follow CCP

Recent blog entries
popular papers

What Is the "Science of Science Communication"?

Climate-Science Communication and the Measurement Problem

Ideology, Motivated Cognition, and Cognitive Reflection: An Experimental Study

'Ideology' or 'Situation Sense'? An Experimental Investigation of Motivated Reasoning and Professional Judgment

A Risky Science Communication Environment for Vaccines

Motivated Numeracy and Enlightened Self-Government

Ideology, Motivated Cognition, and Cognitive Reflection: An Experimental Study

Making Climate Science Communication Evidence-based—All the Way Down 

Neutral Principles, Motivated Cognition, and Some Problems for Constitutional Law 

Cultural Cognition of Scientific Consensus
 

The Tragedy of the Risk-Perception Commons: Science Literacy and Climate Change

"They Saw a Protest": Cognitive Illiberalism and the Speech-Conduct Distinction 

Geoengineering and the Science Communication Environment: a Cross-Cultural Experiment

Fixing the Communications Failure

Why We Are Poles Apart on Climate Change

The Cognitively Illiberal State 

Who Fears the HPV Vaccine, Who Doesn't, and Why? An Experimental Study

Cultural Cognition of the Risks and Benefits of Nanotechnology

Whose Eyes Are You Going to Believe? An Empirical Examination of Scott v. Harris

Cultural Cognition and Public Policy

Culture, Cognition, and Consent: Who Perceives What, and Why, in "Acquaintance Rape" Cases

Culture and Identity-Protective Cognition: Explaining the White Male Effect

Fear of Democracy: A Cultural Evaluation of Sunstein on Risk

Cultural Cognition as a Conception of the Cultural Theory of Risk

« Three models of risk perception -- & their significance for self-government . . . | Main | Don't be a science miscommunicator's dope (or dodo) »
Thursday
Feb202014

Democracy and the science communication environment (lecture synopsis and slides)

Gave a talk earlier in the week at Cardiff University, the last stop on my fun "cross-cultural cultural cognition road trip." Cardiff's Understanding Risk Research Group features a '27-Yankees equivalent lineup of risk perception scholars--including Nick Pidgeon, Wouter Poortinga, Adam Corner & Lorraine Whitmarsh (I decided not to use that metaphor during my talk)--who are surrounded by top-notch sociologists studying technology and society. They also have a high-charged group of science communication scholars. I had an amaizing few days there & felt very sad when the time came to leave!
 
Slides from my talk are here. I can't quite remember how I put things, but it was something like this . . . .

0. What is this “science of science communication”?  The science of science communication can be understood as a remedy for two fallacies.

The first is res ipsa locquitur (“the thing speaks for itself”): the validity of valid science is manifest, making scientific study of it neither interesting nor necessary.

The second is ab uno disce omnes (“from one, learn all”): the scientific knowledge necessary to enable a doctor to meaningfully advise a patient on a complicated treatment decision is the same as the knowledge necessary to enable a science journalist to edify a curious member of the public, an empirical researcher to advise a policymaker, an educator to teach a high school student the theory of evolution, etc.

My remarks are mainly directed at the ab uno fallacy. I want to describe the distinctive species of SSC that is most likely to evade comprehension if one makes the mistake of thinking it’s only one thing. It is also the one that is arguably most important for the well-being of democratic society. 

The aim of this species of SSC is to protect the science communication environment.

1. The puzzle of cultural polarization over risk

Members of the public in the U.S. are highly divided on all manner of fact relating to climate change. So are members of the public in many other nations, including the UK.

There are other risks—from GM foods to nuclear power to gun ownership to vaccination against infection by HPV or other contagious diseases—that fracture the members of some of these socieites but not others.

Not to be struck by the puzzling nature of this phenomenon is to admit a deficit in curiosity. It’s not surprising at all that people with different values would disagree about what to do about a societal risk like climate change or gun possession. But there’s nothing in how much one weights equality relative to wealth, or security relative to liberty, that determines whether the earth is heating up as a result of human activity or whether permitting citizens to carry concealed handguns in public deters violent assaults.

