Are Biodata Tests a New Frontier for Employment Litigation?

By Robert Otinger and Carrie Kurzon

06-27-2007

Corporate America is changing its hiring practices, and Google is at the forefront.In the rush to hire talent, innovative companies like Google are abandoningtraditional employment practices. With research showing that grades andinterviews are not reliable predictors of success, the use of "biodata" is becoming moreprevalent. This new trend has serious implications for human resource departmentsand employment law attorneys as there will inevitably be a surge of discriminationclaims arising from the use of biodata as these innovative practices potentially test thelimits of established law.

Biodata is the term used in industrial andorganizational psychology for biographicaldata. Biodata probes deep into an individual'sbackground and consists of a person'slife and work experiences, as well as beliefs,values, opinions and attitudes. This informationis then presumed to be related topersonality structure, personal adjustmentor success in social, educational or occupationalpursuits. Employers are now usingbiodata as a means of obtaining a more scientificapproach to the interview process.

GOOGLE'S USE OF BIODATA

Google's prior method for selecting itsown employees consisted of the traditionalmodel: hiring candidates from top schoolswith outstanding grades. That model isnow considered outdated. Today, insteadof reviewing transcripts, Google uses an algorithm that selects employees who arebetter fits for the company, and it accomplishesthis selection faster.[FOOTNOTE 1] Google reportedlyhires over 100 people per week andwith that figure expected to double nextyear, Google has determined that biodatais the most efficient way to predict performance.

While no one can say with certaintywhether the use of biodata is better than thetraditional interview process, it does seemto work for Google, which purports to haveless than 4 percent employee turnover.Google's system, as the leading model ofthis new wave, deserves closer examination.Google created an automated system thatcollects biodata from the 100,000 job applicationsit receives each month. The systemasks job applicants to fill out an elaborateonline survey that examines their personality,behavior, attitudes and other personaldetails going back to high school. The surveycovers a massive range of topics includingwhether the applicants like to work aloneor in groups, if they like dogs, if they everpublished a book, started a club in schoolor set a world record.

Google's army of mathematicians createda set of formulas that computes the applicant'sresponses and scores each applicanton a scale of 0 to 100. The score is designedto predict how well the person will fit intoGoogle's corporate culture. Although Googleis an innovator, it is not the only companyjumping into the biodata arena.

USE IS ON THE RISE

Although biodata surveys have beenaround for decades, mainly in governmentagencies, during the 1960s the tests spreadto the private sector. However, they fell out of favor afterpassage of the Civil Rights Act, as companiesfeared they could have a disparate impact and therefore be discriminatory.[FOOTNOTE 2] A survey[FOOTNOTE 3] of 348 companies conducted in 1988 foundthat only 6 percent of companies had usedbiodata. A similar study of human resourcespecialists by the Bureau of National Affairsreported that only 4 percent used biodata.Notably, 40 percent of respondents said thatprivacy concerns prevented them from usingbiodata and some cited fear of litigation. Currently,the use of biographical surveys similarto Google's new system is on the rise.

In the past few years companies haverealized that increased competition makesefficient hiring of productive employeesimperative. In 2002, Proctor & Gamble contractedwith the Performance AssessmentNetwork Inc., to provide a biodata-basedhiring system for the company's prospectiveemployees. The biodata system wasintegrated throughout Proctor & Gamble'soffices in Asia, Europe, North America andLatin America. Other large companies havefollowed suit, including SER Solutions Inc.,a communications company in Dulles, Va.,which also utilizes the personal assessmenttests as a more effective means to findingthe right employees. There is no doubt thatcorporate America is using biodata-basedsurveys. The question for employmentlawyers is whether U.S. laws can balance acompany's need to quickly and accuratelyselect suitable employees and also preventdiscrimination.

POTENTIAL IMPLICATIONS

On one hand, the use of biodata mayreduce discrimination claims by removingthe inevitable subjectivity that comes witha traditional interview. On the other hand,certain algorithms that analyze a person'sbiodata can also adversely impact protectedcategories such as race, gender, age andnational origin. Courts have repeatedly heldthat a selection device should measure theperson for the job, not the person in theabstract. Griggs v. Duke Power Co., 401 U.S.424, 431, 91 S.Ct. 849, 28 L.Ed.2d 158 (1971).If a job applicant's past is examined, the useof biodata can arguably measure a personin the abstract, or subconsciously favora protected class, thereby subjecting thecompany to liability.

Title VII of the Civil Rights Act of 1964,42 U.S.C. §2000, prohibits "procedures ortesting mechanisms that operate as 'built-inheadwinds' for minority groups." Connecticutv. Teal, 457 U.S. 440, 102 S.Ct. 2525, 73L.Ed.2d 130 (1982). In analyzing hiring proceduresand mechanisms similar to biodatamodels, courts focus on the procedure itselfto determine if any discriminatory barriersexist within both the process and theresults of the process to determine if theprocess has an adverse impact on minoritygroups. Employee selection systems such asGoogle's biodata algorithm may find itselfunder close scrutiny.

Employment discrimination claims are traditionallybased on either disparate impactor disparate treatment. Legal challenges toemployment decisions based on biodatawould invoke the disparate impact theory.The disparate impact approach invalidatesemployment policies that appear on their faceto be unbiased, but impact one group morethan another and cannot be justifiedby business necessity. Griggs v. DukePower Co., supra, 401 U.S. at 431.

The individual componentsof a biodata-basedhiring process may constituteseparate andindependent employmentpractices. Theseparate componentswill be subject to TitleVII even if the overallresult of the processdoes not have an adverse impacton a protected group. Smith v. Xerox,196 F.3d 358, 370 (2d Cir. 199 ). Whena ranking mechanism dictates whatcandidates can be considered foremployment, the components ofthat mechanism will be scrutinizedto determine if it acts as a barrierto potential protected employees.Such a barrier can mean the systemhas a disparate impact on minorityhiring. Waisome v. Port Auth. ofN.Y. & N.J., 948 F.2d 1370, 1378 (2dCir. 199 1). As such, even if a biodata-based selection system doesnot ultimately favor one group overanother, the system would still besubject to legal dissection so thateach part of the process can betested for discrimination.

An employer's overallbiodata-based employeeselection system is subjectto invalidation ifit adversely impacts aprotected group and theselection criteria are notclosely linked to predictingjob performance. Inorder to challenge a biodata system,a plaintiff must identify the challengedemployment practice such asa biodata survey or testing program.Next, the plaintiff must demonstratethat this survey or testing programis the cause of an adverse impacton a protected group such as raceor sex as defined by Title VII. If theplaintiff can make this showing, theburden then shifts to the company.Bradley v. City of Lynn, 443 F.Supp.2d145, 156 (D. Mass. 2006).

Once the burden shifts to thecompany, the company has severaloptions. First, it can challenge theplaintiff's proof directly by arguingthat no survey or testing policyexists or that the policy does existbut it does not produce anadverse impact on the protectedcategory at issue. Or in the alternative, thecompany may admit thatits policy does have adisparate impact, butargue that it is job-relatedand consistentwith business necessity. Ifthe company fails in its effort torebut the claimant's argument, thenthe company may be found liable.

Although the risk of liability isvery real for a defendant using abiodata-based employee selectionprocess, thus far, courts have failedto address this issue substantively.For example, in United Statesof America v. City of Garland, 2004 WL741295 (N.D. Tex. 2004), the plaintiffalleged that the city's use of itswritten examination for entry-levelpolice officers and firefighters hada disparate impact against African-Americans and Hispanics, was notjob-related and, thus, violated TitleVII. The court, in granting judgmentto the defendant, merely held thatthe plaintiff failed to prove by apreponderance of the evidencethat the employmentpractices violatedTitle VII. Id.

Some courts haveimplied that the use ofbiodata would have helpedthe defendant's case in adisparate impact case. InBradley, supra, the plaintiffsin a class action in the District ofMassachusetts argued that thewritten exam used to fill firefightervacancies had a disparate impact onAfrican-Americans and Hispanics, inviolation of Title VII of the 1964 CivilRights Act.

In agreeing with the plaintiffs,Judge Patti B. Saris concluded thatthe "statistical evidence showedclear differences in scores as a functionof race for both the 2002 and2004 examinations." She determinedthat the exam was not properly validatedor job-related. Saris, however, stated that the defendantcould have used a biodata test incombination with the written cognitiveexamination to reduce the disparateimpact of the examination.Id. at *174.

The use of biodata has also beenalleged to be a violation of the EqualProtection Clause of the 14th Amendment.In Antonelli v. State of NewJersey, 419 F.3d 267, 96 FEP 491 (3dCir. 2005), the plaintiffs brought acivil rights action, alleging,inter alia, that themethod used by NewJersey to administerand score the biodatacomponent of a firefighters'exam violated theirrights under the EqualProtection Clause. Id.

The Court of Appeals, inaffirming the lower court'sdecision granting summary judgmentto the defendant, held that the biodatacomponent of an examinationfor entry-level firefighters had nodisparate impact on non-Hispanicwhites. Since the exam was found tobe facially neutral, the plaintiffs neededto show that the defendants actedwith discriminatory intent and theexam had a discriminatory impact.According to this finding, plaintiffsfailed to meet this burden. Id.

