By allowing ads to appear on this site, you support the local businesses who, in turn, support great journalism.
Tradeoffs and payoffs:
Washington voters asked to make murky choice on class size
EdNext Chart


Washington voters will be asked in November to lower school class sizes, following a pathway forged by Florida voters in 2002. The push comes in the midst of education funding chaos in Washington, which currently has the fourth largest average class sizes in the country.

Class sizes nationwide have been creeping up since 2008, according to the Education Commission of the States, as financial pressures force districts to shave costs and the high personnel costs of small classrooms prove daunting.

While class-size reductions remain very popular with the public, borne out in poll data and at the ballot box in states like Florida and California, in elite policy discourse, the contrary viewpoint has become quite mainstream, even dominant.

Billionaire education crusader Bill Gates has voiced opposition to reducing class size, for example. And Malcolm Gladwell echoes that same view in his latest book, as does President Obama’s Secretary of Education Arne Duncan.

“We spent billions of dollars to reduce class size,” Duncan told ABC’s Andrea Mitchell in 2011, when we could instead give teachers higher salaries in exchange for larger classrooms, thereby attracting much more talented teachers.

“As a parent, we all love small class size,” Duncan said. “But the best thing you can do is get children in front of an extraordinary teacher. So other countries have higher class sizes but extraordinary talent in those rooms.”

In Washington, reducing class size would cost the state $4.7 billion over five years, with additional local property tax hikes also likely.

The state Supreme Court held a hearing Wednesday to decide whether to hold the legislature in contempt after it failed to provide increase education funding as the court had ordered in January, an order based on a state constitutional provision that requires “ample” education funding.

“The debate is over,” says Stanford professor Eric Hanushek. “Almost nobody, except those with some ax to grind, thinks spending money on large reductions in class size is the best use of funds.”

Jeremy Finn, a professor at SUNY Buffalo, dismisses Hanushek and his allies as ideologically “ultraconservative economists” who employ sketchy regression analysis to reach biased conclusions.

Will smaller class sizes juice student performance? Or would those funds be better spent on other priorities? It depends on which side you talk to — and both are equally emphatic they have the goods.


Duncan voices the central critique of large and costly class-size reduction, which is that tradeoffs matter and are going to happen.

Of course the ideal class size is one,” said Matthew Chingos, an education policy expert at the Brookings Institution. But that isn’t going to happen. The question, he argues, is whether one specific investment outperforms another.

For his part, Chingos would rather pay more for better teachers in slightly larger classes. He would also consider paying teachers more for a longer school year. Hanushek likewise suggests that additional funds should be spent buying out bad teachers’ contracts, raising average teacher quality.

Chingos points to a 2011 study published by the National Bureau of Economic Research, co-authored by Northwestern University professor Diane Whitmore Schanzenbach, a leading defender of smaller classes.

That study estimates how much it costs using specific interventions to nudge an underprivileged child into college. Class-size reduction proved to be among the most expensive interventions. Using an iconic class reduction experiment from Tennessee, the study landed on $400,000 per child, compared to Head Start at $133,000 and others ranging much cheaper than that.

Defenders of class-size reductions point out that getting a kid into college is not the only outcome to consider. But the larger point, Chingos says, is that tradeoffs are inevitable.

Tradeoffs also surfaced in a recent Education Next survey. Some respondents were asked whether they would rather invest in technology, teacher salaries or smaller classes, but they were not given any real numbers. In this case, teachers and the general public both favored smaller class sizes by large margins.

Another set of respondents was asked to reduce class sizes by three students, raise teacher salaries 13 percent or annually spend $10,000 in new books and technology. With this frame, the public split evenly among the three choices, while teachers favored higher salaries by a 10-point margin.

Even the authors of the Washington ballot initiative could not escape tradeoffs. The Washington Research Council points out that in a survey in which 45 percent of Washington schools responded, two-thirds reported they would need additional classroom space if I-1351 passes.

But I-1351 has an emergency outlet. Schools that can’t find the space for smaller classes can instead use the funds for other “school-based personnel who provide direct services to students.” Thus, many would end up spending the money on additional staffing without reducing class sizes at all.

