Appendix 6: Supplementary written evidence
on electoral system options by Dr Alan Renwick and Professor Iain
McLean
Electoral System Options
Paper prepared for the Joint Select Committee
on the Draft House of Lords Reform Bill
Dr Alan Renwick, University of Reading and
Prof. Iain McLean, University of Oxford,
11th January 2012
In connection with our oral evidence session with
you on 19 December 2011, we have been asked to answer a number
of questions concerning the operation of electoral systems—either
an open-list proportional system (open-list PR) or a single transferable
vote system (STV)—that satisfy two conditions:
- they allow voters the option
of casting a simple party vote;
- they allow voters to express preferences among
individual candidates across as well as within parties.
Before answering the specific questions, we think
it would be helpful to outline various forms that such systems
could take. We will outline two versions of open-list PR and two
versions of STV that would satisfy the two conditions.
Open-List PR Systems
We are aware of two countries that presently use
open-list PR and allow voters to express preferences across party
lines: Switzerland and Luxembourg.[459]
The systems used there allow voters to fill in their ballot papers
in a great variety of different ways: they can shift names between
lists, create new lists, delete names, and so on. Such complexity
may make sense where it has evolved over time, but we suggest
that it would not be desirable when designing a new system. In
the UK context, it would create great confusion and open the procedures
to ridicule.
Besides these cases, an attempt was made in the Australian
Capital Territory in 1989 to combine the principle of a list election
with that of the transferable vote, but the electoral system produced
was probably the most complex ever implemented. It took over two
months to count the votes and the system was quickly scrapped.[460]
Again, we suggest that this is an example not to follow.
We suggest two simpler ways in which open-list PR
could be combined with cross-party
preferential voting. The first
is a simplified version of the Swiss system. The second looks
(at least to voters) more like STV. These are only
illustrations of the sorts of system that could be adopted: much
more work would need to be done in evaluating options before a
precise recommendation could be made.
Option 1
The first system would give voters as many votes
as there are seats available in their region. Voters could cast
these votes in either of two ways: either by placing an X next
to a party (in which case all their votes would count for the
party's preferred list order; or by voting for up to seven candidates.
The ballot paper might be laid out roughly as shown below (for
the example of a constituency electing seven members).
When it came to the count, the first step, as in
any list system, would be to count all the votes cast for each
party, whether directly for the party or for the party's candidates.
This would determine the total number of seats allocated to each
party. Then the number of votes cast for the party directly and
for each of its candidates would be used to determine the order
in which the candidates were elected. As we mentioned in our oral
evidence on 19th December, there are several ways in which the
party's preferred order and the voters' preferences could be combined
to determine the final list order. The method selected is very
important: some methods give greater weight to the party's preferences,
others to voters' preferences. We would be happy to give further
guidance on this if the Committee wished.
This system would allow voters to express some preference
ranking among candidates: in a region where seven members were
being elected, for example, a voter could give three votes to
one candidate, two votes to another, and one vote to each of two
more. But it does not allow a full ranking. Giving voters the
opportunity to rank all candidates in order of preference would
require something like Option 2.
Option 2
Under the second form of open-list
PR, voters would have two options as to how to vote: they could
vote for a party or rank the candidates in order of preference.
The layout of the ballot paper would be
|
You can vote in one of two ways:
- EITHER vote for a party by placing an X in one of the boxes above the thick black line;
- OR indicate your preferences among candidates below the line by placing a ‘1’ next to the candidate you most favour, a ‘2’ next to your second favourite, and so on; you can express as many or as few preferences as you wish.
|
something like the following:
In counting the votes, as before, the first step
would be to count up the votes for each party. The simplest option
here would be to say that a voter who expresses preferences among
candidates is deemed to have voted for the party of their first-preference
candidate. But this would have the undesirable effect of allowing
a voter to influence the order of the candidates on a party's
list without giving support to that party. An alternative would
be to give each preference a fractional value such that the fractions
summed to 1. If a voter expressed two preferences, for example,
their first preference could give that candidate's party 2/3 of
a vote and the second preference 1/3. If a voter expressed three
preferences, these preferences could yield, respectively, 4/7,
2/7, and 1/7 of a vote for each candidate's party. The same fractions
could then be used in determining final list order.
This would allow voters to express a full set of
preferences. But it would be necessary to make assumptions about
the relative weight of these preferences in order to count them—assumptions
that might or might not express the genuine nature of voters'
preferences.
