Definition
Plain language
A statistical method for data where each person contributes many measurements.
As stated in the literature
Mixed-effects regression with random intercepts (typically per participant) plus fixed effects for conditions, used to separate within-subject variation from treatment effects in repeated-measures designs.
Also called: mixed-effects model, mixed models
Why it matters: Treating repeated measurements from the same person as independent inflates confidence and produces findings that don't hold up.
For example, if ten people each rate twenty conversations, a mixed model accounts for the fact that some raters are simply harsher overall before deciding whether one condition really beat another.
Heard on the show
“The mixed model gives p equals point zero eight two, with a confidence interval crossing zero.”Episode 249 — The Chatbot Knows Your Facts And Still Won't Mention Them