Climate Science Glossary

Term Lookup

Enter a term in the search box to find its definition.


Use the controls in the far right panel to increase or decrease the number of terms automatically displayed (or to completely turn that feature off).

Term Lookup


All IPCC definitions taken from Climate Change 2007: The Physical Science Basis. Working Group I Contribution to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Annex I, Glossary, pp. 941-954. Cambridge University Press.

Home Arguments Software Resources Comments The Consensus Project Translations About Donate

Twitter Facebook YouTube Pinterest

RSS Posts RSS Comments Email Subscribe

Climate's changed before
It's the sun
It's not bad
There is no consensus
It's cooling
Models are unreliable
Temp record is unreliable
Animals and plants can adapt
It hasn't warmed since 1998
Antarctica is gaining ice
View All Arguments...

Keep me logged in
New? Register here
Forgot your password?

Latest Posts


Climate Hustle

AGU Fall Meeting sessions on social media, misinformation and uncertainty

Posted on 16 July 2012 by John Cook, Stephan Lewandowsky

We have proposed several sessions for the AGU Fall Meeting in San Francisco on 3-7 December: on social media, misinformation and uncertainty. AGU members are invited to submit abstracts for the sessions - the deadline to submit an abstract is August 8. Details of the sessions are:

PA013: Social Media and Blogging as a Communication Tool for Scientists

Session Abstract:
Social media and the Internet has become an increasingly indispensable tool for scientists and communicators. This session will feature key figures in the climate blogosphere who have adopted novel and effective methods of communicating climate change science on the Internet. They will discuss the risks and rewards of new media, covering issues such as the challenges and advantages of crowd sourcing, viral marketing, Internet marketing and traffic generation, the use of smartphones, the management of online communities and social media sites such as Twitter and Facebook.

Convenor: John Cook

Invited Speakers:

  • Michael Mann
  • Michael Tobis
  • Peter Sinclair
  • Zeke Hausfather

ED042: Understanding & Responding to Misinformation

Session Abstract:
Scientists face many challenges in effectively communicating science to the public, not the least being the presence of misinformation. However, there are actually positive educational opportunities available in the correction of misinformation. This session will explain the psychology and origins of misinformation, the cognitive processes at play when correcting misconceptions and recommended approaches to effective myth debunking. The session will be relevant to scientists seeking to explain their science, to communicators wishing to effectively outreach to the public and educators who may need to respond to misinformation in the classroom.

Convenors: Stephan Lewandowsky, John Cook

Invited Speakers:

  • Daniel Bedford
  • Max Boykoff

GC014: Construing Uncertainty in Climate Science

Session Abstract:
Uncertainty forms an integral part of climate science, and it is often cited in connection with political arguments against mitigative action. However, the implications of uncertainty are not always well understood. In particular, uncertainty is often misunderstood to imply that the risk from climate change may be minimal, whereas in fact greater uncertainty translates into greater risk. This session will examine how uncertainty can be misconstrued, how such misconstrual can be avoided, what the implications of uncertainty are for risk management, and why the notion of uncertainty plays such an important role in cognition and decision making as it relates to climate change.

Convenors: John Robert Hunter, Stephan Lewandowsky, James Risbey

Invited Speakers:

  • Naomi Oreskes
  • Gerard Roe
  • Paul Baer
  • Mike Raupach

AGU Members, click here to submit an abstract for either session.

0 0

Bookmark and Share Printable Version  |  Link to this page


Prev  1  2  3  4  Next

Comments 101 to 150 out of 155:

  1. Re the Lewandowsky research and paper and all the discussion and analyses of it on this thread and elsewhere in the blogosphere.

    "Much ado about nothing!"

    Let's get back to discussing and analyzing stuff that really matters.
    0 0
  2. John Cook has written a good article relevant to posts in this thread. It's just been published today on The Conversation (click here).

    Maybe he'll post it on SkepticalScience as well.
    0 0
  3. Tom Curtis #95
    Of course there's a good reason for randomizing the order of questions. No doubt Kwiksurveys offers that feature. That's not what happened. Climate Audit was offered a different survey with different questions and a different ID. So Lewandowsky can claim he was fully justified in announcing the results of his n=1100 survey at Monash, because the survey Climate Audit was invited to participate in was a different one. Or he can continue to claim that 5 sceptic blogs were invited to participate in this survey, (-Snip-), since his n=1100 survey was likely to be swamped by a totally different type of respondent at any moment. He can't logically do both.

    You say "the reason for using bins is that no single question clearly demarks those who accept AGW from those who reject it". True enough,and it's a serious criticism of the survey that they couldn't even devise a question to do that. The four questions are logically linked, so someone who strongly agrees with one and strongly disagrees with another is not undecided, but more likely seriously confused. No wonder they believe in conspiracies.
    I can see the use of a battery of questions with the general public, because a single question which adequately distinguishes the warmist from the denier is likely to be too complicated for the average punter to understand. ( -Snip-). But this question is aimed at climate obsessives who weigh the meaning of every word (well, I know I do). A perfect opportunity to devise THE climate question wasted.
    0 0
    Moderator Response: [DB] Accusations of dishonesty and sloganeering snipped.
  4. Tom,

    I have not seen any references in your posts to justify your analysis. The Lewandowsky paper provides copius references to support their analysis. Please provide a reason I should listen to an amateur analysis without any peer reviewed input? I am dissapointed in your posts here and I think it affects your reputation.

