Class Size and Student Outcomes, by Matthew Chingos, sums up what we have learned about the importance of class size.
Lesson 1: Parents, teachers, and the public at large are all convinced that small classes are wonderful and are prepared to spend oodles of money to get shrink class size.
Lesson 2: Small class sizes probably do have a small positive effect on student outcomes–emphasis on the word “small.” They’re not worth the money.
Chingos reviews the pile of studies done on the topic. Since the number of really good studies is limited we can’t be entirely sure what reducing class size does. The clear weight of the evidence is that the class size effect is small, but the evidence is not completely conclusive.
These studies are hard because you can’t just compare student outcomes and class size. Well-off communities can afford to buy smaller classes and well-off communities have easier to educate kids. An apparent correlation between class size and outcomes might just reflect the confounding effect of “well-offness.” (You can probably think of a bunch of other factors that could have a similar confounding effect.)
Chingos puts serious weight only on studies in which class size was varied by intentional or unintentional randomized experiment. In the early part of the 20th century there were a number of small scale class size experiments. They generally found no effect of reduced class size. But perhaps enough has changed in nearly a century that these should be discounted. The one large, more recent, intentional class size experiment was the Tennessee STAR experiment. STAR did find an advantage of smaller classes, probably a large enough effect to justify spending money to reduce class size. (Chingos reports that the STAR experiment had a direct political impact on class size decisions in several states.)
Chingos then walks through the results of several “natural experiments.” As an example, suppose a small school in a district with a strict limit of 25 students per class has one 25-student class one year and then one additional student enrolls the next year. The school would have two classes of 13 students each the following year, and we’d know that the reduction really just reflected random chance. Chingos looks at several studies which look at variants of this kind of natural fluctuation.
The summary natural experiment finding? Probably not much effect of smaller class sizes.
Here are Chingos words of summary:
…the weight of the existing high-quality evidence indicates that although smaller classes may represent a cost-effective investment in some circumstances, many school systems in the U.S. have overinvested in class-size reduction. In other words, there are likely many circumstances in which modest increases in class size would benefit students if the resources were reinvested in more cost-effective interventions…