Despite the lack of rigorous research on the subject, reform proposals to reduce school size are quite popular these days. What does the best evidence say about the benefits of small schools? This event will feature Christopher Berry of the University of Chicago and Martin West of Harvard University, who will present their newly published findings on the relationship between school size and earnings later in life.
Grover Whitehurst, director of the Institute of Education Sciences, and David Ferrero, director of research and evaluation for education programs at the Bill and Melinda Gates Foundation, will join the presenters in a discussion about what we know, what we need to know, and what kind of research will answer key questions about the effects of school size. Frederick M. Hess, director of education policy studies at AEI, will moderate the discussion.
| 10:15 a.m. | Registration | |
| 10:30 | Introduction: | Frederick M. Hess, AEI |
| Presenters: | Christopher Berry, University of Chicago | |
| Martin West, Harvard University | ||
| Discussants: | David Ferrero, Bill and Melinda Gates Foundation | |
| Grover Whitehurst, Institute of Education Sciences | ||
| Noon | Adjournment | |
September 2004
Are Smaller Schools Better?
The idea of that small schools are better than large schools has been at the center of reform suggestions for the past several years and continues to enjoy strong support. While no one seems particularly opposed to the idea of small schools, the lack of rigorous research on what benefits can be specifically attributed to size cause the more skeptical to question the immediate application of policy. At a September 22 AEI conference hosted by Frederick M. Hess, director of education policy studies at AEI, Christopher Berry of the University of Chicago, and Martin West of Harvard University presented their recent report offering statistical data on the small-school issue. Grover Whitehurst, director of the Institute of Education Sciences, and David Ferrero, director of research and evaluation for education programs at the Bill and Melinda Gates Foundation, responded.
Christopher Berry
University of Chicago
Our research on school size and the returns to schooling takes a historical perspective and focuses on the middle of the twentieth century with the Consolidation Movement. Between 1930 and 1970, 115,000 school districts were eliminated; 130,000 schools closed; the average district size increased fourteen-fold; and the average school size increased five-fold. Approximately two-thirds of American schools closed in a reform effort aimed at giving all children the perceived benefits of a larger school. Along with the Consolidation Movement, there were also increases in both enrollment and population, adding to the size of schools. Concurrently, we also see state governments taking a more direct role in the effort and offering financial benefit to consolidating schools.
In evaluating data on school size, we should remember that current data is mostly taken after the 1970 increase and is plagued by a selection bias with parent choice influencing the comparative results from large and small schools.
Following the methodology of David Card and Alan Krueger (1992), we have analyzed three cohort groups of white men between the years of 1920-1949 who were educated between the years of 1926-1966. Controlling as many factors as possible, we have attempted to draw out the estimated returns and relate those changes to the Consolidation Movement. These results suggest that a hundred-student increase in school size leads to a 3.7-percent drop in earnings after high school graduation.
Martin West
Harvard University
In an effort to check these results we evaluated the criticisms that school size might be a mere proxy for other early environmental factors. However, this is unlikely given the then-current opinion that larger schools were better. Empirically, we have also attempted to control for average income and share of population in rural areas so that results were not adversely affected. In attempting to confirm that school size is an influential factor in attainment, we noted that a hundred-student increase in school size led to a 2.5-percent increase in drop-outs and a 2.5-percent decrease in college participation, while there was no noticeable change in college graduation rates.
Our results shed a positive light on small schools, but we would like to raise a few cautions for contemporary policymakers. First, this report does not address any data after the 1970s. Second, our results are based on state averages and are not directed towards advice for any particular school or location. Third, our report does not present a method for deducing why school size matters, but there are three possible reasons: 1) the over-bureaucratization of larger schools, 2) curricular changes that accompanied larger schools, and 3) the loss of a sense of community or even the increased transportation time. While this may not directly relate to contemporary policymakers, these general observations and findings suggest these cautioned skepticisms: 1) Conventional wisdom may be wrong. 2) We need to ensure rigorous evaluations of research before accepting policy conclusions. 3) We need to explain why school size matters. 4) And above all, there must be an emphasis on student outcomes--school reform is not an end in itself but only as it provides benefits to the students themselves.
