Steps to reduce 'cybervetting' bias in hiring

A new paper on cybervetting says that organizations need to develop and implement clearly defined rules regarding how they use online information about job candidates. Failing to better regulate their use of cybervetting can introduce bias into an organization’s hiring processes, invade the privacy of job seekers and ultimately hurt the organization’s bottom line.

What the researchers say: “Companies seem to assume that cybervetting offers some sort of advantage to their hiring process,” said the first author of the paper. “But our analysis of the research on cybervetting finds that these presumed benefits are unclear – and it is increasingly apparent that cybervetting creates opportunities for biases to influence the hiring process.”

“Our previous work highlights the scope of these biases,” a co-author. “But this paper addresses actions that hiring agents and organizations can take to limit the potential harms associated with cybervetting. We also outline things job seekers can do to try to limit their exposure to bias, though job seekers have far less control over the process than the people who are doing the hiring.”

For this paper, researchers drew on all of the available research on cybervetting and analyzed it through the lens of sociology of work and labor markets. Broadly speaking, that means they looked at the science of how hiring processes work, how cybervetting is changing that, and what people can or should do about it.

The researchers outlined how cybervetting is important because it can adversely affect society in several ways. First, cybervetting can have a pronounced impact on the workforce by broadening discrimination and making bias more pronounced in hiring. These can be biases related to race, religious affiliation, gender, sexual orientation, age and so on.

Second, cybervetting can reduce diversity in organizations if the people doing the hiring are focused on finding a good “culture fit” for the organization. And a lack of diversity hurts the bottom line of many businesses – particularly those that rely on innovation.

“And cybervetting also promotes privacy invasion, which is deeply problematic,” noted the lead author. “I think we’ve already reached a point where we are accepting far more invasion of our personal privacy than we should be comfortable with.

“While job seekers have no control over whether prospective employers engage in cybervetting, or how potential employers use that information, there are a couple of things that job seekers can do to reduce potential risks related to cybervetting.

“One: remove information employers may view negatively. That could include posts about drug use, drinking or profanity. Remove posts about religious affiliation. Remove posts that convey negative attitudes about work, etc. Two: consider using privacy settings to make it more difficult for potential employers to access your information.”

The researchers also outlined steps that hiring agents should take in regard to cybervetting. Hiring agents are hiring managers, people who work as recruiters in an organization, or people who organizations hire to work as third-party recruiters or headhunters.

“Research tells us that, many times, cybervetting is done without any clear idea of how it relates to the job or the nature of the work,” the researchers said. “We urge hiring agents to consider how their cybervetting relates to specific work tasks or competencies.

“Ultimately, our research makes clear that if organizations are going to use cybervetting, they need to develop guidance to reduce the risk of cybervetting being used in a way that introduces bias to the hiring process. There needs to be a systematic, rigorous, informed process in place, with clearly defined goals. And what we’ve found is that few organizations have this sort of guidance in place.”

While the researchers note that developing meaningful rules governing cybervetting will be a big step in the right direction, there will still be a host of questions and challenges surrounding the practice.

“Once guidance is developed, you run into a follow-up challenge, which is how to enforce cybervetting guidelines,” the lead author concluded.

So, what? The study adds to a now large body of research showing the dangers of making judgements on the basis of AI—in health, in law enforcement, in school grading and in hiring. As all of the studies have shown, these cyber technologies are not unbiased and are very frequently simply inaccurate.