Andrew Goddard writes: Given the events in Canterbury diocese set out in my previous articles (here and here) I was expecting the result of the election of the Canterbury Vacancy in See Committee (ViSC) might raise further concerns—but I did not foresee just how many and how serious they would be.
Thankfully few people have the sad level of interest I have in voting procedures. But that is not necessary to think something must be odd when a voting system (one specifically designed to try and secure a more democratic, proportional result in which the share of seats reflects the expressed preferences of the voters) ends up with such outcomes as:
- Two candidates managed to get elected despite nobody having them as their first preference.
- Four were elected with only one person having them as their first preference.
- One candidate who got 10 first preference votes (behind one with 16 and two with 11 and ahead of all the other 19 candidates) was not elected.
- The majority (8) of the 15 candidates elected had gained two or less of the 88 first preferences and another 3 had received only 3 or 4 first preferences.
- In fact 11 of the 15 finally elected had between them only 21 of the 88 first preference votes.
- The only ordained woman whose first preference votes were counted (the other on the ballot was excluded as explained below) was standing to represent the clergy of Ashford Archdeaconry and got 10 votes but a clergyman from that Archdeaconry with 8 votes got elected instead of her.
How did this happen? The journey from the first preference votes to who gets elected is a complex one under the Single Transferable Vote (STV) and as yet we do not have the details though they are required to be published. It would, however, appear that the anomalies arise from two key factors:
- A failure to follow the STV Rules that should have been applied, with potential consequences in terms of who was elected; and
- The existence of multiple constraints on who could be elected and in particular the new constraint (and how it was applied) in relation to not electing candidates with a “relevant connection” with each other (i.e. part of the same worshipping community).
As explained more fully in a longer article, the process appears to have been marked by the following five features.
Firstly, the exclusion of two candidates on the ballot paper (a lay man and an ordained woman) because of the new “relevant connection” rule meaning they could not be elected as they were part of the same worshipping community as one of the unelected ex officio members of the ViSC. Any first preference votes for them (we do not know how many there were) would have been moved in full to their supporters’ second preference (assuming they had given one). This appears to be the proper application of the STV Rules given the new “relevant connection” rule although this rule was not in place when the election was called and nominations opened.
Secondly, and contrary to the STV Rules, 5 of the 23 candidates should not have been on the ballot paper because they should have been declared elected unopposed given how the count was then processed. This is because there were only 4 candidates for the 4 lay seats in Ashford Archdeaconry and only 1 candidate for the 2 clergy seats in Maidstone.
Thirdly, the 4 lay candidates for Ashford who should not have been on the ballot were then recognised at the start of the count as needing to be elected if all 4 seats were to be filled (what is technically known as “guarded candidates” to ensure a “conformant result”). This is how one of the two elected with no first preferences was able to be elected. As one of the 4 had a relevant connection with one of the 4 clergy competing for the 2 Ashford clergy seats that clergywoman was then judged unable to be elected (technically “doomed”). This was despite the fact that she had 10 votes compared to only 1 for the layman connected to her. She therefore had already received enough votes to be deemed elected (making the lay person the “doomed” candidate of the two connected candidates who could not both be elected). Presumably her 10 votes were all then transferred in full to any second preferences of those who wanted her elected.
This demonstrates that the new “relevant connection” constraint is being implemented without reference to the relative support the connected candidates have among the voters. Even if nobody at all wants one them elected they can be elected at the expense of another candidate even if that person is the most popular candidate on the ballot.
The outcome also then depends on the number of candidates in different categories. If, for example, there had been 5 lay candidates in Ashford and not only 4, then that simple change (even if the 5th candidate also got no first preference votes) would have meant the election of the clergywoman with 10 votes, rendering the lay person with 1 vote to whom she was connected the doomed candidate and ensuring the election of the other 4 lay candidates.
It is not clear to me what would happen were the connection to be between candidates in the same House ie both lay or both clergy in the same Archdeaconry (or different Houses in different Archdeaconries) and both connected candidates need to be elected to fill all the seats. Nor is it clear what would have happened if there had been only 2 clergy standing in Ashford instead of 4 and so both the connected candidates being elected unopposed to fill all seats. Here the “relevant connection” rule and “the election of the required number of candidates fulfilling all the constraints in force” would not both be able to followed and one would have to take precedence.
