Re: Response to Jamieson regarding 'More on the Bayesian Approach and the LR'

摘要

Dear Editor, We accept that, for a scientist, a subjective probability should be rational. It is subjective, or personal, none the less. That is true whether or not the probability is informed by data. The correspondence refers to a text elsewhere [iv] which introduces the novel concepts of comparative and absolute evaluative opinion. Unfortunately, in the example given to illustrate the definition of'comparative evaluation', the likelihood ratio has been confused with the odds ratio. Furthermore, offering 'absolute evaluation' as a type of assessment of weight of evidence when there is only one proposition is problematic. As a general rule, we believe that it is not rational - indeed, not possible - to assess weight of evidence when considering a single proposition. What the authors describe is not what would generally be seen as 'assessing the weight of evidence'. Their definitions and examples deal with the probability of propositions - a completely different notion from evaluating the weight of evidence.