Vincent J. Curtis
4 Apr 2026
Ross McKittrick is a professor of Economics
at the University of Guelph, and he has become famous for his interventions in
climate science. This is possible because the mathematical methods often used
in climate science, regression analysis, are the bread-and-butter of economic
analysis; and someone well-versed in statistical analysis is well-positioned to
spot errors in the application of economic analytical methods to climate data
analysis. McKittrick is an expert in a field of poorly trained amateurs
applying mathematical methods they clearly don’t understand.
Dr. McKittrick presented a paper at the
October, 2021 ICCC conference that had been published in the August edition of Climate
Dynamics, in which he, and his co-author Richard Tol, proved invalid a
method of analysis used in climate science since 1999, and begun in a paper referred
to as ‘AT99.’
The science of attribution begins with the
propositions that; (a) we can observe patterns and changes in climate data; (b)
it can be hard to tell if there’s been a statistically significant change in
the climate; and (c) it’s hard to prove that those changes were caused by GHG
emissions. The attribution methodology
works by comparing observed patterns to patterns from climate models. This
requires that we assume that climate models provide a true and complete
representation of the climate, and that involves a statistical methodology
known as regression analysis.
The beginnings of a logical fallacy are
already evident. A model is a representation of reality, and if the model fails
to accurately forecast the behavior of reality, it is assumed here that there’s
something wrong with reality. This is a form of begging the question. A second
problem arises from the data used to construct the model: is it truly
representative, or is there a sampling problem unrecognized by the modellers? Sampling
could be a problem if not a long enough period of time was sampled for data; in
addition, there’s an implicit assumption of stasis in the climate and model. ‘Alteration/ is not a change deemed within
the principle of the climate. Attribution
tries to get around these logic problems by adjusting the parameters of the
model to fit observed reality, and if more GHG is required by the model to fit
the new data, that is grounds for attributing GHG as the cause of the deviation
of reality from the expectations of the model. There is a fallacy in this also,
but it goes unobserved by the Attributors.
Attribution was begun in 1999 with a paper
entitled, “Checking for model consistency in in optimal fingerprinting” by
Myles Allen and Simon Tett, published in the journal Climate Dynamics
(AT99). The authors claimed to have provided an unbiased and ‘optimal’
(precise) measurement of the size of effects of GHGs on climate patterns (the
‘Fingerprint’), which allowed researchers to check that the statistical models
are properly specified (the RC or residual consistency test). Theirs was a
regression method. The IPCC grew to love the method, and claims it gives
results that show unequivocally that “human activities” or “anthropogenic
emissions” were responsible for the “observed warming over the last fifty
years.”
Another logical error lies hidden in these
words. By “human activities” and “anthropogenic emissions” they mean carbon
dioxide, and hence “warming”, “climate change”, and alleged increases in
incidents of bad weather were caused by carbon dioxide. There is no effort to
correlate in detail rising concentration with rising temperature, extreme
weather events, or other effects, as one would expect in a linear regression
analysis; blaming carbon dioxide is the broad assumption.
McKittrick and Tol examined the certainty
in the IPCC-approved fingerprinting method. AT99 specifically invoked the
Gauss-Markov theorem to prove that their method of regression estimation was
unbiased and efficient, or as precise as possible. (BLUE: best linear unbiased
estimator)
McKittrick observed that AT99 made a number
of errors with respect to the Gauss-Markov theorem: the authors didn’t state it
correctly, and failed to noticer that it did not apply to their estimator!
Hence, there is no basis for confidence in the estimator that others thought
there was. AT99 results are biased, completely misleading, and largely
meaningless. McKittrick found that it was unclear what the RC test actually
tested, the test wasn’t derived in the usual way, no mathematical derivation of
the test was presented, the RC test doesn’t relate to the Gauss-Markov theorem,
and it doesn’t actually test what the authors said it tested. The RC test is a completely meaningless
statistic, said McKittrick.
The errors in the use of the Gauss-Markov
theorem jump off the page to those experienced in using it and statistical
analysis in general; and so the question arises, why didn’t anyone in the field
notice this before McKittrick? He speculates that people with a lot of
experience in statistical analysis haven’t been looking at this literature, and
hence the theoretical inaccuracies have gone hitherto undetected.
McKittrick sums up the situation with a
pair of quotes from his co-author, Richard Tol, who tweeted on Aug 21, 2021,
“Allen & Tett introduced an FGLS estimator that is completely wrong and a
test to show that they’re right, when they’re not.” And another on Aug 20,
2021, “Another literature that is entirely wrong…” There is, in short, no reason to believe that
AT99 fingerprinting method produces valid results; and a lot of the literature
attributing changes in atmospheric moisture, snow cover, forest fires, etc. are
based on a method that gives wrong and meaningless results.
Thus, the basis for claims of certainty of
around climate attribution is mathematically flawed and invalid.
Those of a philosophical bend will find
interesting a back-door introduction of Thomism into the “science of
attribution”. To say that one’s model is a true and accurate representation of
a climate is to say that the model is a true understanding the form, or
essence, of the climate. Following upon that are efficient and final causality,
or end-directedness, of the climate; the form directing the climate towards a normative
range of ends, which may be weather events. If ends other than these normative
weather events occur, these anomalies or perturbations may be caused by “human
activities” and “anthropogenic emissions”, which are nothing but efficient
causes of change to the end-directedness, to the essence, of the climate being
perturbed. The climate is expected to behave in accordance with the essence or
form understood in the model. “Alteration” is not a change deemed within the
principle of the climate, and hence, when reality doesn’t follow the model, the
cause of that anomalous behavior, or alteration, is deemed due to external
forcing, possibly by man’s emissions of carbon dioxide into the air, being the
efficient causes of that alteration. The
element of morality follows upon putting the efficient causality of ‘climate
change’ on the activities of man, even as the proximate agent of change is
carbon dioxide. Following upon that, the attributors attribute efficient
causality to carbon dioxide - for bad weather events!
Since Newton, physical sciences have
attempted to eliminate efficient and final causality, forms and essences, from
the picture, and places all the causal eggs in the matter and in mechanisms.
The science of attribution offers no mechanisms by which small increments in
carbon dioxide cause changes in large atmospheric processes, or “weather
events” (a possible violation of the principle of proportionate causality). Climate science, in attribution, adopts a
Thomist view of the physical world, including that of immorality in its
efficient causes.
From its early days, the climate “crisis”
has been a means of casting moral opprobrium upon the United States and Europe
for their cultural and economic successes, and was used as a means of
undermining a foundation of those economic successes, namely cheap and abundant
energy. Newtonian science, with its materialism and mechanisms, does not
readily lend itself to casting moral aspersions; but a Thomist view of the
world, with its recognition of formal, efficient, and final causes, as well as
material causes, enables the casting of moral blame upon its efficient causes,
namely “anthropogenic emissions” and “human activities”, or, in short, America
and Europe. Hence, the peculiar adoption of a Thomist analysis in what should
be a Newtonian science
-30-