Welcome Guest [Log In] [Register]
Sky Dragon is up and running. We need people to sign up and add content. Introduce yourself at the General discussion forum. Speak up to welcome a few others, start a thread, or contribute to someone else's thread.
Welcome to Sky Dragon.

This site has been set up to allow convenient discussion of Earth's atmospheric greenhouse effect. The main topic for discussion here is the physics of how (and if) the atmosphere helps to determine Earth's current temperature. Other spin off topics are also supported.

Registration is not required to join in. Anyone can join in immediately. Anyone may also register as a user, and gain access to some additional features. Registration is simple, fast, and completely free.




Username:   Password:
Add Reply
The Drowned-Polar-Bears Paper - Monnett & Gleason; Rebuttal
Topic Started: Feb 12 2012, 03:16 AM (1,051 Views)
Sean McHugh

An examination of the drowned-polar-bears paper and the investigation surrounding it:

Observations of mortality associated with extended open-water swimming by polar bears in the Alaskan Beaufort Sea, by Charles Monnett and Jeffery Gleason

The Monnett/Gleason paper can be found here.

Introduction:

It would nearly impossible to exaggerate the role of the Monnett/Gleason paper in placing polar bears as the icon of global warming theory. This is from the Guardian, which is known for its catastrophic anthropogenic global warming orthodoxy:

Quote:
 
The paper quickly heightened public concern for the polar bear. Al Gore, citing the paper, used polar bear footage in his film Inconvenient Truth. Campaigners focused on the bears to push George Bush to act on climate change, and in 2008, the government designated the animal a threatened species.


It was the first animal to be classed as a victim of climate change. And from Nature Magazine:

Quote:
 
The idea that polar bears could drown like this became a rallying point for advocates of action on climate change, most notably appearing in former US vice-president Al Gore's film An Inconvenient Truth (2006).


A plethora of similar descriptions could be cited that speak of its public and political impact. The study was based on the sighting of three specific, supposedly dead, polar bears, during flights over the Arctic's Beaufort Sea, following four days of strong winds and rough seas. The drowning of the bears was attributed, by Monnett and Gleason, to global warming. This attribution was made on the assumption of the bears being weaker and having to swim further due to melting ice.

But did this paper deserve such recognition and influence? Is the science in it sufficiently strong? Are there issues with the integrity of the study? Did the paper receive proper peer review and endorsement? I wish to examine those points. In February 2011, investigators interviewed Monnett and Gleason. Though the investigation was fundamentally with regard the possibility of Monnett inappropriately assisting a government grant going toward one of his paper's peer reviewers, the interviews were very interested in the method of the data collection and the conclusions drawn from it. One hears the statement that the science was not the issue but in reading the interview transcripts, it becomes quickly clear that the science was a very major part of the investigation:

Quote:
 
Although the director of BOEMRE, Michael Bromwich, has stated that the investigation has "nothing to do with [Monnett's] scientific work, or anything relating to a 5-year-old journal article, as advocacy groups and the news media have incorrectly speculated," yesterday's 3-hour session focused exactly on those issues, says Jeff Ruch, the director of the Public Employees for Environmental Responsibility in Washington, D.C.


In the first interview, Monnett is told that he has been accused of scientific misconduct (p3) and that it relates to his statistics (p83). It would reasonably be the case that one would want to examine whether the paper stood on its own merit or required "highly inappropriate" assistance, gained by steering grant money to obtain a favourable review and promotion of the paper.

Monnett's first interview, February 23, 2011, is here.

Gleason's interview is here.

Monnett's second interview, August 9, 2011, is here.

Criticism of the Interviewers:

The defenses for the paper, that I have been given, is that it is 'fine', that the interviewers were incompetent, that the paper received no peer criticism and that nothing has come of the investigation. Let us first look at the Green criticism of Monnett's first interview:

Quote:
 
May then tried to argue that there were actually 63 polar bears, because 4 live polar bears + 3 dead polar bears / 11% = 63. Dr. Monnett quickly grew frustrated at the innumeracy and illogic of this calculation.


