A recent release from the Education Commission of the States reminds us that the term “virtual school” refers to several different types of educational options, and that the ecosystem—more important now than ever before—requires specific attention and support from policymakers.
The four main types of virtual schools are charter schools, single-district schools, multi-district schools, and state schools. Governance structures vary across the typologies, with a variety of agencies and actors responsible for oversight and implementation, including state education agencies and state boards of education, charter school authorizers, local education agencies, and third-party providers.
Charter schools constitute nearly half of all full-time virtual schools in the U.S. and have the largest enrollment share. As of January 2020, more than twenty states permitted virtual charters (with more potentially on the way). Their operation and governance are generally akin to brick-and-mortar charters, with a mix of nonprofit and for-profit management organizations. Single- and multi-district virtual schools are rarer and clearly different from charters. Virginia and Tennessee, for example, permit a district or group of districts to contract with entities that meet state standards to serve as the online provider while the districts maintain oversight. The Florida Virtual School (FVS) is the prime example of a state-sponsored virtual school. It is run by a governor-appointed board with oversight from the state department of education. And the pandemic pivot to remote learning has allowed it to expand far beyond the Sunshine State.
A quick rundown of recent research on virtual education leads with the bad news that a raft of studies have shown that students in virtual charters experience weaker academic growth, increased mobility, and lower graduation rates than their brick-and-mortar peers. Specifically, virtual schools that focus on independent study and asynchronous instruction with limited student engagement and teacher contact time, high student-teacher ratios, and a reliance on family support for learning all experience poorer academic outcomes. However, evidence also suggests that course content and student motivation matter greatly. A Michigan study showed that part-time virtual students—those supplementing in-person learning with specific online content—fared better than their full-time virtual counterparts, while a study of FVS found far stronger positive outcomes for non-traditional students working on credit recovery versus traditional, first-time course takers.
The guide ends on a reasonably positive note by providing evidence that state-level policy levers are an effective means by which to implement improvements to virtual schools. State-level standards for school—and in the case of charter virtual schools, for their authorizers, too—can support research-driven initiatives, such as differentiated instruction and student-teacher contact time; flexible definitions of attendance and progress monitoring; teacher training and licensure requirements designed specifically for remote instructors; and funding mechanisms that prioritize virtual-specific needs, such as one-to-one student technology access, in-person testing, and at-home enrichment activities.
No state is pulling out all of these various stops, but best practices from across the country are described in the guide. Prior to the plague, virtual schools were expanding in a slow and organic way. The policy infrastructure to properly support them was, perhaps, lagging behind, but the right moves were being made in that regard. While the recent explosion of virtual schooling was involuntary, exponential expansion is inevitable, even after the Covid danger has subsided. Now is the time to make sure we understand how virtual schooling works best and make sure states guarantee the highest functionality of virtual options for all families who want it.
The state should force Boston Public Schools into a receivership, as it has “poorly served Black and brown students.” —Boston Globe
For two years, America has harmed children by miscalculating risk and sacrificing their experience and growth to assuage adult fears. This must stop. —The Atlantic
This is how the research explains charter schools’ success with traditionally disadvantaged students. —Greg Richmond
Jeers
Ignoring the harm that disruptive and violent students inflict on their peers, this article focuses only on disciplined students missing school. —The Nation
“[Rhode Island] Governor McKee’s office says ‘School Choice Week’ proclamation issued in error.” The governor supports charter schools but not publicly-funded private school vouchers. —WPRI
Ironically, the ACLU is opposing bills that promote government transparency in what curricula states and districts are using. —ACLU
This month’s sudden switch to remote learning is troubling news for kindergarteners. —Cincinnati Enquirer
Rick Hess interviews Doug Lemov, author of Teach Like a Champion, on what wisdom he has to share for teachers in 2022. —Education Week
A study from the University of Pennsylvania finds that “cash aid to poor mothers increases brain activity in babies.” —New York Times
“How it feels to be an Asian student in an elite public school.” —New York Times
For the first time in two decades, graduation rates dip as the pandemic slows seniors down. —Chalkbeat
GreatSchools’s popular rating system is biased and often inaccurate, according to research by leading economists. Tracking student-level growth is the better way to rate schools. —Chalkbeat
It is, purportedly, a new day in the three Ohio school districts still nominally overseen by Academic Distress Commissions. But as you read this utterly ridiculous description of the recent shenanigans surrounding Youngstown City Schools’ elected board, you will see that despite a couple of new names and a slightly different administration dynamic, things seem very much back to the bad old days up there. (Vindy.com, 1/26/22) That’s right: the same old accusations of secret phone calls, bad motives, power plays, and the utter desecration of Robert’s Rules of… Quick! Forget all that! Look over here! (Mahoning Matters, 1/26/22)
Speaking of traditional school districts going off the rails (were we?), the treasurer of Findlay City Schools will be hosting a four-part series of presentations for the public collectively titled “School Finance 101”. It is supposed to be about “transparency” so the punters will know more about how the money flows in their district. Madam Treasurer has, it seems, done this dog and pony show before in lots of different districts but it always seems to have followed the same general pattern. I would love to know what the official agenda is for the session on charter schools (in person event; no Zoom - sorry), but all I can find from the past is a single slide. Lots of things have changed in terms of charter school funding in Ohio in the last six months. Hope the presentation is clear on that score. (Findlay Courier, 1/24/22)
Crain’s Cleveland is the first – and so far only – media outlet to note that Ohio jumped from 24th to 12th in this year’s NAPCS ranking of state charter school laws. Hopefully other outlets will celebrate National School Choice Week with this news, and acknowledge the recent policy changes enacted by the General Assembly and Governor DeWine that lead to the jump. But I’m not holding my breath. (Crain’s Cleveland Business, 1/25/22)
Here at last is a proper celebration of National School Choice Week in Ohio, brief though it may be: the story of an early childhood education specialist in Columbus who not only helps other parents find the best educational fit for their children, but who has also recently embraced school choice for her own children. Huzzah! (10 TV News, Columbus, 1/26/22)
And finally today, going back to the topic of educational policy changes, let’s talk about Ohio’s Dyslexia Law. HB 436 was enacted back in January 2021, mandating universal dyslexia screenings starting in Kindergarten. The first year of implementation will also include kids in grades 1-3 just to make sure everyone who needs to be caught early is caught. Nice! That first implementation, however, does not start until next school year – which seems to me like a ton of lead time. But since the Ohio Department of Education (and the state board) need to get all the rules and policies together before then to ensure screenings, referrals, and service provision are handled properly, affected can get the interventions they need, perhaps 18 months or so is prudent. According to this piece, an 11-member committee has been working on the implementation guidebook for the last six months and has nearly 400 public comments to sift through before moving on to whatever is the next stop on the policy-implementation conveyor belt. (IdeaStream Public Media, Cleveland, 1/24/22)
Did you know you can have every edition of Gadfly Bites sent directly to your Inbox? Subscribe by clicking here.
