Posts Tagged ‘statistics’

I really enjoyed an Editorial in the Guardian about the unreliability of statistics – here’s the phrase which really made me laugh –

Research in 2005 suggested that only 36% of people think official statistics are accurate

I don’t know, what do you think……..do you believe it?

Read Full Post »

I have a real love hate relationship with statistics. I’ve heard it said that all men love figures. Well, I’m not any different that way. Numbers interest me. On the other hand, however, I hate statistics! Actually, that’s not really true. What I hate is when statistics are given as “the truth”, or are given greater weight than human experience. My preference is always for stories. I am completely hooked on stories. The work of Gigerenzer really impresses me and I was reminded of his work yesterday when I read a piece on the BBC site about statistics. The piece is written by Michael Blastland and I enjoyed his style. He started off by picking up on a news item which claimed that vitamin E increased the risk of death by 14%. As he rightly points out the risk of death for all of us is 100%, so what point was the journalist trying to make? That the risk of death if you take vitamin E is 114%?! The thrust of the article is that we need to reconnect figures to human experience. He suggests we do this in two ways –

First, we need to remember that not much in life is either/or. According to the research, there’s something in the claim that Vitamin E supplements can be harmful. But, as with the consumption of salt, or even water, much that can kill is also essential to good health. The world does not divide easily into what’s toxic and what’s not, what’s safe and what isn’t. Risk is simply a way of measuring where we stand on the messy middle ground – which is almost everywhere. What matters in that messy middle is the relevant human quantity: how much supplementary vitamin E? A little won’t do any harm (or, probably, much good). A lot, especially if you are getting on in life, might. So a 14% increase in risk of death does mean something, but only if you say at what dose (high), for which group (the elderly), over what period (a single year, not in a lifetime).

The second common problem with any percentage increase like this, also crying out for a dose of real life is: what’s it increased from? Because 14% might be a lot if you start somewhere big, next to nothing if you start somewhere small. A 100% increase from one in a million becomes two in a million. So what? A 100% increase in the number of bullets in a revolver – if you are playing Russian roulette – well, that makes a difference.

I loved his concluding paragraph –

A percentage is not really a number, it is a share. The simple question to keep in mind is one that always strives to put it into a proper, human context: “A share of what? A share of a lot – or a share of a little?” Better still: “A share of who?” Keep it real.

Read Full Post »