Are teachers biased against poor students? Experimental evidence from Peru

This study tests whether teachers use students’ income to evaluate their scholastic aptitude, behavior, and education potential, using experimental data from teachers in elementary schools representative of the public sector in Metropolitan Lima, Peru. The experimental design was adapted from a landmark study from psychology (Darley and Gross, 1983), in which subjects viewed a video of a child and teacher in a testing situation, where the child’s performance provided a very noisy signal of scholastic aptitude (the child correctly answered some difficult questions and incorrectly answered some easy questions). Prior to the testing video, subjects were randomly assigned to two different priming videos that showed the same child playing at home, either in a working class neighborhood or a middle class neighborhood. We augment this experimental set-up with another variant in which the child provides a very clear signal of high scholastic aptitude. We test to see whether teachers assigned to the different experimental variants rate the child differently with respect to their estimates of the child's grade level, educational potential, and behavior.