Gareth  Stevens

If you are like me, you have probably noticed an indisputable correlation between a downturn in your mental wellbeing and incessantly cruising the internet and repeatedly checking your social networking apps. Come on – hands up if you have noticed this, it can’t be just me? So is this just coincidence or do these two variables have a causal relationship? If they do, which way does it flow? Does waking up with a sense of purposelessness predispose you to being wedded to your smartphone all day, or does the endless web-surfing and feed-checking bring you down? Or is it that both perniciously combine to push you to scroll more and feel more morose?

Tyler Vigen has a great website called ‘spurious correlations’ that instantly convinces us that to rely on correlation alone without in-depth analysis can lead to nonsensical and superficial thinking. When looking at the chart below you could be forgiven for thinking that cutting down on your cheese intake is a prophylactic against dying in a knot of your own bedsheets. Absurd right? Sure, detecting correlation can be a good starting point, but it is
only basecamp. When further investigated, the initial correlation between smoking and lung cancer turned out to be causal, but it is important to know the difference between coincidence and a causal relationship.

Vigen has written software that trawls through disparate databases and reveals very close correlations between data sets from different and unconnected domains. There is, he found, a stunning inverse correlation between the number of honey producing beehives in the USA and the incidence of street arrests for the possession of marijuana, but surely only a fool would immediately suggest that beekeepers lobby for the decriminalisation of weed, or that federal government subsidise beekeepers as an efficacious strategy for preventing drug crime?

What Does Crime Have to Do with Abortion?
In 2005 the book ‘Freakonomics’ by Levitt and Dubner popularised the notion that we need to dig deeper to find root causes of new phenomena and behaviours. More than that they opened the way to considering oblique, obscure and multi-stage causal chains. In the early 1990s the crime rate in the US dropped precipitously after the previous fifteen years had seen it rise by 80%. Lots of research casually harvested the low fruit. Different bodies claimed a range of causes had led to the sudden drop in crime. Some said it was a direct result of tighter gun laws, others that it was to do with more police being on the street. Other arguments claimed that the reason was recent changes in crack and other drug markets and/or increased use of capital punishment and higher rates of imprisonment.

The authors of Freakonomics offered a previously unconsidered cause of the crime rate drop. In 1973 the growing number of US states in which abortion became legal extended to the whole country. This, they said, was a very important component cause for the downtrend in crime in the early 1990s. Their argument went like this: once abortion became legal and acceptable, young, poor and unmarried couples would abort where previously their child would have been born into a desperate situation, and would in their teens have been more likely to resort to crime. The debate concerning Levitt and Dubner’s claim still rages, but the point is is that some correlations and causal chains are hidden in a complex mire: a flux of causal relationships which are not immediately obvious or domain specific.

Domain Lock In
When I was about sixteen I was sitting across the table from my Dad as he read his broadsheet at breakfast. I noticed a story on the front page that faced me. It described how the UK government were going to finance research that would try to ascertain why you were more likely to be in a road traffic accident if you were driving a black or white vehicle. This research, the article said, would focus on whether there was any optical reason why this might be the case. I remember thinking “Noooooo …… it’s obvious, you are more likely to drive like an idiot if you own a white or black car!!!”.

The psychologist Susan Blackmore explores this way of thinking further. She asked “Suppose it’s been discovered that children who eat more tomato ketchup do worse in their exams. Why could this be?”. Again it doesn’t sound like a revelatory question does it? Try asking a large group of people this question as I have, and you will quickly see that you have opened up a proverbial can of worms. Those with a more scientific bent will quickly assert the possibility that some chemical component of ketchup might be impeding cognition or people’s ability to focus. Others however, will argue for a more convoluted causal mechanism involving ideas about socio-economic groups and the extent to which they value education or consume junk food. What does this immediately tell us? That we should be beware of domain lock in: that is that we often need to combine disciplines and think beyond correlation to arrive at the truth.

How Does My Meat Consumption Cause Flooding?
When Somerset Levels flooded in December of 2014, the deluge left 600 homes uninhabitable. There was an immediate squabble to identify the cause. People still seem to be wedded to the idea that there must be one, usually obvious cause, and so recriminations and superficial debate followed. Nothing is binary and there is rarely a singular cause for floods like this. The arguments and counterarguments that followed were not designed to
get to the truth, they were more about absolving the arguer from any culpability, or to apportion blame elsewhere. Confirmation bigotry was rife. Some said it was climate change, others unprecedented rainfall. It was also claimed that the environment agency had not carried out necessary dredging work on the rivers in the area. We should also not forget that the tidal range in the Bristol estuary is the second biggest in the world and that large areas affected by the flood are below sea level.

The activist and writer George Monbiot had another idea. He claimed that the fact that so many of us eat too much meat was a prime cause. Before you read on, try to work out the causal chain that he identified. He writes that mass animal agriculture causes the UK government to subsidise the production of maize, which is a main ingredient of livestock feed. In turn, huge areas of arable land are given over to maize production. Maize root systems are very shallow and so during heavy rain, instead of being slowly absorbed into firm ground, the rainwater simply runs off and takes a lot of topsoil with it to become silt in nearby streams and rivers.

So beware, not only is correlation not always causal, but also an outcome may result from a confluence of causes, the main ones of which may be out of sight and difficult to discern.


We hope you have enjoyed reading this article from Hastings Independent. The future of this volunteer led, non-profit publication would be far more secure with the aid of a small donation. It only takes a minute and we would be very grateful.