Abstract

"In recent controversies concerning the statistical significance of the results of card-guessing in parapsychology, the possibility of using the guesses themselves to provide a criterion of significance has not been adequately considered. Arguments are put forward in favour of employing for this purpose elementary and obvious methods in preference to the conventional statistical methods which assume a strictly random distribution in the target sequence. A worked example of a simple procedure is described; and a proposal is made to test the so-called P.-K. effect, by means of the \'Galton board\' (a pin-table which, in the absence of any such influence, gives a Gaussian distribution)."