Eric May, the interviewer, was then rudely accused of being unable to do fifth grade cross multiplication. May remained congenial during the extended barrage of insults. In reading the Monnett interview, I also thought that May seemed a bit slower at that point, which was surprising given that elsewhere he performed a cross multiplication calculation quicker than Monnett. He answered, "thirty-six". Monnett replied, "whatever that number was, yeah, 36..". That's on page 56. A little after is when Monnett insults May with repeated words like, "stupid" and "goofy". It has been my experience, and the experience of others, that warmists tend to be more rude and vitriolic than sceptics. The insults were doled out through pages 1/60 and 1/61 of the first interview. This, from page 1/60, is what brought it on:

Quote:
 
ERIC MAY: “If seven total bears, four swimming, uh, and 3 three drowned represents 11 percent of the population”

CHARLES MONNETT: It doesn?t.

ERIC MAY: Okay, and we?ll – let me, let – “of bears before the storm, then the total number of bears after the storm is 63,”


After that Monnett began insulting May's maths. But the maths per se wasn't what was wrong. If 7 represents 11% of the whole, then the whole is 63. May worked that out quickly too. It didn't come from the paper. There appeared to be nothing wrong with his mental cross multiplication. Monnett showed poor form in making that particular charge. The question is, why did Eric May first add the 4 live bears to the 3 allegedly dead bears? The answer might lie in the Monnett/Gleason paper itself. Quoting (emphasis by SM):

Quote:
 
Limiting data to bears on transect and not considering bears seen on connect and search segments, four swimming polar bears were encountered in addition to three dead bears.If these bears accurately reflect 11% of bears present under these conditions, then 36 bears [4 live bears/.11 = 36] may have been swimming in open water on 6 and 7 September, and 27 bears may have died as a result of the high offshore winds [3 dead bears/.11 = 27].


That actually does sound like he is talking about 7 bears in a concurrent sample, 4 alive and 3 dead. In fact, it is arbitrary whether one considers there to be two separate data samplings or one larger one. Though Monnett intended the former, one can see the possibility of someone temporarily inferring the latter. That this was the simple misunderstanding is made more evident when Monnett finally says to Eric May that the live and dead ones were seen on separate samplings, one before the storm, one after. That appeared to sufficiently resolve that matter.

The transcript shows the investigators as being quite unconvinced with the Monnett/Gleason study. As might be expected, searches for comments on the Monnett/Gleason work result in a flood of post-investigation articles. Sifting eventually located this one from 2008, which was written before the investigation started:

Quote:
 
Where Are All The Drowning Polar Bears?

. . . . . . .

Last fall, as a massive media campaign reminded us, the extent of Arctic ice was at an all-time (since 1979) low, yet we cannot recall a single report of a drowned polar bear as a result. Surely, with all the attention on polar bear well-being that arose as the Interior Department considered its ESA decision, if there were evidence of polar bears drowning last summer, it would have been held up front and center. But it wasn’t. Because they weren’t.

. . . . . . .

In 2004, the researchers saw four, that’s right 4, polar bear carcasses floating at sea where they had never seen any in previous surveys.


And many years further on, the same question remains. If the paper's science is right, why have there been no subsequent occurrences? Note again, how the "massive media campaign" was spawned by Monnett's and Gleason's study. Here is the kernel of the paper:

Quote:
 
Limiting data to bears on transect and not considering bears seen on connect and search segments, four swimming polar bears were encountered in addition to three dead bears. If these bears accurately reflect 11% of bears present under these conditions, then 36 bears may have been swimming in open water on 6 and 7 September, and 27 bears may have died as a result of the high offshore winds. These extrapolations suggest that survival rate of bears swimming in open water during this period was low (9/36=25%).


In other words, 4 bears found swimming, 3 later found drowned, represents 25% survival across the "population". That's it. That's Monnett's E=mc^2. During his investigation interview, Monnett makes this derivation painful to the point of cruelty. I believe that that contributed to the communication problems during the interview. I will now present points from the interviews, followed by my own:

Points From the Interview

Very Poor Recollection of Supposed Greatly Significant Observations:

It is interesting, that in his first interview, Monnett merely thought it was on more than one flight that the four dead floating bears were seen (25-26). By page 31 it becomes two or three flights. In his paper, Table 2 indicates that it was with four flights! You think he would remember his ground-breaking observations better than that. This also raises the question of how many transects (assigned flying runs) were involved and/or how many kilometres were completed during the surveys when the live swimming bears were counted, compared to how many were completed in obtaining the dead count. In this regard it is interesting to note that, whereas Table 1 has the flights finishing on the 18th, Table 2 has a 'dead' bear being counted on the 22nd, nine days after the high winds ended and two weeks after the 'four' bears were seen swimming on the transects.

