The National Academy of Science (NAS) Investigation


The National Academy of Science (NAS) Investigation


     Mann’s complaint against Mark Steyn, the National Review, Rand Simberg and the Competitive Enterprise Institute tells us that numerous organizations have investigated Mann’s behavior and concluded he did not commit academic fraud. This is true. Mann’s complaint continues with the following, which is misleading:


     “every such investigation - and every replication of Dr. Mann’s work - has concluded that Dr. Mann’s research and conclusions were properly conducted and fairly presented.” (Mann M., 2012)


     This part of Mann’s complaint stretches the truth a bit too much. All the investigations found serious problems with Mann’s work. This section will only discuss the National Academy of Sciences investigation. The National Research Council is the working arm of the United States National Academies, which includes the National Academy of Sciences. The National Research Council conducted the investigation and wrote the report.

     While Mann did not commit academic fraud, his hockey stick was incorrect, and his methods were explained poorly in his papers. His refusal to share his data and the details of his methods slowed the natural scientific process and allowed time for the IPCC and Al Gore to continue to promote his flawed temperature reconstruction.

     The National Academy of Sciences concludes that little confidence (National Research Council, 2006, p. 4) can be placed in Mann’s primary conclusion that “the 1990s are likely the warmest decade, and 1998 the warmest year, in at least a millennium” (Mann & Bradley, 1999).

     Academic fraud includes intent, that is Mann would have to have been found to intentionally falsify his data or his statistical analysis to deceive. In the definition from the University of Virginia:


     “False data is the fabrication or alteration of data to deliberately mislead. For example, changing data to get better experiment results is academic fraud.” (University of Virginia, 2020)


     Thus, absent proof that Mann did the data manipulation deliberately, he is not guilty of academic fraud. The National Science Foundation makes this clear in their final “closeout” assessment of Michael Mann’s hockey stick work:


     “Although [Dr. Mann's] data is still available and still the focus of significant critical examination, no direct evidence has been presented that indicates [Dr. Mann] fabricated the raw data he used for his research or falsified his results. Much of the current debate focuses on the viability of the statistical procedures he employed, the statistics used to confirm the accuracy of the results, and the degree to which one specific set of data [bristlecone pines] impacts the statistical results. … Such scientific debate is ongoing but does not, in itself, constitute evidence of research misconduct.” (National Science Foundation, 2011)


     One of the original scientific papers that Mann published about the hockey stick was published in When the evidence of data tampering was published by McIntyre and McKitrick (McIntyre & McKitrick, 2003) they submitted an extended critique of the errors and misrepresentations in Mann’s 1998 and 1999 papers to Nature magazine.

     The papers by McIntyre and McKitrick, Soon and Baliunas and the Nature submission led to the National Academy of Sciences (NAS) investigation. Their results were announced on June 22, 2006 (North, et al., 2006). In this report they found that the final 400 years or so of Mann’s temperature reconstruction were reproducible and they had confidence that the mid-to-late-twentieth century rise in temperatures was unprecedented since 1600AD. This is uncontroversial, since 1600AD was in the deepest part of the LIA, but this conclusion is quite different than the “warmest … in at least a millennium” claimed by Mann and Bradley.

     The hockey stick from 1000AD to 1600AD was very much in doubt according to the Academy (National Research Council, 2006, pp. 18-20). Prior to 1600, there are too few temperature proxies, in too few locations, to tell whether the world was warmer or cooler than the late century. The scarcity of precisely dated proxies before 1600, and the short length of the instrumental record, used to calibrate the proxies, made the hockey stick, and Mann’s conclusions, questionable.

     The National Academy of Sciences’ review of Mann’s work barely mentions data manipulation, but state that systemic uncertainties in climate records from before 1600 were not communicated in his papers as clearly as they should have been. We will often refer to the National Academy of Sciences report as the “NAS report.” Some references will also refer to it as the National Research Council report or the NRC report. NAS did not feel the hockey stick should have been featured so prominently in the IPCC third assessment report (TAR). One of the NAS panel members, Kurt Cuffey said the following about the prominence of the hockey stick in TAR:


     “I think that sent a very misleading message about how resolved this part of the scientific research was.” (Brumfiel, 2006)


     Nature (Brumfiel, 2006) reported on the NAS study under the misleading title, “Academy affirms hockey-stick graph.” The Academy did not affirm the graph, they simply said that they had confidence in the last 40% of the hockey stick. NAS points out that others had produced reconstructions like Mann’s from 1600 to the present, this is true, but there are also many peer-reviewed reconstructions that are dramatically different. A good example is shown in Figure

     The NAS Press Release was entitled: “High Confidence That Planet Is Warmest in 400 Years; Less Confidence in Temperature Reconstructions Prior to 1600” (North, et al., 2006). This is true, but misleading because it is obvious, hockey stick or no hockey stick. Directionally, the hockey stick is correct since 1600, but the magnitude is off by a factor of three or more according to historical records. The hockey stick offers nothing we didn’t already know and is quantitatively wrong.

