Seemingly discrepant results across studies and random assignment of subjects: an example by using prenatally...

Four experiments, in which the prenatally X-irradiated effect on open-field behavior was examined, were repeated as identically as possible, and the magnitude of fluctuation of results across experiments was assessed. Contradictory results among the four experiments in some measures were obtained using the conventional significance test results. A computer simulation experiment which attempted to assess the degree of fluctuation in results due to the chance of subject assignment was administered. The simulation result demonstrated that the conclusion in terms of the conventional significance test results is questionable. The easy logic by which difference in results is attributed always to the difference in methods (strain, procedure, etc.) was criticized. Issues concerning requirement for standardization of methods and replication experiments were also discussed critically.