    For example, you are completely wrong about the two "outliers". You have provided no justification for deleting these responses further than "I doubt anyone would answer this". You must provide an objective reason to discard any results. Where will the line be drawn between acceptable and "gamed"? "I doubt it" is completely unacceptable. You have suggested nothing so I cannot even criticize your proposal. I have participated in scientific surveys and you need a reason to delete samples. I regularly talk to conspiracy theorists at my job teaching High School in the USA and I am not surprised by those two results.

    There is a lot of literature on this subject, you have provided no links to support your claim that these points are outliers. Perhaps you could start with Landowsky's references. Your other claims are similarly unsupported.
    0 0
  5. Further to my 90 and 100, here are the updated data for the distributions; the first being in absolute numbers, the second in percentages. In my opinion, the change in data requires no revision to my commentary @90:

    0 0
  6. Foxgoose wrote: "A great many people are sceptical, however, of the politically led and inspired "science" which grew out of the UN inspired IPCC."

    Mmmmm... I might be skeptical of that too. If it actually existed.

    Let's imagine for a moment that the IPCC (heck, the UN) never existed. How would this change the scientific research on global warming?

    The answer is, of course, that it would have no impact whatsoever... because the IPCC does not conduct research. Rather, it collects existing research into a report for policy makers around the world. In short, it is an effort to transmit the scientific consensus to the political establishment. We could debate the merits of that, but the fact is that it has absolutely no impact on the science itself... and the belief to the contrary amongst 'skeptics' is just another demonstrably false conspiracy theory.

    The idea that the IPCC is conducting biased research fails utterly in the face of the fact that they are not conducting any research... yet it is a cornerstone of 'skeptic' beliefs. Again, Lewandowsky's study is redundant. The sky is blue. Earth revolves around the Sun. Many online GW 'skeptics' are conspiracy theorists. Nothing new here.
    0 0
  7. Micheal Sweet @104: May I tentatively suggest that you appear to me to be trying to create excuses to avoid engaging with the evidence?

    Tom is not writing a peer-reviewed paper, he is analysing the data and presenting his results on the fly. I've done the same on SkS comment threads, and so have many others. Furthermore he seems to be the only one on this thread who is analysing the data.

    May I further, and again tentatively, suggest that the sociology of what is occurring on this thread is very interesting in ways which go beyond what has been noted up till now. Yes, skeptics have been throwing up conspiracy theories as if in a bizarre attempt to confirm Lewandowsky's research. But AGW proponents have been responding in a similarly bizarre and self-discrediting manner.

    If I may use the tribal metaphor (hopefully this will be permitted since I am applying it to my own tribe), when one of us applies our skepticism in a way which undermines rather than reinforces tribal values, the reaction of the tribe is to scapegoat that person. That undermines our supposed credentials for supporting an evidence-based approach.

    I'm only a dabbler in social anthropology, but if you want a reference I believe this book by Rene Girard (510 citations) is a seminal work in the field.
    0 0
  8. michael sweet @104, the evidence that the two responses with the strongest support of conspiracy theories are outliers cannot be found in any academic book or paper, but only in Lewandowsky's data. Analysing that data shows them to be 6.49 and 6.28 standard deviations greater than the mean value for responses to the conspiracy theory questions (excluding that regarding climate change). If this were a normal distribution, that would mean they have probabilities of occurring in a single trial of 1 in a billion, and 1 in ten billion respectively. This is not a normal distribution, and there were significantly more than 1 trial, so the probability of their occurring is significantly greater than that. Never-the-less, the occurrence of two such extreme results is improbable - hence the assumption that they are just a statistical aberration has little force.

    This is particularly the case given that both of these statistical aberrations are also aberrations (in this survey) in having very strong disagreement with climate science; and given that they come from the smallest bin and a bin with a narrow distribution. (See my post at 105). If they had been statistical aberration is it approximately 8 times more likely that they would have occurred in the acceptor bin, and more than 2.5 times more likely that they would have occured in the undecided bin. That means that even if two such improbably events should be expected, the probability that both would be in the rejectors bin is less than 0.76% (It is, of course, far less probable than that they should actually occur, although I am unable to determine a reasonable estimate of that probability.) I do not need a reference to know that it is far more likely that the survey has been gamed than that so improbable an event has occurred.

    This is especially the case in that anonymous internet surveys are known to be plagued with attempts to game them, even when they deal with uncontroversial topics. On Lewandowsky's own data, he has excluded 16.8% of responses as having been probably gamed. If this percentage was carried through the rest of the results, then any two results taken at random have a 2.8% chance of being gamed. It should not be supposed that the responses that survived screening are as likely to have been gamed as those that did not - but nor can it be assumed that none of those responses that did survive screening where not gamed. Obviously the more unusual the data in a response, the more likely it is to have come from gaming the survey. Therefore, on statistical information alone it is highly probably that these two responses are scammed.

