On "Declining" Global Temperatures: "This Little-Discussed Fact... Over the Past Several Years... Average Global Temperature... Has in Fact Decreased..."
Of all the strange things in Steve Levitt and Steve Dubner's Superfreakonomics, perhaps the strangest of all is page 186:
with its remarkable claim that that "there is this little-discussed fact about global warming: while the drumbeat of doom has grown steadily louder over the past several years, the average global temperature during that time has in fact decreased."
Back when Superfreakonomics first came out, this cliam puzzled everybody: when an economist says that something has decreased, he or she means that there is some way of estimating a trend over some period of time that produces a statistically-significant declining trend. Nobody could figure out how to do that with global temperature data..
Now it turns out that what Levitt and Dubner meant was something different: that 2005 was the hottest year on record, and that that record high temperature had not yet been broken--so global temperatures had "decreased" from their high back in 2005.
There have been thirteen new global annual temperature records set in the past 130 years--about one per decade. Since a new global temperature record is set once every ten years, Levitt and Dubner's methodology would thus lead them to say that over the past century and a third global temperatures have been decreasing 90% of the time--yet over that interval temperatures in total have risen by almost a full 1C. There have been eight new global annual temperature records set in the past forty years. That's one every five years. So by their methodology over the past sixty years global temperatures have been decreasing 80% of the time--yet over that interval temperatures in total have risen by a full 0.5C.
What odds would you need in order to take a bet that we will not see a hotter year than 2005 before 2020?