Unlike Their Neighbors: Charter School Student Composition Across States
The national debate over charter schools has been fueled by two competing narratives about the kinds of students charters serve. Opponents claim charters unfairly select the most advantaged students, draining resources from traditional public schools and avoiding accountability. Proponents paint a different picture, claiming many charters purposefully serve the most disadvantaged students who have languished in failing public schools. Both characterizations have some merit, but neither accurately describes charter schools writ large. As cataloged in a previous report, Differences on Balance, charter schools nationwide display a variety of balanced differences in student composition, compared with their neighboring public schools.
However, charters are governed by the states, and some states’ charter schools display a much less balanced set of differences. In some states, charter schools look much more like their opponents’ characterizations, serving far fewer historically disadvantaged students than their neighbors. Other states have charter sectors that look like charter proponents often suggest, serving more disadvantaged students.
Several states are on the extremes of this spectrum, but most fall somewhere in between, reflecting the diversity of not only charter schools but also charter sectors across states. These individual reports profile each state’s charter sector to promote a more nuanced national portrait of charters and a more informed discussion of state charter policy.
National debate over charter schools has hit a fevered pitch this year. Teachers unions and many classical Democrats have opposed charters—with this wing successfully stiffening anti-charter language in the Democratic National Committee’s official platform. Even more extreme, the NAACP and the Movement for Black Lives Matter called for a complete national moratorium on charter schools.1
However, the charter debate does not cleave along a simple left/right divide. While those on the right tend to support charters, groups on the left fall on both sides of the issue. Teachers unions, traditional Democrats, and some civil rights organizations oppose them, but President Barack Obama and many civil rights organizations have been strong charter supporters. Other groups, including the Democrats for Education Reform (DFER), the Black Alliance for Educational Options (BAEO), and hundreds of civil rights leaders2, have specifically dismissed calls for a moratorium, with BAEO President Jacqueline Cooper calling the NAACP resolution "ill-conceived and based on lies and distortions about the work of charter schools.3
The divide is rooted in two competing narratives about what charters are and what students they serve. Opponents paint charter schools as "public-private" schools run by independent groups that seek to profit from public funds without accountability and that use various means to select the most advantaged students for their schools.4 Hillary Clinton illustrated that view last year when she said, "Most charter schools, they don’t take the hardest-to-teach kids, or, if they do, they don’t keep them."5
Supporters paint a very different picture. They tend to think of charters as hope-filled alternatives for historically disadvantaged students who have long been failed by traditional public schools (TPSs). As Shavar Jefferies of DFER argued, "In communities of color throughout our country, public charter schools are providing pathways to college and careers that previously were not available."6
A national conversation begs for a simple idea of what charters are and who they serve, and these two narratives are competing to define that idea. That competition has thrived in part because of limited evidence on charter student selectivity. Some evidence has had a national scope but used faulty comparisons between all charters and all TPSs. This is problematic because it ignores the fact that most TPS students have no charter choices. Other studies have used more nuanced methods but in small areas, which means that the results do not generalize to all charters.
I tried to bridge this divide in a recent report that compared the student composition of charter schools to that of the TPSs that neighbor them.7 This approach affords a viable look at student selectivity in all brick-and-mortar charters nationwide by removing the majority of TPSs—whose students do not have viable access to charter options—from the comparisons. By examining how often, how much, and in what directions charter schools’ students differed from those in neighboring TPSs, I found that, nationally, charters frequently differ from neighboring TPSs on many characteristics, but not in uniform ways.
Student poverty is a perfect example. Charters are often assumed to serve more poor students than TPSs. Compared to their neighboring TPSs, many charters do serve substantially more poor students; however, just as many serve fewer poor students, and both groups differ by comparable amounts.
Clarifying these national differences between charters and TPSs is worthwhile for a national debate. However, that clarity may be even more important at the state and local levels, where charter policies are articulated. Across the nation, charters differ from both their neighboring TPSs and other charter schools. Some of those differences are balanced nationally, but that is not the case in all states. Looking across states can also show how the charter sectors in one state can look very different from another.
In addition, charter schools are accountable to state and local authorizers, which granted their charters and can take them away. Clear evidence on how charter schools differ from their neighboring TPSs in a given state are vitally important when high-stakes decisions are being made, such as the upcoming voter referendum in Massachusetts that will decide whether the state cap on charter schools will be lifted.
This report provides that state-specific context. It compares charter and neighboring TPSs at the state level, using the same methods as the national report. The first section provides a brief summary of the methods used to identify neighboring TPSs and compare them to charter student populations and explains how to interpret the findings. The next section contains reports for each state that had at least 50 charter schools that could be matched to five neighboring TPSs.
The primary data for this report came from the 2011–12 Common Core of Data (CCD) from the National Center for Education Statistics (NCES), which includes information on school type, location, and the percentage of students by race and eligibility for reduced-price meals for every school in the nation.8 Data from the 2011–12 Civil Rights Data Collection from the US Department of Education’s Office of Civil Rights supplied percentages of students with disabilities and students who were English language learners, as well as data on out-of-school suspensions. EDFacts data from the NCES provided school student proficiency data.
Identifying Neighboring TPSs. I developed a straightforward means of identifying neighboring TPSs, which are non charter public schools whose students could have enrolled in a charter school, using three criteria: distance, jurisdiction, and grade range.
The first matching criterion was distance. Based on the assumption that students in the closest TPSs are the most likely to attend a given charter school, I included the closest five TPSs as the comparison for each charter.9 TPSs located more than 30 miles from a charter school were considered too far away to be reasonably close neighbors and were excluded.
The second criterion was the charter school’s relationship to the school district in which it is located. Charter schools authorized by a school district were matched only to TPSs in the same district. Those authorized by an entity other than the school district were allowed to match with any TPS in the state.
