The Evidence on Teaching

How much survives?

Most education research can’t tell teaching apart from genes and selection. Don’t take that on faith — apply the skepticism yourself: three bars (design rigor, who ran the evaluation, how long effects had to last), and every verdict recomputes live from the studies that survive.

Before you drag anything: predict how many of the 51 decisions survive with every bar at its strictest. Our own retrieval-practice verdict says the guess will make the answer stick.

drag →

every study in the base (A–D)

Who ran the evaluation?

every study, regardless of who ran the evaluation

How long must effects last?

including effects measured only at end of treatment

1042 / 1042
studies still standing
88 / 88
verdicts still supported
The corpus · 1042 studies · click any square
The decisions · 88 · click one to open its evidence

These bars are the archive’s published default weights — the methodology calls them defaults, not theorems.