Grover Whitehurst
Institute of Education Sciences
Although the data presented today does have an aggregate bias because it is research based on state-level data, there is still a perceived wisdom about small schools. This perceived wisdom that in small schools students are making more rapid progress toward graduation, behaving better, dropping out in lower numbers, and generally more satisfied than in larger ones leads one to the conclusion that smaller is better.
Equally important, however, is that there is a small but legitimate number of studies and reasoning that are inconsistent with this perceived wisdom.
1) In a 2003 dropout study conducted by Valerie Lee and David Burkam, the lowest percentage of students dropping out was found in medium-sized schools (600-1500 students.) Also, there was no significant difference between small schools (<600 students) and large schools (>1500 students) and their dropout percentages.
2) A California study of 127 small and large high schools showed that the large schools actually had better academic achievement in terms of higher SAT scores and number of students taking the SAT than their smaller peer schools.
3) The Department of Education randomized trial study of Career Academies-- schools within a larger school focused on different academic and career goals-- showed that although there was no change in dropout percentages in these programs, there was a correlation between these academies and earnings after graduation. This could indicate that the dependent variable of earnings after graduation is particularly sensitive.
4) A study by Craig Howley illustrates that as school size increased by adding middle-income students, achievement rates actually increase. The opposite was also found. When school size increased by adding low-income students, achievement rates dropped.
Again, although these studies are not consistent with the majority of school-size studies, they cannot be ignored. These studies also support the fact that there are deficiencies in the data that everyone uses for studies, and often we cannot adequately control selection bias. Studies in general, however, narrow the range of uncertainty about school size, and smaller schools still seem to have the advantage.
Even with the most up-to-date information and research that we have, there are still questions about small schools that need to be addressed:
1) How small should schools be? Can they be too small, and if they can, how small is too small?
2) What are the cost benefits? Are we getting enough returns from building and operating small schools?
3) What is the active ingredient? Is "small" defined purely on structure, or are we referring to what is offered inside of the smaller schools?
4) What is the counter-factual? We are going to see change when we research changes in smaller schools, but what do we actually want to compare this information to?
In conclusion, we need to look past structural data like class size, school size, and status as a charter school or single-sex institution because structure is a context for which learning occurs. We need to explore the process of what is happening inside these schools not just the contextual data.
David Ferrero
Bill and Melinda Gates Foundation
The small size of a school alone is not enough to automatically consider it to be a good school. In the United States today, a little over 50 percent of students attend schools with less than 1,000 total students, and many of these schools, just like many large schools, are mediocre at best. Also, many schools put out data on increased attendance records, or lower dropout rates, which make them appear to be breakout schools, when in reality the entire school as a whole may not be improving significantly.
What is important, however, is that small schools are a good starting point for reform, especially schools that are specifically designed to be small in size. These schools have strong organizational, structural and cultural strategies. These schools have a big complex package comprised of what needs to happen at all levels to make the institution more personalized, cozier, and capable of having alternative curriculums, many elective courses, and better management techniques. When schools are designed to be small and intimate environments there is greater choice and competition between schools and incentive to find uniquely qualified teachers. These schools act as a "Trojan horse" for reform because when we start out with a small school and open up the possibilities for change, many successful outcomes are possible.
One of the most interesting questions regarding small schools is: what is going on besides simply size? Research needs to be conducted on what things are being done at the policy and government levels to make a difference in the school, and what is happening inside the school and classrooms that make the small schools better.
Finally, after hearing the presentation by Mr. Berry and Mr. West, I have a few questions for future research on the success of smaller schools:
1) What can we infer, using the return numbers and attainment rates of small schools, about the quality of what is happening inside the schools to get the labor market outcome?
2) What are the cost efficiencies of the small schools, and are they worth the costs of transition and operation?
3) What level of confidence do we need in our research before we can implement policies and test them?
This summary was prepared by AEI interns Erin Riley and Melissa Silvers.
