Breadcrumb

  1. Home
  2. Meetings of the Commission
  3. Meeting of July 1, 2015 - EEOC at 50: Progress and Continuing Challenges in Eradicating Employment Discrimination
  4. Written Testimony of Donald Tomaskovic-Devey Professor of Sociology, University of Massachusetts, Amherst

Written Testimony of Donald Tomaskovic-Devey Professor of Sociology, University of Massachusetts, Amherst

Meeting of July 1, 2015 - EEOC at 50: Progress and Continuing Challenges in Eradicating Employment Discrimination

Thank you for the opportunity to speak before the US Equal Employment Opportunity Commission and to mark its 50th anniversary. My remarks come from the perspective of an organizational sociologist who studies progress towards equal opportunities in the workplace and economic inequalities more generally. Much of my work has used EEO-1 private sector reports, which I have had access to as a result of Intergovernmental Personnel Act appointments to the EEOC. This position is unpaid, but a rich opportunity to advance both basic social science and to support the EEOC's mission.

In my remarks today, I will focus first on the role of the EEOC in private sector racial and gender employment progress since the Civil Rights Act, followed by a summary for employers and regulators of what current social science research suggests works and does not when it comes to reducing employment discrimination and promoting equal employment opportunities. I will next make some comments on our ability to identify systemic bias, in the absence of, prior to, or in conjunction with discrimination complaints. Finally, I will comment on the EEOC's Strategic Enforcement Plan, especially in regards to emerging opportunities to partner with the social science community. Indeed, these partnerships have already produced great advances in our knowledge of employer best practices as well as helped develop potential approaches to enforcement strategies targeting systemic bias.

Equal Opportunity Progress since 19651

Social distinctions around race, national origin, gender, disability, age, and sexuality are powerful sources of bias, discrimination and resulting inequalities of opportunity. They are infused with cultural meaning, associated with stereotypes and status hierarchies, often accompanied by imbalances in material and interactional resources. But these cultural distinctions are not immutable. The Civil Rights Movement of the 1950s and 1960s dismantled what were essentially apartheid racial hierarchies in US workplaces. The political successes of the Civil Rights Movement spawned other movements to challenge categorical status distinctions, most prominently the women's movement of the late twentieth century, followed by the LGBT movement more recently.

Prior to the passage of the Civil Rights Act of 1964, African American and female subordination, including segregation and exclusion in employment, were deeply institutionalized in the U.S. and elsewhere. Most everyone understood that white men would occupy the most desirable jobs and hold authority over other groups. It was also assumed that women of all races and non-whites of all genders would occupy marginally rewarded jobs with little or no authority over others. These society wide patterns of social relations were both legally codified and normatively sanctioned, ensuring the maintenance of rigid race and gender hierarchies. Age, disability, veteran status, sexual orientation and gender identity were not protected categories, and that they might one day become so was beyond the imagination of most people.

In our research on equal opportunity progress since the Civil Rights Act my co-author Kevin Stainback and I identified three distinct historical eras in EEOC enforcement efforts.

The first, between 1960 and 1972 -- during and directly after the Civil Rights Movement, was particularly powerful in producing employment advances for African Americans, especially for black males. During this period the EEOC was formed and began to define best practices, issue legal opinions, and collect data on workplace race and sex composition. The OFCCP was also formed, putting teeth into the legacy of toothless presidential non-discrimination directives to federal contractors. Both agencies in the 1960s made clear their mandate was to promote racial, but not gender, equality and there was essentially no employment progress for women during this period.

We describe this period as a period of corporate uncertainty. Employers were being pressured by the Civil Rights movement and the EEOC and OFCCP were gearing up, but there was no case law or consensus on appropriate human resource equal opportunity practices. What constituted discrimination and segregation, both prohibited by Title VII of the 1964 Civil Rights Act, were threatening concepts with no clear managerial responses or legal definitions.

Racial progress, especially for black men, nevertheless happened, and it was rapid. Employers were uncertain about what to do, but were fairly certain that they would shortly be held accountable for racial equal opportunity progress. The largest federal contractors were the leaders in this regard, other large firms followed.