    It has been suggested that people adhereing to nearly all conspiracy theories are not uncommon. The two academic papers I have read on the subject, however, never report more than three conspiracy theories held by one person (including Goertzel 94, cited by Lewandowsky). I am aware of just one person who accepts near universal conspiracy theories (cited by Goertzel on a web site), but he did not accept all theories, and would only avow that there was "a kernel of truth" in most conspiracy theories. On that basis, for most conspiracy theories he would answer 3 (agree) rather than 4 (strongly agree). In that way he is like the two other outliers in the survey, of whom I am not suggesting that they are scammed. Here are the two suspect responses along with the next two most extreme responses so you can compare them, and note just how extreme the suspect responses are:

    Very little of this information is new information; and I have argued these points before on this thread in detail. Paraphrasing such detailed analysis as merely saying "I doubt it" is gross misrepresentation. If you have a serious rebutal to make, make it. I would like little better than to be shown that I am wrong. Simply suggesting that I am wrong because you are unwilling to actually analyse the data, however, is hardly convincing.

    Finally, you say that I am damaging my reputation. My reputation at SkS has been built on reasoning in exactly this style, but with "skeptical" arguments and comments as the target. The only difference now is that my target is somebody closely associated with the defense of climate science. It appears, then, that my reputation with you has been built not on my analysis, but on my agreement with your opinion. [inflamatory snipped]

    Perhaps you would like to reconsider that comment; [inflamatory snipped]
    0 0
    Moderator Response: [Dikran Marsupial] [Mildly] Inflamatory snipped. Discussion of this topic is becoming a little too heated. Please can we all keep the discussion as impersonal as possible, and based purely on reason [for which Tom rightly has an excellent reputation IMHO].
  9. Tom,
    I see someone who is not an expert in a subject claiming a peer reviewed paper is wrong based on their back of the envelope calculation. We see this all the time at SkS. We normally rely on peer reviewed data to support arguments. You normally cite peer reviewed data to support your analysis. Here you cited your wife as agreeing with you. In this case you have cited no papers. You have provided no data to support your claim that the distribution should be a normal distribution. The sample size is small so it is difficult to calculate the normal curve, let alone count extremes. There are abundant papers counting conspiracy beliefs, why don't you cite one of them to support your claim that these outliers should be excluded?

    You have not drawn a line where you would include a response and when you would exclude a response. In my experience a line like that is impossible to agree on. In the end all the responses are counted. If you think the responses are gamed you discount the conclusions. Lewandowsky discusses this in the paper.

    If you can cite a paper that says results that appear gamed to the researcher should be excluded from the analysis I will consider your data. My understanding is that all responses are normally included.

    AIUI Lewandowsky's conclusion is that deniers often believe in conspiracy theories. What is so contraversial about that? You don't like his title.
    0 0
    Moderator Response: [Dikran Marsupial] The first paragraph of this comment is verging on ad-hominem territory. The correctness of an argument does not depend the expertise of its source. The absence of a suitable reference does not mean that the argument is incorrect, just that you have to judge it on its intrinsic strength and Tom has been completely candid about that. Please keep the discussion more tightly focussed on the argument and not the individual making it.
  10. Did you actually read Tom's response @108? May I suggest you do so again?
    0 0
  11. Leaving aside for a second the probability of one swimming into Lewandowsky's net, chimeras apparently do exist, presuming various anecdotes such as that related in the link are not all lies.

    To me the most interesting part of Tom's analysis lies in the case he constructs for the improbability of catching a sample of the relatively few very most strange fish swimming in the ocean. On the one hand, Tom shows mathematically how bizarre these creatures are, yet it seems it's not impossible to find examples without an absurd amount of effort. It's nonetheless a leap to say that a properly constructed survey instrument will bump into them as easily as a person steering a browser into particular places.

    Many confounding factors.
    0 0
  12. Reply to Tom Curtis at 07:34 AM on 6 September, 2012


    You're mistaken I'm afraid.

    I think you missed the point that Geoff Chambers and I made in our earlier posts.

    We pointed out that Steve McIntyre has now specifically confirmed that he received the "reminder" email on September 23rd.

    Which rather invalidates the rest of your post.
    0 0
  13. Reply to #91 Bob Loblaw at 12:00 PM on 6 September, 2012

    To put your mind at rest, I did "know exactly what I was doing" when I said Steve McIntyre received the reminder email on the 23rd.

    I was stating the truth.

    You may, of course, draw whatever conclusions you like from that

    Be careful though - maybe telling the truth is part of a cunning conspiracy ;-)
    0 0
  14. Foxgoose:

    Yes, but focusing on the date of the reminder, instead of the date of the earlier original invitation, means you weren't telling "the whole truth". After you get that figured out, perhaps we'll move on to "nothing but the truth".
    0 0
  15. Repy to #114 Bob Loblaw at 07:02 AM on 7 September, 2012

    No - the key date was the final request.

    0 0
    Moderator Response: [DB] Inflammatory snipped.
  16. @foxgoose - is it time to give this a rest?

    First off, as lots of people have pointed out to you it was one slide out of many in a presentation. It was preliminary indication, not a final paper. It was for fellow scientists, not the general public.

