Thirty years ago, Minnesota became the first state to allow charter schools. The idea was simple: give educators freedom from bureaucracy, hold them accountable for results, and let innovation flourish. Today, more than 3.7 million students attend over 8,000 charter schools across 47 states[s]. The question of charter school performance has generated enormous political heat but surprisingly little honest analysis of what the data actually shows.
The honest answer, after three decades of research, is this: charter schools are neither the salvation nor the catastrophe their loudest advocates claim. The most comprehensive charter school performance data we have reveals a messy, complicated picture that should make everyone uncomfortable.
What the Numbers Actually Say About Charter School Performance
Stanford University’s Center for Research on Education Outcomes (CREDOCenter for Research on Education Outcomes at Stanford University, which conducts the most comprehensive national studies on charter school performance.) has conducted the most rigorous national studies of charter schools. Their 2023 analysis covered 2 million students across 29 states. The headline finding: charter school students gained an additional 16 days of learning in reading and 6 days in math compared to similar students in traditional public schools[s].
Charter advocates trumpet this as vindication. But context matters enormously. Those gains translate to moving from the 50th percentile to the 50.4th percentile in math and the 51st percentile in reading[s]. That’s statistically significant but practically modest. It’s not nothing, but it’s not a revolution either.
More importantly, these are averages that mask enormous variation. Only 36% of charter schools outperform their local traditional public schools. Another 47% perform about the same. And 17% perform worse[s]. Roughly one in six charter schools is actively failing its students compared to the alternative.
The Improvement Story
One genuine bright spot: charters have gotten better over time. When CREDO released its first national study in 2009, the results were grim. Margaret Raymond, CREDO’s director, described the findings in blunt terms: “It was the pits”[s]. Charter schools were underperforming traditional public schools in both reading and math.
Over the following 15 years, reading growth in charter schools rose by 23 additional days per year, and math growth increased by 37 days[s]. The sector learned from its failures. But that learning came at the cost of millions of students who attended underperforming schools while the sector figured things out.
Who Benefits, Who Loses
Charter school performance varies dramatically by student population. Black students in charter schools gained the equivalent of 40 extra days of reading instruction per year compared to their peers in traditional schools[s]. Hispanic students and students from low-income families also showed stronger growth. Urban charter schools posted the best results: 29 additional days of growth in reading and 28 in math[s].
But students with disabilities tell a different story. Special education students had significantly weaker growth in both math and reading at charter schools[s]. Traditional public schools, with their mandated services and established support systems, served these students better.
Virtual charter schools represent an outright disaster. Students at online charters lost the equivalent of 58 days of reading instruction and 124 days of math, roughly two-thirds of a school year[s]. These schools have expanded substantially since the pandemic, which should alarm everyone.
The Closure Problem
More than one in four charter schools closes within five years. By year 20, the failure rate reaches 55%[s]. Low enrollment is the primary reason, followed by fraud or mismanagement[s].
Charter advocates argue this is the system working as intended. Schools that fail to attract students or produce results get shut down, unlike traditional public schools that persist regardless of performance. There’s merit to this argument. But it ignores the students caught in the crossfire when their school suddenly closes.
What Should Change
The charter school performance debate needs to grow up. Both sides are cherry-picking data to support predetermined conclusions. The honest position is uncomfortable: charters work well for some students in some contexts, fail others, and aren’t solving the systemic problems in American education.
We need better authorizerGovernment agency or educational body responsible for approving, overseeing, and potentially closing charter schools based on their performance and compliance. oversight to close the 17% of charters that underperform. We need virtual charter schools shut down or radically reformed. We need honest accounting of why special education students fare worse. And we need to stop pretending that school choice alone will fix educational inequality.
The data supports neither dismantling charter schools nor expanding them indiscriminately. It supports doing the hard work of figuring out what actually helps students learn, school by school, context by context. That’s less satisfying than a bumper sticker, but it’s what the evidence demands.
In 1991, Minnesota enacted the first charter school law in American history[s]. The premise was deceptively simple: public schools freed from bureaucratic constraints, held accountable through performance contracts, would innovate their way to better outcomes. Three decades later, charter school performance data allows us to evaluate that premise with unusual precision. The results should satisfy neither charter advocates nor their critics.
The most comprehensive longitudinal charter school performance research comes from Stanford’s Center for Research on Education Outcomes (CREDOCenter for Research on Education Outcomes at Stanford University, which conducts the most comprehensive national studies on charter school performance.). Their 2023 National Charter School Study analyzed data from over 2 million students across 29 states and the District of Columbia from 2014 to 2019[s]. The topline finding: charter school students, on average, gained an additional 16 days of learning in reading and 6 days in math compared to matched peers in traditional public schools.