Questionable Identification of Dead Polar Bears From the Air:

There is also the question of how positively the 'dead' bears were identified. The following talks about the interview with the paper's co-author, Gleason.

Quote:
 
Gleason [the paper's co-author] says the airplane circled all dead polar bears to get a better look. On page 25 of the Monnett transcript, Monnett agrees, but then contradicts himself on page 29, when he says “I know some of them, we didn’t circle on. We just kept going. We, we identified them, um, you know, flying by.” Note that the paper says “Swimming and floating polar bears are difficult to see from the survey’s standard 457m altitude even under ideal conditions.”


Difficult to see a dead bear but apparently not too difficult to diagnose! On page 25 of the interview, Monnett says they might have seen another dead bear on a transect but they couldn't go back to find it. Inclusion of that that bear, in their cross multiplication, would have determined 0% survival. Later I'll discuss this stuff in more detail.

Photos of Only One Bear:

Gleason was asked why he took only (bad) photos of one 'dead bear' and no subsequent ones:

Quote:
 
ERIC MAY: So the first dead polar bear you observed, that's when you took the photos?

JEFFREY GLEASON: We're pretty sure that was the very first, yeah.

ERIC MAY: Then the subsequent one, you just didn't try?

JEFFREY GLEASON: Um-hm [yes].

JOHN MESKEL: Was this a significant thing at the time, to observe these dead polar bears?


'Dead' Bear Photo So Bad, it's Useless:

Gleason was asked why his whale photos were clear but not the bear photo:

Quote:
 
JEFFREY GLEASON: Yeah, those are just bowhead whale pictures.

ERIC MAY: Was this around the time you observed the dead polar bears, because this is pretty clear.

JEFFREY GLEASON: Yeah.

ERIC MAY: And I obviously have the dead polar bear photo. Why is this [whale photo] so much more clear than the polar bear shot that you took?


Here is the photo.

Posted Image

Manipulated Photo:

The investigators asked why the photo appeared to be manipulated:

Quote:
 
ERIC MAY: Did Mr. Monnett try to manipulate these photos to make it more clear?

JEFFREY GLEASON: Not that I'm aware of.

JOHN MESKEL: So was that you that was using that program, whatever it was you referred to, to try –

JEFFREY GLEASON: Yeah, like I said, it was just something on the computer, Paint or one of the – in the hopes of trying to make it clear. And it allows you to do some things, but I could never get it to where I thought it was worth including in the manuscript. And like I said, we took probably that file to an image processing place to see if they could enhance it.


Zooming onto the bear reveals more creativity than enhancement. That is not to say that it isn't a bear.

Gleason Asked Why There Have Been No Recurrences:

Gleason was also challenged as to why the 2004 findings were not reflected in subsequent surveys:

Quote:
 
ERIC MAY: Well, the reason I ask is because I did do some research on the sightings. And we found that through 2007, it appears there were no subsequent sightings of dead polar bears during the surveys conducted after your survey of 2004.


A Misleading Abstract and the Global-Warming Gold Card:

This is from the second interview (p71):

Quote:
 
Eric May: What were the main points you wanted the reader to understand after reading -- after reading your manuscript?

Charles Monnett: The main points were that, in a -- as part of a long-term study that had been going for 25 years at that point, that we had seen a change that was for polar bears that we had seen floating that we assumed had been drowned, and that we thought that was associated with a storm which we documented in the paper increased wind


But on page 77 May points out that the abstract made no mention of the storm. Neither did the introduction that contained the abstract. He speculated , correctly I believe, that the omission was to maximise the global warming pitch, without any distractions. That the agenda of playing the CAGW gold card was further demonstrated during Monnett's second interview:

Quote:
 
Eric May: Okay. Dr. Monnett, here's an email sent to you from Jeffrey Gleason, dated September 28th, 2004, which was written approximately eight days after your observations of the dead polar bears. Can you please read this email out loud, please. Charles Monnett: "Chuck, just got off the phone with my co-supervisor from my Ph.D. who is an Arctic ecologist and I mentioned the dead polar bears. He thought we might be onto something with a global warming angle.


