RULES OF EVIDENCE - A LARGER CONTEXT FOR THE STATISTICAL DEBATE

Statistical tests define rules of evidence in psychological science. A greed-upon rules of evidence ar-e needed to discuss what is true. Stat istics are probability statements about the reliability of association s. In public and journalistic debates about psychological topics, the need for statistical rules is evident. Anecdotes, intuitions, and sens ational and unlikely events often dominate popular discussions. Psycho logical science demands that the probability of events and association s be evaluated by statistical means. Understanding of statistical test s can be improved by abandoning the term significance, which means rel iability to psychologists, but in everyday English, means importance. The importance of a result depends on magnitude of association, nor on whether the event has a nonzero chance of recurring. Given large samp les and multiple measures, statistical rests are used too often to dra w misleading conclusions about the importance of research results. Bet ter uses of statistics would focus on the magnitude of effects and err or estimates.