Problemas de cómputo de la corrección por sesgo en el caso lognormal

Abstract

When a logarithmic transformation of the dependent variable is made in order to obtain a linear relation with the independent ones, to apply the general linear model the lognormal case is reached. The predictions of the dependent variable values using such estimation are biased; so, it is necessary to correct them. This paper deals with the problems of such a correction, that appear when the computation program finds the inverse of the moments respect to the sample means matrix, or the correlation matrix instead of the matrix of the moments respect to the origin.