Services on Demand
Article
Indicators
Related links
- Cited by SciELO
- Similars in SciELO
Bookmark
Revista Varianza
Print version ISSN 9876-6789
Abstract
COA CLEMENTE, Ramiro. Consecuencias de alta multicolinealidad en un modelo de regresión lineal. Revista Varianza [online]. 2019, n.16, pp. 22-27. ISSN 9876-6789.
Abstract In this article, some consequences of the high multicollinearity among covariates present in the systematic part of a linear regression model are reviewed and illustrated. For this purpose, two models are compared. In the first there is no problem of multicollinearity, that is, the covariates are linearly independent. In the second model there is the problem of high multicollinearity, that is, the covariates are very linearly associated. Analyze four types of consequences: (i) on the magnitude of the regression coefficients, (ii) on the sums of additional squares, (iii) on the magnitude of the standard errors for the coefficient estimators and (iv) on statistical tests of the coefficients. In the presence of high multicollinearity among the covariates of the model, these consequences can lead to erroneous statistical inferences and consequently to incorrect conclusions.
Keywords : Multicollinearity; perfect multicollinearity; high multicollinearity or imperfect multicollinearity.