The Spin Cycle is a reoccurring feature based upon just how much the latest weather or climate story, policy pronouncement, or simply poo-bah blather spins the truth. Statements are given a rating between 1-5 spin cycles, with less cycles meaning less spin. For a more in-depth description, visit the inaugural edition.
Today’s press buzz is about a new paper appearing in this week’s Science magazine which concludes that the “hiatus” in global warming is but a byproduct of bad data. The paper, “Possible artifacts of data biases in the recent global surface warming hiatus,” was authored by a research team led by Director of the National Oceanic and Atmospheric Administration’s National Climatic Data Center, Dr. Thomas Karl. Aside from missing the larger point—that the relevant question is not whether the earth is warming, but why it’s warming so much slower than the computer model projections—the paper’s conclusions have been well-run through the spin cycle.
The spin was largely conducted by the American Association for the Advancement of Science (AAAS), publisher of Science magazine, through its embargo campaign and the courting of major science writers in the media before the article had been made available to the general public (and other scientists). Given the obvious weaknesses in the new paper (see below and here, for starters), there seems the potential for more trouble at Science—something that Editor-in-Chief Marcia McNutt is up to her eyeballs with already.
One major problem with the new Karl and colleagues paper is that the headline-making finding turns out not even to be statistically significant at the standard scientific level—that is, having a less than 1-in-20 chance of being due to chance (unexplained processes) alone.
Instead, the results are reported as being “statistically significant” if they have less than a 1-in-10 chance of being caused by randomness.
More and more we are seeing lax statistical testing being applied in high profile papers (see here and here for recent examples). This tendency is extremely worrisome, as at the same time, the validity of large portions of the scientific literature is being questioned on the basis of (flawed) methodological design and poor application and interpretation of statistics. An illuminating example of how easily poor statistics can make it into the scientific literature and produce a huge influence on the media was given last week in the backstory of a made-up paper claiming eating chocolate could enhance weight loss efforts.
But, as the Karl et al. paper (as well as the other recent papers linked above) shows, some climate scientists are pushing forward with less than robust results anyway.
Why? Here’s a possible clue.
Recall an op-ed in the New York Times a few months back by Naomi Oreskes titled “Playing Dumb on Climate Change.” In it, Oreskes, a science historian (and author of the conspiratorial Merchants of Doubt) argued that since climate change was such an urgent problem, we shouldn’t have to apply the same 1-in-20 set of rigorous statistics to the result—it is slowing down the push for action. Climate scientists, Oreskes argued, were being too conservative in face of a well-known threat and therefore, “lowering the burden of proof” should be acceptable.