It is widely accepted that empirical research methods and statistics belong together. This is no different in phraseology. In contrast to the natural and social sciences however, no standardized statistical culture has yet emerged in this field. Empirical research in German phraseology during the last 15 years shows a variety of approaches and analysis methods, which often led to contradictory results. Part of this dilemma can be explained by the fact that statistical methods were ignored, too few participants were involved in the studies, the participants chosen were not representative for the intended group or the statistical tests were improperly used.
This article gives a basic introduction to the main statistical tests needed in phraseological research (i. e. chi-square test, t-test, analysis of variance and regression) and clear instructions on when and how to use them. Special attention is drawn to the importance of interactions between different factors (that can be tested with the help of analysis of variance).
With this article I hope to have been able to demonstrate that statistics are a necessary tool in phraseology and other linguistic disciplines: on the one hand to organize and to clearly present the data and on the other hand to provide more evidence for regularities in the data.