[EM] 12/27/02 - Northern Ireland Assembly:

[James Gilmour]

Donald

The most important points in your post are at the end, so I am going to start
there.  Much of the rest is detail.

>You wrote
> I am not advocating only one area elections, what I am saying is that there
> are methods that will provide near 100% party proportionality for
> constituency elections.  Top Up MMP is one of those methods and the only
> method in use today.

You may not be advocating "one area elections", but you are quite clearly
advocating two other features of voting systems that have profound political
effects, ie "near 100% party proportionality" and national aggregation of votes.
These raise issues that go far beyond the details of the voting system.  I have
expressed my views on them both before, but they have been studiously ignored.

Electing a national parliament or a city council is, or should be, about far more
than securing "party proportionality".  If that is what you want, or more
correctly, if that is ALL you want, there are several methods available by which
it can be secured.  But I, and many others, do not share this narrow view of
political representation.  To me, representation, and proportional representation,
is about much more than party PR.  At the very least, it is also about getting PR
WITHIN each party.  (If you need any illustrations of the importance of this, you
need look no further than the two main political parties (Conservative and Labour)
in the UK over the past 50 years.)  None of the Party-PR systems presently in use
give PR within the parties.  Open lists are better than closed lists, but they
still fall far short of giving PR within the parties.  Top UP MMP does not give PR
within parties.

But elections are for electors, not for political parties, essential though
parties may be.  What the electoral process should be about when electing a
parliament or a council, is obtaining PR of the voters' views, as expressed by
their responses to the candidates who are prepared to offer themselves for
election.  That goes well beyond party PR and indeed, often cuts across the
parties, much to the annoyance of the party apparachiks.  But elections are for
electors.  But maybe you do not agree.

We know from practical experience of preferential voting, especially with STV-PR,
that voters are motivated to cast their preferences by factors other party,
important though party is to most voters.  But locality (place of residence),
religious affiliation, ethnic origin, gender, views on specific national or local
issues are also very important to many voters.  Given the chance to express their
views (through preferential voting), electors do.  The fact that most voting
systems suppress the expression of such diverse and complex views, does not mean
those views do not exist.  But party PR voting systems make us all behave as
though those differing views did not exist beyond the differences
institutionalised in the political parties.  That is a severe limitation on
democratic representation and one I would not readily accept.


You also advocate national aggregation of votes to determine the party PR.  While
this may not be a problem when electing a city council with not more than, say, 20
members, it has extremely undesirable consequences when electing a national
parliament with 100 to 650 members, in that it gives representation to many, very
small political groups.  The undesirable political consequences of this have
generally been recognised and an arbitrary threshold imposed for seat allocation
to exclude the smallest groups in most countries where party PR is in use.  This
applies also to Top UP MMP as implemented in Germany, where there is national
aggregation of votes to maximise party proportionality AND an arbitrary threshold
to exclude parties that get less than 5 per cent of those votes.  I can understand
the political rationale for this contradictory procedure, but it is a totally
illogical.  The point at which these thresholds are set, is of course, completely
arbitrary and determined in practice by political expediency.


One smaller point that you have also ignored, is that MMP elects two very
different types of member - those from single-member constituencies and those from
the parties' regional top-up lists.  This, too, has had undesirable political
consequences, at least in countries like the UK and New Zealand.  I would predict
similar problems in any country with a broadly similar political culture.


> Thank you for posting imformation about the Northern Ireland Asssembly.  I
> prefer to discuss information from real elections.  There are too many
> concocted elections presented on this list.
>
> You wrote: "Northern Ireland Assembly
> ===================
> 108 members elected from 18 constituencies each returning 6 members.
>
> Overall summaries based on first-preference votes must be interpreted with care.
> STV-PR guarantees PR only WITHIN individual constituencies.
> Although constituencies of 6 members guarantee representation only to
> groups that
> can secure at least one-seventh of the vote within a district, groups with much
> less overall support usually do gain representation in STV-PR elections..."
>
> Donald here:  A group with much less overall support will be able to gain
> representation only if the group has most of its support bunched up in one
> or two constituencies, but if their support is spread even over all the
> constituencies, then the group will not elect any members.

