DOCUMENT RESUME

04860 - rB03452Z1]

Problems with Federal E'u~al Employment Opportunity Guidelines onEmployee Selection Pqeddures Need To Be Resolved. FPCD-77-54;5-167015. February 2 , 9i78. S8 pp. + 10 appendices 128 pp.).

Report to the Congress; by Elmer B. Staats, Couptroller General.

Issue Area: Personne-" n&1agement and Compensation (300);Non-Discriniaation and Equal Opportunity Pr=-.Lams: FederalAgencies' Achievement of Equal Gppcr;nitty andNondiscrimination. bQlectives (1010).

Contact: Federal Personnel and Compensation Div.Budget Funct on: Gen r-a" Government: Central Personnel

ranagemes.t (805) Gseeral Government: Other GeneralGovernment (806) '-

Organization Concerned: Civil Service Commission; Department ofLabor; Department of Justice; Equal Employment OpportunityCommission.

Congressional Relevanhd'e House Committee on Education and Labor;Senate Committee' o¥r Human Resources; congress.

Autthority: Equal Empflyzent Opportunity Act of 1972 (P. .92-261). Civil RHigts Act of 1974. Executive order 11246.Griqgs v. Duke Pouer Co., 401 U.S. 424 (1571). AlbermarlePaper Co. v. Moody, 422 U.C. 405, 425 (1975). McDonnellDouglas Corp. v. Green, '411 U.S. 792, 802 (1973). Washingtonv. Davis, 429 U.S. 229 (1976). Douglas V. Hampton, 512 F.2d976 (D.C. Cir. 1S5).

The Equal Employment Opportu..1ty Coordinating Council,consisting of the Secretary of Labor, the Attorney General, andthe Chairpersons of the Civil Service Commissicn, Civil RightsCommission, and Egual Employment Opportunity Commission cr theirrespective delegates, was established in 1972 to cocrdinateFederal egual employment opportunity enforcement efforts. Earlyin 1973, the Council set out to develop and adopt uniformguidelines for determining the proper use of tests and otherselection procedures Gonsistant with the equal employmentopportunity requireibhts of Federal law. After 5 years, thiswork is still not completed. Findings/Conclusion:: Longstandingdisagreements on guideltine requirements have arisen among the

embPer agencies of the Council. Their vieus have differed on thelegal and technical standards for judging the proper use oftests, and they perceive their mandates differently and havepursued different op ting responsibilities. 5he Coun:il lacksauthority to compel 'eib'er agencies to change their policies andouidelines or to adojt ned ones in the interest of developinguniform positions on,. ,atters relating to equal employmentopportunity. Other i~portant unresolved issues include;eliminating misunder£taandings about what the Federal guidelinesrequire employers to oq, obthaiing financial and professionalresources to put the;qidelines into practice, gettingconsistent agency iqt9jretation and enforcement of uniform

guidelines, evaluating 'Ie costs and benefits cf guidelines andtheir effect on sele~COi decisions and minority and femaleemployment patterns, ail' reconciling the competing social valuesof individual merit. and group equality. Recommendations: ThePresident should direct the Equal Employment OpportunityCoordinatinq Council 'td establish a means by which membeLagencies can agree uplgand puFt into practice consistent equalemployment opportunity policies and procedures withoutunreasonable and leigt4ky delays and adopt and use uniforsguidelines on employee selection procedures. The Council should,in its annual report to.the =resident and the Congress, presentcurbAnt, reliable information on the actual costs and effects ofputting into practice and enforcing uniform guidelines onceidopted. Ic should also: develop and issue documents vhickclearly explain tha Guidelines and show how to follow them,develop enforcement st&thards that allow public and privateemployers to meet thd guidelines, train agency enforcementpersonnel to apply the guidelines in a consis'zent marner, reviewthe agencies' use of the guidelines and make changes a.s neededto maintain consistent enforcement, and encourage menmeragencies to fund research and provide technical assistance tsemployers for developing 'cost effe:tive methods of achievingequal opportunity in imployee selection. (Authcr/$i)

.- 4; .

c:.4~ :4

4 4

ok\

REPO.RT TO THE CONGRESS

o" BY THE COMPTROLLER GENERALOF THE UNITED STATES

Problems With Federal EqualEmployment Opportunity GuidelinesOn Employee Selection ProceduresNeed To Be Resolved

Uniform guidelines on employee s'eection areneeded to

--eliminate conflict, duplication, and in-consistency in the operations of theFederal equa. employment opportunityenforcement agencies and

-promote efficiency and credibility inthe Federal Govemrent's managementof enual employment opportunity pro-grams.

This report examines problems associatedwith developing, putting into practice, and en-forcing the guidelines.

FPCD-74 FEBRUARY 2, 1978

COMTMOLLN R QENERAL OF TH UNITED STAtSWAIa..le DA. muOS

,--167015

To the President of the Senate and theSpeaker of the House of Representatives

This report discusses the problems involved in devel-oping,. putting into practice, and complying with Federalequal employment opportunity guidelines on employee selec-tion. It also discusses other factors which influence theconcept of equal employment opportunity and how it can beachieved in employee selection.

In June 1973, the Chairman, Senate Committee on Laborand Public Welfare, requested that we review the implementa-tion of the Equal Employment Opportunity Act of 1972, as itapplied to Federal employees. This report is one of a ser-ies resulting from the Chairman's request.

We made our review pursuailt to the budget and AccountingAct, 1921 (31 U.S.C. 53), and the Accounting and Auditing Actof 1950 (31 U.S.C. 67).

We are sending copies of this report to the President;the Acting Director, Office of Management and Budget; theSecretary of Labor; the Chairmen, Civil Service Commissionand Civil Rights Commission; the Attorney General; and theChair, Equal Employment Opportunity Commission.

ptroller Generalof the United States

COMPTROLLER GENERAL'S PROBLEMS WITH FEDERAL EQUALREPORT TO THE CONGRESS EMPLOYMENT OPPORTUNITY GUIDELINES

ON EMPLOYEE SELECTION PROCEDURESNEED TO BE RESOLVED

DIGEST

The Equal Employment Opportunity CoordinatingCouncil, consisting of the Secretary of Labor,the Attorney General, and the Chairpersons ofthe Civil Service Commission, Civil RightsCommission, and Equal Employment OpportunityCommission or their respective delegates, wasestablished in 1972 to coordinate Federal equalemployment opportunity enforcement efforts.Early in 1973 the Council set out to develop andadopt uniform guidelines for determining theproper use of tests and other selection pro-cedures consistent with the equal employmentopportunity requirements of r'ederal law. After5 years this work is still not completed.

Why? First, longstanding disagreements onguideline requirements have arisen amongthe member agencies of the Council. Theirviews have differed on the legal and techni-cal standards for judging the proper use oftests, and they perceive their mandatesdifferently and have pursued differentoperating responsibilities.

Second, the Council lacks authority tocompel member agencies to change theirpolicies and Guidelines or to adopt newones in the interest of developing uniformpositions on matters relating to equal em-ployment opportunity. (See pp. 17 to 21.)

Other important unresolved issues include:

-- Eliminating misunderstandit.s about whatthe Federal guidelines require employersto do. (See pp. 25 to 27.)

-- Obtaining financial and professional re-sources to put the guidelines into prac-tice. (See pp. 27 to 29.)

-- Getting consistent agency interpretationand enforcement of uniform guidelines.(See pp. 29 to 30.)

mmwhn Upmnemu. Owrpwt i FPCI-77-54

-- Evaluating the costs and benefits ofguidelines and their effect on selectiondecisions and minority and female em-ployment patterns. (See pp. 30 to 31.)

-- Reconciling the competing social valuesof individual merit and group equality.(See pp. 32 to 34.)

RECOMMENDATIONS TO THE PRESIDENT

The President should direct the Equal Em-ployment Opportunity Coordinating Councilto (1) establish a means by which memberagencies can agree upon and put intopractice consistent equal employment op-portunity policies and procedures withoutunreasonable and lengthy delays and (2)adopt and use uniform guidelines on em-ployee selection procedures.

RECOMMENDATIONS TO THECOORDINATING COUNCIL

The Coordinating Council should in its annualreport to the President and the Congresspresent current, reliable information on theactual costs and effects of putting intopractice and enforcing uniform guidelines,once adopted. It should also:

-- Develop and issue documents which clearlyexpFain the guidelines and show how tofollow them.

-- Develop enforcement standards that allowpublic and private employers to meet theguidelines within a reasonable time andyet not exreed their financial and pro-fessional resources.

--Train agency enforcement personnel to ap-ply the guidelines in a consistent manner.

--Review the agencies' use of the guidelinesand make changes as needed to maintainconsistent enforcement.

--Encourage member agencies to fund researchand provide technical assistance to em-ployers for developing and testing cost-effective methods of achieving equal op-portunity in employee selection.

ii

AGENCY COMMENTS

Tte Department of Justice, the Civil Serv-ice Commission, and the Equal EmploymentOpportunity Commission said that their dis-agreements on the uniform guidelines werenot as serious as GAO suggested. Theyalso said that the present leadership ofthe four agencies could soon reconcile thedifferences, develop a uniform position,and draft a set of guidelines. (See apps.VII, VIII, and X.)

The Department of Labor agreed, in general,with the report. (See app. IX.)

INK gh~t iii

ContentsPage

DIGEST i

CHAPTER

1 INTRODUCTIONTesting for employee selection 1

Use of employment tests 2Usefulness of written tests 2

Legislative prohibitions against dis-criminatory selection practices 4

The 1964 Civil Rights Act 4Judicial decisions in testing cases 5

2 ESTABLISHING UNIFORM GUIDELINES ONEMPLOYEE SELECTION PROCEDURES 9

Need for uniform guidelines 10EEO Coordinating Council actions 10

Public comments on proposed guide-lines 11

Continuing efforts to develop uni-form guidelines 13Areas of disagreement 13

Effects of inconsistent guidelines 15Advantages of uniform guidelines 17

Reasons for delay 17EEOC mandate 18CSC mandate 19EEO Coordinating Council lacksauthority 21

Conclusions 21Agency comments 22Recommendations 23

3 OTHER PROBLEMS IN REGULATING EMPLOYEESELECTION TO ACHIEVE EEO 25

Requirements difficult to understand 25Conclusion 27

Lack of resources hampers compliance 27Conclusion 28

Problems with multiagency standards 29Conclusion 30

Overemphasis on selection 30Conclusion 31

Professional concerns about Federalguidelines 32

Conclusion 34Agency comments 34Recommendations 37

CHAPTER Page

4 SCOPE OF REVIEW 38

APPENDIX

I Glossary 39

II Professional testing standards and terms 42

III Other EEO laws and executive orders 46

IV Our reports issued on the subject of theEEO Act otf 172 48

V Areas in which guidelines agree 50

VI Departments, agencies, organizations, andindividuals contacted 53

VII August 30, 1977, letter from Chairman,Civil Service Commission 55

VIII August 31, 1977, letter from Consultantto the Chair, Equal Employment Opportun-ity Commission 56

IX September 12, 1977, letter from AssistantSecretary for Administration and Manage-ment, Department of Labor 57

X October 26, 1977, letter from AssistantAttorney General for Administration,Department of Justice 59

ABBREVIATIONS

BNA Bureau of National Affairs

CSC Civil Service Commission

EEO equal employment opportunity

EEOC Equal Employment Opportunity Commission

FEA Federal Executive Agency

GAO General Accounting Office

CHAPTER 1

INTRODUCTION

On March 24, 1972, the Congress amended title VMI ofthe Civil Rights Act of 1964 by enacting Public Law 92-261,the Equal Employment Opportunity (EEO) Act of 1972. The1972 amendments, among other things, added section 715 totitle VII, which established a five-member Equal EmploymentOpportunity Coordinating Council consisting of the Secre-tary of Labor, the Attorney General, and the Chairpersonsof the Civil Service Commission (CSC), Civil Rights Commis-sion, and Equal Employment Opportunity Commission (EEOC) ortheir respective delegates. The Council's objective was todevelop and implement "agreements, policies, and practices"which would

"* * * maximize effort, promote efficiency, and eliminateconflict, competition, duplication and inconsistencyamong the operations, functions and jurisdictions ofthe various departments, agencies and branches of theFederal Government responsible for the implementationand enforcement of equal employment opportunity legis-lation, orders, and policies."

In its June 29, 1973, report to the President, the Councilemphasized the development of a uniform set of guidelineson employee selection procedures as its "primary projectfor the year."

Our report provides an overview of the importantissues that have developed since the project began. Thes~include legal, administrative, technical, and practicalissues associated with developing and putting into practiceuniform Federal guidelines on employee testing and selec-tion and on complying with such guidelines. Information isalso provided about factors influencing the concept of EEOin employee selection and how to achieve it.

TESTING FOR EMPLOYEE SELECTION

Debates about equal opportunity in employee selectionhave focused attention on the proper use of written tests, sothat written tests have become almost synonymous with theselection process itself. (The term "test" as used in this re-port denotes any one of a variety of employee selection proce-dures or measurement methods, such as evaluation of infor-mation on application forms, job interviews, performancetests, training programs, and so forth.)

Use of employment tests

Employers spend millions of dollars for outside pro-fessional testing services and in-house testing programs todevelop and put into practice effective selection programsto help hire the best employees and avoid the costs ofpoorly selecting and placing employees--low efficiency andproductivity, poor morale, high absenteeism, frequentturnover, and so forth.

Written tests have been a frequent and popular means ofscreening job applicants, particularly among Federal, State,and local governments having merit system coverage. CSC,one of the largest producers and users of written tests, hasnoted that in many instances such tests are the only avail-able method of meeting the Federal employment requirementsfor job-related and valid examinations; these requirementscall for rankincr applicants in order of merit so that se-lection can be made from among the best qualified. Forcertain types of positions, particularly clerical andoffice work, public and private employers have reliedheavily upon written tests to evaluate applicants.

