Summary: The internet overflows with commentary about climate change, much by scientists. Yet all this talk generates more heat than light. Here’s a small but telling example, about an IPCC finding that should be a standard note in articles about Obama’s Clean Power Plan — but is ignored by journalists and so little known.
“Ignorance and confidence are constant companions.”
— Into the Heart of Truthby John McAfee (2001).
Obama’s sweeping Clean Power Plan rests on a finding in Chapter 10 of Working Group I of the IPCC’s latest report, AR5 — something important and little known. See page 884, emphasis added…
“We conclude, consistent with Hegerl et al. (2007b) {i.e., chapter 9 of AR4}, that more than half of the observed increase in GMST {global mean surface temperature} from 1951 to 2010 is very likely due to the observed anthropogenic increase in GHG {greenhouse gas} concentrations.”
AR4’s statement about the effect of GHGs was similar (although put in its Summary for Policy-makers, not page 884): “Most of the observed increase is global average temperature since the mid-20th century is very likely due to the observed increase in anthropogenic greenhouse gas concentrations.”
In both AR4 and AR5 the IPCC defines “Very likely” as having a “likelihood of the occurrence/outcome” at “>90% probability”. That’s below the 95% standard usually used in both science research and making of vital public policy decisions (e.g., by the EPA and FDA).
This finding about the effect of GHGs is relatively little known compared to AR5’s better known finding in the Summary for Policymakers about all anthropogenic forcings…
“It is extremely likely {95%+ certainty} that more than half of the observed increase in global average surface temperature from 1951 to 2010 was caused by the anthropogenic increase in greenhouse gas concentrations and other anthropogenic forcings together.”
I learned of the AR5’s lower level of confidence about the effect of GHG from a comment by attorney Tom Curtis in a comment at Skeptical Science, There are few mentions of this on Google. I asked two climate scientists if they know of this (they didn’t) — understandable since it’s buried on page 884 of AR5. This factoid about the effect of greenhouse gases has several kinds of significance.
First, this is the IPCC finding most relevant to President Obama’s Clean Power Plan to reduce US CO2 emissions. Such sweeping policies usually demand 95% confidence or the equivalent in the underlying research. This doesn’t have it. It has 90%. Journalists and scientists should report it correctly.
Second, a central tenet of the publicity campaign about climate change has been the consensus of scientists about their strong confidence about the effect of greenhouse gases on global temperatures since 1950. So it’s important that we understand that confidence.
Surveys of varying quality have shown strong agreement among climate scientists with the IPCC’s findings. The most recent and detailed was a survey about this of approximately 6,550 scientists studying climate change published as “Scientists’ Views about Attribution of Global Warming” by Bart Verheggen et al in the 19 Aug 2014 issue of Environmental Science and Technology.
Unfortunately, the public has been misled about the consensus of scientists and their confidence by the propaganda campaign to enact strong public policy changes. Absurdly broad claims have been made about the “95%” or “97%” of scientists, with little or no factual support. Also the meaning of statistical confidence has been grossly distorted. For example, see the NYT op-ed “Playing Dumb on Climate Change” by Naomi Oreskes (Prof, History of Science at Harvard). As many have pointed out, much of this is incorrect…
About the meaning of these statistics
For a clear explanation of these matters see “The New York Times Op Ed page needs a statistician” by Michael Lavine (Prof Statistics, U of MA-Amherst) and “Significance Levels are Made a Whipping Boy on Climate Change Evidence: Is .05 Too Strict?” by Deborah G. Mayo (Prof at VA Tech). I strongly recommend reading the discussion in the comments of Mayo’s post.
For a more advanced discussion of these issues — which apply to all sciences — see “The fickle P value generates irreproducible results” by Lewis G. Halsey et al in Nature Methods, March 2015. It’s part of a larger and growing debate among scientists about proper use of statistical tools, driven by the growing realization that much of today’s research cannot be replicated.
“Error is a delicate concept; for if we can call on it at will, or willfully, then
it no longer explains anything or accounts for anything. And if we can’t
call on it when we need it, none of our theories … will stand up.— Philosophy of Science: A Formal Approach
by Henry Ely Kyburg (1968).
