22 December 2011 Last updated at 06:12 ET

Go Figure: Are we related?

 

By Michael Blastland GO FIGURE - Seeing stats in a different way

Sales of wide-screen TVs have risen, so has male life expectancy. Is there a connection? In his regular column, Michael Blastland asks why we feel the need to join the dots.

One hot summer, hard to recall in this week's chill, Go Figure invited readers to fill in the details of a story about changes in ice cream sales and shark attacks.

When one goes up, so does the other.

This is true. There really are more shark attacks when ice creams sales rise. The question was why?

We didn't want serious explanations, we wanted facetious ones. And boy did we get them.

In stats terms, this is an easy example of mixing up causation and correlation. Two numbers - for ice cream sales and shark attacks - rise and fall together and we ask if one causes the other. The real reason they're in synch is… well, you work it out.

Did uncertainty over the outcome of the UK general election cause the stock market to fall?

I call this the dimputations game, as in dim-imputations. One for all the family at Christmas? You have to be that kind of family. Here's another to try after the turkey.

The fatter turkeys get, the more people go to church. Why?

I'm making these up, by the way, though this one is probably true - as sales of wide-screen TVs have risen at Christmas, so has male life expectancy.

The connection? All hanging on for the next series of Strictly, obviously.

You tell us.

There's an almost-serious point to all this. Namely, how stories about real-world facts can be compelling - even when nonsense. Because they work as stories, they work on our imagination in a similar way to the just-about truth of not being able to out-swim a shark one-handed.

This isn't a dig at fiction. I wouldn't be so unwise as to suggest that art is a poor imitation of life, just that stories about fact in the media and elsewhere often resemble bad storytelling or bad art in the way that they offer us little more than vague ideas suggestively stuck together.

For a sober real-world example, take the period just after the indecisive UK general election when, as readers will recall, everyone felt uncertain about who would form the next government and what policies they would follow. Would it be an alliance of Nick and Dave or Gordon and Nick or - unlikely this one - Gordon and Dave?

Once we had convinced ourselves that the story was uncertainty, the media set out to find evidence to confirm that story. Among this evidence was a fall in the UK stock market. There's uncertainty for you, we said, putting two and two together - markets are nervous following the uncertain outcome of the election.

Except that just about the whole world saw stock market falls that day. Frankfurt, Paris, New York, Moscow, Johannesburg…

All worried about UK uncertainty? One of the few exceptions was Sri Lanka, where the Colombo All Share Index rose, proof I guess that the Sri Lankans took a more benign view about whether Nick and Dave would get it together.

Mind you, there was something going on with Greece at that time.

UK political uncertainty and the UK stock market fall were correlated, though possibly not causally connected. But once we have a story in mind, it's easy to find evidence and impute connections to confirm it.

This is not lying or playing with statistics. It's simply the human instinct to seek links and fill in the details of a plausible story to make it whole and coherent. It is almost involuntary.

Which is why you know exactly what explains the typical rise in heating bills, just after peak consumption of mince pies.