Monday, December 24, 2007

Anatomy of a lie

Well, perhaps not exactly a "lie," but a clear and deliberate misrepresentation of very carefully chosen partial facts intended to mislead. This is only a single example of a massive amount of mis- and dis-information circulating in the climate change debate; it is however a very represetative example.

Ol' Tom Nelson, formerly of Ivorybill Skeptic fame, writes:

"Note that according to RSS MSU satellite data, November 2007 was a whopping 0.915 °C colder than April 1998."

He provides a supporting link for his claim. If you actually go to the link and look at the data, you find several interesting things:

1. The data set for satellite temperatures runs from 1979 to the present.

2. November 2007 was all of 0.014 C below the 28-year average.

3. November 2007 was the first month in this Century to show a below average global temperature; the last month with a below average temperature was January 2000 (yes, folks, the year 2000 was in the 20th Century). So last month broke a 93 month run of above average temperatures.

4. The month chosen for comparison, April 1998, was the warmest month of the entire data set. It was in fact exceptionally, freakishly warm: its anomaly of +0.901C is 0.169 C higher than the second warmest month (February 1998). Indeed that run of hot weather in 1998 is unparalleled before or after in the dataset; every single month with an anomaly greater than +0.509C occured in 1998.

5. If you look at the data other than for 1998, you see in fact that the aughts have indeed been significantly warmer than the '80s. During the 1980s, a majority of months (79/120) had below average temperatures. I suspect anyone who applies a non-parametric test of the trend of negative versus positive anomalies will find a significant upward trend. And, as the large positive anomaly in 1998 skews the data away from a normal distribution, the non-parametric test would be more appropriate than a linear regression model which assumes normal distribution of residuals.

Conclusion: The claim as stated is grossly misleading to an extent that borders on malicious deceit.

Now, where have we seen this pattern before? Hmmm... selective data presentation deliberately chosen to bolster an a priori conclusion, ignoring or deliberately obfuscating the larger context and pattern of the data.. sounds vaguely familiar...

OK now to put all this in its own larger context:

1. (Scientific fact) Even if a 28 year dataset does show a warming trend, this is still within the range of normal long-term climate cycles so it is only consistent with, not proof of, anthropogenic global warming.

2. (Personal opinion) The sin of selective use of data and trying to prove long-term trends from isolated events is being committed left and right by all sides in the climate debate. But such blatant and apparently malicious examples of it do seem to be coming primarily from the "global warming skeptics."


Post a Comment

<< Home

Site Meter