La Trobe University Postgraduate Association Inc. (LUPA)

Multiple regression is generally used to study non-causal relationships between measures gathered from a single group of participants. The Pearson product-moment correlation (r) is the most common description of the relationship between two interval-scaled variables. It is important to understand the basic assumptions underlying this descriptive statistic because it is fundamental to more advanced correlational analytical techniques. Simple regression is used to illustrate the linear equation model (i.e., Y = bX +c) where one (independent) variable, X, to predict the variation in the dependent variable, Y. Errors of prediction, called residuals, are analysed to determine the goodness of fit of this prediction. The linear model is generalised, in multiple regression, to use several independent variables to predict one dependent variable. There are several kinds of multiple regression techniques. The technique we choose depends on the purpose of the data analysis is exploratory or confirmatory.