11.07.2015 Views

statisticalrethinkin..

statisticalrethinkin..

statisticalrethinkin..

SHOW MORE
SHOW LESS
  • No tags were found...

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

322 12. MONSTERS AND MIXTURESdispersed the beta distribution is. When θ is large, the beta density is bunched up around themode. When θ is small, the density is more spread out or, at very low values, even bunchedup on the edges of the probability space. Starting search at log-odds equal to zero and θ = 2defines a flat Beta distribution.Let’s compare the fits of these model estimations, before taking a look at the estimates:R code12.23compare(m12.4,m12.5,nobs=sum(d$density))k AICc w.AICc dAICcm11.5 2 273.09 1 0.0m11.4 1 574.49 0 301.4Okay, so the beta-binomial model fits the data much much better, even aer accounting forhaving one more parameter. Now how do its inferences differ, such that it does such a betterjob? Taking a look at the estimates from both models:R code12.24coeftab(m11.4,m11.5)m11.4 m11.5a 0.84 0.92tau NA 0.96nobs 48 48So the log-odds estimates, a, aren’t so different. For the binomial model, the mean posteriorprobability and 95% HPDI are:R code12.25post

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!