For More Information
If you liked this post, like us on Facebook and follow us on Twitter. For more information see The keys to understanding climate change and My posts about climate change. Also, see these posts about the IPCC…
- Climate scientists speak to us. What is their consensus opinion?
- Another disturbing article about climate change. Fortunately we have the IPCC!
- Is our certain fate a coal-burning climate apocalypse? No!
- The 97% consensus of climate scientists is only 47%.
To help you better understand today’s extreme weather
To learn more about the state of climate change see The Rightful Place of Science: Disasters and Climate Change
.
I think you should look at Figure 10.5 in AR5. My first thought was why would the GHG attribution be > 90% while the full anthropogenic contribution is >95%, when the other anthropogenic contributions likely had a cooling influence (anthropogenic aerosols). The answer – I think – is that both the anthropopgenic GHGs and other anthropogenic influences individually have quite large uncertainties. Hence, if you ignore the other anthropogenic influences it becomes possible that anthropogenic GHGs only contribute to less than 90% of the observed warming, with something else producing the other more than 10%. However, it is still the case that the anthropogenic GHGs probably contributed to more than 100% of the observed warming.
However, if you combine the anthropogenic GHGs and the other anthropogenic influences, you can constrain their uncertainties by considering the uncertainties in the other influences (solar, internal variability) which are small. Hence we can make a stronger stament about anthropogenic influences overall than we can about anthropogenic GHGs alone. So, to argue that we shouldn’t be acting yet because we’re only > 90% certain about GHGs, when we’re > 95% certain about anthropogenic infuences overall and the evidence suggests that GHGs alone probably contributed to more than 100% of the observed warming, seems to be giving undue weight to a somewhat arbitrary p-test.
I muddled the above a little. When I said
I really should have said something more like
I’m also not quite sure why you think that the Naomi Oreskes piece was wrong. As I understand it all she was really suggesting was that scientists are typically inherently conservative and will only accept something if they are pretty certain. Hence they will tend to make Type II errors (failing to reject a false null hypothesis) rather than Type I errors (incorrectly rejecting a null hypothesis).
In that light, consider the above. We are extremely certain (> 95%) that more than half of the observed warming between 1950 and 2010 was caused by the anthropogenic increases of GHGs and other anthropogenic influences. This means that we have rejected the null that less than half could be anthropogenic. However, we think that it is only very likely (> 90%) that more than half of the observed warming between 1950 and 2010 was caused by the anthropogenic increase of GHGs. If we were being statistically strict, this would mean that we would accept the null hypothesis that less than half was caused by anthropogenic GHGs.
However, all the non-GHG anthropogenic influences produce cooling (land use, aerosols), so how can it possibly be that more than half of the warming was anthropogenic while less than half was caused by anthropogenic GHGs? Therefore accepting the latter null is would almost certainly be a Type II error (we will have failed to reject a null hypothesis that almost certainly true given the other null hypothesis that we rejected).
ATTP,
(1) Do you get papers published that don’t meet the 95% test? Some examples would be interesting.
(2) I am familiar with the FDA’s drug approval process (as an interested observer). Your reasoning is clever, but would get your drug denied. Total washout.
(3) “I’m also not quite sure why you think that the Naomi Oreskes piece was wrong”
Was this text unclear to you? You have read enough posts here to know the drill: my statement, followed by links to supporting analysis by experts. I’m sure they’d like to see your rebuttal to their analysis.
No idea what this has to do with what I said. There are many papers published where this question would be irrelevant.
I too understand the standards of drug approval. I wasn’t trying to get a drug approved. Not all statistics has to be frequentist.
I read some of these, most of them seemed to simply be lengthy posts about statistics, but I couldn’t find what Oreskes had actually got wrong. She seemed to have correctly described Type I and Type II errors and the point in her article seemed pretty simply; scientists tend to avoid making strong claims unless they’re pretty certain. I guess one can disagree with what she was suggesting, but that’s different to her getting something wrong.
ATTP,
“Not all statistics has to be frequentist.”
Excellent. I look forward to your cites of major findings not supported at the 95% level.
“I wasn’t trying to get a drug approved”
OK, you don’t see how public policy decisions work. There is a big literature on this, but outside the scope of this to provide.
