Statistics According to Bayes
Gudmund R. Iversen · Sociological Methodology · 1970
Thomas Bayes may not have foreseen that a paper he wrote and friends of his published in 1763, after his death, would serve to tie his name to a branch of statistics that has come into prominence today. Bayesian statistics gets its name from the use and special interpretation of what is known as Bayes' theorem, a simple identity in conditional and unconditional probabilities, contained in the paper by Bayes (1763). The rise of Bayesian statistics in the last couple of decades has been prompted mainly by dissatisfaction with what sometimes is called classical statistics. This is the statistics of hypothesis testing with errors of two kinds and the statistics of confidence intervals, as developed by Neyman and Pearson in this century. Their formalizations were a major achievement, but in practice difficulties are found with classical statistics. One such difficulty rests with the specification of the value of the parameter that is tested in the null hypothesis. In the case of a correla-