Statistics for diagnostic procedures. III. Philosophic and research design considerations.

Statistical analysis of data presupposes that properly designed research has been performed. Improperly performed studies are worse than none at all because they cause readers to assume erroneously that useful information has been obtained when it has not. Important points include thorough pretrial planning without mid-course alterations, clear identification/definition of relevant parameters, correct interpretation of data, and willingness to publish negative results.