Any issue that involves race, gender, economic inequality,
children, and public funding is bound to bring with it some controversy; the
debate over charter schools is no exception.
The idea of a charter school is simple.
Along with funding comes a whole host of procedural requirements that
public schools have to abide by, restricting the ability of public schools to
adapt to their neighborhood. While still
publically financed, charter schools are granted much more operational autonomy
in return for more stringent accountability.
Many charters have taken advantage of the autonomy by structuring their
administrative organization and teaching practices in very different ways than
public schools. In particular, much
attention has been given to the fact that they, on average, employ younger
non-unionized teachers, while public schools work exclusively with unions. But stripping away some aspects of the
controversy, the question stands: does the empirical evidence suggest that
charter schools work at improving student achievement.
Initially, this question runs headlong into the problem of
selection. An ideal study wants to
compare one student who goes to charter school with an otherwise identical
student who goes to a public school.
Then, any difference in their academic achievement is solely due to the
differences between the schools the two of them attended. While it is relatively easy to control for
observable differences between a student who attends a charter school and one
who attends a public school, the very fact that they chose different schools
implies differences between the two that researchers cannot control for. Studies that naively compare charter school
students with public school students might reflect a difference in the type of
students who attend the two schools, rather than a difference in the ability of
the two schools to educate students.
Because charter schools are publically funded, if more
students want to attend the charter school than the school has room for, the
school must hold a lottery to determine which students can attend the school; they
must choose this way instead of “hand picking” which students can attend, the
way a private school like Colby does. In
addition to drama, as documented in Waiting
for Superman, these lotteries provide the randomization needed to
consistently estimate the effects of attending a charter school. In particular, it allows researchers to
compare two students who not only are observably identical, but also made the
same choice to apply to a charter school (or multiple charter schools).
These lottery studies all point to a similar conclusion:
students who win a lottery to attend a charter school do better than identical
students who lost that lottery. This is
true for just about every observational type of student, regardless of his or
her gender, race, prior test scores, or family income. (See, for example, Hoxby
et. al., 2009) In addition, this is true
for a variety of ways to measure “do better,” whether it is high school test
scores, SAT scores, AP scores, or enrollment at four-year colleges. (See
Angrist et. al, 2013) The one caveat to this result is that it holds only for
urban charter schools; the results for rural or suburban charter schools are
much more mixed.
After reading these studies, the policy implications seem
clear: remove, at least in urban areas, some of the many existing barriers that
currently make it difficult to found new charters. But while the research suggests that
oversubscribed charter schools work well, the policy implications require more
thought than it at first might seem. First,
these studies only estimate the effect of charter schools that hold
lotteries. Assuming that more children
apply to the best charter schools, these studies are implicitly looking at the
effect of the best charter schools, not of every charter school. Non-experimental studies, which compare
students who are demographically identical, but who differ in their school choice,
can estimate the effect of all charter schools.
But they must assume that the selection issue, described in the second
paragraph, does not matter. There is
some evidence that this is true in among the schools that are oversubscribed,
but I think it is safe to say that this issue is not well understood. (As an aside, this is a plus for me
personally, because it means I can study it.)
The main issue, however, is that even though it is now known
that current charter schools are (on average) more successful than public
schools, no one has a great sense of why. This makes it tough to know whether or not
the positive effect of charter schools will extend to the next set of
charters. Suppose, for example, that motivated
students apply to charter schools and that being around motivated students
increase academic outcomes. In that
case, students who attend charter schools will do better academically than
identical ones who do not, but increasing charter schools is not necessarily a
good policy. This is both because the
next set of charter schools will not as successful and because the presence of
charter schools hurt public school students by lowering the motivation of their
peers. On the other hand, if charter
schools lead to better results because they often have a longer school year and
more flexibility in hiring teachers, increasing the number of charters is
advisable.
In fairness to charter schools, the available evidence
suggests that the being surrounded by motivated peers is not the reason why
charter students do better than public school students and that getting around
restrictions imposed by teachers unions is one of the main advantages of
charter schools, but the point still stands.
Without knowing why charter
schools work, it is impossible to know what the effect would be of increasing
the number of charters. Knowing why they
work is the next step, one that hopefully I’ll be able to help with.