It’s not surprising either that ordinary members of the public would disagree with one another on facts the nature of which turns on evidence as technically complex as that surrounding climate change, nuclear power, or gun control. 

But if complexity were the source of the problem, we’d expect disagreement to be randomly distributed with respect to cultural and political values, and to abate as individuals become progressively more comprehending of science. 

Not so: on the contrary, the most science comprehending members of the public are the most culturally polarized! (At least in the U.S.; I’m not aware of resarch of this sort with non-US samples & would be grateful to anyone who fills in this gap in my knowledge, if it is one).

What’s the explanation for such a peculiar distribution of beliefs—and on facts that not only admit of investigation by empirical means but that have in fact been investigated by expert empirical methods?

2. The cultural cognition thesis

The answer (or certainly a very large part of it) is cultural cognition.

Cultural cognition is a species of motivated reasoning, which refers to the tendency of people to conform their assessment of all manner of information (empirical data, logical arguments, brute sense impressions) to some goal or interest independent of forming a correct judgment. 

The cultural cognition thesis holds that people can be expected to conform their perceptions of risk and like facts to the stake they have in maintaining their connection to and status within important affinity groups.

The nature of these commitments can be measured by various means, including right-left political outlooks, but in our research we ordinarily do so with scales patterned on the “worldview” dimensions associated with Mary Douglas’s “group-grid” framework.

3. Some evidence

Studies conducted by myself and my collaborators have generated various forms of evidence in support of the cultural cognition thesis—and against rival theories that are often used to explain political conflict over societal risks.

a. Cultural cognition of scientific consensus. In one study, we performed an experiment that showed how cultural cognition influenced formation of public perceptions of what expert scientists believe. The results showed that how readily individuals of diverse cultural outlooks identified a scientist as an “expert” on climate change, nuclear power, or gun control depended on whether that scientist was depicted as espousing a position consistent with the one that prevails in the individuals’ cultural groups.

If individuals selectively credit and dismiss evidence of “expert” opinion in this fashion, they will become culturally polarized over what scientific consensus is in disputed issues.  And, indeed, the study found that in all cases the vast majority of subjects perceived that “scientific consensus” on the relevant issue—climate change, nuclear power, and gun possession—was consistent with the one that prevailed in their cultural group.

The study findings were not only consistent with the cultural cognition thesis, but also inconsistent with two alternatives.  One of these attributes political conflict over societal risks to one or another group’s hostility to science. In fact, no group subscribed to a position that it perceived to be contrary to prevailing scientific opinion.

The second alternative explanation sees one or another group as more attuned to scientific consensus than its rivals. But in fact, all groups were equally likely to view as the “consensus” among expert scientists the position contrary to the one endorsed as the “consensus” position by the U.S. National Academy of Science.

b . “Feeling” the heat—and the hurricanes, floods, tornados etc.  A common theme—indeed, the dominant for commentators who derive their explanations from syntheses of general literature rather than by original empirical research—attributes popular conflict over climate change to the public’s overreliance on heuristic, “system 1” as opposed to more reflective, dispassionate “system 2” information processing.

Those who advance this thesis typically predict that individuals will begin to revise upward their perception of the seriousness of climate change risks as they experience climate-change impacts first hand.  “Feeling” climate change, it is argued, will create the emotionally vivid impression that those who form their risk perceptions heuristically will require to start taking climate change seriously.

This prediction is also contrary to the evidence. 

It’s true that individuals’ perceptions of climate-change risk correlate with their perception that temperatures in their area have been increasing in recent years. But their perception of recent local temperatures are not predicted by what those temperatures have actually been.

Rather, they are predicted by their cultural outlooks, suggesting that individuals selectively attend to or recall weather extremes in patterns that reflect their groups’ position on climate change.

Nor do individuals appear to uniformly revise their perception of climate-change risks as they experience significant extreme-weather hardships. A CCP study of residents of southeast Florida found that the number of times a person had been forced to evacuate his or her residence, had been deprived of access to drinking water, had suffered property damage, etc. as a result of extreme weather or flooding had a very modest positive impact on the perceived risk of climate change for egalitarian communitarians—the individuals most culturally predisposed to credit evidence of climate change—but none on hierarchical individualists—those most culturally predisposed to dismiss such evidence.

In other words, people don’t “believe” in climate change when they “see” it; they see it only when they already believe it.

Cultural cognition predicts this—although so does elementary logic, since individuals who experience such events can’t “see” or “feel” the cause of them. What they see extreme weather as evidence of (climate change, tolerance of gay marriage, nothing in particular, etc.) necessarily depends on their assent to some account of how the world works that they are not themselves in a position to verify. And that’s where cultural cognition comes in.

c. Motivated system 2 reasoning. The popular “thinking fast, thinking slow” account of climate-change controversy also implies that the members of the public most disposed to use reflective “system 2” reasoning can be expected to form perceptions of climate risk more in line with scientific consensus. 

Again, the evidence does not bear this claim out.   In fact, they are the ones who are the most polarized.

That’s what the cultural cognition thesis tells us to expect.  Those who possess the skills and habits of mind necessary to critically evaluate complex arguments and data have more tools at their disposal to fit their assessments of evidence to the beliefs that are predominant in their identity-defining groups.

4. A polluted science communication environment

The spectacle of intense, persistent political conflict can easily distract us from the state of public opinion on the vast run of facts addressed by decision-relevant science. The number of risk issues that divide members of the public along cultural lines is infinitesimal in relation to the number that don’t but could.  There’s no meaningful level of political contestation over the health risks of unpasteurized milk, medical x-rays, high-power transmission lines, fluoridated water, etc. On these issues, moreover, culturally diverse individuals do tend to converge on the best-available evidence as their capacity for science comprehension increases.

The reason that these issues do not provoke controversy, moreover, is not that individuals understand the scientific evidence on the relevant risks more completely than they understand the evidence on climate change or nuclear power or the HPV vaccine or gun control.

Individuals (including scientists) align themselves appropriately with a body of decision-relevant science much vaster than they could be expected to comprehend or verify for themselves. They achieve this feat by the exercise of a reliable faculty for recognizing insights that originate in the methods that science uses to discern the truth.

Their everyday interactions with others who share their cultural worldviews are the natural domain for the use of this faculty.  Individuals spend most of their time with others who share their values; they can exchange information with them readily, without the friction that might attend interactions with individuals whose fundamental outlooks on life differ fundamentally from their own; and they are more able to read those with whom they share defining commitments, and thus to distinguish those of their number who know what they are talking about from those who don’t.

All the various affinity groups within which individuals exercise their knowledge-recognition faculties are amply stocked with people high in science comprehension, and all fully equipped with high-functioning processes for transmitting what their members collectively know of what’s become collectively known through science. So while admittedly (even regrettably) insular, the ordinary interaction of ordinary individuals with those who share their cultural worldviews generally succeeds in aligning individuals’ beliefs with the best available evidence relevant to the decisions they must make in their personal and collective lives.

This process breaks down only in the rare situation when positions on particular issues become entangled in antagonistic cultural meanings, effectively transforming them into badges of membership in and loyalty to one or another competing group. At that point, the stake that ordinary individuals have in forming and persisting in beliefs consistent with others in their group will dominate the stake they have in forming beliefs that reflect what’s known to science: what she personally believes—right or wrong—about climate change, nuclear power, and other societal risks won’t have any impact on the level of risk she or anyone else faces; the formation of a belief at odds with the one that predominates in her group, however, threatens to estrange her from those on whom her welfare—material and psychic—depends.

These antagonistic cultural meanings are a form of pollution in the science communication environment.  They literally disable the ordinarily reliable faculty ordinary individuals rely on to discern what’s known by science.

Engaging information in a manner that reflects their individual interest in forming and persisting in group-convergent beliefs, diverse citizens are less likely to converge on the best available evidence relevant to the health and well-being of them all.

