Some Remarks on Compliance Testing

Abstract : We discuss several issues concerning compliance testing. First we discuss the F-number and provide an alternative definition which is identical to the common definition when sigma sub w is known but which is applicable to a broader range of testing situations. We discuss for compliance of a single new event given a collection of events for which we have both CORRTEX and seismic readings. These tests are considered in the case which sigma sub sei is assumed to be known and in the case in which sigma sub sei is unknown but the ration sigma cub sei/sigma sub cor is known. We examine the robustness of these tests under situations for which we do not have perfect knowledge of sigma sub sei or the ratio and show that the test based on the ratio is more robust to errors in specification of the variances. We also discuss the estimation of sigma sub sei based on the events for which we have both CORRTEX and seismic data . A brief discussion of the use of the mixture-of-normals model for purposes of estimating the seismic variance is also given.