How Bias Is Ruining Scientific Research If it bleeds (or has a miracle result), it leads. by Michael Fumento
https://spectator.org/science-progress-publishing-journals/
“We must choose between science and science fiction.”
When you hear “Science says … ,” there’s no such thing. It’s bullying rhetoric, like “Everyone knows … ” or “It’s common wisdom that … ” Sometimes we can turn to authorities like the CDC or the WHO, but they are hardly omnipotent nor without political agendas. At one time they were at least essentially apolitical, but no more. Likewise for smaller institutions like colleges or universities, or for individuals. Of late, Anthony Fauci has come to mind.
Fact is, there’s strong evidence the progress of science is slowing — masked by relatively few high-profile advances (such as cell phones) and click-bait journalism with “breakthroughs” that never materialize. And there’s evidence of a decline in the quality of the scientific literature, even as it’s accompanied by an explosion in quantity. Publication seems to be doubling every nine years. Yet as I’ve noted previously here, the Alzheimer’s arsenal is effectively empty. Cancer progress has been achingly slow — the “War on Cancer” goal was to cure all types by 1976. And now over 50 years since Apollo 11, we actually don’t have the ability to return people to the moon.
The standard in observable advances is supposed to be published, peer-reviewed literature. But never mind that sometimes, as with COVID-19, it can take a long time for that to appear. Further, a recent study indicates that there may be a bias that makes publications in prestigious journals less reliable precisely because of what it takes to get in one, plus the desire of those publications to print the most cutting-edge or simply exciting material.
We have long known that the vast majority of published studies are never replicated because of such culprits as fraud, bias, negligence, and hype. In 2005, Stanford Professor John Ioannidis published “Why Most Published Research Findings Are False,” the most downloaded paper ever in the Public Library of Science, better known as PLOS. Not incidentally, he has also become notorious for his minority views on COVID, including criticisms of case prediction models and the efficacy of lockdowns. Insisting on empiricism can make you very unpopular.
But a recent study from the University of California San Diego’s Rady School of Management published in Science Advances would seem to emphasize the hype aspect. Just as “if it bleeds it leads” in the popular media, you’ll probably get a lot more attention if you say something cures cancer than if you say it’s a dead end — never mind that reaching and publishing dead ends is a vital part of science. Or if you write that the Earth is doomed versus that it’s not doomed, or if you claim you’ve discovered a new electric vehicle battery that will wipe out internal combustion engines in five years. People like science most when it appears to be magic, whether white or black.
We know “that experts can predict well which papers will be replicated,” wrote the authors Marta Serra-Garcia, assistant professor of economics and strategy at the Rady School and Uri Gneezy, professor of behavioral economics at the same institution. “Given this prediction, we ask ‘why are non-replicable papers accepted for publication in the first place?’”
First, Serra-Garcia and Gneezy looked at data from three influential replication projects that tried to systematically replicate the findings in top psychology, economic, and what are considered the two top general science journals, Nature and Science. Whether you know it or not, a huge amount of what you read or hear regarding science and medicine comes from just those two outlets. In psychology, only 39 percent of the 100 experiments successfully replicated. In economics, 61 percent of the 18 studies replicated, as did 62 percent of the 21 studies published in Nature and Science.
Then the researchers used Google Scholar to test whether or not papers that failed to replicate are cited significantly more often than those that were successfully replicated, both before and after the replication projects were published. By far the worst offenders were (Drumroll please!) Nature and Science. They cited non-replicable papers 300 times more than replicable ones.
When the authors looked for explanations such as the number of authors, the rate of male authors versus female, the details of the experiment (such as location, language, and online implementation), the relationship between replicability and citations was unchanged.
Are there efforts to acknowledge this? Pretty much… nope. “Remarkably, only 12 percent of post-replication citations of non-replicable findings acknowledge the replication failure,” the authors wrote. Once it’s out there, it stays out there. It becomes an “everybody knows” situation, and if you claim otherwise you may be labeled a “denialist” and sent to a ring of Hell that Dante overlooked.
How does that matter to you? Much more than you can imagine.
