Statistical discrimination is the theory that when employers do not have full information about the productivity of job candidates, they rely on their beliefs about observable characteristics, such as race or gender, to assume candidates’ productivity and make hiring decisions. In economics, statistical discrimination theory is often discussed as a rational, profit-maximizing, and commonplace strategy in the face of incomplete information. This article argues that because statistical discrimination tends to be presented this way, exposure to it can lead employers to justify and use stereotypes in hiring. In this survey experiment, participants who read about statistical discrimination theory had a stronger belief in stereotypes, stronger acceptance of stereotyping, and engaged in gender discrimination in a hiring simulation to a greater extent, compared to control groups. Reading a critical commentary of statistical discrimination along with the theory lessened these effects. This study suggests that how people are exposed to ideas about discrimination can affect whether they act discriminatorily, in turn shaping labour markets.


Statistical discrimination theory is a dominant social science framework for understanding discrimination. It appears often in policy papers, popular media, and economics courses. The theory suggests that when employers have incomplete information about candidates, they will make “educated guesses” to infer who the most productive candidate will be, using their beliefs about observable characteristics as proxies for quality. For example, they might believe older workers are less innovative on average than younger workers, and hire a young worker based on this belief. However, although employers may not be making consciously discriminatory decisions, their beliefs about certain demographic groups are often shaped by stereotypes rather than accurate statistics. Statistical discrimination thus harms candidates belonging to these groups.

Nevertheless, statistical discrimination tends to be framed in economics textbooks, lectures, and media as profit-maximizing, efficient, and even commonplace when information about the future productivity of job candidates cannot be fully known. Statistical discrimination then becomes easy to justify. It stands in contrast to taste-based discrimination, i.e., overt prejudice towards a certain group, which is not considered as societally acceptable.

This study predicted that people exposed to statistical discrimination theory would show greater belief in stereotypes and their usefulness, and that they would engage in more hiring discrimination, compared to other groups. It also predicted that reading a critical commentary of the theory would mitigate these effects. The author conducted a survey experiment with 2500 participants, all of whom were U.S. residents with managerial experience recruited from an online survey platform. Participants were exposed to one of four texts: 1) statistical discrimination theory, 2) unrelated materials, 3) taste-based discrimination theory, or 4) statistical discrimination theory with critical commentary. Exposure materials were created from college textbooks, articles, and other accessible descriptions about these theories.

After exposure, participants were asked about their attitudes towards stereotyping and undertook a hiring simulation, where they had to hire four of ten job candidates based on limited information (first name, university, major, GPA, and summary of work experience).


Results showed that participants who were exposed to statistical discrimination theory without critical commentary perceived stereotyping as more acceptable and stereotypes as more accurate compared with participants in control groups. They also hired fewer women in the hiring simulation. On average, participants in control groups selected gender balanced teams, with two women in teams of four. Those who read about statistical discrimination without critical commentary created a gender gap of about 10%, with their average teams comprising of 55% men and 45% women—even when controlling for factors such as university major, work experience, and GPA. However, these effects were mitigated for those who read about statistical discrimination theory with critical commentary.


The way we talk about discrimination has impacts on the labour market and can reinforce inequality – Statistical discrimination tends to be framed and discussed as profit-maximizing, rational, and commonplace. This study demonstrates that exposure to this theory without including critiques of it may lead to increased discrimination in the labour market. Economists, professors, and other leaders have a responsibility to ensure that statistical discrimination is not normalized, whether in classrooms or workplaces. Critical commentary about statistical discrimination—such as that people’s beliefs about specific groups are often not based in fact, and that it harms individuals who do not fit stereotypes—should be communicated.

Managers should carefully consider whether they are using statistical discrimination in hiring and mitigate this behaviour – It may be tempting for managers to make hiring decisions based on statistical discrimination, as people tend to believe that their perceptions about different demographic groups are accurate. Hiring managers and other decision-makers in organizations must ensure that they are not normalizing selecting candidates based on unfounded assumptions about characteristics that are unrelated to the job, such as race, gender, and age. This behaviour could reinforce inequality within organizations and continue to normalize statistical discrimination for future hiring.

Research brief prepared by:

Carmina Ravanera

Tilcsik, A. (2020). Statistical Discrimination and the Rationalization of Stereotypes, American Sociological Review 86(1): 93-122.

See more research briefs


Statistical Discrimination and the Rationalization of Stereotypes


András Tilcsik


American Sociological Review






Research brief prepared by

Carmina Ravanera


Sign up to our email newsletter to receive information about research, events and education.