Regression Bias in Using Solar Wind Measurements

Simultaneous solar wind measurements from the monitors, WIND and ACE, differ due to spatial temporal structure of wind. Correlation studies that use these as input may infer an incorrect correlation uncertainties arising this structure, especially at extreme rare values. In particular, regression analysis will lead a function whose slope is biased towards mean value measurement parameter. This article demonstrates bias by comparing simultaneous ACE measurements. A non-linear between them leads perception underestimation values one on average over other. Using numerical experiments, we show popular techniques such linear least-squares, orthogonal are not immune bias. Hence while using parameters independent variable in or analysis, random uncertainty can create unintended biases response dependent variable. More generally, effect impact both event-based, statistical magnetospheric forcing.