Perhaps the most damaging non-replicable paper in history was a 1983 article in JAMA by an unknown epidemiologist claiming that AIDS appeared to be transmissible through casual contact. Even before the discovery of the causative virus HIV, it was obviously nonsense. We knew at least a year earlier it was being spread through blood and semen and indeed the first name for AIDS was “Gay Related Immune Deficiency” (GRID). But the JAMA paper served the political purpose of reducing the stigma of homosexual males and intravenous drug abusers. The sole author was rewarded with appointment to America’s top infectious disease position, which decades later he would use as a bully pulpit to wreak absolute havoc during the COVID-19 crisis. Yes, his name was Anthony Fauci.
Later, the crafty Fauci consolidated his power in great part through massive citation padding, as I’ve detailed elsewhere. Basically, it appears he makes a deal with researchers to put his name on papers (often as either the writer or the lead researcher) with the understanding that the Fauci name will bump the paper up the list of prestige publications. Even when he was chief of a government agency that should have absorbed his attention full time, he was apparently spending 24/7 in various labs regarding an incredible array of subjects. Call it “plagiarism by consent.”
(In case you were wondering, the infamous projections from Neil Ferguson and Imperial College that more than anything else led to the lockdown fever that yet grips us was never published in a journal at all, merely on the Imperial College Web site. It was deemed too important and urgent to await publication. By Ferguson, that is. Still, he most certainly could have gotten it published in a prestigious peer-reviewed publication.)
In the long run, however, hysteria over “global climate change” or GCC will ultimately probably cause more damage than that over COVID. Perhaps you’ve heard that about “97% of published papers say that GCC is real and to at least an extent man-made” or similar wording. In any event, it’s wrong and you can read in detail why here.
But to the extent there’s truth to this claim, it represents publication bias. If you don’t come to the “correct” conclusion, Nature and Science just aren’t interested. Nature even has a whole spinoff entitled Nature Climate Change. (Actually, lots of science publications have GCC spin-offs.) So they need a constant influx of new papers on every last aspect of the alleged phenomenon. Thus in May, Nature Climate Change published a paper concluding that “Across all study countries, we find that 37.0% (range 20.5–76.3%) of warm-season heat-related deaths” can be attributed to GCC. Wow! “This is the worst of the worst catastrophes in the world!” to quote the WLS reporter at the Hindenburg crash.
Actually, the Hindenburg crash killed only 36. And take heed of that heat death article as well. The lead researcher Antonia Gasparrini is from London’s Center on Climate Change and Planetary Health. So GCC fear pays his salary in what’s presumably a cushy job. But if you want to keep such positions, you still have a lot of competition in the publish or perish race, with citations both directly and indirectly counting for getting grants. The higher up the publishing ladder, the better. You can tell just by the name that Nature Climate Change is pretty high up in that more prestigious publications tend to have shorter names, even to a point of going to acronyms to gain prestige as with The Journal of the American Medical Association becoming JAMA.
Sure enough, popular media such as France 24 picked up on the paper with “Global Warming Blamed For 1 In 3 Heat-Related Deaths,” as did popular science media like NewScience “Climate Change To Blame For 37% of World’s Heat-Related Deaths,” and so did lots of medical and science journals. Well, not precisely that article necessarily. You see, the paper’s author, Antonio Gasparrini, has turned this theme into a printing mill. A search of the PubMed database of medical and science journals finds that since 2002 he’s co-authored 41 papers from this basic data set, albeit perhaps with some updating. Indeed, a lot of citations for this latest article come from his own earlier articles.
Mind, Greens like recycling, so apparently they would have no problem with that. But we should. The same stuff just gets repackaged, and if there are any criticisms, at best they will be published in the letters to the editor. And who reads those?
Replications of Gasparrini’s findings? Sure, plenty! But apparently all from Gasparrini. Yet to understand that GCC advocates suffer from various biases such as feedback loops, grant needs, savior complexes, the need to feel important, a desire to rein in the “excesses” of capitalism, and even the law of inertia, you can see that GCC papers are really pretty close to perpetual motion machines — physics be damned.
More recently, of course, the health/science issue has been COVID-19. But how much that’s been published in medical and science journals on that has been replicated or in any case is trustworthy? Probably the single greatest “should-be” issue with COVID is the “died from” and “died with” issue. Given that a new Italian analysis finds only about 3 percent of “COVID-19” deaths didn’t have cofactors, that would convert the “worst pandemic of modern times” into something less than seasonal flu. (The CDC has found it’s about 6 percent.) Since autopsies can’t look at a person whose proximate cause of death was a heart attack but also was infected with COVID and say for a fact that the virus was not a contributory factor, we will never know the actual death toll from this virus. But the medical literature, unless it focuses on this point precisely, will simply assume a “with” is a “from.”
