I've spent a great deal of time and effort in evaluating the predictions made by Jim Berkland in his monthly newsletters. He predicts for 4 areas and each area has the same magnitude range every month. His predictions are well suited to the effort because they are identical except for dates. This means that the probability of success is identical which is necessary for statistical evaluation of probability.
I've tried several ways to evaluate his success and finally settled on the z-binomial test of significance. I've used two different methods of determining probability. One is simple probability, the amount of time covered by his windows as a percentage of the total time involved. The other is the Jones Observed Probability which counts the actual number of like-sized windows which contain quakes in the time span.
My results show clearly that his predictions are no better than chance. His methods are without merit. Despite this he continues to publish his preditions each month.
I can no longer post on his website so to avoid having all my work go to waste I'm getting it published in the Skeptical Inquirer. It is in the September-October 2006 issue.
|
