BAYESIAN ARCHAEOLOGY.

BAYESIAN ARCHAEOLOGY. Bayesian analysis deals with probabilities. Even when I have provided specific dates, there is in fact a probabilistic range for those dates. A rough way of thinking about how a Bayesian analysis would work is that the archaeologists adopt a model which incorporates their initial beliefs about the chronology. For example, they might begin with a range of perhaps 250 years given by radiocarbon dating. Then they use hundreds of bits of additional data to modify the probabilities in the initial chronology. As David Keys says in the article I linked to, “When literally hundreds of pieces of radiocarbon and stratigraphic data from a given site have been systematically analysed by a Bayesian computer program, the precision can typically be improved to a 25 year rather than 250 year date range.” Here is a Google books link to an article on Bayesian archaeology by Alex Bayliss and Christopher Bronk Ramsey in a book on statistical methods edited by Caitlin E. Buck and Andrew R. Millard. Bayesian analysis is controversial. Bayesians and frequentists have a long-standing disagreement, as I posted on here. Frequentists think that you can’t have subjective probability and that probability estimates should properly be based on sampling data. The Bayliss and Ramsey article says (at page 35) that while there are those statisticians who “don’t like adding any arbitrary assumptions to an analysis …[such as by choosing a model]…fundamentalist Bayesians might want to apply a probability to everything and let it all come out in the wash.”