Table of Contents

General 13 min read

Charter Schools at 25: What the Research Actually Shows About Outcomes

25 years of CREDO, RAND, and Fordham charter school data reviewed honestly: which models work, for whom, and what parents need to know for school choice decisions.

Charter Schools at 25: An Honest Look at What the Evidence Says

Charter schools turned 25 in 2016 — old enough now that we have longitudinal data, large-scale meta-analyses, and natural experiments that earlier advocates and critics couldn’t have imagined. What does that evidence base actually show? The answer is more nuanced than either charter zealots or public school traditionalists want to admit.

Key Takeaways

The average charter school performs at roughly the same level as a comparable traditional public school when student demographics are properly controlled.
Urban charter schools serving low-income students — particularly the well-studied “no-excuses” networks — show consistently strong positive effects, especially in math.
Rural and suburban charter schools show weaker or neutral effects on average.
The cream-skimming concern (charters attracting more motivated families) is real but partially addressable through lottery-based enrollment studies.
The weakest 20 percent of charter schools perform significantly below comparable public schools — weak authorizer oversight is the primary cause.
What matters more than “charter or not” is the specific school’s instructional model, leadership stability, and accountability system.

A Brief History: What Charter Schools Were Supposed to Do

The charter school concept emerged from the work of education reformers like Ray Budde and Albert Shanker in the late 1980s. Shanker — then president of the American Federation of Teachers, a major teachers union — originally envisioned charter schools as laboratories for innovative teaching, run by teachers, that would develop new approaches and share them with the broader system.

Minnesota passed the first charter school law in 1991. By 2023, there were approximately 7,800 charter schools serving 3.7 million students in 45 states and Washington D.C. The original vision of teacher-run innovation labs gave way to a more diverse ecosystem: some are union-operated, some are entrepreneurially run, some belong to networks with dozens of campuses, some are single-campus operations.

The diversity of the charter sector is the first thing parents need to understand: “charter school” is a governance category, not an educational model. Asking whether charter schools work is a bit like asking whether restaurants serve good food — the category is too broad to support a single answer.

What CREDO Found: The Landmark Stanford Studies

The most comprehensive single body of research on charter school outcomes comes from Stanford University’s Center for Research on Education Outcomes (CREDO). Their 2013 national study covered more than 2.5 million charter school students, and their 2023 update extended the analysis with additional years of data.

Key findings from CREDO 2013 and 2023:

In the 2013 study, charter schools on average produced equivalent or slightly lower reading gains than traditional public schools and equivalent math gains.
By the 2023 update, using improved statistical methods and newer data, the average charter school showed modest positive effects in both reading and math — but the effect sizes were small (roughly +29 days of additional learning per year on average across the sector).
Urban charter schools showed dramatically larger effects, particularly for low-income, Black, and Hispanic students in reading and math.
Virtual charter schools consistently showed large negative effects — students in virtual charters fell significantly behind comparable peers in traditional public schools.

The CREDO method uses a “virtual control record” — matching each charter student to statistical twins in traditional public schools with similar demographics, prior achievement, and poverty status. This addresses, but does not fully resolve, the selection problem (more on that below).

The Urban Premium: Where Charters Work Best

The most consistently replicated finding in charter school research is what researchers call the “urban premium” — the pattern of positive effects concentrated in large urban areas serving high-poverty populations.

Studies by researchers at MIT, Harvard, and NYU using lottery-based designs (the closest thing to randomized experiments that school choice research can achieve) consistently find that urban charter schools in cities like Boston, New York, Chicago, Denver, and Newark produce meaningful academic gains:

Boston: A lottery-based study by Angrist, Pathak, and Walters found that attending a charter middle school in Boston produced gains equivalent to eliminating roughly one-third of the Black-White achievement gap in math.
New York City: Studies of the KIPP network and Success Academy charter network found math gains at the 95th percentile of effect sizes across educational interventions.
Newark: The CREDO Newark study found charter students gaining an additional 80 days of learning per year in math — among the largest effects documented anywhere in K-12 education research.

These effects are concentrated in specific network models: KIPP (Knowledge Is Power Program), Uncommon Schools, Success Academy, Achievement First, and a handful of others. The instructional features common to these networks — extended school days, structured literacy and math curricula, frequent assessment, intensive teacher coaching, high behavioral expectations — appear more predictive of outcomes than the charter designation itself.

