Guest essay by Eric Worrall.
JoNova reports on a hilarious error in Lewandowsky’s paper “The Role of Conspiracist Ideation and Worldviews in Predicting Rejection of Science” . The calculated age of survey respondents has reportedly been skewed by one “outlier” who claimed to have been born in the palaeolithic, 32,757 years ago.
Raw data – line 607, http://websites.psychology.uwa.edu.au/labs/cogscience/documents/PLOSONE2013Data.csv
Lewandowsky was informed of this error over a year ago – and has reportedly done nothing to address this glaring problem with his calculations and conclusion.
According to Jo Nova;
“Lewandowsky, Gignac and Oberauer put out a paper in 2013 which was used to generate headlines like ‘Climate sceptics more likely to be conspiracy theorists’. The data sample is not large, but despite that, it includes the potential Neanderthal, as well as a precocious five year old and some underage teenagers too. The error was reported on Lewandowsky’s blog over a year ago by Brandon Shollenberger, then again by Jose Duarte in August 2014. Nothing has been corrected. The ages are not just typos, they were used in the calculations, correlations and conclusions. The median age was 43 but the mean age was a flaming neon 76. One wildly old person in the data skewed the correlation for age with nearly everything:
That one data point – the paleo-participant – is almost single-handedly responsible for knocking out all the correlations between age and so many other variables. If you just remove the paleo-participant, leaving the minors in the data, age lights up as a correlate across the board. Further removing the kids will strengthen the correlations.”
Worst of all, the bad data has apparently led to a demonstrably erroneous conclusion. According to Jose Duarte,
“This would be a serious problem in any context. We cannot have minors or paleo-participants in our data, in the data we use for analyses, claims, and journal articles. It’s even more serious given that the authors analyzed the age variable, and reported its effects. They state in their paper:
— “Age turned out not to correlate with any of the indicator variables.”
This is grossly false. It can only be made true if we include the fake data. If we remove the fake data, especially the 32,757-year-old, age correlates with most of their variables. It correlates with six of their nine conspiracy items, and with their “conspiracist ideation” combined index. It also correlates with views of vaccines – a major variable in their study. See the graph below.”
http://www.joseduarte.com/blog/how-one-paleo-participant-can-change-the-outcome-of-a-study
Lets hope that Lewandowsky finally takes action to correct the error, and amends the erroneous conclusion of his paper, which is inferred from Lewandowsky’s analysis of the grossly defective data. Of course, while he’s at it, I’m sure we could suggest a few other defects with Lewandowsky’s work which he could correct.
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
For what it is worth, 32,757 is one off the size a small integer in computer science which is 32,767…
Ten off? I suggested on Jo’s page the cause might have been a birth date erroneously entered for 11 years in the future with the low 15 bits (why not 16?) saved in the database.
It could also be a valid response from a very prescient ovum.
Ric, you’re only half correct, you’ve neglected the requisite spermatozoon
Possibly last bit determines -ve number if set to 1. Also 15 bits may include 0 which would actually = 16 bits 0 to 15
Yes, 10 off. Brain not fully engaged before my morning walk…
It’s a time traveller come back to laugh at Lewandowsky!
Bit-ters are good before walks..You must have a very large bit bucket..Har-dee har har..
int( -11) == uint(32757). Age = Round(current_date – birth_date). If someone entered their birth year as current_year + 10, a single digit error or if the entrant was born in 2000 + something and flubbed the decade digit, this would result. The first case would require two mistakes on the part of the entrant, the second, one mistake on the part of a preteen entrant. Any other scenario would seem to indicate fraud ( too many error needed to create).
DesertYote
If 10 years old, that would be current year – 10. Not current year + 10.
if about 20 years old (born in between 1993 and 1994 for example) that would still be current year -20
RACook,
What does that have to do with what I posted? I said “flubbed a digit”, sheesh.
What is the job of peer review?
To enforce the consensus?
Who did the peer reviews? (These should be three “qualified” climate scientists, right?)
Plus the senior “qualified” editor who selected those three people, plus the (junior) editor who did the physical reading and editing of the paper itself.
But, of course, anonymous star-panel “peer review” is the absolute guarantee of accuracy and unbiased results in climate science scientific papers, isn’t it?
when you’re a member of the Church of CAGW, you suggest reviewers who are also blindly-adherent true believers.
note: When you submit a paper to PLoS, you submit names of suggested reviewers, and also of research area “competitors” who might be adverse. It is up to the reviewing editor then to use judgement in assigning at least two reviewers who can objectively review it.
