This chapter discusses different statistical procedures available for the analysis of stylistic and sociolinguistic variation in corpora. It reviews different approaches to variation, pointing out the common connection to the notion of ‘style’ understood as a particular way of speaking and using language. The statistics discussed include the t-test, ANOVA, Mann-Whitney U test, Kruskal-Wallis test, correspondence analysis and mixed-effects models. Each of these focus on a different type of sociolinguistic analysis and require a specific research design: for group comparisons, the t-test, ANOVA, Mann-Whitney U test and the Kruskal-Wallis test are used. Individual linguistic style can be explored using correspondence analysis. Traditional (Labovian) socioliguistic analysis with a focus on variation in a carefully defined linguistic context can be carried out using mixed-effects models, a technique that is still fairly new in corpus-based sociolinguistics.