Equal-percentile cutoffs produce deeply unequal measured categories

Evidence

Because the underlying data used by the UNDP are highly skewed, percentiles produce quite uneven categories. Even normal distribution data would produce unequal cutoffs based on equal percentiles. This is why second-order linear thinking using percentiles produces first-order, quite unequal categories. — Gary Goertz, Social Science Concepts and Measurement, 899

Categorization

• SEED: Angle: why the most ‘neutral’ statistical move — cutting data into equal percentiles — can manufacture the most lopsided categories.