Tuesday, 11 February 2014

Predictions & Forecasts - the danger of statistics

One of the things our technological world does is produce lots of numbers, too many in fact. Every day in our press we see Labour, the LibDems and the Tories picking different ones to justify their position or trash the other side. Too much information can lead to clever folk playing the stats for their benefit.

There are, out there, people who specialize in understanding which numbers are worth looking at (signals) and which ones simply pack the sample and can, in the wrong hands, distort it (noise). The trouble is those with an axe to grind tend to dislike these folk; whilst they get the right trends and debunk bad economics, poor thinking or trash science, they are rarely popular amongst pressure groups.

There are, of course, two things you can do with statistics. You can predict or you can forecast. Prediction means you have seen the trends, taken out all the noise and are willing to make absolute statements about what will happen. Forecasts are gentler, suggesting what might happen. Deciding what statistics tell us is a science in itself but the mathematicians who have the skills to do it hardly ever get asked for their opinions.

Let me give you an example of how statistical trends can be misread.

In 1916 the British Army started issuing tin helmets to their troops. After six months the statistics showed an increase in head injuries and at a War Cabinet meeting a ‘specialist’ contended that the troops were getting over confident because of the helmets – he suggested they be withdrawn. It took a professor of mathematics to show that they were looking at the noise not the signal. Deaths by head injury were actually massively reduced and so reported head injuries that were none fatal were increasing. Signal and noise and the inability of a scientist to see through the fog almost caused the biggest mistake possible.

Scientists know all about this phenomena. In fact they often use it to their advantage. Take the passive smoking debate for instance. In its early days there were no supporting statistics, just a huge amount of sympathy for Roy Castle’s wife and a lot of businesses (particularly airlines) who saw advantage in an excuse to ban smoking. Then along came GP’s to oblige those in favour of a ban. They were told to classify every respiratory problem in a child whose parents smoked as ‘passive smoking’. There was not a shred of evidence for this assertion, they just did it – and hey presto science emerges that can allow scientists to make preposterous claims about the effects of passive smoking. Bad, dishonest science wins the day. Yet the signal, the fact that particulate concentrations in exhaled smoker’s breath are too low to represent a threat to others is ignored. It is ignored because the ones following the noise say they have data – they don’t, they invented it, but it sticks.

But the best, the very best use of noise has to be Anthropogenic Global Warming (AGW) or man-made global warming as it’s known.

There is so much data about; caused by scientists knowing that if they utter the AGW phrase governments will throw money at them (even scientists have to eat). And scientists are good at turning the statistical data their own way. Even when the mathematicians are screaming at them about finding the signal within the noise. They don’t want to look. They select the data that fits their views and anybody who dares to say otherwise is called a flat earther, an idiot or worse. It doesn’t even matter when the maths bods say they are just trying to point out facts about the data – dare to question and you are a Luddite.

But the AGW mob made a fatal mistake. If they had stuck to forecasting, and screaming about the likelihood that the world would end, they would probably have been fine. But they chose instead to predict. In fact for the last 10 years they have been making hard predictions, even as mathematicians told them they were not working hard enough with the data and that there was no possibility of making predictions with so much noise in the system.

Of course, when you get carried away like that and start reading things into the data that isn’t there, you will ultimately hit problems. According to the predictions made in 2003 the earth should be 0.3 degrees warmer than it is now (that’s a lot evidently) but it isn’t. In fact it’s cooler than it was in 2003. Interestingly when this prediction was made two mathematicians pointed out that the Pacific Carbon Sink, which stores heat and carbon had switched itself off in 1980, that data showed it did that sort of thing and that if it came back online nothing (REPEAT NOTHING) would happen. These two guys barely got out with their lives!

The Pacific system turned back on in 2004. Since then it’s been storing carbon and heat (as it does) and dissipating that heat though deep flow systems, cooling as it goes.

It may continue to do this for 50,000 years or 5 years – nobody knows because we don’t understand how it works. It is, however, a major signal in amongst the noise of the data collected by scientists who then bleated about their findings. As I say, their mistake was in making predictions.

2003 – the Arctic ice mass would be gone completely by 2010; the ice is thicker this year than since records began
2004 – 70% of UK rivers would dry up by 2014 – take a look outside!
2005 – the glacial waters that feed California with it’s water supply would be gone by 2010; today they are thicker than ever

and then you get the crazy stuff

2006 – there would be a 60% increase within 10 years of the incidence of scale 9 earthquakes – nobody can predict earthquake likelihood, they can’t even say how the after shocks will work
2007 – the warm water currents that keep the UK’s climate temperate would shut down by 2015 as a result of increased ice melt in the Arctic – today this weather system is stronger than in recorded history and again, nobody knows how it works
2008 – the jet stream would head south creating polar conditions behind it (in the UK). The jet stream has sat further north than ever before for the last 4 years and nobody knows how it works!!

And this stuff continues. Every major weather event is put down to AGW (even the floods we are having now which seem to be directly linked to the jet stream speeding up as it does on a 300 year cycle) without science, facts or in many cases the slightest attempt to prove the statement.

The message for the AGW scientists is stick to forecasts. Predictions using bad science tend to lead to egg on faces. Of course it would be better to invite the mathematicians in to find the signals and eradicate the noise from the statistics. Unfortunately the work in that area done so far would suggest what many think; that we are insignificant in the weather systems of this big rock we live on and there’s nothing we can do to stop it managing itself.

Sobering thought given all those expensive wind turbines, solar panels and the billions spent on ‘green issues’. It can’t be true surely? That governments and their scientists are conning us for their own benefit and spending our money pointlessly?