a lot of experimental results that belong over on the unpublishable side of the p= .05 boundary have been cajoled, prodded, tweaked, or just plain tortured until, at last, they end up just on the happy side of the line. That’s good for the scientists who need publications, but it’s bad for science.
