The Peer Effect in Education
In 2014, the US spent an average of $16,268 a year per student, well above the global* average. Since education is universal and compulsory at lower levels of schooling, spending may be influenced by a country’s ability to pay. However the US spends a larger proportion of its gross domestic product on education as well, with 6.2% of GDP going towards primary to tertiary education (5.2% global* average). However this may not be translating into better results for US students. One of the biggest cross-national tests, the Programme for International Student Achievement (PISA), ranked the US 38th out of 71 countries in math and 24th in science. Additionally, in the Trends in Mathematics and Science Study, US fourth graders ranked 11th in math and 8th in science out of 48 countries, while eighth graders ranked 8th in both math and science, out of 37 countries.
To help explain how the US is being outpaced in educational outputs despite above average spending, it is important to understand the inputs of education. Education inputs include the number and quality of teachers, school facilities, teaching materials and supplies. However one input in particular is harder to understand and measure, the effect of peers. If a child’s peers impact outcomes into adulthood, than the importance of student composition can not be overstated.
A paper published in the American Economic Journal by Bifulco, Fletcher and Ross reports that higher percentage of high school classmates with college educated mothers decreases the likelihood of dropping out and increases college attendance. To estimate their effects, they used a set of controls designed to break systemic differences in the quality of students in a specific cohort, along with a placebo test that measured the student composition for a student’s actual cohort against the same measures for a randomly selected younger or older cohort from the same school. They estimate that a one percentage point increase in the percent of students whose parents are college educated is associated with a decrease in the likelihood of dropping out of about 0.3 percentage points and an increase the likelihood of attending college between 0.4 and 0.5 percentage points.
A paper published in 2018 in the American Economic Review by Carrell, Hoekstra and Kuka also seeks to identify the peer effect but goes further to estimate the effects in the long run. In the paper, they focused on the negative effects of “disruptive” peers. They characterize students whose families are affected by domestic violence as “disruptive”. Previous research by Carell and Hoekstra has shown that exposure to these peers significantly disrupts achievement and behavior. Secondly, exposure to domestic violence is unrelated to the students’ classmates, an important improvement for causal identification compared to the Bifulco, Fletcher and Ross paper. To distinguish the peer effect from other factors including variation in educational resources, they focus on variation within schools.
Their results show that exposure to disruptive peers has important long run consequences for both educational attainment and earnings into adulthood. They found that exposure to one additional disruptive student in a class of 25 throughout elementary school reduces math and reading scores in grades 9 and 10 along with declines in college enrollment. They also found that exposure to an additional disruptive peer led to a 3 percent reduction in earnings from ages 24 to 28. These findings correspond to the same change in earnings as about one half of a standard deviation in teacher quality and imply that one year of exposure to a disruptive peer in elementary school reduces the present discount value of classmates’ future earnings by around $80,000.
The findings from the Bifulco and Carrell papers mentioned above estimate robust effects of classroom composition on both academic achievement and long run earnings. The sum of these findings suggests that there is significant social benefit waiting for policy that could diminish the impact of disruptive peers and maximize the positive effects of student composition on learning.
Note: * OECD data is only available for 34 OECD member countries, so the global* average represents the average for member countries. These are: Australia, Austria, Belgium, Canada, Chile, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Israel, Italy, Japan, Korea, Luxembourg, Mexico, the Netherlands, New Zealand, Norway, Poland, Portugal, Slovak Republic, Slovenia, Spain, Sweden, Switzerland, Turkey, the United Kingdom, and the United States.
SOURCES:
National Bureau of Economic Research
The Effect of Classmate Characteristics on Individual Outcomes
The Long Run Effects of Classmate Characteristics on Individual Outcomes