by Peter Cappelli | 8:00 AM September 8, 2014
The 2002 movie Minority Report told the story of a future in which law enforcement could tell who would commit crimes in the future. The police then arrested those people before they could commit the crimes.
A good deal of work in the social sciences tries to do the same thing, albeit without clairvoyance or Tom Cruise. The idea is to identify the attributes of individuals that cause them to act in certain ways in the future: What causes some students to do well in school, why are some patients bad at taking their medicine, and, for our purposes, what causes some candidates to perform well in jobs?
Most of the studies in the workplace have been done by psychologists. Despite the new hype about big data as a means to build such models, psychologists have been studying questions like what predicts who will be a good performer since WWI. Over the generations, we’ve gotten used to the fact that tests most applicants don’t understand, examining attributes such as personality, determine who gets hired.
Of course, there are lots of other tests that are not as well known but sometimes used by HR in the workplace, such as “integrity” tests that try to determine who will steal at work, something very much like the Minority Report movie.
But a few things have changed with the rise of big data. The psychologists have lost control of the effort. It’s now done by economists, data engineers, IT operatives, and anyone who has access to the data. It also migrated outside the firm to an ever-growing crowd of vendors who offer enticing claims about the benefits of their prediction software. Rather than giving tests, the new studies look for associations with background data. The better ones worry about actual causation.
As data has gotten easier to access and software has made analyzing it simpler, we can examine every aspect of employee behavior. My colleague Monika Hamori and I did a study of what determines whether executives say “yes” when headhunters call to recruit them; in work underway, a colleague recently identified the attributes of individuals who get laid off in a consulting firm based on email traffic; another is looking at the attributes of supervisors that predict which ones do the best job of training. You name it, it’s being studied now.
The promise of big data means that we are likely to get better at prediction in the future. Even a small improvement in predictive accuracy can be worth millions to companies that hire tens of thousands of people per year. These tools are especially attractive to retail and service companies because they have so many employees and such high turnover, which means they are hiring all the time.
Here’s the issue, which is not new but it has grown more important with the developments above: Many of the attributes that predict good outcomes are not within our control. Some are things we were born with, at least in part, like IQ and personality or where and how we were raised. It is possible that those attributes prevent you from getting a job, of course, but may also prevent you from advancing in a company, put you in the front of the queue for layoffs, and shape a host of other outcomes.
So what, if those predictions are right?
First is the question of fairness. There is an interesting parallel with the court system where predictions of a defendant’s risk of committing a crime in the future are in many states used to shape the sentence they will be given. Many of the factors that determine that risk assessment, some of which include things like family background that are beyond the ability of the defendant to control. And there has been pushback: is it fair to use factors that individuals could not control in determining their punishment?
Some of that fairness issue applies to the workplace as well. Even if it does predict who will steal, what if, for example, being raised by a single parent meant that you did not get a job, all other things being equal?
Second is the effect on motivation. If I believe that decisions about my employment such as promotions, layoff decisions, and other outcomes are heavily influenced by factors that I cannot control, such as personality or IQ, how does this affect my willingness to work hard?
And finally, unlike the Minority Report movie, our predictions in the workplace are nowhere close to perfect. Many times they are only a bit more accurate than chance, and they explain only a fraction of the differences in behavior of people.
The field of psychology has long thought about the ethical issues and moral consequences of their tests. At least as of yet, the new big data studies and the vendors selling them have not. How we balance the employer’s interest in getting better employees and making more effective workplace decisions with broader concerns about fairness and unintended consequences is a pretty hard question.
PETER CAPPELLI
Peter Cappelli is Professor of Management at the Wharton School and the author of several books, including his latest, The India Way (Harvard Business Review Press, 2010).
No comments:
Post a Comment