    Secondly, let's say McIntyre did finally decide to post the survey - a few days later (unlikely) and another dozen or two people responded. Say this included one full blown conspiracy theorist (probably not even that), half a dozen extreme right wingers, a dozen or so luke warmers of various political leanings and half a dozen who accept climate science, again with a spread of various political leanings. How much difference would it make to the survey results? Very little. At best, they might have grabbed a few more extreme right wingers to help the analysis - which is probably why they tried again to see if anyone from sites that are likely to attract a greater proportion of extreme right wingers would respond.

    Thirdly, on a blog if something isn't done quickly it probably won't get done at all.

    Fourthly, on most blogs, articles quickly drop down the list (okay, not on CA so much but I think it was a bit more active back then). If Steve or anyone else had decided to post the link then responses would probably come within a day or two and drop off rapidly.

    Fifth - McIntyre didn't post the link.

    Sixth - on the internet, downloads are almost instantaneous. The slide could have been prepared the morning of the presentation.

    Seventh - no, it was not the final results - and nor was it a car sale.

    Eighth - it was a single slide on the survey, not a whole presentation about the survey. And it was not the final result.

    9 and 10 - it was not the final result. The paper is being published two years later. McIntyre didn't post the link.
    0 0
  17. Tom,
    I am sorry, I did not intend to come across so strongly.

    It is my understanding from working on a single public survey that no samples are deleted unless they self contradicted. The deviation from the mean was not considered.

    I respect that you put in a lot of time to research your posts and cite regularly from the literature.
    0 0
  18. Just in case we forget from one minute to the next:

    Fifth - McIntyre didn't post the link.

    And again, lest we forget and as the wag elsewhere suggested, too busy with the "other matter" of rummaging in other people's email to read his own email.

    Obviously it's Lewandowsky's fault.

    Some features of this affair are not worth exploring, for anybody.
    0 0
  19. Foxgoose @113, I was incorrect in assuming you were talking about the relevant first invitation rather than the irrelevant reminder. Beyond that it still remains the case that McIntyre had plenty of time to put up the survey before the 23rd of Sept; and that you have no evidence that the survey was actually terminated on or before the 23rd of Sept. Indeed, the sending of a reminder notice strongly suggests that it was not terminated at that point.

    I wish to confine myself to rational criticisms of Lewnandowky, in press. Not wild goose chases.
    0 0
  20. Foxgoose@115

    If you've offered a car for sale, and posted a reminder, and I have never responded at all to your offer or indicated a desire to buy, then you are under no obligation to me whatsoever. You are free to do whatever you want with the car. Do you really think that you'd feel you should have sold me the car when I turn up months later with the reminder, saying that I was cheated?

    McIntyre was in his rights to ignore the invitation to participate. He has no right to complain (nor do you) that Lewandowsky proceeded with his analysis using the responses he did get without waiting for a reply from McIntyre that was unlikely to ever arrive.
    0 0
  21. This is turning into the kind of buffoonery that is normally the exclusive territory of Monckton.

    Let's see: all the temperature records show consistent increases, surface and satellite, the radiative properties of CO2 have been known for fr#%^ing ever, the Arctic sea ice is on a death spiral, extreme weather events are showing up exactly like you'd expect and Foxgoose is arguing about dates. Not even about the meager substance of that marginally interesting Lewandosky item, just some dates. Really Foxgoose, that's all you've got? Seriously? That's where you're taking the argument? And you don't see anything wrong with that?

    The only thing I take from it all is that there is absolutely nothing of interest on McIntyre's blog; even when something almost interesting comes his way, he doesn't post the link. What was he afraid of, I wonder. And I don't buy that he missed it. It's not like he's busy doing real science, teaching etc.

    This, among other kinds of nonsense, is why I knew early on what kind of "debate" we're looking at here. Complete intellectual bankruptcy against reality. Who wins?
    0 0
  22. Michael Sweet @117, that's OK. I am a little prickly at the moment as I am under a lot of pressure, so it is quite possible I have over reacted.

    With regard to your experience in a public survey, just from Lewandowsky's paper we know that he omitted 71 responses because their IP was the same of some other response. That seems a reasonable precaution against the survey being gamed, but some of those responses may have been from other family members, or from work mates. The probability that they are gamed responses is sufficient to omit them, even though it is no guarantee. We also know that another 161 were omitted because either the participants age was considered implausible (95); the provided inconsistent responses to the consensus questions; or because the responses where incomplete. The first two are, again reasonable precautions against the survey being gamed although there is, again, a small probability of their excluding legitimate responses. At least one survey response had consensus responses of 100/100/0.9. It is entirely possible that the 0.9 was intended as a 90%. A similar slip up may account for some of the improbable ages and/or inconsistent consensus responses.

    The deletions due to incomplete answering are more complicated. Had I taken the survey (I did not), there was one conspiracy question that I would not have answered because I simply do not know, and do not know enough to even assess reasonable probabilities. Because of that, my response would have been excluded. However, I certainly fit the profile of people that the survey was intended to sample, ie, "... individuals who choose to get involved in the ongoing debate about one scienti c topic, climate change." If similarly principled people were deleted for not answering one or tow questions, the sample has been biased towards the more dogmatic or manipulable members of the target population. (Please note, that is a bias of the population, and does not in anyway warrant inferences about any individuals who completed the survey and answered every question.)

