Scientific method of elections.

To home page.

Links to sections:


Preface on evidence to the Independent Commission on Voting Systems.

This is evidence presented, but not alluded to, by the Independent Commission of 1998, other than by letter, as an 'extremely interesting submission.'
The commission's report refers to organisations, politicians and academics, as well as several pages listing 'interlocuters' from foreign countries.
The others were held to be too numerous to mention. The number was not even mentioned. I could not find even the name of one ordinary member of the public -- a mere voter -- acknowledged by the Voting Commission ( in the Stationary Office edition of the Jenkins report ).

The commission was supposed to be open to the general public. But not so much as one extract from their written submissions was thought worthy of including in their 'Key Evidence'.
This consisted entirely of the comments or studies from those of power or influence, which evidently was all that weighed in the minds of the commission, despite the report's repeated protestation they gave 'most' or 'very serious consideration' to the dedicated support for STV. Not that there wasn't more than ample, in the Key Evidence, to point the commission in that direction.

The effect of this commission on ordinary people, for whose benefit it is supposed to be concerned, must be to conclude that taking an interest in official studies is a waste of time and effort.

Scientific method of elections.

Summary of: How To Do It.

To top.

The four main measurement scales define the single transferable vote as scientific method of elections.
Party-proportional counts generally lack the middle two, the ordinal scale, given by a preference vote, and the interval scale of transferable votes in surplus of a quota. This lack explains why party list systems have no coherent principle of electing individuals.

Likewise, combining simple majorities with lists uses only the first and fourth scales, the classificatory scale ( for one person one vote ) and ratio scale ( for proportion ) without the logical progression thru the other two scales.

STV contains four logically possible kinds of PR: 1) with respect to parties, preferring partisanship, primaries, coalitions, referenda.
And likewise: 2) with respect to quotas between, within, across and without constituencies.
As proposed, STV satisfies 6 ( and potentially all 8 ) of these kinds of PR.

The single member system has one kind ( PR between constituencies ).
( Localised multi-member constituencies offer the most convenient and accurate 'geometry' of historic communities. )
List systems have two kinds of PR: between parties ( partisanship ), and, without constituencies.
So, combined systems average between one and two kinds of PR.

STV offers comprehensive PR proper, as a theory of relative choice that consistently generalises the vote and count, from the simple majority system of one-preference vote for one-member majority count, to many-preference vote ( of ordered choice ) for many-member majority ( Droop quota ) count.

Mill's democracy goes beyond absolutist schisms from 'maiorocracy' or proportional partisanship.
A deductive explanation of elections offers a practical conciliation of unity ( or fraternity ) in liberty, thru the principle of equality in transferable voting, that proportionally represents all group attributes by individual preference.

Summary of: How Not To Do It.

To top.

Additional Members were an ad hoc attempt to save disproportionate first past the post elections as a constituency system, that failed because of the anomalies from second ( or third ) past the post MPs.
So, the established AMS of the Double Vote was fallen-back on. A second X-vote for party list portions of seats reduces the number of single-member monopolies that become more disproportionate and more polarised. This combined system of inconsistent axioms is inherently unstable ( most clearly failing the test of Euro-elections ).

This compensated parties system denies the public knowledge thru freedom of choice from PR of individuals and thereby other groups in society ( failing, for instance, Independents and independence, most obviously in local elections ).

For general elections, the Double Vote has an ad hoc threshold to suppress the consistent application even of its party-privileged proportional count, lest small parties hold coalitions to ransom.

The Double Vote is a particular patch-up not a truly general theory of choice. Separate 'personal' and 'party' votes deny a proper 'control' in the electoral test of personal choice in relation to party choice.

In terms of Britain's 'safe seat' system, the additional list members are a second safety net for failed candidates and failed parties, tardily and unstably going thru all the ( unpreferred ) coalition combinations.
Or, just a third party, perhaps, is squeezed by the two-party system of single members. Yet the main parties cannot govern without this confirmed stooge in the Double Vote system. Unless they cut out the middleman, this vulnerable, because narrowly undemocratic, base for a king-maker chooses which unstable course to take, by way of successive left or right wing coalitions.

The Double Vote denies basic electoral rights of individual choice and electing a government.


To top.

'Science Is Measurement'.

And to give the game away right away, only one voting system follows the four main scales of measurement, known to science.
For Sidney Siegal, the usual problem of the behavioral sciences is, like physics, to go beyond the first two scales ( which his book on non-parametric statistics is mainly about statistical tests for ). And coming to the hundreds of voting methods, that have been invented, all -- but one -- miss out, typically two of the measurement scales.

But the four scales are built-up by logical progression, and the usual omission, of the middle two scales, renders those voting methods unscientific. The practical criticisms justly brought against them, stem from that missing mensural information.

