It does appear that the traditional UNS methodology for seat projections cannot cope with a Lib Dem surge.
I have done some calculations on a retention swing. That is a calcalation on the basis that in each seat the national transfer of votes is replicated. However, if Labour have lost a third of their votes nationally they lose a third of their votes that were cast previously in each seat rather than the same percentage of the total vote in each seat.
This is a more complex calculation, but it produces more reliable results.
It needs more work, but a first cut of the data is available here
I have done some calculations on a retention swing. That is a calcalation on the basis that in each seat the national transfer of votes is replicated. However, if Labour have lost a third of their votes nationally they lose a third of their votes that were cast previously in each seat rather than the same percentage of the total vote in each seat.
This is a more complex calculation, but it produces more reliable results.
It needs more work, but a first cut of the data is available here
Comments
What I think is needed is to apply swings to some monotonic function of votes, that results in behaviour like uniform swings at low swings, and like proportional swings at high swings. If that makes sense.