A pet peeve, and believe me, there are plenty, is the use of statistics in newspaper journalism. Basically if you see the word 'increase' or 'percentage' written anywhere, it's generally safe to assume the paper has it wrong. You'd expect this in the tabloid papers, but this report comes from the Guardian.
So yes, I'm ranting against the Guardian.
Titled, on the website, Sharp Rise in Number of Women Killed by Violent Partners (paper version: Murders of partners by men rise sharply), the article claims that there has been a dramatic rise in deaths by partners. Reading the article provides the following:
2009 - 101 deaths
2008 - 72 deaths
As a raw percentage increase, this is, on the face of it, shocking: a 40% increase over one year. Which is obviously how the refuge charity 'Refuge' was trying to sell this.
However, let's put this in context, looking at the UK population figures (reference here):
2009 - 61,792,000
2008 - 61,398,000
Putting the death figures as a percentage of the total population:
2009 - 0.000163452 %
2008 - 0.000117268 %
I realise I shouldn't be using percentages here, but without the raw data to work from, it should do to make my point.
There is very little difference between these two figures, In fact, if I round to 3 decimal places, they are identical. This is due to the total UK population being so large that a number of 101 becomes insignificant from 72.
Indeed, run any statistical analysis of those numbers and you'll find there is no statistical difference between the two years.
there has been no dramatic rise in murders by partners when put into context of both population. What has happened is that there has been an increase in absolute numbers.
What might be more interesting would be to analyse the increase over a longer time period, as the data, as presented, can be interpreted as ' a 0.6% population increase appears to lead (but not cause?) to a greater increase in partner murders', or you could interpret the data as 'a 0.6% population increase leads to a 0.00004 (rounded)% increase in partner murders'. What you cannot say is that there has been a dramatic rise in murders.
B
10 December 2010
Subscribe to:
Post Comments (Atom)
2 comments:
Well said... the same actually applies to the road toll... stats only seem to be used to drum up revenue....
TLP
the nz road toll is, statistically, far more important. it's around 400 from a population of 4m, which is 0.01% of the population. Far more significant than the number of women killed by partners.
moral: in a relationship in the Uk is safer than driving in NZ.
Post a Comment