Instrument Like a Learning Scientist

Short summary

The article argues that edtech platforms should log dosage (countable units of work completed) and outcomes (pass/fail scores) from day one, not engagement vanity metrics. A Dartmouth study found multiple-choice quizzing produced no measurable learning while constructed-response did — a finding only visible because dosage was tracked against exam performance. The author provides a minimal SQL schema and argues this data asset compounds into the real moat for institutional edtech sales.

• •Log dosage and outcomes from commit one, not engagement vanity metrics
• •Dartmouth study found constructed-response quizzes correlated with exam scores while multiple-choice did not
• •Two database tables are enough to start accruing an efficacy evidence base that compounds over time

Generated with AI, which can make mistakes.