    However, despite that potential bias, the number omitted due to incompleteness may have been small; the unanswered questions may have been relating to age or satisfaction in life, in which case the omission would not bias the sample; or the omitted responses may have been massively incomplete, ie, all but a few questions not answered, in which case their inclusion would not have aided analysis in any way. So, there is a possibility of a problem here, but no certitude, in which case Lewndowsky should be given the benefit of the doubt.

    That is a side note to the main issue, however. Clearly Lewandowsky was prepared to omit responses given a reasonable suspicion that they were gamed. He was correct to do so. But it follows that he has no grounds in principle to not omit these two suspect responses, given equal reasonable suspicion. Indeed, it is certainly not an unusual practice in science to omit extreme outliers, which, given the sample size, samples more than 6 deviations from the mean in a unimodal distribution certainly are. Of course cutoffs in such pruning must be, to a certain extent arbitrary. That does not mean they are inappropriate.

    However, you are mistaken if you think I am calling for Lewandowsky to just exclude the data. I think at a minimum he should acknowledge the outliers and state his reason for including them, or excluding them. At a minimum, he should also explicitly acknowledge that the inclusion or exclusion of the outliers significantly affects the conclusions of the paper with regard to conspiracy theory ideation. It would be better if he also calculated the results of the paper both including, and excluding outliers. However, replication is not auditing, so that is not compulsory.

    Of course, it is disputed that those outliers do significantly affect the result. It should be noted, however, excluding the suspect outliers, that if just 4% of responses to individual questions by AGW rejectors had been one point lower, AGW rejectors would have had the same measured propensity to accept conspiracy theories as AGW acceptors. With a sample size of 96, that is a very weak result. I doubt it is statistically significant (although I cannot do the maths to work that out).
    0 0
    Moderator Response: TC: minimally edited for accuracy 10:06am
  23. doug bostrom @111, while gladly acknowledging that some snark's are boojums, your boojum has just four acknowledged conspiracy theories attached. One of those was excluded for methodological reasons, while a second does not appear in the paper. Assuming that he would answer Strongly Agree (4) for the two relevant acknowledged conspiracy theories; and split between five agrees (3), and six strongly agrees on the other 11, this boojums mean score in the conspiracy response would have been 3.6, a massive 5.5 SDs above the mean. That would place them above the next highest (below the two suspect proxies), but still significantly below the two suspect proxies.

    Showing me boojums does not prove the existence of super-boojums.
    0 0
  24. They're out there somewhere, Tom, like Moby Dick! Now, pass me my spyglass, and call me Ishmael.
    0 0
  25. doug_bostrom @124, that's a rather self destructive metaphor you've got going. You will forgive me, I hope, if I wish that you neither kill your Moby Dick, nor are killed by them.

    Besides, which, there was only one Moby Dick, and he was fictional.
    0 0
  26. Well, Ahab's already spoken for.
    0 0
  27. Tom,
    We disagree. Landowsky has used an agreed upon standard to delete 20% of his samples. He clearly is willing to delete samples if a reason exists. Outliers are not necessarily gamed, they may exist. The data may be bimodal. We do not know, that is why the data was collected. There is a standard in this research as to how to delete samples and I believe Landowsky has followed the standard. If the standard is to not delete without a reason the statistical analysis is irrelevant. I think we need an expert to tell us how these issues are handled.

    In the survey I worked on we only deleted samples that were contradictory (for example saying they had smoked 1 cigarette in one question and 100 in another question). Reviewing the data afterward, I realized that the nature of the questions made it very difficult to game since the participants did not know what we were interested in (we were interested in the urge to smoke in people who just started smoking). If you claimed you smoked a lot of cigarettes (the most obvious game)you were eliminated. Lewandowsky had a lot of questions he deleted from analysis. Perhaps these additional questions disguised the intent of the survey.

    Psychology surveys are more difficult to parse than temperature analysis. Humans are tricky animals. There is the possibility of gaming. On page 13 of the paper this is discussed. The two samples you do not like are not mentioned specifically in the paper. Lewandowsky claims the results relating to life satisfaction and conspiracy are similar to previous work, which suggests that the results are honest.

    Lewandowsky claims that belief in conspiracy theories in general is correlated with denial of AGW. Is this a surprise to readers who have visited WUWT? I find this conclusion easy to believe, even though the number of responders is low.

    I thought a poster on this forum regularly performed surveys. Can they come on here with expert advise?
    0 0
  28. Reply to #120 Bob Loblaw at 09:05 AM on 7 September, 2012

    How could Lewandowsky have known that the reply from McIntyre was "unlikely ever to arrive" I wonder?
    0 0
  29. Michael Sweet voices at #127 something that has been bugging me for days.

    What a priori evidence is there that indicates that any index of the psychology of conspiracy belief should follow a gaussian distribution? Much in psychology is non-gaussian, and indeed is multimodal (with highly discrete modes), and it's one reason why non-parametric statistics are often used in the discipline. Think about religious beliefs for a start.

    I'd certainly be interested in extreme responses in a survey, but I wouldn't be discarding them out-of-hand. Indeed, in a survey that I conducted for a government project the distribution of responses was bimodal, with the two modes at either ends of the response continuum. I suspect that in the survey that Michael alluded to, similar extremes would also exist, and that these would not imply that a majority view lies somewhere in between.