A Measure Of Representation.

1) The Nominal Scale: The Spot Vote.

The first scale of measurement is the classificatory or nominal scale. 'One person one vote' classifies all adults as voters. Everyone is at least nominally represented. Hence, the chivalrous tradition that a single member will work for all his constituents, including those who did not want to be saddled with him. He is their nominal or named representative but not necessarily who they actually voted for.

X marks the spot gives rise to a two-party system because it can only express a single preference between one of two candidates, in a system of single member constituencies. But British voters have sought more than the simplest two-way choice, which is all this system affords.

Even a mere two-party choice of government suffers already from a fatal defect. Equal electorates are no guarantee against disproportionate results. One party may pile up huge majorities. The other party may slip first past the post to win more seats with fewer wasted votes. ( See Enid Lakeman: How Democracies Vote. ) That was how the minority won twice in South Africa, resulting in the unconstitutional Apartheid regime.

2) The Ordinal Scale: The Alternative Vote.

To top.

More than two candidates poses the dilemma of giving a single-preference X-vote either to one's first preference or a second preference with a better chance of keeping out a least preferred candidate. It is the familiar poor choice between a wasted vote or a tactical vote.

Elections are a test of public opinion. Scientific method applies to the electoral test as to any other scientific experiment. Experimenters control any chance factors that might obscure the law, which explains the results they are looking for. In the experiment of an election, the law to be observed is the law of a majority count, in terms of which the results of the voting are electively understood.
( We will show the concept of a majority count generalises to include proportional counting, and become a general law of the count. )

Experimental control is given the voters, by replacing the one preference spot vote with a many-preference vote for candidates in number order of choice: 1, 2, 3, etc. This is simply called preference voting. An ordered choice is on an ordinal scale of measurement.

The 'control' is of the chance of split voting between two candidates, one of whom might be more preferred than a third candidate, winning more votes than either but less than both. To know the truth of the matter, it is necessary to control such contingencies, which obscure the law of majority election, that doesn't depend on the chance of one of the losing candidates not standing down.

Indeed, the Second Ballot system obliges the candidate, with the fewest spot votes, to stand down, to allow his voters to contribute to another candidate's reaching an overall majority. This allowance of an ordered choice is, in democratic terms, a preferential suffrage, and, in scientific terms, a preferential control.

The Alternative Vote effects a Second Ballot, or as many successive redistributions of the votes from eliminated candidates as it takes, to find the overall majority candidate.
Churchill called this process of elimination 'the worst votes for the worst candidates'.

The short answer is that the same argument applies with more force to the spot vote. A choice between a wasted vote or a tactical vote is between using the spot vote as a first preference or a second preference. Denying the public a preference vote does not abolish the reality of relative choice. It only degrades the public's freedom to express it. Thus the X-vote makes some voters use a worse vote for a worse candidate than they need do with a preference vote.

So, Churchill's famous phrase applies more to the spot vote than the preference vote. ( As a Liberal, he admitted to the Proportional Representation society that the Single Transferable Vote is 'excellent in theory' but meanwhile would support the Second Ballot. Tho, returning to the Tories, he opposed the Alternative Vote. )

Maude and Szemerey ( Why Electoral Change? ) note a further objection to preference voting as 'circular'. Some permutations of preference voting may cancel each other out:

Voter lst 2nd 3rd
A Labour Conservative Liberal
B Liberal Labour Conservative
C Conservative Liberal Labour

This set of preferences reversed would also cancel each other out, making two self-cancelling sets. These cover all six permutations of preference between three candidates. But taking each preference from one set, and replacing it by each of the preferences, in turn, from the other set, you get three times three new sets of preferences, in both the original sets (or 3 × 3 × 2 ) equals eighteen non-cancelling sets of preferences.

Compare that with having only a single-preference spot vote. This merely minimises the amount of preferential information about voters' choices. The truth is not improved by suppressing parts of it one doesn't like, to reduce it to a spot vote for a two-party system:

Voter A for lst: Labour, 2nd: Conservative.
Voter C for lst: Conservative, 2nd: Labour.

Here there is only one possible set of preference permutations and it is a self-cancelling set. So, contrary to objection, the more preferential the voting, the vastly decreased the probabilities of turning up circular or self-cancelling sets of preferences.

3(i) The Interval Scale ( Assumed ): Borda's Method.

Borda's method anticipated Churchill, by giving the least weight in the count to the lowest preferences or 'worst votes'. Unlike the Alternative Vote, Borda's method ( 1770 ) makes standard mathematical use of all the preferential information, weighted relative to order of importance.
Weighting with the harmonic series counts 1/1, 1/2, 1/3 etc for each lst, 2nd, 3rd etc. preference, respectively. These are assumed values at which the strength of voters' feelings for candidates perhaps falls off.