The period from 1972-1980, which we call the short regulatory decade, included the rise of the women's movement and the development of forceful EEOC enforcement of the new equal opportunity laws. This short period produced the most widespread equal opportunity employment progress, extending now to white women and black women, as well as black men. During this period case law became established and tended to favor plaintiffs. The EEOC was given the power to bring discrimination cases to court, targeted prominent national firms as well as employers in the US South, and continued its active promulgation of best practices and legal interpretations of EEO law. The OFCCP began to audit federal contractors. Early in the decade legislation and court decisions clarified and expanded the scope of discrimination complaints. Regulatory uncertainty for corporations nevertheless persisted, produced by the expanding definition of discrimination, coupled by real regulatory and legal threats to business as usual.

Progress, however, was not to last. By 1980 the Civil Rights Movement was essentially dead, a national politics of racial backlash took its place, legal decisions shifted toward favoring defendants in employment discrimination complaints, and both funding and political support for the EEOC and OFCCP withered. There was still substantial political power in the women's movement and the two major parties began to compete for white women's votes. After 1980 there were very limited aggregate gains for black men and women, but continued gains for white women. As the women's movement lost steam, white women's employment gains slowed in the 1990s, stalling altogether by 2000.

Table 1 illustrates the shifts in administrative, legal, and political pressure on employers. Early pressures focused on race, but not gender. By the mid-1970s most new pressures for EEO progress were produced by the women's movement and focused on gender equality. The political environment around race became indifferent or hostile after 1980, and increasingly punitive policing practices put additional pressures on black communities. Racial progress stalled, while gender progress decelerated as the women's movement quieted.

Table 1. The Changing Movement, Legal and Political Climate for Equal Employment Opportunity Progress.

1950s and 1960s 1970s 1980s 1990s
Civil Rights Movement 1971 Griggs v Duke Power 1980 EEOC issues sexual harassment guidelines Civil Rights Act, 1991
1961 Executive Order 10925 EEOC rules pregnancy discrimination is sex discrimination Meritor v Vinson, 1986 Family Medical Leave Act, 1993
1963 Equal Pay Act 1972 Equal Employment Opportunity Act Supreme Court Shifts Burden of Proof to Plaintiff, Limits Discrimination Claims
1964 Civil Rights Act 1972 Equal Rights Amendment Reagan/Bush Weaken OFCCP, EEOC; politics of "Welfare Queens" and Willie Horton; rise of mass incarceration, racial profiling in policing Clinton on race and affirmative action: "Mend It Don't End It"; mass incarceration, racial profiling continue
1965 EEOC and OFCCP Voting Rights Act 1978 Pregnancy Discrimination Act

Table 1 does not extend into this century, where we are now experiencing a vibrant civil rights movement, with attendant legislative and legal progress, around the rights of sexual and gender minorities. The EEOC is again playing a leadership role providing legal interpretations of existing law and promoting the employment rights of the LGBT community on multiple fronts. For LGBT rights the EEOC is identifying discrimination and equal opportunity, while providing leadership for courts and legislators.

At the same time, the US has become increasingly restrictive of the rights of immigrants and has had a disastrous three decade policy of mass incarceration. Both have led to new employment inequities that require the EEOC to act. Racial and gender discrimination are no longer legally or normatively sanctioned. The same is true of age, disability and increasingly sexual orientation and gender identity. Discrimination against persons with criminal records, the unemployed, and immigrants remains largely undisturbed.

Although in some protected categories we see slow or no progress nationally, there remains progress in other areas. Rather, we have moved from the nationally institutionalized racial and gender apartheid of the 1960s and earlier, to an era of locally produced equal opportunity practices. In our research we have found that some firms, industries and localities continue to make progress, while others show patterns of increased segregation and exclusion.

To craft an effective enforcement strategy we now need to be asking where is discrimination and segregation high or rising? In order to promote normative goals for all employers we should also be asking who are the equal opportunity employers?

I can provide a quick summary of some things we have learned about these trends. Black Men, Black Women, and White Women have made continued gains into jobs with clear educational criteria, especially in the professions. Consistently, firms that screen new employees on the basis of education related job requirements tend to hire more women and minorities, and fewer white men. On the other hand, women and minorities make little progress in high wage industries and remain disproportionally represented in low wage sectors. The rural, black-belt South continues to have particularly bad outcomes for African Americans in terms of both higher segregation and lower access to good jobs, while the urban South does comparatively well. White women have lower access to good jobs and higher segregation from white men in more rural, predominantly white areas.

