Hotelling and Wilson on Statistical Inference: Local Attitudes and Universal Dreams

Abstract

During the early 1920s, a scattered group of American scholars, among whom were Harold Hotelling and E. B. Wilson, started a joint effort to develop statistical instruments in economics. They became interested in disparate topics such as the statistical determination of demand curves, business cycles, yield forecasting, population growth, and many others. Hotelling and Wilson shared much: both were trained in mathematics, both were convinced of the importance of developing inferential statistical tools based on probability theory, and both were at odds with the common sentiment of the community of statisticians contemporary with them, a community that was highly skeptical about statistical inference grounded on probabilities. Their trajectories, however, differed, diverted by their entanglement with Stanford for Hotelling and Harvard for Wilson. Confronted with different local realities, they came to develop profoundly different approaches that characterized the debates of the 1920s and 1930s on statistical inference.