ABSTRACT
This chapter introduces a statistical model as a description which provides answers in the form of intervals each with an associated probability. Statistical regularity represents an article of faith, since it can never be demonstrated to hold for any actual situation. For instance, statisticians cannot toss a coin an arbitrarily large number of times. Any real coin would wear out after a finite number of tosses. In statistical problems statisticians do not have just a single model to describe their measurements, but rather a class of models which is assumed to include at least one suitable model. Statistically, statisticians might want to determine which of the regression coefficients appear to be 0, or they might want confidence intervals for some of the regression coefficients, to get an understanding of how each independent variable influences the dependent variable. This class of models can represent a much greater variety of relationships than might imagine.
