A misadventure of the correlation coefficient

Abstract

The correlation coefficient gauges linear association between two variables. However, interpreting its value depends on the question at hand. This article argues that relying on the correlation coefficient may be irrelevant for many neuroscience research tasks. When the experimental dataset is contextually suitable for binning-averaging, other indicators of statistical association could prove more suitable.

Keywords: Pearson’s R; binning; brain imaging; colocalisation; linear regression; statistical association.