In the wake of the biggest education crisis in living memory, the need for transformational change is palpable and urgent. Accordingly, this report takes a fresh look at a question that is fundamental to the goals of many education reformers: Can a rising tide of charter schools carry students in America's largest metro areas—including those in traditional public schools—before it? And if so, how far?
To address these questions, Fordham's associate director of research, David Griffith, analyzed a decade of data on reading and math achievement at the metropolitan level, as well as nearly two decades of data on charter and traditional public school enrollment. The results are summarized in three findings:
1. On average, an increase in total charter school enrollment share is associated with a significant increase in the average math achievement of poor, Black, and Hispanic students, which is concentrated in larger metro areas.
2. On average, increases in Black and Hispanic charter school enrollment share are associated with sizable increases in the average math achievement of these student groups, especially in larger metro areas.
3. On average, an increase in total charter school enrollment share is associated with a significant narrowing of a metro's racial and socioeconomic math achievement gaps.
For a more detailed account of the study’s methods, the full results, and the implications for policy, read the full report below. For the full report and Technical Appendix, click “DOWNLOAD PDF.”
FOREWORD
By Amber M. Northern and Michael J. Petrilli
The education reform engine known as results-based accountability—which was sputtering in the pre-pandemic period—has now all but stalled out. In contrast, the engine known as school choice is firing on all cylinders. Across the country, private school choice programs are proliferating, charters are burgeoning, new models that defy traditional classification are being invented, and a nontrivial percentage of families is leaving the traditional public school system—perhaps for good.
Renewed demand for more effectual alternatives to traditional school districts is one reason that billionaire Michael Bloomberg has doubled down on his investment in urban charter schools. That demand—and his promise of additional supply—is warranted in light of rigorous studies that show attending an urban charter school is associated with faster progress in reading and math, greater odds of enrolling and persisting in an institution of higher learning, and voting—plus a lower likelihood of being incarcerated.
Nationally, charters account for an ever-growing share of total public school enrollment, and they remain the most consequential school choice offering in most metropolitan areas, with 240,000 U.S. students newly enrolled since the start of the pandemic.[1]
Yet much remains poorly understood about the wider implications of charter growth. After all, most estimates of charters’ effects on students are driven by schools that serve a subset of a community’s students—or of a racial or socioeconomic minority—whose families have self-selected into this alternative form of public education. And, like any other promising education innovation, charters may struggle to replicate the successes of early exemplars as they expand to serve a larger and/or more representative population of students and staff. We don’t know whether charter schools can close achievement gaps at scale or whether urban charters can serve as a rising tide that lifts all boats in large metro areas, including student performance in traditional public schools.
Of course we need to be cautious about “gap closing,” for this can also happen when the achievement of higher-scoring kids takes a nosedive. Our preoccupation with learning gaps has also narrowed the scope of schooling and thwarted their transformation. Still, given the mountains of research showing that students who fall behind early have a hard time catching up, it’s worth asking whether education’s greatest innovation has improved the success rates of all children.
As readers may recall, a 2019 study conducted by Fordham’s associate director of research David Griffith found that an increase in the share of Black and Hispanic students who enrolled in charter schools was indeed correlated with a district-wide increase in Black and Hispanic students’ reading and math achievement. That important finding led the Wall Street Journal editorial board to declare that “Charter Schools Ace Another Test.”
The story doesn’t end there. Good reasons exist to keep asking about the impact of growing charter school enrollments on the achievement of all students, whether they attend charters or traditional (district-operated) public schools. First, the data continue to improve. Second, the boundaries that one uses to define a charter “market” could make a big difference. For example, the Phoenix metropolitan area contains at least thirty school districts, and students move relatively freely among them,[2] so any analysis that is limited to individual school districts risks missing the forest for the trees. The metro area is where the market for better schools ultimately operates.
In addition, we recently published an analysis (and interactive website) examining school quality in the nation’s larger metro areas. That study addressed which metro areas were performing well in the pre-pandemic era, not why. One possible explanation is that metros with more charter school attendees perform better. A third purpose of the present study is to test that hypothesis.
So, with David on board again, we decided to revisit in greater depth and with better data the impacts of charter school enrollment share at the metropolitan level.
Our new sample includes 400 metropolitan statistical areas and 534 micropolitan statistical areas—all of which are referred to as “metros” in the report and most of which include not only the urban cores that are the stuff of “I heart New York” postcards but also the surrounding suburbs and exurbs.
Here’s what the new study found:
First, an increase in total charter school enrollment share is associated with a significant increase in the average math achievement of low-income, Black, and Hispanic students, especially in larger metro areas.
This is a big deal. After all, the country’s hundred largest metro areas enroll two thirds of its K–12 pupils, so any policy that works for poor students and students of color in “large” metros has game-changing potential. And because charter schools’ share of total enrollment averages 7 percent in these places, even a conservative reading of the results suggests that most large metro areas would benefit from an increase in charter school enrollment.
Second, increases in Black and Hispanic charter school enrollment share are associated with sizable increases in the average math achievement of these student groups, especially (again) in larger metro areas.
In the average “large” metro area, roughly one in ten poor kids, one in eight Black students, and one in fifteen Hispanic students is currently enrolled in a charter school. To repeat, policies that allow those proportions to rise could yield significant gains for these student groups.
Some of the report’s most compelling results are for Hispanic students, consistent with a 2015 CREDO study that found Hispanic youngsters enrolled in urban charters gained twenty-two days of math learning per year.[3] Yet despite their seemingly transformative outcomes, predominantly Hispanic charters are woefully understudied. To us, this suggests that more attention should be paid to better understanding charters’ successes with Hispanic students.