Jimbo, I think looncraz has the answer.
What a strange notion. Reviewers haven’t got time to check the actual data!! They’ve got a world to save dontchyaknow.
As long as it smells right…..https://hro001.wordpress.com/2012/02/02/phil-jones-keeps-peer-review-process-humming-by-using-intuition/
If the “scientist” doesn’t care about the data, why should the reviewer?
/sarc
A peer reviewer is a peer — a scientists in the same field or with some knowledge of the subject — who reviews a paper. If you’ve published in the subject area before, the editor might ask you to review a paper but I don’t think an editor ever pays for a peer review. If you work for someone who will let you review the paper while you’re “on the clock” then you could paid to do the review. In my experience, it’s a professional courtesy. You review other papers because someone had to review yours. Usually the reviewer will make some recommendations about how the paper could be made better, more clear but they could recommend against publishing; seems like that is what should have happened in this case. I don’t think it’s normal for a reviewer to look closely at the raw data but I would if the conclusions seemed odd or unsupported. It’s a “review” of the paper, one is not expected to run the experiment and see if one gets the same results.
One peer reviewer could have been his colleague in the Psych Dept at UWA, Prof. Carmen Lawrence, a warmist former Labor Party Premier in Western Australia.
“Peer review” is based on the assumption that the reviewers are at least equally knowledgeable as the author(s). The system is therefore vulnerable because the reviewers may not be.
The whole AGW farce will go down in history as a textbook example of a massive infarct of the review system.
A short integer which is 16bits has the range of 0-65536 if unsigned and
-32768 to 32767 if signed. I wouldn’t be surprised if the fellow whindid that survey did a bit of programming in his life.
The largest number you can fit into a 16-bit unsigned integer is actually 65535. There are 65536 possibilities but that includes zero. 😉
Anyway, I think the explanation below and elsewhere that it’s a full date of birth (March 27, 1957) is probably the answer. Next they’ll be calculating TCR by multiplying by the change in anthro forcing instead of dividing. 😉
More likely someone entered DoB (032757) instead of age – an American, since the rest of the world don’t start with the month.
Well the link to the original dataset / paper within the above article as linked here:
http://websites.psychology.uwa.edu.au/labs/cogscience/documents/PLOSONE2013Data.csv
Seems to have been removed. Any theories as to why this might have happened? or am I being too conspiratorial in assuming it has been removed to hide the evidence?
The PLoS link is now reporting “Not Found” as well.
One can hope that this post is the proximate factor…
… In which case; Congratulations Mr. Worrall & Watts!
OK, now the download link is working. Maybe I rushed to a conspiracy,or……
The PLoS link, the “official” published paper, is still Not Found. This is the first link at the top of the Post.
– the 32,757-year-old paleo-participant –
That’s no way to talk of my mother !!!
…in-law.
+1!
The participant has been found. Hurray.
http://www.nhm.ac.uk/resources-rx/images/1008/neanderthal-woman-200-109116-1.jpg
Sou might object to you posting her picture.
I’m trying to figure out what she’s holding.
I’m trying to figure out what she’s holding.”
A broken hockey stick
A scapula. From her former boyfriend.
A birch bark Erlenmeyer flask.
You know, some things can never be unseen!
Maybe that’s an old guy cave-man with Mann-boobs?
ManBearPiG!!
I think Lew is right keeping the paleodata. The great wisdome of Ötzi’s bigbigbigbigbigibigbigbig,etc-brother tops any opinion coming from present living, minor or not.
But Ötzi lost a serious game of ‘Cowboys & Indians’.
How can a loser be a serious role model for anything (except dodge those arrows).
As for his (putative) bigbigbig (etc to the fortieth power, squared) brother – not a lot of common DNA.
That said, I agree that our Best Buddie Forever, Lew [Doctor, good with animals and small children, and the rest – I assume], may need to review, adjust and re-publicize his – ahhh – ‘data’.
Auto
A 32,757 year old sounds about as right as the rest of their data. Maybe it’s the candles on his birthday cake that is contributing to AGW!
Sorta has the Frank Zappa stare. http://ecx.images-amazon.com/images/I/51FceUcTWLL._SY355_.jpg
Maybe J Z Knight was the respondent, channeling Ramtha.
“This is grossly false. It can only be made true if we include the fake data. If we remove the fake data, especially the 32,757-year-old, age correlates with most of their variables.”