“I couldn’t find what Oreskes had actually got wrong”
OK, that’s interesting. Odd, but interesting. Since those were explicit critiques to NO’s op-ed, I suggest you post comments there asking for additional explanation. We’ll let readers here come to their own conclusions.
Are you just pissing about? It’s a bit irritating if you are. Try reading what I said again.
I realise this. Maybe read what I said again. My simple point was that if you were to accept the null hypothesis that less than 50% of the observed warming could be due to anthropogenic GHGs then you’d almost certainly be making a Type II error. Why? Because other information such as GHGs producing more warming than anthropogenic influences overall, and that it’s extremely likely that more than 50% of the observed warming is anthropogenic. Maybe it’s good that policy decisions make such Type II errors, but that doesn’t change that it will almost certainly be a Type II error.
Your second link ended the article with a comment from the blog owner saying I’m not sure Oreskes is guilty of any misinterpretation of p-values, or statistical methodology (except in too closely connecting statistical and substantive conclusions), never minding the bit about Fisher, or the problems with trials on second-hand smoke. I take her to be alluding to an informal standard of proof, at a substantive, not a formal statistical level which is essentially how I see it. A bunch of formal statistical critiques about an NYT oped just seems like pedantry and the authors should probably look up the meaning of strawman.
ATTP,
Personally I found this sentence by Prof Mayo more interesting — and going to the heart of the issue: “I take her to be alluding to an informal standard of proof, at a substantive, not a formal statistical level, and at the level of reaching policy.” I doubt that one in a million laypeople reading the op-ed would see that, the correct context of NO’s analysis. By design (NO is a skillful propagandist).
NO believes we should revamp the world economy based on “informal standards of proof”. It is a commonplace mode of analysis in the climate wars. One of the most incompetent publicity campaigns, ever.
Fabius,
I think what Naomi Oreskes is saying is more thoughtful than that, but – even so – you seem to be illustrating what I was suggesting. It’s one thing to disagree with what she has said, but another to claim that what she said was wrong. Even if it has been an incompetent publicity campaign, it’s not going to change what actually happens in the future. My great hope is that the majority of climate scientists are wrong, because if they aren’t we may regret not paying more attention to these campaigns – incompetent, or not.
ATTP,
Thank you for your comments. Time will provide answers to these questions.
ATTP,
I wasn’t going to bother with this, but on reflect this is too large to ignore.
“It’s one thing to disagree with what she has said, but another to claim that what she said was wrong.”
Prof Mayo’s defense of NO was to reposition what NO said from “formal statement” to “informal”. While generous of Mayo to lower the bar, that’s a gross misread of NO’s op-ed — which was presented as a formal statement of statistics. To consider that a validation of her analysis is absurd, imo. But this is, as I and others have repeatedly shown, typical of the climate wars — and why so many have lost confidence in them.
This game-playing and dancing is not how scientists behave when they believe they have a strong case. Details here: http://fabiusmaximus.com/2015/08/04/climate-science-public-policy-broken-87951/
I didn’t say validation and nor did Mayo. My suggestion was simply that Naomi Oreske was arguing that scientists are typically cautious. Pedantry apart, I still fail to see what she got wrong in her description of Type I and Type II errors.
Also, if you think you’re NOT playing this game, you need to look in a mirror. I tend to lose faith in people when they start to tell me how scientists behave. What is remarkable is how often people who tell me how scientists behave are not actually scientists. What they’re typically describing is some caracature of how scientists are meant to behave in some ideal world where everyone behaves as they should. They’re not describing how scientists actually behave.
ATTP,
“I didn’t say validation and nor did Mayo.”
That’s an odd objection, since I didn’t put “validation” in quotes. It was a description, and imo exactly describes what you and Mayo were doing.
“I tend to lose faith in people when they start to tell me how scientists behave.”
That’s a weird objection. “Faith”? I give examples — with links — and describe their significance in the public policy debate. Logical objections are to either the accuracy of my description or to specifics in my analysis. I don’t see you doing either.
“What is remarkable is how often people who tell me how scientists behave are not actually scientists.”
Are you saying that we’re supposed to be passive watchers of the public policy dance, unable to draw conclusions about the actions of our betters? One of the worst publicity campaigns ever.
