This technical note presents a minimal diagnostic analysis of the relationship between calibration and cross-context comparability. Using a controlled synthetic setup and a small empirical illustration, the note shows that locally calibrated models or measurements may remain internally stable while failing to support transferable comparability across regimes, datasets, or interface conditions. The focus is not on proposing new models or improving calibration procedures, but on clarifying a structural distinction: local calibration does not guarantee the existence of a shared descriptive currency across contexts. The analysis highlights conditions under which: – local agreement remains stable, – apparent consistency persists under restricted evaluation, – but cross-context comparison becomes structurally underdetermined or misleading. The result suggests that comparability must be treated as a separate, non-trivial achievement, rather than as a direct consequence of calibration quality. The note is intentionally minimal and reproducible, and is provided together with a small code and data package to allow independent verification of the diagnostic behavior.
Danilo Tavella (Tue,) studied this question.