This paper by David J. Zimmerman from Williams College examines peer effects on academic outcomes using data from the college's housing system. The study employs a quasi-experimental approach, leveraging the random assignment of first-year students to roommates based on their preferences and housing forms. The author argues that this random assignment allows for unbiased measurement of the impact of roommates' academic characteristics on individual grades. Key findings include: 1. **Verbal SAT Scores**: Peer effects are more strongly linked to verbal SAT scores than math SAT scores. 2. **Middle SAT Students**: Students in the middle 70% of the SAT distribution may perform worse if they share a room with a roommate in the bottom 15% of the verbal SAT distribution. 3. **Top SAT Students**: Students in the top 15% of the SAT distribution are least affected by their peers' SAT scores. 4. **Statistical Significance**: While the effects are not large, they are statistically significant in many models. The study also considers the impact of "academic environments" within clusters of rooms and the effects of "academic advisors" associated with these social units. The results suggest that peer effects can influence academic performance, particularly in the context of residential housing at Williams College. However, the findings must be interpreted with caution, as they are based on a highly selective school environment.This paper by David J. Zimmerman from Williams College examines peer effects on academic outcomes using data from the college's housing system. The study employs a quasi-experimental approach, leveraging the random assignment of first-year students to roommates based on their preferences and housing forms. The author argues that this random assignment allows for unbiased measurement of the impact of roommates' academic characteristics on individual grades. Key findings include: 1. **Verbal SAT Scores**: Peer effects are more strongly linked to verbal SAT scores than math SAT scores. 2. **Middle SAT Students**: Students in the middle 70% of the SAT distribution may perform worse if they share a room with a roommate in the bottom 15% of the verbal SAT distribution. 3. **Top SAT Students**: Students in the top 15% of the SAT distribution are least affected by their peers' SAT scores. 4. **Statistical Significance**: While the effects are not large, they are statistically significant in many models. The study also considers the impact of "academic environments" within clusters of rooms and the effects of "academic advisors" associated with these social units. The results suggest that peer effects can influence academic performance, particularly in the context of residential housing at Williams College. However, the findings must be interpreted with caution, as they are based on a highly selective school environment.