Circularity and the Hockeystick: coming around again 4


The recent posts at climateaudit and WUWT show that climate scientists Gergis and Karoly were willing to manipulate their study to ensure a hockeystick result in the Southern Hemisphere, and resisted advice from editors of the Journal of Climate to report alternative approaches to establish robustness of their study.

The alternative the editors suggested of detrending the data first, revealed that most of the proxies collected in the Gergis study were uncorrelated with temperature, and so would have to be thrown out.

A false finding of “unprecedented warming” is a false positive. False positives are a characteristic of the circular fallacy. The circular logic arising from the method of screening proxies by correlation was written up by myself in a geological magazine “Reconstruction of past climate using series with red noise” DRB Stockwell, AIG News 8, 314 in 2005, and also occupies a chapter in my 2006 book “Niche modeling: predictions from statistical distributions” D Stockwell Chapman & Hall/CRC.

It is gratifying to see the issue still has legs, though as McIntyre notes in the discussion of his post he has been the only one to cite the AIG article in the literature, its been widely discussed on the blogs, but a nettle not yet grasped by climate scientists.

Because the topic is undiscussed in climate science academic literature, we cited David Stockwell’s article in an Australian newsletter for geologists (smiling as we did so.) The topic has been aired in “critical” climate blogs on many occasions, but, as I observed in an earlier post, the inability to appreciate this point seems to be a litmus test of being a real climate scientist.

Its now fourteen years since the publication, with great fanfare, of Mann, Bradley, and Hughes “Global-scale temperature patterns and climate forcing over the past six centuries” with the “premature rush to adoption” that followed, the creation of research agendas in multiple countries and institutions devoted to proxy study, amassing of warehouses of cores. In any normal science the basics of the methodologies would be well understood before such a rush to judgment.

Considered in the context of almost a decade of related public blog discussion of the issue, that screening proxies on 20th century temperatures gives rise to hockeysticks is a topic apparently only discussed in private by climate scientists:

The Neukom email from 07 June 2012 08:55 “…I also see the point that the selection process forces a hockey stick result but: – We also performed the reconstruction using noise proxies with the same ARl properties as the real proxies. – And these are of course resulting in a noise-hockey stick.”

Of course, the problem is that rigorous analysis of many studies would fail to confirm the original results, many of the proxies collected and used by their colleagues are shown to be useless, and the abandonment of the theory that contemporary warming is “unprecedented”.

In light of all the data and studies from the last decade, I am convinced of only one thing – that the fallacy of data and method snooping is simply not understood by most climate scientists, who tend to see picking and choosing between datasets, ad hoc and multiple methods as opportunities to select the ones that produce their desired results.

This highlights the common wisdom of asking “What about all the catastrophe theories we have seen adopted and later abandoned over the years?” And while climate scientists dismiss such questions denial, after you have witnessed the rise and fall of countless environmental hysterias over the years, you become more circumspect, and adjust your estimates of confidence to account for the low level of diligence in the field.

  • LogicalChemist

    David, how come you haven’t been asked to work on any of the IPCC reports or to testify before Congress?  This blind ignorance on the part of the climate science community certainly hasn’t made the news.

  • Pingback: pokazy gotowania()

  • Pingback: zobacz oferte()

  • Pingback: zobacz tutaj()