The CSO has released a multivariate analysis of the 2007 National Employment Survey which shows broadly similar results to the recent ESRI study (which used data from the 2003 and 2006 surveys). I imagine that next week’s ESRI seminar should stimulate some interesting debate on this important topic.