As person, I’ve always been a contrarian. To quote Bruce Springsteen: when they said sit down I stood up.
As a scientist, I’ve wedded my contrarian nature with a healthy dose of skepticism. When the majority of people in my field believe one thing, I’m the one looking for reasons they are all wrong.
So when scientists began to conclude that human beings were contributing to global warming, I was dubious. However, as I read various reports on global warming, my skepticism began to wane. A wide range of studies all seemed to be pointing in the same direction.
Then the Wall Street Journal came out with an editorial appealing to my contrarian instincts, warning readers to: “beware claims that the science of global warming is settled.” In an editorial from February 5, 2007, the Journal claimed to show just how much the Intergovernmental Panel on Climate Change, or IPCC, had been backpedaling on key predictions.
This wasn’t minor backpedaling in the view of the Wall Street Journal, but “startling revisions of previous UN predictions.” In the 2001 report, “the UN’s best high-end estimate of rising sea levels by 2100 was 3 feet.” The newest report has a high-end estimate of only 17 inches.
At this point in the editorial, my contrarian desire to join the Journal in questioning global warming slammed into another of my personality characteristics: I hate when people insult my intelligence. I noticed that the Wall Street Journal failed to mention what the low-end estimate was from 2001. And in this omission lies the fallaciousness of their thinking, or more sinisterly, the deviousness of their ways.
It has to do with confidence. Confidence intervals to be precise. Scientific estimates are not perfect, and therefore when scientists begin to collect data on a new topic, their estimates come with margins of error attached — confidence intervals, the plus or minus statements so familiar in political polling. For example, after a couple basketball games, it is difficult to predict what the season-long scoring average is going to be for, say, LeBron James. If he gets 10 points the first game and 30 points the second game, does that mean he’ll have a season scoring average of 20? We can statistically estimate what his scoring average will be, but we need to put a wide confidence interval around that estimate. As the season progresses, we’re going to have a much better idea of the average number of points James is going to be putting up per night. Our high-end estimate of his scoring average will come down, and our low-end estimate will come up.
Confidence intervals, you see, get narrower both from the top end and the bottom end.
Climate change of course is a much more complex thing to predict than basketball scores or election polls. There are many more sources of uncertainty. So we can expect wide confidence intervals, as scientists begin to grapple with the complex problem. If we want to know whether people are reversing their scientific predictions, we need to look not only at how they have revised their high-end estimates but also how they have revised their low-end estimates.
Somehow this balanced approach seemed to escape the Wall Street Journal editorial writers. They only mentioned the high-end estimates. And they seemed to act as if scientists, when revising their estimates over time, are somehow reversing their opinions.
Can we blame the Wall Street Journal editorialists for overlooking the importance of confidence intervals? After all they’re not scientists or statisticians, they’re just reporters.
But I think it is hardly possible that they could have been unaware of confidence intervals. After all, these people have advanced knowledge of financial and investment matters. And any knowledgeable investor knows that the more you diversify your investments, the more you minimize your risks. If you buy one or two stocks, both could crash and burn this year or both could take off. Your predicted yield might be, say, 10% per year, but this prediction would come with a wide confidence interval–you might lose everything, or you might make a fortune.
By contrast, if you buy lots of stocks, or if you buy mutual funds that invest in hundreds of companies, your average return will remain around 10%, but your confidence interval will narrow. Not all of these companies can tank at the same time, and not all of them can quadruple in value. So even though your expected return will be the same, your high-end and low-end estimates will be dramatically different.
So how could the Wall Street Journal, of all publications, have made such an egregious mistake? Did they overlook parallels between market diversification and global warming forecasts? Or did they purposely mislead readers? Are they fools or liars?
To avoid casting aspersions on them, perhaps it is best to simply say that they were full of hot air!
Peter Ubel is Professor of Medicine and Psychology at the University of Michigan, author of You’re Stronger Than You Think, and is currently writing a book on capitalism and human nature.