An Overview of Partial Least Squares

Abstract

Partial least squares analysis is a multivariate statistical technique that allows comparison between multiple response variables and multiple explanatory variables. Partial least squares is one of a number of covariance-based statistical methods which are often referred to as structural equation modeling or SEM. It was designed to deal with multiple regression when data has small sample, missing values, or multicollinearity. Partial least squares regression has been demonstrated on both real data and in simulations (Garthwaite, 1994, Tennenhaus, 1998). It has been very popular in hard science, especially chemistry and chemometrics, where there is a big problem with a high number of correlated variables and a limited number of observations. Its use in marketing has been more limited although data has similar problems (Ryan, Rayner, & Morrison, 1999). This paper provides a brief overview of partial least squares (PLS) and its use as an analytical method in marketing research.

Keywords: partial least squares, structural equation modeling, lisrel, methods, soft modeling