Statistical Codology

Tim Harford, in his column last Saturday in the Financial Times, laments the innumeracy which pervades the popular press and much of the political debate. There are people unable to remember the difference between a million and a billion constantly pontificating on weighty economic issues of all descriptions in both print and broadcast media. My favourite category of statistical codology is the university ranking tables now produced in profusion by various self-appointed scorekeepers, notably the Times Higher Education Supplement which ought to know better. The Irish newspapers reported last week the sad news that Trinity College Dublin had fallen from 117th best university in the whole wide world to a mere 120th according to this venerable source. It has fallen behind the University of York, the shame of it, and has managed to stay just an inch ahead of the University of Oslo (phew!).
Unreported was the even sadder news that the Cork City football team, last season’s Irish champions, are now ranked a humble 160th in the world, down from the heady heights of 157th this time last year and just a corner-kick ahead of Croatia’s Rijeka FC. This vital info can be gleaned from, who must be wondering why their number-crunching attracts so little coverage.
There is a simple reason. Football fans know that these rankings mean nothing whatsoever. If anyone really needed to know whether Cork or Rijeka boasts the better team, say if they were drawn against one another in some competition, the matter takes ninety minutes to resolve. How though do you check if Trinity or Oslo has the better university? A ninety-minute showdown between the staff of the two institutions would be quite a spectacle but could turn ugly.
The attraction of the university rankings for journalists is precisely that they are meaningless and accordingly incontrovertible. The winner in the latest league table for world’s best university was none other than the University of Oxford, whose distinguished alumni include Boris Johnson, Theresa May, David Cameron and just about all the other folks who have been doing such a spiffing job on Brexit. Which does not mean that Oxford is a poor university. It just means that the concept of university league tables is for the birds. Football league tables (based on the results of actual matches between the teams in each league) are at the upper end of respectable scientific practice by comparison. It looks like Dundalk will relieve Cork of their title as the season draws to a close. But since each team will have played 36 games against the rest this really does suggest that Dundalk have the best team.
However daft the concept and dodgy the statistical methodology, the university tables have their uses. The recent declines in the rankings for the Irish colleges have fuelled demands from their presidents for extra taxpayer cash. A few years back the same tables were showing an advance up the rankings, proof positive, according to the same people, that public spending on universities was delivering the goods and should be increased.
Brexit provides another example of the innumeracy which annoys Tim Harford. The UK’s annual and recurring net contribution to the EU budget has recently been running at about £9 billion per annum, a large number with lots of zeroes. This number, not to be confused with the once-off exit bill, has been ventilated by Brexiteers as, quite properly, a potential ongoing benefit to the Treasury of a full exit. Another figure in circulation is the population of the EU-27 which comes in around 450 million, not quite so large a number but lots of zeroes too. There have been regular assertions that the loss of the UK’s money will cause great damage to the finances of the EU-27: the political editor of the Sunday Express Camilla Tominey ventured that it would ‘bankrupt the European Union’ on a BBC programme, without challenge, a few months back. Nobody at the BBC, it would appear, has thus far bothered to divide the larger of these two big numbers by the smaller. The answer turns out to be £20 per head per annum. Total government revenue in the EU-27 is close to £4,000 billion. The UK’s £9 billion will be missed, but not noticed.
Some people are good at figures but most are not and are lost with very large numbers. Tim Harford’s solution is a call for better statistical training in schools and universities. Even at Oxford this looks a wee bit optimistic. There is however no excuse for the sloppiness about statistical matters in well-resourced media organisations. Here’s another very large number: the BBC gets about £3.5 billion per annum from the license fee, surely enough to engage the services of a statistician, or to buy everyone a pocket calculator. (courtesy the Farmers Journal).

11 replies on “Statistical Codology”

I must admit I do despair about the grasp of numeracy exhibited by many of the Brits I meet and the perverse pride so many seem to take in exhibiting innumeracy. So it should be no surprise that this is reflected in the media and politics. (However, I think you should relate the UK’s net annual contribution to the EU’s annual budget.) I reckon that, on average, Irish people have a better grasps.

But we spin far better yarns and are able to suspend disbelief and to project optical illusions almost indefinitely. Some of these yarns and tales are found in most advanced economies, but we have turned their generation in to a true art form. For example, there’s the fiction that a government is analagous to a household. Margaret Thatcher was big on this and Angela Merkel has her sensible Swabian housewfe, but Irish pols could bring tears to your eyes. So when times are hard, a government should tighten its belt. In most instances this is the very opposite of what it should do. Another lovely fiction is that governments shouldn’t spend until they’ve identified the matching tax revenues. However, governments use their central bank accounts to generate money by magic, spend it and then tax. And there’s the fiction that banks have to accumulate deposits before they can lend. Banks like governments create money by magic when they issue a loan and rely on deposits rolling in subsequently or bondholders liking the quality of loans they’ve issued.

But we have peculiarly Irish fictions, like the cost of living is very high in Ireland, but it’s the result of things out of our control. It’s like the rain, there’s nothing we can do about it. For example, only the Finns pay higher average household electricity bills than us (they need a lot of energy to keep warm), but the story goes that our bills are high because we use a lot of fossil fuels to make our electricity and the prices of these are very volatile. (Of course this means that final prices should reflect some of the volatility, but we’ll use a mental reservation on that.) Another reason is that electricity costs are high becasue houses are further apart than in other countries and we use an awful lot of poles and wire. And another good one is that we’ve been investing hugely because demand has been growing – but that one has been weakened a bit by the fall in electricity consumption since 2008 (though we seem to be getting back on to trend. And a lot of the investment is due to the madcap decision to use the electricity industry to meet our EU renewable target for 2020.

Another popular fiction is that if something is funded from general government expenditure it’s free.

