The basics of determining the coefficients of a linear correlation

Abstract

The aim of the paper is to present the basic measures related to the analysis of relationships between quantitative variables used in econometric modelling and their selected applications. The following measures are discussed: the Pearson correlation coefficient, the multivariate correlation coefficient, coefficient of determination, partial correlation coefficient and semi-partial correlation coefficient. A homogeneous approach is applied to the measures presented. Each is defined as a linear correlation coefficient of relevant vectors derived from regression equations. Additionally, mutual relations between the coefficients are described. Bordered matrices have been applied to the calculations, which significantly simplified the process, while the Statistica 13.3 PL program was used to verify the correctness of the calculations. The issue is illustrated in the model of regression of salary growth in Poland in the years 2001–2019 with four covariates, estimated using the least squares method.