Why this is mistaken
A clean output can feel like a measurement because it is specific and reproducible. Reproducibility proves what the stated inputs and rules produce; it does not prove that the inputs describe the world or that omitted mechanisms are immaterial.
Require four labels: identity, model implication, empirical estimate, and observation. Then ask which assumptions generate the output and which data could test or calibrate it. A model may still be useful when its value is to isolate a mechanism rather than predict a precise outcome.
The correction is not “models are false.” It is to match the claim to the model's purpose, assumptions, evidence, and domain.
Where to watch
When this mistake may appear
- A diagram yields a clean equilibrium or comparative result.
- A spreadsheet or simulation returns a precise number.
- The model omits institutional or distributional details.
Check your work
Your work may contain this mistake if:
- Uses predicted, estimated, observed, and caused interchangeably.
- Omits the conditions held constant and the model domain.
- Treats numerical precision as evidence of empirical accuracy.