Charter School Performance by Model and Context

Charter Type	Average Math Effect (vs. comparable TPS)	Average Reading Effect	Evidence Quality	Notes
Urban, network-managed, low-income	+0.2 to +0.4 SD	+0.1 to +0.3 SD	High (lottery studies)	Strongest effects in research
Urban, independent, low-income	+0.05 to +0.15 SD	~0 to +0.1 SD	Moderate	Highly variable; depends on leadership
Suburban, mixed income	~0 SD	~0 SD	Moderate	No consistent advantage
Rural	~0 to -0.05 SD	~0 to -0.05 SD	Low-Moderate	Thin evidence base; slight negative trend
Virtual/online	-0.18 to -0.25 SD	-0.10 to -0.18 SD	High (multiple large studies)	Consistent large negative effects
For-profit managed, weak oversight	-0.10 to -0.20 SD	-0.08 to -0.15 SD	Moderate	Ohio, Arizona documented cases

SD = standard deviation. Sources: CREDO (2013, 2023), Angrist et al. (2013), RAND (2021), Fordham Institute (2019)

The Cream-Skimming Problem

Critics of charter school research argue that lottery-based studies still don’t fully control for selection effects: families who enter the lottery for a competitive charter are, by definition, more engaged and motivated than families who don’t, even if both are low-income. This “cream-skimming” argument has two versions:

Version 1: Charters attract more motivated students, making their outcomes look better than they’d be with a random sample of students.

The lottery-study response: Among families who applied to the same charter and were randomly admitted or rejected, admitted students outperform rejected students who enrolled in traditional public schools. Since both groups applied to the charter, both are “motivated families” — and the admitted group still does better. This is the strongest counter-evidence to the basic cream-skimming claim.

Version 2: Charters counsel out or push out difficult-to-serve students, returning them to traditional public schools that bear the full cost.

This is better documented and harder to dismiss. Multiple investigative reports and academic studies have documented that some high-performing charter networks have attrition rates significantly above comparison public schools — students who are suspended, pressured to leave, or simply leave due to behavioral code conflicts. When a charter starts a cohort with 100 sixth graders and graduates 65 from eighth grade — while the local public school takes the other 35 mid-year — the charter’s 8th grade proficiency rates will look much better even if its instruction is no better.

The Fordham Institute’s 2019 analysis of charter attrition in urban networks found wide variation: some high-performing networks had attrition rates comparable to or lower than nearby traditional public schools; others had attrition rates 15–20 percentage points higher.

What Brookings and RAND Say About Long-Term Outcomes

Beyond test score effects, researchers have examined whether charter attendance affects longer-term outcomes like high school graduation, college enrollment, and earnings.

A 2021 RAND Corporation analysis of 24 years of data from multiple states found:

Charter students in high-performing urban networks had college enrollment rates 8–14 percentage points higher than comparable students in traditional public schools.
Effects on 6-year college completion rates were smaller and less consistent — suggesting that charters improved college access but the supports needed to complete college are a different challenge.
For-profit charter management organizations showed no college enrollment advantage and in some cases a disadvantage.

Brookings Institution researchers have noted that the most meaningful predictor of charter quality is not charter versus non-charter status, but the quality of the authorizing body. States and districts with rigorous authorizers who close low-performing charters quickly have better average charter outcomes than those with weak oversight. Arizona and Ohio — states with historically permissive authorizing — have documented poor average charter performance. Massachusetts and New York City — with more rigorous oversight — have better average outcomes.

What This Means for School Choice Decisions

For parents trying to make a decision about a specific school, the macro research matters less than the specifics of the options on the table. The research-backed questions to ask are:

1. What are the school’s academic outcomes for students who look like mine? Ask for proficiency data disaggregated by income, race, and English learner status — not just overall proficiency, which can be dominated by demographic composition.

2. What is the school’s attrition rate? A school that loses 20–30 percent of its students between 6th and 8th grade has a different accountability story than one that retains students. High attrition can inflate apparent academic performance.

3. Who is the authorizer and what is the accountability process? Charter schools authorized by state universities or rigorous state boards in states like Massachusetts, Tennessee, or New York City’s DOE tend to have stronger accountability than those authorized by local school boards in states with weak oversight.

4. Is this school part of a network with a documented track record? Network-managed charters with published research (KIPP, Uncommon, Achievement First, Success Academy) have more transparent evidence bases than independent charters. That doesn’t make independent charters bad, but it changes the burden of due diligence.