OFCCP oversight was associated with increased minority male and lower female employment before 1980, but has had little or no impact since 1980.2 Federal contractors today are more likely to display patterns of resegregation and declining representation of women and minorities in managerial jobs. On the other hand, industries with a high level of OFCCP oversight continue to make slow gains, suggesting that the influence of the OFCCP is now normative, rather than regulatory. Similarly, industries and localities with high densities of EEOC enforcement tend to have more positive EEO profiles, again suggesting a normative causal mechanism.3

Best Practices?

Many firms where top management has endorsed equal opportunity principles still struggle to achieve increases in workplace diversity and to control bias in employment decisions. There are no magic tools to erase implicit bias or eliminate explicit animus, but I can say a few things with confidence about managerial practices.4

Research is clear that external political and market pressure can produce equal opportunity progress in private sector firms.5 Internally, the most efficacious organizational practice seems to be simply creating organizational responsibility roles and structures to monitor human resource outcomes.6 Taken together these results suggest that EEO progress happens when it becomes part of normal business practice.

Equal-opportunity experts have long promoted managerial discretion-control reforms, including diversity training, diversity performance evaluations, and discrimination grievance mechanisms, to prevent managers from discriminating. Earlier in its history the EEOC often endorsed these practices as efficacious. Courts have grown to view these practices as evidence of nondiscrimination and they often appear in EEOC negotiated settlements.7 Unfortunately there is little evidence that any of these practices actually work, and some that they can actually produce worse outcomes for protected classes.8

The general lesson that social scientists have learned is that attempts to control managers tend to provoke resistance, while human resource practices that involve managers, increase market transparency, and are supported by external regulatory or market uncertainty tend to be more efficacious.9 Here are a set of general findings from studies that compare organizational outcomes:

  • There is no evidence that corporate equal opportunity statements are associated with increased employee diversity.
  • Diversity training is not usually efficacious, and often harms racial minority's employment representation. There is also evidence that when training has legal content it leads to decreased employment opportunity for women and minorities.10
  • Increased representation of protected groups in influential managerial roles leads to stronger implementation of equal opportunity policies and higher employment representation of protected groups.
  • Accountability and transparency in manager's employment decisions leads to increased representation and more equal pay for protected groups.
  • Identity based networking groups are not efficacious, but mentoring programs with senior managers are often effective in promoting increased representation of protected groups.
  • Formalization in hiring procedures, job posting/advertising and job descriptions tend to increase protected group representation.
  • Formal evaluation systems are not associated with improved outcomes for protected groups.
  • Human resource practices that exaggerate inequalities, such as pay for performance systems, performance based merit pay or tall job ladders, tend to increase the disparities between majority and minority employees.
  • Downsizing based on workers' tenure or position as formal layoff criteria disproportionally harms black and female managers, unless there is internal monitoring of EEO consequences before final decisions.

In general, policies, even normally ineffective ones, are strengthened when external regulatory or market forces or internal accountability structures reinforce EEO goals.

Identifying Systemic Bias

The EEOC has recognized the need to develop techniques for identifying systemic bias in employment associated with particular industries, localities and firms.11 This is essentially a social science, rather than a legal, question. How can we identify firms, industries and localities with particularly bad (or good) employment records?

One strategy, already common in discrimination litigation, is to identify appropriate baselines for representation and segregation and then identify particularly good and bad employment profiles. This can be done prior to receiving charges and used to prioritize case selection or in order to target particular firms, industries or localities for commissioner charges.12

This poses a big data problem, but not a very hard one to resolve given modern computer power. I have prepared such reports ranking both industries13 and large firms14 on gender disparities using 2012 EEO-1 surveys and Census data. The basic strategy is to grade industries on a series of equal opportunity dimensions relative to appropriate labor market baselines and then generate report card-like grades on employment equity across multiple dimensions. In my exercise I looked at total employment representation, managerial representation, occupational segregation, and pay gaps. These were then used to create an overall summary grade.15

Here, as an example, is what we find for women's representation by industry in 2012:

