Re: #329

What is the distribution of a show of hands? It cannot be less than zero so perhaps a Poisson?

With a finite electorate of independent voters, it would be binomial, and hence bounded by the number of voters. The Poisson distribution is unbounded, but closely approximates the binomial when the number of voters is large and the probability of a hand is small.

But I think we are veering off topic. So how were the dinners, Steve? Does Julien, Judith or Kim play a decent game of squash?

Three questions that I can recall. Steve spoke a little longer then 45 minutes, so the ~15 minute Q&A period was compressed to 5 or 10 minutes.

1. There was one question from the back of the room, from whom I am guessing was a student, regarding weighting of different PC’s for the hockey stick. Steve expanded a little beyond what he said during his presentation, but echoed that whatever weird statistical method (my terminology, not his) you use, eventually you end up with weighting factors for the different proxies. Whatever weighting factors you end up with, you ought to publish them along with your results. No argument from the audience. They had all just seen Steve present the great lengths he went to, to derive the MBH weighting factors. Steve went on further to explain that whatever future Mann output surfaces, he expects it to weigh heavy on the same proxies; if Mann wants to produce another HS.

2. There was a short question from the front right side (audience perspective), from whom I am guessing was a professor, about Mann claiming that he had his financial records subpoenaed during the congressional testimony on the HS. Steve expressed that he didn’t think that was the case, that Congress issued a pro forma question, as they sent to Steve, regarding who was funding his work. It was not like checking bank account statements or anything. Steve stated his MBH research was self funded.

3. There was a question from the front left (audience perspective), on other evidence that there is significant GW, like artic ice melting, rising sea levels, etc… I believe Steve started to reiterate his narrow scope that the MBH HS calculation was bad, he wasn’t claiming that it wasn’t getting warmer, but went further to say that the proxies suck (again, my terminology) so how do you really know over the last 1000 years or so what happened. The questioner followed up stating that there should have been oral history from 1000 years ago (or so) from northern societies if significant warming occurred back then. Steve disputed that, he sited some specifics about Inuit and other northern societies and how long they have been in place, and whether you could trust the oral history to be accurate to remember a significant warm period. The question came from a gentleman that was sitting between JEG and I; it is possible that JEG knows him.

#311, Frank, anything you can add?

As you said Kenneth, hopefully Steve will come back with a summary. That should clean up any of my misunderstandings or misrepresentations.

Thank you for your response about multiparameter modelling re my # 188.

My reasoning is:

In climate science there are many paramaters that extend back 100 years, some much more, for which measurements have been made. Some are treated as proxies for temperature as by Craig Loehle. There is argument about the confidence that should be put on these proxies. Some of the argument arises from statistical methods.

In real life there are many parameters for which records have been kept, also some for a century or more. Prima facie some of these parameters are unrelated; on the other hand, many might be related to a common attractor as yet unidentified well (to use the term wrongly).

Since you mentioned you had a data collection, I thought it might be a profitable exercise to hide the identities of each number series, assume they were all climatic factors, then do a multiple regression or similar correlation study to see what came out in the wash in a blind test.

This would be another way to estimate the credibility of variance claims; and perhaps another way to demonstrate how hockey sticks can be made. Make a meta-stats study of a whole heap of variables then try to explain the correlations.

I did not intend my # 188 to question your background or experience. I was merely interested in the fact of your collection and the hope that a prominent ststistician might want to take on this exercise and give it a run.

these IPCC confidence statements may in fact just be subjective judgments by the cronies of the lead authors rather than the outcomes of actual computations: “Likelihood [as defined in the above IPCC scale] may be based on quantitative analysis or an elicitation of expert views.” (”Guidance Notes,” p. 4)

The Delphi technique?–sweet.

330, I recall one such study (unrelated to climate), I believe a distribution in the gamma family was chosen.