Autocorrelation - Prevalence of identification of collinearity cause

The paper demonstrates that autocorrelation is an accidental statistical phenomenon, whose origin is the incomplete data base. It also shows that the attempts to redistribute factors interactions have focused on the development of methods of solving the effect rather than identifying the cause that generates collinearity. Three possible methods for collinearity removal are analysed comparatively. The premise for two of these methods is autocorrelation redistribution, and the third reveals the cause of collinearity and, implicitly, its cancellation. The three methods are named as follows: 1. Classic method [1,7]; 2. Method of Merce E., Merce C.C.[6]; 3. Method of Merce E., Merce C.C.[5] 3. Method of Merce E., Merce C.C.[5; It is demonstrated that the first two methods are conventional approximations on the distribution of factors’ interaction, with possible subjective consequences. The ideal solution is the use of a complete data base. If this is not possible, as is often the case with databases of economic or sociological research, solving can be the completion of information with theoretical values, obtained by adjusting the causal relationship, in the hypothesis of a certain regression model, a procedure that represents, in fact and implicitly, a way of redistributing the interaction on the influence factors included in the causal model