With Science and Nature being unabashedly liberal, one of their preoccupations is what mental instabilities contribute to the making of a political conservative. All the more so given that according to a New York Times survey (which has become a popular meme) that while 46 percent of all medical doctors consider themselves Republicans, a mere 24 percent of psychiatrists do. (Only epidemiologists ranked lower, which may explain much about how COVID has been handled.) Presumably psychologists similarly skew left. So they must be wondering, what makes a conservative tick?
In 2008, Science ran an article claiming that political conservatives have stronger physiological reactions to threatening images than liberals do. Knee-jerk right-wingers, if you will. The Science article was still influential a decade later, and some researchers decided to try to replicate it. As the liberal publication Slate noted, the follow-up came to no such conclusion. Figuring that since Science had published the original paper it would be the ideal place to publish an important follow-up, the second set of researchers submitted it there. Nothing doing, said Science. Try someone less important than us.
Was the Science refusal political? Was it an attempt at face-saving? Only The Shadow knows what lurks in the hearts of men. But if the purpose of Science is science, it certainly was irresponsible — and only the latest in a huge series. After all, this was the same journal that in 2011 published “Peak Oil Production May Already Be Here.” Never mind that peak oil predictions have been made since 1875 and while it was fracking that has actually made America a net oil and gas exporter, the first hydraulic fracturing experiment for oil and gas goes back to 1947. There was even a nuclear frack in Colorado in 1969. That is, “peak oil” was always a nonsense notion.
Still, we’re told, while journal biases may exist, they’re not political. Or so declared a study from just a few months ago that found “no serious evidence of a liberal (or conservative) bias with respect to replicability, quality or impact of research.” Never mind that it was published in a psychological journal — remember the ones with such notoriously low reliability? This is a version of the insistence that “Yes, we journalists are overwhelmingly liberal, but it doesn’t show up in our reporting.”
What’s the answer to this bias problem? Well, with apologies to the Steve Miller Band, sometimes “Oh, oh, there’s not a solution.” Still, the Wikipedia entry on the replication crisis has a nice list of potential remedies at its end — if we had the will to apply them.
Absent is the vaunted “peer-review.” After all, every paper discussed here was peer-reviewed. Yes, that Fauci JAMA paper was sent out to, well, somebody. But it may actually have been a group he recommended! Far from a magic bullet, peer review actually gives bad science what the military calls “cover and concealment.” The editor of the BMJ (remember the rule about shortness of journal names) and indeed chief executive of the BMJ Publishing Group for 13 years once declared it “a flawed process, full of easily identified defects with little evidence that it works.” Still, he said, “it is likely to remain central to science and journals because there is no obvious alternative, and scientists and editors have a continuing belief in peer review.” The highly respected Cochrane Network has also found “little empirical evidence is available to support the use of editorial peer review as a mechanism to ensure quality of biomedical research.” That said, it definitely wants you to know that “Each Cochrane Review is a peer-reviewed systematic review.” (Their redundancy.)
None of this is to advocate scientific nihilism or to bash “the experts.” There are truths on one end and likewise BS or magic on the other. I currently live in a country where people wear black boxes of herbs around their neck with a tiny blue LED to ward off COVID-19, and similarly some businesses use a white “incense” box to sterilize cash with a very impressive button that lights up for five seconds when pushed. Never mind that incense doesn’t kill viruses; the box is empty. Likewise, there are real experts. But in my 35 years of science journalism I’ve learned to quickly weed out the shysters. One trick? A real expert will often admit that “We don’t know,” while the shyster always knows. He’s not explaining, he’s advocating.
Another “trick” suggested by one of the Rady School authors is to essentially distrust “work that is more interesting or has been cited a lot.” Or, as Stuart Ritchie declares in his 2020 book Science Fictions: How Fraud, Bias, Negligence, and Hype Undermine the Search for Truth, “Let’s make science boring again.”
We must choose between science and science fiction.
Michael Fumento (www.fumento.com) has been a science journalist and attorney for 35 years, specializing in trying to find issues that are interesting and important yet real. It’s not always easy.
Comments are closed.
