#A.19:5.2.4.3 Declare “incomparable” when appropriate.

Preface node heading:a-19-5-2-4-3-declare-incomparable-when-appropriate:20486

#Content

If for some critical Characteristic there is no valid NormalizationMethod to translate measurements between two contexts (e.g. the scale types are fundamentally different, or the measurement’s meaning doesn’t carry over), then the framework insists that we declare the states or metrics incomparable rather than attempting any fudge. No comparison should ever default to “close enough by name” or other heuristics. For instance, if one context measures “User Satisfaction” qualitatively and another quantitatively, and no monotonic mapping can be justified, one must simply say a user satisfaction state in context A cannot be compared to one in context B. Mark it incomparable and avoid any misleading conclusions. This rule guards against the natural temptation to compare things just because they have the same label or general intent, when in fact their measurement basis is different.