Page images

enforcement agencies will expect at a appropriate by race and national minimum additional empirical re- origin; and search evidence meeting the standards (c) The selection procedures utilized of subparagraphs section 14B (2) and (either standardized or not standard(3) above for the additional jobs or ized). groups of jobs.

These records should be maintained (b) Determination of common work for each race or national origin group behaviors. In determining whether (see section 4 above) constituting more two or more jobs have one or more than two percent (2%) of the labor work behavior(s) in common, the user force in the relevant labor area. Howshould compare the observed work ever, it is not necessary to maintain behavior(s) in each of the jobs and records by race and/or national origin should compare the observed work (see § 4 above) if one race or national product(s) in each of the jobs. If nei- origin group in the relevant labor area ther the observed work behavior(s) in constitutes more than ninety-eight each of the jobs nor the observed work percent (98%) of the labor force in the product(s) in each of the jobs are the area. If the user has reason to believe same, the Federal enforcement agen- that a selection procedure has an adcies will presume that the work verse impact, the user should maintain behavior(s) in each job are different. any available evidence of validity for If the work behaviors are not observ- that procedure (see sections 7A and 8). able, then evidence of similarity of (2) Information on impact.-(a) Col. work products and any other relevant lection of information on impact. research evidence will be considered in Users of selection procedures other determining whether the work than those complying with section behavior(s) in the ćwo jobs are the 15A(1) above should maintain and same.

have available for each job records or

other information showing whether DOCUMENTATION OF IMPACT AND the total selection process for that job VALIDITY EVIDENCE

has an adverse impact on any of the

groups for which records are called for $ 60-3.15 Documentation of impact and

by sections 4B above. Adverse impact validity evidence.

determinations should be made at A. Required information. Users of least annually for each such group selection procedures other than those which constitutes at least 2 percent of users complying with section 15A(1) the labor force in the relevant labor below should maintain and have avail. area or 2 percent of the applicable able for each job information on ad workforce. Where a total selection verse impact of the selection process process for a job has an adverse for that job and, where it is deter impact, the user should maintain and mined a selection process has an ad have available records or other inforverse impact, evidence of validity as mation showing which components set forth below.

have an adverse impact. Where the (1) Simplified recordkeeping for total selection process for a job does users with less than 100 employees. In not have an adverse impact, informaorder to minimize recordkeeping bur- tion need not be maintained for indidens on employers who employ one vidual components except in circumhundred (100) or fewer employees, and stances set forth in subsection other users not required to file EEO-1, 15A(2)(b) below. If the determination et seq., reports, such users may satisfy of adverse impact is made using a prothe requirements of this section 15 if cedure other than the "four-fifths they maintain and have available rec rule,” as defined in the first sentence ords showing, for each year:

of section 4D above, a justification, (a) The number of persons hired, consistent with section 4D above, for promoted, and terminated for each the procedure used to determine adjob, by sex, and where appropriate by verse impact should be available. race and national origin;

(b) When adverse impact has been (b) The number of applicants for eliminated in the total selection prochire and promotion by sex and where ess. Whenever the total selection process for a particular job has had an ad complete and organized manner to verse impact, as defined in section 4 permit direct evaluation of the validity above, in any year, but no longer has of the selection procedure. Previously an adverse impact, the user should written employer or consultant remaintain and have available the infor ports of validity, or reports describing mation on individual components of validity studies completed before the the selection process required in the issuance of these guidelines are acpreceding paragraph for the period in ceptable if they are complete in regard which there was adverse impact. In ad- to the documentation requirements dition, the user should continue to col- contained in this section, or if they lect such information for at least two satisfied requirements of guidelines (2) years after the adverse impact has which were in effect when the validity been eliminated.

