tag:blogger.com,1999:blog-738485353871431380.post2733613919407789612..comments2024-07-28T12:54:28.323-07:00Comments on Crystal Prison Zone: Intuitions about pJoehttp://www.blogger.com/profile/10825531253125205466noreply@blogger.comBlogger2125tag:blogger.com,1999:blog-738485353871431380.post-69170505647287215002017-01-03T07:21:46.569-08:002017-01-03T07:21:46.569-08:00Well, when the null is true, the probability of p ...Well, when the null is true, the probability of p < .05 is just 5%, whereas the probability of p > .05 is 95%. My point is that when the null is true, the probability of p > .95 is just 5% -- p does not tend towards 1 when the null is true.<br /><br />As far as "How do I know whether statistical significance is due to chance," you don't -- that's Type I error. You can protect yourself from Type I error by lowering the alpha threshold (which will increase Type II error, assuming the same sample size), by conducting replications, and by abstaining from practices that increase the alpha threshold (e.g., p-hacking).Joehttps://www.blogger.com/profile/10825531253125205466noreply@blogger.comtag:blogger.com,1999:blog-738485353871431380.post-90641824642552662202017-01-03T07:08:51.629-08:002017-01-03T07:08:51.629-08:00If I get this right, you are saying that one is as...If I get this right, you are saying that one is as likely to get a p=.05 or less than any other p value when the null hypothesis is true and there is no effect, right? But then... in one-time experiments, how to know if this is what is happening (if the p-value is due to chance or not?).Gahariethttps://www.blogger.com/profile/16644828499487888629noreply@blogger.com