Can Machines Make Better Hiring Decisions Than Humans?
Are you about to be outsourced? A recent study suggests machines can make better hiring decisions than managers.
But first—a disclosure. BrightMove delivers software solutions that support and ease challenges that face HR professionals. Our background and expertise is in recruiting and staffing—and we design software that does a lot of the work formerly done in HR units. Staffing, recruiting, onboarding, and back office support—we design tools that make your life easier. But don’t replace you.
So what does the study say? A recent paper from the National Bureau of Economic Research (NBER) is bound to start a few conversations about whether we should turn hiring decisions over to algorithms. The findings note, “Our results suggest that firms can improve worker quality by limiting managerial discretion.” Instead, metrics, algorithms, and test results should have the final say.
Algorithms and Big Data make the headlines a lot these days—the next new algorithm could be the next billion dollar unicorn. We all know what algorithms do for Google—but are computer conducted processes, and personality tests, superior decision-makers when compared to experienced recruiters or a hiring manager?
What the NBER study does—and does not—say
The NBER study is a meta-analysis of approximately 300,000 employment records, from lower skill service areas that involved 15 different companies. While workers in these jobs typically stay in their position under four months, the study found workers who were hired based on rank assigned by an algorithm stayed on the job longer—upping the retention rate for the company.
Increasing retention rates is important for any company. As we discussed earlier, high turnover is costly—in money, time, energy, knowledge, and employee morale. Some key points of the NBER study include:
- About one-third of the study group was evaluated using an online test that assessed their personality, technical skills, and basic fit for the job. The algorithm rated and ranked potential employees based on the results.
- In some cases, hiring managers were allowed to override test rankings.
- The results? Candidates given the highest rankings by algorithm stayed on the job approximately 15 percent longer than those hired by humans. Plus, hiring off the grid by managers did not lead to higher productivity, causing researchers to note, “exercise of discretion is strongly correlated with worse outcomes.”
Study author Dr. Mitchell Hoffman, of the University of Toronto, said, “It definitely suggests that more decision making powers should be given to the machine relative to the humans.”
Now let’s consider what the NBER study does not address.
The focus of this hiring study was the efficiency of decisions made to hire workers in low skill jobs, like call centers, and other highly repetitive jobs. Metrics used for evaluation included job test scores and personality measures. It is well known that all people—even hiring managers and recruiters—have conscious and unconscious biases that cloud judgment and can deter good decision making.
The use of analytics in certain hiring decisions could be useful. The practice is already well enmeshed in the recruiting process—for better, or for worse. Using keywords, ATS routinely parse candidate cover letters and resumes, ruling in—and potentially ruling out—well-qualified candidates on the turn of a phrase. The volume of job applications calls for the use of algorithms.
But on the hiring end, are machines, or people, better equipped to make employment decisions? Maybe it is both.
The real goal of hiring the right talent—better retention and engagement
The great benefit of machine learning, and cognitive computing, in recruiting is the ability to harvest and analyze Big Data and social information to create analyses, profiles, and side-by-side comparisons to aid with interview, hiring, and onboarding decisions.
While machine-made hiring decisions may be a trend of the future, there remain compelling reasons to understand how humans can make better hiring decisions—and why that is important.
The NBER study focused on a specific employee type. It is convenient, and perhaps easy, to relegate rote hiring decisions of rote workers to a machine—but will that really work?
The NBER study comes to important but narrowly applied conclusions. The unique factors that drive retention and engagement for many employment positions are probably best assessed by a human—with the support of well-designed staffing and recruiting tech.