<i>The effects he recorded were small but statistically significant.</i><p>And this, ladies and gentlebeings, is why you should be very wary of ever believing studies with small but statistically significant effects.<p>Psi is the control group for modern science: they use the same methods in a case where no actual phenomena exists, and yet it looks like a regular field of scientific study, dealing with "small but statistically significant effects", and the larger the sample size, the smaller the effect, but it remains "statistically significant".<p>And not only that, if a psi study finds that someone can guess which cards someone else has drawn from a deck, using telepathy, you will next find that someone can predict cards <i>before</i> they are taken from the deck, using precognition, and then that subjects can <i>influence</i> which cards are taken from the deck, using telekinesis, and <i>then</i> you will discover that they can influence which cards are taken from the deck, <i>two weeks later going backward in time</i>.<p>With small effect sizes that are statistically significant, of course.<p>Not because people can actually control which cards are taken from the deck two weeks after the fact.<p>But because <i>bad statistics are symmetrical in the directions of time and causation</i>, and it makes no difference to <i>bad statistics</i> whether the temporal distance is two seconds or two weeks.