About some misconceptions and the discontent with statistical tests in psychology1

The function and potential importance of statistical tests in examining and evaluating substantive and psychological hypotheses is discussed. Psychological hypotheses are sharply distinguished from statistical hypotheses. Decisions on statistical hypotheses must be separated from decisions on psychological hypotheses. Some differences between both kinds of hypotheses are addressed, and the question is attacked whether they are complementary or not. The answer to this question is negative. The use of the modus tollens in theory corroboration is discussed and it is argued that evaluations of substantive hypotheses always is accompanied by some inductive aspects. A further reason for the discontent with statistical tests is identified: Most of these tests are rather insensitive the differential patterns of predictions and of data, whereas differential patterns of data can be derived from nearly every psychological hypothesis or theory. To test for these differential patterns statistical tests should be applied thoughtfully, not routinely.