( At any rate, weighting with the arithmetic series perhaps over-emphasises, as the geometric series would under-emphasise, the importance of lesser preferences, as the number of candidates increases. )

Borda's method was the first attempt to overcome the 'Condorcet paradox' ( which has a bearing on Churchill's aphorism ). With equally weighted preferences, the least preferred of three candidates, in the first round, might be the most prefered on the second preferences of voters for the leading two candidates. This can produce a contradictory result to the Alternative Vote or Exhaustive Ballot, including its limited version, the Second Ballot, which eliminates candidates with the least first preferences.

3 (ii) The Interval Scale ( Real ): Gregory's Method.

To top.

In simple statistics, interval weighting means taking into account the number of items, in each class of an ordered table of data, to arrive at a rational average. When it is not known how many items there are in each class, the number may be assumed. This is akin to Borda's assumed count. When it is known, we have a real interval scale, as is the case with Gregory's method ( 1880 ), used as 'the Senatorial Rules' in some Commonwealth countries.
This is why Gregory's method supercedes Borda's.
Unlike Borda's method, second or lesser preferences cannot count against the election of first or greater preferences.

In single member constituencies, a winning candidate usually has a margin of redundant votes over those just needed to elect him first past the post. This surplus vote is the interval between his total vote and his elective vote. Gregory's method makes surplus votes transferable for the election of second or next preferences in a multi-member constituency.

But all the first-elected's voters are equally entitled to a preferential say in the redistribution of his surplus vote. For example, suppose 100 votes is enough to elect candidates but the most popular candidate received 160 first preferences, a surplus of 60 votes. There are 160 voters equally entitled to prefer how the surplus weight of 60 votes should be redistributed. So, all 160 voters' second preferences are weighted at 60/160 = 3/8 of a vote, their 'transfer value'.

Say 80 out of 160 transferable votes went to the same second preference, giving him votes to the value of 3/8 × 80 = 30 votes. If that candidate already had 70 or more votes, he would now have the 100 votes to become the second elected candidate in a multi-member constituency.

The surplus votes from the most preferred candidates, with transferable voting, is the logical opposite of deficit votes from the least preferred candidates, with the Alternative Vote, 'the worst votes for the worst candidates'. Then, transferable voting must be the best votes for the best candidates, for Churchill's catchphrase to have any meaning.

4) The Ratio Scale: The Droop Quota.

To top.

Given best votes for best candidates, by transferable voting, Churchill's other cleverest of electoral sayings implies its multi-member constituencies: I would rather be one-fifth of the representatives for the whole of Leeds than one representative for a fifth of Leeds. This preference is worthy of the meaning of the House of 'Commons' or communities. Whereas the single member system has turned parliament into a House of Monopolies.

Electing one member with half the votes ( or just over ) is only the first in a rational series of possibilities. Two candidates can be elected on quotas of one-third the votes each, for a proportional representation of two-thirds the voters in a two-member constituency. Three constituency seats are won on quotas of a quarter the votes each, proportionally representing three-quarters the voters. And so on.
This rationalisation of representation is known as using 'the Droop quota' ( 1869 ).

Votes in surplus of a quota are transferable by Gregory's method. If all surplus votes have been transferred and any seats remain to be filled, then the candidate with least votes stands down for redistribution of his next preferences.

This standard procedure eliminates all but the most residual kind of tactical voting. For formal correctness, Brian Meek and later D R Woodall ( Representation, Journal of the Electoral Reform Society, Vol. 23 no 90 ) worked-out an exhaustive computer count for transferable voting. The result can be checked with a calculator.

Roger Penrose ( Shadows Of The Mind ) warned of computer fraud in elections.
Transferable voting should be least vulnerable, as a national complex of individual preferences. Whereas list totals of 'party' votes might only need a nudge one way or the other to falsify the result decisively.

To sum-up, we have related a coherent electoral procedure by following the four main scales of measurement, in: 1) a classification in terms of one person one vote; 2) the one vote is in order of preference; 3) each single vote is transferable in that order, to the interval value of surplus votes over the quota, which is 4) the Droop quota, an elective ratio of votes to seats in a multi-member constituency.

In fact, we have just defined an existing electoral system known as proportional representation by the single transferable vote. Therefore, STV/PR is the scientific method of elections.

Mensurally Deficient Non-Transferable Voting.

The logic of measurement shows-up the deficiencies of non-transferable voting systems. An X-vote for one of two parties in a single member system is the least measured system of elections, offering the least choice.