I’m not trying to “validate” what Naomi Oreskes has said. I’m suggesting that her description of statistical tests and Type I and Type II errors seems reaonsable and that the criticisms of her description of statistics appears to be pedantry at best.
No, I’m suggesting that there is a difference between criticising the behaviour of some individuals, and making claims based on some kind of caracture of how you think scientists should behave. Personally, I find the whole “scientists don’t behave like that” claim a little irritating given that there isn’t some generic definition of how scientists should behave (well, not one that is a reasonable descriptor of reality).
Take the example of your post. It’s extremely likely that anthropogenic influences overall contributed to more than 50% of the warming since 1950, but only very likely that anthropogenic GHGs contributed to more than 50%. From a strictly statistical point of view, that would suggest that we accept the hypothesis that anthropogenic influences contributed to more than 50%, while rejecting the hypothesis that anthropogenic GHGs contributed to more than 50%. However, given other information, I think that those two positions are inconsistent and hence we have to either be comitting a Type I error (anthropogenic influences can’t be contributing to more than 50%), or a Type II error (anthropogenic GHGs are contributing to more than 50%). Does that make me a bad scientist?
ATTP,
“I’m suggesting that her description of statistical tests and Type I and Type II errors seems reaonsable and that the criticisms of her description of statistics appears to be pedantry at best”
Definition of validate: “demonstrate or support the truth or value of”. Looks to me like you are validating her analysis. And ignoring the two critiques I gave from experts in statistics.
“I’m suggesting that there is a difference between criticising the behaviour of some individuals, and making claims based on some kind of caracture of how you think scientists should behave.
I don’t know to what you refer. The following quote makes no sense to me as support for that, certainly not with respect to anything I’ve said.
“Does that make me a bad scientist?”
First, that conclusion has zero relevance to anything I’ve said. It certainly has no relevance to the quote you provide, in which I refer to your statement about the public’s right to assess scientists’ role in the public policy debate. People can draw their own conclusions; that doesn’t imply you are a “bad scientist.”
Second, you appear to be conflating two posts about two different subjects (I am unclear what you are saying, so this m/b incorrect).
ATTP,
Sidenote: The FM website is under attack by trolls, so I’ve turned up the anti-spam defense. Somehow your posts are affected, going into moderation as an unintended effect. I am of course approving them ASAP.
Okay, this is getting too convoluted to really continue, but
No, I’m not ignoring them. I’m suggesting that I can’t see anything in those critiques that really invalidates what Naomi Oreskes has actually said, unless one choose to base one’s critique on pedantry.
The relevance is that part of Naomi Oreskes point is that an understanding of the complexities of a physical system may allow one to draw a different conclusion to what one would draw if one rigidly stuck to stastistical analyes. Your post points out that the IPCC document use extremely likely when refering to all anthropogenic influences but uses very likely when referring to anthropogenic GHGs only. If we were rigid about the statistical analysis we would reject that GHGs have contributed to more than 50% of the warming, while accepting that anthropogenic influences overall contributed to more than 50% of the overall warming. Well, these positions seem inconsistent, so assuming that we should treat these independently and stick to the independent statistical analyses seem wrong, whatever some statisticians might try to tell you.
Anyway, it’s getting late here and I suspect this discussion has run its course.
ATTP,
“I’m not trying to “validate” what Naomi Oreskes has said.”
“I’m suggesting that I can’t see anything in those critiques that really invalidates what Naomi Oreskes has actually said.”
The linguistic theory you employ was described to Alice by a famous philosopher:
“The relevance is that part of Naomi Oreskes point is that an understanding of the complexities of a physical system may allow one to draw a different conclusion to what one would draw if one rigidly stuck to stastistical analyes.”
From that you conclude that if I disagree with your analysis I believe you to be a “bad scientist” (that’s the sentence to which I was replying). The concatenation of logic to get your conclusion is certain be fascinating, but probably also best described in Through the Looking-Glass.
“whatever some statisticians might try to tell you.”
I’ll rely on statisticians about statistics rather than you, and doubly so more than History of Science Prof Oreskes.
“this is getting too convoluted to really continue”
Yes, that’s certainly true. You are shadow-boxing with yourself rather than anything I say, or with what the experts I cite say.
