Skip to main content

Welcome, the Hub connects all projects

Voices From The Field

MSPnet Blog: “Virtual schools: flying with broken wings”

See All Blog Posts

posted April 13, 2017 – by Brian Drayton

The National Education Policy Center ( is out with the latest in a series of “Virtual Schools Reports”, presenting data and analysis on the state of virtual and blended schools around the country. The 103-page report has three parts: [1] Full-time virtual and blended schools: Enrollment, student characteristics, and performance; [2] Still no evidence, Increased call for regulation: Research to guide virtual school policy; [3] Key policy issues in virtual schools.  Note that the “virtual schools” for the purposes of this report have no face-to-face interaction between students and teachers, while “blended schools” have varying proportions of in-person vs. on-line activity. In what follows, I will focus on the “virtual schools” strand of the report.
The increased emphasis on “choice” in education, championed with fresh energy by USED Sec. DeVos, but part of the education agenda in the Obama years as well, is likely to add further energy to the virtual/ blended school “market,”  one of many sectors in the “education market.”   (Sorry for the scare quotes, but I continue unreconciled to the construal of education as just another economic process.)   Especially in areas where there are not lots of alternatives to public schools, but authorities do not wish to invest in school improvement, virtual schools are increasingly advocated as the solution of choice.   The growth in such schools is documented in this report:  in 2015-16, they write, there were 528 full-time virtual schools, in 34 states, with an enrollment of 278,511 students.   Moreover, there is “virtual school legislation” under consideration in a large number of states (37 in 2016), and this trend is likely to continue at least for a while.

Now, as with labor-force projections, it’s important to keep track of what is being counted in reading accounts of the shape and structure of the virtual schooling phenomenon.  Case in point:  Who’s in charge?  Well, if you look at percentages of schools, it looks as though private “Education Management Organizations” (EMOs) are a small portion of the whole — just under 30% of the total.  However, the private EMOs account for 70% of the student enrollment across the country.

There are significant implications just related to this point alone, since increasingly these private institutions (managed by not-for-profit and for-profit entities of various stripes) are recipients of taxpayer funds,  as public education money is allowed to “follow the child”  in one of various ways.

Meanwhile, accountability to state or local entities is (as has been widely reported and is documented in this report) either lax or non-existent (as with charter schools). Indeed, “many states have frozen their accountability systems,”  or are loosening whatever financial or operational oversight is being exercised on charters as well as virtual schools.   The authors of the report were only able to get school performance data on schools form 18 out of the 38 states that have virtual or blended schools.   The news that emerges from the available data, however, is not reassuring:
• teacher:student ratios are higher in virtual schools (1:34) than in regular public schools (1:16), and in for-profit virtual schools it’s even higher (1:44).  Meanwhile, other data suggest that in many cases the teacher’s role is more circumscribed than the typical public school teacher’s, for example because the EMO purchases curriculum and associated assessments, and expects parents to take on some of the teacher’s role in overseeing and coaching the student.
•  For-profit virtual schools have lower proportions of lower-income and minority students than do mainstream schools;  not-for-profits seem to be closer in their demographic make-up to the mainstream school population.
• Available data about on-time graduation rates show virtual schools with much lower success than mainstream schools (about 44% for virtual schools vs 82% for public schools), with some for-profit virtuals even lower (schools operated by K12 Inc at about 37%).
• Some states rank school performance, on the basis of variables relating to academic performance.   Data from 19 states were available for the study period.  Looking across all full-time virtual schools, about 37% received an “acceptable” rating.  Interestingly, district-operated virtual schools had a much higher percentage of “acceptable” ratings (56%) than did for-profit schools (26%).

There are data on teacher quality and the sources of teachers;  more details on demographic factors; and an interesting examination of available data on the qualifications of principals or other administrators of virtual schools.

Another broad area is in accounting for costs, and here the data are also murky — indeed, the authors suggest that virtually (sorry) no state has required consistent accounting of the cost-structure of these schools, especially for-profits, which are no less expensive then brick-and-mortar schools.  The authors suggest some possible reasons that virtual schools might not cost less per pupil than mainstream schools — but these are conjectures, and cannot be tested as yet with actual data.

One of the many unfortunate consequences of market-driven innovation is that markets tend to be created for new products before there is any strong evidence that the new thing is a better thing.  So we try one thing after another, using teachers, students, and everyone else in the system as experimental subjects quite as a matter of course, and quite without consent.  Of course, education is not the only arena in which we allow such trial-and-error intervention — nor the only area in which we can carry on the intervention (one cannot call it an experiment) for long periods of time without even collecting reasonable data about the outcomes and impacts.  It has become common-parlance to use terms borrowed from economic modeling such as “cost-benefit analysis,” but a model is only as good as the data used to test and refine it, and I have seen no educational modeling yet that seems adequate to the multi-variate problem presented by almost any educational policy.


Note:  The opinions expressed in this blog are those of the author alone, and do not necessarily represent those of MSPnet, TERC, or the National Science Foundation. 


Discuss this post in the “Opinions Worth Debating” forum