Hoverable Dropdown

Hoverable Dropdown

Hoverable Dropdown

Hoverable Dropdown

Hoverable Dropdown

Hoverable Dropdown

Hoverable Dropdown

Hoverable Dropdown

Friday, 3 July 2020

The Single Transferable Vote has arrived. What does this mean for the Labour Left?

Earlier this week, the Labour Party's National Executive Committee (NEC) voted in favour of changing the way in which Constituency Labour Party (CLP) reps on the NEC are elected. In this summer's elections, the Single Transferable Vote (STV) will be used for the first time to elect the 9 CLP reps on the NEC.

In this article I'll look at what STV is, what it means for the election, and discuss what the Labour Left's strategy should be.

In short:

  • STV does not mean that vote splitting no longer matters
  • STV does not mean that everyone will just transfer their vote to other Left candidates

We should not be relaxed or complacent about these elections, because if we fail to adjust our strategy to account for the new electoral system, there is a serious chance of us winning no seats on the NEC whatsoever.

What system have we been using?

Prior to 2020, the CLP reps on the Labour Party NEC were elected using a system known as the "Bloc Vote". Under this system, which is used to elect many local councillors in England and Wales, voters have as many votes as there are positions to be filled: so in the case of the NEC, members would cast 9 votes for 9 different candidates. The 9 candidates who recieve the most votes would then win.

So, what is STV?

The Single Transferable Vote is an electoral system that is very similar to the Alternative Vote (AV), the system that Labour uses to elect its Leader. With AV, voters rank the candidates in order of preference: 1, 2, 3, 4 etc. Under AV, if no-one wins over 50% of 'first preferences', losing candidates are eliminated, and their voters' preferences redistributed, until one candidate wins over 50% of the vote.

However, whilst AV is used to elect individual positions (like a Leader) STV is instead used to elect multiple positions at once - in this instance, to elect 9 NEC members at the same time.

Instead of needing over 50% of the votes to win, like with AV, candidates in an STV election must win a minimum number of votes in order to be elected - this is known as the 'quota' . In most cases, this quota is calculated by looking at the total number of first preference votes, then dividing it by the number of available seats, plus 1.

So, for example, if 304,332 members voted in the NEC election, the minimum number of votes needed to be elected would be 30,433.2 votes:

304,332 / (9+1) = 30,433.2

If 9 candidates all win the minimum number of votes solely from first preferences (in this case, 30,433) then they are all elected. However, if this does not happen, then voters preferences are redistributed until 9 people do win the necessary number of votes. This is done through redistributing the votes of candidates who came last in each round of counting, and also through redistributing the votes of candidates who have already won - though votes for candidates who have already won are weighted at a smaller value to ensure that their vote does not count twice.

If this seems complicated, that's because it absolutely is. There are countless ways to count STV. For more information check out the Electoral Reform Society's briefing.

STV is used to elect local councillors in Scotland and Northern Ireland, as well as to elect the Northern Ireland Assembly. The Republic of Ireland also uses it for local, general and European elections.

What does this mean for the NEC elections?

The immediate, definite outcome of STV being introduced is that no faction will be able to win all (or even a majority of) the 9 CLP seats that are up for election. For comparison, in the 2016 elections, the 2017 by-elections and the 2018 elections, the Labour Left won all of the available seats; in the 2020 by-elections, the Labour Right won both available seats.

In short, switching to STV means that a faction won't be able to win all of the seats simply by winning 31% of the popular vote (as the Labour Right did in the 2020 by-elections).

However, this does not mean that the Labour Left doesn't have to worry about splitting the vote.

What does this mean for the Labour Left?

Splitting the vote

In the 2020 NEC by-elections, there were at least three different leftwing slates standing 5 different candidates. This meant that there were 5 leftwing candidates running for just 2 places on the NEC, and the Left vote was split.

As a result, although 46% of voters supported leftwing candidates and just 31% supported the Labour Right, the Labour Right won both seats. Why? Simple: only two candidates from the Labour Right stood, and they both won 15.6% of the vote. Meanwhile, because the Left vote was split five ways, no leftwing candidate managed to win more than 15.5% of the vote - and the Labour Right won both seats.

Lots of leftwingers have been arguing that this is no longer a problem, as STV allows members to transfer their vote to another leftwing candidate if their preferred candidate doesn't win. Lots of leftists now seem relaxed about the possibility that multiple leftwing factions could all run 9 candidates of their own for the 9 positions. But this would be an absolutely catastrophic mistake.