Third, an increase in total charter school enrollment share is associated with a significant narrowing of a metro area’s racial and socioeconomic math achievement gaps.
It’s possible this result is partly due to declines in the achievement of higher income and/or white students (though the estimates for these groups aren’t statistically significant). That wouldn’t be good. Still, critics who allege that charters exacerbate educational inequality for low-income students and students of color will find no support for that allegation in this report. Rather, the findings suggest that large increases in charter school enrollment share could yield similarly large reductions in longstanding racial and socioeconomic math achievement gaps.
***
It’s not hard to connect the dots: the United States is reeling from a pandemic that has widened and deepened achievement gaps that were already pernicious while depressing the achievement of most students. Getting more children into charter schools could help reverse those dire trends.
INTRODUCTION
This study builds on a 2019 Fordham Institute report that examined the relationship between charter school enrollment share—that is, the share of students in a community who enroll in a charter school—and the average achievement of all the students in that community, including those in traditional public schools.
Like its predecessor, this report seeks to understand the systemic effects of charter schools as well as the potential for diminishing returns as their enrollment share increases. However, unlike the first report, which focused on school districts, this one focuses on metropolitan statistical areas (MSAs), which are an order of magnitude larger and typically encompass multiple school districts (see How is this study different from Fordham’s first Rising Tide report?).
As noted in the Foreword, the broader context—a global pandemic that has scarred the educational experiences of students in every corner of the United States—makes the quest for academic accelerants unusually pressing. But of course, policymakers needn’t start from scratch. Nor should they, if making the most of the current window of opportunity is the goal.
Accordingly, we are taking this opportunity to revisit one of the central questions of education reform: Can a rising tide of charter schools carry all students—including those in traditional public schools—before it? And if so, how far?
Specifically, this study seeks answers to the following research questions:
1. How does an increase in total charter school enrollment share—that is, the percentage of students who enroll in charter schools—affect a metro area’s average reading and math achievement? Are these effects bigger for certain student groups?
2. How do increases in the percentages of Black, White, and Hispanic students who enroll in charter schools affect the average reading and math achievement of each of these student groups?
3. How do increases in charter school enrollment share affect the size of a metro area’s racial and socioeconomic achievement gaps in reading and math?
The report is organized as follows: Background discusses prior research on charter schools and student achievement. Data and methods describes the data, the sample, and the methods. Findings presents the findings. And Takeaways discusses the implications for education policy and practice.
How is this study different from Fordham’s first Rising Tide report?
Although the overall approach is similar, this analysis differs from the first Rising Tide report in several ways.
First, it examines the effect of charter school enrollment share at the metropolitan level rather than the school district level, thus dramatically increasing the student population and physical size of the average unit. For example, Figure 1 shows the enormous difference between the physical boundaries of the San Francisco Unified School District and those of the San Francisco–Oakland–Fremont MSA.
Figure 1. The difference between a school district and a metro area
Source: GoogleMaps
This change in unit size has several consequences for the analysis: First, because charter schools are disproportionately concentrated in urban centers, it reduces the average level of and change in charter enrollment share that is observed within units during the study period. For example, Figure 2 shows charter enrollment share in the Los Angeles, Detroit, and Orleans Parish school districts in 2017–18 alongside charter enrollment share in the surrounding MSAs.
Figure 2. Charter school enrollment share at the district and metro levels
Notes: Charts are based on the authors’ own estimates of charter school enrollment share, which were generated by linking school-by-grade-by-year-level data on charter and traditional public school enrollment from the National Center for Education Statistics to host school districts and MSAs as defined by the Stanford Education Data Archive using the SEDA 4.1 crosswalk.
Because the average metro area is far larger than the average school district, focusing on metros also reduces the number of “district switchers” who cross jurisdictional boundaries to attend charter schools, thus mitigating their impact on estimates of the charter school effect (see Limitations). Furthermore, switching from districts to metro areas reduces the number of observations in—and thus, in some specifications, the weight that is assigned to—states with smaller school districts (e.g., Arizona and Michigan), while increasing the weight that is assigned to states with larger districts (e.g., Florida and Georgia).
In addition to those changes, this report improves on its predecessor by estimating the effect of average charter school enrollment share in every grade level between Kindergarten and the grade level in which an assessment was administered, rather than the effects of charter enrollment share in that grade level exclusively.[4] Finally, instead of restricting the sample to geographic units with zero to 50 percent charter school enrollment share, this report presents estimates for the full range of charter school enrollment share. As discussed in the Methods section, the downside of this approach is that it allows a small minority of metro areas to dictate the shape of some of the graphs; however, it has the virtue of transparency.
BACKGROUND
Numerous studies have found that enrolling in urban charter schools boosts the academic achievement of low-income, Black, and/or Hispanic students.[5] Other research has found that charter schools’ effects on the achievement of students in traditional public schools in their vicinity are neutral to positive.[6] Together, these literatures imply that the equilibrium effects of charter schools—that is, their effects on all students’ average achievement after accounting for whatever spillover effects are associated with charter school enrollment—are also positive, at least in major urban areas and for the student groups in question.[7] Yet direct evidence on this point is limited, and even observers who believe that charters are having a positive impact acknowledge that we know little about the extent to which returns diminish (or increase) as charter schools’ enrollment share increases within communities—that is, whether a community should expect the same benefits from its fiftieth charter that it derives from its fifth.
To date, only two studies have addressed that last question: The first, a 2019 Fordham report entitled Rising Tide: Charter School Market Share and Student Achievement found a positive relationship between the percentage of Black and Hispanic students who enrolled in a charter school at the geographic school district level and the average achievement of students in these groups—at least in the largest urban districts (see How do this report’s findings compare to those of the first Rising Tide report?).[8] The second study, by Chen and Harris, used similar data but different methods. It found a positive relationship between the percentage of “all students” who enrolled in charter schools and the average achievement of all publicly enrolled students, especially in math.[9]
Still, as anyone who has worked in or attended a charter school in Arizona or Michigan can attest, the boundaries of a traditional school district don’t necessarily define the boundaries of the local education market. And regardless of the unit of analysis, important questions remain unexplored. For example, no extant study has examined the effect that an increase in charter school enrollment share has on the achievement of economically disadvantaged (ECD) students or the racial and socioeconomic achievement gaps that exist within communities. Accordingly, this study utilizes new and better data to examine the relationships between charter school enrollment share and a broader set of achievement outcomes at the metropolitan level (rather than the district level).
