It's interesting to note that the class size effect has been measured for CA's 5,000 elementary schools. Buried in a technical report are the regression coefficients of selected variables that predict the School Characteristic Index (SCI). Among the socioeconomic variable that schools have no control over (parent education, free/reduced price lunch, ethnic composition) are two variables that a school can control - Average Class Size for K-3 and for 4-6. The coefficient for K-3 is -0.1644 meaning that a school's SCI will increase 0.1644 points if the average class size is reduced by one student.
See Coefficient Model 1 on page 4 of
But we're really not concerned with SCI. We'd really like to know how class size might affect a school's Academic Performance Index (API). No problem! There is a formula for moving from the SCI to the predicted API. Predicted API=8*SCI -600. Therefore the expected API gain for reducing class size by one student (20 to 19) in grades K-3 is a miniscule (8*0.1644) 1.3 points.
The gain is even smaller for class size reduction in grades 4-6.
For the PA members - CA's 1.3 point gain is about the same as a 3 point gain on the PSSA.
I'm always amazed that the popular wisdom says that class size reduction is important for improving educational outcomes. If I were a school board member I'd be investigating whether it might make sense to let class sizes increase at the elementary level and invest the resulting savings on some other more effective educational initiative. (If I remember correctly, CA had a referendum passed a decade or so ago mandating class size reduction which was later reversed.)
No, this is not an argument for increasing class sizes. It might be an argument for spending class size reduction money on more efficient methods of improving student performance, however.