Many employers, however, do not use written tests forselection. In 1975 the American Scciety for PersonnelAdministration and Prentice-Hall queried 2,500 companiesabout their employee testing and selection procedures.According to the published results, almost 4 out of 10 ofthe survey respondents did not use written tests. Thepoll and 4 1975 Georgetown University Public ServicesLaboratory study of surveys of selection practices in localgovernments showed that interviews (oral examinations) aremore consistently used by public and private employers thanany other selection device.

Usefulness of written tests

Discussions about the usefulness of tests in employmenthave traditionally centered on the extent to which testscores correlate with, are predictive qf, or provide infor-mation about future job performance. 1/ A comprehensive

I/The statement that a test is correlated with job perfor-mance means that ranking applicants according to theirtest scores provides $~me information about their re-lative standing in job performance. The stronger the cor-relation, the more closely the ranking in test scores re-sembles that in job performance. A perfect correlationindicates a one-to-one correspondence between rank orderon the test and rank in terms of job performance. Inother words, a perfect correlation means that test scorescan be used to exactly predict performance.

2

analysis was made of the correlation bet¥ en test scoresand job proficiency or training success._ It examined theresults of several hundred published and unpublished stud-ies conducted between 1919 and 1971. For a number ofreasons the analysis produced conservative estimates of thestrength of the correlations. Nevertheless, it showed thatfo- all jobs and all tests taken together,

-- cn the averager test scores provided about 39 per-cent of the training success information and 22percent of the job proficiency information;

--when only the strongest relationships were counted,test scores provided between 28 and 65 percent ofthe training success information and 24 to 46 per-cent of the job proficiency information; and

-- on tihe average, when only the strongest relationshipswere counted, the test provided 45 percent of thetraining success information and 35 percent of theperformance proficiency information. (Based uponthe sample sizes of these studies, these percentagesare statistically significant.)

Once a degree of correlation is established betweenscores on a particular test and measures of job performance,a number of factors determine the usefulness of that test:the strength of the correlation, the number of employees tobe selected from the pool of applicants, the percentage ofpresent employees considered satisfactory, and the costs ofthe test. Tables showing the effects of some of thesevariables on test usefulness have been available for almost40 years.

Appendix II discusses testing terms and professionallyrecognized standards for developing and validating testsand other selection procedures. These terms and standardshave been incorporated in Federal guidelines on employeeselection and referred to by the courts in determiningwhether tests are job related. Familiarity with theseterms helps in understanding the technical issues involvedin fair test development and use.

1/Edwin E. Ghiselli,"The Validity of Aptitude Tests inPersonnel Selection," Perspnnel Psychology, 26, 1973,461-477.

3

LEGISLATIVE PROHIBITIONS AGAINSTDISCRIMINATORY SELECTION PRACICES

The legal aspects of personnel testing and selectionare now prominent in personnel management in public andprivate employment. A variety of Federal, State, and localstatutes and executive orders contain EEO provisions regu-lating personnel selection. E' 9 legislation as it relatesto employee selection seeks to insure that all candidatesfor a job, training, or other employment opportunity areconsidered on the basis of factors relevart to job per-formance rather than such discriminatory 'actors as race,color, religion, sex, national origin, and age. The statusof legal issues involving EEO "nd testing is determined bythese laws, the court decisions, and the decisions andguidelines of regulatory agencies which interpret theselaws in given situations.

The 1 -4 Civil Rights Act

Civil Rights Act of 1964, as amended, is the mostimp'r-z.nt law directly addressing discrimination in employeeselection. Title VII of the act is specifically directedat eliminating discrimination in employment by, among otherthings, making it unlawful for employers

"(1) to fail or refuse to hire or to discharge anyindividual, or otherwise to discriminate againstany individual with respect to his compensation,terms, conditions, or privileges of employment,because of such individual's race, color, reli-gion, sex, or national origin; or

"(2) to limit, segregate, or classify his employeesin any way which would deprive or tend todeprive any individual of employment opportuni-ties or otherwise adversely affect his status asan employee, because of such individual's race,color, religion, sex, or national origin."

Several provisions of title VII are intended to clarifyits restrictive language. One of these, section 703(h),formulated by Senator Tower of Texas and referred to as theTower Amendment, addresses employee selection procedures andlinks employment testing to EEO:

"Notwithstanding aay other provision of this subchapterit shall not be * * * an unlawful employment practicefor an employer to give and to act upon the results ofany professionally developed ability test provided that

4

3uch test, its administration or action upon theresults is not designed, intended or used to dis-criminate because of race, color, religion, sex ornational origin * * *."

The Tower Amendment was intended to insure the con-tinued use of tests which select applicants solely on thebasis of job qualifications so that employers would not beforced to hire unqualified applicants.

The EEO provisions of several other Federal laws andexecutive orders prohibit discrimination in employee selec-tion and enable Federal agencies to establish and enforceguidelines on testing and selection practices. AppendixIII contains a brief discussion of some of these laws andexecutive orders.

JUDICIAL DECISIONS IN TESTING CASES

The courts have been actively involved in interpretingand enforcing the EEO provisions of Federal laws. In largemeasure it has been left to the Federal courts to construewhat title VII means and to stipulate what is meant bysection 703(h), which provides for the use of "profession-ally developed ability tests" in employment situations. Indetermining whether a test or other selection practiceillegally discriminates, the courts have generally dealtwith two basic questions: (1) whether the test has anadverse impact (see p. 39)--that is, disproportionatelydisqualifies a group on the basis of race, sex, or nationalorigin--and (2) if adverse impact is shown, whether theselection practice having such results is job .-elated--thatis, measures for knowledge, skills, or abilities validlyrelated to actual performance of the job. If an employercannot convince the court that the selection procedurehaving adverse impact is job related, the court can concludethe procedure illegally discriminates.

In challenges to employment testing and selectionpractices, the courts have often had to deal with highlytechnical professional questions about test construction,fair test use, and validation. Predictably, the courtshave referred to the guidelines on employee selection pro-cedures issued by the Federal EEO enforcement agencieswhen considering these technical issues.

Among the best-known legal challenges under title VIIto employee testing and selection practices is the case ofGrigqs v. Duke Power Co., 401 U.S. 424 (1971). In itsdecision, the Supreme Court said that a test or otherrequirement for employment which operates to exclude

minorities--irrespective of whether it is intended to dooo--is prohibited unless it can be shown to be related tojob performance. As section 7C3(h) was intended to inaure,employers can use tests which happen to select a dispropor-tionately higher percentage of white applicants, but theburden is on the employer to show that the selection is onthe basis of job qualifications.

The EEOC testing guidelines received Supreme Courtapproval in Griqgs v. Duke Power Co. The decision stated:

"Since the Act and its legislative history supportthe Commission's cor.structicn [of section 703(h)],this affords good reason to treat the guidelines asexpressing the will cf Congress." (401 U.S. at 434.)

The Suprere Court in Albemarle Paper Co. v. Moody, 422U.S. 405, 425 (1975), concisely reiterated the standardsfor title VII Challenges to employee selection procedures.

"In Griggs v. Duke Power Co., 401 U.S. 424 (1971),this Court unanimously held that Title VII forbidsthe use of employment tests that are discriminatoryin effect unless the employer meets 'the burden ofshowing that %ny given requirement [has] * * * amanifest rel..tLonship to the employment in question.'Id., at 432. rhis burden arises, of course, onlyafter the complaining party or class has made outa prima facie case of discrimination, i.e., hasshown that the tests in question select applicantsfor hire or promotion in a racial pattern signifi-cantly different from that of the pool of applicants.See McDonnell Douglas Corp. v. Green, 411 U.S. 792,802 (1973). If an employer does then meet the burdenof proving that its tests are 'job related,' itremains open to the complaining party to show thatother tests or selection devices, without a similarlyundesirable racial effect, would also serve theemployer's legitimate interest in 'efficient andtrustworthy workmanship.' Id., at 801. Svth ashowing would be evidence t-hat the employer wasusing its tests merely as a 'pretext' for discrimi-nation. Id., at 804-805. * * *" (Footnote omitted.)

In the Albemarle case, the Court dealt with the ques-tion of what constitutes a sufficient showing of job-relatedness with respect to an aptitude test shown to haveadverse impact. The Court, noting its statement in GrigqSthat the EEOC guidelines are "entitled to great deference,"again endorsed them by measuring the company's testvalidation study against the technical requirements of the

6

EEOC guidelines. Both Chief Justice Burger and JusticeBlackmun, in separate opinions, criticized the Court's"wooden applica ion" of tte EEOC guidelines. The ChiefJustice advocated that the guidelines be given the sameweight as well-founded expert testimony.

There have been few title VII actions against the Fed-eral Government. In one case dealing with Federal employeeselection, Douglas v. Hampton, 512 F. 2d 976 (D.C. Cir. 1975),the Court expressed the opinion that the test validationstandard to be applied to tests used for Federal employmentor advancement should be the same as that already recognizedin the private sector and expressed in the EEOC guidelines.

This reliance on private sector test validation standardsas expressed in the EEOC guidelines may have been due in partto the Court's professed uncertainty about exactly what theCSC standards meant, since they had not bee interpreted in anadjudicative proceeding. Nonetheless, the LeOC preference forcriterion-related validation studies (see app. II) was held tobe among the standards which CSC must follow.

The most recent Supreme Court case regarding testingdiscrimination is Washington v. Davis, 429 U.S. 229 (1576).Although it does not touch upon the specific problem of thedifferent standards being rpDlied by the Federal EEO agen-cies and while it is not a title VII case, a statement inthe opinion addresses the question as to what methods oftest validation are judicially acceptable.

Regarding test validation methods, Justice White notedin a footnote to the opinion:

"It appears beyond doubt by now that there is nosingle method for appropriately validating employ-ment tests for their relationship to job perform-ance. Professional standards developed by theAmericain Psychological Association in its Standardsfor Educational and Psychological Tests and Manuals(1966), accept three basic methods of validation:'empirical' or 'criterion' validity (demonstratedby identifying criteria that indicate successfuljob performance and then correlating test scores andthe criteria so identified); 'construct' validity(demonstrated by examinations structured to measurethe degree to which job applicants have identifiablecharacteristics that have been determined to beimportant in successful job performance); and'content' validity (demonstrated by tests whosecontent closely approximates tasks to be performedon the job by the applicant) * * ." 42 U.S. at 247.

7

The Supreme Court placed great reliance on the EEOCguidelines in the early landmark title VII testing cases.However, the decision in Washington shows a willingness tomake independent evaluations and judgments about the job-relatedness of employee selection devices.

8

CHAPTER 2

ESTABLISHING UNIFORM GUIDELINES ON

EMPLOYEE SELECTION PROCEDURES

A great deal of debate about regulating employee se-lection practices for EEO purposes has focused on the dif-ferent requirements contained in various Federal guidelineson employee selection procedures and the slow, time-consuming, and tedious process of getting Federal EEOenforcement agencies to develop and adopt uniform guide-lines. The EEO Coordinating Council has been working atthis process for several years. Its member agencies havehad a common objective--requiring employers to show thejob-relatedness of selection procedures which adverselyaffect the employment opportunities of women and minori-ties. They have disagreed, however, about what an employermust do to justify the continued use of these selectionprocedures. These disagreements are the result of basicdifferences in the mandates and responsibilities of theagencies.

During the late 1960s and early 197Cs EEOC, theDepartment of Labor, and CSC, under separate legal author-ities, each developed and issued guidelines on the properuse of tests and other employee selection procedures. Theimpetus for these agencies to develop and adopt uniformFederal guidelines on employee selection procedures beganwith the establishment of the five-member Equal EmploymentOpportunity Coordinating Council under the 1972 amendmentsto title VII. (See p. 1.) In its June 29, 1973, report tcthe President and the Congress, the Council emphasized the im-portance of developing uniform guidelines. The report stated:

"The Council recognized in view of the overlappingjurisdictions of various Federal agencies and thedifferences in their approach to testing that oneof the most critical needs to which it could addressitself to achieve improved coordination was thedevelopment of a uniform approach to testing foremploy-ant in State and local governments. Coor-dinathon here was essential since these jurisdic-tions by law were required to respond to thedirectives of the Federal agencies involved."

The effort to develop uniform guidelines was clearlyin accord with the mandate of the Council to eliminateconfl ict, competition, duplication, and inconsistencyamong the Federal EEO enforcement agencies. The Councilmembers agread that the principles of EEO and employment

9

on the basis of merit were consistent with one another,and on this basis a uniform set of guidelines was to bedeveloped.

NEED FOR UNIFORM GUIDELINES

Nt the time the Council began its work in 1973, itwas uncertain what direct affect inconsistent sets ofguidelines had on test users, test development, courtdecisions in testing cases, and the overall enforcementof EEO laws. This uncertainty remains, even today. How-ever, it was recognized that having more than one set ofFederal guidelines on the same subject is a poor Govern-ment policy and EEO enforcement posture. It leads toconflicting EEO enforcement strategies, inconsistent infor-mation on minimum legal requirements for employee selectionand procedures for test validation, and misunderstandingof the purpose, appropriateness, and results of the FederalGovernment's involvement in this area.

Private and governmental employers criticized theFederal EEO agencies for having inconsistent guidelineswhich imposed different requirements for using and vali-dating tests on the same employer. Many persons andorganizations pointed out the unfairness involved, sincethe requirements for some groups of employers were morestringent than for others. Establishing and enforcinginconsistent requirements on the same subject was alsoconsidered a needlessly expensive and inefficient use ofGovernment and employer resources. In addition, it led togeneral misunderstanding and doubts about the Government'smanagement of EEO enforcement efforts.

