Definition
Plain language
A statistical adjustment that soaks up baseline differences between groups — like separate countries or time periods — so they don't muddy the comparison.
As stated in the literature
Model terms (e.g., per-country or per-wave dummies) that absorb group-level confounds in a regression; adding or omitting them is one of the defensible analytical forks whose choice can flip a study's conclusion.
Also called: country fixed effects
Why it matters: They keep baseline differences between groups from masquerading as real effects, but choosing whether to include them can flip a study's conclusion.
For example, when comparing income across several countries, fixed effects soak up each country's overall wealth level so it doesn't get mistaken for the effect you're actually studying.
Heard on the show
“Next: introduce country and wave fixed effects.”Episode 196 — AI Agents Reached Opposite Conclusions From the Same Data — and Passed Review