study was completed. If they are not (c) When data insufficient to deter complete, the required additional docmine impact. Where there has been umentation should be appended. If an insufficient number of selections to necessary information is not available determine whether there is an adverse the report of the validity study may impact of the total selection process still be used as documentation, but its for a particular job, the user should adequacy will be evaluated in terms of continue to collect, maintain and have compliance with the requirements of available the information on individu. these guidelines. al components of the selection process (c) Completeness. In the event that required in section 15(A)(2)(a) above evidence of validity is reviewed by an until the information is sufficient to enforcement agency, the validation redetermine that the overall selection ports completed after the effective process does not have an adverse date of these guidelines are expected impact as defined in section 4 above, to contain the information set forth or until the job has changed substan- below. Evidence denoted by use of the tially.

word “(Essential)” is considered criti(3) Documentation of validity evi. cal. If information denoted essential is dence.-(a) Types of evidence. Where a not included, the report will be considtotal selection process has an adverse ered incomplete unless the user affirimpact (see section 4 above) the user matively demonstrates either its unashould maintain and have available vailability due to circumstances for each component of that process beyond the user's control or special which has an adverse impact, one or circumstances of the user's study more of the following types of docu- which make the information irrelementation evidence:

vant. Evidence not so denoted is desir(i) Documentation evidence showing able but its absence will not be a basis criterion-related validity of the selec- for considering a report incomplete. tion procedure (see section 15B, The user should maintain and have below).

available the information called for (ii) Documentation evidence showing under the heading “Source Data” in content validity of the selection proce- sections 15B(11) and 15D(11). While it dure (see section 15C, below).

is a necessary part of the study, it (iii) Documentation evidence show- need not be submitted with the report. ing construct validity of the selection All statistical results should be orgaprocedure (see section 15D, below). nized and presented in tabular or

(iv) Documentation evidence from graphic form to the extent feasible. other studies showing validity of the B. Criterion-related validity studies. selection procedure in the user's facili- Reports of criterion-related validity ty (see section 15E, below).

for a selection procedure should in(v) Documentation evidence showing clude the following information: why a validity study cannot or need (1) User(s), location(s), and date(s) not be performed and why continued of study. Dates and location(s) of the use of the procedure is consistent with job analysis or review of job informaFederal law.

tion, the date(s) and location(s) of the (b) Form of report. This evidence administration of the selection proceshould be compiled in a reasonably dures and collection of criterion data, and the time between collection of quantified, should be provided (essendata on selection procedures and crite- tial). If rating techniques are used as rion measures should be provided (Es criterion measures, the appraisal sential). If the study was conducted at form(s) and instructions to the several locations, the address of each rater(s) should be included as part of location, including city and State, the validation evidence, or should be should be shown.

explicitly described and available (es(2) Problem and setting. An explicit sential). All steps taken to insure that definition of the purpose(s) of the criterion measures are free from facstudy and the circumstances in which tors which would unfairly alter the the study was conducted should be scores of members of any group provided. A description of existing se- should be described (essential). lection procedures and cutoff scores, if (6) Sample description. A description any, should be provided.

of how the research sample was identi(3) Job anlysis or review of job infor- fied and selected should be included mation. A description of the proce (essential). The race, sex, and ethnic dure used to analyze the job or group composition of the sample, including of jobs, or to review the job informa- those groups set forth in section 4A tion should be provided (Essential). above, should be described (essential). Where a review of job information re This description should include the sults in criteria which may be used size of each subgroup (essential). A dewithout a full job analysis (see section scription of how the research sample 14B(3)), the basis for the selection of compares with the relevant labor these criteria should be reported (Es market or work force, the method by sential). Where a job analysis is re which the relevant labor market or quired a complete description of the work force was defined, and a discuswork behavior(s) or work outcome(s), sion of the likely effects on validity of and measures of their criticality or im- differences between the sample and portance should be provided (Essen- the relevant labor market or work tial). The report should describe the force, are also desirable. Descriptions basis on which the behavior(s) or of educational levels, length of service, outcome(s) were determined to be and age are also desirable. critical or important, such as the pro- (7) Description of selection proce. portion of time spent on the respective dures. Any measure, combination of behaviors, their level of difficulty, measures, or procedure studied should their frequency of performance, the be completely and explicitly described consequences of error, or other appro- or attached (essential). If commercialpriate factors (Essential). Where two ly available selection procedures are or more jobs are grouped for a validity studied, they should be described by study, the information called for in title, form, and publisher (essential). this subsection should be provided for Reports of reliability estimates and each of the jobs, and the justification how they were established are desirfor the grouping (see section 14B(1)) able. should be provided (Essential).