The third criterion was grade range. Neighboring TPSs were considered comparable only to a charter serving overlapping grade ranges. Matching grade ranges avoided inappropriate matches, such as between a high school and a nearby charter elementary school.10
Not all charter schools could be matched to five neighboring TPSs. Of about 5,700 charter schools in the CCD in 2011–1211, 890 were excluded because they were special-purpose schools, too small, or virtual charter schools12. About 4,800 charter schools (84 percent) were matched to at least one neighboring TPS, and 4,280 (89 percent) were matched to five13. In addition, Louisiana charters were not included in this analysis because all the states charter schools were identified as alternative schools on the CCD.14
Comparing Charters to Their Neighboring TPSs. Even with an ideal comparison group of TPSs, comparing schools’ average characteristics can be misleading. Comparing averages assumes charters are reasonably uniform, which national comparisons in the report Differences on Balance demonstrated to be false.15
The comparisons in these state reports look at the distribution of differences between each charter school and its neighboring TPSs. They show how much each charter school’s student composition differs from the average for its five neighboring TPSs and in what direction. The distributions of differences reveal how often, how much, and in what direction charter schools differ from TPSs in terms of the school’s suspension rates and students’ race, poverty, special education and limited English proficiency (LEP) status, and proficiency.
Individually, each report gives a clear and concise description of how the students served by that state’s charter sector differ from the students served by neighboring public schools. The differences among states are substantial. In some states, such as Ohio, the charter sector as a whole serves far more black and poor students and far fewer white students than the public schools located nearest to charters. The fact that Ohio charters serve a relatively higher proportion of historically disadvantaged students is reflected in their relative proficiency rates, which, although mixed, are predominantly lower than neighboring TPSs. In other states, such as North Carolina, the charter sector serves much smaller proportions of black, Hispanic, poor, and LEP students, and far more white students, than the schools located near them. Unsurprisingly, given these differences in student composition, far more North Carolina charters have markedly higher proficiency rates than their neighboring TPSs. Most states fall somewhere in between these extremes, and in most, charters defy a simple description, as more charters than TPSs serve both higher and lower percentages of multiple student types. A few, notably New York, both serve more historically disadvantaged students and have higher relative proficiency rates.
Collectively, these reports are more than the sum of their parts, and readers are encouraged to look at all the state profiles because they provide context for individual states. The entire report gives readers a comparative view of charter sectors across states, which shows that charter sectors, like charter schools themselves, are quite diverse. In total, they reflect variety in charter laws and regulations, diversity in charter authorizers and operators, and differences in the families that are attracted to charter schools across the states.
1. The NAACP call is pending on a vote from the national committee.
2. David Craig et al., letter to the NAACP board, September 21, 2016, http://www.publiccharters.org/wp-content/uploads/2016/09/ NAACPResponse_FINAL_9212016.pdf.
3. Democrats for Education Reform, “Statement from DFER President Shavar Jeffries on the NAACP’s Resolution Calling for a Moratorium on Charter Schools,” press release, August 3, 2016, http://dfer.org/statement-from-dfer-president-shavar-jeffries-on-thenaacps-resolution-calling-for-a-moratorium-on-charter-schools/.
4. Charles P. Pierce, “Don’t Believe the Charter School Hype,” Esquire, September 19, 2016, http://www.esquire.com/news-politics/ politics/news/a48717/massachusetts-charter-schools/; and K. G. Welner, “The Dirty Dozen: How Charter Schools Influence Student Enrollment,” Teachers College Record 17104 (2013).
5. Kimberly Hefling, “Hillary Clinton Rebukes Charter Schools,” Politico, November 9, 2015, http://www.politico.com/story/2015/11/ hillary-clinton-charter-schools-education-215661.
6. Jason Russell, “Black Leaders Push Back Against NAACP’s Charter School Moratorium,” Washington Examiner, August 5, 2016, http://www.washingtonexaminer.com/black-leaders-push-back-against-naacps-charter-school-moratorium/article/2598729.
7. Nat Malkus, Differences on Balance: National Comparisons of Charter and Traditional Public Schools, American Enterprise Institute, August 16, 2016, http://www.aei.org/publication/differences-on-balance-national-comparisons-of-charter-and-traditional-publicschools/.
8. Findings are based on universal administrative data, so observed differences are not tested for statistical significance and are assumed to be true.
9. Researchers have used a variety of geographic boundaries to compare charter and nearby traditional public schools, including states, school districts, Census tracts and block groups, and school attendance boundary areas. Generally, smaller geographic units more accurately reflect differences. E. Frankenberg and C. Lee, “Charter Schools and Race: A Lost Opportunity for Integrated Education,” Education Policy Analysis Archives 11, no. 32 (2003); C. Gulosino, “Circles of Influence: An Analysis of Charter School Location and Racial Patterns at Varying Geographic Scales,” Education Policy Analysis Archives 19, no. 8 (2011); and S. Saporito and D. Sohoni, “Coloring Outside the Lines: Racial Segregation in Public Schools and Their Attendance Boundaries,” Sociology of Education 79, no. 2 (2006): 81–105.
10. For a full description of the matching method, see Malkus, Differences on Balance.
11. All school counts have been rounded to the nearest 10 per data disclosure rules.
12. Schools not considered “regular” public schools on CCD were excluded, as were special education, vocational, or alternative schools. For a full description of the sample, see Malkus, Differences on Balance.
13. For a detailed description of this methodology, see Malkus, Differences on Balance.
14. On the original CCD data file, all Louisiana charter schools were identified as "Other/Alternative schools." Though this is likely due to identification errors, Louisiana charters were dropped from this analysis because regular schools could not be separated from alternative charter schools.
15. Malkus, Differences on Balance.