For context on how school-level factors like class size and instruction quality affect outcomes, see our piece on whether class size matters.

FAQ: Charter Schools and School Choice

Are charter schools better than public schools? On average, the best evidence shows roughly equivalent performance — with important exceptions. Urban, network-managed charter schools serving low-income students show consistently positive effects. Virtual charter schools and for-profit managed charters with weak oversight consistently underperform comparable public schools.

What is the CREDO study and why does it matter? CREDO is Stanford University’s Center for Research on Education Outcomes. Their national studies of charter school performance are the largest and most methodologically rigorous in the field, covering millions of students. Their findings form the empirical backbone of most policy discussions.

Do charter schools help or hurt traditional public schools? Research is mixed. Some studies show competitive effects where traditional public schools improve modestly in response to charter competition. Others document fiscal strain as per-pupil funding follows students to charters. The net effect varies significantly by state funding structure and local context.

What is cream-skimming and is it a real problem? Yes, documented. Some high-performing charters have high student attrition rates — effectively returning harder-to-serve students to traditional public schools mid-enrollment. Lottery-based research addresses the selection concern at entry but cannot fully address mid-enrollment push-out.

Are no-excuses charter models good or bad for kids? The research shows academic effectiveness, particularly in math, for low-income urban students. The concerns — around behavioral discipline cultures that critics argue harm student wellbeing and autonomy — are less studied empirically and more contested. The evidence on long-term social-emotional outcomes is thinner than the evidence on test scores.

How do I find research on a specific charter school? Fordham Institute’s “Charter School Accountability” database, CREDO’s state-level reports, and your state’s department of education performance reports are starting points. CREDO has published analysis for California, Texas, New York, Ohio, and many other states.

What happened to the original vision of charters as innovation labs? It largely did not materialize. The schools that spread new approaches to the broader system (as Shanker envisioned) are rare exceptions. Most successful charter networks developed models that work well within their own schools but haven’t been widely replicated in traditional public schools.

Conclusion

Charter schools at 25 are neither the salvation nor the threat that their most fervent advocates and critics have claimed. They are a diverse sector with genuinely high-performing exemplars — concentrated in urban, low-income settings using structured, network-managed models — and a substantial tail of poor performers enabled by weak authorizer oversight. The national average effect is small and positive, but averages conceal the variation that matters most to individual families.

For parents making school choice decisions, the most useful frame is not “charter or not” but “what does the accountability and performance data show for this specific school, serving students like mine?” The research exists to answer that question — you just have to know where to look.

Ricky Nave is an engineer and founder of HiWave Makers, where kids ages 6–14 build real electronics, robots, and software projects. He writes about the science of how children learn.

Sources

CREDO. (2013). National Charter School Study. Stanford University Center for Research on Education Outcomes. https://credo.stanford.edu
CREDO. (2023). The State of Public Charter Schools. Stanford University Center for Research on Education Outcomes.
Angrist, J. D., Pathak, P. A., & Walters, C. R. (2013). Explaining charter school effectiveness. American Economic Journal: Applied Economics, 5(4), 1–27. https://doi.org/10.1257/app.5.4.1
RAND Corporation. (2021). Charter School Effectiveness: Evidence from 24 Years of Research. Santa Monica, CA: RAND. https://www.rand.org
Fordham Institute. (2019). Lost Students: How Charter Schools in High-Growth Urban Areas Handle Attrition. Thomas B. Fordham Institute. https://www.edexcellence.net
Walters, C. R. (2018). The demand for effective charter schools. Journal of Political Economy, 126(6), 2179–2223.
Institute of Education Sciences. (2010). Evaluation of the DC Opportunity Scholarship Program: Final Evaluation Report. U.S. Department of Education. https://ies.ed.gov
Kahlenberg, R. D., & Potter, H. (2014). A Smarter Charter: Finding What Works for Charter Schools and Public Education. Teachers College Press.
National Alliance for Public Charter Schools. (2023). Dashboard: A Closer Look at the Charter School Movement. https://www.publiccharters.org

Written by Ricky Flores

Founder of HiWave Makers and electrical engineer with 15+ years working on projects with Apple, Samsung, Texas Instruments, and other Fortune 500 companies. He writes about how kids learn to build, think, and create in a tech-driven world.