(8) Techniques and results. Methods (4) Job titles and codes. It is desir- used in analyzing data should be deable to provide the user's job title(s) scribed (essential). Measures of central for the job(s) in question and the cor tendency (e.g., means) and measures responding job title(s) and code(s) of dispersion (e.g., standard deviations from U.S. Employment Service's Dic and ranges) for all selection procetionary of Occupational Titles.

dures and all criteria should be report(5) Criterion measures. The bases ed for each race, sex, and ethnic group for the selection of the criterion meas- which constitutes a significant factor ures should be provided, together with in the relevant labor market (essenreferences to the evidence considered tial). The magnitude and direction of in making the selection of criterion all relationships between selection measures (essential). A full description procedures and criterion measures inof all criteria on which data were col. vestigated should be reported for each lected and means by which they were relevant race, sex, and ethnic group observed, recorded, evaluated, and and for the total group (essential). Where groups are too small to obtain ing the method for operational use, reliable evidence of the magnitude of and the evidence of the validity and the relationship, need not be reported utility of the procedure as it is to be separately. Statements regarding the used (essential). The purpose for statistical significance of results which the procedure is to be used (e.g., should be made (essential). Any statis- hiring, transfer, promotion) should be tical adjustments, such as for less then described (essential). If weights are asperfect reliability or for restriction of signed to different parts of the selecscore range in the selection procedure tion procedure, these weights and the or criterion should be described and validity of the weighted composite explained; and uncorrected correlation should be reported (essential). If the coefficients should also be shown (es- selection procedure is used with a sential). Where the statistical tech- cutoff score, the user should describe nique categorizes continuous data, the way in which normal expectations such as biserial correlation and the of proficiency within the work force phi coefficient, the categories and the were determined and the way in which bases on which they were determined the cutoff score was determined (esshould be described and explained (es- sential). sential). Studies of test fairness should (11) Source data. Each user should be included where called for by the re- maintain records showing all pertinent quirements of section 14B(8) (essen information about individual sample tial). These studies should include the members and raters where they are rationale by which a selection proce used, in studies involving the validadure was determined to be fair to the tion of selection procedures. These group(s) in question. Where test fair- records should be made available upon ness or unfairness has been demon request of a compliance agency. In the strated on the basis of other studies, a case of individual sample members bibliography of the relevant studies these data should include scores on should be included (essential). If the the selection procedure(s), scores on bibliography includes unpublished criterion measures, age, sex, race, or studies, copies of these studies, or ade- ethnic group status, and experience on quate abstracts or summaries, should the specific job on which the validabe attached (essential). Where revi- tion study was conducted, and may sions have been made in a selection also include such things as education, procedure to assure compatability be training, and prior job experience, but tween successful job performance and should not include names and social the probability of being selected, the security numbers. Records should be studies underlying such revisions maintained which show the ratings should be included (essential). All sta- given to each sample member by each tistical results should be organized and rater. presented by relevant race, sex, and (12) Contact person. The name, mailethnic group (essential).

ing address, and telephone number of (9) Alternative procedures investi the person who may be contacted for gated. The selection procedures inves- further information about the validity tigated and available evidence of their study should be provided (essential). impact should be identified (essential). (13) Accuracy and completeness. The The scope, method, and findings of report should describe the steps taken the investigation, and the conclusions to assure the accuracy and completereached in light of the findings, ness of the collection, analysis, and should be fully described (essential). report of data and results.

