ABSTRACT
A GREAT DEAL of geographical analysis involves studying the relationships between two or more variables. This chapter focuses on those statistical techniques that enable us to measure and determine the strength of a statistical relationship between two variables. A number of different methods of correlation analysis are available to us, and Chapter 4). The power of the different types of correlation analysis varies, and it is generally recognised that the product-moment or Pearson’s correlation coefficient (r), which uses interval/ratio scale data, is the best in that respect. For example, it has been calculated that Spearman’s rank correlation coefficient, which uses ordinal data, is only 91 per cent as efficient as Pearson’s. This means that if in a sample of 100 cases from a bivariate normal population, the product-moment coefficient is significant, it will require a sample of 110 cases of the same population to achieve the similar reliability using the Spearman coefficient.
