A common critique of charter schools is that the publicly funded but privately operated schools of choice systematically “push out” students who require specialized services or are struggling academically. As Adam Kho, Andrew McEachin, and Ron Zimmer recently discussed on this site, prior studies have failed to find evidence that such students are more likely to exit from charter schools than from traditional public schools. However, while informative, these studies have at least two limitations. First, they do not use research designs capable of distinguishing the causal effect of attending a charter school on student mobility from the effect of pre-existing differences between students who enroll in the charter or traditional public school sector. Second, they do not speak to whether any across-sector differences in student attrition are specific to charter schooling or rather a product of students attending a school they chose instead of a school to which they were assigned.
In a new study released by the Wheelock Educational Policy Center, we use data from Newark, New Jersey to address both of these limitations. Our results represent the clearest evidence to date contradicting the common claim that charter schools systematically push out hard-to-serve students.
To carry out this study, we take advantage of Newark’s use of a universal enrollment system that assigns students to all traditional public and the large majority of charter schools (from here, “participating schools”). The centralized system offers students seats according to an algorithm that takes into account the student’s rank-ordered school assignment preferences, priorities that schools have for certain students (e.g., sibling at the school, student lives in a particular neighborhood), and a randomized lottery number used for breaking ties between students with similar preferences and priority status. Our analyses rely on our ability to observe school placements and students’ rank-ordered school assignment preferences revealed within the school assignment process.
We apply a recently developed strategy that uses the randomized component of the enrollment system to estimate the causal effect of enrolling in a charter school on later student outcomes. As in a more conventional random assignment study, this research design compares the later outcomes of students who enrolled in a charter school to students who enrolled in a traditional public school who were equally likely to be offered a charter school seat but were not because of how the random lottery numbers happened to fall. The study looks at 13,868 students who participated in the city’s universal enrollment system over a span of four years, 2014-15 through to 2017-18.
We find that attending a participating Newark charter school substantially reduces student mobility. On average, students were about 22.4 percentage points less likely to leave their school in the next two years if they enrolled in a participating charter than if they had enrolled in a traditional public school.
A complicating factor when interpreting this result is that Newark’s charters are in such high demand that most students attending one are also attending one of their most preferred schools. That is, perhaps the effect we have unveiled is not specific to attending a charter school but rather also occurs for students who enroll in a traditional public school they preferred. Our ability to observe students’ school preferences allows us to investigate this issue in a novel way. In particular, when we add to the regression controls for the student’s rank order of the school, the estimated effect of enrolling in a charter school is cut by more than half but remains a statistically significant 9.6 percentage points. Thus, both charter schooling and enrolling in a preferred school independently reduce student attrition.
We find that enrolling in a charter school not only decreases student mobility for the average charter entrant but also for students within subgroups that charters are often accused of pushing out. Enrolling in a charter school rather than a traditional public school causes a 10.8 percentage point reduction in the probability that students with a disability leave their school within the next two years. English language learners are about 16.3 percentage points less likely to exit their school if they enroll in a charter than if they had enrolled in a traditional public school. Enrolling in a charter school doesn’t significantly impact the mobility of Hispanic students. And the effect of enrolling in a charter school on student mobility does not depend on the student’s test scores the year before they entered the school.
Finally, we take the opportunity to dig deeper into the question of why the percentage of students within charter schools who receive special education services is lower than in traditional public schools—a common pattern in cities like Newark with a sizable charter presence. In particular, using our causal research design we show that enrolling in a participating charter school increases the likelihood that a student with a disability is later declassified out of special education, but it does not impact the probability that a student is newly identified for special education services. These results mirror those from Boston reported in a recent paper by Tuft’s Elizabeth Setren that was also highlighted in Education Next. The results provide further evidence suggesting that this enrollment gap between charter and traditional public schools is in part explained by differences in how the sectors classify students.
Our study of Newark’s charter sector provides new insights and challenges some common beliefs about the factors that drive student mobility and demographic differences between charter and traditional public schools. To be clear, our results do not exclude the possibility that charter schools have inappropriately excluded students in some specific cases. Rather, our finding that, on average, enrolling in a charter school reduces the likelihood that a student exits their school strongly suggests that such behavior is not systematic within participating charter schools in Newark. And the fact that this result holds also for students with disabilities and English learners implies that charter schools do not systematically push out such students who enroll. Rather than focusing on push-out, policymakers seeking to reduce enrollment gaps between charter and traditional public schools should instead look for ways to increase demand for charter schools within targeted subpopulations.
Marcus A. Winters is an associate professor in Boston University Wheelock College of Education & Human Development. He is also faculty director of the Wheelock Educational Policy Center. Allison Gilmour is an assistant professor at Temple University College of Education and Human Development. Colin Shanks is a PhD candidate in economics at Boston University.