London Murders: Stats Theory Shows Numbers Are Predictable

Found an interesting report on my science RSS feeds this morning, which ties in with something which tends to irritate me slightly about how the media runs with “scare stories”.

We all tend to focus on big one-off events, but in doing so tend to ignore the smaller more frequent events which have a bigger overal impact.

For example, if a multiple car accident occurs, the media will be all over it – but the death rate in the accident would actually not be noticeably higher than the daily average for road culls in the UK – which is about 10 deaths per day.

What is it about 10 deaths in one place which makes it more newsworthy than 10 deaths dotted around the country?

Are the other 10 people not worthy of our attention?

Obviously, if the media reported the news based on statistical impact – then every newspaper front page would proclaim the deaths of dozens of people every single day – a “Baby P” every week (yes, they actually are quite regular events), etc.

OK – Joseph Fritzel is an exception ;)

Anyhow, in the release I read, statistician Professor David Spiegelhalter claimed that the number of murders in London last year was not out of the ordinary and followed a predictable pattern. Spiegelhalter’s report, published today in Significance, the magazine of the Royal Statistical Society, argues that shocking headline numbers are not as surprising as one might think.

The study, which looked at the pattern of murders between April 2004 and March 2007, also shows that on around 64% of days we can expect there to be no murders at all. Each year the London Metropolitan Police record around 170 homicides and there has been little change over the last five years.

You can read the full press release (warning – the study was funded by the government, so caveat emptor may apply) at AlphaGalileo

Whats's on in London: today or tomorrow or this weekend

Posted in Politics, Science Tagged with: , , ,

Leave a Reply

Your email address will not be published. Required fields are marked *