Philip Stark of UC Berkeley is the frequentist in question. I pointed
him to this discussion and he replied as follows (forwarded with his
permission).

mt

---------- Forwarded message ----------
From: Philip B. Stark <...>
Date: Sep 13, 2007 1:24 PM
Subject: Re: frequentism discussion

Hi Michael--

Thanks for pointing me to this and for your interest.

My argument--applied to earthquake forecasts by the USGS--is in
this preprint:
http://statistics.berkeley.edu/~stark/Preprints/611.pdf

Slides from Monday's talk are here (in OpenOffice format):
http://statistics.berkeley.edu/~stark/Seminars/sandia07.odp

My points about the frequentist coverage probability of Bayesian
credible regions were not connected to the Chicxulub story...
Luis and I are planning to write something up for the conference
volume on that topic.

The gist of my argument about whether the chance the KT impactor
came from Baptistina Asteroid Family is that the probability
ultimately comes from stochastic assumptions about the collision
that formed the BAF (Gaussians for some things, truncated Gaussians
for some things, uniform for others, etc.), simplified simulations
about how orbits evolve to move objects into the resonances with
Jupiter and Mars, how objects are ejected from resonance, and on and
on, plus ad hoc corrections to astronomical catalogs, extrapolation
of empirical scaling laws well beyond the data, assumptions that
things like albedo, density and specific heat capacity are constant,
etc.  There's a huge chain of physical and probabilistic assumptions
comprising an extremely complex--and tenuous--model.  Buy the model,
buy the 90% figure.  But why buy the model?

This is not like an urn model, and my skepticism is not because this
is a one-off event, or even that it either happened or didn't. (We
could rephrase it to be prospective, rather than retrospective,
and I'd have much the same problem with it.)  My main complaint
is that the model is far fetched and only weakly tied to observation.

Best wishes,
Philip


On Thu, 2007-09-13 at 12:47 -0500, Michael Tobis wrote:
> There's an interesting discussion of reasoning under uncertainty on a
> public discussion list which I manage. I mentioned the Chixculub
> argument you presented at the Santa Fe workshop and that has proved an
> anchor of the discussion.
>
> However, I am doing a lousy job of defending your position because of
> some combination of not understanding it and/or not agreeing with it.
> So far nobody else has showed up to advocate for a frequentist
> perspective.
>
> You may at least be interested in reading the discussion so far. Your
> input would be most welcome.
>
> http://groups.google.com/group/globalchange/browse_thread/thread/cd703ecce94de4d4
>
> regards
> Michael Tobis
> http://www.ig.utexas.edu/people/staff/tobis/
--
Philip B. Stark | Professor of Statistics | University of California
Berkeley, CA 94720-3860 | 510-642-1430 | www.stat.berkeley.edu/~stark




--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups 
Global Change ("globalchange") newsgroup. Global Change is a public, moderated 
venue for discussion of science, technology, economics and policy dimensions of 
global environmental change. 

Posts will be admitted to the list if and only if any moderator finds the 
submission to be constructive and/or interesting, on topic, and not 
gratuitously rude. 

To post to this group, send email to [email protected]

To unsubscribe from this group, send email to [EMAIL PROTECTED]

For more options, visit this group at 
http://groups.google.com/group/globalchange
-~----------~----~----~----~------~----~------~--~---

Reply via email to