Academic Peer Effects with Different Group Assignment Policies : Residential Tracking versus Random Assignment

Garlick, Robert

World Bank, Washington, DC

关键词: ACADEMIC OUTCOMES;    ACADEMIC PERFORMANCE;    ACADEMIC YEAR;    ADJUSTMENT;    ASSISTANT PROFESSOR;   

DOI  :  10.1596/1813-9450-6787
RP-ID  :  WPS6787

学科分类：社会科学、人文和艺术（综合）

来源: World Bank Open Knowledge Repository

【 摘 要 】

This paper studies the relative academicperformance of students tracked or randomly assigned toSouth African university dormitories. Tracked or streamedassignment creates dormitories where all students obtainedsimilar scores on high school graduation examinations.Random assignment creates dormitories that are approximatelyrepresentative of the population of students. Trackinglowers students' mean grades in their first year ofuniversity and increases the variance or inequality ofgrades. This result is driven by a large negative effect oftracking on low-scoring students' grades and anear-zero effect on high-scoring students' grades.Low-scoring students are more sensitive to changes in theirpeer group composition and their grades suffer if they liveonly with low-scoring peers. In this setting, residentialtracking has undesirable efficiency (lower mean) and equity(higher variance) effects. The result isolates a pure peereffect of tracking, whereas classroom tracking studiesidentify a combination of peer effects and differences inteacher behavior across tracked and untracked classrooms.The negative pure peer effect of residential trackingsuggests that classroom tracking may also have negativeeffects unless teachers are more effective in homogeneousclassrooms. Random variation in peer group composition underrandom dormitory assignment also generates peer effects.Living with higher-scoring peers increases students'grades and the effect is larger for low-scoring students.This is consistent with the aggregate effects of trackingrelative to random assignment. However, using peer effectsestimated in randomly assigned groups to predict outcomes intracked groups yields unreliable predictions. Thisillustrates a more general risk that peer effects estimatedunder one peer group assignment policy provide limitedinformation about how peer effects might work with adifferent peer group assignment policy.