The end-of-year festive dining season will soon be upon us, and perhaps it is finally time to be careful.
The world has always been bombarded with news, data and information about our dietary habits, which are then countered with fake news, bad research and disinformation about our food. This is curiously funny to observe, even though it is a sobering topic.
One example is the recently published paper from the American College of Physicians countering the World Health Organisation’s (WHO) recommendation for people to avoid processed and red meats.
Processed meats are categorised as Group 1 carcinogens (compounds known to cause cancer) and red meats are recognised as “probable” Group 1 carcinogens.
The actual recommendation in the new paper is as follows:
“The panel suggests that adults continue current unprocessed red meat consumption (weak recommendation, low-certainty evidence).
“Similarly, the panel suggests adults continue current processed meat consumption (weak recommendation, low-certainty evidence).”
What is curious and laughable at the same time is the amount of publicity this paper received, even though it readily admits that its “weak” recommendations are based on “low-certainty evidence”.
The only people who would cheer this report would be those involved in the production of processed meat products and red meats.
That is because the media would use the paper to confuse the public into thinking processed and red meats are fine to eat again.
But this is far from the reality of the situation. It may be safe to continue eating such foods for certain people if their consumption levels are minimal, but it is certainly not safe to continue “current consumption” levels if, for example, one is eating half a kg of fried bacon daily.
That is another weak point of the report, in that it does not establish a consumption threshold before the probable onset of various forms of cancer.
Such confusion arises from the interesting use of statistics. It has been estimated by Macleod in 2017 that perhaps up to 85% of research funding is wasted by inappropriate research questions, irrelevant endpoints, faulty study design, flawed execution, poor reporting and non-publication.
In short, not all research is proper science. If a conclusion is required by some industry, the industry may sponsor “research” which is targeted to arrive at a required conclusion. The media is then engaged to disseminate the “findings” to customers.
Sometimes, industries do not even bother with proper in-depth research when pressured by consumers.
An example may be BPS (bisphenol S, or 4,4’-sulfonyl-bisphenol) which was hurriedly adopted by the plastics industry to replace BPA (bisphenol A, or 4,4’-dihydroxy-2,2-diphenylpropane) after serious consumer scares in the United States and Europe.
Some tests AFTER the wholesale replacement of BPA by BPS established that BPS may be an even more active endocrine (or hormone) disruptor than BPA. (Read more online at https://bit.ly/360KvhK)
This sounds bizarre but quite often there are separate outcomes which may be suggested based on different interpretations of the same data.
A humorous study which caught my eye years ago was published in the Journal of the American Medical Association in 2013.
It claimed fatter people lived longer than thin people, and this made headlines everywhere as millions of people obviously welcomed the “longevity benefits” of being somewhat overweight.
It turned out to be a statistical hoax.
People who were identified as “skinny by choice or behaviour” and hence more likely to die were actually wasting away and really dying from serious diseases such as cancers.
The statistics were indicating that people get thin when they are gravely ill and about to die, hardly a recommendation for maintaining a few extra kg. But the facts were grossly distorted into some form of “benefit” which was then hyped up by the media.
The Texas sharpshooter fallacy
This is a statistical fallacy which is becoming more common due to the vast amounts of data available these days.
It is used not only in science, but also to confer credibility to political arguments, business decisions, superstitious beliefs, etc.
How it works is simple, especially if either a lot of data is available, or if only sparse data is possible.
It is called the Texas sharpshooter fallacy because it is similar to an amateur gunslinger firing 100 bullets into a wall and then drawing bulls’ eyes only around the clusters of bullet holes randomly distributed in the wall due to his inaccuracy.
If only sparse data is available, then the targets are painted over the fewer bullet holes in the wall. The Texas sharpshooter then claims that he had intentionally meant to shoot at those targets.
The reason why it can appear plausible is because of three factors:
i. Data, even random data, tend to “clump” at various points and appear to be significant even when there are no real correlations to any factors;
ii. If the research already has a preconceived outcome, then only a few choice data points are needed;
iii. There is so much research going on that it is impossible to check every published paper.
There is probably no worse example of the Texas sharpshooter than the huge and weird controversy over vaccines causing autism. This is because Andrew Wakefield not only fired a lot of bullets into a statistical wall, he also faked the bullet holes as well.
His actions have led to the preventable deaths of hundreds of innocent children because their parents believed in his fake “research”. Even today, there are cranks still promoting his fake study, even though he has been jailed for his fraud.
Another way the Texas sharpshooter works with real data is deceptively simple and almost plausible.
Studies in Sweden in 1992 seemed to suggest that children living within 300m of high-voltage power lines had four times the rate of leukaemia than the national average.
As the claim was based on a rich set of data (spread over 25 years), the Swedish Government felt compelled to act.
The reality, though, is over 800 ailments were monitored and in any such large monitoring exercise, there would be a high probability that at least one ailment would appear to be statistically significant, purely due to chance.
The meat consumption trend
Another curiosity is, although we can observe a trend towards eating less meat, the reality is 2019 will very likely exceed 2018’s world record for the production of meat; it looks very much like the amount of meat produced globally in 2019 will exceed 335 million tonnes, despite the swine flu epidemic in China.
How is it possible that we are eating more and more meat, even as more and more people are turning into flexitarians, vegetarians and vegans?
It cannot make sense. Or can it?
The reality is simple. If we investigate the average meat consumption pattern in Germany, statistics show a 25+% drop in the consumption of meat in 2018 compared to 1990, down from over 80kg a year to just under 60kg. This fits in with the observable trend in much of Western Europe.
However, in developing countries, economic progress has led to an accelerated demand for meat.
In 1960, the average annual consumption of meat per person in China was less than 5kg. However, in 2018, this has risen to over 60kg per person despite the population growing by about 600 million people.
China now consumes 28% of the world’s production of meat, which is twice as much as in the United States. However, the US has only just over a fifth of the population of China – and consumption of meat there is (amazingly) also still growing, albeit at a slower pace than China.
Now we know why over 259,000 sq km of the Amazon jungle was burnt and cleared for meat (mostly beef) and livestock feed production. The destroyed jungle area is roughly twice the size of Peninsular Malaysia.