Electing the House of Lords: Should there be above-the-line voting?

Alan Renwick

The parliamentary select committee that has been examining the government’s proposals for reform of the House of Lords will be publishing its report in a couple of weeks’ time.  Rumour has it that they want an electoral system different from the one proposed by the government.  Nick Clegg and colleagues argue that the Single Transferable Vote (STV) form of proportional representation should be used.  But the committee has been interested in finding a system that will give voters a choice between voting for individual candidates and for a single party ticket (see the transcript of their oral evidence session last December, when the quizzed Iain McLean and me on this subject).  According to the Guardian, the committee is going to recommend the form of STV used in many Australian elections, where voters can vote ‘above the line’ for a party or ‘below the line’ for individual candidates.

The Electoral Reform Society is crying foul over this.  Calling the proposals a ‘dog’s breakfast’, they say that STV with above-the-line voting will return power to the parties, rather than allowing voters to determine who gets elected.

What should dispassionate observers make of this?  I think three questions need to be considered.  First, how much power would the inclusion of a party voting option give to parties and to voters?  Second, how much power should parties and voters have in determining which candidates are elected?  Third, are there any other considerations that we should take into account before deciding whether we think that possibility of above-the-line voting should be welcomed?  Most of this rather lengthy (sorry) post will focus on the first of these questions; I’ll say a little about the other two at the end.

Thinking about the logic of STV

The key issue when thinking about the effects that above-the-line voting will actually have concerns the order in which a party’s candidates are elected.  How likely it is that voters’ preferences could upset the party’s preferred ordering, such that one candidate gets elected ahead of another candidate who was ranked higher by the party?  Those of you with a penchant for playing around with STV counting scenarios will be able to work out that, if no votes are exhausted, the minimum number of personal votes that a candidate needs to gather in order to rise up the party’s ordering is just over half of one quota.  Say, for example, we are in a district in which five seats are to be filled.  The quota in his district – the number of votes that a candidate must surpass in order to be guaranteed election – is one sixth of the vote, or 16.67 per cent.  So a candidate needs at least just over half of this – just over 8.33 per cent – to have any chance of being elected ahead of someone else higher on the party’s ticket.

Let me give you an example to see why this is.  Let’s stick with a district from which five members will be elected and say that a total of 1,000 valid votes are cast.  Furthermore, let’s focus in on one party, which receives 334 of these votes.  Of these votes, 250 are cast above the line for the party ticket.  The party has three candidates.  The first two of these win no personal votes, while the third wins 84.

The first step is to calculate the quota.  In a five-member district, the quota is calculated by dividing the total number of votes by six, giving 166.7.  A candidate must exceed this – gain 167 votes – in order to be guaranteed election.

Now we begin to count the votes.  The 250 votes cast for the party ticket are initially allocated to the first candidate on the ticket (Candidate A).  Candidate A had no personal votes, so her vote total is 250 (250 + 0).  She has passed the quota and is therefore immediately elected.  Using standard STV counting procedures, her surplus of votes (the number of votes she had above the quota, i.e., 83) is then transferred according to the second preferences of the voters who cast these votes.  Given that all of these votes were ticket votes, the second preference is in all cases the second preference of the party – that is, the second candidate on the list, Candidate B.  So 83 votes transfer to Candidate B.  Candidate B, like Candidate A, had no personal votes, so his total is now 83, just short of Candidate C, who still has 84 votes.  There are, of course, other parties with other candidates, and various elections and eliminations will happen among these.  Assume, however, that there are no vote transfers to the party we are focusing on, so the vote totals for Candidates B and C remain unaffected.  At some point, Candidate B will be the remaining candidate with the lowest vote share and will therefore be eliminated.  His votes will transfer to Candidate C (following the party ordering).  Candidate C therefore gains 83 votes in addition to her 84 personal votes, giving a total of 167 votes.  This just meets the quota, so Candidate C is elected.


Count stages in a hypothetical STV election with above-the-line voting


Votes   as cast

First   count

Second   count

Third   count

Above the line




Candidate A





Candidate B





Candidate C






A is elected

C is elected


Candidate C has therefore been elected ahead of Candidate B, even though B was higher on the party’s preferred ranking.  This was possible because C won a number of personal votes equal to just over half the quota.  But you can see from the example why this is the minimum share of the vote making such a result possible (still assuming no exhausted ballot papers).  Had C won only 83 personal votes, she would not have secured the quota and would not have been elected.  Had there been any more ticket votes, the surplus transferred to B would have been larger and C would have been eliminated before B (in the event of a tie between the two of them, presumably some random mechanism would have determined which was eliminated).  Had either A or B secured any personal votes, the same thing would have happened: in such situations, C would have needed more than the minimum number of personal votes to secure election.  Thus, in any realistic scenario, in which A and B would have won some personal votes, half a quota of votes would not have been enough to win C the seat.  Half a quota is a minimum, but normally more will be required.

Two qualifications to this are required.  First, Candidate C could have picked up some of her personal votes as second or lower preferences: it’s not the case that she needs to capture all of these votes as first preferences.  So long as C receives the vote before anyone ranked above her on the party list, and no candidate below her on the list secures more votes than she does, she can benefit from that vote.  Nevertheless, a candidate needs to receive at least just over half a quota of preference votes of some kind early enough in the counting process not to be eliminated.

Second, as I have noted a couple of times, I’ve been assuming that there are no exhausted ballot papers: that all voters express a full ordering of candidates (either implicitly through their above-the-line vote or explicitly by ranking all the candidates below the line).  But this is not realistic.  All of the preferential voting systems that have been used or proposed in the UK (the supplementary vote system used in mayoral elections in England; the STV systems used in local elections in Scotland and in all non-Westminster elections in Northern Ireland; the STV system proposed by the government for the reformed second chamber; and the AV system put forward in last year’s referendum) have allowed voters to rank as many or as few candidates as they wish.  This means that some ballot papers are exhausted – all the candidates for which preferences have been expressed have been eliminated – before the end of the count.  In this scenario, some candidates are elected even though they do not meet the quota (just as, under AV, as discussed extensively last year, some candidates are in fact elected with less than 50 per cent of the vote).  This reduces the minimum number of personal votes that a candidate needs in order to leapfrog those above them on the party list.

Adding all of these pieces together, how many votes does a candidate need to be elected out of list order?  Alas, there is no straightforward answer: it depends on how the votes are spread.  A few candidates might sometimes secure election out of list order having gathered less than half a quota in personal votes (whether as first or lower preferences).  But in most cases, candidates will rise up the list only by collecting more than half a quota of such votes – and often they will need much more than that.

Evidence from Australia and around the world

If abstract logic won’t give as all the answers, we can look for real-world evidence.  STV systems with above-the-line voting are currently used only in Australia.  The first such system was introduced for the Australian Senate elections 1984 and has been retained ever since.  Similar systems have now been adopted for upper house elections in South Australia, New South Wales, Western Australia, and Victoria.  In most of these cases, voters can either vote for one party above the line or rank the candidates below the line.  New South Wales now has a variant under which voters can either rank parties above the line or rank candidates below the line.

As the Electoral Reform Society points out, the proportion of voters in fact voting above the line in Australia is very high: as the following table shows, it was higher than 95 per cent in the most recent election in each of these jurisdictions.  With such high ticket voting rates, it is virtually impossible for a candidate to be elected out of rank order.  In fact, not one candidate has been elected out of rank order since above-the-line voting was introduced (I’m grateful to Antony Green, Australia’s leading elections expert, for confirming this by email).  This means that Australian has, in effect, something very close to a closed-list proportional system.


Rates of above-the-line voting in recent Australian elections

Australian Commonwealth Senate (2010)

New South Wales Legislative Council   (2011)

South Australia Legislative Council   (2010)

Western Australia Legislative Council   (2008)

Victoria Upper House (2010)






Sources: Australian Electoral Commission and relevant state electoral commissions.


At the same time, the Electoral Reform Society is wrong to imply that we should expect the same pattern to apply here.  In most of these Australian cases, if a voter chooses to vote below the line, she or he must rank every candidate – of whom there may be several dozen – in order to cast a valid vote.  In the extreme case of New South Wales in 1999, there were 264 candidates.  This makes voting for individual candidates much more onerous than voting for a party ticket, so it is hardly surprising that the vast majority of voters choose the latter option.

In the UK, by contrast, a vote would be valid even if only one candidate preference were expressed, so the effort required to vote below the line would be no higher than that required to vote above the line.  We can therefore expect more personal votes to be cast than in Australia.

But just how many more?  The Australian evidence here isn’t very useful.  The most permissive Australian case is Victoria, where a valid below-the-line vote in state Upper House elections requires only five preferences.  The table above suggests that this makes no difference to the number of personal votes cast.  But we shouldn’t infer too much from this.  First, Victoria introduced this system only in 2006.  Voters had become accustomed by then to voting above the line in Senate elections, and it is hardly surprising that they extended their habit to the new venue.  Second, providing five preferences is still a lot more onerous than casting one party vote.