DATA AND METHODS
Data for this study come from two sources: The first is the Common Core of Data collected by the National Center for Education Statistics, which includes school-by-grade-by-year-level information on total and subgroup enrollment for every school and year for which states reported such data. The second is the most recent version of the Stanford Education Data Archive (SEDA 4.1), which includes nationally comparable estimates of average metro-area-by-year-by-grade-level achievement in Reading Language Arts and math for ten school years (2008–09 through 2017–18) and six grade levels (grades 3–8), plus a host of other variables.[10]
Per the SEDA documentation, a unit increase in reading or math achievement can be thought of as the progress made by the average student in the average metro area in the average school year—or, for the purposes of this report, as one “year of learning” or “grade level.”
Importantly, SEDA’s achievement estimates reflect the performance of essentially all regular public schools that were physically located within a metro area, including nearly all “brick-and-mortar” charters not classified as “special education” or “alternative” schools;[11] however, more recent versions of SEDA don’t include data on charter school enrollment. Consequently, the variable of interest (“charter school enrollment share”) was constructed by (1) downloading school-by-grade-by-year-level enrollment data from NCES for every charter and traditional public school and every school year from 2000–01 through 2017–18, (2) merging the resulting dataset with SEDA 4.1 at the school-by-year level, and (3) calculating a cohort’s average exposure to charter school enrollment share between Kindergarten and the assessed grade level by dividing total charter school enrollment across relevant grade levels by total public school enrollment across those same grade levels.
Taking the steps outlined above for the “all-students” group yields a “total charter school enrollment share” variable that ranges from zero to 47 percent across a total of 100,477 metro-area-by-grade-by-year-level observations, 5,596 metro-area-by-grade-level units, and 934 metro areas, of which 450 had at least one charter school in the study period (in addition to 400 metropolitan statistical areas, SEDA 4.1 includes 534 micropolitan statistical areas, which are included in the analysis; for simplicity’s sake, we refer to both groups as “metros”).
As Table 1 illustrates, this variable is highly skewed. Still, we do observe total charter school enrollment share in excess of 10 percent 5,064 times across 117 different metro areas, including twenty of the largest one hundred metro areas in the country. Similarly, the distributions of our three measures of subgroup charter school enrollment share—which are based on the percentages of Black, Hispanic, and White students who enrolled in charters—are skewed. However, there are still enough metro areas with white, Black, and Hispanic charter school enrollment share above 10 percent to permit a meaningful analysis of the implications for the average achievement of the students in these groups.
Table 1. Distribution of metro areas with nonzero total or subgroup charter school enrollment share
0–10%
10–25%
25–50%
50–100%
Total charter school enrollment share
440
117
19
0
White charter school enrollment share
431
127
25
2
Hispanic charter school enrollment share
414
118
23
2
Black charter school enrollment share
330
154
33
4
Notes: Columns show the number of metro areas with at least one observation in the specified range of total or subgroup charter school enrollment share for which contemporary reading and math estimates exist for all students and the relevant subgroup. Because many metro areas experienced increases or decreases in charter school enrollment share during the study period, some metro areas are reflected in multiple columns.
To isolate the relationship between charter school enrollment share and average reading language arts and math achievement, this study relies on a combination of MSA-by-grade fixed effects, MSA-by-year fixed effects, and state-by-grade-by-year fixed effects, plus a rich collection of observable MSA-by-grade-by-year characteristics (see Technical Appendix). In a nutshell, this "triple differences" model compares different grade levels within the same MSA to see if those where charter school enrollment share grew or shrank experienced relative increases or decreases in achievement—after taking into account observable changes in student demographics and statewide factors that could have impacted scores in the relevant years and grade levels (e.g., an easier standardized test).
Standard errors are clustered at the metro level to account for potential autocorrelation. Unless otherwise noted, data are weighted by the number of tested students in a metro area, grade, year, and subject (see Technical Appendix for unweighted and variance weighted estimates). Finally, because one goal of the report is exploring the potential for diminishing returns, rather than assuming a linear relationship between charter school enrollment share and average achievement, some specifications allow the slope and curve of this relationship to change at the tenth, fiftieth, and ninetieth percentiles of charter school enrollment share.
FINDINGS
Finding 1: On average, an increase in total charter school enrollment share is associated with a significant increase in the average math achievement of poor, Black, and Hispanic students, which is concentrated in larger metro areas.
On average, a one-percentage-point increase in a metro area’s total charter school enrollment share is associated with a 0.025 grade-level increase in the average math achievement of its economically disadvantaged and/or Hispanic students, and there is suggestive evidence that it is associated with a similar increase in Black students’ math achievement (Figure 3).
Figure 3. On average, an increase in total charter school enrollment share is associated with a significant increase in poor and minority students’ math achievement.
Notes: Solid bars denote significance at the 95 percent confidence level. Striped bars denote significance at the 90 percent confidence level. Empty bars denote estimates that are not statistically significant at conventional levels. ECD stands for economically disadvantaged students.
In contrast, the relationship between total charter school enrollment share and average reading achievement is positive but statistically insignificant for most student groups, with the notable exception of economically disadvantaged students (Figure 4). Per the figure, there is suggestive evidence that, on average, a one-percentage-point increase in total charter school enrollment share is associated with 0.015 grade-level increase in economically disadvantaged students’ reading achievement. And, despite failing to achieve statistical significance, the estimate for Hispanic students’ reading achievement is notably large and positive.
Figure 4. There is suggestive evidence that an increase in total charter school enrollment share is associated with an increase in economically disadvantaged students reading achievement.
Notes: Striped bars denote significance at the 90 percent confidence level. Empty bars denote estimates that are not statistically significant at conventional levels. ECD stands for economically disadvantaged students.
To grasp the implications of these estimates, it helps to consider the potential effects of larger increases in charter school enrollment share. Accordingly, Figure 5 shows the relationship between total charter school enrollment share and economically disadvantaged students’ average math achievement for the full range of charter school enrollment share that we observe—that is, from zero to 47 percent.
