How Online Personality Assessments Could Revolutionize Hiring
Usually, even a psychology study that makes a considerable splash in the scientific community fails to interest the mainstream media. But back in March 2013 -- when Michal Kosinski, along with two co-authors, published a study in the Proceedings of the National Academy of Science that found that what you “Like” on Facebook can reveal extremely personal (and statistically valid) information about you - everything from your race, to your IQ, to your sexuality – it made big waves everywhere.
Working with a dataset of over 58,000 Facebook users’ Likes through an app called myPersonality, Kosinski and company developed a set of statistical models that were able to predict personality traits with uncanny accuracy. The study went viral, covered by media outlets from CNN to Gawker.
The idea that our public Likes – those Facebook pages we often click casually – could expose such personal depth (intelligence, sexuality – even slippery traits like extroversion, neuroticism, conscientiousness) may be innately unsettling, but according to Kosinski, it’s just the tip of the informational iceberg. A researcher at Cambridge University and a research consultant for Microsoft, Kosinski is convinced that individuals will soon have access to extensive, highly accurate psychological profiles compiled from their online fingerprints. There would be many implications to such a large-scale roundup of personal data including, Kosinski believes, a complete overhaul of standard corporate hiring procedures.
He insists that fast and automated psychological assessment could revolutionize recruitment by enabling employers to evaluate millions of applicants within seconds.
“Our model,” Kosinski says, “is not about predicting personality from Facebook Likes – it’s about predicting personality from your behavior. And this behavior can be expressed just as easily from browsing the internet, as it can by hanging around, visiting a bookstore, or any other kind of offline expression of preference.”
From LinkedIn profiles to psychological profiles
As we spend more and more of our lives online – shopping, communicating, and browsing – we are providing huge stockpiles of data that researchers like Kosinski could potentially collect, feed into an algorithm, and use to produce a nuanced psychological profile.
“There is no magic in terms of what’s being done, because our model follows the established rules,” Kosinski says. “The magic is in the scale at which we can now do it.” He predicts that in a few years most people will have a psychological profile online, the way they have a Facebook or LinkedIn account.
It’s already partially possible. You can check out what your Facebook Likes say about you at websites such as YouAreWhatYouLike.com and ApplyMagicSauce.com. If you log in through your Facebook account, both sites automatically perform a personality test that rates your openness, stability, agreeableness, extroversion and conscientiousness. While the results are fairly primitive (primarily because they only work with your Facebook data), Kosinski says that corporations could easily develop more sophisticated models that paint an accurate picture of applicants’ psychometric profile that takes into account all of their online behavior.
Job seekers would then be able to obtain their profiles, and post them on LinkedIn or any other recruitment site, allowing employers to scan thousands of potential applicants and find the most promising candidates for a particular position in seconds. “Instead of approaching hundreds of people, employers could approach two or three finalists who are a really good fit, and just interview them,” says Kosinski.
Currently, companies often ask job candidates to take a psychometric test, typically in the form of a traditional questionnaire. But because the process is expensive and time consuming, it’s usually reserved for individuals applying for high-level positions.
As Kosinski points out, this is a self-selecting group. “These applicants generally have graduated from a good university and have proven success elsewhere,” he says. “You’re choosing between the good and the very good.”
Kosinski believes psychometric tests would have the most impact when a company is looking to fill lower level positions. They are rarely used in this capacity, which he finds impractical. At a company like Walmart, which has hundreds of entry-level jobs, candidates are rarely asked to take a personality test. However, each position requires specific skills and, one could argue, psychological traits. For example, good cashiers, given their interaction with customers, benefit from being outgoing.
Theoretically, the relationship is mutually beneficial: applicants are placed in roles that suite their personalities – and are presumably happier for it – while the employer is able to fill vacant positions with competent individuals. The process could also be viewed as a more meritocratic approach to traditional hiring methods, which sometimes favor the well-connected over the qualified. Promising applicants, who may not have otherwise been identified, can be elevated to higher positions in the company, or encouraged to return to school for additional training.
While he can’t predict a date, or even a year, when this recruitment revolution will play out, he is certain there will be a tipping point. "It will dramatically change from one day to the next, which is characteristic of many new, transformative technologies."
Privacy on the line?
While this may be an efficient way to hire, it raises thorny questions about discrimination based on previously invisible attributes like IQ and personality type. And it smacks of a Big Brother-style intrusion into the most personal part of ourselves.
Kosinski agrees that when it comes to this amount of personal information, there is the danger of abuse. That’s why he is adamant that individuals should have complete access and control of their own data.
“No one would agree that a company should have the right to look at a Facebook profile without consent, and make a psychological profile,” he says. “No one wants that. If anyone wants to do something with it, they need your permission. You can give your permission allowing them to use it forever, but they have to ask first.”
Lawmakers, he feels, need to address the issue by designing policies and tools to minimize the associated risks that come with such extensive, highly personal information.
If properly regulated, he believes that this system will ultimately empower job applicants. With traditional psychometric tests, company psychologists have access to the results before you do. “Whereas nowadays,” Kosinski says, “technology allows the participant to see her score first, and then she can decide whether or not she wants to share it. It sounds like a detail, but it changes the rules of the game.”
There are a seemingly endless number of questions surrounding this idea. For instance, how “voluntary” will it be? Will employers require applicants to submit their profiles to be seriously considered for the job? And if so, will people respond by consciously changing the way they behave online?
When pressed, Kosinski admits that “obviously if you don’t reveal your data, the company can say, ‘Ok we’re not hiring you because you don’t want to give away your score,’” but he maintains that it’s important nonetheless for individuals to “retain the right to say no.”
Kosinski’s prediction also raises the question of job applicants attempting to game the system by modifying their online behavior to fine tune their psychometric results. While he concedes this is a possibility, he doubts it will have much of an impact: gaming the system would mean fundamentally changing the way you operate online. As our offline lives and online lives continue to merge, this essentially means drastically changing the way you live. “If you can fake being open minded for five years,” Kosinski says “then you’re probably just open minded.”