The Limits of Ordinal Skintone Measurement: A Population-Based Evaluation of Guide-Based Scales Using RGB Data

Research on skin tone inequality increasingly moves beyond broad racial categories, yet most empirical studies continue to rely on fixed skin tone scales that discretize a continuous phenotypic trait. This study evaluates whether widely used categorical skin tone instruments validly represent real human variation. Using averaged red–green–blue (RGB) values extracted from yearbook photographs of every high school senior in a single U.S. county, the study constructs a population-level, continuous skin tone distribution grounded in observed data. Principal component analysis identifies a dominant axis of tonal variation, and kernel density estimation reveals a smooth, unimodal distribution with no natural breaks. The Monk Skin Tone Scale and the Massey–Martin Color Guide are embedded into the same RGB space and evaluated against the population distribution using spacing analysis, projection onto the dominant component, and Mahalanobis distance tests. Results show that both instruments are ordinal in structure, unevenly spaced, and poorly aligned with population density. Many palette-defined color points exhibit near-zero probability of arising from the observed population covariance structure and violate the correlated channel relationships that characterize real human skin pigmentation. These findings demonstrate that ordinal skin tone scales and fixed color encodings, including hex-based representations, do not approximate actual people and are unsuitable for gradient-based inference. The study supports replacing categorical and display-oriented palettes with continuous, population-derived measurement frameworks grounded in observed human data.