How Statistics Get Abused in the Media
Televised sport has gone stats-crazy; now we have to learn which we can trust
When Google’s chief economist Hal Varian remarked that ‘the sexy job in the next ten years will be statisticians’, the producers of sports shows sat up and paid attention.
Nowadays, you can’t go five minutes of couch-dwelling without a slick graph or pie chart telling you what you simply need to know about Team X and Team Y — with the intention of guiding the viewer towards forming a conclusion. Unfortunately, a lot of the time, these statistics don’t paint the full picture. We’re going to take a little look at why and how the media mislead the general public with statistics.
Picture this (made-up) scenario:
FULHAM 9th in table after 34 games
ASTON VILLA 17th in table after 34 games
I am a big believer in utilising statistics when considering potential bets. They are a vital component of decision making. So when John Motson tells you that Fulham are unlikely to lose to Villa at their home ground because they have gone five games unbeaten there, it’s a lock that they’ll avoid defeat. Right?
In a Word…
No. It’s not a lock by any stretch. In this hypothetical example, what Mr Motson has stated is a fact. But then, there are facts and there are facts. He has entertained us with a catchy little nugget, whose relevance lies somewhere along a scale that ranges from so what? to great tip!
What if…?
For example, what if their previous five home games were three draws with other mid-table sides like Swansea, West Ham and WBA, plus two narrow wins against lower-league Bolton and Barnsley in the cups? And what if the two home matches prior to this sequence were defeats to Sunderland and Cardiff? Technically, they are five unbeaten at home, but we cannot infer that they are on a formidable run of form. The commentator has implied a conclusion that is as flimsy as a rag-doll in a rain-storm.
Statistics can be dressed up in a number of ways to suit all kinds of objectives. After all, Motty could just as easily have said ‘Fulham have lost two and drawn three of their last five league games at home’, including the Wigan and Sunderland fixtures while discounting as irrelevant the cup victories against lower-league opposition. No wins in five, versus unbeaten in five, depending on the evidence that the commentator chooses to impart. The facts are the same, however, their context and delivery can be manipulated to encourage the viewer to make a certain inference.
Statistically Speaking, a Lot is Irrelevant
The stats shared by commentators and pundits tend to be deemed worthy of broadcast precisely because of their unusual nature. Facts that are prosaic are not engaging and, simply, don’t get the public’s attention. As such, certain stats that make for snappy soundbites are proffered, with context being manufactured out of largely innocuous, unrelated facts. Despite the fact that they have been framed in a seemingly-meaningful manner, it is important that the viewer understands that not all statistics are relevant.
Which Statistics Can I Trust?
There is no hard-and-fast rule, but I tend to deem as more reliable the stats that are not conjured out of the air. If a number of facts and figures are displayed, rather than a solitary, attention-grabbing one, then it is logical to infer that the statisticians are more interested in trying to give the viewer the full picture, rather than cherry-picking those that will make a lightweight hypothesis appear robust (called selection bias).
When only one is stated, with no context or supporting evidence, then my default stance is one of scepticism until I have delved beneath the surface for more information. After all, ‘Fulham have no home league wins in five’ and ‘Fulham are undefeated at home in five’ are engaging and suggest meaningful patterns. ‘Fulham, a mid-table side, have no real significant run of home form, and anybody trying to splice together a pattern is probably selecting their evidence craftily’ is more accurate, but would probably get Motty into trouble with the BBC. Match of the Day loves a novel statistic, and so should the viewers. However, gamblers should be vigilant against rogue stats: don’t fall into the trap of believing that, because it’s stated on air, it must be significant!
So we have established that plenty of stats are meaningless at worst, and flimsy at best. Do us a favour and help us get our Medium stats up by recommending this article, following us, and spreading the word on Twitter and Facebook ;)
The Sportsrated team write extensively about the psychology of sport, competition, and gambling. Follow them on Twitter, and download their e-book for free.
