For many of us, finding the “right” career can feel like an impossible feat. When my little sister was in her last year of high school, she took a career aptitude test. Her top career? Chimney sweep. Luckily, things look to be working out for my sister. She’s now training to become an engineer, but who knows whether she will move on to something else: The average person today will hold about 12 jobs in their lifetime. At any given moment, one in three employees are underqualified for their current roles, while one in four are overqualified. Globally, the vast majority of people feel disengaged at work. Winding up in the wrong profession seems to be a surprisingly common experience.
While career tests can help, they’re no magic bullet. Most, like the one my sister took, rely on self-report surveys, which are not only time consuming, but can also be faked. It’s also debatable whether the results from these tests are meaningful. The popular Myers-Briggs Type Indicator, for example, has been critiqued for offering a “ridiculously limited and simplified view of human personality.” A new paper offers an alternative to the imperfect career test: the Twitter feed. The researchers applied machine-learning approaches to Twitter data to offer what they describe as a “21st-century approach to matching one’s personality with congruent occupations.” Marian-Andrei Rizoiu, a lecturer in computer science at the University of Technology in Sydney, and one of the study’s co-authors, said, “The study was pretty fascinating because it took this big data approach to personality and career.”
Your job-personality fit can even affect your income.
Rizoiu and his colleagues analyzed tweets from almost 130,000 Twitter users. By identifying patterns in users’ language, the researchers were able to capture what Rizoiu calls personality profiles—collections of traits and values that form who we are. He and his colleagues compared these profiles to the careers that users mentioned in their Twitter bios. The link was strong—so strong that the personality profiles could be used to “predict” professions with more than 70 percent accuracy. “That means,” Rizoiu explained, “that if you give me a random personality profile, without any other information, in three in four cases we can correctly guess their profession.”
If you dive deep into the data, those errors—the one in four cases where the application gets it wrong—are often less problematic than you might think. According to Rizoiu, many of the errors occur when the application confuses synonymous professions, like school principal and superintendent, or jobs that involve similar skills, like data scientist and software engineer.
The study could have exciting implications for future job seekers. “I think it can open up…possibilities,” said Peggy Kern, an associate professor at the University of Melbourne’s Center for Positive Psychology, and the lead author of the study. “My hope is that it would allow people to explore more and think about going beyond the typical way that we make [career] decisions.”
To recommend occupations, Kern and her colleagues’ method relied on two aspects of user personalities: traits and values. Traits, she explained, are common characteristics like extraversion or neuroticism. Values, on the other hand, represent what we truly care about, like helping others, respecting tradition, or achieving success. Both are core aspects of our personality, and both, it turns out, are deeply linked to how we feel and perform in our careers.
Past studies have connected personality to everything from hireability and career potential to job performance and professional success. Your job-personality fit can even affect your income. In a 2017 study of almost 8,500 employees, people whose personalities were well-suited to their professions were more likely to earn up to 10 percent more. The consequential significance of personality is becoming more widely recognized. Last year, the American Psychologist published a paper titled, “The Policy Relevance of Personality Traits.”
Finding the right match could even impact your health. “If what I’m doing [at work] fits within my value system, it has a positive impact upon my wellbeing,” Kern said. By using data to make career choices that feel authentic and meaningful, we might all be better off—professionally, financially, even physically.
Kern believes that a refined version of the method used in the study could be developed to help people identify career options that truly fit their traits and values, based on the digital traces that they leave through their online behaviors. Rather than pursuing whatever career their parents recommend, for example, people could use their own social media data to identify a variety of options that might be a better fit and more fulfilling.
Using big data—like this study’s large collection of tweets—can be helpful for uncovering trends that would otherwise be difficult to identify. But the method also relies heavily on the status quo, for better or for worse. “I think the paper makes perfect use of the social media data we have available, where you have a massive data set of people’s personalities in different locations,” said Kiki Leutner, a business psychologist and data scientist at University College London, who was not involved in the study. “But, of course, you can only look at the current state of things. You can see where extroverted people are more likely to work, but it’s unclear whether that means that extroverted people are better in those roles. It doesn’t necessarily mean that people’s vocational choices are right.”
It’s not clear whether there are meaningful differences in how diverse groups of Twitter users communicate on the platform—and whether that would affect the career options they’re recommended. But if demographics like race, gender, or age do play a role, then machine-learning approaches like the one Kern and her colleagues used could end up perpetuating existing biases in the workplace. “One of the biggest concerns I have is about how it could be misused,” she said. “I could very easily imagine some businesses taking it [as] a new way to screen for employees…I think that would be quite dangerous.”
Machine learning is only as good as the information it learns from. So, if a particular minority group was underrepresented in the data that the team analyzed, the analysis could end up contributing to some real biased outcomes. Leutner agrees. “If you wanted to use it in an [employee] selection context, you would have to make sure that it’s very reliable,” she said. “You would really have to prove that having those traits is beneficial to performance in that role.” Until researchers know whether these digital personality profiles truly represent the general population, this technology should remain an exploratory tool—more informational than prescriptive.
Still, the possibilities the research offers are enticing, especially as the modern workplace becomes increasingly automated. “Think about how many jobs that have always been there are disappearing,” said Kern. As more and more careers become redundant and workers are tasked with finding new ones, data-driven recommendations could be extremely helpful. “This could keep people learning on their career path, and that is going to be very important as we think about the constant change of jobs that is going to continue to happen throughout the 21st century.”
Career choices could soon be driven less by happenstance and more by data. When I asked Kern about how she ended up in her own career, she laughed. “I never imagined I’d be living in Australia, an assistant professor in academia,” she said with a shrug. “And yet, it’s been a good fit for me. I love what I do, but I feel like I’ve stumbled into it.”
Alice Fleerackers is a freelance writer and a doctoral student at Simon Fraser University, where she studies how controversial science is communicated in the digital sphere. Find her on Twitter @FleerackersA.
Source: Nautil blog