Federal contractors can use computer algorithms that analyze job candidates’ digital footprints and other data to make hiring decisions. But will doing so shield them from allegations of unintentional discrimination during Labor Department audits?
“There lies an interesting equal employment opportunity and legal issue,” said Eric M. Dunleavy, an industrial and organizational psychologist and director of personnel selection and litigation support services for DCI Consulting Group Inc. in Washington.
An open question exists about whether so-called big data analytics can be validated as a defense against alleged disparate impact discrimination found by the DOL’s Office of Federal Contract Compliance Programs during audits, observers said. Other unsettled issues revolve around whether the tools can conform with a decade-old OFCCP rule about contractors’ record-keeping and analysis requirements for internet applicants.
Employers can use big data tools to “predict” whether a job candidate will succeed in a given position. The tools, which can be developed in-house or purchased from outside vendors, take many forms.
Some algorithms scan resumes for a job seeker’s knowledge, skills and abilities and compare them to current successful employees. Others digitally record and analyze applicant interviews. They can even require individuals to problem solve their way through a video game-like interface.
They also can analyze social media or other available online information of individuals who aren’t actively seeking employment.
Big data proponents argue that algorithms can reduce the risk of intentional discrimination in hiring decisions. But will use of the tools pass muster under current federal contractor laws intended to protect against unintentional bias?
“No one really knows at the moment,” Adam T. Klein, a class action plaintiffs’ attorney with Outten & Golden in New York, told Bloomberg BNA. “There are concerns about whether these kinds of systems are defensible.”
Although big data was discussed at an October Equal Employment Opportunity Commission meeting, the issue has largely been “flying under the radar,” Klein said.
“I don’t think there’s enough discussion about it from the employment community,” he said.
OFCCP Is Asking About Big Data, Attorney Says
The OFCCP audits roughly 2 percent of about 200,000 federal contractor locations annually.
The agency has begun to ask federal contractors during compliance audits if they’re using big data tools and, if so, how they developed them, Heather A. Morgan, a management attorney with Paul Hastings in Los Angeles and global chair of the firm’s workplace data and technology practice, told Bloomberg BNA.
The OFCCP has yet to publicly discuss big data, let alone offer any official guidance to federal contractors about its use, although the agency’s previous director, Patricia A. Shiu, attended the EEOC’s October big data meeting prior to her November departure from the agency.
“To our knowledge, the current administration at OFCCP hasn’t done anything to address this,” said David Cohen, president of DCI Consulting and a co-chair of the OFCCP Institute, a national nonprofit employer association in Washington.
Whether the incoming administration of President-elect Donald Trump will tackle big data in the OFCCP context remains to be seen.
Big data might not be “front and center” on the agency’s agenda, Morgan said. The OFCCP already has a number of enforcement priorities on its plate, among them pay equity issues.
But contractors may begin to see a curious agency ask about big data more regularly, especially during audits focused on alleged discrimination in recruitment and selection, she said.
“I think eventually we’ll see that as part of the norm,” Morgan said. “And where that leads is a little uncertain. The current legal standards, namely the Uniform Guidelines on Employee Selection Procedures and the Internet Applicant Rule, do not provide contractors a clear or manageable legal framework within which to operate.”
But big data discourse will continue and, according to Morgan, members of the contractor community hope to be able to join the OFCCP in active dialogue “to develop thoughtful and practical solutions given the challenging compliance implications that arise from employers’ use of big data predictive analytics.”
Morgan is a member of Bloomberg BNA’s Labor and Employment Technology and Innovation Board. The board’s goal is to provide feedback to enable Bloomberg Law to improve products and workflow tools for labor and employment lawyers.
The Labor Department declined to comment for the story.
Is Big Data a ‘Selection Procedure’?
During its audits, the OFCCP analyzes contractors’ hiring data based on a number of protected characteristics, such as race or sex.
If a contractor uses a facially neutral test or other selection procedure to make hiring decisions—that is, a test or procedure that doesn’t explicitly discriminate against a protected group—the OFCCP will determine if that procedure nevertheless has a statistically significant disparate impact on a protected group.
If the OFCCP finds such statistical indicators, the contractor then has an opportunity to explain that its hiring decisions weren’t based on discriminatory reasons.
Imagine a contractor tells the OFCCP that it used big data algorithms to pick candidates for hire. One question sure to pop up is whether the algorithms would be considered selection procedures that fall within the scope of the 1978 Uniform Guidelines on Employee Selection Procedures (41 C.F.R. part 60-3), which is enforced by both the OFCCP and the EEOC.
