Definition
Plain language
A leftover alternative explanation that the experiment's controls didn't fully remove.
As stated in the literature
Variance in a measured signal still attributable to a nuisance variable after contrast construction and control projection — e.g. injury content persisting in a pain direction tested on explicitly pain-free injury sentences.
Also called: residual confounds
Why it matters: Unnoticed leftovers like this can turn a clean-looking result into a claim about the wrong concept entirely.
For example, a pattern meant to capture pain may still fire on "she broke her arm but felt nothing," suggesting it is partly tracking injury rather than pain itself.
Heard on the show
“The authors call it a residual confound.”Episode 267 — A Pain Axis, a Relief Button, and the Control the Paper Skipped