The Climate Horse Race is a Statistical Tie

The past couple weeks have seen the Japan Meteorological Agency (JMA) and the National Oceanic and Atmospheric Administration (NOAA) release their latest datasets on global temperatures. Both the JMA and NOAA reported that June 2014 was the hottest June on record.

Here is Seth Borenstein's report over at the Associated Press:

The globe is on a hot streak, setting a heat record in June. That's after the world broke a record in May. The National Oceanic and Atmospheric Administration announced Monday that last month's average global temperature was 61.2 degrees, which is 1.3 degrees higher than the 20th century average. It beat 2010's old record by one-twentieth of a degree. While one-twentieth of a degree doesn't sound like much, in temperature records it's like winning a horse race by several lengths, said NOAA climate monitoring chief Derek Arndt.

---

Several points of confusion here. First off, according Arndt's profile at the American Public University, "he is the Chief of the Climate Monitoring Branch at NOAA's National Climatic Data Center in Asheville, North Carolina. He holds a Bachelor's and Master's Degree in meteorology from the University of Oklahoma, and is currently enrolled in OU's Ph.D. program in Adult and Higher Education Administration." It seems odd that NOAA's "climate monitoring chief" lacks a terminal degree in this field (aka a Ph.D. in climate science) and yet heads up this unit, and that he is apparently choosing to pursue a Ph.D. in "Adult and Higher Education Administration" rather than in climate science. Wouldn't it make more sense to have NOAA's Climate Monitoring Branch headed by an individual with a Ph.D. in climate science that was completed long before he/she assumed the position? And according to Arndt, this one-twentieth-of-a-degree (Fahrenheit, of course) difference between the global temperatures in June 2010 and June 2014 is "like winning a horse race by several lengths." Really? That clear a difference? Not likely. NOAA's global temperature anomaly history for June is available here. June 2010 was 0.69°C, and June 2014 was 0.72°C, for a difference of only 0.03°C (or 0.05°F – i.e., "one-twentieth of a degree"). Here is the problem: NOAA fails to include error bars on its estimates of global temperature. And make no mistake: there should be error bars on any and all discussions of local, regional, national, and/or global temperatures. Not to include error bars is simply bad science. The U.K. Met Office includes error bars on its global average temperature anomalies, and they are quite large. The 95-percent confidence range on the standard HadCRUT4 and HadCRUT3 datasets are plus-or-minus 0.09 to 0.10°C. For example, the HadCRUT4 global average temperature anomaly for 2012 was 0.45°C, with a 95-percent confidence range from 0.35 to 0.55°C. In other words, there is a 95-percent probability that the global temperature anomaly for 2012 is somewhere between 0.35 and 0.55°C. Back to the NOAA-NCDC data that Arndt is using. One reasonably presumes that NOAA's global temperature anomaly has about the same error range as the U.K. Met Office. Thus, the June 2014 global temperature from NOAA likely has a 95-percent confidence interval of about plus-or-minus 0.10°C. The difference between the June 2010 and June 2014 global temperatures was, according to NOAA, just 0.03°C. This hardly seems like "winning a horse race by several lengths." Actually, it appears that the global temperatures for these two months are a statistical tie – otherwise known as no significant difference. If you want to see what this looks like graphically, here are NOAA's June global temperature anomalies for 2010 and 2014 along with the 95-percent confidence intervals taken from the U.K. Met Office's global temperature datasets. If anyone thinks this is June 2014 "winning a horse race by several lengths," they need new prescription lenses. And if NOAA doesn't agree with my using the U.K. Met Office's error bars, then NOAA can provide error bars of its own for its datasets that we can all look at and compare with the U.K. Met Office's global temperature error bars. I have also previously discussed the significant differences between among global temperature anomaly datasets, which I termed anomalies in the global temperature anomaly. In some cases, the differences among various climate science organization datasets are massive. The Australian Bureau of Meteorology has 1998 as the warmest year on record for the planet, and by a mile, whereas other datasets such as NASA-GISS have 2010 as much, much warmer than 1998. Over at Slate.com, Eric Holthaus claimed that the "Earth just finished its warmest quarter-year ever." I guess an error analysis was just too much work to support this claim. If we use NOAA-NCDC data, the last three months were only 0.02°C warmer than the same period in 2010. Add in 0.10°C error bars plus-and-minus on both periods, and we have another statistical tie. I was already doing error analyses within first-year science classes at college, but somehow the climate science community can't be bothered on arguably one of the most important science policy discussions of all time? Just ridiculous. Thus, we see a general absence of statistical analyses by NOAA and other climate scientists/entities, and especially supposed science journalists, when discussing trends in climate data and when comparing two or more climate data points. In other words, we have an undeniable war on rigorous science by the same climate science community that continually screams about a purported war on science against them by conservatives. Psychological projection is near ubiquitous among the climate activist community. Whatever they accuse others of failing at are very often their own failings. The past couple weeks have seen the Japan Meteorological Agency (JMA) and the National Oceanic and Atmospheric Administration (NOAA) release their latest datasets on global temperatures. Both the JMA and NOAA reported that June 2014 was the hottest June on record. Here is Seth Borenstein's report over at the Associated Press:

The globe is on a hot streak, setting a heat record in June. That's after the world broke a record in May. The National Oceanic and Atmospheric Administration announced Monday that last month's average global temperature was 61.2 degrees, which is 1.3 degrees higher than the 20th century average. It beat 2010's old record by one-twentieth of a degree. While one-twentieth of a degree doesn't sound like much, in temperature records it's like winning a horse race by several lengths, said NOAA climate monitoring chief Derek Arndt.

Several points of confusion here. First off, according Arndt's profile at the American Public University, "he is the Chief of the Climate Monitoring Branch at NOAA's National Climatic Data Center in Asheville, North Carolina. He holds a Bachelor's and Master's Degree in meteorology from the University of Oklahoma, and is currently enrolled in OU's Ph.D. program in Adult and Higher Education Administration." It seems odd that NOAA's "climate monitoring chief" lacks a terminal degree in this field (aka a Ph.D. in climate science) and yet heads up this unit, and that he is apparently choosing to pursue a Ph.D. in "Adult and Higher Education Administration" rather than in climate science. Wouldn't it make more sense to have NOAA's Climate Monitoring Branch headed by an individual with a Ph.D. in climate science that was completed long before he/she assumed the position? And according to Arndt, this one-twentieth-of-a-degree (Fahrenheit, of course) difference between the global temperatures in June 2010 and June 2014 is "like winning a horse race by several lengths." Really? That clear a difference? Not likely. NOAA's global temperature anomaly history for June is available here. June 2010 was 0.69°C, and June 2014 was 0.72°C, for a difference of only 0.03°C (or 0.05°F – i.e., "one-twentieth of a degree"). Here is the problem: NOAA fails to include error bars on its estimates of global temperature. And make no mistake: there should be error bars on any and all discussions of local, regional, national, and/or global temperatures. Not to include error bars is simply bad science. The U.K. Met Office includes error bars on its global average temperature anomalies, and they are quite large. The 95-percent confidence range on the standard HadCRUT4 and HadCRUT3 datasets are plus-or-minus 0.09 to 0.10°C. For example, the HadCRUT4 global average temperature anomaly for 2012 was 0.45°C, with a 95-percent confidence range from 0.35 to 0.55°C. In other words, there is a 95-percent probability that the global temperature anomaly for 2012 is somewhere between 0.35 and 0.55°C. Back to the NOAA-NCDC data that Arndt is using. One reasonably presumes that NOAA's global temperature anomaly has about the same error range as the U.K. Met Office. Thus, the June 2014 global temperature from NOAA likely has a 95-percent confidence interval of about plus-or-minus 0.10°C. The difference between the June 2010 and June 2014 global temperatures was, according to NOAA, just 0.03°C. This hardly seems like "winning a horse race by several lengths." Actually, it appears that the global temperatures for these two months are a statistical tie – otherwise known as no significant difference. If you want to see what this looks like graphically, here are NOAA's June global temperature anomalies for 2010 and 2014 along with the 95-percent confidence intervals taken from the U.K. Met Office's global temperature datasets. If anyone thinks this is June 2014 "winning a horse race by several lengths," they need new prescription lenses. And if NOAA doesn't agree with my using the U.K. Met Office's error bars, then NOAA can provide error bars of its own for its datasets that we can all look at and compare with the U.K. Met Office's global temperature error bars. I have also previously discussed the significant differences between among global temperature anomaly datasets, which I termed anomalies in the global temperature anomaly. In some cases, the differences among various climate science organization datasets are massive. The Australian Bureau of Meteorology has 1998 as the warmest year on record for the planet, and by a mile, whereas other datasets such as NASA-GISS have 2010 as much, much warmer than 1998. Over at Slate.com, Eric Holthaus claimed that the "Earth just finished its warmest quarter-year ever." I guess an error analysis was just too much work to support this claim. If we use NOAA-NCDC data, the last three months were only 0.02°C warmer than the same period in 2010. Add in 0.10°C error bars plus-and-minus on both periods, and we have another statistical tie. I was already doing error analyses within first-year science classes at college, but somehow the climate science community can't be bothered on arguably one of the most important science policy discussions of all time? Just ridiculous. Thus, we see a general absence of statistical analyses by NOAA and other climate scientists/entities, and especially supposed science journalists, when discussing trends in climate data and when comparing two or more climate data points. In other words, we have an undeniable war on rigorous science by the same climate science community that continually screams about a purported war on science against them by conservatives. Psychological projection is near ubiquitous among the climate activist community. Whatever they accuse others of failing at are very often their own failings.