[openspending-dev] Measuring coverage (was: Re: Data)

Tue Mar 6 14:42:54 UTC 2012

Hey,   

On Tuesday, March 6, 2012 at 10:36 AM, Lucy Chambers wrote:

> Is there any way that we can help Paul with this query? This is something I've been thinking about for a while. Is there an easy way e.g. that we can draw up a matrix or something with time along one axis, dataset along another, then showing e.g. 100% coverage of all transactions over 25k, partial coverage, little -> no coverage?  
>  
>  

Yes, we need to develop a useful way to talk about coverage. Of course this problem is not specific to spending data, but lets try to come up with a local solution first.  

For the EU's data, I'd previously made this Matrix: https://docs.google.com/spreadsheet/ccc?key=0AvoV_cBqwo28dE8wRWV4YzV4QUlKRWpuSTBZaWQyR3c#gid=3 - shall we make a similar one for the UK as well? Of course it would be cooler to get this into OpenSpending itself, but I think we first need to find out what the specific pieces of metadata are that could be most useful to give:  

* Start and end of reporting period
* Geographical coverage
* Sectoral/topical completeness (is this only about agriculture?)
* Reporting bodies providing data
* Granularity
* Censorship

This is interesting because we'd end up with a pretty large block of info for many cases. For the UK departments, we'd essentially need to have a big table with every public body in the UK for each month after Nov 2010 and each entry can have a couple of values (exists, incomplete, unparseable, …).
> If journalists can't tell what we have, and whether it gives them a complete picture of the state of affairs, then I don't think they are going to report on spending. :(  
>  
> Alternatively, could we spin this the other way and get Paul to do a story on how it is impossible to write about spending because of the incompleteness of the data (unless this is our fault...)... ?  
I think its a mixed thing, most of the open data projects collecting this stuff (Chris, and us on a CG level, as well as other projects in differnet countries) have better and more complete data than whats available to many governments. Not sure if a story about the everyday drama of bad gov IT works though, although I'd read it :)  

- Friedrich  

>  
> Lucy  

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.okfn.org/pipermail/openspending-dev/attachments/20120306/ec7363aa/attachment.html>