EEO COORrINATING COUNCIL ACTIONS

In February 1973, the EEO Coordinating Councildirected its staff committee responsible for drafting uni-form guidelines, as a starting point, to work with the 1970EEOC guidelines on employee selection procedures and makethe changes it considered appropriate. From August 1973to September 1975 the Council circulated drafts of proposeduniformr guidelines for comments. When revising the variousdrafts, the Council considered coranents filed by industry,State and local governments, Federal agencies, psycholo-gists, and civil rights groups.

EEOC reviewed the September 1975 draft agreed upon bythe staff committee and determined that it did not repre-sent EEOC's position. It opposed circulation of thisdraft for prepublication comment, but a majority of theCouncil disagreed.

10

The EEOC Commissioners subsequently voted not toendorse the September 1975 draft. They instead directedtheir staff to review EEOC's 1970 guidelines to see if anyrevisions were necessary in order to incorporate new devel-opments in the law and industrial psychology. A revisionof the 1970 guidelines was proposed by the EEOC GeneralCounsel on February 25, 1976; the Commissioners took noaction on it. They indicated that EEOC had no plans, atthat time, to update or revise its 1970 guidelines.

Following the prepublication comment period, theDepartment of Justice, Labor, and CSC revised the Sep-tember 1975 draft; the proposed guidelines which resultedwere published in the Federal Register on July 14, 1976,the beginning of a 45-day comment period.

One member agency, the Commission on Civil Rights, didnot participate in the decisions to circulate and publishthe proposed guidelines. In its annual report to thePresident and the Congress dated July 23, 1976, the EEO Co-nrdinating Council indicated that it did not seek or receivethe concurrence of that agency because one of its roles isto analyze critically the efforts of other Federal agenciesin the enforcement of civil rights laws.

Public comments on proposed guidelines

The EEO Coordinating Council received over 150 com-ments on the July 1977 draft of the guidelines from Federalagencies, State and local governments and private employ-ers, the American Psychological Association and otherprofessional groups, civil rights organizations, andothers. These comments indicated that:

--Federal agencies opposed the guidelines becausethey were difficult to understand, and imple-mentation would be costly and require profes-sional staff (psychologists) unavailable to theagencies.

-- State, local, and private employers generallyfound the guidelines much more workable andprofessionally sound than the 1970 EEOCguidelines.

-- The American Psychological Association found theguidelines professionally sound and flexible.

-- Civil rights organizations opposed the guidelinesas a "retreat" from the EEOC guidelines in

11

both substance and tone, which would weaken theexistina legal safeguards against discriminatoryemployee selection procedures. They cited theSupreme Court's acceptance of the EEOC guidelines.

The comments were reviewed and modifications werenade to the draft guidelines. Despite EEOC's oppositionto the draft, on November 23, 1976, Justice, Labor, andCSC adopted this common set of guidelines, referred to asthe Federal Executive Agency (FEA) Guidelines on EmployeeSelection Procedures. The three agencies cited the follow-ing reasons for adopting guidelines:

-- They better represent professionally acceptedstandards for determining validity than anyexisting set of guidelines.

--They are more consistent with the Supreme Courtand the authoritative decisions of the otherappellate courts than any set of existing guide-lines.

-- They app.y the saLme standards to the Federal Gov-ernment as well as Government contractors andothers subject to Federal law.

-- They are more practical and realistic and will domore to provide actual equality of opportunity thanany existing set of guidelines.

--Their adoption is a step toward achieving a uniformFederal position and uniform guidelines.

Currently. Labor uses the FEA guidelines in enforcingExecutive Order 11246, as amended, which prohibits jobdiscrimination by Federal contractors. CSC incorporatedthe guidelines as appendixes to existing Federal PersonnelManual supplements governing the development and use ofselection procedures. CSC adopted the guidelines for usein its examining and certification activities, in review-ing the hir:ng and placement practices of Federal agen-cies, and in carrying out its responsibilities for helpingState and local governments improve their personnelpractices. Justice follows the guidelines in enforcingFederal EEO requirements. EEOC, however, retains its 1970guidelines, which were republished in the Federal Registeron November 24, 1976.

12

Continuing efforts todevelop uniform guidelines

In June 1977, Justice, Labor, CSC, anc EEOC renewednegotiations to develop and adopt uniform guidelines. Inthis latest effort, the four agencies have been working withprovisions of the FEA and EEOC guidelines in an attempt todevelop a set of guidelines which all the agencies willadopt. A new draft of proposed guidelines agreed to by thefour agencies during October 1977 was circulated Informallyfor comment. Comments received from representatives ofState and local governments, psychologists, private employ-ers, and civil rights groups were taken into account inpreparing the draft of proposed uniform (uidelines publishedin the Federal Register on December 30, 1977, for a 60-daycomment period.

Briefly discussed below are the major differences betweenprovisions in the EEOC and FEA guidelines; recently proposeddraft guidelines dated October 1977 which attempt to resolvethese differences are also discussed. (Some of the basicsimilarities in the EEOC and FEA guideline provisions aredescribed in app. V, together with the October draft provisionon each subject. It should be noted that the October draftand the draft of proposed uniform guidelines most recentlypublished in the Federal Register have similar provisions andrequirements regarding the issues discussed below and ilnapp. V.)

Areas of disagreement

1. Evidence of adverse impact--The EEOC guidelines donot define adverse impact but suggest that its existence bedetermined by comparing the rates at which different appli-cant groups pass a particular selection -ocedure. On theother hand, the FEA guidelines require employers to collectdata showing the overall effect of their selection proce-dures on persons of different racial, sex, and ethnic groups;so that a determination of the existence of adverse impactis based on the total selection process (rather than itsindividual components). The FEA guidelines also provide aformula for measuring whether the adverse impact is signifi-cant: A selection process which selects a particular race,sex, or ethnic group at less than 80 percent of the rate atwhich the most successful group is selected generally willbe considered to have an adverse impact. The EEOC guide-lines contain no such formula or guidance.

The October 1977 draft requirements for determiningadverse impact are similar to those in the FEA guidelines.The October draft reiterates that the Federal enforcement

13

agencies sill not take enforcement action based upon theadverse impact of any component of the selection process,provided that the total selection process for a job hasno adverse impact. The draft also states that when a testuser has not maintained data on adverse impact, the Federe'enforcement agencies may draw an inference of adverse impactif the user has an underutilization of a group in the jobcategory, as compared to the group's representation in theapplicable work force.

2. Acceptance of validation strategies--The EEOCgui'elines Indicate a preference for criterion-relatedvalidation. The FEA guidelines and the October 1977draft permit any of the three validation strategies to beused, as appropriate.

3. Additional coverage and requirements--The FEAguidelines and the October 1977 draft extend the definitionof employment decisions covered by the guidelines to in-clude licensing and certification boards to the extentthese practices may be covered by Federal law. The E'EAguidelines and the October draft establish technicalstandards for validity studies, and contain detaileddocumentation requirements applicable to all coveredemployers, including the Federal civil service. CSCis responsible for validating procedures it uses orrequires; each Federal agency is responsible for validat-ing any selection procedures it establishes or requires.There are no similar provisions in the EEOC guidelines.

4. Studies of test fairness or differential vali-dity--The EEOC guidelines require validation data to begenerated and reported separately for minority and non-minority groups, wherever technically feasible. The FEAguidelines recommend collecting data separately for allgroups when criterion-related validation studies areperformed--if a selectio-n r;oedure has an adverse impacton one racial, ethnic, or sex group and there are enoughpersons in each group (not less than 30) for findings ofstatistical significance. The October 1977 draft generallycalls for studies of test fairness when criterion-relatedvalidation studies are performed. The draft states thatthe concept of fairness or unfairness of selection proce-dures is a developing concept. Noting that fairnessstudies generally require large numbers of employees inthe job or groups of jobs being studied, the October draftindicates that the obligation to perform such studiesgenerally will be upon large users, consortia of smallerusers, or test developers.

5. Search for alternative procedures--Under the EEOCguidelines, a test user is responsible for validating a

14

selection procedure in which an adverse impact has beenfound and for demonstrating that alternate selection proce-dures with less adverse impact do not exist. Under theFEA guidelines, when equally valid alternate selectionprocedures are available, use of the one demonstrated tohave less adverse impact is recommended. An employer isnot required to prove that no alternate valid procedureswith less adverse impact exist. The October 1977 draftstates that when a validity study is called for by theguidelines, the user should investigate suitable alternateselection procedures which have as little adverse impactas possible, to determine the appropriateness of using orvalidating them in accord with these guidelines. Theuser is expected to make a reasonable effort to becomeaware of such alternate procedures.

6. Definitions and explanations--The FEA guidelinesand tVe October 1977 draft contain a definitions section ofterms used. The EEOC guidelines define a few major terms,but there is no special section for definitions. The FEAguidelines were supplemented in January 1977 by a question-and-answer document. The EEOC has not issued a correspond-ing explanation of its guidelines.

In summary, the similarities between the FEA and EEOCguidelines indicate that the four agencies have continuouslyshared the objecti;e of requiring that job-relitedness(validity) be shown for selection procedures having an ad-verse impact on the employment opportunities of minoritiesand females. (See app. V.) The agencies have disagreed,however, on what employers must do to demonstrate validityand justify continued use of selection procedures whichhave adverse effects. The December 1977 publication ofproposed uniform guidelines in the Federal Register showsthat the agencies have reached tentative agreement on thesepoints.

EFFECTS OF INCONSISTENT GUIDELINES

Employers have criticized the Federal EEO agenciesfor adopting different sets of guid&_ines, thereby notproviding uniform information and consistent advice onthe minimum legal requirements for selecting employeesand validating tests. Since the jurisdictions of thevarious agencies coincide in certain areas, sometimes twodifferent sets of guidelines are imposed on the same em-ployers. For example, employers with Government contractssubject to supervision by the Department of Labor's Officeof Federal Contract Compliance Programs are also subjectto EEOC jurisdiction under title VII. Similarly, EEOCwould have authority to process complaints against Stateand local governmental units at the administrative level,while the Department of Justice has authority to carry the

15

proceedings into court. CSC, in a somewhat different situa-tion, might set the standards for State and local governmentalunits to use in screening employees in cases in which Federalmoney is used to upgrade personnel management; but EEOC wouldhave authority to process complaints of discrimination basedon testing practices.

The Chief Deputy Director of Persc,nnel of the Countyof Los Angeles has described the situations in which apublic employer encounters different Federal requirementson testing and employee selection as follows:

"Currently, employers are caught in a web of con-flicting enforcement strategies of the differentcompliance agencies. Compliance officers of thedifferent agencies sometimes have different viewsabout the interpretation of regulations. Moreover,different agencies frequently have different report-ing requirements which require the employer to main-tain complex and, in some cases duplicate record-keeping systems. Finally, employers sometimes haveseveral concurrent roles: grant-in-aid administra-tor, prime contractor, and employer with generalliability for observation of equal employmentopportunity laws. The same employer may be subjectlo regulations of OFCC [Office of Federal ContractCompliance] concerning his prime contractor respon-sibility; the Office of Revenue Sharing, the CivilService Commission, and other departments of theFederal Government concerning his grant-in-aidadministration responsibility; and EEOC as theemployer's decisions are affected by title VIIof the Civil Rights Act. Employers often need todevelop separate programs to manage each role."

An employer could devote time and resources to meetingthe requirements of one set of guidelines and yet fail tosatisfy the requirements of a competing set of guidelines.If a Federal contractor, for example, satisfied the Officeof Federal Contract Compliance standards (FEA guidelines),the company would be protected against sanctions or suitunder Executive Order 11246. (See app. III.) However, thecompany would still be vulnerable to suit under title VIIif its tests did not adhere to the EEOC guidelines.

When one Federal agency approves or requires a givenemployment practice, this approved course of action in noway protects an employer from violating a requirement ofanother Federal agency. The reaction of employers to thissituation, besides a mixture of anger and confusion overwhich directive to follow, may include dropping a lot oftests.

16

Advantages of uniform guidelines

In addition to their criticisms of the inconsistentagency standards, public and private employers have. citedvarious advantages which would result from the adoptionof uniform guidelines. Among those most frequently men-tioned are:

--Fairness. The some standards for test use andvalidation would be applied to the Federal Gov-ernment, private employers, and State and localgovernments.

--Efficiency. Test users could more readily developand validate tests if they were required to under-stand and follow only one set n' guidelines.

--Credibility. Providing one standard against whichtest users, regulatory agencies, and the publiccan measure and evaluate the results of a selec-tion program would lessen misunderstanding of thepurpose, appropriateness, and results of theGovernment's efforts to regulate employee selec-tion.

-- Consistency. In the case of Douglas v. Hampton,EEOC and CSC opposed each other on how to validatetests. Reconciling the differences in the guide-lines would enable the Government to develop andmaintain a consistent enforcement posture andlitigation position in testing cases. Withoutconflicting directives from the EEO enforcementagencies, it would be easier to achieve consis-tency in the way courts evaluate an employer'sselection practices.

REASONS FOR DELA!'

The EEO Coordinating Council members have been workingto develop and adopt uniform guidelines for the past 5years. Why is it taking so long for these agencies toreach agreement?

First, the agencies have had longstanding disagree-ments about guidelines requirements. This is becausetheir views have differed on the legal and technical stan-dards for judging the proper use of tests; also, the agen-cies perceive their mandates differently and have pursueddifferent operating responsibilities. No agency has beenwilling to adopt uniform guidelines which it believedwould seriously impair Its ability to operate and meet

17

program goals. EEOC, Labor, and CSC each had its ownrationale for developing its particular set of guidelinesand its own approach to putting them into practice andevaluating the results achieved.

The guidelines developed to carry out one of theseagencies' objectives can frustrate the objectives andrequirements of another, making it difficult to developa set of guidelines compatible with the varied operatingresponsibilities of all EEO enforcement agencies. Forexample, it appears that basic differences in the mandatesand responsibilities of CSC and EEOC hampered efforts toreach agreement on a common set of guidelines.

EEOC mandate

The main task of EEOC is to enforce the mandate oftitle VII to eliminate employment discrimination againstindividuals and groups. Finding the seluction processresponsible for more discrimination thain perhaps any otherarea of employment practices, EEOC developed its guide-lines on testing to help correct this. The guidelinesestablished the enforcement posture of the agency. Theymade it clear that EEOC required that tests resulting inadverse effects on the employment opportunities of racial,sex, or ethnic groups be validabed so that members ofthese groups would not be discriminated against.

The overall goal of the 1970 EEOC guidelines was toget employers to use selection devices and procedureswhich met their business or operational needs and yet hadthe least adverse impact because of race, sex, religion,or national origin. The short-term goal was to eliminatethe use of unvalidated tests which had adverse effects.If, in the long run, tests were validated but still hadadverse effects, then the goal was to see if employerscould develop alternate means of selection which satisfiedtheir needs but lessened the impact.

EEOC indicated that issuance and enforcement of theagency's guidelines brought important results whichincluded:

-- Reducing the use of discriminatory unvalidatedtests.

--Increasing employment opportunities for minoritiesand women.

--Encouraging dne testing community to develop morejob-related selection procedures.

18

The EEOC guidelines have been strongly supported byvarious civil rights groups, psychologists, and employers-for providing effective barriers aqainst the discrimina-tory use of selection procedures. The guidelines have beenrecognized by the courts in many testing cases. (Seech. 1.) The guidelines have given EEOC a sound enforcementposition and have helped achieve its mandate of eliminatingemployment discrimination. Adopting the FEA guidelines,which do not incorporate the definition of discriminationand many of the principles endorsed in the EEOC guidelines,would in the opinion of EEOC and others have impairedEEOC's ability to reduce employment discrimination againstracial, sex, or ethnic groups.

CSC mandate

The merit system laws and selection procedures underwhich CSC and many State and local governments operatewere not designed and have not been administered to achievea representative work force in terms of race, sex, orethnicity.

The Civil Service Act of 1883, which established CSC,called for bringing the lanst qualified individuals intopublic service by ranking candidates on the basis of eval-uated ability and fitness and use of a selection processthat honors this ranking. In the period since 1883, CSChas established an extensive recruiting and exanining net-work for the purpose of carrying out the requirements thatappointments be based on individual merit and that everyonebe given an opportunity to compete.

CSC is also responsible for overall supervision andenforcement of EEO laws in the Federal civil service.There is no inherent conflict between a merit selectionprogram and EEO; but nterit system procedures have, for themost part, not dealt with the meaning of equal opportunityfor groups nor the extent and appropriateness of actions toachieve the goal or a representative bureaucracy. This hasespecially been the case when the methods for obtainingthat representation have differed substantially fromcurrent merit procedural requirements, which cell forranked registers, selection on the basis of r.anking, andequal opportunity for all individuals to be considered.Regarding this, an article in the Civil Service Journal,a CSC publication, concluded:

"For Federal, and also for other employers operatingunder merit system requirements, the emphasisremains on individual merit. The mandate is for

19

valid, job-related procedures with economicvalue. Scientifically, these conditions canbe met, for psychometric theory is based uponindividual differences. A test can fairly andaccurately provide equal opportunity for indi-viduals to demonstrate ability to perform ajob. What the psychological measurement cannot dois provide a valid procedure that assures equalprobability of success for members of groups basedon characteristics unrelated to performance ability,when real ability levels differ among members ofthe groups."

Whereas the EEOC guidelines set forth the technicalrequirements employers had to meet to justify continueduse of tests shown to have adverse impact, the CSC testingstandards specified requirements for the development anduse of all Federal examinations, even if the procedureshad no adverse impact on any group.

The CSC standards made it clear that Federal examin-ing practices were based on the idea that tests which werereasonably related to job requirements did not discriminateon the basis of race or sex. The policies and proceduresset forth by CSC were intended to bring about the system-atic review and collection of evidence to verify the job-relatedness, validity, and conformity with merit principlesof Fecaral examining standards and practices. Compliancewith the CSC standards was intended to help maintain highquality staffing at reasonable costs and insure thatFederal examining practices were not affected by nonmeritfactors, such as discrimination.

The competitive examining and selection programs CSCuses and approves for Federal agency use are subject toreview and litigation under whatever guidelines are appli-cable to the Federal Government. CSC has contended thatthe Federal Government and other merit system employersdo not have the resources to operate under the definitionof discrimination and the validation standards containedin the 1970 EEOC guidelines. On the other hand, whileFederal agencies are concerned about the practical aspectsof putting into practice the FEA guidelines, CSC believesthe Federal Government and other merit system employerscan follow these requirements.

The different mandates under which CSC and EEOCoperate have promoted them to perceive and pursue EEO indifferent ways, to apply different standards in judgingthe proper use of tests, and to delay development ofuniform guidelines.

20

EEO Coordinating Council lacks authority

There is another reason for the long delays in adopt-ing uniform guidelines--the EEO Coordinating Council lacksauthority to carry out its mandate to develop and put intopractice agreements, policies, and practices among the EEOenforcement agencies.

The Council has no power to require member agencies tochange their regulations and guidelines or to adopt newones in tha interest of developing uniform Federal posi-tions on EEO matters. When agencies have disagreed, theEEO Coordinating Council has had no means of forcingcompromise and final agreement. Nothing that affectedthe significant interests of a member agency could be donewithout its consent.

EEOC believed the FEA guidelines would significantlyweaken its ability to enforce title VII; CSC, Justice, andLabor perceived the EEOC guidelines as imposing unrealis-tic, professionally outdated, and costly operating stan-dards on employers which in some cases could not be met.The Council could not get the agencies to resolve theirdifferences in the interest of timely development andadoption of uniform guidelines. These differences haveprecluded prompt development of uniform guidelines. Theadoption of uniform Federal guidelines is still not com-pleted.

CONCLUSIONS

When the EEO Coordinating Council members began meet-ing in 1973, they agreed that uniform guidelines could bedeveloped, since the principles of EEO and merit employ-ment were not contradictory. The two principles can, nodoubt, be defined in a consistent manner; the agenciesneed to strike a balance between the CSC approach to equalopportunity for the individual (individual merit), underwhich it has operated its examining and selection system,and EEOC's approach to equal employment opportunity, whichplaces greater emphasis on the need for employers to demon-strate that there is no discrimination against racial, sex,and ethnic groups. We believe this conflict in the waysCSC and EEOC operate has hampered development and timelyadoption of uniform guidelines.

The 5-year effort to develop uniform guidelines onemployee selection demonstrates that the mandated objec-tive of the EEO Coordinating Council--to develop and putinto practice agreements and policies which promote

21

efficiency and eliminate duplication and inconsistency inthe operations of the EEO enforcement agencies--is notlikely to be achieved in a tirlely manner, if at all,unless the niumcber aqencies compromise and reach aqreement.

AGENCf COMMENTS

Justice, CSC, and EEOC said that the differencesseparating them were not as deep seated as we have sug-gested; their view is that the present leadership of thefour agencies can soon reconcile these differences, developa uniform position, and a draft set: of guidelines. (Seeapps. VII, VIII, and X.)

We recognize that development of the October 1977draft oic uniform guidelines and publication of a draft in theFederal Register on December 30, 1977, indicate that the fouragencies can, with time, reconcile differences among them-selves. However, the fact that it has taken the agencies 5years to reach tentative agreement on guidelines provisionssuggests that the differences in their views have not beenminor. Also, the CSC statement (see app. VII) that detailedcomments on our report by each of the four agencies "wouldhave the effect of exacerbating differences thus making itmore difficult to achieve a uniform position" suggests thatmajor differences among the agencies may still exist.

We believe that basic differences in their mandatesand operating responsibilities prompted the EEO enforce-ment agencies, particularly EEOC and CSC, to perceive andpursue EEO in different ways, Which hampered efforts toreach agreement on uniform guidelines. Since the EEOCoordinating Council cannot force member agencies tochange their views, policies, or practices in order topromote uniform Federal positions on EEO matters, theprocess of resolving disagreements among the agenciesabout uniform guidelines has been a voluntary one. It hasbeen a slow, time-consuming process which is not yet com-plete. To carry out its mandate, the Council needs toestablish a means by which member agencies can agree on andput into practice consistent EEO policies and practices ina timely manner.

In its comments, Labor said that the report shouldhave discussed in greater depth the practical implicationsof having more than one set of guidelines. (See app. IX.)However, as we pointed out earlier in this chapter, it isuncertain what direct impact more than one set of guide-lines has on test users, test development, court decisionsin testing cases, and overall enforcement of EEO laws.Nevertheless, the existence of more than one set of guide-lines on tha same subject is poor Government policy. It

22

leads to confusion over what an employer must do to meetthe requirements of Federal EEO laws; it also causesconflicting EEO enforcement strategies and general mis-understanding and doubts about the Government's objectivesin regulating employee selection and the management ofEEO enforcement activities.

Justice believes the report

"does not adequately deal with the ramificationsof the Issue of achieving professionally accep-table standards for test validation that areconsistent with the intent of the Civil RightsLegislation passed by Congress." (See app. X.)

We wish to reiterate that the purpose of our review was todelineate some of the administrative, technical, and prac-tical considerations involved in developing, putting intopractice, and complying wi.th Federal guidelines on employeetesting and selection. All existing and proposed gu.de-lines are intended to be consistent with existing Federallaw and professional testing standards. We did not set outto determine which agency's approach to EEO was "correct."Rather, we have pointed out that the guidelines developedto carry out the requirements of cne law could frustratethe requirements of another, and make it difficult todevelop a set of guidelines compatible with the responsi-bilities of all the EEO enforcement agencies. Justice'srecommendation that steps be taken to establish a uniformposture concerning enforcement of civil rights legislationhas been accepted by the Congress, as shown by the creationof the EEO Coordinating Council, and this report, as notedabove, deals with pzoblems encountered in putting intopractice this congressional mandate.

RE, ENDATIONS

Jntil the Federal civil rights agencies are reorgan-ized or merged so that there are no overlapping programsand inconsistent enforcement and compliance standards, theagencies will need to coordinate their EEO enforcementpolicies and procedures to help insure uniformity.

The 5-year effort to adopt uniform guidelines onemployee selection procedures demonstrates that the objec-tive of the EEO Coordinating Council--to develop and putinto practice agreements, policies, and practices whichpromote efficiency and eliminate duplication ard incon-sistency in the operations of the EEO enforcement agen-cies--cannot be achieved in a timely manner unlessthe member agencies compromise and reach agreement. There-

23

fore, we recommend that the President direct the EEO Coordi-nating Council to establish a means by which member agenciescan agree upon and put into practice consistent EEO policiesand procedures witkout unreasonable and lengthy delays. ThePresident should also direct the Council members, in the spiritof such an agreement, to adopt and follow uniform Federalguidelines on employee selection procedures.

24

CHAMTER 3

OTHER PROBLEMS IN REGULATING EMPLOYEE

SELECTION TO ACHIEVE EEO

We looked at other concerns employers and testresearchers have expressed about Federal regulation ofselection practices, aside from the issue of uniform guide-lines. Private and governmental employers are directlyregulated and affected by the Federal guidelines on employeeiselection; we believe their views on the costs, benefits,and problems associated with following Federal guidelinesand achieving equal opportunity in employee selectiondeserve the careful attention of the Federal EEO enforce-ment agencies and the Congress. We did not verify theaccuracy of the information provided by individual employ-ers and employer interest groups.

REQUIREMENTS DIFFICULT TO UNDERSTAND

The subject matter discussed in the Federal guidelineson employee selection is technical and complex. While theguidelines make clear the need either to change or validateany test having adverse impact, many employers, especiallyin small organizations, complain that they cannot under-stand the Federal guidelines' technical standards forvalidating tests well enough to implement them. To complywith the guidelines, these employers believe they areforced into dropping tests which have any adverse impact,even though they may be job related, since they cannotmeet the guidelines' validation requirements.

The 1975 American Society for Personnel Administra-tion/Prentice-Hall survey indicated that the majority ofthe smallest firms polled (those employing 1 to 99 employ-ees! "encountered their biggest problems just trying tokeep current on EEO requirements." Six out of ten respond-ents in this group said they found the EEOC guidelines"somewhat confusing" or "very confusing." The majority ofthe firms in this group had not conducted validationstudies.

The larger firms surveyed generally said that theyunderstood the Federal guidelines on testing. Largerfirms were also more likely to validate their tests. Asimilar situation exists among governmental employers.Compared to large city and State governments and Federalagencies, small municipal governments and Federal agenciesare less likely to have professional personnel staffs with

25

the technical abilities necessary to understand the Federalguidelines.

Employers often have difficulty comprehending thetechnical requirements of the guidelines because neitherthey nor their personnel staffs have training in the areasof knowledge with which the guidelines deal (for example,psychometrics, validation strategies, advanced statistics).All sets of guidelines refer employers to American Psycho-logical Association Standards for guidance on how to per-form validation studies. These standards are designed asreference tools for persons who have advanced training ineducation or psychology. Since it is estimated that onlyabout 10 percent of all industrial testing programs aredirected by people who have graduate degrees in psychol-ogy, it is likely that most personnel practitioners havedifficulty understanding the "guidance" provided by theAmerican Psychological Association Standards.

Some employers have suggested that since most person-nel decisiuns are carried out by such practitioners, addi-tional guidance in the form of easy-to-read handbooks ormanuals on how to enforce and follow the guidelines wouldbetter insure that both test users and agency compliancepersonnel understand specific guideline requirements. Onthis point, one public personnel specialist has suggested:

"What is needed is a step-by-step description ofthe validation process with specific examples ofappropriate and inappropriate methods. Examplesof acceptable statistical p ocedures would beparticularly helpful for the inexperienced testadministrator and references for additionalinformation should be made to works that aredesigned for comprehension by the layman, notgraduate specialists in psychometrics, as is nowthe case."

The Federal EEO enforcement agencies do help employersinterpret guidelines' requirements and improve theiremployment practices. For example, to interpret andclarify various provisions, the three agencies issuing theFEA guidelines jointly published in the Federal Registerof January 21, 1977, a series of answers to commonly askedquestions about the guidelines. Some test users haveindicated that the questions and answers were not helpfilin their present form because they confused the technicalissues and seemed to create new requirements.

26

Conclusion

Some of the public criticism of Federal guidelines maybe the result of general confusion and misunderstandingabout what the guidelines specifically require an employerto do. Providing materials that clearly explain the guide-lines' requirements would probab'y improve agency enforce-ment and employer compliance efforts.

LACK OF RESOURCES HAMPERS COMPLIANCE

To our knowledge, there are no reliable estimates ofthe average cost of validating tests or otherwise followingand enforcing the mandatory requirements of Federal guide-lines. Test validation, in most cases, is a costly andtime-consuming process. Many employers, particularly gov-ernmental employers, have stated that they lack the finan-cial and professional resources to comply with the valida-tion requirements of the Federal guidelines.

The 1975 American Society for Personnel Administration/Prentice-Hall survey reported that the majority of the re-spondents (71 percent) spent less than $5,000 validatingselection procedures per job studied. Large firms tendedto spend more than small firms. The survey did not reporton the types of validation strategies used by the respond-ents.

In 1974 the city of Hartford, Connecticut, estimatedan average cost of about $5,000 per validation study (or$2 million total) to validate its selection proceduresalone, excluding other personnel actions. Hartford's per-sonnel director estimated that the city spent $100,000validating selection procedures for police officers.

The reported costs incurred by public employers inconducting criterion-related validity studies ranged from alow of $6,000 to validate an examination for firefightersin Minneapolis to $401,000 to validate a State police exam-ination in New York. Criterion-related validation studiesare generally more expensive to conduct than content valid-ity studies. We obtained no information concerning thequality of these studies and do not know whether thestudies met professional validation standards.

CSC has estimated that the costs of implementing andenforcing the FEA guidelines in the Federal agencies andat the State and local government level would run into sev-eral millions of dollars. These costs would be in additionto present spending, not in place of it. According to CSC,money must be budgeted for new data systems, increased

27

recordkeeping, training materials, additional staffing, andso forth.

Some public personnel officials have indicated thatthe budgetary support in State and local governments forpersonnel management is already severely limited. Theybelieve taxpayers will not approve the additional expendi-tures necessary to follow Federal guidelines and carry outextensive validation programs. Without the necessaryfunding, smaller personnel departments may have to eitherdiscontinue testing when there is evidence of any adverseimpact or else corcentrate scarce resources into just onearea--employee selection procedures. Federal agency per-sonnel officials have also expressed doubts about obtainingthe funding neede. to comply with the FEA guidelinesthrough the Government budgetary process.

Clearly, employers are concerned about how co obtainand allocate resources to be spent meeting guidelinerequirements in relation to resources needed for otherpersonnel management priorities. Employers have alsoexpressed reservations about obtaining and paying for theservices of competent researchers to conduct validationstudies. Generally, to meet the technical standards ofthe guidelines requires the services of qualified testingpsychologists. The American Psychological Associationhas stated that there is "a very limited supply of profes-sional industrial psychologists qualified to design andcarry out validation research." The professional resourcesneeded to develop and validate measurement devices thatwill meet the guidelines' technical standards are inade-quate. Reportedly, individuals who are competent in theseareas can earn salaries well above $30,000 annually.These services, even on a consulting basis, are oftenconsidered too expensive for many public and nrivateemployers to use.

Conclusion

It is clear that some employers believe the Federalguidelines have been developed without emphasis beingplaced on whether or not it is feasible to follow them,in terms of either costs or the availability of profes-sional resources. Public personnel officials have sug-gested that financial implications and the lack ofsophisticated personnel practitioners at all levels ofgovernment--local, State, and Federal--preclude effectivelyimplementing the guidelines.

Some employers may overstate the real costs involvedin trying to follow the guidelines and val dating thei.

28

tests; perhaps they do this to create a reason wh:y valida-tion cannot be done or other guidelines' requirements metin an administratively feasible way. Since the key factorin Federal guidelines for requiring test validation isadverse impact, some employers may use alternate selectionprocedures or privately impose race or sex hiring quotasin order co achieve representative employment among appli-cant groups -nd thus avoid validating their tests. Indoing so, objectivity, reliability, and validity of theselection process may be overlooked or sacrificed. Suchactions contribute to the view expressed by some employersthat the emphasis Federal EEO agencies place on validtesting is primarily a smokescreen to promote greaterhiring of women and minorities, rather than a legitimateeffort to promote development of job-related tests whichhave as little adverse impact on race, sex, and ethnicgroups as possible.

PROBLEMS WITH MULTIAGENCY STANDARDS

Since the adoption of the FEA guidelines by Labor,Justice, and CSC, some test users have expressed concernsabout the practical problems involved in following mulLi-agency standards. With adoption of any common set ofguidelines, questions arise about their consistent enforce-ment and interpretation and the time allowed to comply withthem.

There is no provision within the FEA guidelines for asingle source of information on and interpretation of theguidelines. Each agency decides how it will use them inits EEO enforcement activities. Some test users believethe compliance personnel of the three agencies vary agreat deal in the ways they interpret and apply guidelines.They point out that if the agencies place different empha-sis on important features of uniform guidelines, thenconsistent interpretation and enforcement will be hampered.

In addition, the FEA guidelines do not discuss acommon time limit for requiring employers to meet theminimum recordkeeping and documentation requirementsimposed by the guidelines. CSC has indicated that puttingthe FEA guidelines into practice Government-wide willrequire careful planning and much leadtime. CSC hasestimated that State and local governments collectivelyalso need several years to meet FEA recordkeeping anddocumentation requirements. Time is needed to change ordevelop new data systems, validation strategies, trainingprograms, and so forth. Labor, however, has been eval-uating validity studies submitted by Federal contractors

29

in accordance with the FEA requirements since December 1976.

Conclusion

Many factors influence tha willingness, ability, andthe time it takes an employer, particularly a publicemployer, to validate tests or make changes in selectionpractices for EEO purposes. A 1975 report by the PublicServices Laboratory of Georgetown University on affirmativeaction in city personnel systems noted:

"While there is some progress toward and interestin the elimination of artificial barriers, poorlocal EEO statutes, little personnel office auton-omy, low staffing levels, and questionable techni-cal competence in the majority of personneldepartments--all make intelligent compliance withaffirmative action mandates and test validationrequirements problematical if not impossible."

It may be useful for the EEO enforcement agencies toevaluate the factors which influence the ability ofpublic and private employers to follow the minimum record-keeping and documentation requirements and develop realis-tic time limits for requiring compliance with variousaspects of the guidelines. In this way, certain groupsof employers, such as Federal contractors, would not beforced to move to meet the guidelines'requirements fasterthan others, nor would they be permitted to lag behind.

To avoid the problems of inconsistent enforcementstrategies and interpretation of guidelines, it may behelpful, as one municipal personnel director has suggested,to continuously review the ways the agencies apply theguideline provisions. Additionally, training the agencies'enforcement staffs to interpret and uniformly apply theguidelines would more readily assure consistent enforce-ment efforts and similar agency explanations about how tocomply with the guidelines.

OVEREMPHASIS ON SELECTION

The controversy over developing uniform guidelineshas focused almost exclusively on the pros and cons ofvalidation techniques and the need to include or omitspecific technical requirements. Some employers contendthat the Federal EEO agencies have placed too great anemphasis on technical issues, and have failed to developan evaluation component to help insure that guidelinesdo, in fact, have a positive impact on selection programs

30

and minority and female employment patterns.

According to some employers, many organizations areencouraged by Federal guidelines to spend their personnelresources on indepth test validation studies even thoughthis may not likely produce an increase in the number oftests, especially written, that do not have an adverseimpact on minorities. These employers believe thatresources spent on test validation to follow Federalguidelines would in many cases be better spent elsewherein the personnel system (for example, on recruiting, jobrestructuring, training, and so forth) for EEO purposes;the resources might also be used to make changes to civilservice laws and regulations (veterans' preference,apportionment, and so forth) which may restrict the employ-ment opportunities of women and minorities in Federal,State, and local governments. Employers believe that anincrease in resources to improve testing generally meansthat fewer resources are available to make improvements inother aspects of personnel management which would contri-bute to the hiring and promotion of women and minorities.

A 1976 Bureau of National Affairs survey of companies'EEO programs indicated that three-fifths of the companiesresponding had made some changes in their selection tech-niques for EEO reasons. Statistics in the same studyshowed increases in the percentages of minorities andfemales employed ty these companies. Such survey findingsmay indicate that the visibility of enforcement activitiesinvolving Federal guidelines has prompted increasing num-bers of employers to validate their selection proceduresor otherwise ramove arbitrary barriers to minority andfemale employment.

Conclusion

In order to adequately evaluate the costs, benefits,and impact of guidelines on selection programs and employ-ment patterns, the Federal agencies responsible for enforc-ing the guidelines would have to collect and report currentand reliable data on the actual costs and results ofadministering, putting into practice, and complying withthe guidelines. Federal efforts to regulate selectionpractices should not overshadow the need to investigateand improve other parts of a personnel system or civilservice laws which are not specifically regulated by guide-lines but offer substantial EEO rewards. Employersbelieve, and we agree, that the Federal EEO enforcementagencies have an obligation not just to regulate employmentpractices, but to assist in developing and testingcost-effective methods of achieving EEO.

31

PROFESSIONAL CONCERNS ABOUT FEDERAL GUIDELINES

In attempting to develop uniform guidelines, some ofthe biggest disagreements among the agencies have focusedon the proper interpretations of professional testingconcepts. Early in the professional debate over the impli-cations of title VII and the development of Federal guide-lines, minority group selection was considered a fairlysimple problem. Those in the psychological profession nowappear to be moving toward a view that minority groupselection is a complex social problem whose remedy is notinherent in "better" selection measures.

Initially, psychologists believed that scientificanalysis of the relationship between test scores and jobperformance might support a hypothesis of differentialvalidity--a hypothesis that tests valid for the majoritygroup were less valid or invalid for the minority group,thus unjustly rejecting minorities by erroneously predict-ing their inability to perform on the job. Tests, it wasthought, simply did not predict performance as well forsome groups as for others. If the hypothesis was supported,the solution was considered simple: Valid tests would besubstituted for invalid tests, or those which were lessvalid would be weighted differently in making selectiondecisions for minorities.

By 1968, two studies had been published which appearedto support the hypothesis of differential validity inindustrial selection, although a larger number of studiesindicated no such support in college and military settings.The industrial studies were sufficient for a requirementof separate validations to be incorporated into the 1968Labor Department Testing Order and the 1970 EEOC guidelines.

Professional research studies since 1968 have calledinto question the methodological and mathematical sound-ness of the earlier studies which supported differentialvalidity. More sophisticated analysis has since shownthat studies supporting the differential validity hypothe-sis are best explained by chance and thus prove nothing.Consequently, most concerned professionals today expectthat a test found valid for majority group members Wouldbe just as valid for minorities. In essence, when testsare valid predictors of majority group performance andminorities score lower on the tests as a group, they alsodo less well as a group on the job.

32

Psychologists have developed a number of statisticaldefinitions and models of "fair" selection tests. Onecommentator, Nancy Cole of the University of Pittsburgh,has divided these models into three classes. The firstclass simply says that fairness is accomplished whendeserving groups are adequately represented among selectedapplicants. Test scores can be used to aid in selection,but are not permitted to interfere with achieving adequaterepresentation. This quota-based model has never beenaccepted by those in the psychological profession sinceit does not permit selection on the basis of job-relatedtraits when such selection would lead to disproportionaterepresentation among applicant groups.

The second class of models says that fairness istreating individuals alike according to their likelihoodof performing well. Until about 1971, this class of modelswas largely unchallenged by psychologists as the properprocedure for selection. These models incorporate suchsocial values as maximizing prcductivity, basing competi-tion on merit and equal opportunity for the individual,and minimizing state intrusion in the affairs of business.

The third class of models says that fairness is treat-ing groups alike according to how well they have actuallyperformed. Models in this class differ from those in thesecond in that they focus on equality of opportunity forgroups rather than individuals, and on actual as comparedwith predicted performance. These models incorporate thesocial value of compensating members of certain groupsfor inequality of opportunity.

In most testing situations, models of the second andthird class lead to opposite conclusions about test fair-ness. One professional research study points out thatwhen a model of the second class was applied to publishedstudies on test validity, minority group job performancewas predicted to be higher than it actually turned out tobe. Thus, tests are either unbiased or biased in favor ofminorities according to this model. When a model of thethird class was applied to the same studies, the researchersfound many tests would be judged unfair to minority groupssince fewer of them were selected than should have been,based on job performance.

The EEOC and FEA guidelines hold that a test is fairif test scores accurately predict future levels of jobperformance for all groups. These guidelines and theOctober 1977 draft of proposed uniform guidelines call forstudies of test fairness. Some psychologists believe that

33

research evidence is now very strong that tests which arevalid and fair for whites are just as valid and fair forminorities, so that there is no basis for continuing therequirement to perform test fairness studies.