The factual presuppositions of policy choices having become symbols of opposing visions of the best life, debates over risk regulation become the occasion for illiberal forms of status competition between competing cultural groups.

This polluted science communication environment is toxic for liberal democracy.

 5. The science of #scicomm environment protection

The entanglement of positions on societal risk in culturally antagonistic meanings is not a consequence of immutable natural laws or historical processes.  Specific, identifiable events—ones originating in accident and misadventure as often as strategic behavior—steer putative risk sources down this toxic path. 

By empirically investigating why a putative risk source (e.g., mad cow disease or GM foods) took this route in one nation but not another, or why two comparable risk sources (the HPV vaccine and the HBV vaccine) travelled different paths in a single nation (the U.S.), the science of science communication enables us to understand the influences that transform policy-relevant facts into divisive markers of group identity.

The same methods, moreover, can be used to control such influences.  They can be used to forecast the likely development of them in time to enable actors in government and civil society alike can act to avoid their occurrence. They can also be used to formulate and test strategies for disentangling positions from antagonistic meanings where such preventive measures fail.

The vulnerability of risk regulation to cultural contestation is not a consequence of one or another groups’ hostility to science, of citizens’ “bounded rationality,” or of some inherent drive or appetite on the part of competing groups to impose a sectarian orthodoxy on society.

It is the predictable but manageable outgrowth of the same conditions of political liberty and social pluralism that make liberal democracy distinctively congenial to the advance of scientific knowledge.

By using the hallmark methods of science to protect the science communication environment, we can assure our enjoyment of the unprecedented knowledge and freedom that are the hallmarks of liberal democracy.

 

PrintView Printer Friendly Version

EmailEmail Article to Friend

Reader Comments (12)

A quick read - and this caught my eye:

Nor do individuals appear to uniformly revise their perception of climate-change risks as they experience significant extreme-weather hardships. A CCP study of residents of southeast Florida found that the number of times a person had been forced to evacuate his or her residence, had been deprived of access to drinking water, had suffered property damage, etc. as a result of extreme weather or flooding had a very modest positive impact on the perceived risk of climate change for egalitarian communitarians—the individuals most culturally predisposed to credit evidence of climate change—but none on hierarchical individualists—those most culturally predisposed to dismiss such evidence.

Fascinating. And not something I'd expect. I fully have expected that despite the existence of cultural cognition, and it's importance to how people formulate views about climate change, short-term weather phenomena would have a significant impact on public views related to the risks posed by ACO2-caused climate change.

This is a problem because I may have to make a major adjustment in my views. Or, I could just find some way to reconcile that evidence in ways that fit with my predispositions and biases. :-)

Dan - isn't there some evidence that the public's views on this risks posed by climate change are, at least to some degree, associated with short-term weather phenomena and the economy (even if that evidence doesn't prove causality)? If so, is your one finding sufficient to dispel a hypothesis that there would be some causality there? Are you saying that if there were a spell of completely unprecedented extreme weather over a short period of time, it wouldn't move the needle at on on the public's views on the risks posed by climate change?

February 20, 2014 | Unregistered CommenterJoshua

@Joshua:

There have been several studies now showing that the effect of "recent" & "extreme" weather on perceptions of climate change risk interact with cultural predispositions & the like. I think the best is still Goebbert, K., Jenkins-Smith, H.C., Klockow, K., Nowlin, M.C. & Silva, C.L. Weather, climate and worldviews: The sources and consequences of public perceptions of changes in local weather patterns. Weather, Climate, and Society (2012)..

SE Florida is interesting. It's plenty polarized on climate change but clearly of 1 mind in viewing protection of the region from climate impacts as a public good demanding aggressive govt action. Really, to ask them how they feel about "climate chnage" is to measuring their support for using the best available science to protect the region is like asking someone whether he or she "believes in evolution" is to measuring their science literacy.

