The problem is epistemology, not statistics: Replace significance tests by confidence intervals and quantify accuracy of risky numerical predictions

Author(s): Harlow, L. L., Mulaik, S. A., Steiger, J. H.

Journal/Book: What if There Were No Signifi. 1997; 10 Industrial Ave, Mahwah, NJ 07430. Lawrence Erlbaum Assoc Publ. 393-425.

Note: Article Meehl PE, Univ Minnesota, Minneapolis,MN 55455 USA