Files
Abstract
Multicollinearity is one of several problems confronting
researchers using regression analysis. This paper examines the
regression model when the assumption of independence among
Ute independent variables is violated. The basic properties of the
least squares approach are examined, the concept of multicollinearity
and its consequences on the least squares estimators
are explained. The detection of multicollinearity and alternatives
for handling the problem are then discussed. The alternative
approaches evaluated are variable deletion, restrictions on the
parameters, ridge regression and Bayesian estimation.