February 20, 2014 | Registered CommenterDan Kahan

Dan - that link didn't work for me but I assume you were linking to this paper?

http://journals.ametsoc.org/doi/abs/10.1175/WCAS-D-11-00044.1

Only indirectly does it inform about how perceptions of recent weather affect perceptions about climate change.

I will offer a prediction: I think this winter's weather will show a "signal" in assessments of perceptions about "global warming" in the U.S. - with the signal being a drop in belief that climate change poses a risk. That is, assuming that any substantial polling will be done across a time-frame where it is likely to capture and isolate the impact over such a relatively short period of time. If this summer will be an unusually hot summer, the impact from this winter will likely disappear, IMO. (I'm still in confirmation bias mode, and will hold out for more evidence before I shift my views on the impact of short-term weather phenomena on views related to climate change).

February 20, 2014 | Unregistered CommenterJoshua

The first of the "Southeast Florida" graphs is interesting. I don't think you showed that at your Nottingham talk. It seems to show that, as one might expect, the cultural cognition effect is stronger among communitarians than individualists. More here.

February 21, 2014 | Unregistered CommenterPaul Matthews

@Joshua:

Sorry about messed up link on Goebert et al.

Why do you think evidence is supplies is indirect? On the temperatures, the finding in Goebert et al. was that recent temp -> perceptiont temp -> perception GW risk didn't fit data, whereas cultural worldview -> perception recent temp -> perception GW risk did.

Another recent study finding similar effect -- but using confirmation bias (prior belief in AGW) as mechanism: Howe, P.D. & Leiserowitz, A. Who remembers a hot summer or a cold winter? The asymmetric effect of beliefs about global warming on perceptions of local climate conditions in the u.S. Global Environmental Change 23, 1488-1500 (2013).

I think your prediction could be consistent with findings like these. Extreme weather might bump up perception of those predisposed to believe enough to generate net effect even it it doesn't affect the views of the predisposed not to believe.

It goes w/o saying, of course, that those whose perceptions are related to personal experience are not reasoning very well; no one is in a position to tell from personal observation whether global climate is changing, much less whether it is as a result of AGW. Those who think recent temperature changes "bear out" AGW probably are unaware that global tempmeratures didn't increase in last decade.

The only way to promote "belief" is to give people the sort of evidence that they usually rely on to figure out what's known by science -- not personal experience, but meaningful cues consisting mainly in examples of those whom they treat as knowing-what's-what are acting on belief that AGW is happening. AT that point, they'll "see" confirming evidence all around.

Meanings 1st, facts 2d. Communicators who ignore this are going to concuss themselves as they continue to bang their heads against the wall

February 21, 2014 | Unregistered Commenterdmk38

@Paul:

You are right: I didn't report on any of the SE Fla data in my Nottingham lecture. I almost did -- conversations w/ Warren Pearce & Reiner Grundmann before the lecture tempted me to add the material, but I was committed & eager not to have my lecture go too much longer than it was supposed to!

But I decided to include for Cardiff. I don't like to give same talk twice -- I try to change frame/ theme, even when evidence remains the same basic body of work!

Thanks again for your contribution to making my trip to NOttingham happen; I really enjoyed it!

I assume you don't view the flooding in UK as a reason to revise your assessment of what the best evidence signifies about climate change & human contribution to it?

February 21, 2014 | Unregistered Commenterdmk38

Dan -

You say: "perception GW risk didn't fit data"

But it seems to me that the relevance to views on GW are indirect because from the abstract the study didn't seem to actually assess views of climate change" per se, but how views of recent weather patterns matched actual weather patterns. The conclusions w/r/t how views on climate change might be affected can only be inferred. Is there something I'm misunderstanding?

"Extreme weather might bump up perception of those predisposed to believe enough to generate net effect even it it doesn't affect the views of the predisposed not to believe."

I would think that the recent cold weather might alter the views of people who are relatively non-aligned, temporarily. Similarly, a spate of very unusually hot weather might have the opposite effect. So I don't think that 100% of any signal will be attributable to those whose views are confirmed by the particular type of short-term weather.

