Credit: Illustration by Slug Signorino

As the parent of a 6-year-old in kindergarten, I’m trying to find out about optimal class sizes. I live in San Francisco, and since the mid-’90s, California has mandated minimal class sizes based on a 20-plus-year-old study from Tennessee. I don’t know if this is the best way to either improve children’s education or save money, since there seems to be a lot of infighting on both sides of the argument. So what’s the deal—is a class size of 22 first-graders better than a class size of 26? How about 20, or 16? —Rory H.

Interesting phenomenon at work here. We’ll call it dollars versus common sense.

Education experts had long conjectured that young kids learn better in small classes but lacked the research to prove it. In the 1980s, Tennessee Gov. Lamar Alexander, who would later become education secretary under George H.W. Bush, decided he’d fix that. The one government type in this whole sorry story who seems to have had a clue, Alexander persuaded the legislature to pony up several million dollars for Project STAR (Student-Teacher Achievement Ratio), a multiyear experiment involving roughly 6,500 kids and 330 classrooms in 80 schools throughout the state. The idea: Reduce class size for some kids in grades K through 3 from 20 to 25 pupils to 13 to 17 and see if they do better on standardized tests.

The consensus is they did. On average, reading and math scores for kids in small classes increased nearly 10 percentage points compared to those of students in regular-sized classes. What’s more, the improvement persisted even after the kids moved into larger classes, and minority kids’ numbers improved twice as much as everybody else’s.

Terrific, right? That’s what educators in California thought. Concerned about lousy reading scores and crowded schools and hoping to emulate Tennessee’s success, they pushed through the Class Size Reduction (CSR) initiative, which launched in 1996.

Common sense, had it been applied, might have suggested approaching CSR cautiously. Among the problems:

  • First, even if the Tennessee experiment did work as designed, nobody really knew why. Stanford economist Edward Lazear hypothesized that smaller classes are less prone to disruption, enabling teachers to spend less time maintaining discipline and more time teaching. If so, two obvious corollaries might be (a) kids in larger classes will learn just fine if everyone is well-behaved and (b) a cheaper way to improve achievement would be to add teachers and reduce class sizes in only the most chaotic schools.
  • Second, California wasn’t Tennessee. The small classes in Project STAR had on average about 15 kids compared to 22 in the regular classes, whereas the CSR initiative reduced the average California class size from 29, the highest in the nation, to 20. Sure, big reduction, but the research leading up to the Tennessee study suggested you didn’t see much benefit till class size got down to 15 or below.
  • Third and most important, the STAR program was just one study—the cheery outcome might have been a fluke.

A reasonable approach, therefore, would have been to start out with a pilot program to see if California could replicate Tennessee’s findings on a small scale before betting the ranch. With a state treasury flush with tax revenue from the 1990s tech boom, California educators launched CSR as a full-bore statewide program, ultimately hiring 23,500 additional teachers at a cost of well over $1 billion annually.

What happened? Let’s review:

  • If you want smaller classes, you need more classrooms, and California schools were already full. Many of the districts with the worst crowding couldn’t find space for more primary-grade classes and so got no benefit from CSR.
  • Hiring so many teachers so fast meant standards had to be relaxed, and the professional qualifications of California teachers noticeably declined.
  • Demonstrable impact on test results? By and large, none.

That’s not to say reducing primary-grade class size is useless, merely that it’s not a proven approach. One appreciates that parents clamoring for instant results don’t like waiting while the social scientists patiently try A, then B. But there’s really no other sensible way to do it.

Don’t worry, though. Californians, and Americans generally, won’t make the same mistake twice. Not because we’ve learned anything, but because then we had money, and now we don’t. —Cecil Adams

Have something you need to get straight? Take it up with Cecil at