Fourthly, it is not clear what happened to the 4 first preference votes spread among the 4 lay Ashford candidates. There appear to be two options. In terms of the STV rules they should have continued as a “guarded” candidate. This means those who had voted for them wasted their votes as they should not have been on the ballot in the first place. Alternatively, given the rule had been broken and they were on the ballot and been voted for when they should have been elected unopposed, the votes could have been transferred in full to any second preferences. This would, however, have been another breach of the STV rules.
Fifthly, and likely more significantly for the outcome, it is not clear what happened to the 11 first preference votes for the Maidstone clergy candidate. According to the STV rules given they were on the ballot they should then have been deemed elected immediately on the basis of passing the quota. Any surplus votes beyond that quota should be transferred to any second preferences but at a reduced value. However, if the STV rule had been followed and they were not on the ballot, those second preferences would have become the voters’ first preferences and had a full not a discounted value for a different candidate. In other words, those voting for this candidate seem to have had less than a full vote due to the breach in the STV rule that led to them being put on the ballot.
In addition to these concerns about how the ViSC vote was run, there are other oddities and questions going forward:
- One of the 3 CNC seats is for an ordained woman but no ordained women have been elected to the ViSC as all 5 elected clergy are male (ironically because of the new “relevant connection” rule). There are only two unelected women clergy on the ViSC able to fill this seat so there may be no election required.
- The 10 male non-episcopal clergy on the ViSC are not eligible to be elected to CNC if one of the two ex officio women clergy stand.
- It is unclear how the elections within the ViSC will be run given how the ViSC was run, the STV rules and the new regulations. For example, if a clergyman and a clergywoman stand for election will there still be a contested election with both on the ballot and if so how will it be run?
Conclusion
There are clearly major questions about the process of election to the Canterbury ViSC. These make it important for the result sheet to be published as soon as possible so there can be transparency and greater clarity which may yield some answers. There are also questions as to how other ViSC elections will be affected in future given the new Regulation about a “relevant connection”.
The election of candidates to a ViSC should be such as to secure a ViSC which is properly representative of the diocese and the preferences of those electing its members. It also needs to be seen to be representative. There are questions as to whether these outcomes have been achieved in relation to this ViSC election.
Arguably even more important is that the ViSC election of members to serve on the CNC should also be–-and be seen to be—producing a result which is representative of the diversity of the diocese. For that to happen there now needs to be not only proper compliance with the voting rules but much greater transparency as to how those rules will be implemented. In particular there must be clarity, in advance, as to how the new constraints in relation to electing if possible both a clergywoman and a lay woman to the 2 lay and 1 clergy places on CNC will be implemented.
Revd Dr Andrew Goddard is Assistant Minister, St James the Less, Pimlico, (where his wife *Lis Goddard is vicar) Tutor in Christian Ethics, Westminster Theological Centre (WTC) and Tutor in Ethics at Ridley Hall, Cambridge. He is a member of the Church of England Evangelical Council (CEEC) and was a member of the Co-Ordinating Group of LLF and the 2023 subgroup looking at Pastoral Guidance.
I have never understood the reason for the ‘quota’ in the STV system. As far as I can see, it can lead to a result which depends upon the order in which the votes are counted. A number of voters exceeding the quota put the same candidate as their first preference. With the number of candidates in this election, it is possible that each of these voters had a different candidate as their second preference. So, which of the voters’ second preferences get carried forward to the next round, and which do not?
To me a fairer process would be to eliminate the candidate with the fewest votes in each round, and transferring their votes to highest preference candidate on each ballot paper who remains in the contest. This way, those who vote for the least preferred candidates get more of a chance of getting someone they like elected.
The merit of STV is supposed to be that it eliminates (or at least severely degrades) tactical voting. If people vote with their preferences, then how the all the candidates are preferred will be reflected in the results. So, if there is a candidate who is wildly popular, under STV the supporters of that candidate are able to transfer the “excess” of their support to their second, third etc. choices. If you only eliminate low scoring candidates, and transfer votes until you’re left with the number of candidates for places, then tactical voting is back: those who back a very popular candidate are effectively punished, and you encourage campaigns to try to organise ‘efficient’ voting making sure that just enough of your supporters back you to get you elected, and the rest know to vote for their second preferences. It creates a political machine.
It is quite possible for STV systems in multi-seat elections to get results where those with 0 first preferences are elected because they do well with second preferences etc.. That’s not a bug, it’s a feature. If you object to that then you need to advocate for a very different system.