Peer Review or Pal Review?:

On page 35 Monnett is asked about the peer review. Note well that he was asked, specifically if the paper was "peer-reviewed". He replied in the affirmative and then nominated the following 'peer reviewers' He indicated that the first peer review was by his wife, Lisa Rotterman. Then on page 36 he says it was reviewed by Cleve Cowles, who he said, read it thoroughly. How would he know how thoroughly Cowles read it? He also submits Paul Stang, but says that that would have been more of a political correctness review. How appropriate! He says he sent it to the head of the ICUN (International Union for Conservation of Nature), Andy Derocher. This is the person for whom Monnett is suspected of doing a favour, by way of a grant. Lastly he cites Ian Stirling, whom he describes as the all-time most famous person polar bear guy in the world. In his second interview, in August 2011, Monnettt denied (p67) that Derocher was a peer reviewer. On page 68 he says that none of his reviews (those he nominated in the first interview) had anything to do with peer review. We will later see the reason for the change of mind. Here are the names and words Monnett first offered when asked about peer review:

Quote:
 
"Lisa Rotterman, my wife, who is a, you know, Ph.D. ecologist,"

"Cleve Cowles, um, gave it a thorough read."

"And, uh, then 4 we sent it to, um – well, we sent it to Andy Derocher, who's 5 internationally – he's the, he's the, the head of the IUCN"

"Paul Stang, did, who's a manager, and I wouldn't call that a peer review. That's a, that's a political correctness review."

"Ian Stirling, who's probably the senior, like the dean, you know, the, the all-time most famous polar bear guy in the world."


Later I'll discuss these nominations.

Evidence of Corruption:

Quote:
 
The reason I read those emails, because it clearly establishes on the same day he [Derocher] writes you two emails, one regarding the $1.1 million contract, and one regarding your paper, regarding the polar bear drownings. In looking at that, it appears that the two of you created a relationship wherein you were giving him a $1.1 million contract, no strings attached, and in return he was helping you publish your observations. And such relationships perforce, create a situation wherein his credibility as an objective peer reviewer is damaged, and your credibility as an unbiased Government scientist is also damaged.


It was then that Monnett said that Derocher was not a peer reviewer, contradicting what he said during his first interview. So why, in the first interview, did Monnett nominate Derocher and other above names the peer reviewers? I believe it was because the first interview was mainly directed at the science and his scientific integrity. In fact, Monnett was told it was. He would therefore not have wanted to draw attention to the reviews he obtained from those he would later say were the real peer reviewers. Those reviews were critical of the paper. But when the second interview was querying the fact that the paper and the $1.1 million grant were being discussed by Monnett and Derocher on the same day, it became prudent for Monnett to distance the previously nominated Derocher as a peer reviewer. That's when he nominated the three anonymous reviewers as the actual peer reviewers. The second interview also established that, not only did Monnett help Derocher draft his application for a grant, he, as TPEC chair, was tasked with reviewing Derocher's proposal. The following quotes are from p56 and p44, respectively, of the second Monnett interview transcript:

Quote:
 
Richard Larrabee quoting Monnett: "I believe she is back today so things should start to move at our end shortly. Email the draft to me when you are happy with it and together we can work out any rough spots. You will get the official RFP from the contracting officer after everyone is happy with the statement of work.". . . . . . . . .

Richard Larrabee: "Okay. So, as one of your duties as TPEC chair was to review the ultimate proposal submitted by Derocher?".


So, we have the grant, via Monnett to Derocher, being discussed on the same day that Monnett's paper was being discussed. That's the polar-bear paper that was to be reviewed by Derocher. We also have Monnett helping draft the application, from Derocher, that Monnett was to then review. How is that for great teamwork?

The Official Anonymous Peer Reviews (belatedly nominated by Monnett as the real reviewers):

These reviews found serious problems in the Monnett/Gleason paper:

Eric May quoting peer reviewer number one (second interview, p82-82):

Quote:
 
"I would agree that having to swim greater distances will increase the risks to polar bears, but at least where I work on polar bears, when they come off the ice, they are quite fat and, therefore, float quite well. "In my view, the increased risk comes not so much from having to swim greater distances, per se, but from the increased chance of being exposed to high winds and wave action during a longer swimming period to reach them."