“We are all waiting for legal scholars to weigh in on some of the nuances related to this issue,” Dunleavy said.
An intuitive argument can be made that a computer algorithm is a facially neutral tool. In that sense, it’s just like a test, interview or other selection procedure covered by the Uniform Guidelines, Dunleavy said.
Proving Business Need May Be Challenging
If a contractor uses big data tools, and the OFCCP finds that the algorithms produced disparate impact, the Uniform Guidelines require the contractor to show that the algorithms are “valid,” meaning that they’re job-related and consistent with business necessity.
Proving validity may pose challenges for contractors, observers said.
Contractors would have to demonstrate via research that whatever big data tools they use are predictive of work outcomes, such as job performance, absenteeism or turnover, Dunleavy said.
One way to do this would be through a criterion validation study, which examines whether a statistical correlation exists between a specific selection procedure and a particular work outcome, he said.
But that may prove difficult if employers don’t know what’s in the big data algorithms and their vendors classify that information as proprietary, he said.
Algorithms also are “constantly changing,” Morgan said, and “that in and of itself creates additional challenges.”
Additionally, correlation doesn’t necessarily imply causation.
Klein said it’s “hard to imagine” how a big data tool can be shown to be job-related if an employer doesn’t start with a traditional job analysis of a position’s tasks and the specific knowledge, skills and abilities required to perform those tasks—things that algorithms examining an individual’s social media profile or online presence might not necessarily consider.
Open Issues With Internet Applicant Rule
Questions also exist about how the use of big data may affect the application of the OFCCP’s internet applicant rule (41 C.F.R. § 60-1.12).
The rule went into effect in 2006 to address contractor record-keeping and disparate impact analysis requirements for individuals who use the internet to apply for jobs.
Under the rule, a job seeker must satisfy four criteria to be deemed an internet applicant.
- First, the individual must submit an expression of interest in employment through the internet or related technologies.
- Second, the contractor must consider the individual for a particular position.
- Next, the individual’s expression of interest must indicate that he or she has the basic qualifications for the position.
- Finally, the individual must not remove himself or herself from further consideration or otherwise indicate that he or she is no longer interested in the position.
Cohen posed this hypothetical: A big data algorithm scours the internet, finds a potential candidate from a social media site like LinkedIn who meets a job opening’s basic qualifications, and brings that information to a contractor, which then considers that individual for a position.
By merely having a social media account, has that person expressed an interest in employment to trigger the record-keeping and analysis requirements of the internet applicant rule? Cohen asked.
“I don’t know the answer,” he said. “I guarantee you that OFCCP doesn’t know the answer either right now.”
Morgan raised another potential problem: If a contractor runs a big data algorithm on all of the individuals in an applicant tracking system, will the contractor have “considered” those individuals under the internet applicant rule?
If the answer is yes, the contractor potentially could face “huge” applicant pools on which they must run disparate impact analyses, Morgan said. And analyzing big numbers generally makes it easier for the OFCCP to find statistically significant indicators of discrimination in a contractor’s hiring data.
Re-evaluation of the OFCCP’s internet applicant rule to account for big data tools may be necessary, they said.
- November 2020(57)
- October 2020(79)
- September 2020(72)
- August 2020(64)
- July 2020(71)
- June 2020(74)
- May 2020(50)
- April 2020(71)
- March 2020(71)
- February 2020(58)
- January 2020(62)
- December 2019(57)
- November 2019(64)
- October 2019(26)
- September 2019(24)
- August 2019(15)
- July 2019(24)
- June 2019(55)
- May 2019(82)
- April 2019(77)
- March 2019(71)
- February 2019(67)
- January 2019(77)
- December 2018(46)
- November 2018(48)
- October 2018(76)
- September 2018(55)
- August 2018(63)
- July 2018(74)
- June 2018(64)
- May 2018(65)
- April 2018(76)
- March 2018(82)
- February 2018(65)
- January 2018(80)
- December 2017(71)
- November 2017(72)
- October 2017(75)
- September 2017(65)
- August 2017(97)
- July 2017(111)
- June 2017(87)
- May 2017(105)
- April 2017(113)
- March 2017(108)
- February 2017(112)
- January 2017(109)
- December 2016(110)
- November 2016(121)
- October 2016(111)
- September 2016(123)
- August 2016(169)
- July 2016(142)
- June 2016(152)
- May 2016(118)
- April 2016(60)
- March 2016(86)
- February 2016(154)
- January 2016(3)
- December 2015(150)