You can give people a list of features in a democratic voting system that they will agree are all desirable, yet they are inconsistent. That’s Arrow’s theorem.
My experience is with STV in Ireland. A candidate with 0 first preference votes would be eliminated at the end of the first count unless there was a surplus to be distributed on the second count.
In a party system they might receive enough transfers to move them above the next lowest candidate, but that’s unlikely.
An interesting extreme example of a surplus bumping up a poorly performing candidate is Cyprian Brady in Dublin Central in 2007 who was the beneficiary of then Taoiseach Bertie Ahern’s substantial surplus. https://www.electionsireland.org/result.cfm?election=2007&cons=85
That’s because in a Dail election you’d expect to have substantially more candidates that seats, making it unlikely that anyone is over quota (i.e. has a surplus) on the first count. Of course, the Taoiseach standing in his party’s stronghold might be an exception to that general expectation.
A CofE election like this is quite a different beast, even if using the STV system. The voting patterns, campaigning, candidate-seat ratio, existence of parties are
(I’d suggest) just too different.
That’s partly why the Hare quota is used in the C of E instead of the Droop quota. But making quota on the first count still happens quite a lot in Ireland. The big difference is party management is significant in Ireland and embryonic in the C of E!
The quota is designed to spread valid votes across candidates so as many preferences as possible are effective. The order of distribution of surpluses and elimination of candidates is decided entirely mathematically.
As I understand it, the quota is calculated differently in the C of E and Ireland. The former uses the Hare quota [total votes/total seats] and the latter the Droop quota [(total votes / (total seats + 1)) + 1].
In Ireland, the distribution of surpluses arising on a first count is decided by examining the second preference on all the ballots and ensuring a representive bundle is distributed. Where a surplus is accrued during counting, a random selection of transferred papers are distributed.
In Dáil elections, only whole votes are transferred. I understand fractional votes are used in C of E elections. (Similar to Seanad elections where votes are multiplied by 1,000.)
> So, which of the voters’ second preferences get carried forward to the next round, and which do not?
This is a reasonable question, but there have been at least two good answers ever since STV was created.
Firstly, you can use a sample (either random or representative) of the ballot papers. So if the quota is 8, and a candidate is elected with 10 votes, you pick two ballot papers and use them for the rest of the process. This method is ideal if you are mathematically limited (e.g. if it’s 1925 and you’re doing the maths for an entire county on a single blackboard) or you’re really worried about fraud and want observers to be able to check physical ballot papers from a distance.
Secondly, you can use fractions or decimals. So for that candidate elected with 12 votes, you reduce the value of each ballot paper to 2/10 = ⅕ or 0.2. Then each of the ballot papers counts as 0.2 for their second preference. If you’re using a spreadsheet or specialist software competently, then it’s not much more difficult, so it’s the sensible solution today, because it means every voter’s second & subsequent preferences are perfectly incorporated into every step of the process.
So all the voters need to do is put their preferences honestly, and the STV rules will ensure that their wishes are reflected. The voters don’t need to do any complex maths or try to guess the outcome; the returning officer (or the spreadsheet programmer) does all the hard work for them.
Your proposal is closer to a different system, called the Alternative Vote. It’s less fair than STV because the second preferences of voters who vote for unpopular candidates become more important than those who vote for popular candidates, and then you are back to voters having to vote tactically based on their own maths & guesses about likely winners.
BTW this answer is essentially the same as AJ Bell’s helpful one, but more focused on your specific question.
“The voters don’t need to do any complex maths”
It’s one of the myths around STV that a lower preference can affect the outcome for a higher preference. This is not true. You cannot harm the outcome of your first preference by any of your other preferences, so you should always vote in order of your “honest” preferences.
The only qualification to this is when there is vote management by parties to maximise their seats when they are running at least two candidates. There is a risk in implementing/going along with a vote management strategy that neither or the “wrong” candidate will be elected. For example, in Dublin Mid West in 2002, the Fine Gael candidates finished much further down the field because of the way the vote split. If one of the candidates had been ahead of Labour on the first count, they may have benefitted from Labour transfers instead of the other way round. In truth, they were wildly optimistic to run two candidates in the first place!
https://electionsireland.org/result.cfm?election=2002&cons=95
This mess is the result of a number of things:
1. Justin’s initiative in skewing process because he didn’t like the result the due process was producing.
2. Synod’s idiocy in not thinking this through, compounded by deference to the views of bishops (who I think supported it…?)