     The NAS press release is heavily spun to support the hockey stick as much as possible. It also contains a statement that is clearly false, “None of the reconstructions indicate that temperatures were warmer during medieval times than during the past few decades, the committee added.” In 2006, many reconstructions showed the Medieval Warm Period was warmer than today, one was shown in the First IPCC Assessment Report in 1990 (IPCC, 1990, p. 202). Even more dramatic evidence that the Medieval Warm Period was warmer than today in the Northern Hemisphere is presented in deMenocal’s paper in Science in 2000, six years before the NAS report was published. His reconstructions of temperatures in Greenland, Bermuda and West Africa are shown in Figure


     Figure 21. Temperature reconstructions of the North Atlantic surface. The Greenland reconstruction is from borehole temperature surveys. The Bermuda and West African SST (sea surface temperatures) anomalies are from oxygen isotopic analysis of planktonic (shallow water) foraminifera shells found in subsea sediment cores. Source: (deMenocal, Ortiz, Guilderson, & Sarnthein, 2000) used with Peter deMenocal’s kind permission.


     The full NAS report is entitled: “Surface Temperature Reconstructions for the Last 2,000 Years” (National Research Council, 2006). It is much more nuanced and critical of the hockey stick than either the press release, Scientific American or the Nature articles suggest. So much so, that Senator James Inhofe announced, after it came out, that the “NAS report reaffirms ‘Hockey Stick’ is broken” (Inhofe, 2006). After reading the report, we think this is more accurate than the Nature headline claiming the report “affirms” the hockey stick. The report is detailed, 161 pages long, and clearly says the hockey stick has many flaws.

     The recommended citation says that the “National Research Council” is the author, but inside the report we can see the list of scientists involved. There were twelve authors working under the leadership of Gerald North, the chairman of the Committee. He is a professor of meteorology and oceanography at Texas A&M University.

     Gerald North thinks that humans are causing global warming with their greenhouse gas emissions and that the warming might become dangerous at some point (Dressler & North, 2013). He is biased but has studied climate for decades and has a healthy respect for the uncertainties in the field. He has written that “there remains a large uncertainty in climate sensitivity to a doubling of The range given by the IPCC today is 1.5°C to 4.5°C, which is the same range given in the Charney Report in 1979 (Charney, et al., 1979). He admits that the main sources of error in climate models are that we cannot model precipitation, feedbacks to additional and temperature increases (North G. R., 2013). This is something any informed climate change skeptic might say. Basically, the uncertainty in the data is clear to Gerald North, but his opinion is that climate change is man-made and dangerous. Fair enough.

     The others on the committee are Franco Biondi, an expert on using tree rings to infer ecological changes; Peter Bloomfield, a statistician from the University of North Carolina; John Christy, professor of atmospheric science at the University of Alabama; Kurt Cuffey, professor of geography, University of California; Robert E. Dickinson, professor of atmospheric sciences at the Georgia Institute of Technology; Ellen R. M. Druffel, professor at the University of California, Irvine in Earth sciences, an expert in radiocarbon dating; Douglas Nychka, professor of mathematics at the Colorado School of Mines, an expert in statistics; Bette Otto-Bliesner a scientist at NCAR (National Center for Atmospheric Research) in Colorado, an expert in paleoclimatology; Neil Roberts, professor at the University of Plymouth (U.K.), expert in using isotope data to infer paleo-temperatures; the late Karl Turekian, professor of geochemistry at Yale, expert in using radiogenic isotopes to infer paleoclimate; and John (Mike) Wallace, professor of atmospheric sciences at the University of Washington.

     This is an impressive list of scientists, John Christy was and is skeptical that human-caused global warming is dangerous or significant. Dr. Christy tells us that, in his opinion, Mike Wallace, Karl Turekian, and Franco Biondi were very reasonable and open minded on the issue. So, it cannot be said the composition of the committee was entirely biased.

