As companies seek out the best employees, biases in employment screening can threaten to sabotage their efforts. While the age of open data can provide more information on potential employees, it can also distract companies from the information that matters most.
"When you’re making a hiring decision, you’re trying to discriminate. But you’re trying to discriminate fairly. You’re trying to identify who can do the job versus who can’t," says Frederick Morgeson, a professor of management at Michigan State University who researches personnel selection and staffing.
Sometimes, even the ethnic connotations of a name can elicit improper bias. One 2004 study published in the American Economic Review found that resumes with white-sounding names got 50 percent more callbacks for interviews than the same resumes with black-sounding names.
The sheer wealth of personal information available on social media makes the potential for such bias greater. In one 2014 study, Carnegie Mellon researchers created profiles for fictional candidates on popular social networks and discovered that employers in some geographic areas were significantly less likely to call back a Muslim candidate than a Christian one.
Employers looking to avoid bias may even find that technology unconsciously sways them in unexpected ways. For instance, a 2013 study by Harvard Professor Latanya Sweeney found Google searches involving black-sounding first names were far more likely to be accompanied by ads suggestive of a criminal record than white-sounding first names – regardless of the individual’s actual history.
Just seeing a picture of someone online can easily stoke bias based on race, ageism or appearance. And it’s hard to avoid. According to LinkedIn, profiles on the site are 14 times more likely to be viewed if they include a photo.
"What happens if you see a picture of somebody when you Google them? You say ‘I don’t want to hire somebody who looks like that,’ or ‘I want to hire somebody who’s younger.’ None of that’s related to success in the job," Morgeson says.
Further complicating things for employers is the fact that even some questions that may be helpful in selecting better candidates may also at the same time create inappropriate bias. For example, studies show a link between how far someone lives from a job and their likely job tenure. But screening for this characteristic may discriminate against economically disadvantaged candidates who live in less desirable neighborhoods farther from central business districts.
"In that case, you could ask questions that are obviously related to success in the job," Morgeson says. "Instead of asking how far away someone lives, ask them if they have reliable transportation to get to work at 8 a.m."
To be more accurate and less biased overall, Morgeson recommends that all employers focus on job-relatedness and standardization in all of their hiring criteria. As part of this effort, assessments such as ability or intelligence-based testing, relevant skills testing –such as a typing test for a secretary – and structured interviews can be helpful.
"For structured interviews, the evidence is very clear – they work very well. They work as well as our best tests," Morgeson says. "But the key part of that is that they have got to be structured. The typical off-the-cuff unstructured interview does not work."
Even employment-focused personality tests can have validity in measuring certain personality characteristics that are job-relevant, according to Morgeson. Companies such as Knack and Kronos are also offering game-like employment tests that are gaining traction. But Morgeson cautions that game scenarios, while innovative, are so new that their validity hasn’t been conclusively proven.
To eliminate potential bias during a job interview, employers should ask the same questions of every candidate. Tests and assessments should likewise be given to everyone equally and in the same order. The same standardization should also be applied to procedures for background checks and reference-checking. Controlling hiring managers’ access to outside information can also help them avoid being unduly influenced by information about candidates.
"The advice I would give any employer in relation to hiring candidates is to try to keep your hiring managers or your HR people from Googling a person’s name and just randomly looking at information," Morgeson says. "Because it can bias you and is unlikely to be related to their success in the job."
In the end, it’s easier than ever to learn more about people by looking them up online. But the things you’ll learn by doing that may not be statistically relevant to how they’ll perform in a given role.
"If you have a valid selection process, you want to maintain the integrity of that. You don’t want people to be biased by outside things because that’s going to introduce error or noise into your measurement of your candidates," Morgeson says.