CAUSATION AND STATISTICS

There are particular issues thatwill arise when trying to prove disparateimpact in a case involvingbiodata selection. The first majorissue that is likely to arise in biodatacases will concern causation,namely, how can a plaintiff provethat a testing program is causingan adverse impact on a protectedgroup? The courts have explainedthat a plaintiff must demonstrate asubstantial statistical disparity thatraises an inference of causation. Thismeans that the disparity cannot beaccounted for by chance. EEOC v.Joint Apprenticeship Comm. of theJoint Indus. Bd. of the Elec. Indus.,186 F.3d 110, 117 (2d Cir. 199 ). Inevaluating statistical evidence, theSupreme Court has held that nosingle test controls how disparateimpact claims must be established.Watson v. Fort Worth Bank and Trust,487 U.S. 977, 99 4-95, 108 S.Ct. 2777,101 L.Ed.2d 827 (1988). The U.S.Equal Employment OpportunityCommission (EEOC), however, hasa formula for this.

The EEOC's Uniform Guidelineson Employee Selection Procedures(1978) requires the use of the "four-fifthsrule." The rule providesthat a "selection ratefor any race ... whichis less than four-fifths(or 80 percent)of the rate for thegroup with the highestrate will generallybe regarded by the[EEOC] as evidence ofadverse impact ... ." 29C.F.R. §1607.4(D). Forexample, if a company hires whiteapplicants 40 percent of the timeand African-American applicants 20percent of the time, the selectionrate for African-Americanswould be half that ofwhite employees. This 50percent selection ratiois less than 80 percentand therefore it wouldviolate the four-fifthsrule. This would establishan adverse impacton African-Americansunder the EEOC Guidelines.

Even if a plaintiff can prove anadverse impact, the company canavoid liability by proving that theselection criteria (biodata) is job-relatedand consistent with businessnecessity. The selection criteria arevalid if they are "predictive of or significantlycorrelated with importantelements of job performance." 29C.F.R. §1607.5(B). The strength ofthe relationship between the selectioncriteria and job requirements isdetermined by calculating a correlationcoefficient. Id. §1607.14(B)(6). Acorrelation coefficient of 1.0 meansthat the criteria are highly predictiveof job performance and valid.Williams v. Ford Motor Co., 187 F.3d533, 540 (6th Cir. 1999 ). A correlationcoefficient of 0.0 indicates thatthere is no connection between thecriteria and job performance. A coefficientof 0.3 is the minimum neededto establish a satisfactory relationshipbetween the selection criteriaand job performance. Bradley, 443F.Supp.2d at 161.

Once companies begin using biodatasystems to hire employees, legal challenges can be expected to follow.The law, as described above, providesemployment lawyers with thetools needed to thoroughly analyzethese systems and keep them fair.Companies, however, are likely toargue that litigation over biodata willhinder their ability to compete globally.

In fact, it may be extremely difficultfor companies to determine howtheir biodata systems will impactprotected groups. Even if the biodatasystems do adversely impactcertain protected groups, employerswill argue that it is not logicalto punish them when the impact isunforeseen, unintentional, and simplythe result of our culture.

When evaluating the use andpotential implications of biodata,the "management side" islikely to push for a lawthat only considers jobrelatedness and does notconsider adverse impact.Minority groups, onthe other hand, wouldoppose any attempt tominimize the adverseimpact theory for biodataor similar employeeselection mechanisms. The lawis clear, the argument goes, thatdiscrimination is illegal -- even if itis not intentional, and the burdenof adverse impact discriminationshould not fall on the victims.

CONCLUSION

There is no easy answer to theconflicts that are coming. Discriminationexists and our employmentlaws are designed to prevent it.These new challenges will force usto question these laws. Are therebetter ways to prevent employmentdiscrimination? Are the currentlaws equipped to deal with modernchallenges created by a globaleconomy? Only time will tell whetherthe benefits of using biodata will beoutweighed by the potential legalimplications.

Robert Ottinger and Carrie Kurzon areattorneys at The Ottinger Firm. Their e-mailaddresses are robert@ottingerlaw.com andcarrie@ottingerlaw.com, respectively. CharlesKane, also an attorney at the firm, contributedto this article.

::::FOOTNOTES::::

FN1 Saul Hansell, "Google Answer to FillingJobs Is an Algorithm," The New York Times,Jan. 3, 2007.

FN2 Amy Joyce, "Before Scoring That JobYou'd Better Ace That Test," The Washington Post,June 8, 2006, A1.

FN3 Hammer, E.G., & Kleiman, L.A. (1988)."Getting to know you," Personnel Administrator,34, 86-92.