This is Stephen McIntyre’s response to me, commenting on the
letters from Professor Keith Briffa to the Times in response to my
column on the widespread problem of withheld adverse data. It makes
very clear that my account was accurate, that my account was
mischaracterized by Professor Briffa in serious ways, and that
nothing in his letters refutes my original claim that had a key
dataset not been ignored, a very much less striking result would
have been published. Professor Briffa now says he was reprocessing
the data, but in 2009 he said “we simply did not consider these
data at this time”. Neither explanation fits the known facts
well.
I therefore stand by my story.
My original intention in mentioning this example, chosen from
many in climate science of the same phenomenon, was to draw
attention to the fact that non-publication of adverse data is not a
problem confined to the pharmaceutical industry, but also occurs in
government-funded, policy-relevant areas of academic science.
I have edited McIntyre’s text only to explain acronyms and
abbreviations.
Matt Ridley
27.1.2014
Stephen McIntyre:
Briffa’s new letter does not rebut anything that you had written
in your reply or original article. Instead, it is a pastiche of
comments that are either incoherent, not responsive to points in
your reply, untrue or highly contentious.
Review of Events
Let me first do a brief reprise of events.
Figure 1 shows five different Yamal/Yamal region chronologies
for the period 900 on, converted to Standard Deviation units.
The first three panels show versions from Briffa 2000, Briffa et al
2008 and Briffa 2009 (Climatic Research Unit (CRU) website). All
have pronounced Hockey-stick (HS) shapes. Panels four and five show
the 2006 and 2013 versions, both of which end at elevated but not
HS values.
Figure 1. Five Yamal chronologies. Top – Briffa (2000); second –
Briffa et al 2008; third – Briffa 2009 at CRU website; fourth –
2006 regional chronology mentioned in Climategate dossier and
eventually shown in Briffa et al 2013 Supplementary Material 7;
fifth – regional chronology (“Yamalia”) from Briffa et al
2013. The dotted horizontal line shows, for orientation, the
closing value of the Briffa et al 2013 regional chronology
(smoothed). The first four panels end in 1996 (denoted by a dashed
vertical red line), while the 2013 chronology ends in 2006. The
smooth for 2000, 2008 and 2009 is smooth as used in the original
articles, while the smooth for the bottom two panels is a running
11-year mean.
The Briffa (2000) chronologies for Yamal, Tornetrask and Taimyr
have been widely (almost universally) applied in subsequent
multiproxy temperature estimates. Its Yamal chronology re-processed
the Yamal measurement dataset subsequently published in Hantemirov
and Shiyatov 2002.
In 2006, Briffa and/or Osborn calculated “regional” chronologies
for the Yamal, Tornetrask and Taimyr regions. These regional
chronologies supplemented the measurement data used in Briffa 2000
with other measurement data in the region e.g. measurements taken
by Schweingruber and Vaganov not used in the 2000
calculation. These three regional chronologies were discussed
in a Climategate email (684. 1146252894.txt) on April 28, 2006,
which mentioned the Yamal/Urals regional chronology as follows:
“URALS” (which includes the
Yamal and Polar Urals long chronologies, plus other shorter
ones).
The chronology in this email is shown in the fourth panel above.
It obviously lacks the pronounced HS of the 2000 version; in
addition, its medieval values are somewhat greater than modern
values.
In another Climategate email (780. 1172776463.txt) the following
year (March 2007), Osborn discussed with Briffa the
differences in versions of the Yamal/Urals regional chronology that
were in a CRU presentation that Osborn had sent to Briffa.
Six months after this email (November 2007), Briffa and
coauthors published an article (Briffa et al 2008) purporting to
provide “regional chronologies” for the three northern Eurasian
regions previously considered in Briffa 2000. In the other
two regions (Tornetrask, Taymir), Briffa et al had dramatically
expanded the data sets by incorporating, for example, Schweingruber
datasets in the Taymir region. However, despite the
implication of Briffa et al 2008 that they had used expanded
datasets, the Yamal dataset of Briffa et al 2008 was the same as
the dataset of Briffa 2000, even though the 2000 Yamal chronology
had a much smaller number (12) of samples in the modern period
relative to the other 2000 chronologies, let alone their 2008
expansions. This chronology is shown in the second panel.
