Specialised Statistical Procedures

In this chapter, we explore more specialised statistical procedures as well as a fundamental concept, Bayesian statistical inference. The procedures we discuss and illustrate include: Rasch models and item response theory (alternative approaches for assessing measurement quality); survival/failure analysis (useful for predicting how long a particular outcome event will take to be observed); quality control charts (useful for tracing, measuring and analysing the quality of products, services and processes); conjoint measurement and choice modelling (sophisticated experimental/statistical methodologies for assessing consumer decision making); multi-level models (for estimating multiple regression models at different levels of analysis); classification and regression trees (to facilitate detection of prediction and interaction patterns in data); social network analysis (a quantitative approach to understanding and displaying the nodes, interactions and relational connections in a social network); specialised forms of regression analysis (e.g. robust, multinomial, fuzzy, nonlinear, ridge, generalised least squares, 2-stage, tobit, probit, ordinal, mediated and moderated regression models); data mining (a range of methods for detecting and learning patterns in data, particularly in large data sets); text mining (useful for detecting and learning concepts and patterns in qualitative data); and simulation and computational modelling (useful for building/testing statistical, mathematical or virtual models of the world).