We examined whether larger
networks are more effective than
smaller ones and found that, both with and without correcting for student and peer socioeconomic characteristics and selection bias, students at schools that are part of
networks of three or more schools consistently
outperform students at schools in
networks of only two schools.