Remarks of Commissioner Stuart J. Ishimaru, Opposing

The U.S. Equal Employment Opportunity Commission

Meeting of November 16, 2005, Washington D.C. on Operations in Wake of Hurricane Katrina and Revisions to EEO-1 Report

The decision that faces us today is a test of the commitment we have, as an agency and as a government, to finding employment discrimination and eradicating it. Sadly, when faced with an opportunity to enhance our enforcement tools, enlarge our knowledge about the increasing diversity of the work force, and improve our and private counsel’s ability to examine employers’ practices, the agency has chosen a reporting scheme that will hide data and the diversity that makes this country unique—all while increasing the cost to employers and potentially increasing their liability for discrimination as well. The agency’s proposal to change the EEO-1 reporting form, which collects data from employers about the demographic makeup of their workforces, should not be approved.

The EEOC is the premier federal civil rights enforcement agency. What we do is looked at as the bellwether for civil rights enforcement. Other agencies will look at what we do and follow in our path.

We are being asked to vote on this proposal, which contains several of the most objectionable issues from the draft proposal, at a time when the entire country is being made aware that other federal agencies are sharply changing their views of civil rights enforcement. This past Sunday, the front page of the Washington Post reported that the Justice Department has come under heavy criticism for its lax enforcement of civil rights laws. According to a letter quoted in the article, the employment litigation section of the Justice Department has filed “only a handful of cases in recent years dealing with employment discrimination or discrimination based on the statistical impact on women or minority groups.” The weekend’s news stories also highlighted the Department of Justice’s decision to challenge a university’s fellowship program for minorities and women in Illinois and to approve a Georgia voter identification law that a federal judge found to be tantamount to a poll tax.

The EEOC has not been identified as one of the agencies on a similar path, and, indeed, our case filing numbers continue at historical levels. But I fear that the proposed changes to the EEO-1 form are disturbingly similar to the actions of other federal civil rights agencies in which they appear to be backing away from strong enforcement of our civil rights laws.

There are four major problems with the proposal we vote on today. First, it dilutes a critical enforcement and monitoring tool. Second, it creates a new category for EEO-1 reporting—essentially a mutiracial category—that will lump together all those who are more than one race. Third, it does not collect race data for Hispanics. Fourth, it ignores job segregation in lower wage employment.

I. EEO-1 Data Is A Crucial Civil Rights Tool

EEO-1 serves the work of enforcing civil rights laws in myriad ways.1

  • EEO-1 Data and Employer Records
    EEO-1 is not just a reporting form for our agency. EEO-1 dictates the baseline of data that employers will collect and retain on their employees and applicants. Thus, proposed changes to reporting mean proposed changes to the collection of this essential data.
  • EEO-1 and Litigation
    Private plaintiffs use EEO-1 reports to prove their claims of employment discrimination. Most recently, in Dukes v. Wal-Mart, the private plaintiffs used an analysis of EEO-1 data comparing the number of women in the “officials and managers” category at Wal-Mart and other similar retailers. Indeed, the Lawyers’ Committee on Civil Rights informed us that over 290 cases cite to EEO-1 data. See, e.g. Davis v. General Accident Ins. Co. of America, 2000 WL 1780235, E.D.Pa. (Dec. 4, 2000) (“EEO-1’s, which contain evidence of gross disparity at the officer/manager level positions, are highly probative and relevant to drawing an inference of racial discrimination, and whether [the employer] was aware of this disparity”). The substantive change we vote on today has ramifications for the entire system of private attorneys general upon which Title VII relies for its implementation.
  • EEO-1 as a Research Tool
    EEO-1 data is used by academics, EEOC researchers, and employers to track segregation in the workforce and to study the changes in equal employment opportunity in the aggregate, by industry, and by EEO-1 category. For example, the EEOC produces a publication entitled Indicators of Equal Employment Opportunity, Status and Trends. This publication shows the changes in employment from the 1960’s to the present. Academics use EEO-1 data to analyze employment trends. See Donald Tomaskovic-Devey, Documenting Desegregation: EEO-1 Estimates of US Establishment Sex and Ethnic Segregation 1966-2000, North Carolina University ( n. 2 (“Prior to the year 2000 there was only occasional use of these data [EEO-1 data] by academics (citations omitted)….In recent years other researchers have received access to these data and there are now a series of papers in progress (citations omitted).”)
  • EEO-1 as a Targeting Tool
    Internally at the EEOC and through associated researchers, EEO-1 is used to identify employers whose workforce either does not match competitors or does not align with Census data. This type of analysis has led to Commissioner Charges against employers. See, e.g. EEOC v. Shell Oil, 466 U.S. 54, 80 (1984) (“most of the data on which a ‘pattern or practice’ charge is based are provided by the employer itself in the form of annual reports filed with the EEOC”; EEOC v. Autozone, Inc., 258 F.Supp.2d 822, 825 (W.D. Tenn 2003) (Commissioner Charge based on interviews and EEO-1 Reports). In fact, I have used EEO-1 data to help me determine whether or not to file a Commissioner Charge. Similarly, during my tenure at the Department of Justice, the Employment Litigation Section routinely used EEO-4 data (the EEO form used by public employers) to select targets for its investigations. Unfortunately, this practice seems to have been discontinued. See Dan Eggen, “Civil Rights Focus Shift Roils Staff at Justice” Washington Post, November 13, A1 (noting the low number of employment discrimination cases brought by the Department of Justice in recent years).

