The class-size rule that creates a natural experiment

Question

A school district has a rule: any grade with more than 40 enrolled students must be split into two classes. So a grade with 40 students has one class of 40, but a grade with 41 has two classes of about 20 each. You want the effect of class size on test scores.

Explain how this rule enables a regression discontinuity design, why it is nearly random near the threshold, the key assumption, and the limitation.

Accepted Answer

The rule creates a sharp jump in class size. As enrollment climbs from 40 to 41, class size does not inch up; it suddenly halves, from about 40 students per class to about 20. That abrupt drop at the enrollment threshold is exactly what a regression discontinuity exploits: the "treatment" (small class) switches on discontinuously at 41. Why it is nearly random near the threshold. Compare grades with enrollment just below the cutoff (40) to those just above it (41). Whether a grade has 40 or 41 students is largely a matter of chance, one family moving into or out of the district, a single late…

The class-size rule that creates a natural experiment

Your answer

More Statistics questions