Eric May quoting peer reviewer number two. (second interview, p84-85):

Quote:
 
"The whole exercise in this section seems very dubious to me, and particularly the lack of information on distance from track line to observations of swimming/floating bears and information on the sighting probability function makes the calculations and extrapolation meaningless." [which is what I noted SMc][bold SM]


Eric May quoting peer reviewer number three (second interview, p92):

Quote:
 
I am concerned, however, that too much emphasis has been put on suggesting that the bears died because of loss of sea ice and extended open-water swimming. "The fact is that the causes of these deaths and the circumstances surrounding the mortalities are unknown. I have raised, throughout the document, serious concerns and dangers with respect to the extrapolation of limited data. It is very risky and not supported." The peer reviewer goes on to say, "Much of the introductory material describing climate change and projected loss of sea ice would seem better suited in a discussion of the idea of bears spending more time in open water due to loss of sea ice cover, and therefore, being exposed to increased risk of storms." End of quote. The last one in this area. This -- quote, "This should be revised to give it better flow. While the observations, themselves, are important, I strongly caution the authors about extrapolation and suggesting that a number of bears probably drowned. The data do not support it. [bold SM]


Monnett and Gleason and the publisher apparently chose to ignore those warnings, content to go with the wife, pal and global-warming cheer reviews.

Dead Bears Before the Storm?

Apparently, it has emerged form the second interview with Gleason, that dead bears were spotted by other researchers before the the storm. I presently don't have a transcript of that interview. If this is the case, it provides the final blow to the Monnett/Gleason paper.

My Criticisms:

Association of the Live Bear Count and the Dead Bear Count:

Table 2 in the Monnett/Gleason paper shows that before the strong winds between the 10th and 13th of September, six live swimming bears had been seen - four on transects. These sightings occurred on the 6th and 7th. Table 1, however, shows that the survey flights started on the 1st of September, with no bear-sighting entries before the 6th. That suggests that the 6th and the 7th were unusual. Monnett confirms this in the interview on page 30. This raises the question of how these two days were associated with the dead count that started after the 13th. Setting aside the inadequate sample size for such projections, just because more bears were seen swimming on the 6th and 7th, it can't be assumed there was an unusually high number swimming on the 8th and 9th or more importantly, on the 10th when the high winds and rough seas started. In his paper, Monnett says that they flew on the 8th, which was closer to the date of the high winds. The absence of the 8th from Table 2 implies that no swimming bears were seen on that flight. Table 2 has a 'dead' bear being counted on the 22nd, nine days after the high winds ended and two weeks after the 'four' bears were seen swimming on the transects. That 22nd date presents another problem. It will be next in the list.

A Contradiction Between Table 1 and Table 2:

Whereas Table 1 has the flights finishing on the 18th, Table 2 has a 'dead' bear being counted on the 22nd. Given that the 25% survival rate is based on a count of just 3 dead bears versus 4 earlier live ones, a contradiction with the count of the 3rd and last dead bear matters. Why didn't the cheer reviewers spot this?

Important Information Missing from Tables:

Neither table provides any indication of which bears were seen along the transects and which ones weren't. There is even a note to that affect with Table 1. Table 2 demonstrably has the same problem because it records four dead polar bears with a superscript "a", when we are told that three were on transects. The presentation for these data is strange given that flying transects is supposed to represent the methodology for the surveys and for Monnett's special bear survival statistics of 25%. Also not provided is the number of transects completed and/or distances covered on given dates. Furthermore, dates when no swimming bears were observed, are simply omitted in Table 2. One doesn't know if the there were zero bears spotted during the flight or there was no flight that day. Recall that for for their 25% survival conclusion, Monnett and Gleason limited their arithmetic (of profound global political significance) to sightings along the transects. One wonders how they later determined which bears were spotted during transect runs and which ones weren't. Certainly not from Tables 1 and 2. One asks, if they had the relevant data recorded, why did they omit the same from their paper?

Wife Review, Pal Review, Cheer Review and Peer Review:

Quote:
 
"Lisa Rotterman, my wife, who is a, you know, Ph.D. ecologist,"


In the paper, Monnett's wife, Rotterman, whose surname does not start with an 'M', is not indicated to be Monnett's wife. I suspect, that for most readers, that conflict of interest would remain hidden.

Quote:
 
"Cleve Cowles, um, gave it a thorough read."


Cowles is an environmentalist and was the project supervisor. For environmentalists, bad news is good news. It means work and funding. And how does Monnett know how thoroughly Cowles read the draft and what does that mean anyway?

