Measuring real-time dynamics underlying online rating processes: Models, methods, applications

This thesis collects a set of research articles and working papers produced during my doctoral program. Overall, they refer to: (i) the general problem of collecting measures for the dynamic processes underlying human rating tasks (section Models, methods, instruments) (ii) the proposal of new statistical methods to analyse data structures arising from the dynamic rating measurements (section Statistical techniques). The section Models, methods, instruments contains three chapters on models and methods designed to represent the dynamics of rating evaluations together with some relevant applications. In particular, the latter refer to two different contexts, namely Likert-type questionnaires (Chapters 1 and 2) and two-choice decision making (Chapter 3). Finally, the section also describes two new measurement instruments for dynamic ratings, DYFRAT (Chapter 1) and DYFRAS (Chapter 2). In a similar way, the section Statistical techniques contains three works on novel statistical methodologies to run data analysis in the context of structured data€ , namely fuzzy numbers (Chapters 4 and 5) and interval-valued data (Chapter 6). The justification for structured-data comes directly from the treatment of the dynamic rating components in the sense that such a kind of data can offer a simple and flexible formal representation to model the information related to the rating properties.