3. A failure to think about the reality of the C of E. Rural parishes are increasingly being linked together; an ever higher proportion of Anglicans are attending larger churches. That makes the ‘connection’ rule just stupid.
4. Decision making is dispersed, so no-one can stand up and say ‘This is not working and must be re-thought.’
The central error seems to be imposing restrictions only after the voting has taken place, not before in screening the candidates. That massively complicates the vote-counting and necessarily makes the whole thing opaque.
Hasn’t this led to a fundamentally flawed and thus invalid election?
How can rule changes or new interpretations after the vote be “legal”? They destroyed the process.
Does this election have any basis in law, apart from the morality that should underpin it?
Am I being too simplistic? It further destroys confidence in “the system”
“Hasn’t this led to a fundamentally flawed and thus invalid election?”
Yes!
There needs to be at least one “open” seat to ensure an election with these restraints is fair. And as Andrew noted, any candidates who were going to be elected anyway shouldn’t have been on the ballot.
“transparency and greater clarity” ? You are ‘aving a larff’ !
Firstly, I must point out that I am one of the clergy people not elected in this election. Secondly, my first vote went to the female clergy person who despite receiving a large number of first votes was disqualified because of a lay person being automatically elected from her very large parish.
Secondly, I don’t understand why all Diocesan Synod members vote for all the candidates. If the idea is to elect representatives of the clergy and laity for each of the three archdeaconries, it would make sense to have six separate elections with people only voting in the election for candidates to their archdeaconry and their status (lay or ordained). That would have made for much more straightforward, representative and transparent elections.
Thirdly, another issue was that the candidates were listed in alphabetical order, which when people are asked to order over twenty candidates I think in practice disadvantages those lower down the alphabet.
Your second point is very valid, and as someone fairly well acquainted with elections in Ireland and only very recently introduced to ViSC elections, I’d come to the conclusion quite quickly that would have been a fairer and less complicated way to run these elections.
As I understand it ‘single transferable vote’ has two basic aims – that where there are more than two candidates, people get to express their real choice without feeling their vote will be wasted, and that the person/people elected are those to whom a majority can give significant approval.
Thus if one candidate gets a clear majority of first preference votes, they are elected. If nobody has a clear majority at this point, the second preferences of the losers are taken into account. I’m not sure but I believe there is more than one way of doing this. The point is that you don’t elect, as can happen in UK parliamentary elections, a candidate who got more than others but less than 50%, but is in effect hated by the majority who didn’t vote for him – a minority winner. STV results in a person elected who is at least tolerable to a true – over 50% – majority.
If the CofE can’t arrange that – well I’ll be polite and say they’re incompetent……
Mr Langton, I have written a reply to you that I accidentally posted as a top-level comment. Apologies.
No apology needed – it took me quite a while to work out some functions of this site ….
You’re completely right about the purposes of the Single Transferable Vote (STV). But the analysis in your middle paragraph is muddling two different systems. I think you’ve tried to sensibly simplify your explanation to a single person and been caught by a mathematical consequence that I’ll explain below.
The system you are describing is the one usually called the Alternative Vote (AV) in English-speaking Europe. You are correct that it requires winners to have an absolute majority of votes. For that reason, it’s usually only used for elections with a single winner. It was the system proposed for the House of Commons in the 2011 referendum.
The Single Transferable Vote (STV) is a different system and the one which is most common in the Church of England. You don’t need to have an absolute majority to win; you need to pass a quota. The usual quota is “( votes / ( seats+1 ) ) +1”, though according to other posts in this thread, the C of E uses the older “votes / seats” calculation. Either way, it’s not guaranteed that any one winner has an absolute majority. But it is guaranteed that all the winners, considered together, have an absolute majority. As you say, it lets people express their real choice without their votes being wasted, and it ensures widespread representation. It’s to the C of E’s credit that it uses it.
Now, if you follow the STV rules in an election with only one winner, then the effect is exactly the same as if you’d followed the AV rules, because the STV quota will equal 50% of the votes, essentially the same as AV’s requirement for an absolute majority. Your example only discussed one winner, which is why it’s entirely understandable that it muddles the two systems. In order to gain the full benefits of STV, it needs to be paired with multi-winner elections. In the case of the Canterbury V-i-C election, that should have been easy, because it was electing a committee with numerous members.
