tag:blogger.com,1999:blog-6894866515532737257.post4699790354781080810..comments2022-09-21T20:48:43.728-07:00Comments on Probably Overthinking It: Hypothesis testing is only mostly uselessAllen Downeyhttp://www.blogger.com/profile/01633071333405221858noreply@blogger.comBlogger3125tag:blogger.com,1999:blog-6894866515532737257.post-35283348986222066402015-05-01T17:31:06.813-07:002015-05-01T17:31:06.813-07:00Any good statistician knows about the 4 possibilit...Any good statistician knows about the 4 possibilities you speak of. Hypothesis testing is full of assumptions, and these assumptions are violated every day. The problem is not with the p-value, but with the people who misuse it, and the sheer number of 'experiments' that are performed every day. Don't mess with science!Ralph Wintershttps://www.blogger.com/profile/14548913261473484508noreply@blogger.comtag:blogger.com,1999:blog-6894866515532737257.post-65266689754680154122015-05-01T14:03:23.973-07:002015-05-01T14:03:23.973-07:00Thank you for this clarification. I deliberately ...Thank you for this clarification. I deliberately chose this wording because it is more readable than the more pedantic version, and it is equally correct if we take "apparent effect" to include cases where the test statistic is equal or greater than what was observed. I realize that it is not completely unambiguous, but I stand by my editorial choice. Allen Downeyhttps://www.blogger.com/profile/01633071333405221858noreply@blogger.comtag:blogger.com,1999:blog-6894866515532737257.post-75478987476319483982015-05-01T13:35:50.355-07:002015-05-01T13:35:50.355-07:00"The p-value is the probability of the appare..."The p-value is the probability of the apparent effect under the null hypothesis..." No, still wrong. The p-value is the probability of the observed effect or one *more extreme* under the null hypothesis. All your tables are assuming that what you know is p<.01, not that p=.01.Richard Moreyhttps://www.blogger.com/profile/11319149283079163004noreply@blogger.com