The Impact of Teacher Gender Bias on High School Student Outcomes
Using novel linked data to track educational and employment trajectories in Peru
Joan Jennifer Martinez (PhD candidate in Economics, UC Berkeley) measures the impact of teachers’ gender bias on high school students’ education and employment trajectories using novel data. This study was supported by CEGA’s Psychology and Economics of Poverty Initiative.
Although gender wage gaps have declined worldwide, this pattern has been uneven between higher-income and low- and middle-income countries. Labor supply outcomes continue to differ across genders globally, as unemployment and part-time jobs are more common among women. While recent research suggests that these gaps may be partially driven by behaviors that differ across genders — such as salary requests (Roussille, 2021), self-promotion (Exley and Kessler, 2022), and job-attribute preferences (Le Barbanchon et al., 2020, Biasi and Sarsons, 2020) — such behaviors may be reinforced by preferences and attitudes acquired prior to entering the labor market. Gender bias encountered during school, including students’ exposure to teacher biases may have long-lasting effects on their labor market outcomes.
Measuring the Gender Bias of School Instructors
Using novel data from two sources — linked administrative information on students’ education and employment and nationwide survey responses from public high school teachers and students — I evaluated the long-term effects of gender-bias exposure on 1.7 million public high school students expected to graduate between 2015 and 2019 in Peru.
To construct and validate a measure of teachers’ gender bias, I developed an online government portal that recorded approximately 2,400 teacher responses and 4,600 student responses to an Implicit Association Test (IAT) and gender-attitudes surveys. Data was also collected in person through sessions in Lima during 2022. I constructed an assessment-based measure of teachers’ gender bias comparing the gender differences in teacher-assigned and blindly-graded tests. I veiled this measure with an alternative measure of bias using the IAT. Next, I used Empirical Bayes methods to retrieve teacher-level measures of teachers’ gender bias in assessment, later using this measure as a dependent variable for estimating its long-term effects. I estimated the long-term consequences of bias by comparing students enrolled in the same school, grade and year but assigned to different teachers and, therefore, exposed to different levels of gender bias.
Study Results and the Long Shadow of Gender Bias
Study results suggest that when high school students are exposed to the gender biases of teachers the effects can be deleterious and long-lasting, impacting factors such as whether students finish school, whether they apply to and attend college, and whether they succeed in formal sector employment as young adults.
Math teachers who strongly associate males with scientific disciplines give higher scores to male students, when compared to blindly-graded test scores, while language arts teachers who strongly associate females with humanities-based disciplines award higher grades to female students. Figure 2 shows the distribution of teachers’ IAT scores: math teachers who are men more strongly associate males with science than math teachers who are women, and the converse holds for language arts instructors.
Findings also suggest that female students who are assigned to more biased teachers are less likely to complete high school and apply to college than male students. Assigning female students during one grade to a high school math teacher who exhibits one standard deviation more severe gender bias against girls, compared to the average teacher, reduces girls’ likelihood of ever graduating from high school by 1.3 percentage points (1.6 percent of the group mean) relative to the case in which boys are assigned to this teacher.
Finally, female students assigned to more biased teachers in high school are less likely to hold a job in the formal sector after graduation and/or have fewer paid working hours relative to their male classmates. Female students who are assigned to a math teacher who is one standard deviation more biased during one academic year are 1.3 percentage points less likely to hold a formal-sector job at ages 18–19 (equivalent to 18 percent of the group mean). Further, this bias exposure causes monthly earnings losses of USD 2.6 at ages 18–19. The magnitude of this effect is large enough that it exacerbates the gender wage gap between 11 and 30 percent between ages 18–23, leading female students to suffer cumulative losses during their first two years in the labor market.
Contributions and Policy Implications
Educators’ gender biases in student assessments are a previously undocumented source of gender gaps in formal sector employment and earnings. These results offer insight into the effects of teacher biases on long-lasting human capital decisions and outcomes, especially the adverse conditions women face. Designing policy interventions to reduce the prevalence of such biases in educational settings is urgent. Another important area for potential study is whether students can be trained to recognize teachers’ gender biases in order to prevent these biases from affecting their decisions.