Testing the Consistency of Soviet Data Using a Sequence of Hypothesis Tests

Abstract : The testing of Soviet historical data for consistency is examined in this report via a sequential set of hypotheses by first testing to determine whether or not the apparent seismic variance based on this new data has been reduced. If it cannot be determined that the apparent variance has been reduced, we test to determine whether or not the intercept term in the basic magnitude log yield linear relationship has been inflated on the basis of the historical data. We provide techniques for testing this second hypothesis under the assumption that the variances are known and under the assumption that these variances are unknown but equal. If either of the two sequential hypotheses is rejected, then we reject the consistency of the Soviet data. An extensive set of tables is given for determining the number of CORRTEX events which would be necessary in order to have a sufficiently high probability of rejecting consistency where the historical data is sufficiently inconsistent to produce unsatisfactorily high F-numbers.