People Analytics Help HR Catch Up With Evidence
- by Michael Housman
I have an unusual background in that I spent most of my time as an undergraduate and graduate student studying the intricacies of our health care system. In fact, I was trained as a health economist but with a special focus on organizational behavior within hospitals and health care systems. Over time, my interests evolved. I went from studying organizational culture within hospitals to focusing purely on organizational behavior and labor economics, but within an array of industries. I’ve had the opportunity to study human resources issues across such verticals as call centers, retail, hospitality, finance and so on.
I mention all this because I’ve witnessed firsthand a similar transformation occurring in the health care industry and the human resources space. I first became aware of the phenomenon known as evidence-based medicine approximately 10 years ago when I was just out of undergrad and spent time researching the evidence base supporting treatments like cognitive behavioral theory for traumatic brain injury. I learned that it’s often not the case that there is a clear-cut treatment for a given condition. Research must be conducted with varying levels of rigor until there is a solid enough body of literature to support one treatment or another. Eventually, large-scale reviews and meta-analyses emerged that would evaluate this research within the context of its design and arrive at a conclusion about what the evidence supports for a given diagnosis.
This phenomenon is called evidence-based medicine, and it attempts to systematize the practice of medicine in such a way that doctors will practice in a consistent manner according to what the extensive research suggests is appropriate. When doctors don’t have this sort of consensus available, their practice patterns can vary widely. For example, one study examined a number of hospitals and found that when pregnant women arrived ready for delivery, the likelihood that they received a cesarean section or not varied tenfold depending solely on the doctor and hospital they visited (holding all other factors constant). It makes no sense that the same woman presenting with the same pregnancy and the same vitals might receive a C-section depending largely on the practice patterns and preferences of the doctor in question. Evidence-based medicine attempts to reduce this variance to ensure that the same set of symptoms will result in the same diagnosis and the same treatment regimen.
What I’ve found interesting is that we are now beginning to see this exact transformation occur within human resources. The nascent movement called “people analytics” attempts to remove gut instinct, intuition and human biases from talent management in order to make workforce decisions in an evidence-based and data-driven way. As an analogy to the pregnancy example, if an applicant shows up for a job interview, the applicant shouldn’t be hired or dismissed simply because he or she spoke with interviewer A as opposed to interviewer B. The applicant should be tested in a rigorous and objective manner in order to determine whether their knowledge, skills and abilities fit the role for which they are being hired.
The idea is the same—analyze data and engage in research to generate best practices around the workforce and then disseminate those best practices to ensure that human resources practitioners are making decisions in an evidence-based way.
Let’s consider another practice with a questionable evidence base: reviewing resumes. We know from the research that recruiters spend an average of seven seconds on each resume, and what do they look for? Among other things, they look for an unusual work history: people who are job-hoppers or are among the long-term unemployed. My employer conducted its own research on work history and found that there is almost no correlation between previous work history and future job outcomes. In other words, weeding out job-hoppers and the long-term unemployed is bad practice. It not only hurts these individuals but also removes approximately 2 to 6 percent of viable applicants from the available pool.
Admittedly, our study is just one of many that should be conducted on this resume phenomenon. Once there’s enough evidence to support this finding that previous work history does not predict future job outcomes (or perhaps the opposite), there should be a massive campaign to educate recruiters and ensure that they’re not unfairly penalizing individuals with these sorts of backgrounds. That is the idea behind evidence-based decision-making, and it is slowly but surely making its way to the human resources world and replacing the gut instinct, intuition and cognitive biases that we know affect our objectivity.
As a researcher and an empiricist, I am very excited to see this evidence-based paradigm make inroads in another field. It will benefit employees and employers in the same way that evidence-based medicine benefits patients and health care providers.