5 Comments
User's avatar
eschatom's avatar

How much does lumping the remaining 5% together as "the rest" distort the "true" effective number of parties? Can it be bounded?

Expand full comment
Oliver Johnson's avatar

It's a good question! There are bounds, but they might be quite conservative. That is I think if there is probability p shared between m small parties, the worst case would be that you are throwing away p log m from the entropy. But that's conservative because it's based on the possibility that each of those small parties has an equal share of the residue, whereas in practice I think there may well be some sort of power law drop off in vote share among the more minor parties

Expand full comment
Julian Wells's avatar

Neat stuff, but the UK consists of three countries and one statelet, in some of which there are locally major parties that don't contest seats in the other three countries.

Hence, I'd like to see this analysis done for each of England, Wales, Scotland and the north of Ireland.

Expand full comment
Oliver Johnson's avatar

Yes, I agree it would be interesting, if someone has the time to do that

Expand full comment
Yuly K's avatar

I’d love to give that a go! Sounds like just data-hunting and spreadsheet-fiddling, so if I ever get unburied from studying maybe I’ll take a crack and send it over)

Expand full comment