Journal Highlights

New statistical model emphasizes role of spatial correlation

Editor’s Highlight—


This paper presents a statistical methodology to assess uncertainties in ocean chlorophyll trends. It provides strong evidence that significant differences exist in regional and global trends when compared with linear temporal models that have been widely used in other work. This study’s departure from the typical frequentist approach and its embrace of a Bayesian point of view is timely and welcome. There is growing (albeit far too slowly) awareness that the shortcomings of frequentist statistics, particularly through the incorrect use of null hypothesis significance testing, makes this paradigm ill-suited for scientific applications. Bayesian statistics allows an iterative process of data collection, data and hypothesis based inference, debate. Overall, this statistical model is a new approach and the inclusion of spatial correlation in determining the trends is of particular interest.