In the Obama era, it was routine for press releases to avoid mentioning any margin of error.
It will be hard to notice when President Trump does something worthy of hysteria if everything he does is greeted with hysteria. Take claims that he’s laying siege to the alleged chastity of climate scientists. This is one subject where it might be wise not to rely on the reflexive media narrative.
The year 2016 was the warmest ever recorded—so claimed two U.S. agencies, NASA’s Goddard Institute for Space Studies and the Commerce Department’s National Oceanic and Atmospheric Administration. Except it wasn’t, according to the agencies’ own measures of statistical uncertainty.
Business World Columnist Holman Jenkins Jr. on why the Trump Administration should reform NOAA and NASA. Click on image above to see full Opinion Journal video
Such fudge is of fairly recent vintage. Leaving any discussion of the uncertainty interval out of press releases only became the norm in the second year of the Obama administration. Back when he was presenting the 2008 numbers, NASA’s James Hansen, no slouch in raising climate alarms, nevertheless made a point of being quoted saying such annual rankings can be “misleading because the difference in temperature between one year and another is often less than the uncertainty in the global average.”
Statisticians wouldn’t go through the trouble of assigning an uncertainty value unless it meant something. Two measurements separated by less than the margin of error are the same. And yet NASA’s Goddard Institute, now under Mr. Hansen’s successor Gavin Schmidt, put out a release declaring 2014 the “warmest year in the modern record” when it was statistically indistinguishable from 2005 and 2010.
Nowadays Goddard seems to mention confidence interval only when it’s convenient. So 2015, an El Niño year, was the warmest yet “with 94 percent certainty.” No confidence interval was cited one year later in proclaiming 2016 the new warmest year “since modern recordkeeping began” when in fact the difference versus 2015 was a mere one-quarter of the margin of error.
Commerce’s NOAA makes a fetish of ignoring confidence interval in its ranking of the 12 warmest years. Yet when statistical discipline is observed, 2015 and 2016, the two El Niño years, are tied for warmest. And the years 1998, 2003, 2005, 2006, 2007, 2009, 2010, 2012, 2013 and 2014 are all tied for second warmest.
In other words, whatever the cause of warming in the 1980s and 1990s, no certain trend is observable since then.
Shall we posit a theory about all this? U.S. government agencies stopped mentioning uncertainty ranges because they wanted to engender a steady succession of headlines pronouncing the latest year unambiguously the hottest when it wasn’t necessarily so.
This doesn’t mean you should stop being concerned about a potential human impact on climate. But when government scientists deliberately seek to mislead, it’s a warning to raise your guard.