ABSTRACT
Traditionally, researchers often aim to isolate the infl uence of one variable on another. For example, they might examine whether poor nutrition in childhood leads to lower socio-economic status (SES) in later life. The fundamental diffi culty with this is that other variables which might infl uence a SES are ignored. In real life, variables rarely act independently of each other. An alternative approach is to explore the complex pattern of variables which may relate to SES. Numerous factors may be involved in SES including parental educational level, the quality of teaching at school, the child’s general level of intelligence or IQ, whether or not the child went to nursery school, the sex of the child and so forth. We rarely know all the factors which might be related to important variables such as SES before we begin research; so we will tend to include variables which turn out to be poor predictors of the criterion. Multiple regression quite simply helps us choose empirically the most effective set of predictors for any criterion.