  • Fully 22.8% of industries receive failing grades in terms of women's employment opportunities. If the EEOC were to contemplate targeting industries for systemic bias across all dimensions these would be reasonable candidates to investigate.
  • Industries with failing grades in terms of women's employment opportunities include:
    • Air Transportation
    • Beverage and Tobacco Product Manufacturing
    • Construction of Buildings
    • Heavy and Civil Engineering Construction
    • Mining (except Oil and Gas)
    • Motor Vehicle and Parts Dealers
    • Nonmetallic Mineral Product Manufacturing
    • Oil and Gas Extraction
    • Petroleum and Coal Products Manufacturing
    • Pipeline Transportation
    • Primary Metal Manufacturing
    • Rail Transportation
    • Repair and Maintenance
    • Specialty Trade Contractors
    • Support Activities for Mining
    • Support Activities for Transportation
    • Truck Transportation
    • Utilities
    • Waste Management and Remediation Services
    • Water Transportation
    • Wood Product Manufacturing
  • Only one industry, Nursing and Residential Care Facilities, can be singled out for approaching equal employment opportunities for women compared to men. This industry has high female overall employment and representations in managerial jobs as well as low occupational segregation and (relatively) low gender wage gaps.

Similar exercises can be carried out for racial and ethnic disparity. Firms and local labor markets with particularly bad employment records can be identified for potential commissioner charges or to aid investigators in making decisions about where to commit investigatory attention.

Another application would be to use similar reports to monitor firms after the settlement of lawsuits. Research suggests that discrimination lawsuits often generate resistance and eventual backsliding in targeted firms.16 Issuing report cards on representation and segregation to firms that have conciliated charges or been found responsible in court might be an effective addition to the EEOC enforcement toolkit.

The EEOC charge database is also available for analysis in the same way I have described for employment outcomes. Adjusted appropriately for employment size, we could ask which firms, industries and localities generate the most discrimination charges? Or have high rates of post complaint harassment? Investigators now do such exercises on a case by case basis after a charge has been filed, but these types of analyses do not permit a clear sense of relative frequency and are not necessarily standardized to take into account employment size. You will get more charges from large employers, even if they are relatively good employers, simply because they are large. It would be a shame to target the good employers as a result of naïve metrics.

Importantly, the systematic analysis of good and bad employment profiles will identify the equal employment opportunity leader industries, localities and firms. Such identification could lead the EEOC to issue public report cards praising employers or industries that are equal opportunity leaders. The identification of equal opportunity leaders could be used to promote positive normative models and goals for all firms.

Identifying systemically positive and negative employers need not be limited to the private sector. Such reports could be crafted from EEO-4 and EEO-6 data for local governments and schools. Federal personnel records could be used for the same purpose, in this case grading federal agencies. The EEOC has wage data on these sectors that could be used now to address President Obama's goal to reduce the gender wage gap.

In my analysis of federal agencies gender wage gaps between 1994 and 2008, there was a general decline in gender wage gaps. In 2008, the federal agency with the largest gender pay gap paid women 15% less than men, even after statistical controls for gender differences in experience and education. About half of that pay gap was produced by gender job segregation; the other half through paying men and women in the same job different wages. Once private sector pay data become available, such analyses for firms, industries and localities should be developed. Importantly, in the absence of such pay analyses employers may not even be aware that they have a problem.

The EEOC Strategic Enforcement Plan

The mission of the EEOC is to "stop and remedy discriminatory practices in the workplace." In the past the EEOC has been most influential during periods of social protest, when it produced legal interpretations of existing EEO law, when it concentrated its efforts in particular geographies and industries, and when it promulgated normative guidance on best practices.

When the EEOC was founded most discrimination and segregation was explicit, locally legitimate, and easy to identify through differences in treatment. Today much discrimination is subtle, increasingly produced by implicit bias, normatively illegitimate, and often not clearly apparent to either the perpetrators or targets. Consequently, efforts to stop and remedy discriminatory practices today cannot rely on identifying self-consciously biased employers or discreet acts of discrimination.

The Strategic Enforcement Plan (SEP) identifies a set of priorities for enforcement as well as a set of guiding principles. An efficient use of EEOC resources speaks to the need to target enforcement where it can be most efficacious relative to its mission. The SEP recognizes the need to move toward an integrated approach to enforcement which includes legal initiatives, education, research, and cooperation with multiple stakeholders. I will speak briefly to each.

Legal Initiatives. Much of the early success of the EEOC happened through its interpretation of law ahead of the courts. The EEOC issued opinions that pregnancy and sexual harassment were covered under Title VII long before they were endorsed by the courts. Private sector firms changed their behavior as a result, and so opposition was weaker when the key cases eventually were heard by the courts.

The EEOC is doing this again today around LGBT employment. New initiatives to raise awareness of potential discrimination associated with criminal background and credit checks that are not associated with job performance can play a similar role.

The use of data driven strategies to identify systemic bias at the firm, industry, and locality level are potentially important levers for shifting the employment practices that impact many employees. Since the research suggests that the impact of litigation is much larger on neighbor firms, than on target firms, the choice of cases should not merely be about the size of the employer or merit of the case, but also the potential influence of enforcement on neighbor firms.

Educational Initiatives. Much of the early influence of the EEOC lay in identifying best practices for creating equal opportunity workplaces. Corporate personnel offices followed this advice and later the courts endorsed them as evidence of a non-discriminatory workplace.

The EEOC should revitalize this role today of identifying good and suspect practices. Unfortunately, in the 1960s and 1970s the practices endorsed by the EEOC and courts were never vetted for their efficacy and social scientists have learned only recently that many do not work.17 Thus we have a contemporary problem in which the EEOC's educational function and case law may be promulgating useless or even harmful practices. In order to fix this problem some systematic attention to the research literature on best practices, and perhaps the encouragement of new research should become incorporated into the EEOC's core function.

A particularly promising and cost effective educational initiative would be to use the results of the data driven identification of systemic bias to highlight the best firms, industries, and regions. These could then be held up as normative leaders for other firms to emulate. Doing this will also put the defense bar on notice that you can also identify the firms with particularly bad records.

For EEOC generated data this will probably be limited to gender and race/ethnic employment and earnings, but other efficacious educational initiatives might be possible. The EEOC has already signaled its concern with criminal background and credit checks, perhaps it could raise these issues in such a way as to raise corporate concern. This may change behavior, ahead of case law, a pattern we have seen repeatedly before.

Research Initiatives. The SEP recognizes the need to develop a multi-year research plan for the Commission. Compared to litigation, research is relatively low cost and can provide generalizable guidance to inform the charge processing, litigation and education functions of the Commission. I will briefly outline some ideas that the Commission might consider.

Survey Collection System Modification. The EEOC collects survey data on employment distributions from private and public sector employers. Those survey instruments were designed when data collection and analyses were expensive. They also were designed to track only race and gender protected categories. Additional data collected in the traditional EEOC survey format or in new formats would be appropriate to support the expanded mission of the Commission.

  1. I recommend that EEO-1, EEO-4, and EEO-6 surveys begin to collect data on human resource practices. For example, asking if the workplace non-discrimination policy covers age, disability, or sexual orientation would give you baseline data on employer awareness of these issues as well as signal EEOC seriousness. Questions on the use of criminal background and credit checks could serve the same purpose. Such questions, if repeated over a number of years, could be invaluable in identifying best practices for both the EEOC and employers.
  2. At present EEOC generated data are only available for gender and race/ethnic comparisons. The identification of systemic biases around age, veteran status, disability of LGBT statuses may require new data collections or creative uses of other data sources. For example, a systematic analysis of EEOC charge data may reveal patterns of employer location (specific firms or industries) or behavior (e.g. harassment) that might help in target enforcement efforts.18
  3. Questions on which protected categories are covered by firm EEO policies could also be added to the EEO-1 and EEO-4 forms. These could identify leader firms and signal to all firms the EEOC's interest in all protected categories. This would be a normative signaling strategy of EEOC priorities.
  4. Expanding data collection to cover all protected categories should be considered. This will support the investigation of discrimination charges and will also send a normative message to employers about the active concerns of the EEOC.
  1. The EEODataNet. Over the last decade cooperation between the EEOC and academic researchers has grown tremendously. Over the last two years we have begun to formalize and strengthen this relationship, sponsoring two EEODataNet conferences at the EEOC with support from both the EEOC's Office of Research, Information and Planning and a grant from the National Science Foundation to the University of Massachusetts, Amherst.

    Out of these conferences and the increasingly synergistic ties between the EEOC and the academic community a proposal to found the EEO Research and Data Center has developed. That center would be external to the EEOC, non-partisan, and charged with coordinating relationships between the EEOC and academic partners, facilitate data access and security, help broker research requests between the EEOC and social scientists, and become a stable hub for cumulative expertise in the use of EEOC data products. I hope the commissioners will support such an undertaking.
  1. Strategic Studies. The SEP goal of developing synergies between research, education and enforcement missions suggests that the EEOC should undertake or sponsor strategic studies to support its mission. There are no doubt many such studies that could be undertaken, but some of the most pressing might include investigations of discrimination and vulnerable immigrants, analyses of sexual orientation and gender identity EEO and FEP charges origin, content and outcomes, patterns of harassment in EEOC and FEP charges, EEO scorecards for the federal and private sector, the impact of credit checks, and the development of pay disparity methods and analyses.

