Multiple regression and its alternatives

Abstract
Although multiple regression is popular with social and behavioral scientists, the technique is not appropriate in most cases due to the level of variable measurement or multicollinearity. This article examines seven methods: multiple regression, discriminant analysis, logistic regression, analysis of variance, logit analysis, factor analysis, and multidimensional scaling. It then compares them using computer-generated data for a hypothetical election.

This publication has 6 references indexed in Scilit: