Convincing evidence that home buyers pay a premium to live in particular school attendance zones (see Black (1999)) and the emphasis on education in the national policy debate strongly supports the notion that parents and the public at large believe both that schools make an important difference and that there is substantial variation in school quality. However, the identification of actual differences in school quality has proven to be a more difficult task. The fact that families choose where to live and where to send their children to school impedes the separation of school and family effects on outcomes. Other factors including cost of living differences and the likely existence of compensating differentials (e.g., better working conditions) for teachers add additional complications. Moreover, a finding that specific, measurable school inputs including expenditures are not strongly related to student outcomes does not imply that other aspects of schools are not important.
Several recent papers have taken a different approach to the identification of school effects on achievement by examining within-school variation in teacher quality. Both Sanders and Horn (1994) and Rivkin, Hanushek, and Kain (1998) document substantial variation in teacher quality within schools. Such evidence provides a lower bound estimate of the overall importance of schools and teachers, because it ignores between-school differences in both teacher quality and other factors.
The results from Rivkin, Hanushek and Kain (1998), in particular, provide powerful evidence in support of an important role for teachers and schools. The magnitude of the teacher effects is striking. Consider first the impact on students of moving up the mathematics teacher quality distribution by one standard deviation. This is roughly equivalent to lining up all teachers from least to most effective and passing one third of the teachers. The impact of moving up one standard deviation in teacher quality raises 5th and 6th grade mathematics test scores by at least twice as much (and probably much more) as the very expensive policy of reducing average class size by five students.
The comparison of teacher effects and of family differences provides another perspective. The gain from having a very good teacher (one standard deviation better) rather than an average teacher for five years in a row is at least as large as the average mathematics test score differential between lower income students eligible for a subsidized lunch and higher income students not so eligible. This finding that schools can exert an effect similar in size to family income contrasts sharply with research on education that focuses on specific characteristics such as expenditure, class size, or teacher education.
These results support the notion that schools exert a very important influence on academic development, an influence much more similar in magnitude to the popular perception of the importance of schools. Importantly, this conclusion emerges only following the relaxation of the view of schools as monolithic institutions whose quality is determined by the salaries paid to teachers, the gleam of the laboratory, the size of classrooms, and the availability of the latest computers. While these factors do exert both a direct effect on students and an indirect effect via making the school more attractive to prospective teachers, it is only consideration of the substantial variation in teacher quality within school buildings that leads to the finding that the quality of school instruction is a primary determinant of academic achievement. Importantly, for policy purposes, a key element is that variations in teacher quality occur among teachers who look the same in terms of degree earned, experience, and class size.