Applied Statistics Using Quantiles

The book is devoted to quantile-based methods of analysis. It is divided in three parts. Part I introduces general topics in statistics and sets out the goals of statistical analysis and describes the double-faced nature of statistical distributions, namely probability and quantile functions and how the latter can be used to extract information from the data.

In particular, chapter 3 (location, scale and shape of probability distributions) describes where such information resides; this is a recurring theme throughout the book and is further developed in Chapters 8 and 14. While inferential procedures based on modelling probability functions have been widely described in a number of statistical textbooks, scientific contributions to the development of quantile-based inference are sparse and lack a comprehensive treatment. The main topics of the book are discussed in parts II and III, which introduce methods and applications for unconditional and conditional quantiles.

Each part considers: the distribution-free approach, in which quantile estimation makes no use of parametric probability models; and the model-based approach, in which the quantile function is defined as the inverse of a known distribution function, thus quantile estimation conforms to some statistical model (e.g., Normal, exponential, Pareto). The book emphasises that in a quantile model-based approach the modelling step starts from the quantile function directly (as opposed to modelling the distribution function and deriving the quantiles by inversion).