Thursday, 4 July 2013

“I’m not sure what’s wrong with a statistical offering?”

THIS POST is forced upon me after a Twitter exchange with an aspiring PhD candidate at the University of Toronto.

Amazingly, my innocuous tweet quickly received this in response:-

The ‘every year’ emphasis is based upon a UK Government ‘report,’ to which he links, from the year 2005!

The report itself stands as a perfect example of political spin and the way that statistics are used by the ‘ruling’ classes to have their ‘projections’ accepted as fact. But the interesting thing about all political statistics is that their projections are never once qualified with a ‘confidence level,’ which, as any real statistician will tell you, is the most important component of any mathematical analysis.

Like similar reports, this one too makes conclusions about EU membership from data taken from the period since Britain signed the Single European Act. It does not compare the results of non-membership, prior to joining the EC, or the fourteen years that followed that – and then seek to explain the differing periods in terms of ‘better’ and ‘worse’. Despite the fact that the report concludes membership has been beneficial, its own data base has been specifically engineered to prevent any such analysis – so all its 40 pages contain nothing but pure tosh. Its authors have not compared any facts that could lead to any of its conclusions.

And what about all those ‘projections’ that the politicians and their party trolls would have us believe are fact? Well, they are all based upon the statistical religion that, if you can determine a mathematical progression that best fits a series of data, then you can confidently extend the line into the future and thus predict a ‘trend.’ In a laboratory, under controlled conditions, the technique is indeed useful in determining if a particular result might be worth pursuing; but, outside the laboratory, where conditions cannot be strictly controlled, it has no more value than studying a sacrificed animal’s entrails. (In fact, given the testimony of some historical authors regarding the Roman’s use of the latter to predict the outcome of certain battles: perhaps even less).

But get this: the statisticians know perfectly well that the more variance there is in the data (the greater the distance between the peaks and troughs in a graph) the less confidence there can be in a trend line’s projection. That is why they always qualify their results by expressing that variance mathematically (and stating it in their report).

Just how much one trusts any statistical analysis, it can be shown, largely depends upon one’s age; because the young, more often than not, are prepared to take more chances.

Personally, I would not trust any analysis that purported to be less than 95% confident, and, accordingly, would not devote any more than 95% of my ‘savings’ towards such an ‘investment.’ But the problem with political statistics is that, often, the voters are betting their lives upon politicians making decisions based on confidence levels of very, very much less – and the bastards never have the integrity to tell us what those confidence levels are…

Love his use of ‘primary sources’ there, still refusing to admit that his 2005 argument is just as valid as me claiming that the reason I have a clean driving license is because I insure with Aviva.

There was a time when a nation’s rulers employed astrologers to interpret the heavens and determine the best time for undertaking certain actions, or analyse an enemy’s character to determine their weaknesses; but even they never stooped so low as to imply they could predict the future. Modern day politicians, however, do just that all the time. They really do believe they can predict the future using a statistical trend line; but they also believe they have the power to change its course by simply making other decisions. They fund a horrendously expensive statistical study (and define its parameters) at tax-payer expense – and then they pick the parts of it that ‘agree’ with their predetermined viewpoints, whilst being sure never to publish that confidence level which would show that all its conclusions are meaningless (because the variance in the data is likely to be no more than what is a naturally occurring random fluctuation).

Our secular society has given birth to the Statistics God, and It is alive and well in the Bank of England, HM Treasury, and the Met Office. Furthermore, the Crab’s Statistics God is the biggest con-artist of all.

Financial, Economic, and Climate models are all based upon one idea: provide copious amounts of data and then derive a mathematical formula that provides a line of best fit so it can be extended into the future and a prediction made. But the con is that the accuracy of that mathematical formula is never rigorously tested; because, as new data is fed-in for a further ‘projection,’ the formula is recalculated! In other words: every mathematical prediction has been proven to be untrue once new data has been gathered…

The fact is: humanity has never given-up studying animal entrails in an effort to predict the future: it is just more humane in its methods and has taken to referring to its predictions as facts – but they are not…

The only mathematical facts that can be trusted are those produced by an honest accountant – and the EU’s accounts have not been signed-off for the past twenty years (which just so happens to incorporate much of the data from which that 2005 report was derived).

So I ask myself, and my readers: just how dumb can a PhD student be?..

No comments:

Post a Comment