The Magnitude Problem in Charter School Performance Research
These findings require careful interpretation. Translated to percentile terms, the gains move an average student from the 50th percentile to the 50.4th in math and the 51st in reading[s]. Ron Zimmer, a charter school researcher at the University of Kentucky, characterized these effects as “modest.” By conventional research standards, they represent small effect sizesStatistical measure that quantifies the magnitude of difference between groups, indicating whether research findings are practically meaningful beyond statistical significance..
More significant than the average is the distribution. CREDO found that 36% of charter schools produced stronger academic growth than their local traditional public school comparison, 47% produced equivalent growth, and 17% produced weaker growth[s]. This variance matters: a parent choosing a charter school faces roughly one-in-six odds of placing their child in a school that performs worse than the default option.
Methodological Considerations
CREDO’s “virtual twin” methodology matches charter students with composite students in traditional public schools sharing similar demographics, prior test scores, and free/reduced lunch status. Critics note that this approach cannot account for unobservable factors like parental motivation or engagement[s]. Families who actively choose charter schools may differ systematically from families who accept default assignments.
Alternative methodologies using lottery-based randomization, where charter admission is determined by chance among applicants, generally produce similar results. This convergence across methods strengthens confidence in the basic finding that charter schools produce small positive average effects.
Longitudinal Trends in Charter School Performance
The trajectory matters as much as the current snapshot. CREDO’s 2009 study found charter schools underperforming traditional public schools in both subjects. By 2013, reading results had turned positive while math remained negative. The 2023 study shows positive results in both, with reading growth increasing by 23 days per year and math growth by 37 days over the 15-year study period[s].
CREDO attributes this improvement to existing schools getting better rather than new high-performing schools entering the sector. The mechanism appears to be the combination of operational flexibility and accountability pressure inherent in the charter model. Schools that failed to improve faced closure or loss of enrollment.
Demographic Heterogeneity
Aggregate charter school performance data obscures significant variation across student populations. Black students gained 40 additional days of reading growth annually; Hispanic students and students in poverty also showed stronger effects[s]. White students showed no reading advantage and slightly weaker math growth in charter schools.
Students with disabilities represent a consistent exception to positive findings. Special education students demonstrated significantly weaker growth in both subjects[s]. Traditional public schools’ established compliance infrastructure and mandated services appear to serve this population more effectively than charter schools’ varied approaches.
Geographic variation is equally pronounced. Urban charter schools produced 29 additional days of reading growth and 28 days of math growth. Rural charter schools showed 10 fewer days of math learning than traditional public school peers[s]. The charter model appears optimized for urban contexts with sufficient population density to support school choice.
Virtual Charter Schools: A Categorical Failure
Online charter schools, comprising approximately 6% of charter enrollment, produced dramatically negative results: 58 fewer days of reading growth and 124 fewer days of math growth compared to traditional public school peers[s]. This represents loss of roughly one-third of a school year in reading and two-thirds in math. Virtual charter expansion following the COVID-19 pandemic raises serious policy concerns.
Organizational Sustainability
Charter school closure rates provide another lens on sector performance. Longitudinal analysis shows that more than 25% of charter schools close within five years; by year 20, cumulative closure rates reach 55%[s]. Low enrollment accounts for nearly half of closures, with fraud or mismanagement as the second most common cause[s].
Proponents frame high closure rates as accountability in action. The counterargument: each closure disrupts students who must transition mid-education, and the high failure rate suggests inadequate quality control at authorization.
Resource Allocation
Charter schools operate with approximately 30% less funding per pupil than traditional public schools, averaging $7,147 less during the 2019-20 school year[s]. Differences in student need, particularly lower enrollment of students with disabilities requiring costly services, explain roughly 70% of this gap. Local funding disparities account for most of the remainder.
Segregation Effects
The Urban Institute’s national analysis found that charter school expansion modestly increases racial segregation within school districts. Eliminating charter schools would reduce segregation by approximately 5% in the average district[s]. The effect is small but consistent: parental choice patterns tend toward racial homogeneity.
Segregative effects concentrate in urban districts with high minority populations and suburban districts with low minority representation[s]. Charter schools have not delivered on early hopes that choice would naturally produce integration.
Policy Implications
The evidence supports neither wholesale expansion nor elimination of charter schools. It supports differentiated policy: stronger authorizerGovernment agency or educational body responsible for approving, overseeing, and potentially closing charter schools based on their performance and compliance. oversight to close underperforming schools faster, categorical restrictions or elimination of virtual charters, targeted support for charter schools serving populations that benefit most (urban, low-income, Black and Hispanic students), and improved services for students with disabilities.
The charter experiment has produced genuine innovations and served many students well. It has also failed many others, increased segregation modestly, and created instability through high closure rates. Responsible policy requires acknowledging all of these realities simultaneously, which neither side in the political debate seems willing to do.
Charter school performance data tells us that context matters more than ideology. The question is not whether charter schools work, but which charter schools work for which students under what conditions. That’s a harder question to answer, but it’s the only one worth asking.
