Having collected a set of multivariate data, the first step is to process the data and carry out a preliminary data analysis in order to get a general 'feel' of the data. These preliminary procedures are relatively straightforward, but it must be emphasized that they form an essential part of any multivariate analysis. Most data are recorded manually on data sheets. Unless the numbers of individuals and variables are small, the data will probably be analysed on a computer. This means that the data must go through three stages: coding, punching, and editing. Having recorded the data, it is then essential to look for suspect values and errors of various kinds. The general term used to denote procedures for detecting and correcting errors is data editing. This includes checks for completeness, consistency and credibility. With a univariate set of observations, the analysis usually begins with the calculation of two summary statistics, namely the mean and standard deviation.