Arguments
In a blind test, economists reject the notion of a global warming pause
Posted on 18 September 2015 by John Abraham
Oh how resilient myths can be, even in the face of facts. This past week saw the publication of the third strong refutation of the myth that global warming had somehow stopped a decade or two ago. You would think that with 2014 the hottest year on record and 2015 almost certain to exceed that, and 2016 to potentially set yet another heat record, people would use common sense to conclude that global warming continues. You’d also think with ocean heating breaking records (as discussed here) and loss of ice around the world, any lingering doubts would be put to rest. But alas, for some reason, even more proof is needed.
The first paper, which I covered here looked at the actual temperature trends and found no statistically significant reduction in the rate of warming. The latest paper, just published in the Bulletin of the American Meteorological Society by Stephan Lewandowsky, James Risbey, and Naomi Oreskes, looks at the evolution of the terms “pause” and “hiatus.” The authors find that over the past decade or so, there has been a lot of interest in both the scientific community (as judged by papers covering the topic) and by the general public (as determined by web-search statistics).
In particular, the authors found distinct increases in web searches related to the so-called pause just prior to two major climate-change meetings. The article then asks two questions. First, has there been a pause? Second, why has there been such an intense interest in this so-called event?
The authors show that there is no unique pause in the data. They also discuss biases in the measurements themselves which suggested a slowing in warming that actually did not occur once the data were de-biased. Finally, they reported on recent work that displayed a common error when people compare climate models to measurements (climate models report surface air temperatures while observations use a mixture of air and sea surface temperatures). With this as a backdrop, the authors take a step back and ask some seemingly basic questions.
The authors looked at temperatures since 1970 which they defined as long-term warming. They separately focused on15-year trends which they termed “fluctuations” because they represent short term fluctuations of temperatures about a long-term trend. They find clear fluctuations in the temperature trend, for instance the trend centered around 1999 was larger than the trend centered around 2005. By definition there must be certain time periods that are faster than the long-term average and certain time periods that are smaller than the long-term average. That is the meaning of an average. The real question is, can claims of a “pause” be distinguished from these fluctuations?
The authors looked at all possible 15-year trends in global mean surface temperatures. They first find that every single 15-year period showed warming (all four major datasets). The authors also show that the most recent 15-year trend isn’t very different from other 15-year trends throughout the entire period. To quote from the paper,
For a ‘pause’ to be distinctive, it must deviate below the longer-term trend more than previous periods deviated above the longer-term trend – otherwise, it can be considered just a fluctuation like others observed in the past.
While this sounds compelling so far, the authors went even further. They subjected the data to a blind expert test. They evaluated people’s forecasting judgement because it reveals human perception of a particular set of data. The authors asked a group of economists (each with a Masters or PhD degree in economics or an allied discipline) to evaluate the trend in global temperatures without awareness of the source of the data.
The experts were told that the data referred to agricultural output and were asked questions about whether the agricultural output had “stopped”. In fact, the authors took exact statements from a climate contrarian, except they replaced words associated with global warming with statements associated with agricultural productivity.
In this blind test, the experts strongly rejected the agricultural “pause” conclusion. In fact, they found mention of a pause “to be misleading and ill-informed”. The experts were divided about whether the “pause” statement was also fraudulent. What is particularly convincing is that a blind test like this, which removes the effects of personal biases or preconceived opinions, is the gold standard for many research areas.
Seems strange to me: here in New England today it's in the mid-80s. Again. Temps have been above normal for most of the last two months, and not by just a little. But yet, I haven't seen or heard ANYONE anyone even suggest that this might be due to global warming.
[RH] Please don't use all caps.
@Sunspot,
Well, it's not. Not really. Summer-like periods in fall are pretty common. Still, the NOAA state of the climate report seems to affirm what you are saying. From Long Island to Maine looks like it was about 5°F warmer than usual. Particularly since the June report shows much of New England was maybe a couple of degrees below average, it was a nice swing.
Living in North Alabama, we are used to having hot spells and even warm periods any year. You might google "noaa climate data <your city state>". This will give you insights to compare with the historical record.
Weather is 'the dice' and climate is 'their loading.'
Bob Wilson, Huntsville, AL
I'm a long time New Englander here, just back from a hiking trip with friends to the White Mountains where it was quite warm for this time of year even at altitude.
We need to remember, though, that this summer, at least in the Boston and metro west area, struggled to reach over 90°F until mid July. Below normal conditions. So far it seems to be averaging out.
Long term trends are Climater, that's what matter. Although I believe the world is (clearly) warming, I don't always like this year-to-year or even month-to-month keeping score with the temps. Seems desperate to me somehow.
Let's go over the facts again:
Fact#1 - everyone is predicting that 2015 will be the warmest year on record. Again.
Fact #2 - The temps in New England have been above average for most of the past two months. I don't need noaa to tell me this, I watch the local weather every day, they report on the high for the day and what the historical average high is for that day. Most days we were way over the average, ten degrees or even more. This week it's been more.
So I made the simple observation that this obviously COULD BE could be the result of global warming, but it is NEVER never discussed in the media. I did not say that this warm spell is conclusively because of global warming, if the global average suddenly went up more than ten degrees we'd all be dead soon anyway. But to implicitly imply there can't be any connection by pretending that this weather is business-as-usual sure seems like another form of denial of reality, which is our biggest enemy at this point.
Knaugle - you say it isn't. Then you say that noaa agrees it probably is. Make up your mind...
Bwilson - I assure you I know the difference between weather and climate.
[RH] Please, no use of all caps. And also please try to tone it down a little.
Why is the effect of the current El Nino not mentioned? Does it not have an effect? As for the article to which the links in this piece take the readerwhy is the opinion of economists so crucial to the proponents of anthropogenic climate change as to be published here? They are not climate scientists so why shuld they be consulted? If a Denialist wrote a paper that relied on comments from anyone other than a "Climate Scientist" it would be stridently decried as not coming from a "Climate Scientist". The comments on the piece in The Guardian suggest that not everyone considers it to be a valuable contribution to the debate.
Jenna - year-to -year warming is how global warming is measured. This doesn't mean that the average worldwide temp must increase every single year or global warming is disproven, but when that does happen, as it certainly is currently, it sure looks like global warming to me. And pointing out these facts hardly seems like "desperation". As I said, Jenna, the weather in New England has been running above normal for the past couple of months. And, yes, that can be blamed entirely on the position of the jet stream, that is what determines the weather for the northern latitudes of the northern hemisphere. But scientists who have been studying the jet stream for decades will tell you that its behavior is bizzare and without historical precedent, and that is a likely consequence of global warming as the northern regions of our planet warm faster than the rest due to the loss of reflective ice.
There have been some significantly cooler than normal periods in New England, most notably the second half of last winter. Even though last December averaged a whopping 4 degrees f above normal. But these cooler-than-normal periods are also a likely consequence of global warming due to the slowing of the thermohaline circulation in the northern Atlantic ocean caused by the melting glaciers.
I get my global warming information from climate scientists, and they are currently extremely concerned with what they are observing on this planet. I will certainly listen to what they have to say, and will easily dismiss comments from anyone who presents information contrary to what the experts say. I feel confident that the vast majority of reputable climate scientists will easily agree with any comment I make regarding global warming.
Check and mate. Goodnight...
Anyway, I don't get my global warming information from internet
[RH] This is not an issue that anyone is in competition to win. Please tone it down.
@ryland — the point of using economists is that they have expertise at analyzing time series data, and didn't know that the data represented global temperatures. That's the most interesting part of the paper to me. The economists were told that the data represented world agricultural output, and that it had been claimed that there was a pause in its increase starting in 1998. They were asked to evaluate this claim. The general consensus was that the claim was misleading and/or fraudulent. The implication is that if the strong partisanship associated with global warming is removed, no hiatus or pause is evident from the data.
Some weaknesses I see: there were only 25 economists surveyed. They were told that 1998 was the start of the pause, so they might have suspected the purpose behind the study. It's not clear what dataset they were using ... probably the NASA GISS global temperatures (but starting when?) ... so the results might have been different with different datasets.
[Rob P] - This is the graph presented to the economists. It's NASA GISTEMP data disguised as world agricultural output - in order to eliminate, or reduce, any preconceived biases associated with climate change.
gregcharles @8, the economists where shown the NASA GISTEMP LOTI from 1880-2010 complete with a 5 year running mean, with the running mean value for a given year being the mean of that year plus the four previous years. The data they were shown is shown as Fig 3 of the paper. (Note, because this is a preprint, all figures and tables are shown at the end of the paper after the references.) The economists where told:
The six "test items" the experts where asked to respond to each refer to "the claim" by Mr X, but at least three claims are made in the quoted sentences, ie,:
1) That there was no growth in global "agricultural output" from 1998-2006 (first quote);
2) That there was no trend in global "agricultural output" from 1998-2010 (second quote, first clause); and
3) That 2010 was not significantly more productive "in any way" than 1998 (second quote, second clause).
Depending on how each economist filtered this into one claim, they may have had different responses. In particular economists who parsed the claim as "there was no trend" may well have responded differently to those who parsed it as "2010 was not significantly more productive than 1998". Further, although these claims are in fact claims made by Bob Carter, with "world agricultural output" subsituted for "global means surface temperature" (or what ever equivalent Carter used), they are not the more typical claim that "the trend from 1998-2010 is not statistically significant". This ambiguity raises questions about the interpretation of the economists responses.
More troubling are the introduction of "Mr X" as a "prominent ... critic of conventional economics", which arguably might prime economists to disagree with Mr X; and (most troubling of all) that NASA GISTEMP was used when Bob Carter always uses either HadCRUT or UAH. That is, the economists were not presented with the same data that Carter uses in support of his claims. With HadCRUTv3 (the data actually avaible from the CRU and Hadley center in 2010), the economists might have supported Carter's claims, or not. But this survey does not tell us.
The latter point means we cannot infer from the economists conclusions about incompetence and/or fraud that Carter himself is incompetent and/or fraudulent (although there is far stronger evidence of this elsewhere). Given the assumption that GISSTEMP better represent global temperature variation than HadCRUT, however, we can still draw the weak conclusion that it is either incompetent or fraudulent to conclude that there was no trend from 1998 to 2010 given the best available GMST index. We cannot, however, similarly conclude from the economists responses that it is incompetent and/or fraudulent to conclude that the trend is not statistically significant, or that 2010 was not more productive than 1998, within statistical significance. The reason for the restricted claim is that given ambiguity of the test items, the test results can only support the weakest claim.
Ryland,
It takes time for papers to be published. Looking at the graph given to the economists, the current El Nino you are concerned about had not yet started when the data for this paper was collected. Current data shows even more clearly that there has never been a pause. Economists were consulted because many deniers are economists. Your attempts to deny anything that supports AGW become weaker and weaker with time. Perhaps you should ask why you cannot accept data at face value.
The analysis by Tom Curtis @9 answers your post.
ryland @11, no it does not. It addresses different points entirely. It certainly makes no attempt to justify the breath taking double standard of drawing attention to the El Nino at the end of the data (with 2010 also being an El Nino year) while ignoring the El Nino at the start.
Thanks Tom.
The statistics in table 1 of the paper are interesting, and reveal a couple of problems. To make them easier to interpret, I report them below with the agreement, mean, T-stat and an approximate estimate of the 95% range. That estimate assumes the distributions are near normal, which of course cannot be the case given that only whole numbers can be selected.
The first thing to note is that the level of agreement (first number) of item (1) and item (2) sum to 1.04. That is, 1 in 25 economists both agreed that "the data confirmed the claim" and that "the data contradicted the claim", ie, they agreed with two contradictory statements. Some might cynically suggest that in this survey, the economists performed better then they typically do at avoiding contradiction. More kindly, we may just assume this represents a problem as to how to parse the three effective claims made in the sample quotes into a single claim. (See 9 above)
A similar problem exists between the first, third and fourth items, in which at least 3 respondents both agreed with item (1) and item (3) and/or (4). Presumably the same 3 that agreed with both (1) and (3) also agreed with (4), but that is not certain. In any case, while not strict contradictories, agreement with both (1) and (3) or (4) represents an incoherent position, requiring that you believe both that the data confirm Mr X's opinion, but that Mr X's opinion is either misleading or ill informed given that data.
This most likely indicates that at least 12% of respondents treated the questions as asking about the disjunction (ie, A or B or C) of Mr X's three effective statements rather than their conjunction (A and B and C).
The second thing to note is that although the statistics are correctly calculated (SFAIK), that is consistent with 6 out of 25 respondents considering Mr X's "claim" to not be misleading, and/or to not be illinformed; while 9 out of 25 consider Mr X's "claim" to have been supported by the evidence. That is, the survey results show that the idea that Mr X's "claim" is supported by the data is controversial among economists, rather than that "economists reject" his claim given the data. (Based on this, I think the title of this post should be rewritten by inserting "a majority of" in front of economists to maintain accuracy, with similar corrections throughout the text.)
The third thing to note is that neither of results for items 5 and 6 are statistically significant. (See Table 1 in the paper)
Given these three things of note, it is worthwhile reviewing the claims made in the paper, which states:
Going through these claims, it is clear that "in aggregate" or "the majority of" experts "rejected the idea that the data confirm the statement and instead find that the data contradict the statement". It is certainly not true that "the experts" (without further qualification) did so, for a significant proportion found the opposite. The same is true (to a lesser extent) as to whether or not the experts found "the statement to be misleading and ill-informed".
Overall, I think the paper significantly overstates the nature of the results of the survey.
That does not establish that using economists as a benchmark establishes that there has indeed been a pause in the increase in global temperatures. For a start, at best it establishes that, using economists as a benchmark, such a claim is highly controversial (at best). However, we cannot go even that far due to the poor wording of the survey (discussed @9 above), it is very uncertain how various economists parsed the claims by the putative Mr X into a claim (singular) as required by the survey questions. The inconsistency may be entirely a product of different parsing methods. Ergo, it is entirely possible that the 36% of economists who considered "Mr X's claim" to have been supported by the evidence to have done so because they considered only the best supported of the three claims presented as having been supported. Given that the vast majority of those who thought the claim was supported gave it a rank of 4 out of 6 (ie, the lowest possible level of agreement) that is particularly likely.
