Tactical voting

[jack]

I think I'm sufficiently decided about the euro elections, but I had some left over thoughts.

D'Hondt method

Have I understood this right?

Suppose there's 7 seats available for a number of parties. Then every party whose vote share exceeds 1/7 of the votes, is guaranteed at least as many seats as they have whole sevenths of the vote share? That'd accord with what I'd do so far.

But if I were running it, then I'd say, give the remaining seats to the parties who are closest to one more seat than they already have (whether that's 6 seats but nearly 7 seats, or no seats but nearly 1 seat). But D'Hondt counts your remainder more, in proportion to how many seats you already have?

Why on earth is it like that? Is it supposed to avoid niche parties? Or is it just that it's longstanding and superficially reasonable, so when the voting method was chosen they chose the most FPTP like proportional system they could get away with?

East of England

There were several helpful "tactical voting guides" going around, but several seemed to concentrate more than I'd originally realised on "trust our subjective judgements without asking too many questions" rather than on providing data.

My natural choice would be Lib Dem or Green, mostly as a clear "Remain" vote. Which I prefer depends on circumstance, but I was also trying to figure out if it would make a difference if I wanted to increase the chance of either winning a seat.

Previously Lib Dem have held about 15% of the vote and I think got an MEP, although last euro elections they were at a lower point. The greens have previously been lower, but have been creeping up over time, especially if people are finally starting to pay SOME attention to mitigating climate change.

So it sounds like I'm right, if lib dem are around 15% and greens are lower it would be worth voting lib dem to make sure to push them up over the 1/7 boundary? But if lib dem were already higher than that and green were approaching it, it would be better to vote green to get *them* closer to a seat (unless lib dem were doing so well they were close to 30%).

But that's sufficiently precise it's almost pointless to try to predict, so I should go for whichever I prefer, or whichever I think has the highest support?

Comments

[naath]

If party A had 1000 votes, party B 900 votes and part c 300 votes.

A wins, and gets MEP 1; their vote is now 1000/2=500
B now wins, gets MEP 2; their vote is now 900/2=450
A wins, gets MEP 3; now 1000/3=333 1/3
B wins, gets MEP 4; now 900/3=300
A get MEP 5; now 1000/4=250
B get MEP 6; now 900/4=225
C get MEP 7 since 300 is > 250 and 225.

D'Hondt is annoying. I don't know what you meant by 'closest to one more'.

If A had 100 and B had 600 (so A has 1/7th) B's total is 600 (win 1), 300 (win 2), 200 (win 3), 150 (win 4), 120 (win 5), 100 (er, tied for 6 & 7, I don't know what the tie-break is but either way one each).

If 7 parties each had approximately 1/7th the vote they would each get 1 MEP and then be sent to less than anyone else's votecount.

So yes I think 1/7th guarantees you an MEP, but you clearly don't need 1/7th if the vote is very split. It is known to favour large parties, and the list system favours internal party folks who are likely to be part of/pay attention to list-deciding systems.

[andrewducker]

There are two kinds of ways of dealing with proportional systems - Highest Average (like D'hondt) and Largest Remainder (Hamilton).

Largest Remainder (which I think you're advocating) is good in some ways, but can suffer the Alabama Paradox, where adding a seat means that some of the parties get _less_ seats.

See https://en.wikipedia.org/wiki/Largest_remainder_method for a good breakdown of the good and bad sides.