Online Charter Schools Have Poor Track Record, but They Can Reach Places Other Schools Can’t
They have a friend in Secretary of Education Betsy DeVos and have been growing rapidly in some states, in many cases by successfully beating back political opposition. In a recent interview, DeVos said that, under her tenure, “I expect there will be more virtual schools.”
At the same time, cyber-charters have weathered a blizzard of negative press: Reports in the The New York Times, Education Week, the San Jose Mercury News, and here at The 74 have often castigated e-schools as taking in billions in taxpayer dollars while producing dismal results for students.
Last year, a trio of groups supportive of school choice called for stricter oversight of virtual charters. This seems to have translated into legislative pushes in several states to more strictly regulate online schools, including closer monitoring of student attendance and engagement.
The harsh media coverage would seem to be warranted, as multiple studies, including a widely cited national report, have found that exclusively online charter schools lead to huge dips in student achievement.
Not so fast, say supporters and the schools themselves, pointing out that students who choose to attend an online school are likely quite different than peers at brick-and-mortar schools, making apples-to-apples comparisons of their achievement tricky.
Sarah Cohodes, a professor at Teachers College Columbia University who has studied charter school effectiveness, said that isolating the impact of virtual schools may be particularly challenging.
“We fundamentally don’t know who these kids are, and what’s making them attracted to this particular option,” she said.
These questions enter into a thicket of wonky methodological issues that are crucial for determining whether virtual charter schools serve as a lifeline for students or actively harm them.
The best research to date suggests that virtual instruction produces significantly worse results than in-person classes; however, online courses may provide benefits to students who don’t have access to brick-and-mortar schools.
But finding the right balance for online schools may be politically challenging.
The national study showing that online charter schools significantly harm student achievement was conducted by CREDO, a research group at Stanford University that has produced a series of oft-cited reports on charter performance.
CREDO’s research approach is fairly intuitive. Students at charter schools are matched to “virtual twins” at district schools that feed into charters — as the name implies, the twin is meant to be the same on a number of characteristics, including past test scores, special education status, gender, race, and whether a student qualifies for free or reduced-price school lunch (a rough proxy for family income). Then the researchers compare the test scores over time of charter students against their twins; the difference is the estimated effect of attending a charter.
However, examining the impacts of virtual schools brings a unique set of challenges. Let’s say there are two students, Jamie and Mayra, who appear, on paper, to be similar: they have the same prior test scores, the same poverty level, the same gender, etc. But Jamie ends up attending an online charter because she gets sick and can’t come into a physical school everyday. While she’s at the virtual school, her achievement drops significantly relative to Mayra — not because of the quality of her school, but because she’s ill.
The CREDO results would likely not be able to control for that and would pin the blame for low achievement on the online school. Researchers call this problem “selection bias.”
CREDO researcher Lynn Woodworth says this is a legitimate concern but questions how common the situation would be: “Some of these schools serve 4,000 or 5,000 kids — do we think 4,000 to 5,000 kids got sick?”
Andrew McEachin, a researcher at the RAND Corporation, co-authored a new study showing that Ohio’s online charter schools dramatically reduced students’ test scores. Similarly, the report could only control for readily apparent student characteristics, which McEachin believes might account for some, but not all, of the results.
“I think selection [bias] may … explain some of the negative effects, but the effects are so large that even without selection they’d still be negative and significant,” he wrote in an email.
McEachin also pointed out that in other respects, the findings may understate negative impacts because students who drop out or leave prior to the state test aren’t counted against the school.
Everyone agrees that the best way to isolate the impact of a school is to conduct a randomized trial, in which students who are offered a spot at a school are compared to those who apply but don’t get the chance to enroll. These sorts of studies have been conducted on a number of high-profile charter networks and sectors.
“The problem with all of these [other] methods,” said Cohodes of Columbia, referring to non-lottery approaches, is “that we can’t control for whatever it is that makes people interested in going to charter schools or a particular type of charter school.”
It’s not clear how much this matters in practice. CREDO’s approach has been criticized on this and other grounds, but some research has found that CREDO’s findings match up fairly well with different approaches, including lottery results.