    Without much more in-depth dissection of the conspiracy phenomenon, I'd not be discarding any response out of hand, simply because it's 'extreme'. Beliefs frequently track in a very different way to something like body size...
    0 0
  30. I’m going to stand up for Lewandowsky here.
    Michael Sweet #127 and Bernard J #129 have a point. Sampling the opinions of human beings is not like like pulling different coloured balls out of a sack. There is no reason to believe that their beliefs should obey a Gaussian distribution, and no reason to reject a respondent simply because his responses are several standard deviations out on a limb.
    Imagine someone faced with this questionnaire announcing “I make a point of believing every conspiracy theory I come across”, and ticking the boxes accordingly. You may say he’s irrational, or that he’s an awkward cuss who’s going to screw up your carefully designed survey, but -too bad - if the population you’re sampling is humanity, you have to take the rough with the smooth.
    The point about the two outliers is not that they’re eccentric, but that they’re suspected of cheating. You could even argue that, since a certain proportion of humanity cheats systematically, a representative sample of humanity should keep them in. You could even argue that a warmist pretending to be a sceptic (or vice versa) has as much right to be heard as anyone else.
    What this shows, I think, is that the feebleness of Lewandowsky’s research lies not so much (or not only) in the weakness of the conclusions, which are sensitive to the removal of a couple of respondents, but that the conclusions are absolutely not generalisable to the world at large. We know practically nothing about the sample - whether they are representative of sceptics or believers at large, or even whether they are representative of readers of the blogs from which they were recruited. (Comments would suggest that they were not). The questions asked were so specific in time and space that the answers tell us nothing, even where numbers are large enough to provide a valid sample. What do I know or care about the Oklahoma bombing? Already I’m excluded from the sample.
    There is just so much wrong with this paper. We haven’t got to the bottom of it yet.
    Incidentally, I believe the two outliers Tom identified were close together in the numbered list on the spreadsheet, and that there was a cluster of conspiracy theorists at this point. This suggests that possibly they all came from the same blog. Would someone like to ask Stephan whether he can identify the source blogs for individual respondents?
    0 0
  31. Foxgoose@128

    ...because anyone with any experience in such surveys knows that you never get 100% response - there will always be someone that won't reply. You take what responses you get, and you move on.

    McIntyre is no different from anyone else that was contacted about the survey, and I would be surprised if he were to get any different treatment from anyone else - and to give him special treatment and make extra effort to get him to participate (beyond what was established in the protocols for the study) would likely give reason to wonder about the results.

    You seem to think that McIntyre actually matters as an individual - he's just another one of a group of people invited to participate, and in the context of the study he doesn't mean anything more than "blogger X". He make think he's more important than others, and you may think he's more important than others, but for now all you've got is a tempest in a teapot, and you're still circling the drain of "conspiracy theory".
    0 0
  32. I'm a little confused here. Why do the conspiracy theorists among us actually care that this survey was or was not offered on self-proclaimed-skeptic blogs?

    The results were sorted and analyzed based on the answers, not the source.

    Is the fear that skeptics who visit other blogs aren't representative of "real" skeptics, and so the study was improperly skewed towards a not-acceptable-as-a-real-skeptic population? Or is the problem that someone like McIntyre didn't get to manipulate the outcome by coaching his visitors through the answers?

    I rather suspect that if the survey had been offered at "real" self-proclaimed-skeptic blogs, the conspiracy-signal would have been magnified. It could only have helped to strengthen the results -- unless the inane, bizarre and quite indefensible comments that I see on self-proclaimed-skeptic blogs are somehow not representative of the nutters that frequent/post on them, or they would have suddenly put on their sanity-caps while filling out the survey, then taken them off in order to post more stupid comments on their favorite self-proclaimed-skeptic blog.

    It is what it is. Get over it.
    0 0
  33. Sphaerica #132
    Why do the conspiracy theorists among us actually care that this survey was or was not offered on self-proclaimed-skeptic blogs?
    Because Lewandowsky said it was, and all the major sceptic blogs denied having found it, and when McIntyre did find it, it came from a different person, with a different introduction, attached to a different questionnaire, one week after it had already been publicised and widely discussed on anti-sceptic blogs. All points which need explaining, and which Lewandowsky refuses to address.
    Is the fear that skeptics who visit other blogs aren't representative of "real" skeptics, and so the study was improperly skewed towards a not-acceptable-as-a-real-skeptic population?
    Of course it is. (-Snip-)
    0 0
    Moderator Response: [DB] Ideology snipped.
  34. 133, Geoff,
    Because Lewandowsky said it was...
    So what? He said he did, they claim he didn't... oh, wait, McIntyre found it but ignored it.

    Again, so what? How did this in any way invalidate the study?

    Or do you just feel for McIntyre's poor, trod on sense of importance?
    Of course it is.
    Oh, Jeeze. So you think there was a warmist conspiracy to pretend to be skeptics, just to get them to look bad?

    Holy moly...
    0 0
  35. Because Lewandowsky said it was...

    And Lewandowsky was indeed correct.

    ...and when McIntyre did find it...

    Weeks after it was sent.

