markm reviewed Proving history by Richard Carrier
Review of 'Proving history' on 'Goodreads'
4 stars
This dense and meticulously reasoned argument explains why the author recommends that Bayes’ Theorem be the basic technique for the analysis of hypotheses in history, both for our general education and in preparation for his second volume, On the Historicity of Jesus. I’m no expert, and I mostly used Bayes’ Theorem for demonstrations of medical diagnostic problems about 30 years ago, and then gradually more and more as a replacement of more common frequentist type statistics in everyday work as it became easier to do with computer software and better understood. From my point of view, so distant from Professor Carrier, this method acts as a way to check and compare your data, your assumptions, and your hypotheses, but, frankly, it’s hard for me to imagine thinking this way ab initio. Perhaps it comes with practice. Also, I must admit that regardless of the statistical analysis used, many of the …
This dense and meticulously reasoned argument explains why the author recommends that Bayes’ Theorem be the basic technique for the analysis of hypotheses in history, both for our general education and in preparation for his second volume, On the Historicity of Jesus. I’m no expert, and I mostly used Bayes’ Theorem for demonstrations of medical diagnostic problems about 30 years ago, and then gradually more and more as a replacement of more common frequentist type statistics in everyday work as it became easier to do with computer software and better understood. From my point of view, so distant from Professor Carrier, this method acts as a way to check and compare your data, your assumptions, and your hypotheses, but, frankly, it’s hard for me to imagine thinking this way ab initio. Perhaps it comes with practice. Also, I must admit that regardless of the statistical analysis used, many of the problems that I had or was consulted about in my career mostly benefited in a similar way, i.e. the statistics confirmed why the researchers were correct in their assumptions and hypotheses, either graphically or numerically, but weren’t really necessary for them to know this initially. The great value of these techniques was always in those uncommon cases where the findings could be shown to be counter-intuitive. The classic Bayesian example that most doctors have seen, but probably never really understand, is looking for a rare disease with a sensitive test. The great majority of positive tests are false positives. I think that all of these factors are evident in Carrier’s discussion. I especially liked the flowchart in the appendix that shows the non-numerical use of Bayes’ theorem for the analysis of historical hypotheses.