Figure 5. On average, an increase in total charter school enrollment share is associated with a significant increase in economically disadvantaged students’ average math achievement.
Notes: This graph was generated using the mkspline2 command in Stata. The three red dots are “knots” (i.e., inflection points), which are placed at the tenth, fiftieth, and ninetieth percentiles of total charter school enrollment share. The dotted lines show the 95 percent confidence interval.
Because no individual metro area actually experienced a forty-seven-percentage-point increase in total charter school enrollment share during the study period, this figure should be interpreted carefully. Still, a reasonable interpretation of the graph is that, on average, a move from zero to 10 percent charter school enrollment share—that is, a move that more than a dozen metro areas did make during the study period—is associated with a 0.25 grade-level increase in economically disadvantaged students’ average math achievement.[12] Moreover, the slope of the graph beyond the 10 percent threshold suggests that further increases in charter school enrollment share are associated with gains of similar magnitude (note that the assumption that this relationship is linear has been relaxed for the purposes of this figure).
Importantly because the estimates that are the basis for the figure are weighted by enrollment, these gains are probably a better guide to the experiences of larger metros than smaller metros (though, of course, no specific metro area’s experience will match the graph precisely). And in fact, estimates from alternative specifications suggest that these gains are indeed concentrated in larger metro areas (see Technical Appendix).
Finding 2: On average, increases in Black and Hispanic charter school enrollment share are associated with sizable increases in the average math achievement of these student groups, especially in larger metro areas.
On average, a one-percentage-point increase in Hispanic charter school enrollment share is associated with a 0.016 grade-level increase in Hispanic students’ average math achievement (Figure 6). Similarly, there is suggestive evidence that an increase in Black school charter school enrollment share is associated with an increase in Black students’ average math achievement. Specifically, a one-percentage-point increase in Black charter school enrollment share is associated with a 0.01 grade-level increase in average Black achievement.
In contrast, increases in Black and Hispanic charter school enrollment share are not associated with significant increases in those student groups’ reading achievement (though both estimates are positive), nor is an increase in white charter school enrollment share associated with a significant increase in white students’ average reading or math achievement (though, again, the estimates for both subjects are positive).
Figure 6. On average, increases in Black and Hispanic charter school enrollment share are associated with sizable increases in the average math achievement of students in these groups.
Notes: Solid bars denote significance at the 95 percent confidence level. Striped bars denote significance at the 90 percent confidence level. Empty bars denote estimates that are not statistically significant at conventional levels. RLA stands for Reading Language Arts.
To illustrate the potential implications of these results, Figure 7 shows the relationship between Hispanic charter school enrollment share and Hispanic students’ average math achievement for the full range of Hispanic charter school enrollment share—i.e., from zero all the way to 55 percent.
Figure 7. On average, an increase in Hispanic charter school enrollment share is associated with a significant increase in the average math achievement of Hispanic students.
Notes: This graph was generated using the mkspline2 command in Stata. The three red dots are “knots” (i.e., inflection points), which are placed at the tenth, fiftieth, and ninetieth percentiles of Hispanic charter school enrollment share. The dotted lines show the 95 percent confidence interval.
Like Figure 5, this figure should be interpreted cautiously, as no individual metro area actually moved from zero to 55 percent Hispanic charter school enrollment share during the study period. Still, a reasonable interpretation of Figure 7 is that, on average, a move from zero to 10 percent Hispanic charter school enrollment share—a move some metro areas did make during the study period[13]—was associated with a 0.14 grade-level increase in Hispanic students’ average math achievement. Moreover, the shape of the graph, which is nearly indistinguishable from a straight line, suggests that further increases in Hispanic charter school enrollment share are associated with similar increases in Hispanic achievement (note that the assumption of linearity has been relaxed for the purposes of this figure).
Because the estimates that are the basis for the figure are weighted by enrollment, these gains are probably a better guide to the experiences of larger metros than smaller metros. And, in fact, evidence from alternative specifications suggests they are highly concentrated in larger metro areas (see Technical Appendix).
Finding 3: On average, an increase in total charter school enrollment share is associated with a significant narrowing of a metro area’s racial and socioeconomic math achievement gaps.
On average, a one-percentage-point increase in total charter school enrollment share is associated with a 0.025 grade-level narrowing of the gap between ECD and non-ECD students’ average math achievement, although there is no significant relationship for reading (Figure 8).
Similarly, a one-percentage-point increase in total charter school enrollment share is associated with a 0.032 grade-level decline in a metro area’s Black-White math achievement gap (though, again, there is no significant relationship for reading).
Figure 8. On average, an increase in total charter school enrollment share is associated with a significant narrowing of a metro area’s Black-White and ECD–non-ECD math achievement gaps.
Notes: Solid bars denote significance at the 95 percent confidence level. Empty bars denote estimates that are not statistically significant at conventional levels. ECD stands for economically disadvantaged students.
Once again, it is illuminating to see what effects of this magnitude imply about the potential consequences of larger increases in charter school enrollment share. For example, Figure 9 shows the relationship between total charter school enrollment share and the Black-White math achievement gap for the full range of charter school enrollment share—i.e., from zero all the way to 47 percent.
Figure 9. On average, an increase in total charter school enrollment share is associated with a significant decline in a metro area’s Black-White math achievement gap.
Notes: This graph was generated using the mkspline2 command in Stata. The three red dots are “knots” (i.e., inflection points), which are placed at the tenth, fiftieth, and ninetieth percentiles of total charter school enrollment share. The dotted lines show the 95 percent confidence interval.
Like figures 5 and 7, this figure should be interpreted with care, as no individual metro actually moved from zero to 47 percent charter school enrollment share during the study period. Still, a reasonable interpretation is that moving from zero to 10 percent total charter school enrollment share is, on average, associated with a 0.4 grade-level decline in the Black-White math achievement gap, and there is suggestive evidence that further increases in charter school enrollment share are associated with further narrowing of the aforementioned gap.
Similarly, Figure 10 shows the relationship between total charter school enrollment share and the math achievement gap between ECD and non-ECD students.
Figure 10. On average, an increase in total charter school enrollment share is associated with a significant decline in a metro area’s ECD–non-ECD math achievement gap.
Notes: This graph was generated using the mkspline2 command in Stata. The three red dots are “knots” (i.e., inflection points), which are placed at the tenth, fiftieth, and ninetieth percentiles of total charter school enrollment share. The dotted lines show the 95 percent confidence interval.
