I have been reading the Harari, Krauss/Midford correspondence about the election with interest.

I have been talking to an Oxford doctoral student based at Nuffield college, Roberto Cerina, who has done a forecast of next week's election based on the polling evidence that readers of this forum might find interesting. His results show:

The probability that the Abe led coalition will gain a majority of the seats is around 0.87, making it extremely likely that Abe will be returned as PM.
The probability that he will receive less seats than last time is given at around 0.8, making it very likely he will be returned to a shrunk majority (it will be interesting to see whether the shrinking is due to a fall in support for the LDP or Komeito, something he does not analyse here).

The most likely share of the proportional (block) vote to go to the LDP is 47%; similarly, 47% of the district vote should go to Abe's party. This is interesting as it suggests the LDP will hold on the block level, but will bleed around 2.5% at the SMD level.

The over-time plot shows the optimal forecast for seats given the data at hand at any point during the last year. This suggests the probability of losing a majority seemed high between 4 months and 2 months out of the election, but has significantly recovered since.

Conclusion: Abe will return, he will have a majority, and it will be smaller than his previous one.

I am sure that he would be happy to show his working and to discuss his method etc with any interested parties.

Ian Neary
Nissan Institute
Oxford University