First of all, I start with the demographic composition of the fourteen constituencies taken together. The chart below shows the proportion of each of 69 demographic groups in the fourteen seats (coloured bars). The hollow bars are the proportion of each of these demographic groups across the whole of England & Wales. I have described these 69 demographic groups elsewhere on the blog. See this link: http://election-data.blogspot.co.uk/2014/04/just-who-is-voting-for-ukip-part-ii.html if you want the full descriptions.
Before I go on to describe this composition I want to present the most recent voting behaviours of these demographic groups, in the same type of charts. These following charts show how the vote share for each of the main parties was correlated with demographics in the 2014 local elections. Of course there needs to be some caution attached to these correlations as the local elections are low turnout elections. However the correlations are broadly in keeping with election results from 2005, 2010 and 2012. I will take each party in turn. The scores above the x-axis show a positive correlation (i.e. vote share for the party went up) and scores below the line show a negative correlation.
These charts, taken together, give us some clues to the groups which are being drawn to UKIP from Labour. For example, take a look at the K50 demographic group. This group is made up of older families in low-value housing in former industrial areas. Geographically this group is found most strongly in places like Rotherham, Doncaster, the West Midlands, south Wales, parts of the North-West and North-East. In other words, Labour areas. Unsurprising therefore that they vote Labour (see chart), but a good proportion of them are voting UKIP.
Or how about E21s (Middle aged families living in less fashionable inter war suburban semis)? Some of this group are voting Labour, but many of them are going over to UKIP. Again this group is concentrated in the Midlands, North West, south Wales and the North-East. There are twice the national average of this group in Stockton South for example.
Without any further ado I want to display the demographic profiles of the fourteen Ashcroft marginals, in alphabetical order: