Spinning climate statistics

RE: Climate change

According to NASA’s Goddard Institute for Space Studies and the U.S. Commerce Department’s National Oceanic and Atmospheric Administration., 2016 was the warmest year ever recorded.

Except it wasn’t, according to the agencies’ own measures of statistical uncertainty. This statement was made by Holman W. Jenkins Jr. in The Wall Street Journal. Jenkins goes on to relate that in order to show a continuous increase in average annual temperature the Godard Institute only mentions confidence interval when it’s convenient. They declared 2014 “the warmest year in modern record” but statistically it was indistinguishable from 2005 and 2010. But in reporting for the year 2015, an El Niño year, they said it was the warmest yet “with 94 per cent certainty”. They claimed that 2016 was the “warmest year “since modern record keeping began” however no confidence interval was mentioned and factually the difference was a mere one-quarter of the margin of error. Two measurements separated by less than the margin of error are the same.

Commerce’s NOAA ignores confidence intervals in its ranking of the 12 warmest years to produce a graph that shows continuous warming. But when applying statistical discipline 2015 and 2016, the two El Niño years, are tied for warmest and the remaining 10 years are all tied for second. Another point made by Jenkins is that originally the years 2005 and 2010 were tied, now it is shown as slightly warmer, “just enough to impart an upward slope to any graph that ignores statistical uncertainty.” How’s that for spinning? William Brown, Burlington