Contrary to popular belief, STV does not prevent vote splitting - it simply makes it more complicated. Let's say that 40 candidates stand for the 9 CLP places on the NEC:

-> 20 leftwing candidates from various factions
-> 11 independents
-> 6 candidates from the Labour Right
-> 3 candidates from the Soft Left

With 9 seats up for election, the quota would be around 10% of votes cast.

Even assuming that the levels of support for each faction are the same as in the by-election, this would mean that the Labour Right's 6 candidates would win around 5% of first preferences each - because their 31% share of first preferences would only be split six ways. The Labour Left's 46% share of first preferences, however, would be split twenty ways.

Why does this matter, you ask? After all, won't most of the leftwing members have given their 2nd preference to other leftists? Well, maybe. But even if they did, the sheer number of leftwing candidates splitting the vote would mean that most (if not all) would poll fewer first preferences than the dozens of independents, soft left candidates and other candidates that we can expect to stand.

As a result, those leftwing candidates would be among the first to be eliminated when preferences from more unpopular candidates start to be redistributed. And so they won't benefit from any second or third preferences later on - because they'll already have lost.

Here's an example that helps to illustrate this. As stated above, Ireland uses STV in general elections. In the 2011 Irish general election, the results in the 5-seat constituency of Wicklow were:


Fine Gael 40%, Labour 17%, Fianna Fail 11%, Sinn Fein 10%, Independent 9%, Others 14%

As you can see in the linked results, despite Fianna Fail (11%) winning more first preference votes than Sinn Fein (10%) and the most popular independent (9%), both Fianna Fail candidates were eliminated before they had the chance to win any seats, and the final seat came down to a contest between the Independent candidate and the Sinn Fein candidate (with the Independent winning by 112 votes).


How did this happen? Arguably, it was because Fianna Fail stood too many candidates. A single Fianna Fail candidate would have been in fourth place on first preferences (just behind the three winning Fine Gael candidates), and would have been far better placed to win second and third preferences from independent candidates.

Transfers

Ah, but you might say - would this happen to the left? Wouldn't any Left voters just transfer their votes to other leftwing candidates? Well... maybe. Maybe not. And that "maybe" can make all the difference.

In the 2011 Irish general election, the results in the 3-seat constituency of Cork South-West were:


Fine Gael 49%, Fianna Fail 24%, Labour 14%, Sinn Fein 7%, Others 6%

Proportionally speaking, you'd expect Fine Gael to win 2 of the 3 seats (which they did) and Fianna Fail to win the remaining seat. But instead, Labour won the third seat, and despite winning 24% of the popular vote in that constituency, Fianna Fail won no seats at all.

Why did this happen? It happened because Fianna Fail stood two candidates who won 13% and 11% of the vote respectively, as opposed to Labour who stood only one (winning 14%). This meant that, although most Fianna Fail voters transferred their vote to the other Fianna Fail candidate, not all of them did - so despite polling a joint 10,787 votes on first preferences, the most popular Fianna Fail candidate won just 10,155 votes in the final round, losing to Labour by a very small margin.


Had Fianna Fail only stood one candidate, that candidate would have won 24% of the vote in their own right, topping the poll and almost certainly reaching the quota of 25% before anybody else. And, most importantly, that candidate - even if they had attracted no 2nd preferences - would have had more votes in the final round (10,787) than the Labour candidate (10,754).

So in the event, despite Ireland being very used to STV, enough voters failed to transfer their vote to another candidate from the same party to cost that party a seat.

Are we really confident that Labour members, who have never used STV to elect the NEC before, will be even more consistent at transferring their votes between dozens and dozens of leftwing candidates than an Irish electorate who have been using STV for 99 years?

I'm not.

Conclusion

The Labour Right have immediately recognised the differences in using STV for these elections. Having stood a full slate of candidates in every election under Corbyn (with the exception of some by-elections in 2017), they are only standing 6 candidates for the 9 places in 2020. They clearly understand that in a somewhat-proportional election, they cannot expect to win 9 out of 9 seats with the support of just 30-35% of voters.

The Left has to understand this too. Charlie Mansell estimates that the Left can rely on the support of around 40% of voters in these NEC elections, which would proportionally entitle us to 3-4 seats out of 9.

If that 40% vote share is split between 20-25 leftwing candidates, then the overwhelming majority of leftwing candidates will be eliminated very early after winning less than 1% of first preferences. We would then have to rely on this sizeable and diverse group of voters to rally their preferences behind 3-4 candidates without being told who they are in advance or being asked to do so. That is a difficult task for an electorate that has been using STV for decades, as in Ireland - for an electorate that is using it for the first time, it's nearly impossible.

So what should we do?

My preference is simple: the Labour Left should only stand a united, 5-candidate slate for the 9 places. In the most friendly circumstances, we could only hope to win 5 of the 9 seats (55%) - anything more is simply out of our reach.