Project STAR

The strongest evidence in favor of reducing class size comes from a controlled study in Tennessee during the 1980s, still the only randomly assigned controlled study on class size, called Project STAR.

Students were randomly assigned to either small or normal sized classrooms from kindergarten through third grade. Student performance and life outcomes were then tracked for years after.

Depending on which analyst is speaking, STAR was either a model of social science with clear results favoring small classes -- or it was a robust but flawed effort with ambiguous results.

“Even if you take STAR at face value,” Hanushek said, “it says that big reductions in class size at the kindergarten level may have a positive impact.” For any real results, he argues, the study required huge class-size reductions, and even then the benefits only appeared in very early grades.

“There is no evidence that smaller classes in second or third grade did anything for the kids in the experiment,” Hanushek said.

Design and execution flaws muddied the STAR data, Hanushek says. These included the loss of many kids from the study, often replaced haphazardly by new ones not part of the original group.

But Jeremy Finn, a professor at SUNY Buffalo who did much of the work analyzing STAR in subsequent years, disagrees. He sees STAR as compelling evidence for smaller classes. Finn particularly argues that the impact on students in later years requires attention.

“Kids who were in smaller classes in grades K-3 in project are still ahead academically in grades 6, 7 and 8,” Finn said. In fact, there is evidence that the smaller classrooms actually lowered dropout rates and raised college entrance rates as well.

Other evidence

To look beyond Project STAR, researchers have to piece together natural experiences and data sets. Chingos surveys this evidence in a 2013 journal article.

One of the most famous natural experiments occurred in Israel, where by law a classroom could be no larger than 40 students. If the classroom tipped to 41, it had to split, resulting in dramatically smaller classes.

Proponents point to the data from Israel as strong evidence for smaller classes. Chingos notes that, while the study design is strong, the fact that it is based on comparisons of classes with up to 40 students complicates its relevance to policymakers in the U.S., where classes tend to be significantly smaller.

Chingos also points to a similar study in Connecticut, which used both natural variations and classroom splits forced by maximum class size rules. That study found no evidence of even moderate effects from classroom size changes.

In a recent paper for the National Education Policy Center, Schanzenbach calls the Connecticut study an “unresolved puzzle,” and points to research in Sweden, Denmark and Bolivia with opposite results.

Finn’s own favorite ongoing study is the SAGE project in Wisconsin. “Their results are 100 percent consistent with Project STAR,” Finn says.

Chingos has also studied the Florida experience of gradually phasing in sharp class-size reductions beginning in 2003, concluding that Florida’s experience showed no benefit from smaller classes. Jeremy Finn disputes Chingos conclusion, arguing that he used flawed methods.

And so it goes.

One thing both sides can probably agree on it Chingos’ call for better data and more research. Even if replications of Project STAR are not possible, he says, given the enormous resources at stake, states should consciously stage policy shifts to create better natural experiments.

Key variables

Recognizing that tight budgets are driving larger classes for many states, Finn says he is currently working with a co-author on a book that will try to isolate the variables that drive smaller class size benefits, finding ways to replicate those effects in a larger setting.

“Class sizes are going up due to economic considerations,” Finn said. “There’s nothing we can do about it any more than we can fight the weather. So how can we take these dynamics that make small classes better and make them work in a larger class.”

What are those variables?

Teachers in a smaller class know their students better, Finn said. In the STAR study, he said, researchers expected special education referrals to go down because teachers would be better able to cope. Instead, referrals went up, because teachers were better able to identify signs because they knew the kids better.

Teachers also go into more depth for the whole class, Finn adds, because they have extra time because they are not playing whack-a-mole with divergent learning paces and discipline issues.

Smaller classes mean that the critical mass of disruption is less likely to form, teachers feel less threatened, misbehavior goes down, and student engagement and learning go up.

“Teachers can cope with moderate levels of misbehavior in a smaller classroom without being thrown for a loop,” Finn said. “So the punishment level, the anxiety about being yelled at goes down, and kids can act like kids without fear of being sent to the office.”

Not surprisingly, he adds, teacher morale goes up and retention of better teachers is more likely.

“I would take all of those pieces and say that every one of them may explain a part of what makes smaller classes more effective,” Finn said.