How Big Data may transform pre-employment testing

Password Please?

Let's pretend you're a new college graduate looking for a job. You apply via the web for a number of positions. One of the prospective employers sends you an email that reads:

What would you do?

Most of us would never give away our passwords. But what if there was a safe way to do so, for employer evaluation purposes only, of course? Still sounds a little crazy, doesn't it? But what would your kids do? Would they rather take a test?

Crazier still: What if the information is cached somewhere in the cloud while it's in transit so employer's don't even have to ask for access? Virtually all of the information an employer might care about is sent in an unencrypted format. So ... why not?

It's already started

Over the past few years, look-see's by employers at the Facebook profiles of candidates searching for unprofessional pictures or posts have become commonplace. While some may question whether the practice is ethical, it's become a part of the employment selection landscape and candidates have come to expect it. More importantly, this practice has not deterred would-be job seekers from posting their photos, thoughts, and videos freely. For candidates, having a web identity implies exposing themselves to anyone, including employers. After some initial hubub about public and private lives, we've all moved on.

We think that casual access to a candidate's social network account by a recruiter or hiring manager is only the beginning. As more data about us is collected and made available, it is likely to yield more and more information that is useful to employers making hiring decisions.

A continuous trail of breadcrumbs ...

Today virtually everything we do on the Internet is captured and stored - somewhere. Our social profiles, photos, videos, blog entries, emails, tweets, voice mails, web searches, homework, chats, comments, purchases, friending choices, resume postings, and other interactions have created a unique and personal trail of breadcrumbs that can help others discover who we are and what we're like.

The data-fication of everything

Google's executive chairman Eric Schmidt brings it to a point: "From the dawn of civilization until 2003, humankind generated five exabytes of data. Now we produce five exabytes every two days and the pace is accelerating." (http://www.dashboardinsight.com/news/news-articles/what-really-big-data-and-why-it-will-change-world.aspx) Keep in mind that Mr. Schmidt made that comment more than 10 years ago. 

While we've all been busy leaving our trails of crumbs, the collective amount of stored data has multiplied exponentially. This data is not just about us, of course. It includes everything from weather measurements to stock prices to health measurements to sports statistics. The total amount of data out there has truly exploded. The sheer quantity had grown so much that, as early as the Turn of the Century, technologists coined the term Big Data' to mean a data set so large and growing so fast that traditional relational database systems could not handle it.

Like the expansion of the universe, data-fication is accelerating. From the closed-circuit video cameras appearing at more and more street intersections, to sensors added to smart phones and now smart watches to smart clothing, jewelry, and furniture, the rate of data production is increasing with no end in sight.

Hello Data Science

To deal with the massive quantity of data that has been collected, a new field has emerged called Data Science. This field focuses on the extraction of knowledge from large amounts of data. Data Scientists are part analyst and part artist, according to Anjul Bhambhri of IBM (http://www-01.ibm.com/software/data/infosphere/data-scientist/). They make it possible to extract reliable and meaningful information from the massive quantity of data that is today available. For instance, several real estate companies use signals from cell phones and autos configured to anonymously transmit their GPS locations to measure traffic patterns and estimate commute times - in near real time.And I thought Google map's traffic function was using helicopters!

"He gets on base."

If you've seen the movie Moneyball (http://www.sonypictures.com/movies/moneyball/), or read the book by Michael Lewis, you may remember that line. Moneyball is a true story of how a data crunching Yale-educated nerd helped a professional baseball general manager recruit players based on their statistical performance in measures that had the biggest impact on his team. The result was a dramatic improvement in team performance and a permanent change in the way baseball players are selected. It's a great example of how data science can affect a business.

Data Science meets employment selection

Big Data and Data Science have pervaded most industries and many aspects of normal life. So, what about employment selection? The answer is that it's already happening. For instance, a research study by H. Andrew Schwartz and colleagues from the University of Pennsylvania shows looked at over 700 million artifacts (words, phrases, entries) from over 75,000 volunteers Facebook profiles and found strong evidence of the ability to predict key Big 5 personality factors (http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3783449/). Each participant took a separate personality test and provided access to his or her Facebook profile. Following up in this research, there is now a service that analyzes a person's Facebook profile to draw inferences about their personality (http://labs.five.com).

Looking ahead

So, does this mean we don't need pre-employment tests anymore? Probably not for a little while, at least. Characterizing the Big Five personality factors of a candidate is not enough to predict performance. There are many more factors that need to be measured. However, refinements are coming, and more inferences are likely as the availability of new types of data about a candidate increases.

What if you can scan a new graduate's college email account and records, or his chats and blog postings? Can you draw inferences about the candidate's cognitive ability and writing skills? Or what about analyzing the photo stream in a candidate's Instagram account? Is is possible to draw behavioral inferences? These are problems for the data scientists, as well as for those who worry about protecting privacy on the web. However, the more you think about them, the more 'possible' they become. It's only a matter of time.

So, what about pre-employment tests? Are they about to become extinct? Probably not completely, at least during the next decade. However, if current trends continue, we can expect to see tests supplemented more and more by information gathered and 'inferred' from alternative sources. It's really only a matter of time.