Library
Models
PAX2 reference model
PAX2 · built from the persuasion experiments
Per-feature persuasion differences from a random-effects meta-regression, the spread of true effects, and a leave-one-study-out test of how predictable an ad is from its features.
meta-regressionpriorsderived
Files3 · 1.6 MB
LicenseCC0 inputs
What it can tell you
A prior for what a script's style is likely worth
An honest bound on how little style alone predicts
What it can't
Predict an individual video's persuasion