Like the previous figure, this one should be interpreted cautiously. Still, a conservative interpretation is that a move from zero to 10 percent total charter school enrollment share—that is, a move that at least some MSAs made during the study period—is associated with a 0.25 grade-level decline in a metro area’s ECD–non-ECD math achievement gap. Moreover, the slope of the graph beyond the 10 percent threshold suggests that further increases in charter school enrollment share are associated with similarly meaningful declines.
Because of the glacial speed at which the charter sector has expanded in most places, readers will have to judge for themselves whether it is reasonable to assume that the benefits that some metros reaped by moving from zero to 10 percent charter school enrollment share during the study period can be added to the benefits that other metros seem to have reaped by moving from 10 to 20 percent (and beyond).
Admittedly, some of these changes could be attributable to lower achievement for white and/or non-ECD students, rather than higher achievement for Black and ECD students. But in general, the estimates that are the basis for Findings 1 and 2 suggest the latter is a bigger driver.
For context, the same achievement data that are the basis for these results also suggest that, on average, Black students and ECD students were approximately 2.5 years behind White students and non-ECD students in both reading language arts and math before the pandemic struck.[14]
How do these findings compare to those of the first Rising Tide report?
Broadly speaking, the findings from this analysis are consistent with those of the first Rising Tide report. For example, both reports find that increases in charter school enrollment share tend to boost the average achievement of Black and Hispanic students but not White students. However, in contrast to the first report, which found broader and more consistent gains in reading, this report finds larger and more definitive gains in math.
There are at least three potential explanations for this difference. First, the present study analyzes charter school enrollment share at the metro level instead of the district level, meaning places like Arizona and Texas (where charters have historically performed better in reading) no longer receive disproportionate weight relative to places like Florida and New York (where they perform better in math).[15] Similarly, this report includes three additional years of data on reading and math achievement (for 2015–16 through 2017–18), and there is some evidence that charters’ performance improved more quickly in math than in reading during those years.[16] Finally, this report uses somewhat different analytic methods than the first report.
TAKEAWAYS
1. In general, the growth of charter schools benefits low-income, Black, and Hispanic students academically.
Because this study makes no distinction between the achievement of students in charter schools and that of students in traditional public schools, it’s impossible to say if or to what extent the gains associated with higher charter school enrollment share reflect charter schools’ impacts on enrolled students, as opposed to their competitive effects (or other factors). But regardless of the mechanism, the biggest takeaway from this study is that charters’ effects on the achievement of poor, Black, and Hispanic students are positive, consistent with prior research.
Notably, the results are more consistent and definitive for math than for reading—though even there the estimates for historically disadvantaged groups are encouraging—consistent with the only other study to estimate the effects of increases in charter schools’ enrollment share.[17]
2. More attention should be paid to charter schools’ impressive results with Hispanic students.
Like previous studies, this analysis suggests that many Hispanic students benefit from enrolling in charter schools, especially in the biggest metropolitan areas and in math. Yet the literature on the rapidly expanding supply of predominantly Hispanic charters is shockingly thin.
Because the success of these schools could hold lessons for charters and traditional public schools alike, it would be nice to know how they are approaching things like bilingual education—or at least the education of bilingual students. For example, research suggests that English-language learners in Boston charter schools are well served despite—or perhaps because of—the dearth of specialized programming.[18] But of course, most of the country’s predominantly Hispanic charters are in places like Texas, Florida, and Arizona, all of which have different student populations.
3. Charter schools have the potential to significantly reduce America’s racial and socioeconomic achievement gaps.
Despite the complexity of the data, the salutary effects of increases in charter school enrollment share on racial and socioeconomic achievement gaps are detectable at the metropolitan level. Moreover, the evidence suggests that these reductions are mostly attributable to gains among Black and Hispanic students (though some of the estimates for white and non-ECD students are negative).
Notably, the gains for historically disadvantaged groups seem to be concentrated in the country’s largest communities—that is, in places like New York and Houston, as opposed to Ketchikan and Beeville. Moreover, although caution is clearly warranted, on balance the results suggestthat metro areas with higher baseline charter penetration tended to benefit from further increases.
In the wake of a crisis that has seriously exacerbated America’s racial and socioeconomic achievement gaps, any policy that helps to narrow those gaps deserves policymakers’ consideration.
Limitations
The most obvious limitation of this study is that increases in total and/or subgroup charter school enrollment share aren’t randomly assigned to the metro-by-grade-level units in which they occur, meaning the estimates of their effects on average achievement are ultimately vulnerable to selection bias insofar as different grade levels within metro areas have different achievement trends for reasons unrelated to charter school enrollment share.
Similarly, this study's reliance on metro-by-grade-level estimates of average achievement, as opposed to district- or student-level data, has advantages and disadvantages. On the one hand, a metro-level analysis means some students' effective exposure to higher charter school enrollment share may be limited, particularly insofar as the estimates reflect competitive effects. On the other hand, focusing on metros likely reduces the number of students who cross jurisdictional boundaries unobserved to attend charter schools[19] (or who leave for private schools)[20] to the point where it poses little threat to the findings.[21]
Furthermore, most individual metro areas experienced relatively modest changes in total and/or subgroup charter school enrollment share during the study period. Because of this feature of the data, the “splines” that are the basis for some of the figures in the Findings section should be interpreted cautiously, as it’s not clear to what extent the experiences of metro areas with different baseline enrollment shares can be responsibly combined.
Finally, it’s worth noting that the model that is the basis for this report is somewhat underpowered (though in some cases, it is actually an improvement over simpler models, per the Technical Appendix). In other words, it’s possible that some statistically insignificant results conceal substantively significant effects. For estimates from alternative models and/or specifications that occasionally speak to that possibility, see the Technical Appendix by clicking “DOWNLOAD PDF” on the right.
[2] Jeanne M.Powers, Amelia M. Topper, and Michael Silver, “Public School Choice and Student Mobility in Metropolitan Phoenix,” Journal of School Choice 6, no. 2 (2012): 209–34, doi:10.1080/15582159.2012.673862.