II. The Proposed EEO-1 Form’s Category of “Two or More Races” Impedes Civil Rights Enforcement By Hiding Data and Does not Comport with the Federal Government’s Guidance on Race Data Collection and Reporting.

A. “Two or More Races” Category Hides Valuable Data.

The worst part of this proposal is the decision that we, the premier federal civil rights agency, have made to collect data in a “Two or More Races” category—a category that will hide data so that it cannot be analyzed.

A multiracial category is dangerous because it hides race and pretends that there is no difference between races. In this way, this proposal is hauntingly similar to Ward Connerly’s “Racial Privacy Initiative” in California. The “Racial Privacy Initiative” would have prohibited the state of California from collecting data regarding race with some exceptions. California voters rejected this proposal. I hope that the voters today, my colleagues, make a similar decision.

Lumping all the different people who identify as having more than one racial background together does not comport with the reality of employment discrimination and does not assist with civil rights enforcement. As is clear from our own agency’s series of “Realities and Opportunities” panels and the work this agency does everyday, someone who is African American and white will have different issues and face different biases than someone who is Asian American and white – yet on the EEO.1 those selecting “Two or More Races” will all be treated the same. To give some real life examples, people such as law professor Lani Guinier, newscaster Ann Curry, and actress Tia Carrere might all be in same “Two or More Races” group. Placing these disparate individuals in the same group ignores obvious distinctions. To pretend that Asians, Blacks, Native Hawaiians and Other Pacific Islanders, and American Indians face the same type of discrimination is to ignore the unique history of each group and the history of discrimination in our nation.

If this proposal is implemented, a significant amount of data regarding minority populations will be hidden. According to the Census, the population of those identifying with more than one race is approximately:

54% of Native Hawaiians and Other Pacific Islanders;

40% of American Indians;

14% of Asians; and

5 % of Blacks.

(Source: Two or More Races Population: 2000, US Census Bureau, November 2001. at 7.)

And the hidden data will only increase. Four percent of those under 18 identify as “Two or More Races” which is twice the percentage in the 18 and above age group. Twenty-four percent of Asians under the age of 18 identify as two or more races. Id. at 9; We the People: Asians in the United States, US Census Bureau, December 2004 at 20 (

The amount of data hidden varies depending on location. For example, “Two or More Races” is approximately:

21% of the population of Hawaii;

5% of the population of Alaska;

5% of the population of California;

4.5% of the population of Oklahoma;

4% of the population of Nevada;

4% of the population of New Mexico;

4% of the population of Washington;

15% of Honolulu;

5% of New York City;

5% of Los Angeles; and

5% of San Diego.