STV Systems Allowing a Party Vote
Two ways of combining STV with the possibility of
casting a simple party vote (so-called 'above-the-line voting')
are used in elections currently: the standard form used in Australian
Commonwealth elections and elections in three Australian states;
and an alternative form used since 2003 in New South Wales. We
describe these below as Options 3 and 4. We presume that, if either
of these systems were proposed for the UK's second chamber, voters
would be free to fill in as few or as many preferences as they
wished. In either case, the layout of the ballot paper might again
be roughly as shown above.[461]
Option 3
In the most familiar form of STV with a party vote
option, voters can either express a vote for one party or rank
candidates in order of preference. The instruction on the ballot
paper is the same as under Option 2:
|
You can vote in one of two ways:
- EITHER vote for a party by placing an X in one of the boxes above the thick black line;
- OR indicate your preferences among candidates below the line by placing a ‘1’ next to the candidate you most favour, a ‘2’ next to your second favourite, and so on; you can express as many or as few preferences as you wish.
|
Votes here are always counted as votes for candidates,
not parties. A vote for a party is counted as a vote for the ordering
of candidates determined by the party. The usual STV counting
rules are applied to these votes in the same way as to votes cast
below the line.
In the Australian version of this system, parties
are required to indicate ahead of the election their ordering
not only of their own candidates, but of all candidates: if all
of the party's own candidates are either elected or eliminated
before the count has been completed, votes cast for the party
will continue to transfer, as the party has indicated, to the
other parties' candidates. This requirement fits the logic of
the Australian system, under which a vote is valid only if all
preferences are filled in. Assuming that, in the UK, voters would
be free to express as many or as few preferences as they wished,
it would make sense to apply the same logic to above-the-line
voting and therefore not require parties to express their ranking
of other parties' candidates. It is an interesting question whether
parties should be allowed to express such a ranking.
The system as used in Australia has sometimes caused
controversy when candidates with few first preferences have been
elected because they received preference transfers from others.
Few voters are aware of how their party has ranked other parties'
candidates, so such outcomes can seem to have little to do with
voters' preferences. Concerns such as these prompted the adoption
of the alternative system in New South Wales following the 1999
elections. We describe this as Option 4.
Option 4
In this version, voters can either rank the parties
or rank the candidates. If they rank the parties, then their vote
counts first for the candidates of their first-preference party
(in the order determined by the party), then for those of their
second-preference party, and so on. Thus, it is the voters, rather
than the parties, who determine transfers from party to party.
The parties rank only their own candidates and offer no official
view on where votes should transfer thereafter. We reproduce a
full sample ballot paper here, though, again, it is only the instruction
that needs to change:
As noted above, this system was introduced in New
South Wales because of concerns that voters' preferences were
being transferred in ways that most voters had no knowledge of.
The Joint Standing Committee on Electoral Matters of the NSW legislature
reviewed the new arrangements in 2005 and expressed satisfaction
with them.[462]
Similar concerns to those that prompted the reform
in NSW have been raised in Australia at the Commonwealth level.
The Australian Parliament's Joint Standing Committee on Electoral
Matters advocated the adoption of the NSW system for Senate elections
in its report on the 2004 elections.[463]
The Government rejected this proposal, however, saying it believed
that such a change was "likely to result in increased complexity
and possible confusion for voters, leading to a potential increase
in the level of the informal [i.e., invalid] vote".[464]
The same Joint Standing Committee conducted a further investigation
on the specific issue of preferential voting among parties in
2009. This was in response to a Bill submitted by a Senator advocating
such a system. The Committee rejected the proposal on the grounds
that it would increase complexity and therefore, in all probability,
lead to a rise in the number of invalid votes.[465]
It should be noted, however, that the Bill had proposed that voters
be required to express a minimum of four preferences among parties
if voting above the line. We presume that no such requirement
would be proposed for the UK's second chamber.
General Observations
Before moving to the committee's questions, we should
like to make two general observations.
First, while we have tried to make the systems above
as simple as possible, all are complex compared to most other
electoral systems—including open-list PR systems without
the option of cross-party preference voting and pure STV systems—in
the sense that they increase the range of choice available to
voters. Voter choice is desirable in a democratic election, but
it can also become burdensome or confusing. Particularly when
it is proposed that second chamber elections should be held concurrently
with Commons elections that will use a different system (first
past the post), it is important to bear this complexity in mind.
Second, while much discussion has focused on the
degree of "independence of spirit" of members of the
second chamber, we think, as we said in the oral evidence session
on 19th December, that the electoral system will not be the primary
determinant of this independence, at least as regards the independence
of partisan members from their party. The non-renewable term is
the most important factor here. The likelihood of election of
non-partisan independents is another matter, and we discuss it
below, under Question 1.
The Committee's Questions
Through its Clerks, the Committee has asked us seven
specific questions. In light of the options laid out above, we
now address these questions.
