Regression and causality

Abstract

Regression is a long established statistical tool that is widely used in phenological research for two main purposes i) to detect changes in the timing of phenological events and ii) to relate the timings of phenological events to one or more environmental conditions, typically temperature. Regression provides a flexible and robust method for the analysis of phenological data but can be influenced by start and end years and the length of the data series being analysed. In addition, spurious relationships can arise by trawling through large numbers of environmental variables without consideration of the likely mechanisms driving phenology. This chapter is illustrated using examples of flowering phenology from the United Kingdom.