(Source: Two or More Races Population: 2000, US Census Bureau, November 2001. at 4, 6, and 9.)

B. “Two or More Races” Does Not Comport with Federal Guidance Regarding Race Data Collection.

There are three documents issued by the Office of Management and Budget (OMB) that govern race data collection by the federal government for civil rights purposes: The 1997 Revisions to the Standards for the Classification of Federal Data on Race and Ethnicity; OMB Bulletin 00-02 Guidance on Aggregation and Allocation of Data on Race for Use in Civil Rights Monitoring and Enforcement (March 9, 2000); and Provisional Guidance on the Implementation of the 1997 Standards for Federal Data on Race and Ethnicity (December 15, 2000). 2 All three documents are available at . This administration could have, but has not, rescinded any of these documents. It is clear, however from the content of the proposal before us that their principles may be flaunted at will.

OMB’s policy on the issue of multiracial individuals and civil rights could not be clearer. Multiracial individuals are to be allowed to select the racial groups that represent them and are not to be forced to identify with a “multiracial” group. According to the 1997 Revised Standards,

The method for respondents to report more than one race should take the form of multiple responses to a single question and not a "multiracial" category. When a list of races is provided to respondents, the list should not contain a "multiracial" category.

1997 Revised Standards (D)(1) (emphasis in the original).

Bulletin 00-02 discusses how enforcement agencies should aggregate and allocate data:

Data collected by Federal enforcement agencies often are provided by businesses and institutions in aggregate form. To facilitate agency efforts to work with data on race, an aggregation method is presented below. This method keeps intact the five single race categories, and includes the four double race combinations most frequently reported in recent studies. The method also provides for the collection of information on any multiple race combinations that comprise more than one percent of the population of interest.

A balance category is provided to report those individual responses that are not included [in the aforementioned categories]. Bulletin 00-02 (I).

OMB’s guidance in Bulletin 00-02 specifically mentions civil rights enforcement. It also sets forth how individuals in the dual race groups should be allocated for civil rights enforcement purposes (see OMB Bulletin 00-02 (II)). This allocation procedure requires knowing to which dual race group the individual belongs. Following this allocation guidance is impossible for the EEOC if we use the proposed “Two or More Races” group. No where in this guidance does OMB contemplate or condone a “Two or More Races” group.

OMB’s 2000 Provisional Guidance contains an entire chapter on EEO data collection. It provides “Individuals will be permitted to report one or more races on applications and other forms pertaining to their employment.” Provisional Guidance at 67. It states: “(o)ne of the Federal Government’s most significant uses of data on race and ethnicity is in its efforts to ensure that every individual has an equal opportunity for employment….In implementing the 1997 Standards, the EEO agencies will consider the burden imposed on those required to collect and report racial data to the Federal Government. The 1997 Standards are not intended to diminish the availability and quantity of information collected and made available for civil rights enforcement purposes.” 2000 Provisional Guidance at 64 (emphasis supplied). The Provisional Guidance states “[i]n redesigning EEO forms to comply with the 1997 standards, the following categories will be recommended for EEO data collection and record keeping:

American Indian and Alaska Native


Black or African American

Native Hawaiian or Other Pacific Islander


American Indian or Alaska Native and White

Asian and White

Black or African American and White

American Indian or Alaska Native and Black or African American

Balance of individuals reporting more than one race

Provisional Guidance at 66.

Nothing in the proposal in front of us today explains why OMB’s policy is not being followed or how those in the “Two or More Race” group will be allocated for enforcement purposes. This glaring lack of information is even more troubling because the proposal today does not tell employers to follow OMB’s guidance either—effectively hiding the data that OMB, officially, wants preserved.