The problem here isn’t the voting system, but the attempts to pre-determine the outcome of the election by imposing conflicting additional requirements, and the sheer incompetence of those running the election. In particular, changing the rules partway through the election was bound to cause problems (you should only be doing that if you’re facing one of the four horsemen of the Apocalypse).
Thanks for the clarification. I think I had (roughly) got the impression that this was a case of a committee or panel being chosen not ‘en bloc’ but as individuals each representing as it were a diffferent constituency, and then meeting together to take the decision about the appointment. What was actually going on does seem more messy!
Hopefully between us we may have clarified it a bit for others ….
Presidential elections in Ireland end up being more akin to Alternative Vote as there is a single member “constituency”. Illustrated by Mary Robinson’s victory from second place in 1990 (the first time there was more than one count):
https://electionsireland.org/result.cfm?election=1990P&cons=194
Presidential elections can be more competitive than that. In 2011 it ran to four counts and included a former Eurovision winner:
https://electionsireland.org/result.cfm?election=2011P&cons=194
A better system would be to have a plural amount of Bishops in the role, as well having plural local leaders.
You could call the local leaders a “Session”, and the regional leaders a “Presbytery”.
… I’ll see myself out.
Oh and maybe selecting randomly from a group of qualified people? That’s how we got Matthias.
Would be more transparent that what actually happened 🙂
The closest example to the Canterbury ViSC election is one of the large panels for the Seanad. This includes sub-panels, similar to the constraints of archdeaconry, clergy and laity. But as only two members are required from each of two sub-panels where main panels are between five and 11 members, this rarely impinges on the final result:
https://electionsireland.org/result.cfm?election=2025S&cons=2
(Note that first preference votes are multiplied by 1,000)
I’ve written about STV, and I think (but may be wrong) that there may be some misconceptions in some of the comments above and even possibly in the main article. Others have commented in great detail of course.
It is shocking if candidates were listed in alphabetical order. Since I presume voting was in an online system the right approach would be to present each elector with a different random order. Easily done.
However, that’s not really the point at hand here. The real issue is that if you have an electoral system so complicated that essentially nobody can really get to grips with what’s going on, then you have trouble, because there will be a general loss of confidence in the result. A simple STV election can be explained (by the way, there should be no reason to sample votes for redistribution because the computer can handle fractional votes) but if you overlay with all sorts of constraints you can see what happens.
How you deal with constraints of the sort that there are is an interesting question–maybe split into more elections or more stages of voting. (The election of the Oxford University Chancellor by AV worked by having two stages, one to narrow down a very large field to five candidates, and then another, a couple of weeks later for the final choice. That is much easier for a human to get their head round when voting.) But in this case, all these constraints have become too heavy for STV to bear in practice. At some point in the long process of introducing more and more constraints, someone should have said: Enough!
Thanks for this Bernard – I think your final comment – “But in this case, all these constraints have become too heavy for STV to bear in practice. At some point in the long process of introducing more and more constraints, someone should have said: Enough!” – captures well what lies at the heart of the problem. I hope now there will be a review of the impact of adding more and more constraints within STV.
Elsewhere online someone has wondered about how the software was adapted to apply the new “relevant connection” constraint in relation to the other constraints already in place which raises the question whether/how much this was thought through and if so why it was decided to prioritise filling all seats within each of the effectively 6 elections running rather than ensuring the most popular candidate of any with relevant connections was elected.
You said that there may be some misconceptions about STV in the article and I’d love to know where I’ve got things wrong if you can point to the areas that you were concerned about. Thanks.
I have recently been drawn to the system for electing the Venetian system for electing the Doge !
45 electors were drawn by lot, and then a second lottery reduced them to nine, who nominated forty candidates in all, each of whom had to be approved by at least seven electors in order to pass to the next stage. The forty were pruned by lot to twelve, who nominated a total of twenty-five, who needed at least nine nominations each. The twenty-five were culled to nine, who picked an electoral college of forty-five, each with at least seven nominations. The forty-five became eleven, who chose a final college of forty-one. Each member proposed one candidate, all of whom were discussed and, if necessary, examined in person, whereupon each elector cast a vote for every candidate of whom he approved. The candidate with the most approvals was the winner, provided he had been endorsed by at least twenty-five of the forty-one.”
What happens if Canterbury CNC is unable to appoint which seems to be the case with other Bishoprics such as Ely