Language and prejudice : the "invisible hand" of gender inequality in modern organizations

Abstract

While studies have shown that interactions between workers and employers reinforce gendered allocations of workers within firms, the literature has not yet demonstrated how these interactional mechanisms operate to create gender inequalities even before formal job applications are submitted. Such interactional mechanisms are documented and analyzed using a unique data from a web-based recruiting platform. We conduct a comprehensive content analysis of close to 150,000 jobs posted by approximately 25,000 recruiters reaching more than half million job seekers in the United States to find evidence that gender stereotypes influence recruiters' perceptions of the appropriate selection criteria for potential candidates, thereby affecting the language of job advertisements. Such cognitive biases, in turn, shape the ways in which job seekers of a particular gender inquire about a job. Specifically, when the share of stereotypically masculine words in job advertisements was larger, job seekers that inquired about such jobs were less likely to be female; when the share of stereotypically feminine words was larger, job seekers that inquired about such jobs were more likely to be female, holding all else constant. Further, we find that job seekers themselves were more likely to use stereotypically feminine (or masculine) words in their inquiries the greater the share of stereotypical feminine (or masculine) words increased in job postings. Lastly, we show that job seekers that show interest toward a job were more likely to be of the gender of the recruiter, holding all else constant. Even after controlling for potential within- and between- occupational variations, job seeker responses were strongly influenced by recruiters' use of language.