     However, besides these members, the rest are connected to Mann and his co-authors in one way or another and probably biased, at some level, in his favor. They believe that humans have a significant and potentially dangerous impact on climate. Regardless of the bias in their press release, the committee was honest in the report, and they showed how well they understand the scientific process. In the preface of the NAS report, we find this assessment of the controversy over Mann’s hockey stick:


     “hypotheses are proposed …. Other scientists work on the issue, producing supporting or negating evidence, and each hypothesis either survives for another round, evolves into other ideas, or is proven false and rejected. In the case of the hockey stick, the scientific process has proceeded for the last few years … Critics of the original papers have argued that the statistical methods were flawed, that the choice of data was biased, and that the data and procedures used were not shared so others could verify the work. … The reconstruction produced by Dr. Mann and his colleagues was just one step in a long process of research, and it is not (as sometimes presented) a clinching argument for anthropogenic global warming, but rather one of many independent lines of research on global climate change.” (National Research Council, 2006, p. ix)


     Mann’s hockey stick was a proposal, an idea, or hypothesis. It was the first hemisphere-wide multiproxy temperature reconstruction. Very flawed, statistically inept, but the first, none-the-less. The National Research Council believed, with some justification, that the problem was less with Mann’s work, and more with how prominently the hockey stick was displayed by the IPCC and Al Gore. Mann compounded the problem by withholding his data and code, the big problem is always the cover up.

     Soon, et al., 2003 and McIntyre and McKitrick, 2003 showed the hockey stick was wrong and explained why. The NAS report found the same problems.

     The NAS report confirms the hockey stick had its shape because the statistical methods Mann used forced it. Besides prominently displaying the graph, TAR repeats Mann and Bradley’s unsupported conclusions on page 2 of the Technical Summary:


     “New analyses of proxy data for the Northern Hemisphere indicate that the increase in temperature in the 20th century is likely to have been the largest of any century during the past 1,000 years. It is also likely that, in the Northern Hemisphere, the 1990s was the warmest decade and 1998 the warmest year ([their] Figure (IPCC, 2001, pp. 2-3)


     To substantiate this statement, they point to the hockey stick reconstruction, which is their Figure on page 3 of TAR. Once the graph was shown to be an artifact of flawed statistical procedures, the IPCC was humiliated. It was the last time they tried to show humans are the dominant cause of climate change with direct physical evidence. All subsequent reports try to make their case with climate models.

     According to John Christy, Gerald North said during the press conference, after the report was completed, that the report confirmed the IPCC conclusions. This was clearly an error on North’s part, his report explicitly says otherwise. Christy thought maybe North was using “confirm” in some vague, general sense. Probably he simply misspoke. But regardless the comment was taken to mean the NAS report confirmed the hockey stick which made Karl Turekian furious and he wanted to hold a second press conference to clarify that statement. But it didn’t happen.

     Turekian wanted the following points to be clearly made. First, Mann’s methodology was flawed and second, strip bark bristlecone pine tree rings are not useful for climate reconstructions. The NAS report explicitly does not recommend the statistical methods used by Mann in his study and acknowledges the flaws in Mann’s methods found by McIntyre and McKitrick (National Research Council, 2006, pp. 112-113 & 116).

     The NAS report cites digital experiments that constructed “pseudoproxies” on a known set of temperatures at Mann’s locations and then corrupted them with varying levels of statistical noise. They used Mann’s methods to reconstruct the temperatures. In every experiment the temperature variability over time was attenuated. In other words, the upper temperatures were lessened, and the lower temperatures were increased, relative to the actual values. That means that comparing measured temperatures today to proxy derived temperatures in the distant past is invalid, the proxy derived temperatures will always have less variability than existed at the time (National Research Council, 2006, p. 86). This finding alone invalidates the hockey stick conclusions about century warming.

     This is a problem with all regression-based statistical techniques, like principal components. There are other methods that do not lose as much temperature variability, but they can lose accuracy (National Research Council, 2006, p. 86). Long after the NAS report was published, Bo Christiansen and Fredrik Ljungqvist, built a temperature reconstruction using more recent proxies and a novel statistical technique that retained much more variability than any regression-based technique. They named their reconstruction technique “LOC,” which stands for local (Christiansen & Ljungqvist, 2012).

     The NAS report suggests that the true amplitude of temperature changes over the past 1,000 to 2,000 years is probably at least twice what is seen in a Mann-type reconstruction (National Research Council, 2006, p. 112). Christiansen’s reconstruction suggests the true variability is almost eight times higher than shown in the hockey stick. Christiansen was careful about the proxies he used, and each proxy was tested for significance against local temperature records and rejected if not found to correlate.

     Moberg’s reconstruction, shown in Figure also uses a statistical technique, but one that is designed to lose less variability than Mann’s. Moberg separated his yearly data from the lower resolution proxies and used a novel wavelet analysis technique to retain as much variability as possible in his reconstruction.

     Christiansen and Ljungqvist used 16 proxies in the end (Christiansen & Ljungqvist, 2012). All proxies reached back to 300AD or earlier and all correlated well to local, modern instrumental temperatures.

     Besides the variability lost to proxy noise, the spatial gridding process used in some reconstructions dampens variability. Variability is lost in every step. Christiansen and Ljungqvist explain that this is a systemic bias and, as a result, not included in the confidence intervals provided with Mann’s reconstruction (Christiansen & Ljungqvist, 2011).

     Each of the proxies used in all reconstructions must be calibrated to the modern temperature record for the area where the proxy is located, whether it is an ice core, sediment core, a tree or a coral reef. This calibration typically is done with linear regression. This calibration process removes variability and it suffers from the short length of the measured temperature record, which is generally less than 150 years. Long term, century-scale variability is not present in the measured temperature record. Also, most of the measured temperature record is within the period of human fossil fuel emissions. emissions affect corals, pollen, tree growth, tree rings and other temperature proxies. These problems affect nearly all proxy temperature reconstructions, but not measured temperatures.

     Statistically, the MWP and the LIA anomalies shown in Figures 19 and 20 are significant at greater than the 95% level (Christiansen & Ljungqvist, 2012). Michael Mann’s reconstructions (1998, 1999, 2008, and 2009) all fall outside the 95% confidence limits of Christiansen and Ljungqvist’s reconstruction in the LIA. It is notable that Huang’s global temperature estimates, based on worldwide borehole heat flux data and physical principals, for the past 500 years, are within Christiansen’s 95% confidence limits in the LIA (Huang, Pollack, & and Shen, 2008). Mann’s reconstructions are not supported by the Huang’s borehole heat flux data.

     Significantly, Christiansen and Ljungqvist’s reconstruction shows a MWP that is, statistically speaking, at the same level as the modern warm period. This is also true of Moberg’s reconstruction. This is in great contrast to Mann’s earlier reconstruction (Mann & Bradley, 1999). Mann’s revised reconstruction is shown below in Figure It was created in 2008 (Mann M., et al., 2008) to address the criticism of the 1999 hockey stick in the NAS report. It shows a distinct MWP that is statistically at the same level as the modern warm period, using only proxy temperatures and an apples-to-apples comparison. It is similar in variability to Moberg’s reconstruction. Mann’s 2008 reconstruction invalidates his own 1999 hockey stick.

     Mann published another paper in 2009 that attempts to show that, “while the Medieval Warm Period is found to display warmth that matches or exceeds that of the past decade in some regions,” it falls below recent levels globally (Mann M. E., et al., 2009). An interesting idea, but it still suffers from the serious flaw of comparing modern instrumental temperatures to regression-smoothed proxy data from one-thousand years ago. If Mann is going to convince anyone, he must ditch the short instrumental temperatures and rely only on proxies for his comparisons. Soon and Baliunas make this clear in their 2003 paper (Soon & Baliunas, 2003).


     Figure 22. Mann’s 2008 reconstruction was built to address the problems found in the hockey stick by the National Academy of Sciences (NAS). The original 1999 hockey stick, shown in the IPCC third assessment report, is shown as a heavy dashed gray line. Both reconstructions have been moved to a 1902-1973 zero line. The proxy temperature reconstruction ends in 1995, this is identified with the “splice” line, the instrumental extension that Mann shows in his paper is shown as a very faint line. The end of the reconstruction is at the same level as the Mann, 2008 Medieval Warm Period peak.


     Mann used regression-based methods in 2008, including spatial regression, so some dampening of paleo-temperatures is expected. Whoever told Dr. David Deming that “We have to get rid of the Medieval Warm Period” (U. S. Senate Committee on Environment and Public Works, 2006) must have been very disappointed. Dr. Deming testified that a major climate change researcher told him that in an email but did not identify the individual. The Medieval Warm Period is a major thorn in the side of alarmists because it was likely as warm as today, yet humans could not have caused the warming.

     The NAS report acknowledges the decline in tree ring proxy temperatures relative to measured temperatures in the century as shown in Figures 7 and The origin of the “hide the decline” scandal. They are of the opinion that this “observed discrepancy … reduces confidence that the correlation between these proxies and temperature has been consistent over time” (National Research Council, 2006, p. 117).

     The National Research Council committee found serious errors in nearly every part of Mann’s first two papers. Mann’s conclusions were incorrect, and his statistical techniques were inappropriate. They found that the principal components methods that Mann used, created hockey sticks from autocorrelated (meaning self-correlated) random numbers, exactly as described by McIntyre and McKitrick.

     It was known by most climate scientists in 2006 that Mann’s reconstruction from the present to 1600AD does not show enough cooling. But, since 2006 this has become even more likely. Temperatures, at least in the Northern Hemisphere, were 1.2 to 1.7°C cooler in the LIA than today as shown to be the case by S. P. Huang of Xi’an Jiaotong University and colleague’s borehole temperature flux calculations (Huang, Pollack, & and Shen, 2008). Huang’s borehole flux calculations also show the Medieval Warm Period to be comparable to today’s temperatures. The relative Northern Hemisphere average temperatures between the MWP, the LIA and today are unclear, but larger than the hockey stick shows.