A flexible mixed model for age-dependent performance: application to golf

Abstract

We present a new mixed linear model for the relationship between age and performance. The model allows for random effects at the nodes of a barycentric interpolation, such that performance evolves with age in a non-prescriptive way. We use the model to investigate the effects of age on performance in golf and find that performance peaks in the 30s and then declines after that. We disaggregate performance into its constituent components and find that driving, which tends to require power and speed, deteriorates consistently from the early 20s, whilst putting, which requires touch and finesse, remains strong until the late 40s. Our model can be used in other settings, and requires only that measures of performance exist.