Evidence on what helps
Education is full of confident claims. A small number of practices have solid support. We try to know which is which.
Retrieval beats rereading
Trying to recall something strengthens memory more than reading it again. Low-stakes quizzing is cheap and well supported.
Spacing beats cramming
The same study time spread over days is remembered longer than the same time in one night. Course calendars can be built with this in mind.
Worked examples first
Novices learn faster by studying solved problems before solving their own, then shifting to independent practice as they gain skill.
Measure before you believe
When a tool is introduced, compare it with what was there before, on outcomes the course actually cares about. Be willing to hear that it did not help.
Reading a claim carefully
A new method is announced and the numbers look good. Four questions help. Compared with what: doing nothing, or doing something equally effortful? Measured how: a test the method was built to pass, or one the course already used? On whom: a small motivated group, or ordinary classes? And does it hold when someone else repeats it?
We write summaries that answer those four questions in plain language and label the result: well supported, promising, or not yet shown. Teachers can then spend their limited time on the practices that earn it.
The same discipline applies to our own work. When we introduce something into a course, we plan how it will be judged before it starts.