The package in front of us today goes out of its way not to tell employers what data they must collect versus the data they must report. There is no proposed collection form for employers and no instructions to employers that contain information from OMB’s policy or guidance (instructing employers not to use a “multiracial” category, or instructing employers to allow their employees to mark all the racial groups that apply to them, or instructing employers that “individuals will be permitted to report one or more races on applications and other forms pertaining to their employment”). In other words, there is nothing to stop employers from offering employees a form that asks them to pick one and lists, exactly as our reporting form does “Hispanic, White, Black, Asian, Native Hawaiian or Other Pacific Islander, American Indian, Two or More Races.”

In fact, the 2003 draft proposal contained a suggested questionnaire for employers that, if used, would require employers to at least collect the specific racial combinations to which an employee belongs, even if that information was not reported to us. When it became clear that employers and civil rights advocates were reading the form to require that employers collect the data that OMB mandated, we quickly backed away from that interpretation and have dropped the form from our proposal.

III. Similarly, the Proposal’s Failure to Collect Race Data from Hispanics Will Hamper Enforcement and Does not Comport with Federal Guidance.

A. Failure to Collect the Race of Hispanics

Our current proposal does not require employers to report the race of their Hispanic employees—leading, of course, to employers failing to collect this information. Again, OMB’s prior guidance is clear: “Under the 1997 standards, ‘Hispanic or Latino’ is clearly designated as an ethnicity and not as a race. Whether or not an individual is Hispanic or Latino, every effort should be made to ascertain the race or races with which an individual identifies.” Provisional Guidance at 11 (emphasis supplied for second sentence). While there are exceptions to this requirement, they should not apply when self-identification and the two part question are used—and our form claims to use both. See Preamble at 5, 11 (“The Commission reaffirms its position that self-identification is the preferred method for gathering ethnic and racial information for the EEO-1 report.” “The Commission retains the two-question format…”) See 1997 Revised Standards (“When data on race and ethnicity are collected separately, provision shall be made to report the number of respondents in each racial category who are Hispanic or Latino”)

This part of the proposed rule could result in an over-counting of Hispanics (because those who in the Census would mark both White and Hispanic are forced to mark only one category, Hispanic, or because employers, in their reporting, must select only one category for their employees who select, despite the instructions, both a national origin and a race). It could also result in an under counting (because those who in the Census would mark both White and Hispanic are forced to mark only one category and may choose White). In either scenario, we lose accuracy, information, and data. This format flies in the face of our nation’s improved understanding of the concept of multiracial and ethnic individuals, our emphasis on self-reporting, and OMB’s guidance on this issue.

It is important to note that there would be a cumulative affect on the racial categories in this proposal. Based on Census data, approximately 5% of African Americans will choose to be included in the “Two or More Races” group. In addition, 2% of Hispanics identified as Black. Both of these groups (multiracial Blacks and Black Hispanics) could be hidden from the count of African Americans by the way we are proposing to survey.

B. Lack of Two-Part Question Regarding Hispanics

After extensive field testing, OMB and the Census Bureau found that a two-part question, in which ethnicity was asked first and then race, led to the best responses. We give lip service to this ideal in the preamble to our new form. Preamble at 12. The EEO-1 Form itself, however, contains no requirement for a two-part question. Because we are not asking employers to report the race of their Hispanic employees, there is simply no reason for employers to ask a two-part question. There is no explanation in our proposal as to why this decision was made or what affect we expect it to have on response rates or data accuracy.

III. This Proposal Leaves Employers With Uncertainty.

What employers want is a clear system that will give them useful data and will not change. This proposal delivers neither and may lead to a potential increase in liability for discrimination.

It is in everyone’s best interest that employers accurately collect and report data. For example, in 2003 the Milwaukee district office settled a case against STI Holdings for the use of a written pre-employment test that had an adverse impact on African Americans (and women). Such a case would be simply impossible to bring without an accurate accounting of how many African Americans and Whites took and passed the test at issue. Conversely, employers may find themselves with increased liability in disparate impact cases if some of those who were hired or passed the test are in the “Two or More Races” group and, thus, are not counted as hires or test passers in the affected groups. Employers themselves will be unable to use this category when they try to set up their affirmative action plans and determine underutilization. Furthermore, the creation of the “Two or More Races” group with its undistinguished members will make it impossible to compare performance in past years with the new data because we will not know who is in the new group. The inability to compare data will impede our ability to recognize both employers who have increased diversity in the work force as well as those who have not.