    Minimum lesson learned: Read your own email.
    0 0
  36. GeoffChambers @130, I assessed the distribution of conspiracy theory responses for acceptors, undecided, and rejectors of AGW. Only for recectors is their any hint of a bimodal distribution; and that hint consists solely of the two dubious proxies. Therefore the suggestion that the distribution is bi-modal is ad hoc. Likewise, based on the limited literature I have read on the topic, there is no record of anyone strongly believing more than a few conspiracy theories simultaneously. Based on that, the supposition that such people do in fact exist and came out of the wood work solely for the benefit of this survey is also ad hoc.

    We are faced with two theories about the two suspect responses. The first is that they are attempts to game the survey. The theory that people attempted to game the survey is already confirmed , with at least 5% of responses known to have been attempts to game the survey. That makes the first theory both simple, and a theory with support behind it. The alternative theory is complex because it requires the existence of (so far, unsighted elsewhere) universal conspiracy theorists; and also requires a bimodal distribution, but only among rejectors of AGW.

    Given the choice between a simple theory with supporting evidence, and a complex theory requiring ad hoc hypotheses, I will always choose the former. Never-the-less, I am not saying Lewandowsky should exclude those two responses from the survey. Rather, I am saying he should make people aware of the reasonable doubt, and ideally show that their inclusion makes no difference to the result. (I believe that is indeed what he is claiming, although I have my doubts.)
    0 0
  37. Tom Curtis #136
    I agree absolutely that the probability is that the two outliers were attempts to game the survey, and that Lewandowsky should at least have indicated their existence in the paper. He should also have looked at the distribution of sceptics and conspiracy theorists by source blog - the most basic test of any organised attempt to cheat.
    I’ve been thinking further about the point I made at #130 about survey design and the legitimacy or not of various responses. Imagine the researcher (let’s call her Alice) doing a follow-up interview with a respondent whose questionnaire responses arouse suspicion (call him Humpty Dumpty)
    Alice: “I see you’ve ticked ‘Agree strongly’ to all the questions”.
    Humpty Dumpty: “That’s right. I always tick the left hand box when filling in an online survey.”
    Alice: “So these answers don’t represent your true beliefs?”
    Humpty Dumpty: “Yes they do. It’s my firm belief that I should always tick the left hand box.”
    Alice: “But you don’t really strongly agree...?”
    Humpty Dumpty:”Yes I do. I strongly agree with ticking the left hand box. If I didn’t , I’d tick the right hand box wouldn’t I? That’s logic.”
    ... and so on.
    The serious point is that an interview is an artificial construct which derives its legitimacy from its similarity to other experiences in our lives. If someone stops you in the street and asks you the way to the post office, you don’t think of lying, but answer to the best of your ability. If they added: “Do you think it’s to the left or to the right? You’re not allowed to say you don’t know” you’d probably feel seriously p*ssed off, and refuse to cooperate.
    Telephone interviews, and even more so online questionnaires, are more removed from the face-to-face situation where human beings normally interact, and feel the natural tendency to cooperate. Social scientists are making enormous hidden assumptions about the psychology of their respondents when they put their trust in this sort of exercise.
    0 0
  38. Tom Curtis:

    Recommend that you post a survey on this thread asking our readers if their eyes glaze over when they see yet another GeoffChambers post. I know that my eyes do.
    0 0
  39. Here’s another take on the Lewandowsky survey and the consternation it has generated within Ostrichville.

    Research Links Climate Science Denial To Conspiracy Theories, But Skeptics Smell A Conspiracy”.By Graham Readfearn, DeSmog Blog, Sep 5, 2012

    This graphic is embedded in it.

    Image and video hosting by TinyPic:
    0 0
  40. John! You showed them the secret sign! With the secret logo! Doing the secret hand wave!

    You're going to blow the whole conspiracy wide open! Quick, quick, delete that comment!
    0 0
  41. Sphaerica:

    I figured if I posted a link to Readfern's article, GeoffChambers might not have time to post comments here. It's an obvious dilution tactic. On the other hand, Graham Readfearn may not be pleased.
    0 0
  42. #138 John Hartz
    Tom Curtis:
    Recommend that you post a survey on this thread asking our readers if their eyes glaze over when they see yet another GeoffChambers post.
    Call me a conspiracist, but that sounds like a coded warning to Tom not to fraternise with the enemy.
    Yes, I’ve read the deSmogasbord blog. Fair made my mouth water.
    0 0
  43. Once again we find that a broken clock can occasionally tell us the truth without actually being a working clock.

    Geoff: Social scientists are making enormous hidden assumptions about the psychology of their respondents when they put their trust in this sort of exercise.

    Spot-on, Geoff, you're exactly correct. Social scientists employ methods invisible to and distrusted by you because you've not spent the time necessary to understand how social scientists work.

    Beg, borrow or steal a research methods syllabus from a PhD social sciences program, follow the readings therein, let the shades fall from your eyes. Who knows? Your intuitions might be correct but without getting your game on you'll never be able to do more than guess, let alone usefully articulate your thoughts.
    0 0
  44. @GeoffChambers #142:

    Dang, you broke the code.

    You should seriously consider applying for a job with MI5.

    Then again, you may already work for MI5.
    0 0
  45. #143 doug_bostrom at 06:29 AM on 9 September, 2012
    Spot-on, Geoff, you're exactly correct...

