tag:blogger.com,1999:blog-6894866515532737257.post3735925895027847767..comments2024-03-27T01:01:09.785-07:00Comments on Probably Overthinking It: Some of my best friends are crackpotsAllen Downeyhttp://www.blogger.com/profile/01633071333405221858noreply@blogger.comBlogger2125tag:blogger.com,1999:blog-6894866515532737257.post-80343390080598779312016-10-03T04:54:25.222-07:002016-10-03T04:54:25.222-07:00Great question, thanks! Yes, the chi2 would be co...Great question, thanks! Yes, the chi2 would be correct (and just as easy to implement). I think my normal approximation is pretty close, but when I get a chance, I will run both and see how close. If I did this again, I would use the chi2 distribution, since the normal approximation doesn't provide any advantage for this problem.Allen Downeyhttps://www.blogger.com/profile/01633071333405221858noreply@blogger.comtag:blogger.com,1999:blog-6894866515532737257.post-47825582850333937752016-10-03T00:51:10.618-07:002016-10-03T00:51:10.618-07:00In your book Think Bayes, you use the same example...In your book Think Bayes, you use the same example to illustrate Approximate Bayesian Computation. And you use `scipy.stats.norm.logpdf(s, sigma, sigma/math.sqrt(2*(n-1)))` for the likelihood of the sample standard deviation under the (mu, sigma) hypothesis. And I wonder, won't me more appropiate to use the sampling distribution of the sample variance for that likelihood instead? Something like `loglike += scipy.stats.chi2.logpdf((s**2*(n-1))/(sigma**2),df=(n-1))`.<br /><br />I posted this as question in stackoverflow http://stats.stackexchange.com/questions/238046/what-is-the-likelihood-of-drawing-a-sample-with-standard-deviation-s-from-a-noecerulmhttps://www.blogger.com/profile/02005618894883008001noreply@blogger.com