Question 1. What would be the difference in outcome
between having an STV counting system with the above characteristics
[i.e., allowing voters the options of a party vote and cross-party
preference voting], or an open list counting system with the above
characteristics?
There are some differences between the systems described
above in the nature of the preferences that voters can express.
In addition, two sorts of effect can be noted:
1. Differences in the interpretation put on voters'
preferences. The systems count the preferences expressed by voters
in different ways. Most notably, list systems always count a vote
for a candidate in the first instance as a vote for the candidate's
party, whereas STV systems count a vote for a candidate solely
as a vote for that candidate. Under STV, therefore, a voter can
vote for one candidate from a party without giving any advantage
to any of that party's other candidates, whereas under a list
system a vote for a candidate can help secure election for another
candidate from the same party.
2. Differences in the amenability of the systems
to independents. In our oral evidence on 19th December, we suggested
that STV is more compatible with the election of independents
than are list PR systems. This is because large numbers of votes
cast for a popular independent under list PR can be wasted. There
is some reason to think that this tendency would be weaker with
the forms of list PR discussed here: voters would not need to
put all their eggs in one basket by supporting an independent,
but could rather split their vote. Nevertheless, such voters would
be giving weaker support to the independent than they could (without
risking wasting their vote) under STV, and we would therefore
still expect independents to perform somewhat better under STV.
By contrast, we do not think there is any reason
to expect any significant differences among the systems described
here in the degree of independence of partisan representatives
from their parties. These systems would all give candidates broadly
equal incentives to compete on the basis of their personal reputations.
More importantly, as we suggested above, the non-renewable terms
in the proposed second chamber would leave parties unable to coerce
rebellious members.
Question 2. Would putting the party voting
option below the line, rather than above, have any significant
impact?
We understand this question to relate to the physical
appearance of the ballot paper: whether the option to vote for
a party should come at the top or the bottom. So far as we are
aware, all jurisdictions that currently give voters the choice
of a party vote or a candidate vote place the party vote option
first (either on the top or on the left). It is reasonable to
suppose that reversing this order would increase the number of
voters expressing preferences among candidates, but we are not
aware of direct evidence on this point.
The Committee might remember that there is no reason
to expect rates of "above-the-line" voting to be as
high in the UK as in Australia. While fewer than 4 per cent of
voters voted below the line at the most recent Australian Senate
elections,[466] this
low rate can be explained in significant part by the rule in that
case that a candidate vote is valid only if all candidates are
ranked. In other systems where voters have a choice as to whether
to cast a party vote or a candidate vote, there is considerable
variation in the proportion of voters taking the second option:
around 20 per cent do so in Austria, two thirds in Belgium, and
90 per cent in Brazil.[467]
Question 3. How (particularly with an open list system)
would a party voting option work with independents? Should they
have an "independents" box (then get placed depending
on how many individual votes they received), or would they only
get votes if people voted for them directly under the line?
We suggest that an "independents" box would,
particularly in a list system, be undesirable. As we noted above,
in any system of list PR, a vote for a candidate is in the first
instance a vote for that candidate's party (or, more precisely,
for the candidate's list) as a whole. Thus, voting for a candidate
in an "independents" list could sometimes lead to the
election not of that candidate but of another from the list.
There are two alternatives: it might be possible
to vote for independents only below the line; or independents
might be allowed to register as one-person "lists" appearing
above the line. Independents are able to register this way in
Australia; in practice, some do so while others do not. Given
the bias in the Australian system towards above-the-line voting,
those who do not register to appear above the line are severely
disadvantaged. In the absence of the requirement to fill in all
preferences, however, the difference would be minor: it would
amount to a difference only in independent candidates' visibility
on the ballot paper.
Question 4. Assuming that the party's candidate ordering
has some weight, should the list of individual candidates below
(or above) the line be ordered by party in their order on the
party list?
If the party's candidate ordering has some weight,
this ordering should be transparent to voters. The most sensible
way of doing this is to give that ordering on the ballot paper.
An alternative would be to publicize the parties' orderings widely,
including in polling stations, and then use alphabetical, randomized,
or rotated ordering on the ballot paper. This might permit a purer
expression of voters' preferences among candidates. If the system
is designed, however, to allow parties' rankings to matter, there
would appear little reason to emphasize purity of voters' candidate
preferences, and using orderings that differ from the party's
ranking could create confusion.
Question 5. Should
electors voting for parties still order their votes (i.e. vote
for more than one party), or just put a single x against their
favourite party? Are both possibilities, and if so what difference
would it make to the results? Does the answer vary depending on
whether the chosen party has fewer candidates than there are seats
to be contested?
As described above, the standard form of STV with
above-the-line voting allows voters who choose the above-the-line
option to vote for only one party. But the New South Wales version
allows voters to rank parties in the manner suggested in the question.
