ABSTRACT
This chapter classifies problems to be analysed by correlation and regression analyses from the other problem types because, although they share some similarities with questions of comparison and association, they are very different in other ways, and so would not sit easily within either of the other sections. The chapter demonstrates clearly where each technique should be used correctly, and, in so doing, reduces the amount of confusion and the number of errors in their use. It describes correlation methods to examine sequential relationships in which no causal effect is inferred. The chapter introduces regression methods to analyse cause and effect relationships. Regression analysis is used to characterise causal relationships between one dependent (response) and one or more independent (or predictor) variables. All regression models, whether they involve a single predictor (simple regression) or several predictor variables, must be validated as statistically significant and appropriate before they can be used.