In September 2009, after obtaining measurement data for the
three regions as used by Briffa, I observed that Briffa et al 2008
had incorporated Schweingruber datasets into the 2008 Taimyr
regional chronology, that there were seemingly equivalent
Schweingruber datasets in the Yamal region (noting Khadyta River
KHAD as an example) and that inclusion of such Schweingruber data
would attenuate the very pronounced HS of the Briffa 2008 Yamal
chronology. (BTW this observation has been supported by the 2013
regional chronology.) It seemed highly implausible that Briffa and
associates wouldn’t have done a regional chronology calculation for
the Yamal/Urals region. I speculated that Briffa and
associates must have done a regional chronology calculation, also
obtaining highly attenuated non-HS modern values and, for some
reason not disclosed in the article itself, not chosen not to
report it.
This surmise, which later proved to be true, was sharply rebuked
at the time. In an online article at the CRU website in
October 2009, Briffa conceded that the KHAD site met their criteria
for inclusion in a regional chronology, but claimed that they
“simply did not consider” the data at the time.
Judged according to this criterion
it is entirely appropriate to include the data from the KHAD site
(used in McIntyre’s sensitivity test) when constructing a regional
chronology for the area. However, we simply did not consider
these data at the time, focussing only on the data used in the
companion study by Hantemirov and Shiyatov and supplied to us by
them. We
would never select or manipulate data in order to arrive at
some preconceived or regionally unrepresentative result.
At the time, the Climategate emails were not available and,
while this assertion seemed implausible, neither could it be
contradicted with then available information. However, the
following month (November 2009), Climategate emails became
available. Yamal appears to have been of particular interest
to the hacker/leaker of the Climategate emails based both on the
selection of documents and emails and the limited information on
access times.
Yamal was mentioned in one of the questions in the Muir Russell
Issues Paper as follows. (In passing, it’s interesting to note
that, while the Muir Russell report has been derided on many
counts, its neglect of issues raised in the Issues Paper was
under-discussed.)
Have you been
selective in utilizing tree ring evidence from Yamal in Siberia;
and if so, what is the justification for selectivity and does the
selection influence the deduced pattern of hemispheric climate
change during the last millennium?
The reference in the March 2006 Climategate email to a regional
chronology incorporating Yamal, Polar Urals and “other shorter”
datasets was quickly noticed, as it obviously appeared to
contradict CRU’s October 2009 response. In my
submission to the Muir Russell review, I specifically drew
attention to this email as an example of cherrypicking the Yamal
chronology over a “still unavailable combined chronology attested
in Climategate Letter 1146252894.txt.”
In their submission to the Muir Russell panel in February 2010,
CRU did not address the unpublished regional chronology. They
stated to Muir Russell that the purpose of both Briffa 2000 and
Briffa et al 2008 was to “reprocess” the Hantemirov and Shiyatov
dataset and that they “made no selection of what data to
include”. They incorporated their October 2009 website
article in their submission – an article, which, as noted above,
said that “simply did not consider” the inclusion of Schweingruber
data into a regional chronology at the time. While the
purpose of Briffa 2000 included reprocessing the Hantemirov and
Shiyatov dataset using RCS standardization, this was not the stated
purpose of Briffa et al 2008, which, instead, purported to present
regional chronologies. “Reprocessing” was nowhere mentioned in the
article. Nor is it correct to say that Briffa et al 2008
“made no selection of what data to include”. They decided against
using the expanded regional data used in the 2006 regional
chronology. Even if that were a justifiable decision (as CRU
later argued), it was still a decision about what data to include
and, to that extent, their submission to Muir Russell was
misleading.
As is well known, Muir Russell himself did not even bother
attending the one interview of CRU personnel on Hockey Stick
matters (which was conducted by Geoffrey Boulton.) Following
this interview, Boulton asked CRU to comment on McKitrick’s October
2009 op ed about Yamal – an article published prior to Climategate
and which therefore did not refer to or discuss the 2006 regional
chronology. The panel did not ask CRU to comment on the
regional chronology. In their response to Muir Russell, CRU
re-iterated their implausible claim that the “purpose” of Briffa et
al 2008 was merely to “reprocess” the original Hantemirov dataset.
They also inconsistently said that they had considered the
incorporation of more data (presumably including Khadyta River) and
indeed even “intended to explore an integrated Polar Urals/Yamal
larch series”, but had “felt that this work could not be completed
in time”. Needless to say, this incompleteness had not been
disclosed to editors or readers of Briffa et al 2008. CRU
conspicuously did not disclose to the Muir Russell panel that they
had previously calculated a Yamal/Urals regional chronology or
discuss its supposed defects.