In fact, employers agree that this proposal is unhelpful. According to Bank One “[t]wo or more is so ambiguous that it is not a valid tool for employers who are trying to ascertain workforce under utilization and adverse impact. Accurate race data is an essential element to the implementation of successful affirmative action plans.” Letter from Eloise Gries Cookson, Senior Counsel, Bank One, August 11, 2003. And the Society of Human Resource Management (SHRM) offered these views on our earlier proposal:

Simply stated, the proposed form will be an unreliable and inconsistent data collection tool. Disjointed and incoherent workplace statistics will arise due to the use of a Two or More Race category and a Hispanic/Latino ethnicity category that trumps racial categories. Both of these changes will not only result in masking of the specific demographic of a workplace but will provide ambiguous and vague data as opposed to the clear and crisp data found when employees simply chose race/ethnicity based upon the race/ethnicity with which they primarily identify.

Letter from Wendy Wunsh, Manager, Employment Regulation, August 11, 2003.

Regardless of the outcome of today’s decision, employers will have to spend some amount of money redesigning their Human Resource Information Systems to meet our EEO-1 reporting requirements. Our current course, however, maximizes employer cost. In this proposal, we leave open the possibility of further change depending upon an internal EEOC review of its own charge data and, presumably, the results of the 2010 Census. This suggests that at some point in the near future, employers may have to overhaul and redesign these systems again in order to give a better accounting of the multiracial individuals in their workforce. Considering that the EEO.1 form has not been changed in nearly 40 years, and the burden to employers in revamping their Human Resources Information Systems, we should make a sensible long term approach in what we are requiring from employers.

It is difficult to know based on the information in this proposal how much employers will have to pay, and how much alternatives would have cost. The Preamble estimates the cost but adds “[t]aking the proportion of on-line filers into account, it could be argued that the annual burden of the revised form is actually less than the estimated 599,000 hours.” Preamble at 23-24. The current annual cost estimate for employers for the EEO-1 form is 7.7 million dollars3 , —which works out to $45/form or $171/employer (according to the Preamble, there are 170,000 forms and 45,000 employers). Id. at 25. The proposed form has a projected annual cost of 11.4 million dollars which works out to $67/form or $253/employer. By my calculations, using the assumptions in the Preamble, collecting race data in the 5-4-1 method prescribed by OMB would cost 15.4 million dollars or $91/form or $341/employer. Twenty-four dollars (the difference between the proposed cost and the 5-4-1 cost) for full race information is a bargain.

IV. Alternatives that We Should Have Considered

This proposal was first published in 2003, and there was a public meeting on it that year just prior to my appointment to the Commission. The Chair called a meeting to vote on a final proposal practically identical to the one in front of us today in June 2004. At that time there was an outcry by civil rights stakeholders. That meeting was cancelled to give the EEOC more time to review the issue. Then, well over a year later, this proposal appeared. As should be clear by now, the problems that were identified in the 2003 proposal and the 2004 final proposal are still alive and well today. We have not modified our position on these fundamental civil rights issues in the past two years, despite the civil rights community’s passionate interest in this issue and willingness to work with us. This lack of consultation and lack of willingness to consider other opinions is similar to that used in the planning of the Chair’s repositioning proposal—it did not help the repositioning process, and it does not help the EEO-1 process either.

We have received letters from the Leadership Conference on Civil Rights, the Mexican American Legal Defense and Education Fund, and the Asian American Justice Center reminding us of their concerns with the draft proposal and with our plan to go forward in June 2004. Their concerns have not changed, and they are not addressed by the proposal we consider today.

