In a mixed effects model, the fixed effects are the coefficients estimated as unknown constants that apply to everyone: the population-level intercept and slopes a stakeholder would ask about. They are what remains of ordinary regression inside the mixed model.
The word carries a second, related meaning in econometrics, where a “fixed effects model” estimates a separate dummy coefficient per group instead of drawing group effects from a distribution. That estimator makes no assumption that group effects are uncorrelated with the predictors, which is why it is often preferred for causal readings even at the cost of noisier per-group estimates.
