The statistical bludgeon

 In the dark ages literacy was a secret jealously guarded by the senior clerics. It gave them power in the monopoly of handing down the written and immutable law; and, incidentally, enabled them to conceal their errors (and perversions) of interpretation. The lower clergy were only able to copy documents as arrays of symbols without intrinsic meaning, but God-given, and their errors propagated (such as confusing the Gothic long “s” with “f”).

In these days of almost universal literacy (of sorts) there is an analogy in the case of statistical literacy, though not one to be taken too far. The senior clergy understand statistics (to a patchy extent) and use or abuse them at will. The junior clergy put in numbers and extract them from computer packages, without understanding, and pass them on. The laity know their place, but are impressed.

Uncertainty is not an easy concept to accept. You are accustomed to getting your exercise book back from the teacher, with a tick against the sums that are right and a cross against the ones that are wrong. When someone tells you that there is a ninety percent probability that the answer is A and a ten percent probability that it is B, it is a bridge too far for many. They are grateful when an authority (say the EPA) saves them from crossing this pons asinorum by asserting that 90% = right and 10%=wrong. Therein lies a tale of grand deceit and devilry. The uncertainty has been removed at the cost of understanding.

Just as the mediaeval clergy used their own privileged interpretations of the written laws to bludgeon the laity into conformity, so the modern numerical necromancers use their interpretations of numbers to the same end. In both cases the penalty for indiscipline is the threat of pestilence, hell-fire and damnation; while the cost of conformity is a simple but substantial tithe on your income.

Who are they?

What motivates people to put so much effort into misleading their fellow beings? It is the usual suspects – money, power, fame and zealotry, often in combination.

You could write not just a book but a whole set of encyclopaedias on the statistical abuses carried out by the global warming industry. Globally it is a trillion dollar business, many people (some notorious but most invisible) making themselves multi-millionaires from its manifold branches. It is justified by reference to an alleged consensus, a word that is foreign to scientific argument. The cost is borne by ordinary people, be they Africans locked into primitive and unhealthy life styles or Europeans thrust into fuel poverty by onerous stealth taxes. It also provides a power base for ambitious politicians and bureaucrats. There are virtually no dissenters among the ruling political classes, though there are growing numbers in the general populations.

But the reach of statistical abuse goes way beyond that single issue. Politicians obsessed with targets and league tables are not only distorting the things they purport to measure: they are making the dishonest manipulation of numbers part of the every day work of people in management positions. They are even torturing and killing people in such areas as Britain’s largest industry, its National Health Service, sacrificing them to the gods of targets.

Bureaucrats are great exponents of the abuse of numbers. The SI system of units was one of the most beneficial inventions of the twentieth century, yet it was brought into public disrepute by over-zealous British officials, who used the law to enforce the use of kilograms by the “metric martyrs” (when you could still buy produce by the livre in French open markets). That is just a facile and obvious example, but there are many more complicated ones. Every party out of power promises to simplify the tax system, but in power they do exactly the opposite. A most egregious example is the withdrawal of child support by Britain’s Coalition Government from couples in which just one partner earns above a certain threshold. But what constitutes a couple? Are they married or unmarried; cohabiting or living apart; different sex or same sex; copulating or non-copulating; part of a ménage à trois; joined or separated in their finances and banking; landlord and tenant or joint owners? A fine playground for bureaucrats, but you could almost feel sorry for the revenue men. They are universally hated, for there is nothing more fearful than the combination of power and incompetence. Everyone has a story to tell; yet much of the blame is with politicians who are incapable of thinking things through, but you can confidently predict that they will produce the statistics to prove the success of their legislation.

Journalists are also great traders in junk statistics. A large portion of the modern newspaper is formed by the fillers, small columns of type comprising barely edited press releases from epidemiologists and cod psychologists, usually giving the insignificant results of small ill-controlled studies but with dramaticised headlines. The stories are virtually identical, word for word in several papers. Every now and then a story merits the big treatment, if it is deemed to affect readers personally or conjures a scary headline, but the nature is the same and investigative journalism notable by its absence.

Many people are even confused by simple percentages and graphs, which make them open to exploitation by activists and zealots, and some of those gullible people are journalists, politicians and bureaucrats. Below is a discussion of some of the trends and consequences of statistical bludgeoning, in no particular order.

 Deliberate fraud

Much of the dissemination of misinformation is relatively innocent, people dabbling in something they do not quite understand and motivated by personal ambition. Some of it is less so and knowingly false. At the extreme, however, there are practices that are intended to achieve non-scientific aims by coldly calculated deception under the cloak of corrupt science.

