“England World Cup success sparks water shortage” - As headlines go (which I’ve clearly made up), I think we’re all in agreement, it’s total nonsense. But this sort of headline, or indeed conclusion, is what we do in the media industry on what seems a regular occurrence. Two unrelated events correlated together to drive cause and effect.
As with everything it seems, there is a nice Latin phrase for this, ‘Post hoc ergo propter hoc’ (which incidentally is the title of an episode of the rather brilliant West Wing), which roughly means ‘After it, therefore because of it’. It refers to a logical fallacy that we all succumb to. Simply put, if A goes up, and B goes down, therefore A caused B.
It’s an easy mistake to make, at university I tried to create a model for predicting football attendances based on various factors. It was an unmitigated disaster needless to say (I still have the model if you’re interested) but amusingly, after assessment of all independent variables, fanbase, position in league, goals scored in previous matches etc, the variable that correlated most to gate attendance was the price differential of the pies. The higher the price of the pie, the higher the attendance. Cause and effect. Except of course, it’s total and utter claptrap, and one I clearly did appreciate at the time.
But what makes us want to draw these conclusions though, especially in media? Well, in a world of logic and order, we simply can’t help ourselves. We constantly look for cause, drivers, influencers and for effects. If some behaviours change, then we will leap on whatever contrary behaviour data is available and link the too.
In part, this is because in media we are now awash with more data that we can potentially even comprehend, so of course the opportunity to imply effect on a correlation is only going to get bigger. Where researchers used to sit scripting questionnaires, there are now teams of analytics professionals trying to decipher the millions of data points available to them to understand the consumer. And they are doing some amazing things it has to be said.
But it’s probably in the world of media behaviours where these spurious correlations happen the most, and television often bears the brunt. Headlines came out of the US recently linking the 44% decline in World Cup TV viewing to cord cutting and the rise in online streaming services. Ask any football fan of course and they will kindly point out that the USA weren’t in the World Cup, not to mention the time difference to Russia (compared to Brazil previously) might be an impact too. In the UK we have a constant battle to infer cause and effect on rapidly changing behaviours, mostly from younger audiences. The increase in take up of one type of media platform is the acclaimed death knell for another, as if all choices and behaviours are binary and directly linked.
The truth is simply that we often don’t know. Perhaps cord cutting in the US is driving declines in viewership to the World Cup, but two data points does not make an effect and two disparate data sources at that. And this brings up a continuing problem in measurement, the desire to measure everything across all platforms, devices and media, means we have multiple different measurement techniques collecting data in different formats, to different standards and with different intentions. Measuring cause and effect just got even more difficult.
In the Netherlands, the respective measurement bodies have launched a request for proposals for a single Total Media Audience Measurement service, a desire for a collaborative approach to media measurement, covering readership, listenership and viewership all together in a single service. My initial reaction was that they were insane. How will it work and what it will provide? How on earth will everyone agree? But against the background of my own negativity, it’s inspiring to see such collaboration across competing media and I look on with great interest.
To be fair, we have seen movement towards collaboration on this side of the channel too, PAMCo being a case in point with the integration of UKOM data within readership figures to put print and ‘digital’ side by side. BARB too has worked collaboratively with JICWEBs to use industry standards in online video measurement to ensure like for like comparability in viewership to traditional broadcast. It’s not quite the courage of our Dutch friends but there are signs that as media and consumption become ever more intertwined, collaboration in measurement will become more not less important, especially as we try to understand the true cause and effects of media choices on one another.
We all have a responsibility to play; to make measurement accessible, comparable and understandable. Otherwise we run the constant risk of misleading conclusions and correlations that are damaging to our clients and the industry as a whole. At the end of the day, never discount the stupidity of someone mapping football attendances to burger van prices.
