Lies, dam lies and statistics

I came accross this story this morning (Hospitals 'infested with vermin'). Apparently Nottingham University Hospitals NHS Trust is the dirties in Britian but my comment isn't about that it's about what's behind the figures, where they came from.

So the consertive party in an attempt to stick it to Labor over the state of hospitals (so there's the motivation for the spin in the numbers) used the Freedom of Information Act to find out how many times each hospital had called in pest control. On the surface a useful enough measure but different hospitals and trusts are leikly to have different policies regarding pest control so my question would be how can we tell that those at the top of such a list aren't just 'normal hospitals' with a paranoid policy?

But the thing they did which struck me as really strange and serving only to inflate the numbers for the purpose of what they were doing was they totaled up these numbers to present as trusts rather than hospitals. Now if every trust had the same number of hospitals that wouldn't be so bad but they don't. A smaller trust with slightly more calls per hopspital is leikely to end up looking better than a slightly larger trust with less calls per hospital. So it makes me wonder just how different it would look if they used calls per hospital for each trust rather than calls per trust.

Just goes to show you make statistics say anything, the most important part of the numbers is knowing where they came from.