Monday, May 20, 2013

A Rabett Exclusive

Ethon came flying in the window with a missive from John Cook.  Based on the cross-tabs from the prequel survey, the bunny wondered what the distribution of No Position ratings from the abstracts would be in Cook et al based on the responses of  the authors who responded to the Email survey.

Among papers whose abstracts were rated "no position", according to self-ratings (which is a proxy for the endorsement level of the full paper):
228 endorsed AGW
213 had no position
11 rejected AGW

In other words, most of the climate papers with "no position" abstracts go on to endorse AGW in the full paper. We explicitly mention this in the paper. So whenever the deniers are saying "66% of climate papers have no position on AGW", they're simply wrong and either haven't read the paper or are misrepresenting the paper.
the self ratings being the ratings by the authors.  As Eli remarked before, the No Position abstracts, themselves are attached to very few papers that reject AGW.

Those interested in shark jumping can hi thee over to Lucias.  Eli will not post the link the level of parsing over there is growing dangerous but then again . . . sport:)


Sou said...

With all their puffing and panting and theorising, I can't help but wonder why everyone at the Blackboard doesn't get together and do a similar excercise themselves. They seem to know all about random numbers so I'm sure they could do a mini-exercise with a lesser number of abstracts - or even full papers if they so desired.

Actually, I can take a stab at the answer. It's because Lucia, Schollenberger and co prefer to obsess and whine. They learnt a lesson when someone attempted a 'replication' of Lewandowsky and probably confirmed his analysis (they have buried the results so we'll never know). If they did a quick and dirty themselves they'd lose another talking point.

Sou said...

I fell for using the tribal "they". I'm usually above all that nonsense :)

It was The Auditor, not the folk at the Blackboard, who buried the results of the 'replicated' survey AFAIK.