Inferring causality from relational data and designs: historical and contemporary lessons for research and clinical...

My address to you today is born of frustration with the growing claims, evident in the research literature and in clinical reports that happen to cross my desk, of causation made on the basis of correlational data. I recall, as a sophomore in college, being taught early in my first experimental psychology course, "you cannot infer causality from correlation." This point was made adamant in my first face-to-face critique of an experimental psychology research paper by Dr. Robert T. Brown, who pointed out, in lowering my grade, that I had inferred organic causes to behavior patterns in gerbils based solely on correlational data. This caution was reiterated in my statistics courses until it must have been indelibly stamped upon my then still somewhat plastic brain.