Unfortunately, the ambiguity of the survey cuts both ways so that I think the survey can be used in support of a claim that there was no pause. In particular, even my conclusion @9 must fall by the wayside as the result about "fraudulence" was not statistically significant. Further, those economists who disagreed with "Mr X's claim" may have been testing against the stronger possible interpretations of the claim.
This is all beside the point in that the nonexistence of a "pause" or "hiatus" is already well established by statistical tests. Ambiguous results from economists eyeballing a graph has no bearing on that.
I have a number of serious difficulties with accepting the value of the "blind" test using economists that formed part of this paper. Had I been a reviewer it would have received harsh treatment; I don't know that it could have been salvaged.
1. Based on the text of the paper, the economists appear to have been asked to evaluate the validity of the claimed pause by visual inspection of a graph. What statistical tools can be brought to bear in such a case?
2. The graph of GISS LOTI was disguised as the value of global 1880-2010 agricultural output. However a) the LOTI plot has been widely reproduced and is quite recognizable; b) world agricultural output in constant dollars does not resemble the curve shown; and c) the scale of the plot values 'World Agricultural Output' at ~80% of global GDP. All of these may have led a knowledgeable economist to suspect a ruse, and reduced the 'blindness' of the test.
3. The wording of some of the questions posed to the test subjects is leading, in one case egregiously so. If such terms as 'misleading' and 'ill-informed' are arguably hostile, what can one say about "If incompetence is ruled out, the claim made about the data by Mr. X is fraudulent"?
I can only speculate, but I wonder if some of the hostility that has been directed at authors Lewandowsky and Oreskes by so-called skeptics is being returned, and is reflected in this work. If so, it would be a deeply ironic example of the 'seepage' that three of these same researchers discussed in their recent Global Environmental Change paper.
Magma,
I do not see anywhere in the paper wehre it states that the economists were not given the raw data from the graph to evaluate. It also does not say that they were given the data. It seems to me that you are assuming that the data was withheld without supporting information. It seems to me that in asking for a complete evaluation the data would have to be available if the economists wanted to look at it. It would be simple to convert the temperature data into agricultural data.
Please provide support for your claim that the economists were not shown the data.
michael sweet @17, the paper says:
In that statement, they use "graph" and "data" interchangably; the experts where "shown", not given the data; and the data is referred to as the "plotted data". All of these strongly suggest the data was presented in the graph shown as figure 3, and not also given as numerical values. Further, the stimulus given in such experiments is of the essence of the experiment. Had the experts also been given the numerical data, and that not been explicitly mentioned, that would represent a significant breech of experimental protocol. Therefore, absent clear statement to the contrary by one of the authors, we can safely assume that only the graph was presented.
Magma @16, with regard to your point (3), no "questions" were posed to the experts. Rather, statements were presented with the experts being asked to agree or disagree based on a six point scale. As disagreement is as easy as agreement, it is difficult to see how the survey items can be "leading questions".
In other respects, I agree with your comment.
Michael & Tom, @17-19: I interpreted the study's methodology exactly as Tom did. As well, the sentence before his quote in 18 started "The sample of economists 202 (N = 25) was tested online..." which I interpreted as likely being a survey carried out with limited time and opportunity for analysis. Note that this is different from the Associated Press 2009 article by Seth Borenstein referenced in the paper, which clearly stated that its disguised data was sent to economists for analysis. See Statisticians reject global cooling
Re. 'questions' vs. agreement or disagreement with statements, yes, I was somewhat careless with my choice of words. However I stand by the assertion regarding the leading or biasing nature of the statements the economists were asked to agree with, particularly the fifth one (incompetence/fraud).
As Tom Curtis notes above @9, the "5-year running mean" presented to the economists is a trailing average such as an economist might expect for forcasting purposes, rather than the usual centered average as used for the multi-year averages of temperature anomalies. The 2.5-year lag forced on this 5-year mean changes the visual appearance of the 5-year mean, reducing the visual correspondence of this mean to the various prominent annual extremes in the temperature anomaly record.
For example, the prominent El Nino peak of 1998 followed by the deep La Nina minima of 1999 and 2000 is barely noticeable in the trailing 5-year average compared to that of the usual centered 5-year mean. (Similarly for the strong stratospheric-injecting volcano events of the early 1980s and 1990s.)
A subtle, but very neat touch to further disguise the real identity of the underlying data.
GP Alldredge,
The shape of a 5 year trailing mean and a 5 year centered mean is exactly the same. It is just shifted back 2.5 years. If the "1998 peak" is not visible in the trailing mean it will also not be visible in the centered mean. 1998 is not visible because it is a short weather spike that is averaged out.