Proportional Prejudice: How Diversifying Your Applicant Pool Can Lead to a Less Diverse Workforce
We know certain types of jobs tend to be held by more men than women, or vice versa. Programming jobs, for instance, are notoriously male-dominated, while positions in human resources are often populated by women. The managers we speak with and consult for are often aware of this imbalance and want to remedy it.
One common fix we hear is to recruit candidates more broadly. In other words, encourage more members of the underrepresented gender to apply to particular jobs. We have worked with several startups that specifically provide managers with information about the demographics of applicants so that they are aware of potential imbalances in their talent pools as they attempt to fix them. Conventional wisdom would support the idea that more women applying for tech jobs or more men applying for HR jobs would help even out gender imbalances in these fields. However, our research published in the American Journal of Sociology suggests this strategy can often backfire.
Proportional Prejudice
Most hiring processes begin with the employer viewing the pool of applicants. All subsequent hiring decisions, from screening to creating the short list to scheduling interviews, are made after seeing this pool of applicants.
Because first impressions have a disproportionate impact on the decisions we make, this initial view of the applicant pool can have implications for hiring decisions. We were interested in one particular, and rarely studied, decision: when an employer decides whether to hire any of the applicants or, instead, to let the search fail by deciding not to hire anyone.
To investigate the role of decisions to fail the search, we ran analyses using data from 7 million job applications for more than 700,000 job posts from 292,518 freelancers in an online labor market. Our dataset spanned 249,506 employers and more than 150 different job types, and it contained measures of applicant skills, ratings, and experience, allowing us to control for observable measures of applicant ability.
What our analyses revealed surprised us. We discovered that an employer’s initial impression of their applicant pools may shape their decision whether to hire anyone or not. When employers first view applicant pools, the proportion of women or men can tinge all applicants with stereotypes associated with the numerically dominant gender, a phenomenon we term “proportional prejudice.”
What This Means for Diversity Recruiting Efforts
An employer who receives an applicant pool with a majority of applicants of a gender not stereotypically associated with the job in question is more likely to form a negative impression of all applicants in the pool. This is because the employer will believe the pool does not contain the “kind of people” suited for the job.
In short, the greater the proportion of female job seekers for a technical or programming job, the more likely it was the employer decided not to hire anyone. Put another way: When a lot of women apply to a male-heavy tech job, employers are less likely to hire anyone at all.
The mismatch between the proportion of women applying and employer expectations for what applicants for a programming job should look like gives employers the impression that something is wrong with the pool. Interestingly, we found that this can alter future stages of the hiring process for all applicants, women and men alike. Ultimately, proportional prejudice increases the likelihood employers decide to hire when more typical applicants apply and not to hire when more atypical applicants do.
Though we would like to believe that gender bias in hiring is declining, our work finds that when we account for these failed searches, more women are rejected from technology jobs than men because they represent a greater proportion of job seekers who apply to jobs where no one is hired. We believe that proportional bias is also likely to affect diversity efforts for race as well — for example, when there are more Black applicants for jobs traditionally held by whites.
Our findings have implications for those managers seeking to recruit a more diverse workforce. Encouraging a more diverse applicant pool is one of the most widely touted practices in a manager’s toolkit, and it is one of the least likely to generate backlash from job incumbents. Yet our study shows that having a more diverse applicant pool can backfire if anyone involved in the hiring decision does not know why the pool composition differs from expectations.
Armed with this knowledge, we have been helping managers detect bias in hiring, identify where bias resides in the process, and effectively communicate their efforts with everyone involved. In this way, managers may increase the diversity of their workforces without inadvertently reproducing the bias they are trying to eliminate.
Ming D. Leung is an associate professor of organization and management at University of California, Irvine’s Paul Merage School of Business. Sharon Koppman is an assistant professor of organization and management at the Paul Merage School.