Close Close Comment Creative Commons Donate Email Add Email Facebook Instagram Mastodon Facebook Messenger Mobile Nav Menu Podcast Print RSS Search Secure Twitter WhatsApp YouTube
Protect Fearless Journalism Summer Member Drive Deadline: MIDNIGHT
Donate Now

Methodology on Our Educational Opportunity Project: About the Data and Our Analysis

ProPublica’s analysis tracks from the most comprehensive data set of access to advanced classes and special programs in U.S. public schools—known as the Civil Rights Data Set released by the U.S. Department of Education Office of Civil Rights. As of January 24, 2013, our analysis includes new figures: AP test pass rates and participation in sports programs. You can search the data for yourself in our interactive feature. 

The office is charged with making sure that all students have access to educational programs. The data track access to several programs, including AP, gifted and talented programs and advanced math and science classes. Those factors are aggregated by race, disability status, gender and English proficiency. The data also track whether teachers at schools are certified and how much experience they have.

The data in our analysis were gathered for all schools in districts with at least 3,000 students. The data also track district characteristics such as whether it is under a desegregation plan or whether it has anti-bullying policies.

The data were reported by schools and districts to the Office of Civil Rights. ProPublica spent several weeks verifying the accuracy of the data. Where we were able, we corrected extreme outliers and contacted hundreds of schools to verify their data. Because of some of the problems we found in the initial data, Assistant Secretary for Civil Rights Russlynn Ali said that the office is revamping its process for gathering and verifying their data. We also vetted our analysis with education research experts.

We may not have accounted for every problem in the data and welcome feedback from schools and districts.

Using a statistical tool called linear regression, ProPublica analyzed the data to see if schools with certain student compositions were more or less likely to have students enrolled in those programs. We did not include alternative, magnet or charter schools in our data because they often offer specialized programs and do not draw from the general population. We also included only schools with at least 20 students, to eliminate problems inherent to small populations.

While we found some relationship between the proportion of minority students at schools and access to programs, we found the strongest relationship with the percent of students getting free- reduced-price lunches—a variable often used in education research to estimate poverty at schools.  We found a similar relationship between poverty and AP pass rates. Because the Office of Civil Rights does not regulate access to programs by low-income students, we obtained data on free- or reduced-price lunches from the National Center for Education Statistics.

If the programs and courses in this data were available to all students, there would be no relationship between poverty and the proportion of students in those programs. But our analysis found that in some states, more than half of the change in enrollment in certain programs can be explained by an increase in the percentage of poor students. That means that higher- poverty schools do not have the same access to programs.

For example, in Oklahoma, for every 10-point increase in the percent of low-income students, the percentage of students enrolled in AP drops by three points, on average.

In Massachusetts, for every 1-point increase in the percent of low-income students, the AP pass rate drops by six points.

In many states with little relationship between advanced programs and poverty, specific policies had been put in place to require widespread access to those programs.

For more information about our analysis, please contact Jennifer LaFleur, the Director of Computer-Assisted Reporting.

Latest Stories from ProPublica

Current site Current page