Why Edtech Platforms Should Log Dosage, Not Just Engagement, From Day One

A Dartmouth research team found that tracking how many lessons students completed — rather than simple engagement metrics like logins or time-on-page — revealed that multiple-choice quizzing produced no measurable learning, while constructed-response quizzing did. The key variable in their study was 'dosage,' defined as the number of lessons a student actually completed, correlated against exam performance. Drawing on this finding, the edtech project doerkit is designed from the outset to record two core data structures: an events table and an attempts table, both tied to a student identifier and timestamp. This minimal schema makes it possible to run dosage-versus-outcome analyses on any cohort once exam scores are available, without needing a separate data-collection effort later. The article argues that platforms instrumented this way generate their own efficacy evidence as a byproduct of normal use, creating a compounding data asset that strengthens learning claims over time.