One Size Fits All


When Bill Clinton unveiled his proposal to help public schools hire 100,000 more teachers, he said "every parent already knows" that education improves when class size shrinks. For anyone familiar with the techniques politicians use to discourage critical thinking, the simultaneous invocation of children and a universally acknowledged truth should have set off warning bells.

But most listeners probably just nodded, agreeing that smaller classes are indisputably A Good Thing. Leaving aside the assumption that all good things should be subsidized by the federal government, is it really true, as Clinton suggests, that smaller classes translate into better learning?

That premise, which underlies California's ongoing class-size reduction program as well as Clinton's plan, has strong intuitive appeal but little empirical support. "The evidence gives no reason to believe that reducing overall class sizes will have any effect on student performance," says University of Rochester economist Eric Hanushek, one of the nation's leading education experts. "I think the people who are saying 'let's reduce class size' are taking a very selective look at the evidence."

In a recently completed paper, Hanushek reviews the large body of research on this issue, including aggregate data, international comparisons, econometric analyses, and experimental studies. "The evidence about improvements in student achievement that can be attributed to smaller classes is meager and unconvincing," he concludes. "While policies to reduce class size enjoy political appeal, such policies are very expensive and, according to the evidence, quite ineffective."

The study most frequently cited by advocates of smaller classes is Tennessee's Student Teacher Achievement Ratio (STAR) project. Starting in 1985, STAR randomly assigned more than 6,000 children to small classes (13 to 17 students) or regular classes (22 to 25 students) and followed them from kindergarten through third grade. Each year, the kids in the small classes outperformed the other kids on tests of reading and math skills.

But Hanushek notes that the performance gap did not grow during the course of the study, as one would expect if a low student-to-teacher ratio produced continuing benefits. The advantage for the students in the small classes showed up after kindergarten and remained essentially the same after the first, second, and third grades. Indeed, it was still about the same at end of sixth grade, after three years in regular classes.

This pattern suggests that a small class provides a one-time boost in kindergarten, with no additional payoff after that. If so, hiring more first-, second-, and third-grade teachers, as Clinton proposes, is a waste of taxpayers' money.

It could be worse than a waste, if the drive to shrink classes leads school districts to hire unqualified teachers. "Already," reports The New York Times in a story about Clinton's proposal, "the city's lowest-performing schools, which are often in the poorest neighborhoods, have trouble attracting experienced teachers with full certification." If the choice is between small classes with bad teachers and big classes with good teachers, says Hanushek, the latter is clearly preferable.

In addition to the merits of the smaller-is-better approach, Hanushek questions the wisdom of top-down educational reform in general, whether at the federal or at the state level. "There needs to be a lot more experimentation," he says, and that can happen only if schools are free to innovate.

For trial and error to work, schools also have to be free to fail, which is impossible if they are always propped up by a captive market and a never-ending flow of taxpayers' money. "Enormous funding increases for U.S. public schools have not yielded improvements in student performance," Hanushek notes in the journal Educational Leadership. "Since schools use resources inefficiently, incentives should be linked to performance."

That is the basic idea behind charter schools, private school vouchers, and other attempts to give parents more choice in where and how their children are educated. Other things being equal, most parents probably would prefer smaller classes. But since resources are finite, many would be willing to accept bigger classes in exchange for better teachers, a stronger curriculum, or other features that are important to them.

Because values and preferences vary, these trade-offs are impossible to predict. Identifying the one best class size is therefore an exercise that makes sense only to arrogant technocrats. President Clinton, by the way, says it's 18.