(10) Uses and applications. The C. Content validity studies. Reports methods considered for use of the se- of content validity for a selection prolection procedure (e.g., as a screening cedure should include the following indevice with a cutoff score, for group formation: ing or ranking, or combined with other (1) User(s), location(s) and date(s) of procedures in a battery) and available study. Dates and location(s) of the job evidence of their impact should be de- analysis should be shown (essential). scribed (essential). This description (2) Problem and setting. An explicit should include the rationale for choos- definition of the purpose(s) of the study and the circumstances in which (5) Relationship between the selecthe study was conducted should be tion procedure and the job. The evi. provided. A description of existing se dence demonstrating that the selec. lection procedures and cutoff scores, if tion procedure is a representative any, should be provided.

work sample, a representative sample (3) Job analysis-Content of the job. of the work behavior(s), or a repreA description of the method used to

sentative sample of a knowledge, skill, analyze the job should be provided (es

or ability as used as a part of a work sential). The work behavior(s), the as

behavior and necessary for that besociated tasks, and, if the behavior re

havior should be provided (essential). sults in a work product, the work prod.

The user should identify the work ucts should be completely described

behavior(s) which each item or part of (essential). Measures of criticality

the selection procedure is intended to and/or importance of the work

sample or measure (essential). Where behavior(s) and the method of deter

the selection procedure purports to mining these measures should be pro

sample a work behavior or to provide a

sample of a work product, a compari. vided (essential). Where the job analy. sis also identified the knowledges,

son should be provided of the manner, skills, and abilities used in

setting, and the level of complexity of work

the selection procedure with those of behavior(s), an operational definition

the work situation (essential). If any for each knowledge in terms of a body

steps were taken to reduce adverse of learned information and for each

impact on a race, sex, or ethnic group skill and ability in terms of observable

in the content of the procedure or in behaviors and outcomes, and the rela

its administration, these steps should tionship between each knowledge,

be described. Establishment of time skill, or ability and each work behav

limits, if any, and how these limits are ior, as well as the method used to de

related to the speed with which duties termine this relationship, should be

must be performed on the job, should provided (essential). The work situa

be explained. Measures of central tion should be described, including the

tend- ency (e.g., means) and measures setting in which work behavior(s) are

of dispersion (e.g., standard devi. performed, and where appropriate, the

ations) and estimates of realibility manner in which knowledges, skills, or

should be reported for all selection abilities are used, and the complexity

procedures if available. Such reports and difficulty of the knowledge, skill,

should be made for relevant race, sex, or ability as used in the work

and ethnic subgroups, at least on a behavior(s).

statistically reliable sample basis. (4) Selection procedure and its con

(6) Alternative procedures investitent. Selection procedures, including

gated. The alternative selection procethose constructed by or for the user, dures investigated and available specific training requirements, com dence of their impact should be identi. posites of selection procedures, and fied (essential). The scope, method, any other procedure supported by con- and findings of the investigation, and tent validity, should be completely and the conclusions reached in light of the explicitly described or attached (essen findings, should be fully described (estial). If commercially available selec- sential). tion procedures are used, they should (7) Uses and applications. The be described by title, form, and pub- methods considered for use of the selisher (essential). The behaviors meas lection procedure (e.g., as a screening ured or sampled by the selection pro device with a cutoff score, for groupcedure should be explicitly described ing or ranking, or combined with other (essential). Where the selection proce procedures in a battery) and available dure purports to measure a knowledge, evidence of their impact should be deskill, or ability, evidence that the se- scribed (essential). This description lection procedure measures and is a should include the rationale for choosrepresentative sample of the knowl ing the method for operational use, edge, skill, or ability should be pro- and the evidence of the validity and vided (essential).

utility of the procedure as it is to be

« PreviousContinue »