by Dr Adrian Worton
Despite the fact that the 2017 UK General Election was a few weeks ago, it is only recently that we've been able to get the data with the full list of votes for each seat.
With this, we are going to reprise the analysis we did in 2015, and look at how the election would have looked under a proportional system.
Despite the fact that the 2017 UK General Election was a few weeks ago, it is only recently that we've been able to get the data with the full list of votes for each seat.
With this, we are going to reprise the analysis we did in 2015, and look at how the election would have looked under a proportional system.
The method
In 2015 we outlined how the D'Hondt method (explained below) can be used to group current constituencies into larger groups, keeping the same number of MPs. For example, the five Leeds seats (Central, East, North West, North East, West... apparently Leeds is so northern it doesn't have a south side) can be merged into one constituency, which is represented by five MPs.
Previously, we looked at splitting the current seats in different ways, which were:
We found that even just the first two splits dramatically reduced the level to which the results misrepresented the votes cast. Therefore, we will be focusing on these in this article.
If you are interested in seeing how we grouped the constituencies, and want to see what the results of the new groups would be, the full results are shown in the Appendix at the bottom of this article.
The D'Hondt method
We will quickly go over how the D'Hondt method works - to see the explanation click the heading below - if you aren't that interested, feel free to skip it.
In 2015 we outlined how the D'Hondt method (explained below) can be used to group current constituencies into larger groups, keeping the same number of MPs. For example, the five Leeds seats (Central, East, North West, North East, West... apparently Leeds is so northern it doesn't have a south side) can be merged into one constituency, which is represented by five MPs.
Previously, we looked at splitting the current seats in different ways, which were:
- Split 1: seats are combined into groups of 3-5
- Split 2: groups of 6-13
- Split 3: groups of 15-28 (NI is now represented by one constituency)
- Split 4: groups of 29-49 (Wales is now one constituency)
- Split 5: England is split into 7 groups, Scotland is one constituency
- Split 6: England is split into 3 groups
- Split 7: England, Scotland, Wales & NI are each represented by their own constituency
We found that even just the first two splits dramatically reduced the level to which the results misrepresented the votes cast. Therefore, we will be focusing on these in this article.
If you are interested in seeing how we grouped the constituencies, and want to see what the results of the new groups would be, the full results are shown in the Appendix at the bottom of this article.
The D'Hondt method
We will quickly go over how the D'Hondt method works - to see the explanation click the heading below - if you aren't that interested, feel free to skip it.
The D'Hondt Method
For a given constituency, say we want to allocate σ seats. We add up all the votes cast for each party, which we divide by each number between 1 and σ, and select the highest σ values.
It is easier to follow an example. We will use the one of Leeds above. This new constituency will have five MPs, so the totals for each party (the top row in the table below) is divided by 1, 2, 3, 4 and 5.
It is easier to follow an example. We will use the one of Leeds above. This new constituency will have five MPs, so the totals for each party (the top row in the table below) is divided by 1, 2, 3, 4 and 5.
We simply then look at the table and pick the five highest values. These happen to be the top four in the Labour column, and the top one from the Conservative column.
This means that in this system, Leeds would be represented by four Labour MPs and one Conservative MP. In the real vote, Labour took all five seats.
This means that in this system, Leeds would be represented by four Labour MPs and one Conservative MP. In the real vote, Labour took all five seats.
The results
Below you can cycle between the results of the election - the actual results, then the results using our PR method for split 1 and split 2. Additionally, you can see an overview of the results for all splits together.
Below you can cycle between the results of the election - the actual results, then the results using our PR method for split 1 and split 2. Additionally, you can see an overview of the results for all splits together.
The interesting thing to take away from this is that under no PR scenario can the Conservatives make a majority, assuming that the Lib Dems learn from 2010 and don't support them.
Of course, if an election was run under PR, then the results would be wildly different to those cast under first-past-the-post, due to tactical voting. For example, any voter who voted Conservative, when they really wanted to vote UKIP, would be free to vote how they wanted.
However, we can assume that the balance of left- and right-wing voters would remain fairly similar (for example, whilst we'd expect plenty of votes to go from Labour to the Greens, both are essentially anti-Conservative).
Summary
By repeating our study of the D'Hondt method in UK elections, we can see that the result is significantly different, no matter how constituencies are grouped.
Most notably, under any proportional system, it would be highly unlikely that the Conservatives can form a majority.
Many people see such an outcome as the sign of a weak democracy, as we are told that a government having a majority is a strong government. However, most European nations rule through coalition - for example, Angela Merkel's party in Germany only hold 254 of 630 seats, but work with two other parties to have a combined 503 seats. And Germany hasn't done too badly lately.
Of course, if an election was run under PR, then the results would be wildly different to those cast under first-past-the-post, due to tactical voting. For example, any voter who voted Conservative, when they really wanted to vote UKIP, would be free to vote how they wanted.
However, we can assume that the balance of left- and right-wing voters would remain fairly similar (for example, whilst we'd expect plenty of votes to go from Labour to the Greens, both are essentially anti-Conservative).
Summary
By repeating our study of the D'Hondt method in UK elections, we can see that the result is significantly different, no matter how constituencies are grouped.
Most notably, under any proportional system, it would be highly unlikely that the Conservatives can form a majority.
Many people see such an outcome as the sign of a weak democracy, as we are told that a government having a majority is a strong government. However, most European nations rule through coalition - for example, Angela Merkel's party in Germany only hold 254 of 630 seats, but work with two other parties to have a combined 503 seats. And Germany hasn't done too badly lately.
Appendix - constituency selector
You can use the widget below to select a constituency, using the drop-down list on the grey cell. You will then see the name of each group it is part of for each split, and how each group would have distributed its seats.