ABSTRACT
In this chapter, the author addresses a question that every researcher has to deal with: What is the right statistical test for his purpose? The Binomial model presumes that every member of the set is free to vary. If the set consists of words, the model exaggerates the possible range of variation and renders the test insensitive. In contingency tests, data are expressed in the form of a ‘contingency table’ of frequencies. The independent variable has a discrete set of categories and produces a discrete frequency distribution for each column of the dependent variable. From the perspective of significance testing, confidence intervals tell us about the certainty of observed data points, and help us decide whether points are significantly different. The graph reveals an interaction between the subject and the choice of who or whom for writing not found in the spoken data.
