Chapter 11 A general strategy for statistical modelling

It is important to realise that fitting a probability model to data is just one part of the modelling process. Figure 11.1 gives an outline of a general strategy for statistical modelling of data. The process is iterative, since it is rare that the first model is fitted is adequate. It may be necessary to modify the model several times before it can be used to answer the questions of interest. You should always remember the context behind the data and provide an interpretation of the results in non-statistical terms, particularly if your audience contains people who are not statisticians.

Figure 11.1: A general strategy for statistical modelling.

STAT0002 does not cover all the areas of Statistics you may need to put this general strategy into practice. STAT0003, STAT0005 and STAT0008 will look in greater detail into the theory of statistical inference. Important practical aspects will be covered in other modules, such as: STAT0006: linear modelling with more than one covariate, interactions - where the effect of \(x_1\) on \(Y\) depends on the value of \(x_2\) - generalisations to non-normal responses and flexible regression functions; STAT0023: statistical computing, optimisation, simulation, linear and non-linear regression modelling; STAT0045: design and analysis of statistical experiments and surveys, observational studies, ethical aspects.