    But I do trust social scientists, and I do understand how they work. That’s the basis of my criticisms of this paper.
    0 0
  46. Geoff, you've seen the gold standard for dilettante critique of Lewandowsky. Your "contribution" doesn't overlap with that space at all, has instead been concentrated on chronological trivia.

    You say you understand how social scientists work? Ok, by that I take it you're claiming you have a grasp of some branch of social science at the PhD level, even if you've not actually produced a dissertation. That's not impossible but we have only your word for it. Why not show how you're capable of working at the level of a social scientist? Produce a critique of Lewandowsky's work at an academic level, a comment that might be suitable for publication, something relying on formally established alternatives to the methods chosen by Lewandowsky and employing those to show what's wrong with with his work. Don't skip anything, don't imply "trust me," don't leave anything to our imaginations.

    Failing that, recognize that you don't understand how social scientists work, consider instead being suitably humble and circumspect.
    0 0
  47. This article went up on 16 July and attracted 5 comments in the first six weeks. In the week since I posted comment number 6, there have been a further 140 comments. Nine of these were requests by me for simple information about SkepticalScience’s involvement in publicising the Lewandowsky survey. After at first being told that John Cook was too busy to answer my questions, I was finally told that I should contact him via his private email. Before I could do so, he contacted me, offering to answer my questions. In an exchange of eight emails, I learned that Skepticalscience had indeed received a request from Lewandowsky on 28 August 2010, and that Cook had replied that he would put up a post the same day. But Cook was unable to tell me when the post went up or when it was deleted, and he ignored my questions about comments; were there comments to the post, and if so, were they still available?
    SkS keeps a file of comments deleted by moderators. It seems inconceivable that they would have deleted bona fide comments, even if the post was taken down at the end of the 2/3-month survey period. I ask again; were there comments to the post publicising the Lewandowsky survey, and if so, do they still exist?

    Moderation here is firm but fair. I see nothing in the rules that states that all comments must be of PhD standard.
    0 0
    Moderator Response:

    [DB] The Deleted Comments bin contains the most recent 60 comments not compliant with the Comments Policy. As such, the current list goes back to 5:02 am on August 25, 2012 (Brisbane, Australia time).

    Only those comments not in compliance with the Comments Policy receive moderation (excepting the helpful response to those occasional comments seeking aid, like this one). Every effort is made to avoid deletion of comments, which is a distasteful last resort. Even then, some participants give the moderation staff no choice but to delete comments.

  48. That's right, Geoff, you have wide latitude to create your own reputation, establish the worth readers will attach to your comments. You claim you understand social science but your ability to describe let alone support your misgivings about Lewandowsky's work belies your confidence. That's a problem you can repair but doing so will require a lot of effort, as you'll see if you follow my tip to obtain a research methods syllabus for a PhD social science program.

    The fellow who runs Science of Doom is exemplary in terms of showing the potential of autodidactic learning. I wonder if there's a site similar to SoD that's devoted to cognitive psychology as opposed to geophysics?
    0 0
  49. Igeoffchambers@147: "After at first being told that John Cook was too busy to answer my questions"

    Now that is a distortion of what was said. Early in the comments, someone else had asked what the best way to contact John Cook would be, and a couple of people (neither named John Cook, and neither being people that could be interpreted as speaking directly on his behalf) had the following comments:

    "curiousd, John Cook like most of the contributors to SkS is very busy, and not paid for his contributions to SkS, nor for hosting it. I am sure that he attempts to read the comments on a regular basis, so that commenting here or on the SkS facebook page are probably among the most reliable ways to contact him. There is, however, no guarantee. For what it is worth, I have drawn attention to the fact that you are trying to contact him on another forum I know he frequents. There is no guarantee that he'll read that, either, and even if he does he may well be too busy in any event. "

    [my emphasis]

    "If you want to know how John Cook interacted with Stephan Lewandowsky on his survey, you best send a querry to John via email. He's a very busy person and no longer has time to read all of the comments posted on SkS. "

    By your own admission in comment #147,before you could send him an email he did in fact email you. To try to pretend that anyone implied that you would not get a response, as you seem to have done here, would not be discussing in good faith.

    If you worded that poorly, and did not mean to imply this, then an apology would be a good step to restore some faith. Alternatively, if there is another comment in this thread that you think can be interpreted in such a fashion, feel free to point to it.

    (Note that I have already looked for all occurrences of the word "Cook", and "busy", and checked to see if any of the moderators comments suggest that something has been snipped.)

    0 0
  50. Bob Loblaw #149
    No, I did not mean to imply that anyone implied that I would not get a response. My apologies if I gave that impression. I was surprised that Cook offered to answer all my questions in a private email. I wasn’t after anything secret, but simply wanted to know if the post had gone up and if there had been any comments. The private email and the offer to tell all naturally whetted my appetite. I was sort of hoping he’d say: “Look, I’d rather this didn’t get about, but Tom Curtis tore the survey apart in the comments, so we binned them”, but no. Nada. Just the legendary Australian taciturnity when dealing with strangers. Pity.
    0 0

Prev  1  2  3  4  Next

You need to be logged in to post a comment. Login via the left margin or if you're new, register here.

The Consensus Project Website


(free to republish)

© Copyright 2019 John Cook
Home | Links | Translations | About Us | Privacy | Contact Us