B.5.FM:8 - Common Anti-Patterns and How to Avoid Them

Recognizable failure	Repair
Fit a model whose variables cannot express the proposed intervention.	Recover the participants and relations through which the intervention could change the result.
Copy a familiar analogy while its operation has a different meaning in the present case.	Construct the needed correspondence and test the transferred consequence.
Aggregate cases before performing an operation whose permissions differ between them.	Keep the permission-changing distinction until that operation is complete.
Increase simulation detail while an omitted interaction determines the answer.	Obtain a limiting or contrasting case that exposes that interaction.
Give a specialist a broad modeling request that leaves the needed result unclear.	Supply the provisional account and ask for the contribution that would change the next decision.