[Examples of pitfalls in statistical analysis--2. Parametric and nonparametric analyses].

When conducting statistical analysis, it is essential to use parametric and nonparametric analyses which properly accord with the attributions of data. What is often overlooked, however, is the need to select the most suitable method of spatistical analysis for the model of the studies. Parametric analysis certainly cannot be relied upon in all situations to yield the best resolutions, and even if we strictly distinguish between the use of parametric and that of nonparametric analyses, we still have done nothing to improve the resolutions themselves. In this article, I discuss several frequently asked questions by presenting an example in which I use parametric analysis on data that should be analyzed by nonparametric analysis.