It is fairly safe to assume, for example, that anything coming out of the USA EPA is fraudulent. Ever since their first and greatest coup, the “metastudy” on environmental tobacco smoke (involving at least five gross statistical frauds), was embraced by authoritarian politicians, they have been given carte blanche to ignore the methods and laws of science. They ruthlessly exploit this licence for purely political ends.

The EPA does not now even bother with generation of fraudulent statistics: it makes ex-cathedra pronouncements that defy the laws of science. For example: anything they arbitrarily choose to declare as a poison kills with no lower limit of dosage, contrary to the first law of toxicology (that the poison is in the dose).

The EPA has evolved into a congress-bypass operation, enabling an administration with water melon tendencies to mount attacks on essential indigenous industries, such as energy. It is virtually unmonitored in its activities, yet disposes of nearly ten billion dollars of taxpayers’ money annually. A fine and trenchant essay by  Henry I Miller on Investing in bad science includes the remark:

“For some reason I was favored with periodic reports of the research funded by the epa. The overwhelming majority of it was shoddy, irrelevant, and unpublishable.”

which just about sums it up.

The abandonment of causality

One of the immutable laws of experimental statistics was correlation is not causation. Thus in sound science, in addition to correlation, you have to provide a plausible mechanism linking putative cause to effect before a hypothesis can even be considered. This requirement is openly contested and defied by prominent epidemiologists.

Here is the sort of fictional form of causation you can create by ignoring the law:

Over decades, the exposure to tobacco smoke continually decreased, while the rate of childhood asthma increased. Therefore tobacco smoke is a preventative for asthma.

That is, of course, a deliberately facetious construction, but who knows? It might even be true.

The one in twenty lottery accepted as the norm

Statistical significance normally has two components, the relative risk (RR) and the probability (P) that the result occurred by pure accident. The acceptable limits chosen for these measures in epidemiology and related fields were from the outset regarded as very lax by traditional scientists. In particular, the limit P=0.05, now almost universal in epidemiology, means that there is a one in twenty chance that the result is by accident. If there are a hundred such results in an issue of an epidemiological journal, then by their own admission five of them are completely wrong. It is, however, worse than that. The provenance of P and RR depends on the maintenance of strict conditions (large, blind, randomised, controlled trials etc.). More often than not epidemiological and psychological claims are based on small, ill-controlled observational studies, making them essentially worthless. So why do they persist with this charade? Well, there are simply not nearly enough “discoveries” available to justify the numbers of people employed in the epidemiological scare industry, including publishers of journals, so they have to keep the mouth of the net wide. In consequence, most people now believe that the “scientists” are always contradicting each other, whereby the whole of science is tainted in the public eye. By sheer weight of repetition they have established that their style of lottery is “normal”. It is no more normal than swine jumping over a cliff. The widespread smoking ban was justified by the EPA’s one in ten lottery, an all-time low in standards.

Each enormity justifies the next

Anything hated by a zealot group is subject to campaigns for total ban, but sometimes the motivation is nothing more than a bid for fame by a group of “scientists”. The withdrawal of vioxx, on the basis of a statistical blunder, is used to provide justification of other proposed bans, such as that of diclofenac, on marginal statistical grounds. This is not to say that there is no relation between the drug and heart disease: it is to say that there is no significant evidence either way, and certainly not enough to constrain clinical judgement.

Take it from one who knows, severe arthritic pain is absolute hell. Many sufferers were bereft when their vioxx was taken away, as more will be if the same happens to diclofenac. No ingested substance has just one effect on one part of the body: everything has side effects. Keep the clinicians informed and let them make the decisions.

Playing favourites

Just as some treatments and substances come under attack, either arbitrarily or from pure malignance, so the mirror image occurs. There are others that are treated as holy relics and to question them is to commit sacrilege. 

Statins provide a curious example. Many clinicians state categorically that they have no serious side effects, while others campaign for the compulsory dosing of the whole population above a certain age. Anecdotally, however, there are strong indications of serious and unpleasant side-effects, but public funding of research into them is largely denied.

On reviewing the book that exploded the whole cholesterol myth, over a decade ago, I remarked that “It is a scientific tour de force, for which he will, no doubt, be dismissed as a crank. It says much of the state of the modern world when those who adhere to the scientific method are cranks, while those who flout it win Nobel prizes.” I was wrong. He was not dismissed as a crank, he was just ignored.

