Global warming advocates ignore the boulders

He’s certainly not a scientist, nor, seemingly, is he scientifically literate.

In his latest steaming pile of op-ed on global warming, Mr. Will attempts to call into question the “settled science” of global warming by discussing virtually no science at all. Seriously — a bunch of politicians not being able to agree on a course of action does nothing to question the science. And likewise for businesses making a business decision. But it is the claim that there has been no recent warming that is what really bugs me. George almost gets it right earlier in the op-ed, when he says there has been no statistically significant warming in the last 15 years, but here he (and many others) sin by omission. If one follows the link back to the BBC interview with Phil Jones, one gets a better picture

Do you agree that from 1995 to the present there has been no statistically-significant global warming

Yes, but only just. I also calculated the trend for the period 1995 to 2009. This trend (0.12C per decade) is positive, but not significant at the 95% significance level. The positive trend is quite close to the significance level. Achieving statistical significance in scientific terms is much more likely for longer periods, and much less likely for shorter periods.

In other words, if one has a sufficiently noisy data set, it is always going to be possible to pick a subset of the data where the noise masks any statistically significant trend. It doesn’t mean the trend isn’t there, or that the best fit is a zero slope. When Jones says “Yes, but only just” he’s telling us that one can just draw a flat line through the data, but this means that one could also draw a line with a slope of 0.24C per decade through the data, and it would have the same importance — you can’t exclude warming at that rate, either.

Imagine this question being asked instead:

Do you agree that from 1995 to the present it’s possible there has been global warming at a rate as high as 0.24C/decade?

The answer would have to be essentially identical, i.e. it would have to be yes. You can only statistically exclude warming at a higher rate than that!

What one certainly can’t do (that is, with any intellectual honesty) is conclude that this is an absence of warming. Statistically speaking, if the best fit to the data were a line with no slope, one could rule out neither an increase nor a decrease — one could only quote a limit on those trends. That’s one of the things about science — we try and quantify our results, rather than bandy about generalities. You might force a sound-bite answer out of a scientist (or worse, get there by ripping a quote out of context), but the instinct is to properly qualify the result.

So what if you don’t want both of the above scenarios to fit Dr. Jones’ answer? If you want a statistically significant answer you have to do as he suggests and look at a longer set of data in order to beat the noise down (random noise will average out with the square root of the number of data points). Anyone who does experimental science knows this, and is one of those things that a scientifically literate person should know. So the choice of a short data set is a form of cherry-picking — selecting a data set in such a way as to present a misleading result. If one looks at a longer data set, a statistically significant trend does emerge, and it is one of warming.

George, you’re not a scientist. I had some respect for you in the days I used to read your opinion pieces, because you could and did make cogent arguments, even if I did not agree with you. But science is based on facts, not opinions, and when you have to misrepresent those facts to make your point, your conclusions aren’t worth the paper on which they are printed.

Update: there’s more