Prof. David Butler's evidence to the Jenkins Commission showed that for a single seat election between more than two candidates, the alternative vote is not mensurally acceptable, as I had assumed.
However, Alternative Vote or Exhaustive Ballot offer a full order of choice compared to the Second Ballot or Supplementary Vote ( recommended by the Plant report ) only offering a first and second preference ( or, rather, a two-preference vote, not necessarily first and second choice ).

Borda's method, unlike the Alternative Vote, attempts to account for the relative importance of preferences, with an assumed interval scale weighting. It is worth emphasising that, for developing the scientific measurement of elections, multi-member constituencies are essential, as Gregory's method of real interval scale weighting measures the way voters' later preferences actually do decrease in importance.

Systems of 'PR', that are only based on a party-proportional count of the spot vote, generally lack both the ordinal scale of measurement in preference voting and the ( real ) interval scale in ( Gregory's method of ) transferable voting.

That is why party list systems have no principled means of electing individual candidates. The fault remains when lists are combined, with a single member system, to give Additional Members.

Additional Member Systems, such as the German Double Vote, are unscientific. The first vote classifies one person with one choice for a representative's majority. The second vote counts proportional partisanship in ratios of seats to party votes. The classificatory scale should progress thru the ordinal and interval scales to the ratio scale. Instead, the Double Vote treats its mere two scales as two separate systems.
It brings to mind the proverbially blindfolded visitors pronouncing on the oddest menagerie, really the different parts of an elephant.

The Empirical Rationalism Of Scientific Elections.

The German Double Vote and the Italian constitutional reforms of the 1990s, also introducing single member constituencies, were misconceived attempts to stabilize Continental rationalism with British empiricism.

The one-sided philosophies of British empiricism and Continental rationalism are reflected in their respective Common Law and Roman Law. This separate tradition of understanding their affairs is also found in their respective voting systems. Most English-speaking countries stick to the ( limited ) evidence of an X-vote for an individual candidate first past the post, resulting in rationally 'wasted' votes. Whereas, mainland Europe attempts to solve this problem with a dogmatic rationalism of proportional partisanship, disregarding who the voters' most prefered individual candidates might be.

Science is a working partnership in empirical rationalism. Rational system and empirical investigation work better together. The single transferable vote ( STV ) gives greater evidence in a whole order of individual choice, that is a preference vote. And STV is more rational than a partisan count. The STV count preferentially rationalises choice not only between parties but within parties, across parties and outside parties. This system widens the freedom to know public opinion.

The Explanatory Power Of Transferable Voting:

1) With Respect To Parties.

A good scientific theory of choice is judged by the range and power of its explanations. STV analyses the four logically possible ways to consider individual choice in relation to party support: choice between, within, across and without parties.

The four relations correspond to the four scales of measurement.
Choice between parties classifies choice into party choices.
Choice within parties orders choice of party candidates.
Choice across parties is across party intervals on the political spectrum of uniquely individual 'colors' of candidates. The 'intervals' are the broadly recognised party colors, 'Reds', 'Greens' etc.
Choice without parties, of Independents ( or independent minded nominees ) for non-political ( aspects of ) elections, is possible on a ratio of votes per seats for individual candidates, rather than party lists.

These four relations of party to personal choice may be familiarly named, as:

1)partisanship, by prefering several candidates all of the same party.
Example: A divided community will show little cross-party voting, tho even in Northern Ireland it is not inconsiderable, and STV is the only system that impartially permits the links as well as the breaks to show.

2)primaries, by prefering certain candidates to others of the same party.
Example: The Plant report complained of the high turnover of MPs in each Irish party because of intra-party competition. As Maude and Szemerey say, 'There are no safe seats with STV.'

3)coalitions, by prefering the candidates of the parties most agreeable to each other, ensuring the prefered majority government, by more than one party, if necessary.
The more independent-minded the candidates prefered, the more individualist the parliament, as up to the nineteenth century. The more consensus candidates are prefered, the more national the government, as in emergencies, like war.
Example: Fine Gael and Irish Labour successfully asked their voters to extend their preferences to partner candidates.

4)referendums, by prefering candidates of different parties or Independents on some question of individual conscience or national destiny.
This recognises the growth of single-issue campaigns outside the party system. Included are the civil rights movements for the representation of social groups other than party, such as age, sex, race, class, creed, tongue, tribe or jobs etc, in the desired combinations individually prefered by the voters.

Example: In 1922, regardless of the two main pro- and anti- parties' pact, the Irish people were able to order a choice of several pact and non-pact candidates, in multi-member constituencies, according to their position on the Anglo-Irish treaty. So, a ( large ) majority wish passed into government and the civil war could be ended.

For political activists only concerned with the rivalry for power between parties, party lists or list-combined systems of proportional partisanship, deceptively called 'PR', were as good, or better for their purposes, than a comprehensive proportional representation, not only between but within, across and without parties, afforded by transferable voting.

