1 Field Experiments in Class Size from the Early Twentieth Century

A vast majority of adults believe that class size reductions are a good way to improve the quality of public schools. For example, the Gallup Poll of the Public’s Attitudes Toward the Public Schools (an annual survey representative of the non-institutionalized U.S. population age 18 and over) asked respondents in 1999 what they would do to improve public schools if they could change one thing; reducing class size was the most prevalent choice among parents of public school children and the second most prevalent choice (after increased discipline) among individuals without children. Reviews of the research literature, on the other hand, have provided mixed messages on the degree to which class size matters for student achievement (for example, Glass and Smith 1978; Hanushek, 1986; Hedges, Laine, and Greenwald, 1994; and Krueger, 2003). The literature since World War II has followed two main tracks. A wave of research on class size from the 1950s into the 1990s typically used modern statistical methods to analyze large datasets containing information on a variety of educational inputs and outcomes. However, in the last decade or two, economists have focused on the difficulty of using observational data on class size to draw inferences about causal relationships. After all, class size is likely to be correlate