Association for Public Policy Analysis & Management

School choice policies, particularly charter schools, have been a contentious strategy for school reform for many decades. Despite controversies on the efficacy of these policies and their effect on student segregation, these policies continue to grow in popularity. For example, the charter sector now serves almost 3 million students (Center for Education Reform, 2016).

While past studies have used now dated data to investigate concerns of selective enrollment (Booker et al., 2005; Garcia et al., 2008; Zimmer et al., 2009; Zimmer & Guarino, 2013; Winters, 2015; Welsh et al., 2016), none have moved beyond the scope of large statewide generalizations to examining particular types of schools. For instance, while most studies have not found large statewide evidence consistent with these selective enrollment strategies, many acknowledge that these actions may occur in smaller locales.

This study uses North Carolina as a setting to study student enrollment patterns in charter schools over a seven-year period. In response to Race to the Top, NC removed its charter school cap of 100 total schools in 2011. The number of charter schools has increased by 50% since 2011. In this paper, we use a statewide, student-level longitudinal dataset spanning 2008-09 to 2014-15 to estimate a variety of linear probability models to investigate the student enrollment and exit patterns of charter schools in North Carolina, paying special attention to concerns of selective enrollment, both cream-skimming of high performing students and pushing out low-performing students. Further, we test whether these effects are moderated by important student characteristics.

Our preliminary findings indicate that charter schools are no more likely to push out low-performing students than traditional public schools, and this effect is not moderated by low income status nor race or vary by region types (urban, suburban, town, and rural).

The results of our preliminary analyses on enrollment of high-performing students provides a more interesting story. We define high-performance four ways for each student: above district average, above previous school’s average, above current school’s average, and at least proficient on the state exam. We find no evidence of cream skimming across all definitions except for high-performance defined by above previous school’s average. When we use high performance based on previous school’s ability distribution, we find that high achieving students are more likely to attend a charter school in the next school year (approximately 2 to 3 percent points).

Our preliminary results suggest that this movement is driven by student preferences, not differential charter school admission practices. We find that these results seem to be most prevalent in suburban, town, and rural locales, and of greater concern, that across all definitions of student performance, high-performing black students are about two to three percentage points less likely to enter charter schools.