This study evaluates the algorithms underlying a static (non-dynamic) and an adaptive (dynamic) version of computerized practicing. Dutch secondary school students (N=1021) are randomly assigned to one version for the duration of one school year. Students using the adaptive version receive more difficult exercises, spend more time practicing, but answer fewer questions correctly. Both groups achieve similar test scores, but students performing sufficiently well prior to the intervention perform slightly worse when working with the adaptive version (SD=-0.08). Adaptive practicing specifically aims at better addressing individual learning needs, but can come at the expense of effective test preparation.
JEL Codes: A20, I21, L86
Keywords: Personalized Education, Computerized Adaptive Practicing, Field Experiment