Quote:
 
"And, uh, then 4 we sent it to, um – well, we sent it to Andy Derocher, who's internationally – he's the, he's the, the head of the IUCN"


Andy Derocher is the one to whom Monnett steered a $1.1 million grant. It was being mutually discussed at the same time that the paper was.

Quote:
 
"Paul Stang, did, who's a manager, and I wouldn't call that a peer review. That's a, that's a political correctness review."


Political correctness review? Yes, no doubt. I would suggest that Stang's wasn't the only 'political correctness review'.

Quote:
 
"Ian Stirling, who's probably the senior, like the dean, you know, the, the all-time most famous polar bear guy in the world."


Ian Stirling is a full-on global-warming-polar-bear alarmist. Monnett's paper would have been like manna from heaven.

Quote:
 
“It’s not speculation,” Stirling said in an interview Tuesday. “By the middle of the century, we’re likely to have lost two thirds of the world’s polar bears.”


But the 2009 estimate of 20,000 to 25,000 bears was the same as for 2005. There evidence is that the bear population has grown (probably due to harvesting restrictions).

Quote:
 
The U.S. Fish and Wildlife Service estimates that the polar bear population is currently at 20,000 to 25,000 bears, up from as low as 5,000-10,000 bears in the 1950s and 1960s.


As for the anonymous peer reviewers, those whom Monnett nominated in his second interview, they expressed concerns that were ignored.

Dubious Extrapolations:

I told a friend at work, a software engineer, of this study and how it was based on the spotting of 'four' live bears at one time and 'three' dead ones at another time. I told him how that was extrapolated to conclude a 25% survival rate. He replied, "You're kidding?". His response wasn't rhetorical; it was a genuine query. I recall, decades ago, being driven back from Wollongong. There were two cars ahead of us. One proceeded straight ahead, as we did; the other turned left onto a side road. The driver, who must have been getting bored after numerous late-night Wollongong-Nowra runs, extrapolated and submitted the statistics as indicating that 33% of the traffic traveling south along the highway, turns onto that side road. He was kidding - like the software engineer thought I was.What both of these anecdotes demonstrate is that a layperson's intuition is sufficient to quickly tell one that the Monnett/Gleason paper rests on terribly thin ice.

On page 25 of the first interview, Monnett says they might have seen a 'fifth' dead bear but they couldn't go back to find it. Lucky, doing so could have brought the extrapolated survival rate down to 0%. And if, additionally, the 'fourth' one had been on a transect, the survival rate would have dropped from 0% down to -25%! This, like the previous paragraph, was presented in an earlier thread, before I saw the transcript of the second interview. That interview showed the extrapolations being strongly criticised by anonymous peer review.

I will now present more quantitative criticism of the extrapolations, something I have not seen yet. Though Monnett says his transects covered 10.8% of the total area, he uses a 9:1 ratio in extrapolating. He rounds the percentage off to 11%. But for a 9 to 1 extrapolation, 11.11% is closer. To maintain the 9:1 ratio, that has the sighted 4 and 3 dead representing 36 and 27 respectively, we will use 11.11%.

A spreadsheet table was constructed with columns going from 0 to 144 bears. These represented the total number of bears swimming (S). The rows (0 to 9) in the table represented the number of bears (B) bears found in the 11.11% survey are. The cells in the table contain the probability of B bears being found, in the 11.11% area, with S bears swimming. For anyone interested, the derivation of the formula can be found in the Appendix. Part 1 is hopefully sufficient for understanding it. The spreadsheet produced peaks where expected. For example, the highest probability in the 3-bear row, occurred with 27 swimming bears. The highest probability in the 4-bear row, occurred with 36 swimming bears. The formula was also tested with a small emulation programme that used a random number generator. The results of this matched the probabilities on the spreadsheet to within the accuracy of the random number generator's bias error. I have the zipped and attached spreadsheet .xls file.

So, what is the probability that 36 bears swimming in the wider area will produce 4 in the 11.11% survey area. The probability is only 20.716%. And what is the probability that 27 bears swimming in the wider area will produce 3 in the 11.11% survey area. The probability is only 23.754%. This, I believe, is quite significant.

Now the Monnett/Gleason extrapolation assumes both sightings (4 then 3) to be representative. But, as shown, being out by one or more, is much more likely than not. But let's take it further. Assuming 36 bears, followed by 27 in the wider area, what are the chances that both surveys will see the optimum numbers (4 then 3) in the survey area? The probability is the product of 0.20716 and 0.23754, which is about 0.04921 or 5%. Okay, then what are the chances that at least one will have the optimum number in the survey area? That's 0.20716 + (1-0.20716)*0.23754, which is about 40%. So chances are that neither of the two numbers found in the survey area will be optimum. One or both being out by even 1, dramatically alters Monnett's extrapolated survival rate, which is based on 1-(3/4) or 1 in 4.

What is noticed in the spreadsheet, is that the total of any row's probabilities is the same or almost exactly the same as for other rows. In other words, if one adds up all to probabilities for 4 bears being spotted, with 1 to 144 swimming, one obtains essentially the same total as when one does the same with 0 bears and its row ( 8,999 versus 9.001 ). The tallies for 1, 2 and 3 bears are all 9.001. The remaining rows progressively fall slightly from the 9.000, but it is fairly evident, that if one extended the columns past 144, these tallies would also converge to 9.00. This tells us that, when extended sufficiently, the area under each curve is the same another. Realising this is useful because it allows us to easily work in the other direction and look at probabilities for total bears given a number seen in the survey area. We can compare the relative probability of a slice, under a curve, to the overall, or even to a slice under another curve.

It would be unreasonable to dismiss an extrapolation to 27 if it could be reliable reflect, let's say, an actual 25 to 29. To be fair, we will allow a '27 bear domain' of 23 to 31 and a '36 bear domain' of 32 to 40. So, what is the probability that 3 bears seen in the survey area, reflects the '27 bear domain' for the whole area? The probabilities between 23 and 31 are:

23.037%, 23.403%, 23.639%, 23.754%, 23.754%, 23.649%, 23.447%, 23.158%, 22.790%.

Expressed as a decimal fraction, the total is 2.106. That represents 2.106/9.000 or 23.4% of the whole curve's area. In other words, there is only a 23.4% chance that 3 bears reflects 27 bears plus or minus 4.

Now let's do the same with 4 bears seen in the survey area. The probabilities between 32 and 40 are:

20.256%, 20.489%, 20.641%, 20.716%, 20.716%, 20.646%, 20.512%, 20.317%, 20.066%.

Expressed as a decimal fraction, the total is 1.844. That represents 1.844/9.000 or 20.5% of the whole curve's area. In other words, there is only a 20.5% chance that 4 bears being spotted reflects the projected 36 bear domain (31-40).

With 4 bears spotted, the chance that it represents the projected 36 bear domain (32-40), is nearly the same as it representing 58 or more bears. It's 20.5% versus 20.1%. That's calculated from 1.844/9.000 versus 1.810/9.000.

With 3 bears spotted, the chance that it represents the projected 27 bear domain (23-31), is lower than the chance it represents 22 or less bears. It's 23.4% versus 24.8%. That's calculated from 2.106/9.000 versus 2.234/9.000.

With 4 bears spotted, the chances that it represents a number that's less than the 36 domain (32-40), within the 36 domain, and above the 36 domain, are: 27.97%, 20.488% and 51.54% respectively.

I believe that there is something else that works against the paper's extrapolations. The very fact that the conclusion were so unusual suggests the extrapolations are more likely to be wrong. For example, if you saw someone play 3 holes of golf and get an eagle, a birdie and a par, you would be foolish to bet that, on average, 1/3 of his holes would be birdies, 1/3 eagles and 1/3 pars. With the polar bears, the previous and subsequent years seem to confirm that the Monnett/Gleason extrapolations do not reflect the reality.

Conclusion to Extrapolations: One cannot project from such small data samples, as Monnett and Gleason have done. Just as two of the three 'official' peer reviewers (as belatedly nominated by Monnett) specifically warned, Monnett's extrapolations are dubious and/or meaningless. Monnett, Gleason and the magazine editor disregarded that which they should not have needed to be told. Why did they do this? I submit that 'global warming' was the reason.

The Outcome:

The letter to Monnett from the Investigation included:

Quote:
 
"...During our meeting, you were informed that the OIG asked the CO for this contract and if you had ever informed her that you assisted Dr. Derocher in preparing his proposal for the cotnract. The CO told the OIG that you never informed her you had taken such actions, and if you had informed her she would have warmed you that such actions would be highly inappropriate under procurement integrity policies and procedures. The OIG then informed you that the CO felt that the fact that you served as Chair to the TPEC for the contract made your actions especially egregious, and accordingly, the CO believed your actions warranted immediate action..."


