Take big data out for a spin, but donít be blinded by its charm
By Katie Loehrke, HR Editor, J. J. Keller & Associates, Inc.
Big data. It sounds trendy, legitimate, authoritative. A Google search on the term yields results like ďWhy You Need to Embrace the Big Data Trend in HR,Ē and ďThe Big Data Opportunity for HR and Finance.Ē
With search results like these, employers who havenít yet gotten on the big data bandwagon might wonder whatís taken them so long.
But those same employers might be wondering what big data is, exactly. At the risk of oversimplifying the concept, big data is a large and/or complex mix of information. It could be pulled from a variety of sources, including social media posts, metadata, financial records, web behavior, and more.
Big data for HR
In HR, the term is often used in conjunction with predictive analytics. That is, users accept that the presence of certain data or a certain data set will lead to better decision making. Big data is used by some companies to drive recruiting and hiring strategies (and decisions).
Just like mortgage lenders rely on a personís credit score (a complex calculation aggregating numerous data points) to make lending decisions, big data companies assert that they can help companies more accurately and efficiently identify the best and brightest employees by recognizing patterns in data.
Is it too good to be true?
While it may sound foolish to pass up a magic algorithm to greatly increase the chances of making successful hires, or pinpointing top leadership talent, thereís more to consider about big data than its obvious charm.
One potential problem for employers is born in the nature of the data. For instance, at a recent Equal Employment Opportunity Commission (EEOC) Technical Assistance Program Seminar, representatives of the agency recounted a story of one company that used big data to identify top computer programmers.
The information management company that was helping the company identify top talent had determined that many of the client organizationís most successful workers had visited a popular Japanese anime website. As part of its evaluation of candidates, then, the information management company gave a boost in ranking to those candidates who had visited the same site.
The risk of disparate impact
However, as the EEOC pointed out, employers have to be aware of the potential pitfalls of this type of criteria. Is it possible that a certain type of person is more likely to visit a Japanese anime site (perhaps a young white male) than others? If so, companies have to be aware of the potential bias they might be inserting into their hiring practices by relying on big data.
Even where there may not be a risk for illegal discrimination, companies may want to be aware that leveraging big data (if the process is not monitored closely) could have the unintended effect of limiting diversity in the workplace.
We didnít discriminate, the algorithm did!
Some employers might want to make the argument that any discrimination that might occur as the result of big data is unintentional, and therefore not problematic, but even unintentional discrimination is still illegal. In the end, big data isnít going to be sitting on the defendantís side of a courtroom, the employer is.
The moral of the story is not that employers canít take advantage of what may ó in some cases ó be the magic of big data. Instead, the takeaway is that employers have to do their due diligence to see beyond the magic and understand what is being measured, just as they would with any other employment test or selection procedure.
In short, if big data as a selection tool disproportionately excludes people in a particular race, sex, or another covered basis, the employer must be ready to show that the test is job-related and consistent with business necessity as it would with any other selection tool.
From Mary Borsecnik at J. J. Keller & Associates Inc.