Data-driven categories often measure the data rather than the intended concept.

Evidence

From a semantic point of view these are all problematic because they relatively mindlessly use empirical distributions of data to form categories or do standardization. Hence the categories really are empirical data descriptions which may only be weakly related to what scholars, societies, organizations mean by the underlying concepts. — Gary Goertz, Social Science Concepts and Measurement, 902

Categorization

• DEFINE: Distinguishes empirical data descriptions from the concepts they purport to operationalize.
• QUESTION: How often do index categories drift from their stated concepts without anyone noticing? Worth chasing in concrete index case studies.