|
That's true Steven....the mathematical formulae are what they are. It's the data entered into the equations which is the problem and that's where the bias and manipulation arises from.
That's why any statistical analysis of this subject is flawed right from the beginning. How can you know the probability of any particular outcome when you don't have any prior knowledge of any of the variables of the analysis. What little knowledge we do have is only based on one example and quite frankly that would be rejected in any objective statistical analysis...of any subject let alone this one. Bayesian analysis requires that the calculations are based on a prior distribution of the parameters that are being looked at, in order to arrive at the answers sort. That means if you use a different set of priors, you get a different set of answers. It's loaded straight from the get go. You could make any inference as to what the initial prior was going to be. It could just be based on a hunch and you would be able to plug it into the equations and get any answer you wanted. Science prides itself on its objectivity and yet it could be hardly said that Bayesian analysis is truly objective. That's why you see Frequency Inference statistics being used for the most part, because many scientists are uncomfortable with the subjectivity of Bayesian analysis. That, and you need a powerful computer to do complicated Bayesian because the calculations can become onerous.
In any case, this particular study is moot straight from the start simply because we really have no idea of what the initial conditions were to begin with and the calculations could be made to go either way, just based on your initial set of assumptions. As I have mentioned previously. Meaning it's inherently biased right at the beginning, whether intentional or not.
|