next up previous contents
Next: Prior Knowledge Up: Bayesian Approach Previous: Bayesian Approach

Formal Bayesian Methodology

More formally, the following is the method employed. As mentioned above, statistics is concerned with the estimation of numerical quantities. In the Bayesian context, the quantities of interest will be random variables, and could, for example, be the proportion of rusted cars as referred to above. Before an experiment or survey, the prior knowledge about the quantities of interest are summarised in the form of a probability statement.

Denote the parameter or parameters of interest tex2html_wrap_inline2033 and the state of current experiences to date is denoted H. Such experience might be to do with knowledge of the material properties of cars, the nature of the roads, the weather and indeed knowledge of previous studies. The probability statement about initial beliefs is denoted tex2html_wrap_inline2037 and is termed the prior belief. Since this is a probability statement it takes the form of a probability distribution and is often referred to as the prior distribution, or more simply the prior.





Cathal Walsh
Sat Jan 22 17:09:53 GMT 2000