At the same time, the Left should make it clear when campaigning for this slate that voters need to preference ALL of the Left's candidates. It doesn't particularly matter what order they preference them in - but they do need to rank ALL of them from 1 to 5.

There are three reasons for why we should do this:
1) Having just 5 candidates would make it clear who leftwing members should vote for, especially given that members will have an enormous number of candidates to choose from

2) Having just 5 candidates will minimise the chance of Left candidates being eliminated in the first few rounds of counting, and give them more of a chance to benefit from second and third preferences

3) Standing multiple slates with lots of different leftwing candidates will discourage voters from giving their 2nd/3rd/4th preference to other leftwing candidates, which they absolutely need to do if we are to win seats on the NEC

This may not be what happens. But I hope that this article contributes to the debate and informs folks about the nature of the Single Transferable Vote.

16 comments:

  1. Hi, I'm usually a big fan of your blog, but I'm really struggling with the logic underpinning this analysis. On the Irish seats you're very much assuming that the voters are choosing their first preference PURELY based on party loyalty to reach the unwarranted conclusion that if one of the candidates didn't stand all the votes would go to the other one.

    That might seem like a minor gripe, but unfortunately I think the main conclusions are drawn from that false assumption. Especially when instead of the difference of between 1 and 2 candidates we're talking about the difference between, say, 5 and 6.

    Let's say there's a slate of 5 candidates, and you have the option of a 6th, what impact would that have? Just for argument's sake, let's say they bring along 2000 extra voters to bother voting, of which as few as 1000 transfer to the slate they are pushing, and the majority of the rest just cast a single vote. That's clearly a big win for the slate!

    ReplyDelete
  2. Very useful, but I'd have liked to know more about who counts the votes, and who will be present at the count.

    ReplyDelete
  3. It's a bloody nightmare! I see 6 names have been put forward by a left consortium but apparently others are proposing an additional 3!

    ReplyDelete
    Replies
    1. This needs sorting pronto. If necessary toss a coin to decide who the 5 should be. We cannot have vote splitting - nightmare

      Delete
    2. That was just at the nomination stage to stop three right wingers being nominated.
      The election itself will be just the six.

      Delete
  4. Left has nominated 6 candidates
    Would a further 3 for 7,8 and 9 preferences have any chance

    ReplyDelete
    Replies
    1. No but they will potentially help stop the rightwingers

      Delete
  5. is it a rule that if anyone gets enough votes to be elected, then their surplus must be distributed before any low scorers are eliminated? The system appears open to rigging by removing low scorers early.

    ReplyDelete
  6. Ireland is one STV country. Malta is another. In Malta the effects you show do not apply because there are only two main parties and voters are very disciplined in choosing only candidates from their own party. In one 5 seat constituency, the Maltese Labour Party put up 19 candidates. They got their predicted number of seats.

    ReplyDelete
  7. Furthermore, the aberrations you mention are down to rather small constituencies which are only somewhat proportional. These effect tend to be less relevant in larger constituencies.

    ReplyDelete
  8. So in short, two pieces of advice:
    1. Be disciplined in voting. Vote for all candidates. Do not ever vite for the soft left or the right.
    2. Vote for the candidates in the order of your preference. That is what the system is for. Don't put a candidate that you really want elected 20th. Obvious, really.

    ReplyDelete
  9. The conclusions from the Cork SW example are wrong, in my view. This is an example where SF failed to get a seat, not because they stood too many candidates, but because voters (for their 2nd ranked candidate) failed to transfer to their top ranked candidate. This could be for many reasons but my guess would be that this candidate was a divisive or unpopular figure among local SF voters, hence many SF voters were unwilling to give them even a 2nd preference vote.

    Failure to transfer is always the real reason why things go wrong in an STV election. There is no such thing as 'vote splitting'. Votes either get transferred, or they don't. Ironically, my conclusion is that SF might have won a seat if they had stood MORE candidates in that election, thereby giving their voters a better choice and maximising the chance of transfers. The real problem was that they chose a divisive candidate (almost certainly).

    ReplyDelete
  10. Many people have been circulating a blogpost from ‘Stats For Lefties’ which claims to show that standing too many candidates cost Fianna Fail seats in Irish general elections. They give two examples but they don’t prove what they think it proves. The first example simply shows that more voters preferred the Labour candidate to the Fianna Fail candidates and so they were elected instead. The second example does demonstrate the problem of exhausted ballots but they make the incorrect assumption that minimising the number of candidates standing is a way to reduce the number of exhausted ballots. This is common practice in Irish elections but there is a powerful argument that only running the number of candidates likely to be elected is in fact a strategy that benefits incumbents against challenges from within their own party, not a strategy to benefit the party as a whole. Cohan et al. (1975) show that there is no causal relationship between number of candidates and exhausted ballots.

