Simplified Calculation of a Linear Regression

Abstract

THE “simplified” method of calculating a linear regression put forward by Aldridge, Berry and Davies¹ is the well-known method of orthogonal polynomials which was put on a practical working basis by Fisher^2,3. Unfortunately, in bringing this procedure again to the attention of research workers, the authors reproduce a misapprehension when they deduce that “whereas it is considered that four points are a minimum for determining the slope of the line, the number of observations should always be even”, because observations at the centre point of an odd number of values of the independent variable contribute no information to the linear regression.