Back to basics

Both authors of this book are very visual in their approach to problems, as is evident in the number of figures we have used. It is not surprising, therefore, that we advocate the visual evaluation of every step of the analysis process. Plot the data, plot the results of analysis and, when fitting a model, plot the residuals. There are insights to be gained throughout the process.

The first step in any analysis is to plot the data. Many problems can be avoided by this simple step, combined with an awareness of potential problems and some thought. A common mistake in dealing with spatial data is that the x- and the y-coordinate axes (the columns in a data file) are not used in the right order during the analysis. Indeed, the input format needed varies with the statistical software or GIS package being used; for example, the location of the origin (0, 0) can be either in the upper left corner (as in a matrix array) or in the lower left corner (as sampled in the field). This difference in the location of the origin can result in analysing the mirror image of the data, which is not always an important problem but in some applications may be critical. Furthermore, plotting the data can provide obvious and useful information to guide the choice of which spatial statistics to use and which methods will detect the spatial structure of the data.