Regression analysis. At the three-route crossroad

Abstract

Regression analysis is a rather popular and uncomplicated method of analyzing experimental data. If the results of correlation analysis tell us about the presence of a statistically significant relationship between the studied features, the results of regression analysis represent a graphical description of the nature of this relationship. The article considers the points that authors usually do not pay attention to, but to which the editor should pay attention, and the reviewer simply must do this. Among such points are false predictions, non-Gaussian distribution of residuals, and lack of confidence intervals.