That is true and that is what happened in that NI Assembly election.  But it
reflects the realities of politics - activism is usually concentrated or at least
focused in specific geographical areas.


> James: "...eg the Northern Ireland Women's Coalition, with 2% of the
> first-preference votes and 2% of the seats."
>
> Donald:  The only way this could have happened is if the Women's Coalition
> had most of their support bunched up in one or two constituencies, like
> 0.80% of their support in one constituency and another 0.80% of their
> support in another constituency (or 1.60% of their support in one
> constitucency), if so, then they would elect two seats, which is about two
> percent of the 108 seats.  Had their support been spread out even over all
> 18 constituencies, they would have not elected any seats.

Agreed - see immediately above.

> It is a condemnation of the constituency STV method when a small group must
> have most of its support in one or two consituencies in order for the group
> to receive its fair share of proportionality.

No, it is not "a condemnation".  It is reflection of the practical trade-off
between district magnitude and proportionality.  Most electors in our political
culture want to retain locality (modest district magnitudes) as a feature of the
voting system.  Most give higher priority to locality than to absolute
proportionality.  I would refer you to the more detailed information I have
provided elsewhere on the NI District Councils and the Scottish Education
Authorities.  (I can post it in the Files section of this List if members all have
access.)


> James: "In an STV-PR election the actual proportion of wasted votes cannot
> be calculated simply from the district magnitude (number of members per
> district).  With 6 members, representation is guaranteed to 85.7% of the
> voters (6/7), ie up to 14.3% of those who vote may have their votes wasted.
> But in actual STV-PR elections the proportion of wasted votes is usually
> much smaller."
>
> Donald:  No, it is not much smaller, it only appears to be much smaller
> because the use of the Droop quota has forced most of the wasted votes to
> become exhausted votes (`A Rose by any other name is still a Rose.').

I have cut the rest of your post because the real issue is not in those details
(and anyone who wants, can see them there).

I do not disagree with your analysis of what happens to actual ballot papers in a
standard STV-PR count.  I am very well aware that the standard STV-PR rules are
full of "short-cuts" which were considered expedient when large public elections
had to be sorted and counted by hand.  The way non-transferable votes are handled
when surpluses are transferred is just one of those short-cuts.  The short-cuts
are all designed to minimise paper handling and to maximise speed in obtaining an
acceptable result.

The Droop Quota was adopted as one of these short-cuts, because it most rapidly
identifies the required number of winners (attainers of a quota of votes) with the
minimum of number of stages in the count.

If you find these short-cuts unacceptable, I would suggest you look at Meek's
version of STV-PR.  This was devised to deal with all the problems you see in
STV-PR, and some more.  Meek STV-PR will be used for local authority elections in
New Zealand.  We shall not use it in the UK for the foreseeable future because it
requires computer-only counts.


One further point about the Droop Quota.  You said:
> While I approve of a policy of averaging votes for political parties,
> slates, coalitions, or whatever, the Droop quota is a crude way of
> averaging, for the Droop causes these problems of wasted and/or Type Two
> Exhausted votes.  The Droop quota is the major flaw in current STV,

I am not sure what you meant by this, because there is no averaging of votes for
political parties in STV-PR.  Political parties are irrelevant in an STV-PR
election as each candidate is treated as an individual.  Party PR is usually an
outcome, but it is never the objective of the STV-PR voting system.



> The sad thing about STV is that over the years the politicians have fiddled
> with the rules of STV so that now it is not a good PR method.  Top Up MMP
> is superior to Droop constituency STV.

This statement is untrue.  All the "faults" in the STV-PR rules you have
identified were introduced as practical short-cuts for hand-counted ballots.
Quite separately, we have seen examples of bad implementation, where the district
magnitude was deliberately manipulated for party advantage - the infamous
"Tullymander".

James

----
For more information about this list (subscribe, unsubscribe, FAQ, etc), 
please see http://www.eskimo.com/~robla/em