It’s too bad that Muir Russell neglected this and other issues
in their report. A more competent panel would have settled
some of these issues.
Since the regional chronology was not addressed by Muir Russell,
I submitted an FOI request both for the chronology and for a list
of the sites. In their response, University of East Anglia
(UEA) confirmed the existence of the chronology, but refused to
release it or the list of sites.
I appealed to the Information Commissioner. The
Information Commissioner required UEA to disclose the list of sites
immediately. In negotiations between the Information
Commissioner and UEA, UEA undertook to publish the requested
regional chronology within six months, an undertaking which was a
major improvement, but which was still (in my opinion) a delaying
tactic. The Information Commissioner accepted UEA’s undertaking but
noted my concern, stating that failure on the part of the
University to live up to its undertaking would open up the
possibility of a different position for a fresh FOI request:
The complainant has expressed doubts
as to whether the University really intends to publish the
information by the date specified and believes this to be a
delaying tactic on the University’s part. The Commissioner is not
aware of any evidence to support such a contention, but given the
written assurances which have been received from the University as
to the publication date, he considers that any delay beyond October
2012 will need to be reasonably explained by the University if the
withheld information is to remain exempt from disclosure by virtue
of regulation 12(4)(d), if a further request was made.
Having no confidence in UEA’s undertaking, I appealed the
Information Commissioner’s decision. As matters turned out, UEA did
not make the slightest attempt to live up to their undertaking.
They submitted an article to Nature without the requested regional
chronology. This submission was rejected. So six months later, they
had not even begun to comply with their undertaking to the
Information Commissioner. The appeal at the Information Tribunal
proceeded, with exchanges and submissions to the Information
Tribunal becoming increasingly acrimonious.
Tim Osborn stated that to the Information Tribunal that release
of the regional chronology (shown in the fourth panel above) would
have “adverse reputational consequences” to Briffa and CRU and lead
to criticism that would “damage the reputation of individual CRU
scientists as well as CRU’s reputation as a leading centre of
excellence in the field of climate change research”:
49. Looking at the situation more
narrowly, there would also inevitably be adverse reputational
consequences for the individual scientists involved in this work
and the University itself if disclosure had been effected. This is
because biases in the 2006 chronology, which in CRU’s view limit
its value as evidence of past temperature changes, would doubtless
be seized upon by climate change sceptics as demonstrating that
there were fundamental failings in CRU’s approach to the science of
climate change. Whilst such charges would be entirely unfair in all
the circumstances, they would serve to damage the reputation of
individual CRU scientists as well as CRU’s reputation as a leading
centre of excellence in the field of climate change research.
Meanwhile, CRU had finally commenced preparation of the article
previously promised to the Information Commissioner. The new
article (Briffa et al 2013) presented a new regional chronology
incorporating Yamal and Polar Urals, but not Khadyta River.
This is shown in the bottom panel. In its Supplementary Material 7,
they presented the 2006 regional chronology (fourth panel) together
with a litany of supposed defects. In April 2013, the
Information Tribunal rejected my appeal. While it seemed to me that
they erred in their decision, the issue became moot with the
publication of Briffa et al 2013 and I did not pursue it
further.
Returning to Figure 1, it is obvious that the modern portion of
the 2013 regional chronology is dramatically attenuated relative to
the pronounced HS of the 2000, 2008 and 2009 chronologies and is
much more comparable to the despised and unreported 2006 regional
chronology. The main difference between the 2006 and 2013 regional
chronologies is the shaving of medieval values through a new policy
requiring the exclusion of radially asymmetric root collar samples
from Polar Urals. As I observed at Climate Audit at the time,
strip bark bristlecones have far more dramatic radial asymmetry and
it is this radial asymmetry that causes the extreme Hockey Stick
shape of the Graybill bristlecone chronologies.
Consistent adoption of Briffa’s new policy would require the
rejection of the many reconstructions using strip bark
bristlecones. Otherwise, one simply moves to a new form of
cherrypicking by IPCC paleoclimatologists: accepting radial
asymmetry when it contributes to a HS and rejecting it when it
doesn’t.
Briffa’s New Letter
As noted above, Briffa’s second letter is a pastiche of
incoherency, irrelevancy and disinformation.
First Paragraph Is Incoherent
In his first paragraph, Briffa challenged the following sentence
in Ridley’s reply letter:
Professor Keith Briffa says that he
was “reprocessing” a data set rather than ignoring it because it
gave less of an uptick in temperatures in later decades than the
embarrassingly small sample of Siberian larch trees he
published.
As noted above, Briffa had stated (for example to Muir Russell)
that the “purpose” of Briffa et al 2008 was to “reprocess” the
small Hantemirov and Shiyatov 2002 Yamal dataset and denied that
this decision had anything to do with the adverse (non-HS) results
from their 2006 regional chronology calculations. Although
Briffa objected strongly to the above sentence, I, for one, do not
see any relevant difference between Ridley’s characterization and
Briffa’s new declaration:
This reprocessing was not motivated
by consideration of any “uptick in
temperatures”.
Here Briffa’s animosity has descended to incoherence.
Second Paragraph is Irrelevant
Briffa’s next two sentences are irrelevant to any issue raised
in either Ridley’s original opinion or reply.
Ridley had not suggested that the Yamal chronology was
“dependent” on the Mann bristlecones (or vice versa). Thus Briffa’s
following strident declaration that the two series are
“independent” is simply irrelevant to any actual issue:
in his Opinion Piece he describes my
publication of this version of the Yamal chronology as a
“relaunch” of the hockey-stick graph of Northern
Hemisphere average temperatures. My work was independent of the
so-called “hockey-stick” graph and I and my colleagues have long
ago demonstrated that the conclusions drawn in that work are not
dependant on the inclusion of my Yamal chronology.
As a nit, the original Yamal chronology was published (2000)
some years prior to my entry into the field (2003-2005) and was
therefore not a “relaunch” of the Hockey Stick in response to
our criticisms. More accurately, it was Briffa’s
entry into the Hockey Stick market, as, up to this point, his
primary published reconstructions (Briffa et al 1998; Briffa et al
2001) had marked post-1960 declines, well known through “hide the
decline”.
The Withheld Regional Chronology
Briffa then made a series of highly inaccurate and/or misleading
statements about the unpublished and adverse 2006 regional
chronology as follows:
Ridley then persists in the
repeated claim that a “larger tree-ring chronology from the
same region did not have a hockey stick shape”. Leaving to
one side the questions of what constitutes a “larger
chronology” and what does or does not represent a
“hockey-stick shape”, this statement implies that a
chronology based on more tree-ring data from this region would
invalidate the conclusions from our published temperature
reconstructions. He also insinuates that just such an “adverse”
chronology had been concealed by us and would not have come to
light without a Freedom of Information (FOI) request. He is wrong
again on both counts. This presumably refers to an FOI request made
to the University of East Anglia for a chronology whose existence
was revealed as a result of the theft of emails from the Climatic
Research Unit. Both the Information Commissioners Office and the
Information Tribunal (appeal number EA/2012/0156) rejected this
request, accepting our explanation that this chronology was
produced as part of ongoing research intended for publication.
This chronology was indeed
subsequently published, but as a demonstration of how inappropriate
statistical processing, allied with a failure to recognise and
account for inhomogeneities in the underlying measurement sets, can
produce what is an unreliable indication of regional tree growth
and inferred summer temperature changes in this area. Ridley’s
description of this chronology as a withheld “adverse” result is,
therefore, unjustified.
First, CRU did not disclose the existence of the 2006 regional
chronology or efforts to develop a Yamal/Urals regional chronology
in Briffa et al 2008, their October 2009 website article or in
their submissions to Muir Russell. Does this imply that CRU
“concealed” the existence of the chronology? In this case,
“conceal” is Briffa’s word, not Ridley’s. Ridley’s letter as
submitted used the perhaps more neutral phrase “failed to report” –
a claim that is true.
Second, contrary to Briffa’s assertion, nothing in the
Information Commissioner’s decisions contradicts a view that the
2006 regional chronology would not have come to light without the
FOI requests. In my opinion, while CRU may well have
re-opened the file on Yamal as at the time of my FOI request, I do
not believe that they then had the faintest intention of including
the 2006 regional chronology in any putative publication. I
can’t prove this, but it’s what I think. The Information
Commissioner appears to me to have taken a fairly firm position
with UEA during negotiations. He required UEA to release the list
of sites used in the regional chronology, though this was done
during negotiations rather than in a decision. This was
an important victory for me, as it enabled me to do my own estimate
of the regional chronology – an estimate that was virtually
identical to the then withheld chronology. It also seems to
me that the Information Commissioner took a relatively practical
position on the chronology itself. UEA said that they were
working on the data and undertook to release the data as part of a
publication within six months. The Commissioner accepted this
undertaking, but warned UEA that he would take a different position
on a fresh request if UEA failed to live up to their
undertaking. At no point did the Commissioner opine on
whether the regional chronology would have been disclosed without
the FOI request.
Third, Briffa took issue with the characterization of the
unpublished 2006 regional chronology as an “adverse” result.
But Osborn himself stated that release of this chronology would
have “adverse” reputational consequences for Briffa and other CRU
scientists and would “damage” the reputation of CRU. So, by
their own admission, CRU believed these results to be
“adverse”.
Finally, Briffa et al 2013 (Supplementary Material 7) did indeed
contain a tirade against the 2006 chronology. I entirely agree that
“failure to recognise and account for inhomogeneities in the
underlying measurement sets” is a pernicious problem in the
regional chronology methodology proposed in Briffa et al 2008. All
the more reason why the supposed failure of this methodology on the
Yamal/Urals dataset should have been reported in the earlier
article.
Ridley’s description of the regional chronology as both
“withheld” and “adverse” is completely justified.
“Validated” by Muir Russell
Briffa’s final issue is little more than a cavil. Briffa’s
first letter had stated:
The accusation of “cherry-picked
publication” was investigated by the Independent Climate Change
Email Review, which concluded in 2010 that our “rigour and honesty
as scientists are not in doubt”.
Ridley’s reply, as submitted, stated:
Briffa further claims that
his research was validated by the inquiry chaired by Sir Muir
Russell. Yet Sir Muir did not even attend the only interview
with academics at the University of East Anglia on the Hockey
stick. Nor did the panel interview critics of the UEA group.
Nor did the Muir Russell panel even ask Briffa and Jones about
their destruction of documents to evade FOI requests. The inquiry
did not explore, let alone endorse, the specific data sets in
question.
This was shortened by the editors to:
Briffa claims that his research
was validated by the inquiry chaired by Sir Muir Russell, but that
inquiry did not explore, let alone endorse, the specific data sets
in question.
Briffa now complains that there is a relevant distinction
between saying that the accusation of cherry-picked publication
being investigated by Muir Russell and saying that their research
was “validated” by Muir Russell:
Ridley again misquotes me as
saying “my research was validated by the inquiry chaired by
Sir Muir Russell.” I clearly said no such thing. The
Independent Climate Change Email Review had no remit to
“validate” any research. In my opinion this can be
done only through reinforcement by consistent results produced in
repeated, continuing research and published in the peer-review
literature. What I actually said was that I had not “cherry picked”
my data to produce a desired result, which was the specific
accusation levelled at me in his Opinion Piece. Sir Muir Russell’s
team examined this specific accusation and found that I had
not.
The distinction is really immaterial. Ridley’s letter
could have been rephrased as follows without changing the
point.
Briffa further claims that
allegations of cherry-picking had been settled by the
investigations of the Muir Russell panel. This is not the case.
Sir Muir did not even attend the only interview with academics
at the University of East Anglia on the Hockey stick. Nor did
the panel interview critics of the UEA group. Nor did the Muir
Russell panel obtain the contested regional chronology from Briffa
and/or associates or carry out any investigation of the
circumstances of the contested chronology. Nor did the Muir Russell
panel even ask Briffa and Jones about their destruction of
documents to evade FOI requests. The inquiry did not explore, let
alone endorse, the specific data sets in question.
Conclusion
The issue in the original Opinion Piece was the failure to
report adverse results, with the lugubrious story of Briffa’s
unpublished and unreported 2006 regional Yamal/Urals chronology
being an example. Nothing in Briffa’s letter refutes Ridley’s
original claim.
In retrospect, when one compares the very attenuated blade of
the 2013 regional chronology with the similarly attenuated blade of
the despised 2006 regional chronology, all the past excuses
for withholding the 2006 chronology and all the past attempts to
sustain the superblade of the 2000, 2008 and 2009 chronologies ring
increasingly hollow.