[4]Because of the way the SEDA data were constructed, calculating average exposure to charter school enrollment share for every metro-area-by-grade-by-year-level observation in the dataset is a fraught endeavor; however, while the noisiness of charter school enrollment share makes it harder to detect its effects on student achievement, there is no reason to believe that it is a source of bias. And because charter school enrollment share is highly autocorrelated within cohorts, accounting for prior exposure has little impact on the overall story.
[5] Julian R. Betts and Y. Emily Tang, “The Effects of Charter Schools on Student Achievement,” in School Choice at the Crossroads: Research Perspectives, ed. Mark Berends, R. Joseph Waddington, and John Schoenig (New York: Routledge, 2018), 69–91.
[6]Sarah R.Cohodes and Katharine S. Parham, “Charter Schools’ Effectiveness, Mechanisms, and Competitive Influence” (NBER Working Paper 28477, National Bureau of Economic Research, Cambridge, MA, 2021), doi:10.3386/w28477.
[7] Strictly speaking, this line of reasoning does not account for the possibility that charter schools affect the performance of private schools; however, because those effects are likely positive—or, at worst, neutral—accounting for them is unlikely to affect the bottom line.
[10] For the purposes of this report, we rely on the “gcs” achievement estimates, which allow the estimates to be interpreted in terms of “years of learning.” However, substituting the “cs” estimates yields nearly identical results.
[11]All special education and virtual schools, many alternative schools, and approximately half of schools in highly “remote” or “distant” town/rural locales are not assigned to a specific metro area in SEDA and are therefore excluded from the analysis.
[12] Eight metro areas made this specific move (i.e., moved from zero to at least 10 percent charter school enrollment share). Another ten metro areas made moves of equal or greater magnitude (i.e., experienced at least a 10 percent increase in charter school enrollment share during the study period but from a different starting point). Finally, sixty metro areas experienced five- to ten-percentage-point increases in charter school enrollment share during the study period.
[13] Specifically, at least seven metro areas made this exact move, and at least sixteen others with higher baseline charter school enrollment made moves of similar or greater magnitude.
[14]More precisely, in the average metro, grade, and year in the study period, the gap between White and Black students’ average achievement was approximately 2.1 years of learning in both reading language arts and math; however, when the data are weighted by enrollment, this gap increases to 2.5 years of learning. Similarly, the gap between ECD and non-ECD students’ average achievement was 2.1 years of learning in RLA and 1.9 years of learning in math in the average metro, year, and grade; however, when the data are weighted by total enrollment, these gaps increase to 2.5 and 2.3 years of learning, respectively.
[16] For example, CREDO estimates that students in urban charters gained twenty-nine days of learning per year in 2008–09 and fifty-eight days of learning per year in 2011–12.
[17] As noted in the Background section, Chen and Douglas (2020) also find significant gains in math and smaller and less consistent gains in reading language arts, despite having a different unit of analysis and employing different methods.
[18] Elizabeth Setren, “Targeted vs. General Education Investments: Evidence from Special Education and English Language Learners in Boston Charter Schools,” Journal of Human Resources 56, no. 4 (2021): 1073–112, doi:10.3368/jhr.56.4.0219-10040R2.
[19] Notably, one recent study authored by Kyle Abbott, Eric Houck, and Douglas Lee Lauen at UNC Chapel Hill, estimated that 23 percent of charter school students in North Carolina attended a charter school that was located outside their district of residence. To our knowledge, there is no equivalent estimate for any other state, nor are we aware of any attempts to estimate the share of charter school students whose commutes take them across MSA boundaries. Kyle Abbott, Eric Houck, and Douglas Lee Lauen, “Out of Bounds: The Implications of Non-Resident Charter Attendees for North Carolina,” (forthcoming).
[20] For example, Chen and Harris (2020) estimate that moving from zero to >10 percent charter school market share reduces private schools’ enrollment share by one to two percentage points.
[21] For example, suppose that half of the Hispanic students who attend charter schools within the borders of the largest MSAs in the country are “metro switchers” from neighboring MSAs and that these students outperform their Hispanic counterparts in receiving MSAs’ traditional public schools by an average of three grade levels in math. In this hypothetical scenario, a move from zero to 50 percent Hispanic charter school market share would artificially boost Hispanic math achievement in these MSAs by 0.75 grade levels ceteris paribus—that is, roughly as much as the graph relating Hispanic charter school market share to Hispanic math achievement implies. But of course, this scenario is highly unlikely. After all, some of the largest MSAs are more than one hundred miles across, and most MSAs are drawn around (not through) major population centers—and Hispanic students in charter schools do not outperform those in traditional public schools by three grade levels. More plausibly, we might imagine that 20 percent of Hispanic students in charter schools are “metro switchers” and that these students outperform Hispanic students in receiving MSAs by an average of one grade level. In this alternative scenario, the artificial boost to Hispanic students’ average math achievement as Hispanic charter school market share increased from zero to 50 percent would be 0.1 grade levels. In other words, “MSA switchers” would have to be both implausibly high achieving and implausibly numerous to explain the observed relationships between charter school market share and all publicly enrolled students’ academic achievement.
ABOUT THIS STUDY
This report was made possible through the generous support of the Walton Family Foundation, as well as our sister organization, the Thomas B. Fordham Foundation. We are deeply grateful to external reviewer Martin West, professor of education at the Harvard Graduate School of Education, for offering feedback on the methodology.
On the Fordham side, we express gratitude for the careful work of Fordham’s associate director of research, David Griffith, in conducting the study and authoring the report. Thanks also to Chester E. Finn, Jr. for providing feedback on drafts, Pedro Enamorado for managing report production and design, Victoria McDougald for overseeing media dissemination, Will Rost for handling funder communications, and Jeremy Smith for assistance at various stages in the process. Finally, kudos to Pamela Tatz for copyediting, Jonathan Lutton for developing the original (adapted) cover image, and Dave Williams for creating the report’s figures.
Today, the National Alliance for Public Charter Schools (NAPCS) released its latest review of states’ charter school policies. This year, NAPCS ranked Ohio’s charter law as 12th out of 44 states plus the District of Columbia, a significant increase when compared to the state’s ranking in 2021 (24th). The report specifically noted Ohio’s recent removal of geographic restrictions on startup charter schools, as well as other policy improvements that have occurred within the last year.
