tag:blogger.com,1999:blog-987850932434001559.post8421524113918280793..comments2018-05-25T16:23:11.737-07:00Comments on The 20% Statistician: Prior probabilities and replicating 'surprising and unexpected' effectsDaniel Lakensnoreply@blogger.comBlogger6125tag:blogger.com,1999:blog-987850932434001559.post-74086504814747456142017-09-26T19:43:03.235-07:002017-09-26T19:43:03.235-07:00Dan, just a quick note that the spreadsheet's ...Dan, just a quick note that the spreadsheet's link goes nowhere.Michael Carlozzinoreply@blogger.comtag:blogger.com,1999:blog-987850932434001559.post-48807297510909070032016-06-20T06:44:51.900-07:002016-06-20T06:44:51.900-07:00Thanks for pointing out my error, and taking the e...Thanks for pointing out my error, and taking the effort to leave a comment! I appreciate it! Fixed.Daniel Lakenshttps://www.blogger.com/profile/18143834258497875354noreply@blogger.comtag:blogger.com,1999:blog-987850932434001559.post-71116039395813735102016-06-20T06:43:25.086-07:002016-06-20T06:43:25.086-07:0075% is 3:1 in odds.75% is 3:1 in odds.anonymousehttp://mou.senoreply@blogger.comtag:blogger.com,1999:blog-987850932434001559.post-63513965868174628162014-06-01T12:56:24.596-07:002014-06-01T12:56:24.596-07:00Thanks - really helpful, especially the example of...Thanks - really helpful, especially the example of how with p=0.04, and equally likely a priori H1 is still 26% likely to be wrong. Which is very counter-intuitive.Simonhttps://www.blogger.com/profile/14362211681941540619noreply@blogger.comtag:blogger.com,1999:blog-987850932434001559.post-87106412169025048042014-05-27T23:04:01.851-07:002014-05-27T23:04:01.851-07:00The example focusses on p-values. Effect sizes sta...The example focusses on p-values. Effect sizes stay the same as sample sizes increase, but p-values get smaller when sample sizes increase. Also, for small effects, you need larger sample sizes to get small p-values, wheras for big effects, you can suffice with smaller sample sizes. So, the effect size and priors are independent, but both influence the sample size you need. If that an answer to yout question?Daniel Lakenshttps://www.blogger.com/profile/18143834258497875354noreply@blogger.comtag:blogger.com,1999:blog-987850932434001559.post-14865291410518595702014-05-27T15:35:06.419-07:002014-05-27T15:35:06.419-07:00Great explanation! One question though: since the ...Great explanation! One question though: since the calculation appears to rely on knowing a noncentral distribution for p(h1|d), planned effect size should play a role, yes? So is the size of the expected effect an implicit corollary of the priors? Or is it orthogonal to the priors?johannhttps://www.blogger.com/profile/16428748724803791707noreply@blogger.com