The advantage of the NSW version is that voters can
control how their vote transfers from party to party, whereas
in the standard system these transfers are controlled by parties
in a way that is rarely transparent to voters.[468]
Exactly how the systems compare depends, however, on whether parties
are allowed and whether they are required to indicate ahead of
the election how a vote cast for their list will transfer not
only among their own candidates, but also among the candidates
of other parties. If such extra-party transfers were barred, the
problem identified in Australia would be removed, but rather more
votes would be exhausted before the counting process was completed.
The number of candidates that a party runs relative
to the number of seats available is one factor (though not the
only one) influencing the importance of inter-party transfers.
In so far as such transfers matter, it is important that voters
understand them before deciding how to vote. That could happen
either by ensuring that parties' transfer declarations are well
known or by allowing voters to dictate transfers. As we have suggested,
the Australian experience of the standard system (our Option 3)
is that, though transfer statements are public, most voters are
ignorant of them.
Question 6. Can electors voting for parties also
express individual preferences below the line, or is it an either
/ or situation? What difference would it make?
The four systems that we described above all require
voters to choose between a party vote and voting for candidates.
In STV systems this is necessarily so: under STV, votes are always
counted at the level of candidates: a party vote is simply a vote
for candidates in the order specified by that party. We see no
way of combining this logic with the possibility that voters could
vote both above and below the line.
Under open-list PR, it would be possible to revise
Option 1 such as to permit voters to vote both for parties and
candidates: voters could be allowed to spread their votes across
both above-the-line and below-the-line boxes. This would, however,
weaken the simplicity of the above-the-line option.
In principle, party and candidate votes could be
decoupled under Option 2 as well: voters' party votes could determine
the overall balance of seats across parties and their candidate
votes could determine the distribution of each party's seats among
its candidates. As we suggested above, however, it would be undesirable
to give voters the power to influence the list order of a party
they do not vote for, so we recommend against this possibility.
It might be sensible under Options 1-4 to include
a provision saying that where a voter does mistakenly cast both
party and candidate votes one of these shall be deemed to take
precedence.
Question 7. Should constituencies under either STV
or Open list systems (with the above characteristics) be no more
than six or seven members?
All of the systems we have described give voters
a great deal of choice. Just as under pure STV, we therefore recommend
that constituencies should elect no more than seven members in
any one round in order to prevent the range of options from becoming
overwhelming.
459 See Georg Lutz, "Open Ballot", in Josep
M. Colomer (ed.), Personal Representation: The Neglected Dimension
of Electoral Systems (Colchester: ECPR Press, 2011), pp. 153-74. Back
460
On this, see Parliament of the Commonwealth of Australia, Inquiry
into the ACT Election and Electoral System, Report No. 5 of the
Joint Standing Committee on Electoral Matters, November 1989 (Canberra:
Australian Government Publishing Service). Back
461
For detailed discussion of STV and its variants, see David M.
Farrell, and Ian McAllister, The Australian Electoral System:
Origins, Variations and Consequences (Sydney: University of New
South Wales Press, 2006). Back
462
Parliament of New South Wales, Joint Standing Committee on Electoral
Matters, Inquiry into the Administration of the 2003 Election and Related Matters,
September 2005, paragraph 5.65. Back
463
Parliament of Australia, Joint Standing Committee on Electoral
Matters, Inquiry into the Conduct of the 2004 Federal Election and Matters Related Thereto,
October 2005, pp. 224-9 and 232. Back
464
Government response to the report cited in note 4, p. 18. Back
465
Parliament of Australia, Joint Standing Committee on Electoral
Matters, Advisory Report on the Commonwealth Electoral (Above-the-Line Voting) Amendment Bill 2008,
June 2009, p. 24. Back
466
Australian Electoral Commission, www.aec.gov.au. Back
467
Lauri Karvonen, "Preferential Vote in Party List",
in Josep M. Colomer (ed.), Personal Representation: The Neglected
Dimension of Electoral Systems (Colchester: ECPR Press, 2011),
pp. 119-34, at p. 134. Back
468
The Australian Parliament's Joint Standing Committee on Electoral
Matters has noted that "the effect of the Group Voting Ticket
system is that only the very few above-the-line electors who bother
to inquire will have the faintest idea where their Senate preferences
are going" (Parliament of Australia, Joint Standing Committee
on Electoral Matters, Inquiry into the Conduct of the 2004 Federal Election and Matters Related Thereto,
October 2005, paragraph 9.32); it continued that the system "lacks
transparency, and results in electors ceding their preference
allocation decisions to the political parties themselves"
(paragraph 9.33). Back
| 
