Big Data and the Law: New Tools but a Better Workplace?
More than just numbers, analytics can bring humans and resources together.
Big data analytics promises to save companies millions of dollars by streamlining employment decisions and preserving workforces. But is there risk in relying too heavily on automated workforce decisions?
We encounter the results of big data analytics every day, yet we rarely question the appropriateness of its use. Think about the last time you applied for health insurance — the application likely requested information regarding your health history. Your personal health history can help predict whether you will become seriously ill in the future; a prediction that will partly determine your insurability. The process of applying for a loan is similar. The lender will review your credit history to determine how likely you are to repay the loan based on the lender’s experience with others with a similar credit history. This is analytics at work, and, for the most part, it operates very effectively.
Despite widespread adoption of big data analytics in virtually all aspects of business and business management, use of analytics in workforce management has lagged. This is due in large part to the way human resources professionals have been trained to manage personnel matters.
HR often involves emotions and complex notions of equity and fairness. This is a far cry from the sometimes cold, dry reality of financial transactions. HR professionals are trained to examine each personnel matter individually, talk to the parties involved, review documentation, and consider institutional employment policies in the legal context. Making employment decisions using cold, hard data seems wrong and, perhaps, risky. But is it?
Several reports, including from the U.S. Federal Trade Commission and the White House, have cautioned on the risk of making biased workforce decisions based on big data analytics. Users of HR analytics tools even have been advised to monitor their workforce for any evidence of “discrimination by algorithm.”
The Equal Employment Opportunity Commission held a public meeting on the use of big data in employment in October 2016, during which the Agency explored the benefits and risks of using big data analytics in the workplace.
Do the benefits outweigh the risks? Yes, if the analytics are designed and deployed properly. Appropriate use of analytics allows corporations to predict attrition likelihood, optimize recruitment efforts, gauge employee morale, and focus training and development efforts on what requires the most attention, among other benefits. These use-cases not only can result in a more efficient and better workforce, they can translate directly into company savings.
In Eric Siegel’s “Predictive Analytics, The Power to Predict Who Will Click, Buy, Lie, Or Die,” he points to case studies that demonstrate the value big data analytics brings to the workplace. In one, a well-known tech company used big data analytics to create a scoring system that predicts which of several hundred thousand employees were more likely to leave the organization. The flight risk score empowered company managers to prevent the loss from actually occurring or to plan for the departure. He said the system resulted in $300 million in potential savings.
The benefit is real, but commentators ask how real is the risk? Are the analytics and the algorithms on which they are often based tainted with bias?
Algorithms created to help employers make employment-related decisions certainly could be tainted with bias. For example, one designed to help a talent acquisition team identify successful candidates may take race or gender into account.
Even if race or gender is not used to explicitly identify successful candidates, using the algorithm could unintentionally result in disproportionately excluding a particular race or gender from the preferred applicant pool. Bias is entirely avoidable, however, if the analytics are carefully and correctly designed.
Humans, of course, are not perfect. We have a number of unintentional biases arising from seemingly well-informed decisions based on the experience and intuition of talent management team members. If used correctly, big data analytics can remove potentially biased intuition and, instead, support decisions with reliable and neutral data science.
Why do some commentators remain concerned? Perhaps it’s the feeling that employment decisions are about people, not cold, hard data.
In any event, responsible employers contemplating employment analytics platforms must ensure their algorithms and models do not incorporate protected characteristics such as race and gender as a predictive variable. Moreover, employers must review periodically the models to determine whether certain individuals are being disproportionately excluded or harmed.
The most reliable and effective HR analytics platforms provide guidance to companies faced with employment-related decisions. The algorithms alone should never drive the decision. This point is nicely illustrated in the above example of using flight risk scores. Once flight-bound employees are identified, managers can try to affect employees’ decisions to leave. This hybrid approach leverages the analytics to identify the areas of highest risk and permits companies to focus their efforts and resources in a meaningful and effective way.
Use of big data analytics in the workplace is here to stay. Embrace it. When utilized properly, analytics can have a significant impact on companies’ bottom lines and help preserve their employees.
Eric J. Felsberg is a principal and national director of JL Data Analytics at Jackson Lewis P.C.