I’m looking forward to joining my colleagues Dennis Hirsch and Jay Levine for a roundtable discussion of “Big data, data analytics and the law: What your company needs to know about the next big thing” on May 13. Here is a glimpse into what I plan to talk about from the employment lawyer’s perspective:

Even if we don’t know exactly how big data works, we know what it can do for us in our daily lives. Movie suggestions on Netflix. Targeted coupons at the grocery store. Cheap airfare and hotel rates. Facebook suggestions of people we may know. There is a certain creepiness to all of this but many (most?) of us seem willing to overlook it for the convenience and opportunities it provides.

Human resources departments now are figuring out how to use big data in the workplace. LinkedIn was one of the first businesses to recognize the value that data held for employers. At its most basic level, LinkedIn can steer its individual members to potentially attractive jobs that fit their profile and, for recruiters, it provides a rich database of candidates, including people who aren’t even looking for a new job. But there are a lot more than just recruiting opportunities. Companies like Knack now promote tests like Wasabi Waiter and Dungeon Scrawl that it claims will reveal job applicants’ talents, traits and skills to permit employers to identify the best candidate for their needs. JP Morgan Chase apparently has developed software that analyzes its own employees’ data to try to identify which ones are most likely to “go rogue,” so it has time to stop them before they do.

But with these opportunities come risks. One of the primary risks is not understanding the data being analyzed, where it comes from and how it was collected. Garbage in still equals garbage out. Employers therefore must always recognize that the data being analyzed, at best, identifies correlations and not causation — meaning that the process will be subject to challenge before the EEOC if there is a disparate impact and the test results or other data analysis can’t be validated as predicting success in the job for which the data is being used. In addition, latent biases also can corrupt the data collection and analysis processes leaving the process open to challenge as a discriminatory pattern or practice if for instance the data being analyzed tends to exclude certain protected classes of applicants or employees.

Other potential legal issues abound. Employers court ADA claims if pre-employment tests being used to screen applicants are considered to be psychological in nature and are administered before a conditional job offer is made or if persons with disabilities cannot or are not accommodated in the test taking process.

Finally, the accumulation of so much data is contrary to the primary data protection concept that you collect only the data that you need and keep it only for as long as you need it. As a result, employers contemplating using big data in their workplaces should incorporate privacy and security controls into all aspects of the data collection and analysis processes before implementation of any big data program.