Thread by @PascalLTH

If the EP elections use PR, is tactical voting actually necessary?

People seem to be wondering this and lots of people are saying no. That's the wrong answer and here's a thread to explain why /1

PR isn't just one system. The basic idea of any PR system is that the votes cast should match the seats won, but how that gets implemented varies significantly between systems. Not all systems are totally proportional /2

For EP elections, the UK uses the D'Hondt system when allocating the seats. How does this work? Effectively seats are allocated using successive quotas, until you reach a fraction of a seat /3

This means that there is a lower limit to how many votes you must win in order to win a seat. There is a quota that needs to be hit every time. /4

Imagine there are 10 seats, 5 parties and 200,000 votes cast. The first quota would be 200,000/2. (which equals 100,000). So imagine only one party hit above 100,000 - that party would win one seat. /5

No one else gets any seats because they don't meet the quota. So now you lower the quota by adding 1 (ie, 200,000/3, which gives you 66,666). Now the first party obviously reaches that so they get another seat and then any other party that made the quota will get 1 seat /6

This continues until all seats have been allocated. What do you notice here? Seats are allocated from the top down, with the parties with the biggest scores getting first pick /7

It's therefore entirely possible for those lower down the scale to not win any seats at all even if they win a respectable percentage of the vote /8

To give a more practical example: in 2014 the Greens and Lib Dems won around 13.5% of the vote nationally. Now, if you looked at that and applied a simple formula of 'it's PR therefore it's proportional', their combined seat total should have been 9 seats /9

In reality, their vote was split, meaning they often were too far down to qualify for any seats, and so only ended up with 4 (3 Green, 1 Lib Dem) - they could have won more seats if tactical voting had been used /10

To add to this, you have to keep in mind that the UK is divided into regions for the EP elections (12 regions to be precise) - concentration of votes in particular areas therefore does affect how well a party does, just like it does in a GE /11

So you might get a good score nationally, but if your vote is too spread out then it won't win you many seats /12

The big danger facing us is that we have a number of small remain parties who split the vote across a number of different areas. The negative effect won't be as extreme as with FPTP but Remain parties would still lose out on seats due to not working together /13

So the lesson is: yes, it's PR but yes, tactical voting *does* matter and everyone needs to be aware of this and planning ahead to ensure Remain votes lead to Remain seats /end

@StevePeers @GuitarMoog @ottocrat

PS. as a very rough guide, and based on the 2014 results, if you're below 8% in one of the UK regions, chances are you won't win any seats (only a very rough guide because it does depend on how well other parties do but it gives a sense of the challenge)

It also depends on how many seats are available. The most extreme example from 2014 is North East England.
It elects 3 seats and the results were:
Labour: 36.5%
UKIP: 29.2%
Tories: 17.7%

Two of the seats went to Labour, one to UKIP. Tories got 0.

Reading some comments, I've slightly misrepresented how the system works but rather than redoing the whole thing, let's keep it at: votes are allocated from the top down and there is an effective threshold for smaller parties (ie you can split the vote)

Basically there is a series of successive quotas as I described above but the method to calculate it is different to what I wrote - fortunately doesn't change the point being made

So for those interested, the correct explanation is that the quotient is set by the number of votes each *party* wins, not the total number of votes as I said earlier

So you take all the party totals and divide them by 1 to start. The biggest party is obviously the winner and so they win a seat.

Having won a seat, you +1 ti the divisor. So now that first party's vote share is divided by 2

Now we look at the next biggest party (which has no seats and so is still divided by 1), is their vote share bigger or smaller than First Party÷2? If it's bigger, they get a seat, if smaller then the first party gets a seat again (and gets +1 to its divisor)

This continues until all seats are allocated

The EP explainer shows quite well how this results in having to shift around repeatedly between parties when allocating seats and how the party at the bottom can be left with nothing:

Again, doesn't change the argument, but a frustrating mistake on my part nonetheless

For those jumping in from the top, I didn't explain the allocation method correctly in the thread, added correction at the end:

https://twitter.com/PascalLTH/status/1116218577927979008?s=19

This is also a good visual explainer

https://twitter.com/EmmaOverington/status/1116220849055514630?s=19