Final Comment:

Even when there is little doubt that corruption of procedure had occurred, very often it isn't possible to make the evidence lawyer-proof. It is hard to prove intent. But Monnett got away with this (so far) only in the same way the Michael Mann got away with the 'Hockey Stick' graph. Both carry an albatross around their necks for the rest of their careers. Both are tainted and their icons of global warming are now an embarrassment. The other message that comes from this is that free rides on the global-warming gravy train are no longer assured.

The dirt and fluffiness aside, the bottom line is that the study's conclusions have been denied by the years before it and the years after it. So on that basis alone, how can the paper be 'perfectly okay' and 'fine', as has been put to me?


Sean McHugh

Appendix:

1. Let's assume 36 bears are swimming. What are the chances that there will be 4 bears in the survey area? The chance of any particular bear being in the 11.11% area is 1/9.The chance of it being outside the area is 8/9. The chance of four particular bears being in the survey area and the remaining 32 being outside it is (1/9) ^ 4 x (8/9) ^ 32. But there are 36 bears so there are many possible combinations of four being in the survey area. The number of combinations is 36 x 35 x 34 x 33. That's the same as (36 x 35 x 34 x 33 x 32 x 31 . . . all the way to 1), being divided by (32 x 31 x 30 . . . .all the way to 1). That is the same as factorial 36 being divided by factorial 32, which is represented as 36!/32!. That must be multiplied by the first calculation. But as it stands, those four bears get represented more than once. For example, if it includes 1, 7, 22, 35 it also includes 7, 1, 35, 22. There are 4 x 3 x 2 x 1 or 4! permutations for the distribution of those same numbers. As we want 4 bears in the survey area to represent one possible 4-bear outcome, we must divide the previous calculation by 4! (factorial 4). The equation now becomes: (1/9) ^ 4 x (8/9) ^ 32 x 36! / 32! / 4!. Using that we can determine, that with 36 bears in the total area, the probability of there being 4 bears in the 11.11% survey area, is 20.716%.

2. The above can be turned into a general formula, where the probability of B bears being found in a sampling area, representing A% of the whole are, with S bears swimming in the whole area : (A/100) ^ B * ((100-A)/100) ^ (S-B) * S! / (S-B)! / B!.

3. When B>S, S-B becomes negative and the factorial of that produces an error. This can be solved by changing (S-B)! to (((S-B)+ABS(S-B))/2)!. Now if B>S, the argument gets changed to 0 and its factorial becomes 1. If B<= S, then the argument remains the same. Including this, we get:

(A/100) ^ B * ((100-A)/100) ^ (S-B) * S! / (((S-B)+ABS(S-B))/2)! / B!

or:

(A/100) ^ B * ((100-A)/100) ^ (S-B)*FACT(S) / FACT(((S-B)+ABS(S-B))/2) / FACT(B)

As there can't be more bears found than are swimming, we want a zero result for this situation. This can be achieved by multiplying the whole equation by 1-ISERR(FACT(SIGN(S-B))). In the spreadsheet, SIGN(n) gives a 1 for a positive number, 0 for zero and -1 for a negative number. ISERR produces a 1 if there is an error, otherwise a zero. So, for example, ISERR(FACT(-3)) will produce 1.

Therefore multiplying the whole equation with (1-ISERR(FACT(SIGN(S-B)))) will zero it if B>S. Otherwise it will multiply it by 1 and not affect it. So now the equation becomes:

(1-ISERR(FACT(SIGN(S-B)))) * ((A/100) ^ B * ((100-A)/100) ^ (S-B) * FACT(S) / FACT(((S-B)+ABS(S-B))/2) / FACT(B))

No surprises if someone shows me a less painful way to achieve the filtering.
Attached to this post:
Attachments: BearsMain___Plot.xls.rar (29.99 KB)
Edited by Sean McHugh, Feb 12 2012, 09:53 PM.
Offline Profile Quote Post Goto Top
 
1 user reading this topic (1 Guest and 0 Anonymous)
« Previous Topic · Good and bad in science · Next Topic »
Add Reply