A linear middle of a wealth range differs from the categorized middle.

Evidence

One can see that the two conceptual mappings are asymmetric: one is not the mirror of the other. They are also definitely nonsymmetric in relation to the underlying GDP per capita data. For purposes of illustration, figure 4.5 gives the standard defaults often automatically applied. One can draw a straight line through the data, which is implicitly how poverty–wealth is considered in many statistical analyses; see the line in figure 4.5. The “middle” of the GDP per capita range is not the middle categorization of the World Bank. — Gary Goertz, Social Science Concepts and Measurement, 627

Categorization

• SEED: Angle: how the ‘middle class’ is an artifact of whichever cut points an analyst picks, with political consequences.
• DEEPTHINKING: Quietly undermines the assumption that statistical linearity and conceptual categorization can be reconciled—the mapping choice silently drives every downstream inference.