    Notwithstanding the above objections to their argument about Irish politics, it is not valid to duplicate a strategy from Irish politics to the Labour NEC elections. Whilst the voting system is almost exactly the same, the political context (parties, constituencies, local interests, national government, coalition-building, voter engagement etc.) and therefore voter behaviour will be very different.

    In the 2018 NEC elections the average voter cast 7.8 out of the 9 votes they were allowed. This is now a different voting system but it is likely that Labour members will express far more preferences than the average Irish voter does. It is a self-fulfilling strategy that Irish parties only stand the number of candidates that they expect to win, many voters then only rank this small number of candidates because they only want to vote for one party, which is then used to justify the strategy on the basis that a lot of voters only rank a few candidates.

    If anything is likely to cause more people to end up with exhausted ballots in the NEC election it is spreading misinformation that voters will help the left by only ranking six!

    Bear in mind that the author of the Stats for Lefties blogpost has not put their name on it, and it doesn’t include any references to articles that support their argument. We have looked and have not been able to find any articles supportive of their argument, though we have referenced below a couple which rebut it.

    ReplyDelete
  11. Katz, R.S., 1981. But how many candidates should we have in Donegal? Numbers of nominees and electoral efficiency in Ireland. British Journal of Political Science, 11(1), pp.117-122.

    Gallagher, M., 1980. Candidate selection in Ireland: The impact of localism and the electoral system. British Journal of Political Science, 10(4), pp.489-503.

    ReplyDelete
  12. Thanks for the explanation and analysis. Ballots will start to 'drop' tomorrow. I fear the left haven't agreed a coherent strategy yet, so some questions remain:

    Is it worth updating this article based on what has happened so far? We know the number of CLP nominations but not how the preferences pan out.

    PIDCOCK GV 333
    BLACK OL 301
    Darr GV 292
    BOLTON GV 277
    RAHMAN GV 274
    JAMA GV 250
    HENDERSON GV 241
    BAXTER L2W 223
    JOSAN L2W 204
    AKEHURST L2W 141
    PAYNE L2W 137
    JACKMAN OL 131
    GRIFFIN Tribune 120
    PAUL L2W 119
    SHERRIFF Tribune 113
    TATLER L2W 108
    FLINTOFF Ind 76
    SILVERMAN LLA 65 + possible 150 From other LLA.

    This is how I see this:

    There are two left slates: 6 strong Grassroots Voice (GV) candidates, and 6 weak Labour left alliance (LLA) candidates. 5 of the latter are so weak I suspect they will drop out quickly and pass their votes to make a moderately strong LLA candidate, who is Silverman. Assuming Left wing members mainly rank all candidates by slate, GV and LLA will end up with 6 strong and a 7th slightly less strong candidate who may struggle to stay in the rounds, but will at least pass on his votes. These will be opposed by Centrist Black from Open Labour (OL), 6 Right wing L2W candidates of which I think 2 or 3 are powerful enough to figure in the final 9, and possibly 2 from Tribune. L2W includes right wing Akehurst who is the 3rd most powerful L2W candidates by nominations, but he might attract a lot of 1st choices due to his strong views.

    There are 4 main issues

    1) The 6 GV members are attempting to moderate their votes by asking members to rank them by postcode. I assume they believe they are just strong enough to get 6 of the 9 places between them IF the GV votes are shared. This strategy is debatable IMV because it might make them just weak enough to fail as well! However at present I'm happy to go with this suggestion.

    2) A more interesting question is what happens if Silverman is thrown into the left wing mix to form 7 candidates. Does putting him (or his slate which amounts to the same thing) before GV, reduce the likely number of successful LW candidates? I'll take some convincing it would, providing the two slates are listed sequentially. I might change my view on this especially if someone has a model they could run to prove it.

    3) I think we must ensure our list contains ALL viable candidates listed above (bar our least preferable) because some L2W candidates will inevitably qualify for the final 9, and I think STV will provide left wing members with some influence on WHICH L2W members are selected due to the bottom order ranking.

    4) Finally can we confirm exactly what type of STV is being used? There are at least 3 different variations on how to distribute votes from dropped candidates, which might affect the answers to the above.

    One RW member with far more information than I have, but has some knowledge of the preferences chosen, predicts

    1 Pidcock
    2 Black
    3 Baxter
    4 Josan
    5 Dar
    6 Rahman
    7 Bolton
    8 Akehurst
    9 Payne

    ReplyDelete