    While some of this research can be undertaken by EEOC staff, much will require mobilizing the academic community. The EEOC should begin an explicit program of Requests for Research. Many Requests for Research would not require the expenditures of any money. If asked and supported with appropriate data, interested academic researchers will raise their own resources or do the research as part of their research portfolio. Larger projects, those with shorter timelines, or those that might be expected to be continuing information needs at the EEOC would merit a funding stream to support external research. One of the key functions of an EEO Research and Data Center would be to support these partnerships and broker relationships.

Stakeholder Cooperation. The development of research capacity, both within and external to the EEOC, has implications for other stakeholders in the EEO community. Employers will benefit from a clearer sense of best practices and the ability to benchmark with other firms within their industries and communities. Other federal and state agencies charged with promoting equal opportunity in employment will benefit as well. The Office of Federal Contract Compliance, the Department of Justice, and the Office of Personnel Management are potential consumers of the research envisioned in the Strategic Enforcement Plan. Planning and funding of these endeavors would benefit from cross-agency cooperation as well.

Thank you for the opportunity to speak with you today.


Footnotes

1 Unless otherwise noted, statements in this section are supported by historical and statistical analyses in Kevin Stainback and Donald Tomaskovic-Devey. 2012. Documenting Desegregation: Racial and Gender Segregation in Private Sector Employment since the Civil Rights Act. NY: Russell Sage Foundation.

2 Kurtulus, Fidan Ana. 2015. "The Impact of Affirmative Action on the Employment of Minorities and Women over Three Decades: 1973-2003." Upjohn Institute Working Paper 15-221. Kalamazoo, MI: W.E. Upjohn Institute for Employment Research. http://research.upjohn.org/up_workingpapers/221.

3 C. Elizabeth Hirsh. 2009. "The strength of weak Enforcement: The impact of discrimination charges on sex and race segregation in the workplace. American Sociological Review 74(2):245-271; Blair Druhan Bullock, 2015. "Judicial and Agency Enforcement of Nondiscrimination Laws." Ph.D. Dissertation in Law and Economics, Vanderbilt University.

4 Unless otherwise noted, the research supporting these conclusions are summarized in Stainback and Tomaskovic-Devey (2012) and Dobbin and Kalev (2013).

5 Stainback, Kevin, Tomaskovic-Devey, Donald and Skaggs, Sheryl. 2010. "Organizational Approaches to Inequality: Inertia, Relative Power, and Environments." Annual Review of Sociology 36: 225-247.

6 Kalev, Alexandra, Dobbin, Frank and Kelly, Erin. 2006. "Best Practices or Best Guesses? Assessing the Efficacy of Corporate Affirmative Action and Diversity Policies." American Sociological Review 71: 589-617.

7 Edelman, Lauren B., Linda H. Krieger, Scott R. Eliason, Catherine R. Albiston, and Virginia Mellema. 2011. "When Organizations Rule: Judicial Deference to Institutionalized Employment Structures." American Journal of Sociology117: 888-954; Deitch, Cynthia, and Ariane Hegewisch. 2013. "Title VII Sex and Race Discrimination Litigation Settlements as Opportunities for Organizational Change." Journal of Business and Psychology 28.4: 425-438.

8 The most comprehensive and influential study is Alexandra Kalev, Frank Dobbin, and Erin Kelly. 2006. "Best Practices or Best Guesses? Assessing the Efficacy of Corporate Affirmative Action and Diversity Policies." American Sociological Review 71: 589-617. Meta-analyses of the ineffectiveness of diversity training and performance evaluations can be found respectively in Elizabeth Levy Paluck and Donald P. Green. "Prejudice reduction: What works? A review and assessment of research and practice." Annual review of psychology 60 (2009): 339-367 and , Aparna Joshi, Jooyeon Son and Hyuntak Roh, 2015. "When Can Women Close the Gap? A Meta-analytic Test of Sex Differences in Performance Rewards." Academy of Management Review. A careful, policy by policy, decade by decade review of the impact of human resource practices can be found in Frank Dobbin and Alexandra Kalev. 2013. "The Origins and Effects of Corporate Diversity Programs." Pp. 253-281 in Oxford Handbook of Diversity and Work. Edited by Quinetta Roberson. New York: Oxford University Press. Frank Dobbin, Daniel Schrage, Alexandra Kalev. Forthcoming. "Resisting the Iron Cage: Disparate Effects of Personnel Reforms to Promote Equity." American Sociological Review, find that in the absence of strong managerial oversight many EEO practices can lead to worse outcomes for protected groups.

