Experimenter expectancy and the reassuring nature of the null hypothesis decision procedure.

By employing some form of a null hypothesis decision procedure, 12 of the studies summarized by T. X. Barber and M. J. Silver (see 43:5) were judged to demonstrate expectancy effects. The combined p of the results of these studies was less than 1/1,000,0002. When the results of the studies judged by Barber and Silver not demonstrate clearly the effects of E expectancy were considered separately, the combined p was 6/1,000,000. For all studies combined, p was less than 1/1,000,000(2+x) with a median 2 tailed p < .1. Such results provide little reassurance that the effects of Es' expectancies are not general, robust, nor serious enough to warrant the use of procedures designed to permit their control and/or assessment. Methods for the analysis of data selected by Barber and Silver are discussed in terms of their differential value for the analysis of single experiments and runs of experiments. An appendix is attached to add information omitted from the presentation by Barber and Silver. (PsycINFO Database Record (c) 2006 APA, all rights reserved)