In recent years, two conservative ideas about education have gone mainstream, capturing the imaginations of reformers across the political spectrum. The first is “accountability” for raising students’ test scores, enforced at the level of either the school or the teacher. The second is “choice,” in the form of charter schools.
These reforms are often presented as panaceas for the nation’s educational ills. Serious researchers have claimed that high-quality teachers (hired according to their ability to raise test scores) and charter schools can close gaps between income and racial groups in as little as three to five years. In a National Review Online debate that included me, the Manhattan Institute’s Marcus Winters claimed that these reforms could create an America in which nearly all students are college material.
In The Death and Life of the Great American School System, Diane Ravitch, one of the nation’s most respected education historians and a former advocate of these ideas herself, throws cold water in the face of everyone who’s contemplating these utopian visions. She shows that these approaches will not cure America’s schools — and, much of the time, don’t help at all. However, she fails to prove that her prescribed alternatives would be any better.
Ravitch’s ripest target is No Child Left Behind (NCLB), an attempt to enforce school-level accountability and a national embarrassment. Some of its problems involve the inherent difficulties of test-based accountability. Unless an accountability scheme focuses on individual test scores (rather than group-average test scores, as NCLB does), such a system is bound to reflect a school or classroom’s demographics rather than its performance. To duck “accountability,” principals and teachers will find all kinds of ways to send low-performing kids elsewhere. While tests do a decent job of measuring student performance, they’re far from perfect (even the weather can affect scores). Also, highly consequential tests always distort the teaching process: Teachers will spend valuable time coaching students on test-taking techniques, emphasize tested subjects over non-tested subjects, and sometimes even help the students cheat.
But more often, NCLB’s problems can be attributed to the sheer stupidity behind its implementation. With NCLB, the federal government simply declared that every student in America shall be “proficient” in reading and mathematics by the 2013–14 school year, and it uses schools’ dependence on federal aid to punish them for failing to progress with adequate speed toward that goal.
Schools that make insufficient progress for two years must allow students to transfer elsewhere, and must pay for transportation to the other schools. Schools with five years of insufficient progress have to “restructure.” Ravitch lists the five restructuring options: “convert to a charter school; replace the principal and staff; relinquish control to private management; turn over control of the school to the state; or ‘any other major restructuring of the school’s governance.’” (Most schools, not surprisingly, choose the ill-defined final option.)
Here’s the kicker, though: States get to decide for themselves what “proficient” means. They chose their own tests and set their own cutoff scores, and are free to adjust them as they desire. It should surprise no one that some states lowered their cutoff scores to create the illusion that their schools were improving. Due in part to this phenomenon, and in part to teachers’ focus on test-taking techniques, a disturbing trend emerged: In many cases, children were progressing on their states’ NCLB tests, but not on tests of the same subjects from the National Assessment of Educational Progress (a nationwide, standardized exam). Ravitch is far from the first person to explain the problems with NCLB, but given that this monstrosity is still in effect, pundits and scholars cannot heap too much opprobrium on it.
She also does a good job of deflating the hopes of “teacher quality” advocates. Research has shown that teachers matter — that children assigned to some teachers are far likelier than children assigned to others to improve their test scores year over year. Children assigned to the worst teachers lose about 5 percentile points when ranked against their peers, and children assigned to the best teachers gain about the same amount, making for a gap of 10 percentile points. Based on these numbers, three researchers argued, replacing underperforming minorities’ teachers with top-tier instructors could close the racial gap (34 percentile points) in four years.
Ravitch points to numerous problems with this research (in addition to the inherent problems of testing-based accountability noted above). First of all, only one-fifth to one-third of the gains that teachers impart are detectable a year later, meaning that they won’t accumulate the way the researchers assumed they would. Also, many teachers are inconsistent — they raise scores some years, keep kids on track others, and lower scores still others — and they often improve over time. It takes a few years’ worth of data to get even a reasonably good picture of how a teacher affects test scores. Most troubling is the fact that one study, using the same techniques as the others, found that fifth-grade teachers can raise their students’ third- and fourth-grade test scores. It’s obviously impossible for a teacher to affect how his students did in previous grades, so this calls the entire body of work into question.
#page#These problems, Ravitch claims, are evidence that students’ performance on test scores is not “precise enough” to be used as the sole basis for consequential decisions. But here she steps onto thin ice. Given that consequential decisions must be made — teachers must be hired, fired, and given or denied tenure; schools must be funded, censured for poor performance, and closed — there’s no such thing as a “precise enough” measure. There are only measures (and combinations of measures) that are better or worse than the alternatives. If a given measure chooses the best of two options only 52 percent of the time, it’s still better than a measure that does so 51 percent of the time, which itself is better than flipping a coin.
Ravitch suggests using test scores as only one part of the accountability process, along with such measures as teacher evaluations, grades, student work, attendance, and graduation rates. This method may very well produce better results, but it very well might not. The other measures have plenty of their own problems: Both teachers and students might behave better when evaluators are in their classrooms; emphasis on grades could lead to grade inflation; the quality of students’ work is just as unreliable as their test scores, and more subjective; emphasis on attendance and graduation can encourage teachers to be less strict and demanding. Sometimes, different measures can balance out one another’s flaws and make for a better final decision; but other times, useful but imperfect measures are diluted with even less useful ones, resulting in a worse decision.
Also, some of Ravitch’s objections to accountability practically suggest their own solutions. If teaching to the test is the problem, broaden the test to include more types of questions and more material. If the system is too punitive, make it less so (or even, within reason, provide lower-performing schools with extra resources, as Ravitch suggests — she supports “positive accountability”). If teachers are helping their students cheat, have teachers rotate classrooms on testing day (or give tests to the entire grade at once, with the teachers supervising both the kids and one another).
Ravitch strikes a decent blow against charter schools. While many studies have shown dramatic improvements, she explains, these results are largely due to demographics: The kids who go to charter schools are the kids whose parents cared enough to make them apply, so they score higher than the kids they leave behind. Researchers who take this into account (for example, by studying only students who entered random admission lotteries, comparing winners to losers) still sometimes find gains, but even these studies say nothing about what charter schools do for hard-to-educate children. Also, some of the improvement might come from “peer effects” — that is, getting kids who want to learn away from the disruptive influence of low-performing peers — an advantage that would disappear if everyone, including the low performers, went to charter schools.
We know that good charter schools can help children from dedicated families, but Ravitch points to two costs that come with this benefit: 1) Districts that allow charters end up with bad charters in addition to good ones, and 2) charter schools siphon off the best students, leaving the public schools to deal with the hardest cases.
As for the first problem, some districts probably should heighten their requirements for starting charters, and find ways to identify and close bad charters. As for the second problem, charters could start taking in more low-performing students, as Ravitch suggests, but these kids’ presence might keep the higher performers from making gains, and there’s no evidence that their own performance would improve. It’s worth trying, but if this tactic brings down smart kids without helping dull kids, it would be better to return to the status quo.
All this just scratches the surface of The Death and Life of the Great American School System. The book intelligently and readably addresses today’s education controversies, using a combination of anecdotes, case studies, and statistics. Whatever its flaws, it’s a must-read for education policymakers at all levels of government.