So maybe the Brits have their innumeracy and we have our economic fictions.

Indeed, Colm. But Herr Schäuble, from his new perch in the Bundestag, is no longer able to enforce his fixation on the Schwarze null and on treating all Euros as his domestic currency. Indeed, it was only smaller member-states which could be bullied and threatened who were prevented from using the magic money tree. Herr Scholz, with his SPD neck and neck in the opinion polls with the Greens and the AfD, and his colleagues on the Euro Group may be forced to give the magic money tree a shake. And Signor Draghi may have to resign himself to the fact that his compatriots from the League and M5S will give it a good shake. Even small well behaved members like Ireland who took their puishment and didn’t indulge in back chat may be allowed to give it a little shake.

It’s difficult to push back against what you describe as statistical codology and the media’s uncritical regurgitation of this codology when those who exercise power and influence and their armies of fluunkies and functionaries advance fictions that are convenient for them and their affiliated rent capturers but that rely on statistical and economic codology – or in Prof. Frankfurt’s terms bullshit.

The BBC has been doing a very poor job of reporting on many matters, particularly those that are political and contentious, for some time now. The reality is that it has been cowed by the right. Subject to incessant attacks over the years for claimed left-wing bias, it has over-reacted and over-compensated. In the interests of ‘balance’ it puts up climate denying fools like Dominic Lawson against eminent climate scientists and allows economically illiterate politicians and activists like Rees-Mogg (and pretty much the entire UK political establishment) to spout nonsense, unchallenged. All this is justified, one must assume, only on the basis, as Gove would have it, that the British public has had enough of experts and therefore prefers to be informed, educated and entertained by idiots (given the popularity of Johnson, perhaps on reflection, the BBC has actually got it right).

The BBC remains the premier news source for so many Britons. If it cannot be relied upon for objectivity, accuracy and real (as opposed to spurious) balance, it damages a UK public discourse already under severe pressure from highly-partisan media. And given that when England gets a cold, Ireland gets pneumonia, the outlook for us is hardly bright.

Well, leaving Brexit and Irish domestic monetary policy aside for a moment, the lack of statistical (and basic numerical) understanding extends to all sorts of places.

Even in business, where “big data” is still a current fashion, advanced data tools often simply mean that people – even or especially senior management – can draw madly wrong conclusions from more data faster.

A few years ago, a finance colleague at the time commented “It’s like watching a competition to see who can draw an unsupported conclusion fastest”.And that was with people responsible for big chunks of the business. They had big egos so they each wanted to be first to “understand” the data. Even if the data really didn’t allow drawing any conclusion at all.

As for solutions? Statistics and probability should probably be compulsory requirements in ANY undergrad degree. Even art history or – yes – journalism. Like Mathematics being a requirement to matriculate for any university course in the first place.

A professor of mine once described the theory of comparative advantage as meeting all his requirements for something to be taught in university. It was “important, non-obvious, and true”. Probability and statistics also qualify.

“As for solutions? Statistics and probability should probably be compulsory requirements in ANY undergrad degree.” Hugh S.

Take it easy there Hugh. There are two types of stats (well, sort of ..) the theory stuff and the practical stuff. The theory is always correct – its in the definitions and the selection of premises. But the practical stuff is devilish difficult and messy – problems always start with the sampling and things can only get progressively worse the further the analyst is removed from the base data. One ends up with Meaningless Means masquerading as factual evidence. Intelligent folk are, unfortunately, more likely than most to be seduced by their own numerical analyses: precisely because the Theory is always correct. So, how could their practical efforts be otherwise?

“The practice of quantitative statistical analysis is the attempt to reduce our level of uncertainty in something – rather than attempting to enhance our level of certainty.” Null Hypotheses – and all that!

By all means teach practical quantitative analysis to those undergrads who are more likely to encounter dodgy statistical estimates on a daily basis (think G*Ps etc.) – and leave probability theory to the math nerds, where it belongs. If you are not shown how to recognize dodgy estimates: then how will you know them when you encounter them?

The whole need is to educate people so that they can understand what they’re doing with statistics and probability rather than a situation where smart people are getting it totally and confidently wrong (because probabilities and statistics are often very unintuitive) and they have insufficient education to even understand that it IS often unintuitive. Again, “important, non-obvious, and true”.

Somehow I suspect the author would have written a very different piece if Irish universities – let’s say UCD for instance currently ranked somewhere between 201-250 ) were climbing up the university league instead of tumbling out of the top 100 and continuing their inevitable descent year after year.
Instead he resorts to his usual lazy Brit-bashing shtick that’s as predictable as it is dull.
It was economists like him who confidently predicted,Nostradamus-style,much wailing and gnashing of teeth in the UK economy,huge job losses,recession and a tsunami of jobs and major financial institutions out of the City of London as a result of the Brexit vote.
None of which happened.
Frankly if I was one of those economists I’d tread carefully flinging out jibes about the BBC letting down the academic nature of the Brexit debate.
After all it was the arrogance of those academics and their fellow travellers in the Smugeratti which helped propel the UK out of the EU.

@Professor Pie
None is zero. There certainly have been jobs moved out of London, so this too is somewhat innumerate.

@ Dearg Doom
Actually none was referring to huge job losses not no job losses.
Of which there have been a tiny number and,by most reliable accounts,unlikely to top 5,000 after Brexit.
This is far less than the 75,000 that ” experts such as the Bank of England and Goldman Sachs were predicting.

UK GDP will be about 2% lower this year than it would have been in the absence of the Brexit referendum, and that is before the UK has actually left. Consumer prices are also around 1.5% higher then they would have been otherwise. So it’s not as though there has been no effect so far.

A hard Brexit scenario almost certainly would result in a recession for Britain and possibly Ireland too.

Comments are closed.