The cholesterol/statin myths are an offshoot of the political anti-obesity campaign. Excess body fat is unaesthetic and it is now considered acceptable to medicalise any behaviour of which we do not approve. This has created a bandwagon upon which various interest groups (such as militant vegetarians) have scrambled. Cholesterols are vital substances, absolutely essential to a wide range of structures and processes in the animal body, they are synthesised in the cells, but also obtained somewhat inefficiently from dietary animal fat. There is a consensus (that unscientific word again) that there is a link between cholesterol and atherosclerosis, but there is also a small active international group of sceptics that questions this assumption. Whatever the truth of either view there is no question that (as in other fields) the existence of a consensus suppresses research into alternative hypotheses.

Any graph will do

In the popular mind the graph has been elevated to the status of a religious icon. If there is a graph it must be right. Apart from the fact that a graph is vulnerable to the sort of deceptive manipulation that we have labelled chartmanship, its contribution to the argument can be misleading or even completely irrelevant. A choice example occurred when the salt zealots were challenged to produce evidence for their death-laden claims. They came up with this.

A few thoughts occur immediately:

·         First and foremost, the graph has nothing to do with a putative relationship between dietary sodium and death, which is the nub of the claim.

·         You can fit a straight line of non-zero slope to any random array of blobs. If the trend is not visible in the absence of the line it may be considered to be spurious (the law of trends).

·         Plotting differences rather than absolute values is a noise-increasing process. Here it is done in two dimensions.

·         The choice of which variable is the ordinate and which the abscissa bestows a spurious aura of causality.

·         The range of variation reported is of the same order of magnitude as the random variation in a single patient (as anyone subjected to regular blood pressure testing will confirm).

·         The abandonment of absolute instruments (sphygmomanometers) based on mercury columns (due to EU chemophobia) in favour of secondary instruments that require regular calibration (ironically still in mm Hg) increases potential measurement errors.

·         Small scale studies are worthless (and some here are very small). Combining them by the dubious process of the meta-analysis does not improve matters and introduces other biases (e.g. publication bias).

·         Not only are ordinary readers confused by graphs: as the vioxx case shows, even self-styled scientists cannot interpret their own visual results.

·         Last and hindmost, what were these scatter diagrams supposed to demonstrate? Damned if I know.

 Opportunism

The statistical bigots are ever watchful for “evidence” to promote their causes. In particular they monitor the numbers so that they can publicise any accidental correlations that go in the “right” direction, while ignoring those that go in the “wrong” one.  This activity is at its highest around one of their “coups”; for the march of the zealots must be ever forward, with no reversion.

Thus there is constant selective mining of “before and after events” relative to the smoking ban, which are celebrated by the establishment media. During the writing of this piece there was childhood asthma  in January 2013. Then, even more bizarrely there was Premature birth in February.

In the fifties, none of my many childhood friends at school or in my neighbourhood had witnessed asthma or its symptoms. I was able to demonstrate wheezing to them, which I could trigger by drinking a small quantity of lemon barley water, and it became a sort of party piece. We were constantly exposed to tobacco smoke everywhere, especially in enclosed spaces like cinemas. Now I see many children climbing onto the school bus clutching their Ventolin inhalers and wonder why there seems to be so little interest in finding out why.

Discussion

In the modern age almost everything is political; not just party-political, but also driven by sometimes covert, extra-parliamentary interest groups.

Science has come under total political control and so has lost much of its raison d’être, which is human curiosity.  In common parlance the very word “science” has changed its meaning. A repeated observation is that the existence of a consensus (a concept alien to science) in any field results in the suppression of research into alternative hypotheses. Politicians and bureaucrats nurture research that confirms their prejudices. That promotes a consensus, which inevitably becomes self-reinforcing by the exclusion of alternatives, so alleged scientific progress becomes merely the expression of Government policy.

The EU, USA, Britain and Australia all have left-leaning, authoritarian governments and bureaucracies, who believe it is their right and duty to exercise control over every detail of their victims’ lives. Powerful monolithic media organisations, such as the BBC and ABC, ruthlessly exercise rigorous censorship and selectivity to magnify the apparent weight of zealot causes favoured by the contemporary establishment. Scientific matters are discussed in terms of political ideology rather than the natural language of science, which is predominantly mathematics.

The looseness of trendy modern statistical procedures provides a ready response to political requirements. Results can be virtually manufactured to order. These are then used, with the cooperation of the establishment media, to bludgeon the population into obedience.

John Brignell

February 2013

Index