Ah yes, she of many accents, and very skilled in charging huge sums for bunkers in anticipation of lizard hominid invasion. 35,000 years old. Perhaps we have a match.
She & Loony Loo are a match made in Lalaland.
Wow, 32,757 years old! That beats Mel Brooks’ 2000 year-old man by a long shot. Oy!
I’m just glad I am not around his birthday cake when they light the candles, it would be like Ash Wednesday.
Too funny……
So many people here complain when the raw data is “adjusted”
When it isn’t “adjusted” they make fun of the raw data.
When the data are so obviously, hilariously, false, adjusted or not, they deserve to be made fun of.
Kind of like when a station temperature in San Diego shows -57 degrees F ?
Clearly bogus data cannot be “adjusted” (whatever it means), it needs to be
1) explicitly flagged
2) thrown away
Then, there is the question of how much other, less obvious crap in the data…
eg, the 5 yo.
What we want is correct data. We don’t want adjustments without explanations, and we don’t want data that clearly incorrect, or for that matter data that is incorrect but not very clearly. If there is sensible and logical reason for adjustments we are fine with that, but we want to know what those reasons are and what adjustments were made. That is why we insist on seeing the raw data. Without it there is no way of telling which corrections should be made, and which corrections should not be made.
Adjusted, it’s 65,514. Corrected, it’s 1.
I think 32,757 is too old to include in the data. Now if he had been only 1,000 years old…
I thought the upper bound on this was 969…..
David, I’m not too sure what your point is. Certainly some observed data is spurious. There is also a serious possible problem when committed activists are doing lots of “adjustments” to their data in the dark.
Does that mean there is some gray area here? Sure. Anybody with a brain sees that, but that doesn’t mean there’s anything wrong with jeering at either 32,000 year old respondents, or endless “adjustments” that always trend in the direction of a hockey stick.
There is an estimate in the comment above that corrects it to 10. Would you accept that as the real age or omit it along with the other 25%?
You can’t be that dense, this has nothing to do with adjusting data, the data is wrong. Worse the bad piece of data was from a small set of data and is easily caught and corrected (not adjusted).
Lews paper is a prime example why people are putting less and less faith in science.
Science is dead. What remains is a giant propaganda machine for tyrrany, paid for by money extorted from all of us.
That is exactly how I see it.
Je suis jorgeka….never mind. (:
Misused but not dead.
Science was never going to be without genitals or an anus. And never will be.
Like other institutions, it attracts folks of a certain make-up, and repels others. Like all institutions, it has its peculiar effects & affectations.
Fancy takes on the nature & reality of science are like the Nobel Savage.
It is neither all that awful, nor wonderful.
It was not included in the Constitution, and I submit that it certainly would not be, today. And if it was genuinely scary, or honestly our salvation, it would be front & center in Western Law. Yea or Nay. Instead, we have effete ‘best science directives’.
Luckily, it was peer reviewed. Otherwise a whole heap of rubbish might have gotten through
DC: You really need to understand the difference between ‘adjusted’ and ‘corrected’.
Interesting. I opened the data file “http://websites.psychology.uwa.edu.au/labs/cogscience/documents/PLOSONE2013Data.csv” in Excel, and found the “Age” in column AO. The forrmula “=MAX(AO2:AO1002)” then calculates the max age in the list as “32757”.
I’d be interested in the source of this obviously-wrong number, because it would probably tell us a lot about the quality of the other values in the spreadsheet.
I think they need to multiply by the square root of -1 to correct for all of the other imaginary data…
The fact that it’s 10 away from 32767 (zero-indexed) is pretty fishy, almost like someone stuck a
test outlier in there and no-one fixed it or some algorithm hit a bounds condition. Either way, there’s
no excuse for missing something so far out of range, especially in a “scholarly” study.
It’s pretty obvious they didn’t “study” the data very well.
I would have asked a 32,767 yo how hot was it where he lived when he was a boy and a few other questions about sea level etc… plus if a paleo diet really does work!
A 32,767 must have something to say about climate scares.
The 32767 years old person’s gender is marked as 2, which I presume is a Female !
I wouldn’t be so sure. I suspect that she wasn’t 32767 either. Women lie about their age.
Given his age and the fact he is still in good enough shape to answer this survey, I would count that as very strong evidence paleo works exceptionally well.
How many more skeletons will be found in the cupboards of scientists?
This paper, the analysis of the data, the validity of the conclusions and the quality of the pal review process meets the standard used in climate science – so there’s no problem. Nothing to see here.