In terms of voter experience, the closest analogues to what the select committee is said to be proposing for UK second chamber elections are not the Australian systems, but rather list proportional systems in which voters have the option, if they wish, to express one preference (or perhaps more) among individual candidates.  These systems model much more closely the sort of choice that will be available to voters in the UK.

If we look at such systems, we find that the proportion of voters who actually exercise their right to express a preference varies hugely.  Uwe Kitzinger found that in Austria in the 1950s fewer than 1 per cent did so.  Today, as Lauri Karvonen (in this book) shows, around 20 per cent of Austrian voters do so, while around two thirds do so in Belgium and 90 per cent in Brazil.  These cases paint a very different picture from the Australian evidence.

This wide variation makes it very difficult to predict what would happen in the UK.  Given that British voters don’t much like parties, have very little experience of voting for party tickets, and seem particularly to value independent-mindedness in the second chamber, it seems reasonable to expect that many would exercise a right to vote for individual candidates.  It certainly seems likely that more than half a quota of voters (ranging, in the 5–7-member districts proposed for the reformed second chamber, from 6.25 per cent to 8.33 per cent) will do so.

But, as we have seen, whether these votes make a difference depends on how they are distributed.  In fact, the Australian experience is that most of those voters who choose to vote below the line give their first preference to the candidate who is already at the top of the party’s ranking: 71.5 per cent of such voters did so in New South Wales at the last state election.  That is hopefully a sign that parties and their supporters often agree on who the best candidates are, but it does nevertheless make it even harder for anyone to move up the list.

The list PR systems with flexible lists give us less in the way of useful information here than they did above: they use different mechanisms for translating votes into seats from those used under STV.  But all of the cases that I’m aware of suggest that the party’s ordering generally wins: even in Belgium, for example, where most voters now express candidate preferences, the great majority of the candidates who are elected would have been elected with completely closed lists.

In sum, then, we can say that, while allowing an above-the-line option makes total party domination possible, that is unlikely in the UK given the other rules proposed and the habits of voters.  Nevertheless, above-the-line voting would almost certainly increase significantly the power of parties to determine which of their candidates are elected.

What, then, of the second and third questions that I began with.  What should be the balance of power between parties and voters?  And are there any other considerations that we should bear in mind before coming to a view?

Where should power lie?

It might seem obvious that power ought to lie with voters.  In a parliamentary system, however, parties are very important.  Voters are interested in the composition not just of parliament, but also of the government.  But they can vote not for a government, but only for an individual candidate.  If governments are to be accountable to voters, therefore, voters need to know whether the candidates in their constituency will generally support or generally oppose any particular government.  And that can be achieved if the candidates are loyal to particular parties.  Clearly, a balance needs to be struck here: too much loyalty means parliament fails to do anything useful; too little loyalty means voters lose control and government comes to operate through the delivery of pork.  But an element of party discipline is important.

A straightforward STV system – without above-the-line voting – might get the balance wrong here.  That has certainly been a concern of late in Ireland, where politicians have been criticized for paying too much attention to the baubles they can bring home to their constituencies and too little attention to such national issues as the regulation of the banks.  The debate here certainly isn’t closed, but there are questions to be asked.

Nevertheless, we are talking here about elections to the UK’s proposed reformed second chamber.  This is not the chamber from which the government is formed.  Rather, it is a revising chamber with limited powers in which independent-mindedness is an advantage.  As I have shown elsewhere, party discipline is actually already high there; there is no reason to think it should be higher and much reason to think it could usefully be lower.  As a result, the usual argument against STV doesn’t apply.  And, that being the case, it’s not clear what the value of a more party-based system would be.

Other considerations?

Given the questions asked during its oral evidence sessions, we can expect the select committee to argue that allowing above-the-line voting will simply give voters what they want: if voters want to engage in the complicated business of ranking candidates, they will be able to do so; but they will also have the option, should they wish, of casting one simple vote for a party ticket.

There is certainly merit in the view that some voters would value this opportunity.  But other factors need to be set alongside this.  Allowing above-the-line voting would make the ballot paper more complicated.  Given that the government plans to hold second chamber elections at the same time as elections to the House of Commons, which will continue to use the very different system of First Past the Post, there is a danger of making things more confusing than they need be.  Explaining to voters how STV works will already be challenging enough (though voters don’t need to know all the intricacies to be able to vote intelligently).  Adding an extra element requiring further explanation should only be done with clear justification.

