Association for Public Policy Analysis & Management

Joe Waddington

Ron Zimmer

University of Kentucky

Mark Berends

University of Notre Dame

A pervasive question in the school choice literature is whether schools of choice cream-skim students by selecting higher achieving, non-disadvantaged students in the enrollment process. Similarly, schools of choice may “push out” lower achieving, disadvantaged students. The answers to these questions can provide important insights into the accessibility of choice programs for disadvantaged student populations.

Previous research has focused on the presence of cream-skimming or push out within charter schools (Booker, Zimmer, & Buddin, 2005; Zimmer et al., 2009; Zimmer & Guarino, 2013). Researchers have also examined educational attainment and income levels of parents participating in voucher programs relative to students eligible for voucher programs (Peterson 1998; Howell & Peterson 2000; Wolf, Howell & Peterson 2000), though these findings do not shed light on whether participating private schools are cream-skimming the best students from public schools.

Indiana’s private school voucher program has grown into the nation’s largest statewide school voucher program, enrolling 34,299 students as of the 2016-17 school year. Researchers have begun to unpack the early impacts of the program on student achievement (Waddington & Berends, 2017). However, this research has not specifically examined questions of accessibility. In this paper, we examine whether there is evidence consistent with the claim that private schools participating in the Indiana voucher program cream-skim advantaged students or pushout disadvantaged students. We also explore the variation across private schools (e.g., Catholic vs. non-Catholic) and compare to the cream-skimming and pushout rates within traditional public, charter, and magnet schools.

To examine these questions, we use longitudinal student-level data from the Indiana Department of Education. These data contain information on voucher and non-voucher private school students as well as all public school students from 2010-2016 in grades 3-8. Initially, students in Indiana could only receive a voucher if they were previously enrolled in a public school. Therefore, we examine whether students exiting are above average students relative to students who remain in the traditional public school they exit. Similarly, we examine whether students who exit private schools are below average students. In both cases, we define above average and below average students based on test scores and discipline infractions.

For our analysis, we use a linear probability model to predict the likelihood of a student using a voucher to make a move into or out of a private school. We account for grade-by-year fixed effects, time-varying student characteristics, and a series of indicators for whether a student is from a disadvantaged group or is above- or below-average achieving relative to the traditional public and private school they attended.

Our preliminary results are not consistent with the claim of private schools cream skimming the best students from traditional public schools. However, we do find modest evidence consistent with the claim that private schools are pushing out low performing students as below average students are more likely to exit a private school than above average students.