The thing is my friends, that we are always playing by their rules in this political war. This cretin gets the juicy headlines and then later when we find out about how terrible the study really was. Long after the propaganda damage is done; we discover the travesty that is the “study”. Hell, look at how long the “97% consensus” has held on.
In watching this subject since the “new ice age is coming” days, I have come to believe that science went way wrong in buying into the Jim Hansen theory of CO2 and its supposed effects. The whole thing needs to be redone — a fresh start. But that, my friends, can not happen as long as the government is controlling funding and handing out billions and billions of “free money”. The guy who pays the piper gets to call the tune and we all know what that tune is.
You get it, Mark. Never, negotiate with rules set by your opposition to suit themselves … always seek to have the paradigm to your advantage. Arguing ‘science’ is a waste of time when policy makers, political elites, have already appropriated the junk science for their own ends. Hell, everybody knows that the ‘science’ is junk.
The explanation could be quite simple. The respondent put his birthdate in the age column. He may have been born on March 27, 1957
Likely explanation. . . but it also points out the abysmally inept attempt to processing the data that renders the entire paper meaningless.
My guess as well. And yes, it does indicate the abysmal level of the data processing in the paper.
In Excel, set up for the UK, the numeric value ‘32767’ resolves to ’16 September 1989′ when formatted as a date. It might be that date of birth (and an another case of ‘expert scientists’ not knowing how to use Excel.)
Only if they were American…
10 below signed int overflow limit.
Ha ha
Who uses two byte integers these days? Didn’t they go out of style with the 80386 ?
Those who are limited to 64 KB?
SMALLINT in SQL, PIC 9(1) to PIC 9(4) in COBOL, FIXED BIN(15) in PL/I, short in C, C++, C#, and Java, and probably a few uses that I’ve forgotten. You need to get with the program (so to speak).
Embedded and microcontrollers. Robotics.
If you know the data is going to fit into the size of an int, you generally don’t use something larger. You might be processing millions of these values. You don’t take up more resources than necessary in good design. Computer chips have instruction sets that works with these “machine” size values.
The CSV file is text though isn’t it? If I write “1” , I typed a ASCII value of 49 (decimal), a char character “1”. In other words its considered as text and parsed by an algorithm into its number form when required.
garymount
Most of today’s CPU’s are of a 32 or 64 bit design and access their memory systems in word fashion. So technically, loading and storing a 2 byte integer is a waste of resources (i.e memory bandwidth.)
Primary or Level 1 cache typically range from 2 kilobytes (KB) and 64 KB, which is why you don’t unnecessarily bloat your data.
Access width on the caches is still in word lengths.
And all data reads and writes still have to thru the caches ant go to main memory.
#NeandertalLivesMatter2
Lew embraces dieversity!
GW – and so, why does the paper’s author go ahead and usethis crap ?
Because that let him dismiss correlation with age of respondent, which would have overwhelmed the alleged skeptic correlation. Not part of the narrative, ya know.
I hope it was peer reviewed.
I think it was pal reviewed.
Probably beer reviewed by a pal.
On those mornings I wake up feeling really old, I’ll just think of this paleo-dude and realize I don’t have a thing to complain about. 😉
I, too, will do that.
If I can remember . . . .
Auto
I wonder if his hips are as creaky as mine. 😉
This might be useful. Can we get good historic data from this one individual that can be used in further climate research? If they kept good notes about temps., CO2 concentration, sea levels, solar radiation, cloud cover, etc. there might be much to learn.
Don’t miss the significance of the children in the sample.
It is unethical to exploit minors. The permission of the parent or guardian is needed.
Did the Journal and University check that?
Or do the UWA and University of Exeter not really care much about child welfare?
What’s the name? Sasndusky? Something like that
Isn’t Sadnasty in the State Penn?
He bet Lew obtained preemptive white-wash (written by Lew) from the U and its “Research Ethics Committee”.
Lew studied Lew research behaviour, concluded Lew was a saint.
Yet another example of the “amateur hour” aspect of all of Lewandowsky et al work. Significant errors in data collection, methodology, interpretation of prior data, ethics, and more.
Alleged “explanations” – such as the ridiculous use of Skeptical Science poster data to ‘justify’ Lewandowsky obtaining virtually all of the Moon Landing Hoax data from sites highly biased against skeptics. When in reality NONE of the data was obtained thru the SkS site – AND where their analysis of internet traffic data showed a complete cluelessness – no connection with reality or fact.