Angrist, Joshua D., & Lang, Kevin
Does School Integration Generate Peer Effects? Evidence from Boston’s Metco Program.
Massachusetts Institute of Technology
The impact of METCO (long-running desegregation program) on receiving districts.
Journal Name or Institutional Affiliation:
American Economic Review,
Vol. 94, No. 5, pp. 1613-1634
- Aggregate data on schools strongly suggest the impact of Metco long-running desegregation program is largely a composition effect.
- Micro data show some evidence of an effect of Metco on the scores of minority 3rd graders in Reading and Language.
- School level (District) results: Found a negative relationship between fraction METCO (fraction of students on desegregation program) and test scores. Found no association between fraction METCO and score for Whites.
- School Level (Brookline) results : No effect of METCO students on full sample of non- METCO students.
- No negative effect of increasing the fraction minority on most students.
- Negative effects on non-METCO minority are modest and short lived (Reading and Language Arts for Black third graders).
- Finds no peer effects of METCO program on non-METCO students.
- No minority composition effect on Non-Black, Non- METCO students. Minimal negative effects on minority Non-METCO students.
- Any peer effects from Metco are modest and short-lived.
Journal Article Empirical Research
Busing, Desegregation, Funding, Language, Peer Effects, Reading
Secondary Survey Data
Method of Analysis:
Two Stage Least Squares
Schools in Metco Receiving district
Unit of Analysis:
School, School District, Student
- Aggregate school-level data from the statewide Massachusetts Comprehensive Assessment System testing program.
- Micro level data from Brookline Public Schools (Iowa Test on Basic Skills).
- The METCO program is a voluntary integration program funded by the Commonwealth of Massachusetts through the Racial Imbalance Act.
- Achievement is measures using the ITBS for 3rd, 5th and 7th graders. This analysis uses test scores reported as the national percentile rank, which measures achievement relative to the score distribution in a 1992 reference population.
- OLS regressions
- DV: Average score in the grade/school/year cell
- IV: Class size in cell , fraction Metco, student characteristics
- Instrumental Variables model
- Uses the number of Metco students in a class as instrumental variable.
- DV: Average score in the grade/school year cell
- IV: Number of Metco students , number of non-Metco students, total enrollment