BTW - What do you suppose is the % of people who aren't particularly locked into one orientation or the other?

Suppose there were a spate of very unusually warm weather over the next couple of summers - unaccompanied by other short-term phenomena that might be a counter-balance. I mean events that could not be easily dismissed as just examples of natural variability. I think that such a development would have to affect "skeptics" in some way. I still think that they would have to adjust or compensate - for example, they could switch into an identity-protection mode that led them to believe that they hadn't currently dis-believed that human-induced climate change was occurring, and instead move into identity-protective reasoning that would allow them to move further into identity-aggressive reasoning about policy options. Of course, the "consensus" science says that such a development would be unlikely - so we aren't likely to find out - but the consensus science also says it could be a possibility.


"It goes w/o saying, of course, that those whose perceptions are related to personal experience are not reasoning very well; no one is in a position to tell from personal observation whether global climate is changing, much less whether it is as a result of AGW. "

What's funny about that is that even though I agree, the unusual amount of cold and snow this year (exacerbated by having moved into a colder and snowier region) have had me muttering to myself on occasion "So much for global warming."

"Those who think recent temperature changes "bear out" AGW probably are unaware that global tempmeratures didn't increase in last decade."

As a technical matter, I think that statement is not true. It isn't that global temperatures did not increase, it is that they did not increase at a level of statistical significance.

Anyway, there are quite a few folks out there who are intimately familiar with the temperature records but who think that recent weather events and/or climate developments (if not "temperature changes") are likely to have had some influence from ACO2. I don't think that it is merely a matter of them not reasoning well.

"The only way to promote "belief" is to give people the sort of evidence that they usually rely on to figure out what's known by science -- not personal experience, but meaningful cues consisting mainly in examples of those whom they treat as knowing-what's-what are acting on belief that AGW is happening. AT that point, they'll "see" confirming evidence all around."

Hmmm. I have a hard time believing this. IMO, there will only be an overwhelming prevalence of belief on climate change when either what people experience is very severe or there have been many decades of evidence inconsistent with the theory in all respects (ice melting, sea level rise, surface temperatures, and ocean temperatures). Personal experience and meaningful cues are not mutually exclusive from one another.

"Meanings 1st, facts 2d. Communicators who ignore this are going to concuss themselves as they continue to bang their heads against the wall"

Here I agree - but I guess I have a different understanding of what can potentially comprise "meanings."

February 21, 2014 | Unregistered CommenterJoshua

"As a technical matter, I think that statement is not true. It isn't that global temperatures did not increase, it is that they did not increase at a level of statistical significance."

A technically tricky area to discuss, but much of the sceptical belief is based around the following result:
http://woodfortrees.org/plot/rss/from:1996.6/plot/rss/from:1996.6/trend

Statistical significance depends on what noise model you use. If you use something like AR(1) as the IPCC did, or ARMA(1,1) as Rahmstorf did, you can push it only a little further. If you use ARIMA(3,1,0) as Doug Keenan did, there's no evidence for any significant warming at all. It could *all* be natural background.

This is another case of people seeing what they expect to see. You get back a conclusion that matches the prior assumptions you put in. Personally, I don't think it constitutes evidence on AGW, either way (although it is a serious problem for the climate models), and I disagree as much with sceptics who say it's a disproof of AGW as with believers who say the plateau hasn't happened at all.


"I assume you don't view the flooding in UK as a reason to revise your assessment of what the best evidence signifies about climate change & human contribution to it?"

Discussion in my neighbourhood revolves around planners who have allowed builders to build houses on flood plains, and the Environment Agency, who had recently set a policy to allow those areas to flood, to promote biodiversity. Some sort of EU green initiative, apparently, that the EA upper echelons enthusiastically went along with. They stopped dredging the rivers, maintaining the infrastructure, and super-regulated the disposal of silt dredged from ditches by local farmers.

