This is a Lexicon entry: the mechanism, when it shows up, and the countermeasure. The full correction, with real-world cases and the audit prompt, hasn't been written yet.
Definition
The error of applying a model - statistical, conceptual or formal - to a situation it was not designed for or that systematically omits relevant variables or makes false structural assumptions. Common in economics (models that assume away liquidity constraints or behavioural irrationality), medicine (models that exclude women from drug trials and then apply results universally) and machine learning (models trained on historical data that do not include future distributional shifts). A misspecified model can produce highly confident wrong predictions.
When it shows up
When applying a statistical, conceptual or predictive model to a new context. In financial modelling: when a model built on historical data fails on future data because its assumptions don't generalise. In engineering: when a simulation model is used beyond the conditions it was calibrated for.
Failure mode
Precise predictions in the wrong direction. Economic models that assume away human irrationality. Medical models that excluded half the population.
Countermeasure
Ask: "What assumptions does this model make and which might be wrong in this context?" Actively seek out the cases your model fails to explain.
Related traps
Also connected in the map7 more, locked
Credit & first seen
PopularizedA central concern in econometrics (Hendry, 1995) and the philosophy of science; discussed by statistician George Box and others.
Also revealed
Entry #154 of 631 in The Lexicon · see the full Lexicon