Conclusion

The professional debate over models of test fairnessessentially concerns social values and focuses on how toachieve equal opportunity for all job applicants while atthe same time assuring full minority, ethnic, and sexgroup participation in the Nation's work force. Thesedebates revolve around a conflict in value systems--thosewhich emphasize individual merit versus those which stressgroup equality. Relationships between these concepts havebecome confused so that it is difficult to know what stand-ards to use to evaluate the fairness of selection proce-dures. If one model of test fairness gains explicit favorin the professional community and is endorsed by a FederalEEO agency, it will have an important impact not only ontest development and validation, but also on the conceptand goals of EEO.

AGENCY COMMENTS

Justice was the only agency to comment on the issuesdiscussed in this chapter. (See app. X.) Justice believesthe report does not adequately recognize the "crucialdifference between guidelines application to the publicversus private sectors."

In Justice's view, private employers have greaterflexibility in the procedures they use to make employmentdecisions than do public employers operating under civilservice procedural requirements. Justice believes privateemployers can more readily comply with the Federal EEOguidelines than public employers. One of our previousreports discussed some fundamental problems which affectthe Government's EEO efforts. 1/ We questioned whetherEEO program objectives and efforts might be in conflictwith merit system procedures and practices, and if so,what could or should be done to eliminate or minimize theconflicts.

It came to our attention in this review that somecivil service procedural requirements, such as veterans'

l/"Problems in the Federal Equal Employment OpportunityProgram Need to be Resolved," FPCD-76-85, Sept. 9, 1977.

34

preference, apportionment, and selection among the topthree candidates on a register, are perceived as barriersto the employment opportunities of women and minoritiesin the civil service. In a recent GAO report, we dis-cussed how women's opportunities for Federal employment arediminished by veterans' preference and apportionment. 2/CSC and the President's Reorganization Project are investi-gating these problems and how to resolve them.

The Federal EEO guidelines do not require "an adjust-ment of procedures to result in more minority selections,"as Justice suggests. The Federal guidelines require thatwhen there is adverse impact, thd user must demonstrateand document the validity of the selection procedures,modify the procedures to eliminate adverse impact, orsearch for alternate selection methods which are valid buthave less adverse impact. Civil service merit requirementsdo not preclude following the EEO guidelines. Federalmerit system policy calls for all selection procedures tobe job related and valid. We believe the validation re-quirements of the guidelines reinforce and support meritprinciples and requirements.

Justice believes the report lacks "conclusive informa-tion necessary to perform a knowledgeable analysis of whatmight be required to achieve compliance." We agree thatthere are serious obstacles to following the guidelines.The most serious obstacles appear to be obtaining thefinancial and professional resources necessary to fulfillguideline requirements. To our knowledge, there are noreliable average estimates of the professional and finan-cial resources needed and the costs involved in validatingtests or otherwise enforcing and following any mandatoryguideline provisions.

There are no projections on how cost effective anyset of uniform guidelines will be in terms of achievingEEO. We believe it is important to determine the cost ofguidelines, the ability of employers to obtain the finan-cial and professional resources necessary to put them intopractice, and their effect on selection programs andminority and female employment patterns.

We believe the EEO Coordinating Council should collectand report information on the actual costs and effects ofenforcing and complying with uniform guidelines once

2/"Conflicting Congressional Policies; Veterans' Prefer-ence and Apportionment vs. Equal Employment opportunity,"FPCD-77-61, Sept. 29, 1977.

35

adopted. We also believe the Council should developenforcement standards which take into account the abilitiesof employers to meet guideline requirements in terms ofthe financial and professional resources available to theemployer and the time needed to comply.

Justice also believes the report should address the"state of the art" regarding testing. This topic wasdiscussed in considerable detail in one of our previousreports. Regarding the "state of the art" of testing, wepointed out:

"Practical limitations in the art of personneltesting and measurement restrict the degree ofaccuracy attainable and prevent either theassembled or unassembled examination from beingperfectly reliable or valid. Improved evaluationprocedures can reduce some of this imprecision, buta large part is irreducible. As a result, theexamining process cannot accurately rate andrank comparably qualified applicants in exactorder of competence." 3/

We agree with Justice that there is considerable roomfor the exercise of professional judgment in test valida-tion. The FEA guidelines do establish detailed documenta-tion requirements for test validation studies. In thisrespect, the FEA guidelines follow the style of the Ameri-can Psychological Association testing standards by listingsome documentation requirements as "essential." TheAssociation found the FEA guidelines requirements profes-sionally sound and flexible. The documentation require-ments may, however, appear inflexible and confuse employersand others who are unfamiliar with professional testingterms and methods.

Finally, Justice warns against "the pervasive coverageof the guidelines," which the agency believes "is inconsis-tent with ana in conflict with other existing regulations."While the guidelines may be open to legal challenge onchis basis, we believe CSC and .he President's Reorganiza-tion Task Force are working to minimize conflicts betweenmerit system procedural requirements and EEO programobjectives, such as compliance with EEO guidelines.

3/"Improvements Needed in Examining and Selecting Appli-cants for Federal Employment," B-179810, July 22, 1974.

36

RECOMMENDATIONS

The agreements, policies, and procedures developedand put into practice by the EEO Coordinating Council areintended to "maximize effort, promote efficiency, andeliminate conflict, competition, duplication, andinconsistency among the operations, functions, and juris-dictions t' of the Federal EEO enforcement agencies. Toascertain how well uniform guidelines on employee selec-tion procedures fulfill this mandate, we recommend thatthe Council in its annual report to the President and theCongress present current, reliable information on the actualcosts and effects of implementing, enforcing, and complyingwith the uniform guidelines that are adopted. To promoteeffective enforcement of and compliance with such guide-lines once adopted, we recommend that the EEO CoordinatingCouncil

--Develop and issue documents which clearlyexplain the guidelines and show how to followthem.

--Develop enforcement standards that allow publicand private employers to meet the guidelineswithin a reasonable time and yet not exceedtheir financial and professional resources.

--Train agency enforcement personnel to apply theguidelines in a consistent manner.

--Review the agencies' use of the guidelines andmake changes as needed to maintain consistentenforcement.

--Encourage member agencies to fund research and providetechnical assistance to employers for developing andtesting coat-effective methods of achieving equal op-portunity in employee selection.

37

CHAPTER 4

SCOPE OF REVIEW

To identify legal, technical, and administrativeissues involved in achieving EEO in employee testing andselection, we reviewed the selection and examining guide-lines issued by CSC, Labor, Justice, and EEOC; we also re-viewed professional psychological standards for test develop-ment and use. The legislative background of title VII andSupreme Court and other judicial decisions on employmenttesting were also examined.

We obtained information from representatives of theEEO Coordinating Council on efforts to develop uniformFederal guidelines and the impact of the various Federalguidelines on EEO. We also obtained views of groups rep-resenting public and private employers, civil rights inter-ests, test publishers, and psychologists. We discussed withthem the practicalities, costs, benefits, and problemsassociated with employment testing, test validation, andFederal guidelines. Professional literature on testingwas also reviewed.

This report is one of several resulting from a June1973 request Mrom the Chairman, Senate Committee on Laborand Public Welfare, concerning the implementation of theEEO Act of 1972 as it applies to Federal employees. (Seeapp. IV for a list of our other reports.)

38

APPENDIX I APPENDIX I

GLOSSARY

Adverse impact A lower rate of selection for aracial, ethnic, or sex group com-pared to other groups. The FEAguidelines state that a selectionrate for a particular racial, sex,or ethnic group which is less than80 percent of that of the mostsuccessful group is generally re-garded as evidence of adverse im-pact.

Affirmative action An EEO plan for development of em-plan ployment goals for minorities and

women and timetables for theiraccomplishment.

Criterion (plural: A measure of job performance orcriteria or other work-related behavior againstcriterion which performance on a test ormeasures) other predictor measure is compared.

Fair employment A practice through which peoplepractice with the same expectancies

(probabilities) of success on thejob have the same probabilitiesof being hired. It should berecognized, however, that competingand mathematically incompatibledefinitions of fairness (or itsopposite, bias) are proliferating.It cannot be said, therefore, thatthe psychological profession hassettled on a mathematically precisedefinition of fairness in the use oftests. The definition given hereis chosen because of its priorityin time and familiarity to testingspecialists and because it makesa minimum of mathematicalassumptions.

39

APPENDIX I APPENDIX I

Job analysis A study of work performed to de-termine what is to be done on ajob, the procedures followed indoing it, and the knowledge, skill,and employee behaviors necessary tocarry out these tasks. From jobanalysis one infers the characteris-tics of successful performance(-criteria) and the personal char-acteristics which lead to it(predictors).

Job related A test is job related if it samplesknowledges, abilities, skills, orother characteristics shown tobe necessary or important forsuccessful performance of a job;a test which is significantly re-lated to an appropriate criterionmeasure of job performance.

Merit system, A personnel system which includescoverage an objective, nonpolitical method

of selection and promotion, andprovisions of tenure.

Technically In criterion-related validityfeasible studies:

a. Having or obtaining asufficient number of in-dividuals to achieve findingsof statistical and practicalsignificance.

b. Having or being able to obtaina sufficient range of scoreson the test and job per-formance measures to producevalidity results which can beexpected to be representativeof the total applicant sampleresults if the ranges normallyobtained were utilized, and

c. Having or being able to deviseunbiased and reliable measuresof job performance or othercriteria of employee adequacy.

40

APPENDIX I APPENDIX I

Validation The process of investigation by whichthe validity of a particular type oftebt use is estimated. What is im-portant here is to identify an am-biguity in the term "to validate"which is responsible for much con-fusion in the area of employmenttesting. To validate in ordinarylanguage may mean to mark wit-h anindication of official approval;in this sense it is also possibleto "invalidate" or to indicateofficial disapproval. In thetechnical vocabulary of employmenttesting, to validate is to in-vestigate, to conduct research.Thus in validating a test (moreproperly, in validating a use ofa test), one is conducting an in-quiry. In this context, the term"invalidating" has no meaning atall.

41

APPENDIX II APPENDIX II

PROFESSIONAL TESTING STANDARDS AND TERMS

Professional standards for tests

In 1966, a joint committee of the American Psycho-logical Association (APA), the American Educational Re-search Association, and the National Council on Measurementsin Education published a guide to producers and users oftests and devices for diagnosis and evaluation. The APAStandards, as these came to be known, built upon earlierpublications of the three organizations. The APA Standardswere revised and republished in 1974 partly as a resultof awakened concern over serious misuses of tests, in-cluding employment discrimination.

Growing concern over professional standards for em-ployee selection research also led APP's Division of Indus-trial and Organizational Psychology to publish its own guide-lines in 1975. The Division 14 Principles were meant to beconsistent with the APA Standards and to clarify their ap-plicability to the specific problems of employee selection,placement, and promotion. These sets of guidelines werenot intended as legal documents or to set minimal standardsof professional practice. Rather, they were to provide akind of checklist for test developers and users to considerin the designing, selection, administration, scoring, andinterpretation of tests.

Testing terms and concepts

Tests are sometimes called predictors because theyare used to assess applicant characteristics and makepredictions of future job performance. Validation is theprocess of determining whether and to what extent a testmeasures what it is supposed to measure or the accuracyof inferences drawn from test scores. Thus, in validatinga test, one is conducting an inquiry. Three methods oftest validation are generally recognized as basic: con-tent, criterion-related, and construct validation.

Content validation

A content validation study is designed to determinewhether the content of a test or measuring instrument(the questions asked or the activities required in atesting situation) adequately samples the universe of

42

APPENDIX II APPENDIX II

skill, knowledge, or behavior it was designed to assess.A classical content validity inquiry deals with the ex-tent to whlzh an educational achievement test fairlysamples the content of material presented in a course ofinstruction or an employment test explicitly samples theskills and qualities that will be required in performing ajob.

Content validation is a judgmental process which in-volves a systematic, deductive analysis of a job and pre-senting evidence and making inferences about the testitself as a sample of a knowledge or skill area. Foremployment tests, the determination of the appropriateknowledge or skill area is based upon job analysis. Jobanalvsis refers to any method of obtaining informationabou: jobs.

Criterion-related validation

Criterion-related validation studies determine towhat extent test scores may be used to infer the level ofperformance on an independent variable called a criterion.Criterion-related validity has sought to determine theextent to which an individual's relative standing on atest correlated with his or her relative standing on suchorganizationally relevant criteria as course grades,probability of turnover, performance rating, salesvolume, hourly output and percentage scrap produced, andso on.

Predictive validity is one method of investigatingcriterion-related validity. It requires generally thattest scores be collected from applicants who are hiredwithout consideration of test scores, and that criterionmeasures of their jr)b performance be collected at somelater point in time. The degree to which test scorespredicted criterion performance is then determined.

The other method for investigating criterion-relatedvalidity is referred to as concurrent validity. It in-volves concurrently administering tests to and collectingcriterion performance data for employees. Under certaincircumstances, data on concurrent validity may be used toestimate a test's predictive validity.

A related term, differential validity, refers to thenotion that there are differences in criterion-related

43

APPENDIX II APPENDIX II

validity for particular subgroups of applicants; that atest which is valid for one group of applicants may notbe equally valid for another group. For example, differen-tial validity would be said to ex:ist if a test has dif-ferent validities for blacks and ,yhites, or for men andwomen.

Information about the extent of criterion-relatedvalidity is often presented in terms of correlation co-efficients. Exact mathematical tests determine whether acorrelation coefficient is sufficiently large to indicatea relationship between standing on the test and standingon the criterion. As indicated on page 2, discussionsof testing benefits have historically been couched in termsof correlations.

Construct validation

Construct validity is used when one claims that atest measures observable consequences of a theoreticalidea. Psychologists use the term "construct" to refer toa general trait that is not observable, but is "con-structed" from psychological theory about how peopleperform. Examples of psychological constructs wouldinclude "anxiety," "clerical aptitude," "mechanicalability," and "leadership." Construct validity can bedefined as a relationship between a test and a theorydemonstrated on both logical and empirical grounds. Whencunstruct validity is used in selection, it must be shownthat the test measures the trait it purports to measure,and that the trait is related to job performance andunderlies or explains observable variations in jobperf raiance.

