Social media sites such as Facebook, TikTok, and Instagram have given many organizations a new hiring tool. According to a 2018 CareerBuilder survey, 70% of employers check out applicants’ profiles as part of their screening process, and 54% have rejected applicants because of what they found. Social media sites offer a free, easily accessed portrait of what a candidate is really like, yielding a clearer idea of whether that person will succeed on the job—or so the theory goes.
However, new research suggests that hiring officials who take this approach should use caution: Much of what they dig up is information they are ethically discouraged or legally prohibited from taking into account when evaluating candidates—and little of it is predictive of performance.
In the first of three studies, the researchers examined the Facebook pages of 266 U.S. job seekers to see what they revealed. Some of the information that job seekers had posted—such as education, work experience, and extracurricular activities—covered areas that organizations routinely and legitimately assess during the hiring process. But a significant share of profiles contained details that companies may be legally prohibited from considering, including gender, race, and ethnicity (evident in 100% of profiles), disabilities (7%), pregnancy status (3%), sexual orientation (59%), political views (21%), and religious affiliation (41%). Many of the job seekers’ profiles also included information of potential concern to prospective employers: 51% of them contained profanity, 11% gave indications of gambling, 26% showed or referenced alcohol consumption, and 7% referenced drug use.
“You can see why many recruiters love social media—it allows them to discover all the information they aren’t allowed to ask about during an interview,” says Chad Van Iddekinge, a professor at the University of Iowa and one of the study’s researchers. “But that’s a problem, because one of the hallmarks of legal hiring practices is that they focus on behaviors within the work context. There should be a clear distinction between what people do during work and what they do outside of it.”
In the second study, the researchers explored whether such information affects recruiters’ evaluations. They asked 39 recruiters to review the Facebook profiles of 140 job seekers (obtained from a previous larger study) and rate each candidate’s hireability. The researchers then mapped the recruiters’ ratings against the content in each profile. Although the recruiters clearly took heed of legitimate criteria, such as education and writing ability, they were also swayed by factors that are supposedly off-limits, such as relationship status (married and engaged candidates got higher marks, on average, than their single counterparts), age (older individuals were rated more highly), gender (women had an advantage), and religion (candidates who indicated their beliefs got lower ratings). Factors such as profanity, alcohol or drug use, violence, and sexual behavior lowered ratings; extracurricular activities had no effect on scores.
In their final study, the researchers probed the end goal of social-media mining: hiring better people. They obtained supervisors’ ratings for 81 of the job seekers in the second study (chosen randomly) after six to 12 months of employment and surveyed those employees about whether they intended to stay in their jobs. They then asked a new set of recruiters to assess the Facebook profiles, dividing the recruiters into two groups. One group proceeded without any special instructions. The other was trained in best practices for evaluating social media information: Its members were told to focus on work-related information and avoid job-irrelevant details, use the same criteria to evaluate all individuals, take notes on their observations, and be mindful of decision-making errors and biases, such as a tendency to favor candidates whose interests or characteristics aligned with their own. Neither group’s assessments of the candidates accurately predicted job performance or turnover intentions, indicating that even with careful instruction, hiring officials stand to gain little from probing applicants’ online activity. (LinkedIn, which was outside the scope of the research, seems an obvious exception.)
The participants in the studies willingly granted the researchers permission to view their Facebook pages—but in many cases hiring managers don’t need to ask, because profiles are often public. What’s more, previous research has found that a third of U.S. recruiters request access to candidates’ Facebook pages, and the vast majority of job seekers comply. That’s beginning to change: More than 20 U.S. states now bar employers from asking applicants to pull up their social media pages during an interview or to share their usernames and passwords. EU regulators go a step further, forbidding hiring managers from viewing a candidate’s social media unless that person explicitly consents.
What about using social media solely as a negative screen—that is, to identify any warning signs, such as overt racism or misogyny? “We didn’t study that,” says Liwen Zhang, a lecturer at the University of New South Wales and the research paper’s lead author. “But our research shows that a recruiter will be influenced by everything she sees on a social media site, so if companies want to look for red flags, they should have someone other than the hiring manager do so.”
The researchers suggest that job seekers “clean up” their social media pages, including problematic content that others may have posted, and tighten their privacy settings. Companies and researchers should also explore alternative ways of using social media in the hiring process, they say. For instance, recent studies have found that machine-learning applications might be able to determine certain personality traits from social media profiles—information that could prove useful in managing people once they have come on board.
In the meantime, the researchers recommend that hiring managers resist the temptation to pore over candidates’ social media pages. “We aren’t saying that the information there is useless,” Van Iddekinge says, “but we don’t yet have the tools to find the signal in all the noise.”
About the research: “,” by Liwen Zhang et al. (Journal of Applied Psychology, 2020)