STATISTICAL TESTS OF VAN EARTHQUAKE PREDICTIONS - COMMENTS AND REFLECTIONS

The papers in the debate show that precise consideration of whether va rious events should be considered as having been successfully predicte d is one of the key issues for statistical testing of the VAN earthqua ke prediction efforts. Other issues involve whether the VAN prediction rules have been adjusted retroactively to fit data, how to test the p rediction significance and, finally, the criteria for rejection of the null hypothesis - that the VAN successful predictions are due to chan ce. The diversity of views on these subjects suggests that the seismol ogical community needs a general strategy for evaluating earthquake pr ediction methods.