Is there more than one of you? Sometimes you seem quite reasonable. At other times you seem like a complete arse.
ATTP,
(1) “At other times you seem like a complete arse.”
Insults are the most revealing of comments.
(2) “whatever statisticians might try to tell you”
You don’t agree with the experts I cite. That’s fine, but gives no support for your condemnation of my statements. You are not the Pope of Statistics to rule others invalid.
(3) You’ve frequently ignored or misrepresented what I’ve said, replying with rebuttals to things I’ve not said. This is what I mean by shadow-boxing. Two examples:
(a) “I’m suggesting that there is a difference between criticising the behaviour of some individuals, and making claims based on some kind of caracture of how you think scientists should behave.” Which you never explained, and has no obvious relevance to this post.
(b) “does that make me a bad scientist?” I asked you to explain this weird assertion, and you again replied with many words having no visible relationship to anything I’ve said.
You’ve been playing the man, rather than the ball, most of this thread.
What statements have you made? You’re appealing to authority. Make an actual argument!
You’ve frequently ignored or misrepresented what I’ve said, replying with rebuttals to things I’ve not said. This is what I mean by shadow-boxing. Two examples:
(a) “I’m suggesting that there is a difference between criticising the behaviour of some individuals, and making claims based on some kind of caracture of how you think scientists should behave.” Which you never explained, and has no obvious relevance to this post.
(b) “does that make me a bad scientist?” I asked you to explain this weird assertion, and you again replied with many words having no visible relationship to anything I’ve said.
None of the above suggests that I misrepresented what you said. Try reading the thread again. I’ve got better things to do, but a couple of hints. You said This game-playing and dancing is not how scientists behave.
ATTP,
“You’re appealing to authority”
Yes, that’s how public policy debates are done. Are you serious?
“a couple of hints. You said This game-playing and dancing is not how scientists behave.”
That would have been nice to know hundreds of words ago. Hint: when somebody asks for an explanation of what you’re talking about, explain. Nobody can read your mind.
Well, it would have be nice to have known that you weren’t intending to construct an argument of your own. Would have saved me the trouble of engaging in the first place.
Okay, but I thought that my immediate response to the comment where you made that statement would have made it obvious.
Ok. Out of all this, FM, what directly is your Point? Do you have an idea of what is “so”? Or just a simple, it is unclear still or yet? Help.
Breton
Breton,
I find that people seldom ask for my opinion because they consider it valuable — and even less often because they plan to act on it. Rather it’s the last chapter in the show. Entertainment.
The facts are what they are. Draw your own conclusions. Act as you see fit.
Book and documentary.
Youtube presendations by climate scientist Tim Ball (Prof at University of Winnipeg, retired; see Wikipedia).
Tim Ball’s website: http://www.drtimball.com
Debate between Dr Tim Ball and Elizabeth May — Scroll down to Ian Jessop part 1
http://www.cfax1070.com/Podcasts
Some comments:
1) In answer to FM’s question to ATTP: No you DON’T publish without at least 2-sigma (P=0.95) results. I authored or co-authored more than 60 refereed papers in a physical science (astronomy) and the weakest results were about 2.5 to 3 sigma.
2) ATTP’s comments that somehow he KNOWS that the real significance is better than P=0.90 is unscientific. He needs to come up with concrete estimates for the effects of non-GHGs forcing. Is he really sure that the non-GHGs effects are on the cooling side? (Part of this was written while in the middle of a large asphalt parking lot.)
3) I’m sure ATTP is a much better scientist than this, but too much worry about AGW induces a kind of madness.
Social bill,
Thanks for the simple clear reply.
You are clearly better at this commenting biz than I (not the first time I have noticed this).
Social Bill,
Have you followed the development of the new solar cycle time series? I’ve mentioned it a few times. I can provide cites if you’re interested.
Yes, but it’s not true that every paper published in the physical sciences is presented with results that have at least a 2 sigma significance. You’re thinking of examples where you can actually estimate the significance of a result. There are many examples where this isn’t the case. For example, high-resolution computational simulations. How does one estimate the significance of a single run (or a small number of simulation runs). Of course, if you’re delving into your simulations (think Millenium run, for example) you may well be able to determine the significance of your analysis, but it’s simply not true that every result ever published is presented with a significance. Of course, every result published may well be based on laws/hypotheses for which such a significance has been established, but that still isn’t quite the same as every result having at least a 2 sigma confidence.
