ABSTRACT
We will only use likelihood-based methods in this book. By this is meant both Frequentist, relying on the usual asymptotic approximations, and Bayesian, relying on appropriate priors. There are many useful ad hoc methods for obtaining answers to particular questions, but we will stick to the general approach, which is capable of giving answers to any well-formulated parametric question. However valid the methodology, of course, the data need to be sufficiently informative and the model has to be sufficiently well-fitting. How many seminars have you attended where the conclusion was “Nice method, shame about the data?”