Still, Cohodes, said, we might especially worry that studies of online schools would be biased because of the unique reasons students might attend them.
There are a handful of studies of online education — though not virtual charter schools specifically — that are less likely to be subject to these sorts of methodological limits.
In one study, Jennifer Heissel of Northwestern University examined the expansion of a state-run online algebra course for eighth-graders in rural North Carolina. Since this option was recently added, Heissel was able to compare students with access to the course against students in the same district from earlier years who took the course in person (in ninth grade). She could also look at eighth-graders from other districts who were taught algebra in person. By both methods, students who took the class online did worse on an end-of-year algebra exam.
“No matter which way I do the comparison, there are much lower scores for the kids who took virtual algebra,” Heissel said. The size of the negative impacts were comparable to those found in both the CREDO and Ohio studies of virtual charters.
Another report examined online learning at the higher education level, specifically DeVry University. The study compared college students in online courses to students who were given a chance to take a class either online or in person (since the school offered only some of its courses in person, but all of its classes virtually). Taking the course online reduced students’ grades substantially in that class and also made it less likely they’d stay in college.
These approaches have limitations, but they are less likely to be subject to the same selection bias concerns as matching studies; together they provide a solid case that exclusively online courses reduce student outcomes relative to an in-person option.
The two papers do not focus on virtual charter schools, and they can’t show that CREDO results are correct, since they’re looking at completely different data in different contexts. That they point in the same direction with results in the same ballpark does offer suggestive evidence that the online charter studies are picking up something real.
An older study of an online algebra course offered to eighth-graders in rural Vermont and Maine paints a more encouraging picture. The study randomly gave some middle-schoolers the chance to take an algebra course virtually; other students who didn’t get access were in schools that generally didn’t offer a stand-alone algebra class in eighth grade, though in most cases the normal math course covered algebra to some degree. The study showed that students who took the online class scored better on algebra exams at the end of eighth grade and were more likely to take advanced math classes in high school.
How to square those findings with Heissel’s North Carolina study, which also looked at an eighth-grade online algebra course?
“Virtual Algebra I may increase algebra knowledge more than a [regular] eighth-grade math class, but an in-person Algebra I class may increase algebra knowledge more than a virtual Algebra I class,” Heissel said. “It’s possible that you might score lower in virtual courses than in a regular Algebra I class, but that taking the course early still frees you up to take a more advanced course sequence in high school. There are trade-offs.”
In other words, in-person instruction seems to produce better results than exclusively online classes. But if virtual classes are the only viable option — because a rural brick-and-mortar school can’t offer certain courses or because a student can’t physically attend school due to illness — they might be worthwhile.
It may also be the case that the specific provider examined in the Maine and Vermont study was just particularly good.
At the higher education level, a recent study found that the option to take a computer science class online led to an increase in enrollment among students who would not have otherwise taken the class at all.
Research is hardly settled here, but these findings might hold lessons for online charter schools. Students who simply substitute traditional schools with a virtual program may see large reductions in achievement. Those who can’t attend a physical school may, in theory, benefit from an online option. Still, caution is warranted — it does not appear that any study to date has shown positive, or even neutral, results for online charters specifically.
Translating the existing research into policy will be politically and technically challenging.
Pro-charter groups calling for better regulation of virtual schools argue that they should (somehow) distinguish in admissions between students who are likely to succeed in an online learning environment versus those who are not.
Another option would be removing online schools from under the charter school banner, and allowing state- or district-sponsored online classes in a specific circumstances, like when a school can’t offer a class or when a student has a diagnosed illness.
Politically powerful online charter operators would almost surely oppose either of these changes. To many such schools — which often aggressively market their services — as well as their advocates, the only appropriate judges for whether a student is a good fit for a school are their parents.
Under this view, if a family thinks a child will do a better in an online school than a traditional one, that should essentially be the end of the discussion. DeVos has also espoused a steadfast commitment to allowing parents to choose whichever school they prefer.
K12, the largest operator of online charters in the country, responded to calls for greater oversight last year in a similar way.
“Policies that restrict parent choice, or create perverse incentives for schools to turn away at-risk children or others deemed not likely to succeed, should be rejected. They have no place in the school choice movement.”