Abstract

Notes that 1 approach to randomization of targets is to use whatever methodis convenient, practical, and empirically adequate; such a method isthorough card shuffling. Disadvantages to this approach can be overcome bythe use of random sources which are not merely adequate, but which providecomplete assurance of randomness from a theoretical standpoint. Schmidtrandom number generators are proposed as 1 such source. It is suggestedthat these generators be thoroughly checked for randomness before beingplaced in service (and again at regular intervals in the actual testenvironment) to assure their proper functioning. In psychokinesis tests,control runs can be interspersed with experimental runs in order to detectbias of the generator. All randomness checks should be performed underappropriate protocols to avoid problems of data selection. The problem ofnonindependence of some randomness tests can be overcome by the use of thegeneralized serial test. (26 ref)