A few of the usual suspects have been trying to sell it as 'climate change', but nobody's buying it. And quite a few climate scientists have come out against them saying there's no connection. When people are up to their knees in dirty water, they're not very sympathetic to green attempts to use it to support political agendas, so more of the climate change activists have been keeping their heads down than usual. It's probably only temporary, though.

Just my anecdotal view.

February 21, 2014 | Unregistered CommenterNiV

@NiV:

I thought you had died. Is @Larry w/ you?

February 22, 2014 | Registered CommenterDan Kahan

NiV -

Personally, I don't think it constitutes evidence on AGW,

Well that's good to read from a "skeptic." Unfortunately, it is also something quite rare to read from a "skeptic, in my experience.

and I disagree as much with sceptics who say it's a disproof of AGW ..

Indeed, given that it is only one metric (as discussed in Balmaseda et. al.) and given that for all we know it is a short-term (and temporary) decrease in a longer-term trend

...as with believers who say the plateau hasn't happened at all.

Well, there are some interesting related arguments there, it seems to me (e.g. Cowtan and Way), although I am ill-equipped to evaluate them scientifically. But sure, if someone says that there hasn't been any plateau in the trend reflected within that limited dataset, then there wouldn't seem to be much to agree with, IMO.

So what do you think, NiV - does no statistically significant warming = "no warming at all?" I mean w/o constraining the discussion with specifying CIs - or would you say that the entire discussion is meaningless if you haven't specified CIs? As someone who doesn't understand statistics, I find it to be an interesting discussion.

February 22, 2014 | Unregistered CommenterJoshua

Dan,

"I thought you had died. Is @Larry w/ you?"

My apologies for absenting myself for so long!

Initially I had other commitments and didn't have much time for internet commenting. When I came back to it my interests were mainly drawn elsewhere. And no, I don't know where Larry is!

Joshua,

"So what do you think, NiV - does no statistically significant warming = "no warming at all?""

The problem is that it depends what you mean by "warming" - different people are using it in different senses of the word, and part of the confusion is over not realising that the meanings are distinct.

There are three basic senses in which you might say "global warming". The first is the literal meaning: that the temperature has increased. You can test that very easily, by simply subtracting the temperature at the start of the time interval from that at the end. Or if you want to ignore the high frequency variation in the weather, subtract the average temperature for an interval at the start from an average temperature for an interval at the end. It's very simple and unambiguous, and there isn't much argument about the sign any more, although I think people would still argue about the size of the error bars.

However, that doesn't get us very far. So it's warmed over the past century. Maybe it'll cool over the next century, or warm some more, or stay the same. It's not warmed over the past 17 years, but that doesn't tell us anything about the previous 17 or the next 17.. The two data points don't tell us very much about what's going on. This is where we get our second meaning for "global warming", which now doesn't refer to the actual measured temperature, but to one particular component of a summation of deterministic and random influences, that add up to give the temperature. We build a mathematical model of signal plus noise. The signal is a rising trend. The noise is some sort of weather-related wobble around it. And then there's measurement error on top of that. When people talk about observing "global warming", they mean attempts to estimate that trend component from the signal plus noise of the data. When people say "the warming is still continuing", this is what they're usually referring to.

And this is where the statistical arguments come in, because we're not talking about actual temperatures, but about a parameter of a particular mathematical model of the behaviour, and there's no agreement about what the correct statistical model to use is.

There are many people with set ideas about how things work for which the model is so 'obvious' that they don't even understand that it's subject to question, or that there is a distinction between 'warming' of the observations and of the model parameter.

They were often taught in school how to draw a trend line through data - methods implicitly based on a particular model of straight line plus independent identically-distributed Gaussian noise - and the trend calculated takes on an objective significance for them. They regard it as an observation of the trend, something they can see is 'there', and at the same time an observation (subject to measurement error, of course) of that underlying deterministic component that is "global warming".

(And of course the sceptics are doing the same when they draw a trend line through the last 17 years and say 'look, it's stopped'. There are a lot of ways for an ongoing rising component to be temporarily cancelled by noise. It's stopped in the first sense, but whether it has in the second sense depends on our model assumptions.)

