Abstract

Considers that merely following the text or "cookbook" statistical procedures in analyzing experimental data may lead an experimenter to draw unwarranted conclusions. Such a pitfall exists in an important segment of psi research. In general, the statistical tests for existence of psi are correctly used and interpreted. However, the statistical methods used to affirm relations, comparisons, or correlations between psi and other variables (e.g., personality traits) are unwarranted, since no statistical or probability distribution is known to exist involving psi scores. The chi-square contingency test is an appropriate test, but one must be careful in the interpretations. The t statistic is an example of an inappropriate statistic for tests for psi. (3 ref)