On the Path Analysis Techniques and Decomposition of Correlation Coefficients

Abstract

Path analysis is an extension of multiple regression technique used to evaluate causal model by examining the relationship among a set of variables. The process of decomposition of path correlation coefficients and estimation of path coefficients was investigated in this work. The outcome of a set of interrelated variables was generated using path diagrams and path coefficients. Decomposition of correlation coefficients into different effects was also carried out through the method of structural equations using path analysis theorem. In order to show the proportion of total variation of the dependent variable, the decision coefficient of a specified pathway was constructed from direct coefficient of determination. The path coefficients were estimated from the specified structural equations by ordinary least squares regression method, and it was deduced that each of the coefficients of determination had large effect on some other variables.