Regression to the Mean: More than a Statistic

This article looks at Galton's regression to the mean1 (R to M) from several traditionally unrelated but interwoven venues: (1) as a law of trait heredity; (2) as a statistical artifact infiltrating careless research designs; (3) as an illustration of cognitive bias. Hereditarians argue for the first of these, disputed by biogeneticists, who view R to M as a mere correlate of generational traits (e.g., IQ) decline. Research designers busy themselves with the second perspective, but to explain the concept, cavalierly adduce various organismic states that sum as "" error variance" Research on subjective probability finds R to M in unexpected places. Across these contexts, the weight of argument falls on R to M's confounding rather than causal role in vexing data analysis.