Had we opened this process to other possibilities, we may have found one that would have satisfied civil rights stakeholders concerns and those of the business community. We could have:

  • Followed OMB’s Provisional Guidance as discussed above on page 6;
  • Created a mandatory data collection form, as the Department of Housing and Urban Development has done for its grantees so that the data regarding the race of Hispanics and the “Two or More Races” to which a person belongs is retained, even if it is collapsed for reporting purposes.; and
  • Considered setting up different forms for different types of employers. For example, the Federal Election Commission had different reporting requirements for different sized political action committees.

V. Changes in Job Categories

I support the changes to the job categories at the top end of the wage spectrum and commend the Commission for these changes. However, given the race data collection issues I have identified, I fear that it will be very difficult to gain meaningful information from this higher category, especially for American Indians, Asians, and Native Hawaiians and Other Pacific Islanders.

The same policy that compels that change should be extended to the bottom end of the wage scale as well. Currently, the Service category in EEO-1 contains a wide range of discrete professions with distinct wage patterns. Combining all of the service jobs together hides the job and wage segregation that occurs within this category. The new EEO-1 form should further break down the Service category to collect data on the race and sex composition of workers within the discrete categories of Service workers.

VI. Conclusion

With the proposed changes to the EEO-1 form, we are sending the clear message that we will follow in the footsteps of the agencies that are backing away from strong enforcement of our civil rights laws. We are watering down our own enforcement tools and signaling that measuring race and job segregation is simply not that important to us. I am surprised that my colleagues are willing to approve this form because, while we may strenuously disagree on certain policy matters, spending, and access to information, I have found that all of us frequently agree on enforcement matters. For the good of civil rights enforcement, I urge my colleagues to vote no.


1 The decision today will affect data collection beyond that in the workforces reporting on our EEO forms (private employers, state and local governments, schools, and labor unions). Instructions[0] identical to the EEO-1 proposal in front of us were drafted for federal data recording in Management Directive 715. Despite their controversial content, or maybe because of it, these instructions were never circulated to this body, nor were we given the opportunity to vote on them. EEO-1 is also the template for many other agencies’ data collection including OFCCP and the FCC.

2 From 1977 to 1997 data collection by federal agencies used the criteria in OMB’s Statistical Policy Directive No. 15 (“Race and Ethnic Standards for Federal Statistical and Administrative Reporting”). Under Directive 15, the EEOC collected data in four major race categories (White, Black, Asian/Pacific Islander, and American Indian) and one ethnicity (Hispanic). In 1993 OMB began the process of updating and revising the race and ethnicity categories in Statistical Policy Directive No. 15 in order to address criticisms that Statistical Policy Directive No. 15 did “not reflect the increasing diversity of our Nation’s population that has resulted primarily from growth in immigration and in interracial marriages.” 63 Fed. Reg. 58781, 58782 (1997). OMB used an inclusive, public process to create these changes including soliciting written comments to Statistical Policy Directive No. 15 and holding a series of town meetings. OMB created an advisory panel with representatives of over 30 government agencies, including the EEOC, to determine how Policy Directive No. 15 should be revised. After three years of study which included extensive testing of various formats and evaluation of specific issues and questions, OMB published its Revisions. The EEO-1 proposal in front of us today cites the 1997 Revisions, but it does not heed them and ignores the guidance to implement them.

3 Cost estimates for employer burden on the EEO-1 form are based on the number of hours it takes to fill out the form, which, in turn, is based on the number of cells on the form. The current form contains 90 cells (10 race and ethnicity columns and 9 job rows) and the annual employer cost is estimated at 7.7 million dollars. Annually, the EEOC receives 170,000 forms from 45,000 employers. Preamble at 25. This works out to a cost of $45/form or $171/employer for the current form. The proposed form contains 140 cells (14 race and ethnicity columns and 10 job categories) and has a projected annual cost is 11.4 million dollars. This works out to $67/form or $253/employer. Collecting race data in the 5-4-1 method advocated by OMB and nine job categories would require 180 cells (20 race categories and 9 job rows)—or twice the number in the current form. That works out to a cost of 15.4 million dollars or $91/form or $341/employer.

This page was last modified on November 16, 2005.