An experiment in Peru shows that when prompted by subtle socio-economic cues, teachers exhibit unconscious biases The Project that significantly affect In an experiment involving 600 public school teachers their evaluation of in Lima, Peru, the World Bank tested whether teachers low-income students. who were asked to evaluate the scholastic aptitude, behavior, and education potential of a student named “Diego” were unconsciously biased towards him when prompted by socio-economic markers. To do so, we asked teachers to evaluate Diego’s performance after watching a video where he takes an exam. In addition, to prime teachers to recognize Diego’s socio-economic background, teachers first watched a video that simply introduced Diego and showed what he did outside of school. This introductory video had two versions showing Diego in two different social- economic settings: middle- and low- income. To assess the extent to which teachers use social background as a short cut to assess aptitude and behavior, teachers were presented with one of two Even Teachers are Bested variants of the video in which Diego takes an exam. In the first version, Diego’s performance is ambiguous. by Their Biases He correctly answers some difficult questions but also incorrectly answers easy questions. Sometimes he is paying attention; sometimes he is distracted. In the Deep-rooted biases can inhibit our best intentions. second version, Diego’s performance is less ambiguous. Research has found that we all have unconscious He correctly answers most questions and behaves like a biases and preferences related to gender, race, model student. sexual orientation, socio-economic background, or other aspects of identity. Unconscious bias is just In the experiment, each teacher received a tablet and that – “unconscious,” and often unrelated to our goals were randomly assigned a combination of the two and intentions. introductory and two exam videos. After seeing these videos, they were asked to evaluate the student’s The science behind this is that our brains have evolved scholastic and behavioral performance. If Diego’s social to categorize the world to help us simplify it. We all background does not matter for teachers, then their use mental shortcuts to process information and make assessment of Diego’s aptitude, behavior, and potential decisions quickly. These shortcuts can be useful when should be the same no matter which introductory video making decisions with limited information, bandwidth, or they saw. If, however, teachers do use social background time, but can sometimes lead to harmful stereotypes. So, to assess students, then we might expect them to how much can unconscious bias influence a school teacher’s interpret the same exam video differently, depending on assessment of students’ achievement and their potential? which introductory video they saw. The Results In ambiguity, unconscious bias greatly influences Removing ambiguity is not enough to eliminate how teachers assess students. all unconscious bias. In the ambiguous performance variant, teachers end up Even when Diego unambiguously performs well, teachers being influenced by Diego’s socio-economic background are significantly harsher evaluating his behavioral scores when they assess his scholastic aptitude. “Poor” Diego when they think he comes for a poorer background. For was 22% more likely to be rated as performing below example, teachers score “poor” Diego’s motivation and grade level, a difference equivalent to two months of character 11% and 6% lower, respectively, than the “non- school delays. By contrast, in the high-performance poor” Diego. Thus, it seems that when teachers cannot variant where “poor” Diego answers most questions apply their stereotype to their assessment of Diego’s correctly, teachers assess his performance as better scholastic aptitude, as he unambiguously does very well than that of “non-poor” Diego. in the second exam variant, it gets transferred to another dimension. In this case, to their behavioral assessment. DIEGO’S PREDICTED GRADE LEVEL Difference between poor and non-poor Diego Expectations for poor students are 4.00 always low. Regardless of which variant was used, teachers 3.50 have lower expectations for Diego’s final educational attainment when they are primed to think that he 3.00 is poor. In the ambiguous variant, teachers’ expectation of Diego continuing past high school decreased from 60% to 40% when they were primed to think that he 2.50 was “poor.” The difference is only slightly smaller in * 3.20 3.42 3.90 3.74 the high-performance variant. Interestingly, however, 2.00 teachers’ expectation for “poor” Diego continuing Ambiguous Performance High Performance past high school in the high-performance variant increased from 40% to 70%. Understanding teachers’ Poor Non-Poor expectations of students is critical as they have been shown to influence student’s own expectations Note: Each decimal point can be interpreted as one month of schooling. and aspirations. The * indicates statistical significance at the 10% level. Policy Implications The use of teachers’ unconscious judgments towards and policies that replace negative associations with positive students’ scholastic aptitude and behavior based on their ones. Interventions that use positive role models can help perceived socio-economic background change people mindsets on what a child can affect a child’s lifelong academic One way to counteract from a poorer background can achieve trajectory. It is possible that by if given the opportunity, subsequently unconsciously inferring a student’s unconscious bias is changing pre-conceived ideas about performance using environmental cues to develop interventions intelligence. Similarly, creating a other than actual performance, teachers and policies that replace setting where teachers and students may inadvertently “penalize” students are exposed to positive examples of from poorer backgrounds by extending negative associations with academically successful adults, that pre-conceived notions of learning and positive ones. come from poor backgrounds, can outcomes that, while true on average, provide evidence for teachers and may not be true for the actual student at hand. students to shift their attitudes towards learning and their aspirations. Finally, interventions that promote empathy and Evidence suggests that one way to counteract perspective-taking can help override existing social norms unconscious bias is to develop interventions among teachers that the poor are less qualified. About eMBeD The Mind, Behavior, and Development Unit (eMBeD), the World Bank’s behavioral science team in the Poverty and Equity Global Practice, works closely with project teams, governments, and other partners to diagnose, design, and evaluate behaviorally informed interventions. By collaborating with a worldwide network of scientists and practitioners, the eMBeD team provides answers to important economic and social questions, and contributes to the global effort to eliminate poverty and enhance equity. Stay Connected eMBeD@worldbank.org #embed_wb worldbank.org/embed bit.ly/eMBeDNews Last Update: April 26, 2018