logo

Single Transferable Vote 101 (or, why no-one in Britain knows how to report it properly and what they can do about it)

I am a fan of the Single Transferable Vote as a voting system. Being from Northern Ireland and having lived in the Republic of Ireland, I have voted using STV more often than I have first past the post (FPTP).

Naturally, then, I was pleased to see it introduced in Scotland for local elections, which helped some way to break Labour's hegemony over local politics in the Central Belt by ending the one-party state that existed in most councils in Scotland. (It also probably played a role in giving greater force to the SNP wave in recent years, in the Holyrood, Indyref and Westminster 2015 votes, by enabling the SNP to have a vastly increased number of local representatives on the ground.) It will also likely prove to be a life-support machine to the Liberal Democrats, and prevent Labour councillors going the same way in 2017 that their Westminster party colleagues went in 2015.

2007 local election results

However, the mechanics of STV are not that well known to many people, and it has been frustrating for me to read/listen to coverage of STV elections and by-elections as if they were the same as FPTP elections. This ranges from the terminology used to focus on the wrong elements to drawing erroneous conclusions.

A prime example was today when Mike Smithson (someone for whom I have enormous respect) from Political Betting tweeted this:

With all due respect to Mike, there is absolutely nothing wrong in an SNP supporter reporting the result in that way, because from an SNP perspective it is actually more important than the fact they lost the seat.

That may sound a bit odd, clutching at straws even, but it's not, and here's why.

2012 Aird and Loch Ness result Wikipedia

As you can see, in 2012 the SNP won one of the four seats available in the ward. They won it on the first count, with Drew Hendry getting 840 votes, which was 31 votes over the quota of 809. (The quota is, where n = number of persons to be elected and v = number of valid votes cast, (v/n+1)+1.) So in this case, where four members are to be elected, you need to get 20% of the vote plus one vote, ((4044/5)+1) (in this case, 809). The SNP's second candidate got 7.1% - about a third of a quota. So all in all , the total SNP vote was a little shy of 28%. This is equivalent to 1.4 quotas or just over a quarter of the vote. This would, in pretty much any circumstances, be enough to get one of the two SNP candidates elected, but a long way short of enough to get two elected.

In the by-election to replace Drew Hendry, who took Danny Alexander's Westminster seat at the General Election, there is only one seat to be filled, therefore the quota is going to be 50%+1 vote. As you can see, this is rather a lot more than the 28% the SNP got in 2012. In FPTP by-elections, if a party got the same share of the vote as it did at the main election, the party would almost certainly retain the seat, unless alternative support rallied to one of the other parties standing (the sort of by-election special that the Liberal Democrats used to be particularly good at: "Only the Lib Dems can beat the Tories/Labour!" For all its problems, FPTP is, unarguably, simple). Under STV, however, if the SNP had matched its share of the vote, it still would almost certainly lose the by-election, because of the likelihood of unionist parties' transfers favouring each other. To have had a realistic shot at retaining the seat, the SNP would probably have had to increase its vote share to over 40% (a 12 percentage point or 40% increase over 2012). That isn't impossible, but it is unlikely. Therefore, for the SNP to have increased their share of the vote, and still lost, is very respectable.

In short, it is very difficult to win by-elections under STV (which is one of the reasons why in Northern Ireland councillors' and Assembly members' replacements are co-opted members of the same party, rather than elected through a by-election: there is a high probability that the incumbent party may lose the seat, even if it performs better than at the previous election. This is especially true for smaller parties, i.e. anyone other than the DUP and Sinn Féin. (This is particularly the case where a local electoral district or constituency could have 6 or 7 members, meaning that 6% of the FPV could be enough to see someone elected, but their party might need to get at least five times that support level to have a chance of retaining it at a by-election.)

And since I am on the subject, another pet peeve is the focus, when reporting results, on the "majority" over the first candidate not elected (i.e. the runner-up), either on the first or last count. This is a largely useless statistic, as it tells you very little about how the parties performed, and gives undue importance to something that tells you very little about the election. This should be obvious from the fact that the candidate who leads after the first count may not actually win a seat. So if the candidate who is second after the first count goes on to win the seat, do they have a negative majority?

Nor is there any point in reporting, as this does that

She was the winner at the fourth and final stage of the count. She gained 1,511 of the 3,076 valid votes.

This information is pretty useless, unless we know what the quota is. Also, saying she won on the "fourth and final count" is like saying "she found her keys in the last place she looked." Nor would this phraseology make much sense when there was more than one member to be elected, and would therefore require reporting STV by-elections and general elections in different fashions, which is clearly undesirable. A more useful re-write of these sentences would be to say

She was deemed elected on the fourth count, without having reached the quota of 1,539.

The most important information is the First Preference Vote (FPV) share (by party and individual), with a comparison to the previous election. Next in importance is how many quotas each party and individual has on the first count. And finally, transfer patterns are important, but these will vary from constituency to constituency and ward to ward because, well, people are weird and vote on all sorts of criteria that you often couldn't possibly imagine.

Also, it is the norm to talk about the "first count, second count, third count, etc.", even though technically they are "stages", but stages sounds boring and technical, whereas count gives a better flavour of what is going on.

British political reporters really need to get to grips with the intricacies (not "vagueries" - there is nothing vague about it) of STV and stop falling back on what they know (FPTP) when interpreting results. Otherwise, they are doing a disservice to the public. A trip over to Northern Ireland or the Republic to observe how STV elections are covered and analyzed by people who understand the system well would probably be a good idea. So please, no more talk of "majorities".


Me. Just now.

via GIPHY