Ability Signals and Rigorous Coursework: Evidence from AP Calculus Participation

Authors: Christopher Avery, Harvard University, Joshua Goodman, Boston University

Project Summary

This study examines how one signal of ability—whether a student achieves “Advanced” status on Massachusetts’ 10th grade statewide standardized assessment—effects subsequent enrollment in Advanced Placement (AP) Calculus courses. The findings suggest that ability signals can positively influence choices around student enrollment, either by changing students’ and families’ course choices or by changing teachers’ and counselors’ decisions about whom to assign to such courses.

Key Findings

This study compares future course-taking and performance for students that score just above and below the top achievement category (Advanced) on Massachusetts’ 10th grade math test. Students just above and below this threshold have similar demographics and mathematical skill; their only difference is whether they receive this positive signal. The authors find:

  • There are large gaps in AP Calculus course-taking by race/ethnicity and by income.
  • For Black and Hispanic students, receiving an Advanced score substantially increases their likelihood of enrollment in AP Calculus; the enrollment of White and Asian students is unaffected by this signal.
  • The enrollment effects are even larger for Black and Hispanic students with low reading scores.
  • The Advanced threshold may be set too low.

Implications and Recommendations

That the Advanced signal has little effect on AP Calculus course-taking for the average Massachusetts high school student but meaningfully increases enrollment rates for Black or Hispanic students has important policy implications for state, district, and school leaders. First, as a system, the results suggest that more attention should be paid to the positive or negative signals sent to students about their ability. The results also suggest that states and districts should consider devising signals to students and their educators that are clearer and more predictive of success towards a specific outcome. At the very least, these signals could serve as a check on biases that may exist in the current enrollment and placement processes.