Somehow this passed me by but http://local.uk.msn.com/ now includes some socio-economic data at a national level.

The crime information is worth a look and appears created from Experian data using the British Crime Survey rather than actual crime occurrences – then interpolated some how into a surface. Additionally, when you click the map, the data returned includes a series of the long descriptive profiles for the Mosaic Types or Groups within the “area” (however defined).

I really do worry about these types of commercial representation, specifically given the lack of detail over the methods used and the potential consequences of their erroneous interpretation. The information reported on this page about the crime data: http://money.uk.msn.com/MSN-Local/help.aspx#C appears to be all about perceptions of crime, which is very different from actual crime – as specified on the map. I am guessing that the crime map is created by taking the British Crime Survey, appending Mosaic, modeling weighted values for “perception” / fear of crime – then taking these values at postcode level and interpolating between them, probably with IDW into high / low scores.

The potential for creating spurious values is HUGE given so many uncertainties – at some point in this operation, a categorical value from a black box geodemographic is used to model a continuous point score, not to mention the fact that further values are then interpolated between these points in the conversion to a raster surface. To me, this really doesn’t sound like a set of plausible operations – why not just plot the crime domain of the IMD like we do on LondonProfiler?