Last night, I received the very nice news from the Electronic Journal of Statistics that our paper on the Savage-Dickey paradox was in for revision. One referee recommended acceptance as is and the second referee asked for more details and examples in order to broaden the audience. All reports are very nice and consistent with the response I got from my talk in San Antonio. Maybe giving the talk there helped with the positive decision!
This morning, before going runnin, I took a look by my hotel window and noticed the road below was wet. Since there were warnings of a thunderstorm, I checked on a website the current forecast—taking advantage of an opening in the local networks!— and saw that the local conditons were a “light drizzle”. When I went out, it was hardly drizzling, indeed, and so I went on my “usual” round following the riverwalk for about two miles. At the turning point—an underbridge with huge fish sculptures hanging from the bridge—, rain started to fall rather heavily and on the way back I soon found myself in the midst of a thunderstorm! I had to stop under a bridge to wait for the rain to abate and the storm front to move away. After a few minutes, the strength of the rain went down but by then the riverwalk had turned into a river and stairs into cascades, the river being below street level. At some point, the flow of water falling from the street was so strong that I had to turn back to cross to the other side… This was thus an interesting experience, teaching to pay more attention in the future to storm warnings.
Apart from my (theoretical) Bayes factor session, I attended non-parametric, and two other Bayes factor sessions at Frontiers of Statistical Decision Making and Bayesian Analysis today! (I forgot to mention the talk of Kenneth Rice yesterday that related to my early 1990’s work with George Casella and Juinn Hwang on integrating estimation loss functions into testing losses). I liked very much Ed George’s extension of the g-prior, as well as Elias Moreno’s prior modelling of clustering models (along with Guido Consonni’s and Adrian Raftery’s talks in the same session). As commented by many participants to the conference, one major difficulty was to figure out which of the three parallel sessions to choose… The plenary sessions were superb as well, with Steve Fienberg describing an analysis of a highly (highly) complex model on aging, and Persi Diaconis giving us the flavour of his lastest work with David Aldous on quasi-exchangeability, with exctracts from de Finetti.
Here are the slides for the Savage-Dickey paradox paper that I gave in San Antonio this morning:
(Any suspected coincidence of the first part with earlier talks is for real!) I have tried to spell out as clearly as possible in the second part the issues of version choices that are at the core of the “paradox”.
The first day of the conference Frontiers of Statistical Decision Making and Bayesian Analysis, in honour of Jim Berger, in San Antonio, was quite a day! Starting at 8:30am with a few welcoming talks, we ended up with a poster session till 7pm. The great thing about the theme of that conference is that basically every talk of every session has some degree of relevance for most participants (besides the fact that most people are already acquainted). For instance, most talks I went to today were connected to Bayesian testing, including a plenary session by Larry Brown on the (non-Bayesian) incorporation of the consequences of model selection into the final inference. (I obviously missed two thirds of the talks, but so did everyone, including Jim Berger whose sessions were allocated at random by an R program!) The program is even heavier tomorrow with talks starting at 8pm!
 | David on Bayesian workflow [book review… |
 | Ewan on Tidy up, Claude! |
 | xi'an on Tidy up, Claude! |
| Ewan on Tidy up, Claude! |
| Ewan on Já við viðræðum. Já við Evrópu… |