Lost amid the national election coverage Tuesday night was the defeat of an important ballot initiative in Florida. Amendment 8 would have relaxed the state’s strict caps on class size, adding three to five students per class depending on grade level. (The amendment received about 55 percent of the vote, but it needed 60 percent to become law.) The current caps have proven costly to implement. The Orlando Sentinel described some of the struggles:

To free up money for smaller classes, districts cut electives, combined classes, shifted student schedules, and packed students into elective courses. The Seminole County school district even closed some schools to students in certain grades and bused kids to nearby campuses that had room.

Supporters of the caps argue that class size reduction is valuable enough for the state of Florida to cover the local districts’ added costs.

Conventional wisdom does seem to be that smaller class sizes help students do better at school, but scholars have had an awfully hard time confirming it. Back in 2003, Eric Hanushek searched the literature for rigorous econometric studies of class size impact on student outcomes. He found 276 different estimates of class size impact, but just 14 percent of them were both positive and statistically significant. Another 14 percent were significantly negative, and 72 percent were statistically insignificant—in other words, zero. Zero impact of class size reduction is the modal finding in the literature.

Advocates of smaller classes rely less on the econometric analyses surveyed by Hanushek and more on a couple of elaborate experiments—namely, Project STAR in Tennessee and Project SAGE in Wisconsin. Though both experiments did show moderate improvements for students in smaller classes, they say little about what class size reduction can do on a larger scale. The reductions in the experiment were drastic (often 15 or fewer students per class), the benefits tended to be concentrated in the lower grades and among poorer students, and—at least in the case of SAGE—other kinds of classroom reforms were implemented simultaneously.

At best, we can say that reducing class size may be beneficial for students with particular socioeconomic backgrounds, at particular times in their life, with particular kinds of teachers teaching a particular kind of curriculum. With no evidence that it can bring across-the-board improvements, a uniform statewide cap on class size for all students is a policy unlikely to pass any cost–benefit test.

Nevertheless, more experimentation would certainly be helpful. The more we know about which kinds of students may benefit from smaller classes—or from any educational change, for that matter—the better we can direct scarce resources. In fact, if parents rather than the government directed those resources, experiments with all kinds of different educational models would flourish. Parental choice in education, not statewide regulation, is the better approach to reform.