Association for Public Policy Analysis & Management

Black and Hispanic students in the United States are under-represented in advanced curriculum programs such as Advanced Placement (AP) courses at the high school level. Racial disparities in AP participation are more pronounced in courses involving science, technology, engineering and math (STEM). Differences in test scores or prior preparation do not fully explain their disparate participation levels when compared to white and Asian students. There are also gender differences in STEM course participation with female students less likely to participate than male students. Using an adaptation of an audit study, we examine whether high school counselors exhibit racial or gender biases when recommending students for AP Calculus courses. A sample of counselors are asked to evaluate student profiles that are identical except for the name on the profile and decide whether to recommend each student for advanced coursework.  The names on the profiles are varied randomly to have white or Asian sounding names, black or Hispanic sounding names, and male or female sounding names. Since the only differences in the profiles will be the gender or racial/ethnic origin of the names on top, we would expect no significant difference in the likelihood of these potential gatekeepers recommending a particular student profile for advanced coursework. Any differences can be attributed to bias. We also include a blind recommendation scenario as an experimental intervention designed to reduce the potential for bias.  Results have key policy implications for any program that asks gatekeepers to make subjective student recommendations that may be subject to bias, implicit or otherwise.