Guardian Datablog – data is nothing without analysis

David Brindle posted NCVO’s charity data to the Guardian’s Datablog on Valentine’s Day (gotta love a charity, right).  The data are available by local authority.  When the page loads correctly, you can click on the data table to sort the data by ‘charities per 1000 population’.  The top two local authorities are City of London (118.7 charities per 1000 population) and Westminster (11.34 charities per 1000 population).  Of course, this doesn’t tell us much – both the City and Westminster are home to many charity head-offices (and Camden is high on the list for similar reasons).

City of London also has a risible population of about 7000 souls – many charities spread over few people makes for a big number.  In fact there are quite a lot of low-population local authorities (LAs) at the top of the list.  Six LAs are in the top 20 for charities per 1000 population and the bottom 20 for population size.  So…  Are charities pretty evenly distributed once you get outside London (i.e. much of the variation here is due to the population of the LA not to number of charities per LA)?  There’s plenty of variation to explain: the minimum is 15, the maximum is 2676 and the mean is 386. 

The thing that bothers me about a list like this is how little it can explain.  It’s impossible to draw satisfactory conclusions from this data alone.  It’s great, fabulous, wonderful that The Grauniad is drawing it all together – but it doesn’t really tell us much.

Edit: John Mohan pointed me in the direction of David Kane’s NCVO blog.  His recent post about the data available on the voluntary sector transition fund is another great example of how an open data policy can obscure as much as it reveals.

This entry was posted in Government, Journalism and tagged , , , . Bookmark the permalink.