When Good Science Goes Bad

- Jennifer Stephan

 

Researchers argued for decades about whether class size matters for learning. Then an experiment in Tennessee in the 1980s plainly showed that students had learned more in classes with fewer students. Bingo!

Inspired by the strong evidence, California rushed to adopt a statewide program to shrink classes. How’d it work out? It didn’t. So many things went wrong. Small classes in California were a big disappointment. But confronting the unexpected is one way science progresses.

A Super-STAR Study

Before 1985, plenty of class-size studies existed, but their design led people to draw different conclusions. Then a unique (some might say drool-worthy) opportunity to find a clear answer arose. It was a randomized controlled experiment known as Project STAR (for Student/ Teacher Achievement Ratio). A randomized controlled experiment assigns participants to either a “treatment” or a “control” group based on chance, like by a coin flip. The treatment group receives something (such as a medicine or program) that the control group doesn’t. Then researchers compare the two groups. This process is a big deal because it measures if a treatment really works. The STAR experiment asked if students in kindergarten through third grade would learn more in classes with fewer kids. Over four years, STAR randomly assigned 11,600 kids and 1,300 teachers in 80 Tennessee schools to a regular-sized class (the control group) or a small class (the treatment group). Small classes had many fewer students, about 15 compared to about 22 students in regular-sized classes.