However, the real world has a lot of ways to mess up this nice, school textbook picture. The variations around the line don't have to be independent or identically distributed. The distribution can change, broaden, narrow, shift, skew over time. It can behave differently over different timescales, so it might not vary much day-to-day but vary hugely month-to-month, or decade-to-decade. The noise might not average out over the timescales under consideration - there is a rule of thumb in climate studies that you need 30 years of data to assess climate, but that '30' isn't based on any calculation or measurement. It might take a hundred years, or a thousand, before the noise 'averages out'. We don't know. They probably only picked 30 because when they started doing climatology seriously they had about 30 years of good data to work with.

You can even argue with whether it is really 'noise'. It's not measurement noise, it's actual weather, with actual increases or decreases in temperature. Again, it depends whether the global warming 'signal' you're looking for is about the change in actual temperatures, or the change in the underlying contribution to temperature.

From a statistical point of view, you can't distinguish signal from noise unless you have reasonably accurate models of the statistics of both. And you can't determine the model from the data - while there are many models you can rule out, and while you can do things like find the closest fit within a particular simple class of models, there are always infinitely many alternatives to them that are utterly indistinguishable.

So a statistician always requires 'exogenous information' to select a model. What we need is a validated statistical model of the signal and noise based on the physics. The AOGCMs would seem ideal for this role, except that they don't reproduce the observed distributions and are not validated. But in principle, the right way to do this is to obtain the statistics of 'background weather' from running climate models with no extra CO2, then obtain the distribution with extra CO2, and see which distribution the observations are a better fit to, and by how much. The log-likelihood ratio Log[P(Obs|H1)/P(Obs|H2)] measures the information we acquire in support of H1 over H2 from the observations.

However, as I said, we don't have a validated model of the signal/noise statistics, and so far as I can tell the distributions with and without the change overlap considerably, and are not predicted to be clearly distinguishable until much later in the century. It'll be obvious who's right in 2080, but we probably still won't know in 2020. It's early days yet for the science, but the more empirical estimates of sensitivity seem to be shifting downwards, and the ongoing plateau is not looking good for the high-sensitivity hypotheses. However, I'm still playing it cautious, and my answer is still "We don't know".

It makes sense that there should be some, but we can't really say that we've observed (attributed) it yet. We don't even know what the true confidence bounds are, so we can't say if we've exceeded them. For exactly the same reason, we can't say we've disproved it either. If the noise model should be much noisier than we think - which is what sceptics are supposedly arguing - then the pause could just as easily be the noise.

There is a third meaning to the phrase 'global warming', which is that it refers to the whole 'belief system'. The science, the politics, the campaigning, and the alarm. When people are asked 'do you believe in global warming?' they often read it in the same way as 'do you believe in gun control?' And that leads to even more confusion, especially from social scientists who can't understand why people answer 'do you believe in global warming?' and 'do you believe the world has warmed?' differently. (And of course the question 'do you believe the world is warming?' causes further confusion by not specifying timescale. Do you mean over the last 10 years, or the last 100? Or the last 10,000? What if I think it was warming but now has stopped? Or what if I think the temperature rose but that this was mostly 'noise'? What do you mean?!)

A lot of the argument is over simply not understanding what people mean by the terms they use.

February 23, 2014 | Unregistered CommenterNiV

Dan, glad you enjoyed your trip to Nottingham; as noted on my blog I enjoyed your talk and it helped to make some things clearer.

The floods are just another thing that increases polarisation, via motivated reasoning and confirmation bias. The believers say it's yet more evidence of climate change, as is any weather event, of any sort, anywhere in the world. The sceptics plot graphs showing no increasing rainfall, and point out that the IPCC says there is low confidence in any changes, even in the sign!

February 24, 2014 | Unregistered CommenterPaul Matthews

PostPost a New Comment

Enter your information below to add a new comment.

My response is on my own website »
Author Email (optional):
Author URL (optional):
Post:
 
Some HTML allowed: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <code> <em> <i> <strike> <strong>