9Frank Dobbin, Daniel Schrage, Alexandra Kalev. Forthcoming. "Resisting the Iron Cage: Disparate Effects of Personnel Reforms to Promote Equity." American Sociological Review.

10 Elisabeth Hirsh and Youngjoo Cha (working paper University of British Columbia. "For Law or Markets? Discrimination Lawsuits, Market Performance, and Managerial Diversity.") found that when diversity training was mandated by the courts it enhanced white women's access to managerial jobs, but produced no gains or even lower employment for black men and women. Frank Dobbin and Alexandra Kalev (2015. "Try and Make Me!: Why Corporate Diversity Training Fails" Working paper, Department of Sociology, Harvard University) look more closely at the actual content of diversity training in their panel survey of private sector workplaces. Most diversity training includes both discussions on the legal prohibitions against discrimination, a list of threats posed to the company by lawsuits, and training on cultural sensitivity and psychological self-reflection. They find that training that contains any information about the legal threat associated with discrimination is associated with increased hiring of white men into managerial jobs and decreases in the representation of minorities and women. Given that 76% of training includes at least some legal content, they conclude that the vast majority of diversity training does more harm than good. They also find that mandatory diversity training was associated with decreased equal opportunity progress into managerial jobs.

11 Silverman, Leslie, 2006. "Systemic Task Force Report to the Chair of the Equal Employment Opportunity Commission." Retrieved from the EEOC 2-9-2012 at https://www.eeoc.gov/systemic-task-force-report-chair-equal-employment-opportunity-commission.

12 This approach was first advocated by Mary Graham and Julie L. Hotchkiss in 2009. See "A more proactive approach to addressing gender-related employment disparities in the United States."  Gender in Management. 24:577-595.

13 Industry Gender Disparities: A Preliminary Report on Evidence of Systemic Bias in the Private Sector. Prepared for the U.S. Equal Employment Opportunity Commission. Donald Tomaskovic-Devey, Karen Brummond, Skylar Davidson, University of Massachusetts, Amherst.

14 Firm Grades for Gender Disparities. Prepared for the U.S. Equal Employment Opportunity Commission. Donald Tomaskovic-Devey, Karen M. Brummond, Skylar Davidson, University of Massachusetts, Amherst.

15 The industry report is available on request. The firm report is limited to people with EEOC access to confidential firm data.

16 The general pattern is that while short term gains from lawsuits are common, over the long term white women lose those gains and black men and women are actually worse off. See C. Elizabeth Hirsh. 2009. "The strength of weak Enforcement: The impact of discrimination charges on sex and race segregation in the workplace. American Sociological Review 74(2):245-271 and for evidence on EEOC charges. See Sheryl Skaggs, 2010. "Legal-Political Pressures and African American Access to Managerial Jobs." American Sociological Review. 74:225-244 and 2008a. "Producing Change or Bagging Opportunity? The Effects of Discrimination Litigation on Women in Supermarket Management." American Journal of Sociology 113:1148-82 for class action lawsuits.

17 On the history of the EEOC and the development of EEO human resource practices see Frank Dobbin. 2009. Inventing Equal Opportunity. Princeton University Press. On the failure of many common EEO practices to increase employment diversity see Kalev, Alexandra, Dobbin, Frank and Kelly, Erin. 2006. Best Practices or Best Guesses? Assessing the Efficacy of Corporate Affirmative Action and Diversity Policies. American Sociological Review 71: 589-617.

18 An analysis of this sort has been carried out for age by Vincent Roscigno and colleagues (Vincent J. Roscigno, Sherry Mong, Reginald Byron, and Griff Tester. 2007. "Age discrimination, social closure and employment." Social Forces 86: 313-334.)