Last week we reported on a new paper from NOAA scientists reassessing, and radically cutting, their estimate of atmospheric warming. One of the details that caught our eye was the pains the authors took to point out that near the end of the temperature record the rate of increase appears to rise. So the authors announced an apparent acceleration in warming, possibly as enough of a nod to the climate alarm panic to ensure their paper could get published at all. But in a post at Judith Curry’s Climate Etc. blog, economist Ross McKitrick points out two problems with this claim. First, whenever the data have gone in the other direction, showing an apparent slowdown or even halt, the alarmists are quick to insist that you can’t cherry-pick one little segment, you have to use the whole data set to measure the trend. Sauce for the goose, etc. Second, McKitrick noted that the NOAA authors didn’t do a formal analysis, they just eyeballed their claim of acceleration. So he crunched the numbers and found that there’s no evidence of statistically-significant acceleration anyway.
With any data set, including a record of a series of coin tosses or dice rolls, if you look only at short intervals you can find greater or lesser trends. The important question is whether the deviations are greater than random chance, which requires the kind of well-established statistical modeling economists do all the time, but which climate scientists tend to go to great lengths to avoid doing. While we won’t pretend to follow the mathematical details of McKitrick’s analysis, we understand the conclusion:
“In sum, based on a preliminary analysis the new NOAA data do not support a claim that warming in the troposphere has undergone a statistically-significant change in trend. One of the data sets, from the mid-troposphere, shows evidence of acceleration but it was only temporary.”
So while the offhand remarks about acceleration gave the alarmists a bit of comfort amidst the shock of the STAR data set being revised to show much less warming, in truth it looks like it was an unfounded claim. Which oddly in this case means the dice were loaded, not that they weren’t.