When small isn’t beautiful

Few education policies are more popular than class size reductions. Alongside her faithful friend Laura Norder, Somala Classiz has appeared on the ballot in just about every state election over the past decade. And thus class sizes in Australia have steadily ratcheted downwards, gobbling up more money than any other educational reform.

The logic of class size reductions is easy to see. With fewer children in the room, teachers can spend more time with each student. Discipline challenges can be more easily managed, and lessons can be better tailored to the particular needs of the student.

Yet while smaller classes create the potential for better learning, there is no certainty that this potential will be realised. If teachers do not adapt their teaching style for a smaller group, there may be no improvement in performance. Indeed, it is even possible for a class size cut to reduce student performance. Smaller classes require hiring more teachers - and if the new hires are less effective than the incumbents, students could lose out.

In estimating the impact of class size on student performance, economists are naturally wary about drawing causal conclusions from the correlation between class size and student performance. If schools systematically stream gifted or struggling students into smaller classes, it will be difficult to know the true impact of class size on outcomes.

One way of solving this problem is to exploit class size rules, which grant schools an extra teacher when the number of students hits a given threshold. Across a number of developed countries, this approach suggests that once class sizes get below 30, students gain little benefit when another teacher joins the school and reduces class sizes.

Another research design is to randomly assign students into different sized classes, say by the toss of a coin. In the mid-1980s, Tennessee governor Lamar Alexander agreed with the state’s teachers to conduct a randomised class size experiment. If those in smaller classes did better, then class sizes would be reduced statewide. Although the experiment succeeded in showing that smaller classes raised test scores, some have worried about the incentives that Governor Alexander created. Can we be sure that the promise of across-the-board reductions didn’t influence how teachers behaved in the experiment?

In a new paper (PDF 89KB), Columbia University economist Jonah Rockoff brings some older evidence back into the debate, by reviewing the results of a series of randomised class size experiments conducted in the early-20th century. Prior to World War II, US education researchers conducted 24 randomised experiments to test the impact of class size reductions. In 22 of these experiments, there was either no difference in performance of children in large and small classes, or children in larger classes did better. Only two of the 24 experiments supported the notion that smaller classes improve student performance. Although these experiments might have had methodological shortcomings, they do suggest that modern policymakers should think twice before hitching their wagon to the class size horse.

Are there better policies than class size cuts? A growing body of evidence in the economics of education is pointing to the role of teacher quality as being paramount. And while salaries are not the only factor attracting skilled individuals into the teaching profession, higher pay does buy more effective teachers.

A rarely recognised fact is that if the education budget is not increased, smaller classes translate into pay cuts for teachers. In joint work (PDF 367KB) with my ANU colleague Chris Ryan, we have tracked teacher pay since the mid-1980s. We find that relative to other university graduates (or relative to all employees), average teacher pay has fallen by about 10 per cent. Over the same period, the student-teacher ratio (which closely tracks class size) fell by about 10 per cent. The simple story of the past two decades is that teachers have bought class size reductions from their own wallets.

Put another way, new teachers in the 1980s were paid about a tenth more than the typical university graduate. Today, new teachers earn approximately the same as the typical university graduate. And as earnings inequality has opened up in the non-teaching sector, the uniform salary schedule in teaching looks increasingly unattractive to talented youngsters. It is therefore hardly surprising that the average academic aptitude of new teachers has also declined.

So next time you hear Australian politicians spruiking smaller classes, ask why they’ve never been willing to put their rhetoric to the test with a randomised evaluation of smaller classes. Better-paid teachers don’t look so good on a manifesto, but it might just have a bigger impact in the classroom.