Construct validity is rarely used in employmentselection. There is a lack of substantial literaturedealing with the application of construct validity toemployment practices. No detailed requirements forits use were presented by Division 14.

Reasons for wlidating tests

Test valiaation is costly and time consuming. Thereare, however, important benefits associated with usingvalidated selection procedures. Valid selection proceduresincrease the probability that individuals most likelyto succeed on the job are hired. Validation helps insure

44

APPENDIX II APPENDIX II

that non-job-related factors which unfairly discriminateare eliminated from the selection process. From a costperspective, validation can save an organization muchmoney over the years by improving selection techniques.Use of valid selection procedures can reduce employeeturnover and training costs, increase worker produc-tivity and overall organizational efficiency.

Employees and applicants also derive many benefitsfrom the use of validated selection procedures. As onepublic personnel official pointed out, "a properlyplaced employee is a productive person, a happy person,and generally a contributor not only to the workingsetting but to the community, his home, and elsewhere."

45

APPENDIX III APPENDIX III

OTHER EEO LAWS AND EXECUTIVE ORDERS

Title VI of the 1964 Civil Rights Act,as amended,and comparable provisions in other Federal grant statutesprohibit discrimination in programs and activities re-ceiving Federal financial assistance. While Justice hasresponsibility for coordination, responsibility for en-forcing title VI and the civil rights provisions of othergrant statutes rests with the Federal agencies which ex-tend financial assistance. Each of the agencies can pro-mulgate guidelines to help insure that employment in thegrant program is not subject to discrimination.

Executive Order 11246, as amended by Executive Order11375, prohibits discrimination against employees orapplicants on the basis of race, creed, color, nationalorigin, or sex by Federal Government contractors, sub-contractors, and federally assisted construction con-tractors. It requires contractors to take affirmativeaction to promote EEO for minorities and women. TheOffice of Federal Contract Compliance Programs monitorsthe Government-wide contract compliance program andissues various rules and regulations to carry out thisExecutive order.

Executive Order 11478, issued in 1969, requires non-discrimination in Federal employment and specifies re-quirentents for implementing affirmative action programsin Federal agencies. CSC is responsible for overallsupervision and enforcement of these programs.

CSC is also responsible for administering the CivilService Act of 1883. This law requires that appointmertsto the Federal service be based on merit and fitness,and determined through open competitive examinations.CSC considers the merit principles of civil service lawsand the EEO provisions of Executive Order 11478 andtitle VII to be synonymous operating concepts.

Under the Intergovernmental Personnel Act of 1970,CSC is authorized to provide technical and financialassistance to State and local governments and theiragencies for improving personnel management and employeetraining, with a concerted emphasis on EEO. Under theAct, CSC also develops and administers merit systemstandards to State and local agencies receiving funds

46

APPENDIX III APPENDIX III

under certain Federal programs. In the overall area ofpersonnel selection, the standards have called for opencompetition, test validity, employment of the most com-petent, and affirmative action to help insure EEO.

The 1972 amendments to the Civil Rights Act of 1964fixed the Federal Government's obligation to make allpersonnel actions free from discrimination. CSC was givenauthority to enforce EEO in the Federal Government"through appropriate remedies" and to issue such rules,regulations, orders, and instructions as deemed necessary.The Federal courts were afforded full enforcement powersthrough actions brought by aggrieved persons after finalagency disposition of or failure to act on the complaint.

The EEO Act of 1972 eliminated the exemption ofState and local governments from the requirements oftitle VII. Charges of discrimination by State and localgovernments must be filed with. EEOC. If no conciliationagreement is reached, only the aggrieved person or theAttorney General may bring a civil action.

The 1972 amendments give EEOC authority to filecivil actions against private employers, labor unions,and employment agencies, as well as responsibility for"pattern or practice" litigation previously exercisedby the Attorney General. However, EEOC was denied directenforcement capabilities, such as the power to issue cease-and-desist orders. The EEO Act of 1972 emphatically re-affirmed a preference for ultimate judicial enforcementof title VII.

47

APPENDIX IV APPENDIX IV

OUR REPORTS ISSUED ON THE SUBJECT

OF THE EEO ACT OF 1972

"National Aeronautics and Space Administration's EqualEmployment Opportunity Program Could Be Improved,"FPCD-75-107, Apr. 16, 1975.

"Upward Mobility Programs in the Federal GovernmentShould Be Made More Effective," FPCD-75-84, Apr. 29,1975.

"Equal Employment Opportunity Discrimination ComplaintSystems for Civilian Employees at Selected ArmyInstallations," FPCD-75-118, May 28, 1975.

"Report on Management Information Needs of the EqualEmployment Opportunity Program of the Federal Government,"Civil Service Commission, FPCD-76-71, Sept. 3, 1976.

"General Services Administration's Upward MobilityProgram," FPCD-76-84A, Sept. 13, 1976.

"Review of Upward Mobility Using Job Restructuring,"Department of the Interior, FPCD-76-84, Nov. 2, 1976.

"Review of the Department of the Army's Upward MobilityProgram," FPCD-77-3, Dec. 13, 1976.

"Upward Mobility Using Job Restructuring," Department ofthe Navy, FPCD-77-9, Jan. 7, 1977.

"Upward Mobility Using Job Restructuring," Department ofthe Air Force, FPCD-77-8, Jan. 10, 1977.

"Upward Mobility Program Can Be Improved," Department ofAgriculture, FPCD-77-2, Mar. 21, 1977.

"Progress Made by Agencies in Implementing UpwardMobility Programs," Civil Service Commission, FPCD-77-10,Mar. 28, 1977.

"System for Processing Individual Equal EmploymentOpportunity Discrimination Complaints: ImprovementsNeeded," FPCD-76-77, Apr. 8, 1977.

48

APPENDIX IV APPENDIX IV

"Problems of the Federal Employee Equal EmploymentOpportunity Program Need to Be Resolved," Civil ServiceCommission and other Federal Agencies, FPCD-76-85,Sept. 9, 1977.

"Conflicting Congressional Policies: Veterans'Preference and Apportionment vs. Equal EmploymentOpportunity," FPCD-77-61, Sept. 29, 1977.

49

APPENDIX V APPENDIX V

AREAS IN WHICH GUIDELINES AGREE

1. Similar purpose and coverage

The EEOC and FEA guidelines and the October 1977draft of proposed uniform guidelines are intendedto assist employers in following the requirementsof Federal law and apply to tests and other selectionprocedures used as the basis for making employmentdecisions.

The existing and proposed guidelines define em-ployment decisions to include hiring, promotion,referral, retention, and so forth.

2. Requirement to validate

The EEOC guidelines require evidence of validitywhen the use of any test has an adverse impact onthe employment opportunities of a racial, sex,orethnic group. The FEA guidelines and the October1977 draft state that if the overall selection pro-cess has an adverse impact, the individual selectionprocedures should be analyzed; and evidence ofvalidity is required only for those L£ l:ction pro-cedures which have an adverse effect.

3. Recognition of validity strategies

Both the EEOC and FEA guidelines and the October 1977draft recognize that the three types of validitystudies--criterion-related, content, and construct--may demonstrate job-relatedness.

4. Recognition of APA Standards

The EEOC and FEA guidelines and the October 1977draft contain references to the APA Standards forguidance on how to perform validation studies.

5. Require entry-level testing

The EEOC and FEA guidelines and the October 1977draft require an employer to evaluate applicants forentry-level jobs unless employees can be expected tomove to a higher-level job in a reasonable period of

50

APPENDIX V APPENDIX V

time.

6. Alternatives to validation

If an employer is unable or unwilling to validate,the EEOC guidelines provide the option of changingselection procedures to eliminate conditions sugges-tive of discrimination. The FEA guidelines and theOctober 1977 draft also suggest modifying selectionprocedures to eliminate adverse impact or searchingfor alternative selection procedures which haveless adverse impact if it is not feasible or de-sirable to validate.

7. Cutoff scores

The EEOC and FEA guidelines -nd the October 1977draft state that if cutoff scores are used, theyshould be reasonable and consistent with normalexpectations of acceptable job proficiency.

8. Transporting validation studies

Under certain conditions, both sets of guidelinesand the October 1977 dr-ft permit employers to usevalidation studies conducted in other organizations.

9. No assumption of validity

Both sets of guidelines and the October 1977 draftstate that the general reputation of a test, itsauthor, or its publisher will not be accepted in lieuof evidence of validity.

10. Disparate treatment

Both sets of guidelines and the October 1977 draftstate that employees or applicants denied equaltreatment because of prior discriminatory practicesor policies must be afforded the same opportunitiesaV had existed for other employees or applicantsduring the period of discrimination.

11. Affirmative action

The EEOC and FEA guidelines and the October 1977draft itate that the use of tests validated pursuant

51

APPENDIX V APPENDIX V

to the guidelines does not relieve employers oftheir obligations to undertake affirmative action toassure EEO. The October draft discusses affirmativeaction in much greater detail than the EEOC and FEAguidelines. Regarding affirmative action, the draftstates that the guidelines are intended to encourageadoption and implementation of voluntary affirmativeaction programs. The draft endorses both for privateemployers and governmental employers the EEOCoordinating Council's "Policy Statement on Affirma-tive Action Programs for State and Local GovernmentAgencies" (41 Federal Register 38814, Sept. 13, 1976).The October draft repeats some of the major sectionsof the policy statement, as follows:

--"Voluntary affirmative action to assure EEO isappropriate at any stage of the employment process.The first step * * * should be an analysis of theemployer's work force to determine whether per-centages of sex, racial, or ethnic groups in individualjob classifications are substantially similar to thepercentages of those groups in the relevant jobmarket who possess the basir job-related qualifica-tions. When substantial disparities are found, eachelement of the overall selection process should beexamined to determine which elements operate to ex-clude on the basis of sex, race, or ethnic group.

--"When an employer has reason to believe that itsselection procedures have the exclusionary effectdescribed, it should initiate affirmative steps toremedy the situation. Such steps, in design andexecution, may Le race, color, sex, or ethnic'conscious.'"

52

APPENDIX VI APPENDIX VI

DEPARTMENTS, AGENCIES, ORGANIZATIONS,

AND INDIVIDUALS CONTACTED

Civil Rights Commission, Headquarters, Washington, D.C.

Civil Service Commission, Headquarters, Washington, D.C.

Department of Justice, Headquarters, Washington, D.C.

Department of Labor, Headquarters, Washington, D.C.

Equal Employment Opportunity Commission, Headquarters,Washington, D.C.; Region VII, San Francisco

Dr. Richard S. Barrett, research psychologist; Director,Laboratory of Psychological Studies, Stevens Instituteof Technology

Dr. Jerome Doppelt, Director, Psychological MeasurementDivision, The Psychological Corporation

Lawyers' Committee for Civil Rights under Law

NAACP Legal Defense Fund

American Society for Personnel Administration

Ad Hoc Industry Group (composed of public and privateemployers and associations which represent employers)

Bureau of National Affairs

Robert Garnier, Personnel Director, Civil ServiceCommission, Milwaukee, Wisconsin

Girard Davidson, Personnel Director, Duke Power Company

National Civil Service League

Herbert Kaplan, Deputy Director of Personnel, Los Angeles,California

Robert Krause, Director of Personnel, Hartford, Connecticut

Arnold McDermott, Personnel Director, Denver, Colorado

Dr. Grace Wright, New York State Department of Personnel

53

APPENDIX VI APPENDIX VI

Selma Mushkin, Director, Public Services Laboratory,Georgetown University

William Danielson, Personnel Director, Sacramento,California

Dr. Steven Stanard, Private practitioner, ProjectDirector (formerly), Industrial Systems Laboratory,Science Research Associates

54

APPENDIX VII APPENDIX VII

UNITED STATES CIVIL SERVICE COMMISSIONWASHINGTON. D. C. 20415

August 30, 1977CHAIRMAN

Mr. H. L. KreigerDirector, Federal Personnel andCompinsation Division

United States General Accounting OfficeWashington, D.C. 20548

Dear Mr. Kreiger:

This is in response to your letter of July 22, 1977 requesting the

Ccmmsion's conments on GAO's draft report entitled "Federal Efforts

to Provide Equal Employment Opportunity Selection Guidelines."As the draft report indicates, the four agencies which have issued.selection guidelines (the Departments of Labor and Justice, the

Civil Service ComisLesion and the Equal Employment OpportunityCoemission) currently are engaged in an intensive effort to develop

a uniform set of guidelines. All of the agencies recognize the need

for the Federal Government to speak with one voice in this important

and highly complex subject. Numerous staff meetings have already

taken place and the principals will meet at the end of this month.

Your report makes clear and we recognize the difficulty of the subject

matter involved. However, we believe that the differences separating

the agencies may not be as deep seated as your report suggests. We

are optimistic that the present leadership of the four agencies will

be able to reconcile remaining differences, develop a uniform

position and a draft set of guidelines at an early date. We do not

believe that legislation is either necessary or appropriate to resolve

this problem.

We also think that detailed comments on your report by each of thefour agencies at this time would have the effect of exacerbating

differences thus making it more difficult to achieve a uniform

position on this issue. For that reason we are refraining frommaking detailed substantive comeaants at this time.

55

APPENDIX VIII APPENDIX VIII

EQUAL EMPLOYMENT OPPORTUNITY COMMISSIONWASHINGTON, D. C. 20506

August 31, 1977

OFFICE OF THE CHAIR

Mr. Gregory J. AhartDirectorU.S. General Accounting OfficeHuman Resources DivisionWashington, D.C. 20548

Dear Mr. Ahart:

Your letter of July 22 to Commissioner Eleanor Holmes Nortonconcerning a draft proposed report entitled 'Federal Effortsto Provide Equal Bmployment Opportunity Selection Guidelines'has been referred to me for response.

