “targeted non-monetary relief” for this measure. However, at this point in implementation, computer programs have already been modified and trainings have been conducted and we believe staff will input the correct information. As important as “targeted equitable relief” is in our cases, we also believe it is equally important to measure our success at obtaining monetary relief for victims of discrimination. The laws we enforce provide for such monetary relief and, it is indeed “targeted” as it is provided to specific individuals. Tracking monetary relief, along with equitable relief, completes the picture. Indeed, OGC has and will continue to track this information, and external stakeholders continue to ask for this information. Third, the draft report recommends the measure “percent of litigated cases that ended favorably to the EEOC position. “ We agree that this is an important measure and as the draft report indicates, OGC currently tracks this information and will continue to do so for both internal and external purposes. Fourth, we are concerned about the draft report’s recommendation that the EEOC expand the Strategic Enforcement Plan (SEP) quarterly meetings and use the type of quarterly data- driven reviews that OMB requires for the large federal agencies. Given our dwindling human resources, we believe that our staff needs to be able to concentrate on doing the jobs they were hired to do, and that reporting and process-oriented meetings that require substantial staff input and time, be kept to a minimum. Indeed, we have just completed the first quarterly SEP meeting and the preparation required significant staff time and it is not at all clear that the result of this meeting or the future meetings will lead to programmatic improvements to advance the mission. Further, pursuant to our statutes, the General Counsel, who is Presidentially appointed, has independent authority over the conduct of litigation and any meetings must ensure that the lines of authority and responsibilities within the agency are not blurred. This issue has been of paramount concern to this office throughout the development of the SEP. Thus, we would not endorse the recommendation for expanded meetings. Finally, as a housekeeping matter, the following are the additional names of individuals from OGC who participated in the interview with the Urban Institute: James Lee, Deputy General Counsel Gwendolyn Reams, Associate General Counsel Jennifer Goldstein, Senior Attorney Advisor Please feel free to contact me at 202-663-4702 or Leslie Annexstein at 202-663-4610 if you have any questions. 2
Report 2012-010-PMEV - Evaluation of EEOC's Performance Measures
Published by the Equal Employment Opportunity Commission, Office of Inspector General on 2013-03-19.
Below is a raw (and likely hideous) rendition of the original report. (PDF)