I allude, of course, to the famous categorisation of lies, damned lies, and statistics. Here is a Telegraph subheadline:
NHS doctors are failing to inform up to half of families that their loved ones have been put on a scheme to help end their lives, the Royal College of Physicians has found. (src)
Now, you might think that this was a widespread problem. But in fact, the sub-editor is lying prodigiously.
Firstly, the Liverpool Care Pathway is not a 'scheme to help end life'; it is a palliative care programme designed to give dignity to those who are dying. The sub-editor has phrased this to sound like some kind of crypto-euthanasia, when in fact it is quite the reverse. But what is worse is the blatant twisting of figures. The body text, on which the sub-editor is meant to base their top matter, tells us,
In one hospital trust, doctors had conversations with fewer than half of families about the care of their loved one.
'In one hospital'! A single datapoint! Writing the subheadline on the basis of one datum is not simply poor statistics, it is in fact lying: it is outright damned lying. You could take a single uninformed family and make the claim that up to 100% of families are not informed; or a single informed family, to claim that up to 100% are informed. It is every bit as (il)legitimate.
The real story is that on average no more than six percent of families are not informed. (Since the count is 94% of families are
documented as informed, the figure could be lower than six percent.) That is still, clearly, six percentage points too high (
xkcd) but it is nowhere near the 'up to half' claim. Still, what does the truth matter when you have a subheadline to write, eh?
4 comments:
Vincent Tabak (here) presumably only committed one murder. One would hesitate to call that a "single data point".
I don't see the relevance. What's the statistical claim for which Tabak might be a subject? You wouldn't, surely, make a general claim about murderers on the basis of Tabak alone.
I think the key here is "up to". Depending on which hospital you attend, your relatives have "up to" half a chance of not being informed.
No, it's not great but it's not outright lying, people must have become pretty savvy to 'up to' language since every shop in the world started using it to trumpet price reductions.
What's half a chance? "If I 'ad 'arf a chance I'd give 'em wha' for!"
In Brighton I saw buses that claimed "up to every 6 minutes" for a certain route. I've no idea what that means. It's not even wrong (especially when three come along at once).
Post a Comment