Skip Navigation
Funding Opportunities | Search Funded Research Grants and Contracts

IES Grant

Title: Do Small Schools Improve Student Performance in Large Urban Districts? Evidence from New York City
Center: NCER Year: 2008
Principal Investigator: Stiefel, Leanna Awardee: New York University
Program: Improving Education Systems      [Program Details]
Award Period: 2 years Award Amount: $482,584
Type: Exploration Award Number: R305A080522
Description:

Co-Principal Investigator: Amy Ellen Schwartz

Purpose: This project examines the potential impacts of small high schools, compared to their larger counterparts, on high school outcomes for students overall and across subgroups of students. The analysis also examines characteristics of small and large schools to better understand the relationships between school size and outcomes, and disentangle the possible effect of size from that of other school characteristics associated with size. The small school reform stands out among high school reform programs because of its adoption in many major cities and its substantial public and philanthropic funding base.

Project: A New York City Department of Education dataset containing information on two cohorts of high school students will be analyzed using quasi-experimental techniques to address two research questions. First, does attending a small high school (as compared to a large one) appear to improve student outcomes in high school, particularly for at-risk students? This question has several subparts: Do small high schools deliver better high school outcomes than large ones? Is there evidence that small high schools deliver better college outcomes than larger ones in areas such as applications, matriculation, and GPA? Second, what may explain any findings regarding the first question? How do conditions within high schools of different sizes vary? What hypothesis can be generated about what is "inside the black box" of school size?

Products: The expected outcomes of this research included published journal articles and policy briefs on the performance of small high schools.

Setting: The researchers use a New York University, Institute for Education and Social Policy (IESP) dataset made available by the New York City Department of Education on New York City public school high school students for two cohorts of students; those expected to graduate in 2001 and in 2002, including student and school data from 1997–2001 and from 1998–2002, respectively.

Population: Of the 71,130 high school students in the dataset, 7.6% of students attend small schools, about 72% of students are black or Hispanic, less than 25% are foreign-born, about 60.8% speak English at home, 80.8% of students are eligible for free lunch versus 71.6% in large high schools, 7.5% of students receive ESL services versus 6.0% in large high schools, only 12.3% of students are foreign-born versus 22.1% in large high schools, incoming students in small versus large high schools enter with substantially lower average test scores in middle school, and more incoming students enter with low and middle-to-low eighth-grade math and reading scores.

Intervention: Attending a "small" versus "large" high school.

Research Design and Methods: For the first research question, determining whether there appear to be impacts of small schools on student performance, the researchers (1) perform propensity score matching and (2) establish an instrumental variable estimation using distance to school in an attempt to obtain causal estimates. In addition, they (1) explore non-linearity in the relationship between size and outcomes, (2) assess whether the apparent effects of being in a small school are the same for all students or different for certain groups, and (3) capture the apparent effect of being in small schools on college outcomes.

For the second research question, exploring explanations for the findings, they examine the correlations between school size and other school characteristics (e.g., pupil-teacher ratio, characteristics of the teacher population) using two different models: (1) a regression model to estimate the relationship between these characteristics and school size and (2) an HLM model that treats the intercepts and slopes of key variables as outcomes in the first-level model, introducing school size and other school characteristics in the second-level model.

Control Condition: The researchers compare students in small versus large high schools, using various measures of small and large, and using propensity score matching and instrumental variables to control for nonrandom selection by size.

Key Measures: Measures include 4-year high school graduation rates (52.1% of students in the sample graduate within 4 years, 16.5% dropout, 28.1% continued to be enrolled after 4 years, and 3.3% earned a GED), and college outcomes, including applications, matriculation, and GPA.

Data Analytic Strategy: Propensity-score matching and instrumental variables will be used to address selection bias. The analysis will be done using regression, including multi-level models with high school or college outcomes as the dependent variables.

Publications

Journal article, monograph, or newsletter

Iatarola, P., Schwartz, A.E., Stiefel, L., and Chellman, C. (2008). Small Schools, Large Districts: Small-School Reform and New York City's Students. Teachers College Record, 110 (9): 1837–1878.

Schwartz, A.E., Stiefel, L., and Wiswall, M. (2013). Do Small Schools Improve Performance in Large, Urban Districts? Causal Evidence From New York City. Journal of Urban Economics, 77 : 27–40.

** This project was submitted to and funded under Middle and High School Reform in FY 2008.


Back