So what do I think?

I argued in the Political Studies Association’s briefing paper on House of Lords reform that STV was clearly the best system for electing a reformed second chamber.  I wasn’t trying in that paper to push a particular perspective, but this particular point seemed quite uncontroversial.  A proportional system is needed to make it hard for any one party to secure a majority – a goal that virtually everyone agrees on.  And if independence is valued – both the relative independence of party representatives from the party whip and the presence of independents who take no party whip – then STV is superior to any list-based form of PR.

This remains my view.  Above-the-line voting would be unlikely to bring the party domination seen in Australia, but it would nevertheless enhance the power of parties in a way that is not clearly justified.  It might give some voters an option that they would value, but it would also increase the dangers of confusion.

I make no claims for STV across the board: for first chamber elections in parliamentary systems, it carries significant dangers.  For the proposed elected second chamber in the UK, however, it fits the criteria that most people want very well.

Alan Renwick’s current research into political reform debates in the UK is generously supported by the Nuffield Foundation, to whom he is grateful.  Any views expressed herein and all errors are, of course, his own.

  1. Matthew Shugart’s avatar

    Excellent essay, Alan!

    I have one extension and one quibble.

    The extension is that your overview convinces me that STV with above-the-line option and no requirement for those giving preferences to give more than one (or a few) is likely the best way to balance party and candidate voting. As you rightly note, most “flexible” list systems actually default to the party order most of the time. (I offer this point generally, not specific to the British second chamber, where your case for STV with only candidate-preference voting is persuasive.)

    The quibble is that I have no idea why you mention Brazil in this context. In Brazil there is no party-given order to the list. Only preference votes affect the order. Votes cast solely for the list affect the inter-party balance of seats only. So it is not surprising that 90% or more of Brazilian voters cast a candidate preference. What’s surprising is that 20% of Austrians do so when a party order usually prevails anyway.


  2. Alan Renwick’s avatar

    Many thanks, Matt! I like your extension, but I’ll need to think about the comparison of flexible list PR and STV with ATL voting a bit further. I’m about to get on a train, which could be a good venue for cogitation….

    On Brazil, I’m very much not a Latin America expert, so I bow to you have I have this wrong. But I was looking for systems of any type in which voters have the choice of voting for a candidate or simply for a party ticket, whether final list order is determined wholly or only partly by the candidate votes. My understanding is that Brazil is such a system. What you say about Brazil illustrates the point that whether voters use the opportunity to cast a candidate vote is likely to depend on how much difference these votes make to the outcome of the election.


  3. David Farrell’s avatar

    Dear both
    I’m very persuaded by Rein T’s mantra of keeping things simple and I think the potential dog’s breakfast that is being proposed for the HoL shows why. The point of above the line voting in Australia is that it delivers full preferences as required under their electoral law. The parties agree ticket deals with all the other parties to deliver this. A second contextual points is, as we know, the view that full expression of preferences is the most effective way to maximise the use of your vote in STV, so as to ensure that you make things as hard as possible for the party you dont support.
    Ok. So if the HoL was to mix above the line voting while not requiring full expression of preferences what we’d be left with two extremes in voting behaviour- a large number of voters expressing a handful of preferences and another number of voters ( many of these, I suspect, supporting the main parties) voting above the line and thereby xpressing full preferences. I’m not sure if that is a fair outcome.
    If the parties really want to have some party structuring while allowing independents, why not have a Maltese ballot paper, grouping candidates by party, with parties drawing lots to determine which order. They could set it up so that registered parties appear first above non-party candidates. And they could set a high threshold for ballot access to keep out Nutters.


  4. Francis McGonigal’s avatar

    An alternative to “Above the Line Voting” would be to allow voters to express equal rankings. A voter who liked two or more candidates equally should be able to give them the same preference.
    How the vote would be allocated between equally-ranked candidates would need to be decided. It could be awarded to whichever candidate is leading at the stage that preference is used, or simply divided between them.
    This gets round the problem of ordering a large number of candidates but would avoid the need for Australian-style party tickets. Equal ranking could cut across party lines and would be under the control of the voter, not the parties.


  5. stephen Johnson’s avatar

    Could an Independent Peers’ Commission act as a surrogate party and put forward a slate of approved independent candidates, so that voters could vote above the line for approved Independent candidates (rather than lone independents)?