As the letter to you from Assistant Attorney General Rooneyindicates, the Equal Rmployment Opportunity Commission hasbeen participating in a renewed effort to develop unifornguidelines. We share the views expressed by the AssistantAttorney General with respect to the likelihood of arrivingat a uniform draft set of guidelines by November 1, andaccordingly, believe that legislation is not necessary at thistime on this matter, and that detailed oon ents by each agenoyon the draft report would not be useful at this tims.

Consultant to the Chair

56

APieiWDIX IX APPENDIX IX

U.S. DEPARTMENT OF LABOROmnC or TD Ar&-]rfa SLUCArrY

*.;8AINGTON

Mr. Gregory J. AhartDirectorHuman Resources DlvistsnU.S. General Accounting OfficeWashington, D.C. 20548

Dear Mr. Ahart:

The draft of a proposed report by the General Accounting Office on"Federal Efforts to Provido Equal EmployAment Opportunity SelectionGuidelines" has been reviewed. Generally, we fell that the reportprovides a thorough and comprehensive review of the role that testsand other selection procedures play in personnel oecisions and theefforts of government agencies to eliminate the disc, 4minatory effectsof these selection procedures within the context of Federal "-w. Thereport will, therefore, provide an excellent source of inft , on forCongress in their efforts to bring about a uniform Federal pa1ition onthe non-discriminatory use of selection procedures.

Two mnaor concerns we have with the report are in the discussion of theproblems connected with the existence of separate sets of guidelines andthe proposed solutions to this situation. Compared to the detailed dis-cussion of the other issues, these subjects appear to be dealt withrather briefly. For example, OFCCP's enforcement of the guidelines isbased on a contractual arrangement between the Government and the employer,rather than on the Civil Rights Act. Should an employer's voluntarily-as-sueod obligations under contract law necessarily be the saem as the manda-tory obligations under Federal law? More importantly, what are thepractical implications of having more than one set of guidelines? Is itlikely that a given employer's selection procedures will meet one set ofguidelines and not the other? These issues, as well as others on thesame topic, should be discussed in depth.

(See GAO note 1, p. 58.)

57

APPENDIX IX APPENDIX rX

(See GAO note 2 below.)

In summry, I wish to reiterate that this report is, in general, excellent.I hope our comments will be of use to you.

Sincerely,

stant Setry forinistration aid Management

GAO notes: 1. The deleted comments relate to matters whichwere discussed in the draft report but omittedin this final report.

2. The deleted material suggested minor changesto the report. We have considered thesechanges in this final report.

APPENDIX X APPENDIX X

UNITED STATES DEPARTMENT OF JUSTICE

V*4ITON, D.C. MUM

-,.......... aOCT 26 W197"A 3dm f Idabb sad Nlbi-

Victor L. LoweDirectorGeneral Government DivisionUnited States General Accounting OfficeWashington, D.C. 20548

Dear Mr. Love:

This letter is in response to your request for comsentson the draft report entitled "Federal Efforts to ProvideEqual ployment Opportunity Selection Guidelines.'

Since early sume r, the four agencies which have issuedequal *eploysmet opportunity (N3o) selection guidelines--the Departments of Labor and Justice, the Equal BploymentOpportunity Commission and the Civil Service Commission--have engaged in an intensive effort to develop a uniformset of guidelines. All of the agencies recognise the deedfor the Federal government to speak with one voice on thisimportant and highly complex subject. Numerous meetingshave already taken place and further meetings of the prin-cipals are contemplated. While we recognize the difficultyof reaching agreement on this matter, we believe that thedifferences separating the agencie. are not as deep-seatedaa the report suggests, and that the present leadershipof the four agencies can reconcile them, develop a uniformposition, and draft a set of guidelines before November 1.

Although sincere efforts are being made to reach agree-ment on a uniform set of guidelines, we believe it is essentialthat we reapond frankly and candidly on several issues ofmajor concern to us. We have organized our comments aroundfour msxjor issues which moLe directly affect personnelinterests. In particular, we would like to point out thatwe believe GAO should have investigated and repoted onthese areas with a greater degree of detail and thoroughnecs.There areas relate to:

59

APPENDIX X APPENDIX X

- Valid EEO Guidelines Consistent withLegislative Intent;

- Public Sector Considerations An Imple-menting EEO Guidelines;

- Barriers to Compliance with REO Guidelines!

- The Scope of EEO Guidelines.

Valid EEO Guidelines Consistent with Legislative Intent

We generally agree that it is important to developand implement uniform guidelines concerning the proper andprofessional use of tests. However, we feel it equallyimportant that uniform guidelines in and of themselveshave the same validity as is required of selection procedures,and, moreover, reflect the intent of Congress as expressedin Civil Rights Legislation. Perhaps the clearest expres-sion of this intent is found in section 703(h) of Title VII,the Tower Amendment, which reads as follows.

'Notwithstanding any other provision of thissubchapter it shall not be an unlawful employ-ment practice for an emp2over to give and to actupon the results of any professionally developedability test provided that such test, its adminis-tration or action upon the results is not designed,intended or used to discriminate because of race,color, religion, sex, or national origin."

The GAO report, while suggesting remedies which shouldcertainly promote uniformity, does not adequately deal withthe ramifications cf the issue of achieving professionallyacceptable standards for test validation that are consistentwith the intent of the Civil Rights Legislation passed byCongress. It is especially important that these issuesbe dealt with, because as the GAO report states, 'The reasonsfor disagreements among the EEO Coordinating Council memberson uniform guidelines appear to go beyond differences ofopinion on technical questions to more basic issues of howagencies perceive and pursue their individual operatingresponsibilities." These differences among agencies regarding

60

APPENDIX X APPENDIX X

the manner in which they perceive and pursue their individualoperating responsibilities suggest an inconsistent interpre-tation and application of the Civil Rights Legislationpassed by Congress.

We believe Congress, in passing Civil Rights Legislation,contemplated a uniform posture concerning its enforcement.Accordingly, fundamental differences in philosophy andapproach resulting from agency interpretation of such legis-lation should not be permitted to continue. Steps shouldbe taken to establish such a uniform posture even priorto further attempts to develop meaningful common guidelinesfor EEO enforcement.

Public Sector Considerations In Implementing E30 Guidelines

The report does not adequately reflect certain uniqueconsiderations inherent in test usage by public sectorpersonnel systems. The broad based contacts with publicsector personnel officials made by GAO in the preparationof this report certainly should have surfaced these con-siderations.

Public sector personnel systems operating under civilservice merit requirements mandate the development of validselection procedures. In such systems, merit is perceivedon an individual basis, and tests are used not only todetermine whether individuals can perform a job, but alsoto rank them in terms of their ability to perform it. Theproblem therefore becomes one of designing a selection proce-dure that is equally valid for minority and non-minoritygroup members, and that makes meaningful distinction amongindividuals. If this objective is not achieved, and itrarely is, even after all professionally acceptable techniqueshave been applied, adjustment of procedures to result inmore minority selections is not presently a permissibleremedy to achieve compliance with SEO guidelines. Therefore,in the public sector, guidelines become an ideal wnich isconstantly sought but seldom achieved, while in the privatesector they are a lever through which to enforce EEO objec-tives, i.e., more minority hires.

61

APPENDIX X APPENDIX X

Barriers to Compliance with EDO Guidelines

Assuming a good faith effort on the part of employersto comply with EEO guidelines, important obstacles stillremain to be overcome if compliance is to be achieved.The more serious of these obstacles includes availableprofessional resources, the state of the art as it relatesto the validation of tests, and the cost of validatingtests. While GAO has to an extent attempted to deal withsome of the above issues, the report presentation lacksthe conclusive information necessary to perform a knowledge-able analysis of what might be required to achieve compliance.

(See GAO note 1, p. 58.)

From our own rather limited experience, it is our opinionthat the GAO sample considerably understates the cost ofvalidation. In 1972, an organization within the Departmentinitiated the development of a test for a single occupationat the entry level only. To date, the overall expenditureshave been roughly estimated at $250,000. Conceivably another2 to 3 years may be required to conclude the validationstudy for the single major occupation, bringing the finalcost to perhaps $400,000 or $500,000. Extrapolating fromthese very rough estimates, a cost estimate of 22 to 23millions of dollars would be conceivable considering thefact that the Department has 324 occupations. Even thoughthe actual cost might be less or more than indicated, theexpenditures will easily run into the millions of dollars.We hesitate to attempt any definitive cost estimates in'view of the limited single experience mentioned above,coupled with the fact that time has not permitted any precisein-depth study of this one effort which is still incomplete.

62

APPENDIX X APPENDIX X

At any rate, prior to making further efforts to adoptuniform guidelines, it would seem advisable to secure defini-tive estimates of the costs involved in implementing theprogram being proposed in view of the budgetary implications.It is also essential that a cost-benefit equation be developedto assure that the investment in manpower and money willachieve the desired goals. It would have been such morehelpful if GAO had addressed this issue in greater detail.

The report does not address the issue of whether theprofessional resources available to undertake the task ofvalidation are adequate. We suggest that examination ofsuch resources is crucial to a knowledgeable determination.American Psychological Association estimates indicate thereare approximately 2,200 qualified practitioners of IndustrialPsychology in the United States, 20 percent of whom workfor the government. Given the relatively small number ofpractitioners, and what is sure to be a substantiallyincreased demand for their services should a uniform require-ment to validate tests be developed and enforced, it islikely that resources will not be adequate to the task.

The final barrier that we feel should be addressedby GAO is that of the "state of the art" regarding testing.We appreciate the problems entailed in discussing in a reportof this nature the technical aspects of test validation.However, even after the excellent discussion presented bythe GAO concerning the technical aspects of test validation,we are not altogether clear as to why there is such substantialdisagreement on the subject, not only among the concernedagencies, but among other professionals in the field.

Our one suggestion with respect to remedying thisdeficiency is that GAO obtain further information in thisrespect, exploring in greater depth the differences of pro-fessional opinion that have caused the present impasse.

(See GAO note 1, p. 58.)

63

APPENDIX X APPENDIX X

Our own views in this 'regard are, that while manifoldadvantages could ultimately accrue to the Federal servicein terms of more valid and meritoritorious selection procedures,the fact remains that at best psychometrics is an imprecise&rt. Therefore, since there is considerable room for theexercise of professional judgment, competent practitionersall operating within the parameters of the standards promul-gated by the American Psycholovical Association can arguethe relative validity of any particular system of guidelines.Accordingly, to prescribe in minute detail what must bedone, without allowing latitude for professional judgmentsthat must necessarily be made, does not serve the objectiveof obtaining uniform guidelines, but rather provides a basisfor litigation concerning whatever standard is developed.Moreover, such an absolute approach is inconsistent withthe imperfections inherent in the existing "state of theart. m The GAO report should certainly address this point.

The Scope of REO Guidelines

(See GA( note 2, p. 58.)

We would also hope, should GAO determine additionaldiscussion as to the scope of the guidelines is warranted,that attention will be devoted not only to the types ofpersonnel determinations subject to the guidelines, butto the management level at which the determinations arecustomarily made, and the extent to which capability existsat that level to conduct validation studies. We wouldfurther point out the misconception that the U.S. CivilService Commission's testing program bears the brunt ofvalidation requirements in the Federal sector. This is

64

Ar7ENDIX X APPENDIX X

certainly not the case, as Federal agencies make numerousday-to-day personnel decisions which affect the approxi-mately 2½ million employees in the Federal work force.These decisions, which are directly carried out by agencies,are certainly subject to the validation requirements ofthe Equal Employment Opportunity Guidelines. We estimatethey may well account for 80 to 90 percent of the employ-ment decisions that will be made in accordance with thoseguidelines.

In summary, we believe that the first and foremostquestion has not been adequately addressed! namely, thatof legislative intent. Nor do we believe that the internalvalidity of the guidelines, both with respect to legislativeintent and to the content of the guidelines themselves,has been adequately examined. We are also concerned becausethere appears to have been inadequate recognition of thecrucial difference between guideline application to thepublic versus private sectors. Unlike the private sector,where avoidance of adverse impact is sufficient and wherethe emphasis is in effect on the group, in the Federalsector & merit process is mandatory, effects each individualand every aspect of every individual selection, and mereavoidance of adverse impact will not suffice. Guidelinesof whatever sort, serve to define and delineate the precisenature of the mandated merit selection process.

We have mentioned the all-too-obvious deficienciesof the "state of the art," the failure to obtain adequatecost estimates and the probable enormity of such costs,as well as the question of resource availability--resourceswhich at the present are almost entirely lacking in Federalagencies. Given the potentially enormous budgetary implica-tions, we are especially concerned by the fact the Officeof Management and Budget has never been intimately involvedin this matter. Lastly, we have sought to warn againstthe pervasive coverage of the guidelines which is not onlyin excess of that generally recognized, but is inconsistentwith and in conflict with other existing regulations. Sched-ules A, B, and C, for example, are covered. Yet ScheduleC was established to provide each new administration witha measure of flexibility outside the usual merit processin filling positions of a confidential or policy nature.Schedule A is based on the impracticability to competitivelyor noncompetitively examine, Schedule B on the impracticability

65

APPENDIX X APPENDIX X

to competitively examine. Related thereto, are the largelyunrecognized impact of the guidelines on the agencies them-selves which, in fact, will bear the major burden, and thealmost limitless litigative possibilities which will beopened up by such guidelines.

We appreciate the opportunity given us to comaent onthe draft report. Should you have any further qnestions,please feel free to contact us.

Sincerely,

ein D RooneyAssistant Attorney Genera

for Administration

(964089)

66