Saturday, February 27, 2010

Quote of the day

From Democracy in America

[T]he BBC reporter asked [East Anglia climate scientist Phil] Jones whether he would concede that global warming since 1995 has not been statistically significant. Mr Jones replied: "Yes, but only just," and went on to note that there was a measured global warming of 0.12°C per decade since then, and that it tends to be harder to get statistical significance out of shorter time samples.

This led to a Daily Mail headline reading: "Climategate U-turn as scientist at centre of row admits: There has been no global warming since 1995."

Since I've advocated a more explicit use of the word "lie", I'll go ahead and follow my own advice: that Daily Mail headline is a lie. Phil Jones did not say there had been no global warming since 1995; he said the opposite. He said the world had been warming at 0.12°C per decade since 1995. However, over that time frame, he could not quite rule out at the traditional 95% confidence level that the warming since 1995 had not been a random fluke.

Anyone who has even a passing high-school familiarity with statistics should understand the difference between these two statements. At a longer time interval, say 30 or 50 or 100 years, Mr Jones could obviously demonstrate that global warming is a statistically significant trend. In the interview he stated that the warming since 1975 is statistically significant. Everyone, even climate-change sceptics, agrees that the earth has experienced a warming trend since the late 19th century. But if you take any short sample out of that trend (say, 1930-45 or 1960-75), you might not be able to guarantee that the particular warming observed in those years was not a statistical fluke. This is a simple truth about statistics: if you measure just ten children, the relationship between age and height might be a fluke. But obviously the fact remains that older children tend to be taller than younger ones, and if you measure 100 of them, you'll find the relationship quite statistically significant indeed.

What's truly infuriating about this episode of journalistic malpractice is that, once again, it illustrates the reasons why the East Anglia scientists adopted an adversarial attitude towards information management with regard to outsiders and the media. They were afraid that any data they allowed to be characterised by non-climate scientists would be vulnerable to propagandistic distortion. And they were right.
h/t Deltoid

No comments: