Research indicates that organizations are not meritocratic. Senior executives of large firms weight ambiguous factors like “culture fit” as equally important as critical job skills when making promotion decisions. Criteria beyond the scope of knowledge, abilities, and performance are used in hiring decisions across firms, industries, and management teams. And, as a wealth of studies show, the result is better career outcomes for those already at the top of society’s social ladder.
In recent research, we examined how something as simple as people’s speech patterns can lead to (often accurate) judgments of their socioeconomic status, and confer advantages to people with elevated social standing and disadvantage to everyone else.
Social psychologists have long documented humans’ propensity for thin-slicing, that is, relying on very brief snippets of observable behavior — including appearance, gestures, body language, and speech style — to judge a person’s traits (e.g., conscientiousness, competence, interpersonal skills) and get the gist of what someone is like.
Social class is also a part of this sense-making process, and prior research has shown that a 60-second social interaction between two strangers is enough to lead perceivers to accurately infer the parental income and education of undergraduate college students.
We wanted to test whether people’s speech patterns — how they pronounce words and phrases — would allow perceivers to predict their socioeconomic status and make them more or less attractive as job candidates.
In our first study, we presented participants with audio clips from people saying the same list of seven words out of context (e.g., “yellow,” “imagine,” “beautiful”). We then asked them to guess whether the speaker had earned a four-year college degree or not. Listeners were able to correctly guess the education of speakers 55% of the time, significantly better than chance.
In a follow-up study, participants listened to 75-second recordings of speakers describing themselves. The participants then rated the speakers’ social status based on attained education, occupation, and income. Those ratings tracked with the speakers’ actual status in as little as 30 seconds. When participants read a transcript of the speaker’s description, rather than hearing it, they weren’t able to make such quick and accurate assessments, suggesting that initial impressions were driven more by how people speak not what people said.
In our last study, we enlisted 274 people with prior hiring experience and had them listen to a 25-second sample of speech from applicants for a lab manager position. As in our other experiments, participants accurately perceived the socioeconomic status of those speaking, and, alarmingly, judged those of lower status to be less competent, a worse fit for the job, and deserving of a lower starting salary and signing bonus than their higher status counterparts. Critically, participants made these judgments without any information about the applicants’ qualifications.
That an informal conversation, lasting a mere 25 seconds, can bias hiring decisions in this way is a sobering reminder of how our default organizational practices reproduce societal inequality. Practically, these results mean that people, even those with experience reviewing resumes and interviewing job candidates, instinctively use social class cues as proxies for critical jobs skills — thus discounting those who may still be highly skilled but are from disadvantaged backgrounds.
What can organizations do? A first, and ill-advised, reaction is to suggest that job applicants should learn and practice the speech patterns of the people who populate the jobs to which they aspire. Asking people to “code switch” has clear costs. When individuals cover an aspect of themselves at work, it can deplete cognitive resources and reduce health and well being. And it prevents organizations keen on diversity, equity, and inclusion from reaching those goals, because underrepresented employees are actively monitoring their behavior.
Another possible organizational solution is to use automation and machine learning in hiring to avoid human bias. But algorithmic decision-making still relies on people to decide what data to use and which characteristics are valued.
We advocate for a third solution: Ensuring that hiring decisions both value and incentivize the creation of a workforce that is diverse in terms of race and socioeconomic status. Such a shift will extend to practices such as targeted college recruitment of lower-SES candidates, mentoring programs for employees who are the first in their families to graduate college, and accountability checks implemented alongside organizational diversity programs.
Though our work reveals a disturbing and relatively rapid psychological process that biases hiring in favor of higher status job applicants, we can override it by proactively embracing capable job applicants who exhibit all the requisite skills, but do not look or sound exactly like the people doing the hiring.
This content was originally published here.