an infinite regress of hierarchical priors

An interesting musing posted on X validated about the impact of perpetuating prior models on the parameters of closer priors till infinity. Using a hierarchy of exponential priors and an exponential sampling distribution. If the (temporary) top prior at level d is Exp(1), the marginal distribution of the exponential sample corresponds to a ratio of two independent products of Exp(1) random variables

$X= \frac{\epsilon_{2\lfloor d/2 \rfloor}\cdots \epsilon_0}{\epsilon_{2\lfloor (d-1)/2 \rfloor+1}\cdots \epsilon_1}$

And both terms converge almost surely to zero with d (by Kakutani’s product martingale theorem). Thus ending up in an indeterminate ratio. Hierarchy has to stop somewhere! (Or, assuming an expectation of one everywhere, the variability at each level has to decrease fast enough.)

This entry was posted on October 22, 2020 at 12:20 am and is filed under Statistics with tags Amstel, Belgian beer, Belgium brewing company, Kakutani's product martingale theorem, martingales. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.

2 Responses to “an infinite regress of hierarchical priors”

coreyyanofsky Says:
October 22, 2020 at 3:11 pm

This has been investigated in “Infinite hierarchies and prior distributions” by Gareth O. Roberts and Jeffrey S. Rosenthal, https://projecteuclid.org/euclid.bj/1080004760 .

Reply
- xi'an Says:
  October 22, 2020 at 6:35 pm
  
  uh-oh, don’t tell them!!! (Thanks!)
  
  Reply

Xi'an's Og