IF OUR ESP FINDINGS were to be summarized in a single sen-
tence, it would state that differences in ESP scoring level were
found in relation to certain independently measured psychological
variables, and that in one case these ESP differences attained a
high degree of statistical significance.

Though literally true, such a statement might create the false
impression that the variables associated with high or low ESP
scores in this research should be universally so associated. Any
conclusion stated so broadly omits a crucial factor: the social at-
mosphere set up by subject-experimenter interaction, which we
believe to be an important determinant of the way in which the
personality patterns will influence the subjects' responses.

As a consequence, it is both difficult and time-consuming to
conduct adequate ESP experiments. The investigator who wishes
to repeat our work must be willing to take a large enough number
of cases so that he can go through both a preliminary and test
procedure. He should not necessarily expect to find that all of the
personality traits which were associated with poor scores in our
research will also be associated with poor scores in his own. He
should, instead, examine his subjects' records to find the pattern
that seems to differentiate good scorers from poor ones; and then
repeat the experiments under as uniform conditions as possible
(that is, trying to build up the same kind of rapport that he had
previously) to see if the same personality syndrome will again
differentiate good scorers from poor. We should expect him to find
that this syndrome was different from the one which we found,
if his subjects were more intensely interested or more enthusiastic
than ours had been; or if they were less aware of the intellectual
implications of their responses; or if he, whether deliberately or

-101-