The Explanatory Power Of Transferable Voting:

2) With Respect To Constituencies.

To top.

There are four relations of preference voting with respect to parties. Less well appreciated still, for purposes of electoral reform, is that these four relations also hold for the quota count with respect to constituencies.

PR Between Constituencies means the number of seats per constituency are in proportion to the number of electors per constituency. The 1979 Tory government's equalisation of electorates was a PR between constituencies ( but not within them ). But PR can be between multi-member constituencies. In fact, the constituencies don't even have to be a uniform system.

For example, the PR can be between one-member constituencies of, say, 73,000 electors each. ( Dividing nearly 44 million British electors by about 600 MPs; these figures are only approximate. ) Or, PR can be between two-member constituencies ( the historic English system ) of 2 × 73,000 = 146,000 electors each; or three member constituencies of 219,000 electors each. And so on.

But equally, you could have a mix of one, two, three etc. member constituencies, to match the varying size of communities. And there would still be PR between these constituencies, because the ratio of constituency electorates to constituency seats is the same.

The 1979 Boundary Commission's equalisations caused an uproar all over the country. Local interests contested various excisions of wards. The Home Affairs minister, responsible for the directive, protested against himself with two other local MPs, for the consequences to his own area.

Robert Blackburn ( The Electoral System in Britain ) chronicles Labour government delay and Tory government speed-up, of the Commission's work, to increase their respective parties' number of seats at the next election.
Calculating party HQs know what they are doing. The consequent maximising of Tory seats out of all proportion to their popular support in the 1983 and 1987 elections must be considered as gerrymandering on the grand scale, by following the letter of electoral law rather than its spirit.
Less hypocritical perhaps were Crossman's diary confessions of rigging local government elections first past the post.

Marginal changes to simple majority constituencies are only too liable to change the result. The Boundary Commissioners' contentious and expensive revisions were stalled by the Labour opposition taking them to the High Court.
The single member system has made the Boundary Commission perversely epitomise all that is blurringly unstable and fragmentorily transient to Britain's historic communities. After six years work, in 1983 only 48 out of 635 constituencies were left unchanged. The Tory seats were 19% too many for their share of the vote.

Author of Inside The Third World, Paul Harrison ( Inside The Inner City, 1983 ) gave evidence against the single member system as socially disruptive, p.417-18 ):

Our system of representation by single member constituencies also militates against reforms that could help the urban or regional poor, and, indeed, against any widespread awareness of their problems. Britain's marked social segregation has created constituencies the bulk of which are either predominantly English middle class and Tory, or predominantly working class, Welsh or Scottish, and Labour. Thus the bulk of Tory MPs, and indeed of Tory voters, have no direct experience of widespread and acute need which could awaken their compassion.
For the inner-city MP the problem of lack of widespread awareness is even more acute. Inner-city Labour MPs have trouble convincing even other Labour MPs,.. of the extent and gravity of problems in housing in employment ... law and order.'

(P. 434 ): The British political system cannot be relied on to respond constructively to this challenge. The first-past-the-post system allows governments to be elected with ... as little as one third of the electorate. Voting is largely determined by perceptions of self-interest. The groups most damaged by recent trends -- the poor, the unemployed, council tenants -- still remain geographically concentrated minorities. A sufficient proportion of voters have seen their fortunes improve despite recession to make the continuation of divisive policies possible.

Robert Newland of the Electoral Reform Society and the Liberal/SDP Alliance Commission's first report on Constitutional Reform proposed a normal distribution of one- to eight-member constituencies, about an average of four to five.
This is modest compared to the multi-member constituencies that would be needed today for the larger shires of 'the Commons' from its beginnings over 700 years ago. And most of that time they were without railways or motorways.
( The notion that whole counties would be unmanageable constituencies is laughable by American standards of travel. )

But STV electors for a borough or smaller shire wouldn't have to look into one-seat broken-mirror constituencies to see their representation. Each STV MP would know from surgeries and canvassing returns where his main support is clustered ( urban Labour, suburban Tory or whatever ). So, each MP has his main districts to cultivate within the multi-member constituency. That is the people to whom the MP is literally accountable for his quota.
The voters are, after all, supposed to be choosy about who represents their interests, four or five MPs giving them their choices.

This constituency system offers communities the most convenient and accurate fit: that is the scientific standard of measurement.
In physics, a system of geometry would be chosen for the most convenient and accurate measure of spatial phenomena.

PR within constituencies means each seat in a constituency is won on a quota or elective proportion of its total vote.
For example, we assume ( after Devolution ) about 60 Scottish MPs for the UK parliament. The Scottish electorate is nearly 4 million. And ( nearly 4m )/( about 60 ) equals perhaps 66,000 electors per Scottish seat.

Suppose sparsely populated Scotland averaged constituencies of about three members, with an electorate of some 200,000. ( There'd be 4 to 6 member lots in built-up areas. ) With 75% turnout, the total vote would be 150,000. The Droop quota is ( 150,000 )/( 3 + 1) = 37,500 votes, each of the three MPS had to win.

PR across constituencies means that constituents can also vote for candidates to other constituencies in the region. Some candidates may choose to stand solely on regional issues and not belong to any constituency. All candidates in the region ( or nation within the UK ) are then electable on a regional quota. In Scotland, this would be the Droop quota, taken of the total Scottish vote. On a 75% turnout this would be about 3 million, divided by ( 60 + 1 ), or roughly 50,000 votes.

So, with PR across constituencies, these Scottish ( UK ) MPs would need to win some 12,500 more votes, than in a ( Scottish ) average three member constituency. But they could appeal to Scottish voters outside their constituency. Candidates with personal reputations or party organisations across Scotland ( or equivalently an English region ) would have a better chance of election.

This answers the Plant report's question about STV being proportional enough in Wales with three member constituencies. Actually Wales has been projected to average four member constituencies, enough for four major parties.

PR without constituencies would be essentially the original version of transferable voting by Hare's system. ( The United Kingdom is here not being considered as itself a constituent part of a European parliament. ) The without-constituencies quota would be the nearly 44 million British electors, say, on a 75% turnout about 33 million, divided by ( 600 + 1 ), or about 55,000 votes.

So, a Scottish candidate with a reputation thru-out the UK might hope to pick up another 5000 votes outside Scotland. Most likely he would have enough Scottish support not to need them, for this small increase in the quota.

John Stuart Mill agreed to stand for the constituency of Westminster but on national issues rather than constituency issues. He introduced the two greatest democratic causes to Parliament, votes for women, and Hare's system. Could Mill have drawn on votes outside his constituency, his national reputation would certainly have re-elected him.

Nineteenth century individualists were perhaps less concerned with parties and constituencies than twentieth century collectivists. The lack of communication between these two sides of democracy, the representation of the individual in the community, has sadly impaired it.

In deciding a new constitutional settlement, one has to look to possible futures, as well as the past. This is a question of foresight, not great changes made before people are ready for them.

Providing PR without constituencies could add to STV's powerful array of in-built electoral functions. A computer count of STV could rank MPs in ascending order of choice up to the popular lead. That might confer a presidency or legitimise Britain's alleged 'Prime Ministerial government'.

This leadership count, in a series of rounds, could reduce the national Droop quota's quotient, one seat at a time, to ( 599 + 1 ), (598 + 1 ) and so on, thus eliminating one candidate at a time. This could rank MPs in order of popularity. Those, low down, might be disqualified say, from a Cabinet post, or the Chair of a televised Select Committee. At any rate, until they became better known and proved their worth.

The point is that one voting system can evolve the features of just about any democracy one cares to think of. Because STV is the general theory of elections, it is an amazing multi-purpose democratic tool-kit.

Isaac Asimov said: Don't waste your admiration on warnings, like those of H G Wells, that came true. The surprise is that most people ignore obvious dangers. But, not being stupid, that is surely because, as Shirley Williams said, people think politics is not for them. People wont react if they think they have no effect. Hence, the need for an education in electoral empowerment.

Unfulfilled Possibilities Of Non-Transferable 'PR'.

To top.

A UK quota probably wouldn't be an option till the regions and nations within the UK each had adopted their across-constituency quota. This itself might be done in stages. For instance, Glasgow might rather be one ten-member constituency than two five-member constituencies, north and south of the Clyde. Sparsely populated Scotland would be most likely to adopt a Scottish quota. But that would be for the Scottish people to decide, after the introduction of PR ( within constituencies ).

Overall STV/PR refutes the stock criticism that STV cannot be made proportional enough in the context of a constituency system.
Of the four relations each, of transferable voting to parties, and to constituencies, we can take or leave STV across constituencies and without constituencies. This gives 8, or 6, possible ways to relate parties and constituencies to other voting systems.

Party list systems relate to these 8 possibilities in only 2 ways, namely PR between parties and PR without constituencies. In particular, the latter way is true of national lists. A regional list could be said to give PR between constituencies for Euro-elections. But still only 2 of the 8 logical possibilities are catered for.

Party lists impose dogmatic partisanship on the people. They also impose a constituency-sceptical vote, because lists presume no pattern of voting for local constituency candidates.

Additional Member systems combine these deficiencies of party lists with the limitations of the single member system. All voting systems with any egalitarian pretensions to democracy are to some extent proportional systems.

The single member system, with equal electorates, gives PR between constituencies. But this is only one out of eight logical possibilities for PR. And each half of the combined system only half gives its one or two out of eight kinds of PR, proper.

No rival systems come remotely close to the six- or potentially eightfold way of STV/PR or 'the super-vote', as Joe Rogaley called it.

STV As A Theory Of The Electoral Test.

The Droop quota broadens our notion of a majority. A majority is not simply one member's majority of, say, 50 votes out of 100. ( If two candidates reached the 50 votes quota, that would be a tie-break. ) Using the Droop quota can make two members' majorities of 50 votes each out of 150. Three quotas, of 50 votes each out of 200, elect three members' majorities.

The single member majority count is generalised by the Droop quota into a multi-member majority count.
The vote generalises in a similar way from a one-preference spot vote to a many-preference vote, in order of choice.

First past the post or the simple majority system and STV are the limited and general 'relative majority' systems. First past the post elects one candidate on a majority relative only to the runner-up. Therefore, most voters may go unrepresented. STV elects as many candidates as constituency seats, each with the same majority, relative to all the runners-up put together. Then, the overwhelming majority is represented.

A majority means 'greater than' and, therefore, must be relative to what it is greater than. Choice is also relative, in that we more or less favor some candidates in relation to others. Indeed, a scale of preference is from greater to lesser choice. So, magnitude of choice is translated, from the vote as preferences, into the count as majorities. The count of the community is the measure of greatness that corresponds to the vote of individuals.

A model from physics shows how the vote and count correspond. Einstein favored Relativity as a 'principle theory', which makes logical deductions from a firm empirical base, that motion is relative to a co-ordinate system.
Similarly, choice is relative to a co-ordinate system of the vote to the count. An empirical order of preference 1, 2, 3, etc corresponds to a rational order of 1, 2, 3, etc member majorities.
This is the view of science as empirical rationalism.

Harré characterised model-building by its gradual adaptation to reality. In this case, the reality is the relativity of choice in general, beyond a basic alternative to a whole range of options. STV has the coherence of a good theory of choice.

The theory of relative choice means in practise that transferable voting experimentally controls the relative influence of candidates on each others election. So, the relatively popular candidates do not take too many votes from those who would be next preferred for election. And relatively less popular candidates' voters are not left unrepresented by their next preferences.

STV might become itself a model for experimental measurement, because its ordinal and ratio scales act as controls of its classifying and interval scales, respectively. Ordinal scale preference voting controls wasted voting on failed candidates. Otherwise, it is to no effect for some that everyone is classified as having a vote. The quota count's ratio scale controls the interval, of redundant votes, beyond the votes that just take successful candidates first past the post.

In democratic, rather than scientific, terms, the single transferable vote is both a preferential and proportional suffrage extension.

This general theory of relative choice explains STV as an electoral system. It also explains why STV applies to all elections. And why other voting systems, being less general in their vote or their count, cannot be used for all electoral purposes.

For example, the Irish have used STV at all levels of government, local, regional, national and European.
And the Northern Ireland Euro-elections approach a crucial test, in which all, but the one method used, would probably fail.
A party list system, such as the Seventies Lib-Lab pact's Reggional List, would have split the Nationalist vote in Ulster, because the SDLP and Sinn Fein would not have submitted a joint party list.
Perhaps, John Hume would have become MEP, anyway, but only because of his exceptional standing as a civil rights veteran, and because he was allowed to be an MEP as well as an MP. Nevertheless, Nationalist voters' cross-party preferences ensured a quota for one of their candidates.

In the 1996 Peace forum elections additional to 18 constituencies of 90 members, 2 people were selected from each of the top 10 parties. Judging by this highly contrived arrangement, Ulster/Northern Ireland could do with a provincial STV quota, as explained ( above ) for Scotland.
( Note: Subsequent elections remedied this abberration with a straight-forward system of STV in six-member constituencies. )

Traditionalists, like Maude and Szemerey, say a proposed electoral reform has to be in every way better than the simple majority system. STV could hardly be otherwise, because it is a consistent generalisation, by vote and count, from that most limited system of choice. STV can do everything it can do, and an immense deal more, as shown of STV's explanatory power, for all kinds of electoral requirements.

Deductive Explanation Of Elections.

To top.

Transferable voting puts a principle theory of relative choice to the electoral test as a scientific law. Scientific laws are conditional statements and the conditional principle STV can be said to formulate, is 'unanimity only in liberty'. Across and within party divisions, the degree and kind of national unity, that transferable voting makes possible, is a condition of individual liberty. Or, proportional counting is a condition of preference voting.

We can call this 'the Hare-Andrae law' of scientific elections ( after the Danish Liberal Carl Andrae and English barrister Thomas Hare, who independently discovered it ). 'Proportional Representation' is a condition of 'Personal Representation', as Mill recognised when he gave both names to Hare's system.

Here is another perspective on why party list systems are unscientific. A scientific law is a conditional statement but the pseudo-law of proportional partisanship is not conditional upon an order of individual choice, so it can only express divisions, not unity in diversity.

Conditional statements are part of the deductive model of science, by which explanations can be deduced, in this case, of democracy. The French republican slogan illustrates a deductive explanation of democracy. A law of fraternity or universal brotherhood is given by the individual human condition of liberty, thru representative equality. Thus equality is conclusively explained by its purpose of serving fraternity in liberty.

Scientific elections exemplify the democratic theory in practise. The deductive explanation of elections is that the universality of the proportional count of a community is conditional upon individual liberty of preference voting, implying an egalitarian transfer of the vote in the count. Hence society in its prefered groupings is proportionally represented.

A party-privileged proportional count is unscientific because it destroys this universal principle of true proportional representation, as distinct from proportional partisanship.

For example, the British General Medical Council using STV, gave PR to its social and functional groups. Minorities, such as immigrants, women and specialists were proportionally represented.
A party list-system could not have done this. Having to make 'parties' of all conceivable groupings would have been an arbitrary business and begged the question of the election, in effect, rigging it. ( Party list systems are party-rigged elections. )
Health service voters couldn't order their choices with an X-vote, only vote for a list representing one attribute, say their speciality. Whereas, three X-votes for three attributes, such as immigrant woman specialist, would have counted against each other.

With a transferable vote, the most prefered individuals are elected. Taking the trouble to honor individual liberty dignifies the voters with the power to prefer the candidates in order of their having most of the attributes you wish them to have.

Transferable voting is integrative. Party lists are disintegrative, because, to ensure the single most important thing to you is represented, you have to form a breakaway list to party-proportionally count.
Proportionally partisan X-voting is conducive to a miscellaneous collection of monomania's demanding their coalition terms to tolerate a government.

From Maiorocracy To Democracy.

To top.

John Stuart Mill realised that proportional representation is a continuation of democracy in extending the suffrage from majorities to minorities. Majority rule that stopped short of democracy proper, he called 'maiorocracy'. Yet, the shortcomings of 'elective dictatorship' by simple majorities are often wrongly held to be a failure of democracy.

( Even academics have not learned from Mill. The Plant report claims Arrow's theorem deduces the limitations of democracy. But it is based on a single member majority system, like the Alternative Vote, which is a limited conception of democracy.)

The fragmentation of parts of the world into warring factions is not helped by the fallacy that bare majority rule is 'democracy' and as such the limit to any obligations that a majority has to minorities.
A clear consequence is that minorities seek to escape majority dictatorships by setting up their own separate states, where they can fulfil the dream of being their own majority dictatorships, from which some other ethnic minority may seek escape, if not mercilessly driven out of their homes as refugees or victims of genocide.

Sovereign statehood confers the arbitrary status of absolute majority on some ethnic group, all of which are really minorities relative to humanity as a whole. 'The minorities problem' is ultimately a problem for us all.

Party lists, whether or not combined with majority systems, give minority parties a share of seats. But they impose their own kind of absolutism of party divisions, when people have to vote for 'a party'. The electoral recognition of relative choice, thru transferable voting, makes unity in liberty a practical possibility, for those who desire it.

Plainly, Mill's mature conception of democracy has much to teach the world for peaceful co-existence.
Northern Ireland is a familiar example. Michael Collins was given Churchill's assurance on PR for Northern Ireland. The Irish Free State wrote the single transferable vote into their Constitution. Two referenda failed to remove it. So, the Irish well understand and appreciate their voting system.
This is confirmed by Electoral Reform Society observers.

Before 1929, transferable voting had almost whittled away the Unionists' overall majority. This wouldn't have happened but for the Republican border areas included in Ulster. The simple majority system ensured a one-party state that didn't have to compromise. So, the peaceful civil rights marches, of Catholics in the sixties, found loyalists unready to tolerate them.

British governments, from William Whitelaw onward, appreciated the problem and tried to bring in power-sharing, or that unity in diversity the province had looked to be beginning to move towards.

A fatal obstacle, of course, was that no patient ever took the medicine his own physician dreaded and despised. The point is that until Mill's mature conception of democracy is generally recognised and becomes the accepted convention, there will always be those nominal democrats hoping to get all their own way from maiorocracy.

Naturally, people cannot come together against their will. But give the people the system that lets them unite as well as divide. As one world gradually becomes more of a reality, it is obvious that a world government would never work without power-sharing, as you could never leave-out any of the great continental constituencies or cultural blocs for even one term of office.

Richard Lung.

To top.

Scientific method of elections: How NOT To Do It.

To home page.