I didn’t say KNOW. However, we do have estimates for the non-GHG anthropogenic forcings. If you look at Figure 10.5 the other anthropogenic influences are estimated to be from -0.6 to +0.1C. Anthropogenic GHGs are estimated to be responsible for +0.5 to +1.3C. Therefore if one rejects that anthropogenic GHGs caused more than 50% of the warming since 1950, I’d be pretty sure that that would be a Type II error.
I’m sure you’re a better person than this, but dismissing AGW induces a kind of (okay, I can’t think of a suitable word, but you probably get my point).
Pingback: “very likely” versus “extremely likely” | …and Then There's Physics
To FM: No. References would be appreciated, thanks.
To ATTP: Given that the GHG CO2 is steadily increasing I think we can all agree that at some point AGW will resume. On the statistics issue, if additional information about non-GHG effects becomes known, it should be explicitly incorporated into the error model. Finally, I think we can all agree that much more effort is needed in research (Manhattan Project sized).
Social Bill,
I don’t have time to run this down, but here’s a sketch. The US spends roughly $35 billion/year on non-military foreign aid. In 2015 Africa south of the Sahara is scheduled to get $1.4 billion — pocket change vs its population of almost a billion. Northern Africa gets more money, part of the WOT (and to help Israel).
https://en.wikipedia.org/wiki/United_States_foreign_aid#2000s
ForeignAssistance.gov
AGW has not stopped. Ocean heat content data alone should convince you of this. At best, surface warming has slowed. There is, however, no evidence to suggest that anthropogenic global warming has stopped.
It is essentially known (well, “known” is a poor term, but it’s not unknown). This is kind of the point. If you treat anthropogenic GHGs and other anthropogenic influences independently, it is sometimes hard to distinguish – in a fingerprint analysis – their specific contributions. As I understand it, the problem is essentially that if you want to do a detailed attribution study where you consider anthropogenic GHGs independently, it is sometimes hard to determine if some particular signature is GHGs, or some other anthropogenic influences. When you combine anthropogenic GHGs with the other anthropogenic influences, this problem is reduced and one can make a stronger statement about anthropogenic warming overall, than one can make about anthropogenic GHGs alone.
However, it is still likely that anthropogenic GHGs would – alone – have produced much more warming than observed, with the other anthropogenic effects having a cooling influence.
For perspective, scientists first confirmed (publicly?) the higgs at 4.9 sigma. Before putting trillions of dollars of drag on the world economy and condemning billions more in the generations to come to poverty please be more certain.
How disgraceful that politics was able to hijack the field of climatology and cause the corruption of the scientific method, the peer-review process, the assessment of climate science, scientists, scientific institutions, academia, the media and economists, all in the cause of the ideology of environmentalism and sustainability … disguised as being done to save the planet from a coming climate catastrophe based on flawed unvalidated projections from climate models.
Mervyn,
While I agree, this is an old story. Science is a weak institution, with shallow roots in our society. When science overlaps with powerful social forces, it accommodates them. This is a connecting theme of the books by biologist Stephen Jay Gould. For example, it was important to prove that Blacks were an inferior race, and for a century science obliged.
I believe it’s more useful to understand why this is, and build institutions that can resist these winds. Difficult to do, with occasional success the best we can hope for.
As I have so often written, the “real” solution is for us — the American people — to become more skeptical, less gullible, prizing truth over tribal loyalty. For more about this see the posts in section 4 here.
The old Fermilab had about 2 sigma data on the Higgs, just out of reach. It provided a bit of contributory evidence (Obama shut down their last year which might have provided a bit more.)
SocialBill,
Nice — long, meandering — discussion for laypeople of problems in modern science, the increasing number of retractions: “Science isn’t broken” at 538 (as usual for that house, firmly supporting our institutions.
Also see this at Nature: “Online debate erupts to ask: is science broken?“.
Pingback: Another example of PolitiFact’s climate propaganda | Watts Up With That?