Monday, 28 April 2014

Paul Jones on genealogical confidence

On Sunday afternoon at the St Catharines OGS conference Toronto genealogist Paul Jones will present "Determining how much confidence you should have in your genealogical inferences."
As Paul mentions blog postings of mine I'm giving the presentation a shout out. The abstract reads
Over the past quarter-century, the use of statistics in the sciences has been shifting from a “frequentist” approach (t-tests and the like) to the realm of conditional probabilities and Bayes Theorem, i.e., the use of successive pieces of evidence to progressively modify our estimate of 
the probability of the truth of a hypothesis. This trend is now entering the social sciences and it is a reasonable question to ask whether genealogists should add simple probability calculations to their research tool kit to distinguish between conclusions that are, say, 20%, 80% and 99% likely to be accurate. This presentation illustrates basic probability calculations, corrects some common misunderstandings about probability and suggests how probability could, in some circumstances, augment but never replace the Genealogical Proof Standard (GPS). 
One of the other references for Paul's presentation is Richard Carrier, author of Proving History, who posted If You Learn Nothing Else about Bayes’ Theorem, Let It Be This on his blog on Sunday. The  main he discusses are that Theories Cannot Be Argued in Isolation and Prior Assumptions Matter
Be aware that Carrier doesn't pander in his writing. His post has a fog index of 11.7 where 5 is readable, 10 is hard, and 15 is difficult.

No comments: