Damned Lies and Statistics: 'Climate Change'

“There are three kinds of lies: lies, damned lies and statistics,” a quote which Mark Twain in his Autobiography attributed to Benjamin Disraeli—though it more likely derives from the obiter dicta of the First Earl of Balfour. We all know—or should know—that statistics can be deceptive. Like language itself, they serve a dual function: to tell the truth and to lie—except that, unlike ordinary language, statistical contrivances appear to share the property of pure mathematics, that is, they seem objective, factual, impartial, and irrefutable. People are easily convinced, writes Darrell Huff in How to Lie with Statistics, by a “spurious air of scientific precision.”

The only way to disarm plausible but specious statistical accounts is to dig down into the source data or, when feasible, simply to use one’s common sense. Of course, statistics can be woven out of whole cloth, total fabrications which are easily rumbled with a modicum of attention, but it is their subtlety, their playing with half-truths, that can be most persuasive and damaging. Telling half the truth can be more insidious than a manifest falsehood.

Stars and shadows ain't good to see by.

Global Warming statistics are among the most readily manipulable, delivering factoids that are true and yet false—in other words, in other words. The tactic is to present a lesser truth that disguises a greater one. For brevity’s sake, let’s take just a few examples of how “climate change” statistics can rank among the most effective means of producing assent to outright mendacities, coating whoppers with honey.

Consider the twaddle that came out of the University of Illinois’ 2009 survey that 97.4 percent of scientists agree that mankind is responsible for global warming, a finding which is easily debunked when one accounts for the selection methodology.

As Lawrence Solomon explains in a crushing putdown, the Illinois researchers decided that of the 10,257 respondents, the 10,180 who demurred from the so-called consensus “weren’t qualified to comment on the issue because they were merely solar scientists, space scientists, cosmologists, physicists, meteorologists, astronomers and the like.” Of the remaining 77 scientists whose votes were counted, 75 agreed with the proposition that mankind was causing catastrophic changes in the climate. And, since 75 is 97.4 percent of 77, overwhelming consensus was demonstrated.

The real percentage, however, of concurring scientists in the original survey is a paltry .73 percent. That the chosen 75 were, as Solomon writes, “scientists of unknown qualifications” adds yet another layer to the boondoggle. This sort of thing is not a little white lie or an inadvertent statistical error. Once it reaches the point where a deliberate misconstrual must be maintained by the omission of details, the distortion of data and the suspicious liability to intentional error, we are in the presence of the great statistical charade as it is practiced by our accredited “experts.”

Not to be outdone, the Climate Research Unit (CRU) at the University of East Anglia developed a graph showing the trend to global warming, but neglected to note that it is calibrated in tenths of degrees rather than whole degrees, giving the misleading impression that the world is heating up when there is, in effect, little to no global warming to speak of. Similarly, the British climate journal The Register points out that NASA data have been “consistently adjusted towards a bias of greater warming. The years prior to the 1970s have again been adjusted to lower temperatures, and recent years have been adjusted towards higher temperatures.” Moreover, NASA data sets, as is so often the case, were predicated on omission, so-called “lost continents” where temperature readings were colder than the desired result.

Eureka! It's alive! 

As The Register writes, “The vast majority of the earth had normal temperatures or below. Given that NASA has lost track of a number of large cold regions, it is understandable that their averages are on the high side.” Additionally, NASA reports their global temperature measurements “within one one-hundredth of a degree. This is a classic mathematics error, since they have no data from 20 percent of the earth's land area. The reported precision is much greater than the error bar.”

The problem, warns Joel Best in Damned Lies and Statistics, is that “bad statistics live on; they take on a life of their own.” Their longevity supports their putative truthfulness. And the public is gullible, prey to the baked-in lies that Best calls “mutant statistics,” no matter how implausible.

Similarly, Tim Harford in The Data Detective, a celebration of good and useful statistical models, refers to the tendency toward motivated reasoning, i.e., “thinking through a topic with the aim, conscious or otherwise, of reaching a particular kind of conclusion.” Obviously, such thinking can work both ways, disparaging reliable statistics as well as valorizing dubious ones. The whole point, of course, is obfuscation, to keep people in the dark. Our soi-disant climatologists could just as well have written that climate is defined by a statistical curve in relation to a congruence subgroup of a modular elliptic, and the effect would have been the same. Whatever it means, it sounds official and incontrovertible.

In his essay, “March of the Zealots,” John Brignell comments on such acts of dissimulation. “If the general public ever got to know of the scandals surrounding the collection and processing of data [about global warming]… the whole movement would be dead in the water… It is a tenuous hypothesis supported by ill-founded computer models and data from botched measurement, dubiously processed.”

Examples of data manipulation abound. For more thorough analyses, see Michael Shellenberger’s Apocalypse Never, Steven Koonon’s Unsettled, Tim Balls’ The Deliberate Corruption of Climate Science, and Rupert Darwall’s Green Tyranny, all of which are eye-openers. As Stanford professor Dr. John Ioannidis writes in a much-circulated paper provocatively titled Why Most Published Research Findings Are False, “There is increasing concern that in modern research, false findings may be the majority or even the vast majority of published research claims. However, this should not be surprising.”

Flawed statistical analyses have become the established currency of the climate economy.

Lies, Damned Lies, and the Media

As the saying goes, there are lies, damned lies, and statistics. There’s also a corollary: properly used, statistics don’t lie. But when selectively abused, statistics are meaningless.  The kerfuffle that followed President Trump’s interview with Jonathan Swan which aired on HBO earlier this week is yet another example of the phenomenon.

Actress Julia Louis-Dreyfuss was among those who weighed in on the interview. Dutifully following the “orange-man bad” narrative, a Dreyfuss Tweet seemed to imply a belief that Swan had a masterful command of meaningful pandemic statistics, while President Trump was basically clueless:

What made the president a fool and Swan a genius? Trump highlighted the statistical fact that the United States has been more effective in curing, aka reducing the death rate, among Americans who are diagnosed with COVID-19 than most of the rest of the world. This is clearly a testament to the effectiveness of our overall health care system in treating infectious and potentially fatal diseases.

Swan highlighted the statistical fact that more Americans have died of COVID-19 exposure per capita than have died as a percentage of population when compared to nations like Germany and South Korea. Though he didn’t directly say so, Swan clearly implied that this statistic was far more important than the statistic President Trump had mentioned.

Trump disagreed with Swan’s analysis: “You can’t do that,” he said.

“Why can’t I do that?” Swan responded, rudely.

At this point, neither party to this discussion displayed any sort of expertise about how to properly interpret statistics. Trump was stumbling, but so would every other President at this level of detail, going back to at least Eisenhower. American presidents are not masters of detail. Moreover, can anyone honestly believe that Joe Biden could get to that part of so nuanced of a discussion without his head exploding or threatening to punch somebody?

I believe the point Trump was attempting to make was that it is unsound scientifically to use the per capita death rate as the metric with which to judge the effectiveness of the administration’s response to the pandemic. If that is indeed the correct interpretation of “you can’t do that,” then the President’s point is valid.

If the death rate per person infected is relatively low, but the death rate per capita is higher, then the infection rate is the driver. Consider an example: Both Group A and Group B consist of one million individuals each, each demographically similar to the other. In Group A 100,000 get infected, while 20,000 of the infected sub-group die. The mortality rate per capita is 2%, while the mortality rate per infection is 20%. In Group B 50,000 people get infected, while 15,000 of the infected die. The mortality rate per capita is 1.5% and the mortality rate per infection is 30%. Infections are more prevalent in Group A, but treatment of the infection is much better in Group A than in Group B.

Or, let’s look at the following real-world analogy. In many developing countries the motor vehicle fatality rate per capita is far lower than it is in the United States. Does that mean it’s safer to drive in those nations? No, it means they have fewer cars. When you look at a meaningful statistic – deaths per motor vehicle – the fatality rate in most of the very same developing countries far exceeds that of the United States. As anyone who's ever driven in the Third World knows.

Per capita statistics are thus rarely useful analytical tools when considered in a vacuum. One must understand the underlying causes and how those causes may or may not be influenced before citing a per capita stat. In the case of COVID-19 there are at least two important underlying variables that should factor into any analysis: infection rate and treatment effectiveness.

Clearly, infection rates vary by state because the individual states have been driving different isolation and protection policies at varying speeds and implementing different “get back to normal” recovery programs as well. If Swan believes that the Administration could have and should have done something to implement a national isolation policy and national recovery policy, he should have said so.

Could the Trump administration have done something like that? I don’t see how. The states would scream bloody murder if he tried to interfere with them. The President can’t even get blue states to disperse riotous mobs occupying the streets of major American cities. Any attempt by this administration to impose rigid standards involving public gatherings and personal interactions would have been denounced as a violation of federalism and widely ignored.

It’s clear that stemming the spread of COVID-19 is about isolation and protective gear. The highest rate of new infections is now among the 20-29 year old demographic, many of whom ignore such restrictions. That’s understandable. They are at relatively low risk of dying even if they do catch it, and most of us who remember our twenties will recall that following rules – even rules meant to protect you – are not a high priority at that time of life. But this development emphasizes the simple fact that the infection rate part of the per capita mortality rate equation is about personal behavior, not national policy.

Among the parts of the equation that the administration could and did address was providing care for the sick and protection for health care workers. From getting Ford to produce ventilators, to ensuring there was an equitable distribution of face masks among the states in the early days of the pandemic, the Trump administration focused on those things it could do to facilitate research, to ensure that health care facilities were not overwhelmed, and to save as many lives of the infected as possible. Certainly the states and numerous organizations both public and private played a huge role in the success of that effort, but it’s petty partisanship at its worst to pretend that the president’s actions were unimportant or somehow misguided.

Sadly, Jonathan Swan’s abuse of statistics is business as usual for the legacy media these days. He focused on a statistic over which the Trump had no practical control, presumably because it made the president look bad, while ignoring the stat that demonstrated how effective the administration has been in helping to address those parts of the pandemic it actually could influence.