ON THE PROBLEM OF GRADING FOR ONE-DIMENSIONAL SEISMOLOGIC PREDICTIONS

This paper gives a complete commentary on some popular grading methods which are based on the 22 contingency table, for one-dimensional seismologic predictions. It contains:1. In view of statistical correlation, the Wallen's grading V =QS is the most reasonable one among all gradings.2. Each one of the three gradings, Gou Zhen-chao's grading S =n11/n1.-n01/n0. Obu-hov's grading Q=n11/n.1-n10/n0. and Xu Shaoxie's grading R =n11/n.1-n1./N, has its owninadequacy; although they are so easy to use.3. A set of gradings can be derived from V. The harmonic mean of S and Q,H(S, Q) and the Hedike's grading SH are members of this set.