“Ohio continues to make strides towards a policy environment that supports high-quality public charter schools,” said Aaron Churchill, Ohio Research Director for the Thomas B. Fordham Institute. “The jump in the national rankings reflects these recent policy gains, which include the removal of geographic restrictions and more supportive funding, especially for its highest performers. Kudos to Governor DeWine and the General Assembly for supporting these policy changes, and we encourage them to redouble their efforts to expand great charter opportunities to Ohio students.”
An Ohio-focused, in-depth piece in the Washington Post this week described a troubling web of relationships between sponsors, charter operators, and charter-affiliated for-profit real estate companies. It’s all very disheartening. Here’s hoping there’s more to the story than presented in the Post.
More choice for rural students
A former Ohioan, now the dean of students at a charter school in Idaho, opined in the Idaho Statesman this week, about the lack of school choice options for rural students. Not only in the Gem State, but across the country. “One school in a community does not ensure that students have the option to find a school that is the best fit for them,” he writes. “More needs to be done to get school choice options into our rural communities.”
National School Choice Week begins Sunday and celebrations will be numerous across the country. Andrew Campanella, president of NSCW, kicks things off with an op ed in the Warwick (RI) Beacon, in which he reminds us all “how wonderful it is that kids have different talents” and that “the different learning environments that inspire them” should be the focus of our celebrations next week.
Speaker series to aid teacher effectiveness and wellbeing
Career-themed schools are still a thing in Toledo City School District, including what we’ll call ProMedica High, a medical-themed program supported heavily by the aforementioned health system, which will open its hermetically sealed doors for the first time this fall. And while I’m happy that the young man interviewed is excited about what might lie ahead for him there, let us remind ourselves that a) kid is talking about medical school/being a doctor while the system will be hyping porter and radiography tech just as heavily, and b) every district doing this career immersion thing guarantees no jobs or postsecondary readiness for any student. And I am still skeptical of them all for these reasons. (WTOL-TV, Toledo, 1/19/22)
Finally today, on the eve of National School Choice Week, one of the attorneys who will soon be fighting in court to preserve Ohio’s EdChoice Scholarship Program has some blunt words for the haters in this op-ed published in the Dispatch. (Columbus Dispatch, 1/21/22)
Did you know you can have every edition of Gadfly Bites sent directly to your Inbox? Subscribe by clicking here.
In the latest twist to an ongoing saga that began last September, U.S. Secretary of Education Miguel Cardona came under fire last week for his alleged role in soliciting a six-page letter sent from the National School Boards Association (NSBA) to President Biden. In this letter, the NSBA sought federal law enforcement intervention to threats aimed at school board members across the country. In emails obtained by the watchdog group Parents Defending Education, an NSBA board member wrote that the interim executive director at the time told officers that the letter sent to Biden was “from a request by Secretary Cordona” [sic].
For his part, Cardona quickly denied orchestrating the letter, but it has done little to tamp down the furor. Forty-one House Republicans have since called for Cardona’s head, and twenty-two Senate Republicans wrote a letter to him demanding answers. The hullabaloo comes a year into Cardona’s otherwise unremarkable tenure, during which he’s earned praise from educators, but low marks from parent groups. (The irony is not lost that his predecessor had the exact opposite to contend with.) It should come as no surprise, then, that the name calling and finger pointing trace back to parents taking offense at being labeled “domestic terrorists” in NSBA’s letter, and the perception of having the DOJ and FBI sicced upon them.
Absent a congressional hearing, revealing media leak, or crackerjack investigative journalism, the question of Cardona’s involvement cannot be definitively answered. Although Congressional Republicans would love nothing more than to see him become the fall guy, they would do well to consider how their thirst for blood might cut both ways. Biden could use this dust-up to his advantage, as conservative pundit Jonah Goldberg has observed, if he fired Cardona and publicly repudiated any sub rosa involvement (e.g., “This is not how we do business in my administration”), in effect creating a Sister Souljah moment. Realistically though, it’s hard to see the president throwing any of his friends or allies under the bus, given his track record of passing on other such opportunities, most recently in Chicago, to say nothing of how nothing seems to be going right for him these days. If anything, the White House might have asked Cardona to solicit the letter, which would make it difficult for Biden to disown a scheme of his own staff’s making.
What’s worrying, if not surprising, about the GOP’s eagerness to defenestrate Cardona is how it memory-holes the tireless, behind-the-scenes advocacy that got him installed over a number of less desirable (to Republicans) candidates in the first place. It’s worth remembering that the former head of the nation’s largest teachers union almost became education secretary. Although that didn’t happen, Cardona’s deputy is as establishment as they come when it comes to opposing the flavors of reform typically favored by conservatives. Which is to say, in the extremely unlikely event that Republicans get their way, there’s little reason to believe that whoever follows would be an improvement in their eyes. In fact, chances are that such a person would be more overtly sympathetic to the teachers unions and their distaste for all things reform-related.
To be fair, the effort to tie Cardona to NSBA’s ill-advised missive is not entirely of the right’s doing. In some respects, he brought this upon himself by initially taking parents for granted during the school reopening process, an oversight demonstrated by the absence of parent leaders from last year’s National Safe School Reopening Summit, which Cardona hosted. This omission didn’t go unnoticed, providing fodder for opponents and helping to fuel the narrative that the Biden administration has been ginning up a “war on parents.” In this sense, the pillorying of Cardona can be seen as the latest in a larger attempt to swing Republicans to victory in this fall’s midterms.
Taken together with last November’s gubernatorial contest in Virginia, Republicans have seen how well school debates can be weaponized, and they view the NSBA/Cardona brouhaha as another opportunity to keep banging the drum. Lost in the overcaffeinated fracas is any granularity or nuance. When it comes to the NSBA letter and the DOJ’s response in particular, this means that an informed and good-faith debate about whether threats against school boards require any form of federal government action is completely out of the question.
Following two years of interrupted and intermittent schooling, the anger and frustration being directed at Cardona is understandable, if not somewhat misplaced. These feelings have been amplified by a growing sense that he and his team have prioritized teacher well-being over student recovery. However, without more conclusive evidence that the fix was in on this—in which case, a serious discussion on repercussions is worth having—Republicans should spend less time attacking Cardona and more time pursuing a seriousagenda that responsibly addresses the many hurdles our students and families still face.
FOREWORD
