Few education reforms make as much sense on a gut level as giving teachers fewer students to teach.

The idea is popular with parents and politicians alike — at least 40 states have carried out some kind of class-size reduction in the past 15 years — and the Legislature in Washington has pledged to reduce average class sizes in kindergarten through third grade to 17 students by the fall of 2017.

Initiative 1351 on the Nov. 4 ballot would go even further, lowering average class sizes to 25 for grades four through 12 in Washington’s schools (with smaller sizes for schools where the majority of students come from low-income families).

According to the latest federal data based on teacher surveys, the average class size is 24 in the state’s elementary schools and 30 in secondary schools.

But despite more than four decades of research in the U.S. and abroad, the effects of this simple idea about how to raise student achievement have been hard to isolate and measure, leading to academic squabbles over its value.

Researchers generally agree that lower class sizes, at least in the earliest grades, are linked to positive educational benefits such as better test scores, fewer dropouts and higher graduation rates, especially for disadvantaged children.

They disagree, however, on whether those benefits outweigh the costs — I-1351 would cost nearly $5 billion through 2019 for more teachers and other school staff — fueling a highly politicized debate that becomes more intense when money is tight.

In recent years, researchers have been trying to figure out why smaller class size works, how it works and who benefits most.

Nailing those questions would help educators, policymakers — and the public — understand what else they need to do besides just shrinking classes to get the biggest bang for the buck.

The studies, based on classroom observations and interviews, have revealed some surprising insights:

• The most obvious explanation for why reducing class size works — that teachers give students better, more-tailored instruction in smaller classes — probably isn’t the reason why achievement goes up. Teachers for the most part don’t change their practices automatically when their classes have fewer students.

• Students behave better and pay more attention in smaller groups, and this may account at least initially for the gains. For example, it’s harder for a couple of troublemakers in the back of the room to derail the class when they can’t hide in a crowd.

• Reducing class sizes can have the potential to make a big difference for students only if teachers get the training and administrative support to take advantage of the situation by changing how they teach and how they interact with parents.

Tennessee study

The most persuasive class-size research in the United States comes from a large experiment in Tennessee that found students in small classes outperformed students in larger groups, even when teachers had the help of an aide.

Project STAR (Student Teacher Achievement Ratio) was conducted in Tennessee between 1985 and 1989 and involved more than 1,200 teachers and almost 12,000 students, according to Jeremy Finn, of the University at Buffalo (State University of New York), a statistician who helped set up the experiment and publish its results.

While no study of such a complex subject is immune from criticism, the results from Project STAR are highly regarded because teachers and students were randomly assigned to classes of differing sizes.

That kind of experimental design is considered the gold standard because it gives researchers confidence that the effects they are seeing stem from the change (smaller class sizes) and not some other factor.

The Tennessee results inspired California and Wisconsin to carry out statewide class size-reduction projects in lower grades in 1996, a time when state governments were enjoying surpluses.

There’s scant research on the effects, positive or negative, of reducing class sizes in the upper grades because the variables are much harder to pin down, Finn said.

Wisconsin achieved similar results to Tennessee, but California did not, which showed that simply making classes smaller is not all that needs to be addressed.

In 2003, Finn co-authored a paper that identified a gaping hole in the puzzle:

“Despite the many studies that show positive effects, research has yet to come up with a consistent, integrated explanation for the gains attributable to reduced class size,” according to the paper, published in the journal Review of Educational Research.

The most intuitively satisfying explanation — that teachers give students more individualized instruction in smaller classrooms — didn’t pan out when researchers observed what actually happened in smaller classes.

Several studies have found that while teachers may have more interactions with students, they tend to teach the same way they always have, regardless of the size of the class.

Finn and his colleagues proposed a different explanation, which they believe better fit the evidence from the studies and also jibed with classroom observations: Students behave better and participate more often when they can’t hide in the back of the classroom.

He saw the change himself visiting classrooms in Buffalo.

“In a big class, everybody in the back of the room is talking and giggling, and the little kids are throwing things at each other,” Finn said. “But in a small class, the first thing a teacher says is ‘Let’s all bring our chairs around me here in a circle. ’ ”

Smaller, quieter classes (fewer than 20 students) may have their biggest effect on kids who are inattentive and try to avoid looking the teacher in the eye. That’s because they can’t hide either.

Using Project STAR data, Finn compared the academic performance of fourth-graders considered to be disruptive, inattentive or neither.

“Most people I talk to predict that the disruptive kids are the worst, but they’re not. The inattentive, withdrawn kids are by far and away poorer students than all the others,” Finn said. “If you want to get lost in the back corner, whether you’re disruptive or not ... you disconnect yourself from any instruction at all.”

But if class-size reduction works because students change their behavior, wouldn’t it work even better if teachers and principals changed what they’re doing, too?

That’s the question Elizabeth Graue and her colleagues at the University of Wisconsin-Madison have been studying at schools involved in a project called SAGE (Student Achievement Guarantee in Education).

Wisconsin initiative

The SAGE program began in 1996 with 30 low-income schools around Milwaukee and eventually spread to almost 500 schools around the state as part of a broader effort to combat poverty.

Graue and her team have studied several SAGE schools to get a handle on what it takes for teachers and principals to make the most of small classes.

They found wide variations.

For example, sometimes schools would put two veteran teachers in the same room, which lowered the class size on paper.

But in practice, if one teacher was doing administrative work while the other was teaching all the kids, then the class size had effectively gone up for those kids.

It also didn’t work well if teachers didn’t have enough physical space.

“We saw classes that were in former locker rooms,” Graue said. “I don’t know how many classrooms I saw where they used a row of bookcases as a wall ... and anytime the other class did anything noisy, everything ground to a halt in the class next door.”

Smaller classes worked best when teachers received training in how to better tailor instruction to each student’s needs and when they spent more time getting to know their students’ families.

“Class-size reduction alone will only get fewer children in a class,” Graue said. “It doesn’t translate directly to a change in achievement.”

The same year Wisconsin launched SAGE, California rolled out its own statewide class-size reduction, which wasn’t as successful as SAGE or Tennessee’s Project STAR.

Critics have pointed to California’s rapid hiring of inexperienced teachers and the lack of physical space to accommodate the smaller classes, which affected kids at low-income schools the most.

And even in Wisconsin and Tennessee, there were wide variations in the effect that smaller classes had from building to building.

“Often it comes down to the individual school,” Graue said. “I don’t think any state program can be identified as a slam dunk.”

John Higgins: jhiggins@seattletimes.com or 206-464-3145. On Twitter @STjhiggins.

Four weeks for 99 cents of unlimited digital access to The Seattle Times. Try it now!