Proportional Representation: A Study in Methods of Election

by John H. Humphreys

Considerable discussion has taken place as to which of the list systems yield the most accurate results. It is obvious that as electors do not divide themselves into groups which are exactly one-fourth, one-fifth, or one-sixth of the whole, the utmost that a system of proportional representation can do in the allotment of seats is to approximate as closely as possible to the proportions in which the electors are divided. There is very little difference in the results obtained by the various list systems and by the single transferable vote. The Belgian (d'Hondt) rule slightly favours the larger party; this rule allots seats to parties according to the number of times the party total contains the common divisor, the votes remaining over being ignored. For this reason other advocates of list systems prefer the simple rule-of-three or méthode rationelle.[7] With this system the total number of votes polled is divided by the number of seats. The totals gained by the respective lists are then divided by the quotient thus obtained and the seats allotted to the lists accordingly. If after the allotment of seats to the different lists there remain some seats not allotted, these are awarded to the lists with the largest numbers of votes not utilized. The transferable vote in practice, if not in theory, also awards seats to the various parties according to the number of times the party total contains the quota. If there is a seat not allotted it does not necessarily fall to the party having the largest number of votes not utilized. All the votes not utilized are taken into consideration, and the smaller remainders may, by combination, win the odd seat. For example, suppose that in a six-member constituency five seats have been allotted and three candidates remain in competition for the last seat with votes as follows:—

Candidate A 4,000
" B 3,000
" C 2,000

Then if the supporters of candidate C prefer B to A and have indicated this fact on the ballot papers, the votes given to C would be transferred to B, who would be elected to fill the last seat. With the d'Hondt rule remainders are ignored; with the "rational method" the largest remainders are favoured; with the single transferable vote the last seat is awarded to the majority of the electors not otherwise represented. The transferable vote therefore gives a result at least as accurate as any of the rules devised in connexion with the list systems. But in the majority of cases all three rules will yield the same result.

Panachage.

In the previous chapter reference has been made to the possible abuse of panachage. In order to prevent such practice the Belgian system provides that the elector shall vote for a member on one list only. In Switzerland the elector is permitted to vote for members of more than one list, and any abuse of this privilege is prevented by allowing the elector to cumulate as many as three votes upon any of his favourite candidates. This provision assures the return of the favourite candidates of each party. The problem hardly arises with the single transferable vote; the favourites of each party will doubtless always receive more votes than are sufficient to ensure their election. The elector who desires to advance the interests of his own party as much as possible must indicate his preferences among all the members of his own party before recording any preference for a candidate of another.

Applicability to non-political elections.

The single transferable vote possesses another advantage over list systems. It is not only applicable to political elections, but to all elections in which it is desired that the elected body should be representative in character, but in which party lists are undesirable. The British Medical Association has decided to conduct all its elections so far as possible by the transferable vote; Trades Unions have made use of it in the election of their committees; it has been used in Australia by the Labour party for the selection of parliamentary candidates by members of the party before the date of election. Thus the single transferable vote would produce a much to be desired uniformity in method in different elections.

Bye-elections.

The list systems have an advantage over the transferable vote in the simplicity of their solution of the problem of bye-elections. Under list systems bye-elections are abolished. But the preliminary question, whether it is desirable that they should be abolished, needs consideration. The Report of the Royal Commission on Electoral Systems says: "Neither the single transferable vote nor list systems provide for a solution of the problem of bye-elections which is both fitted to English ideas and practically satisfactory." The Report continues: "Bye-elections are generally regarded as valuable, if rough, tests of public approval or disapproval of the proceedings of the Government, and useful indications of the trend of political feeling. A system, therefore, which would abolish or seriously hamper them is bound to excite opposition."[8] If bye-elections are desirable because of the indications which they give of the trend of political feeling, then the large constituencies which the proportional system demands would add to their value. The opinion of a larger number of electors would be obtained.

Wherever the single transferable vote has been adopted bye-elections have been retained. In Tasmania, whenever a vacancy occurs the whole constituency is polled; the Transvaal Municipal Act allows single vacancies to remain unfilled, but provides for bye-elections when two or more seats become vacant. The Proportional Representation Society, in view of the demand for the retention of bye-elections, suggests that single vacancies should be immediately filled by a bye-election when they occur in a three-membered constituency, but that in larger areas no bye-election should be held until two seats are vacated. But is not the importance of bye-elections overrated? In many respects they are the least satisfactory feature of English elections, and it is noticeable that the change of opinion registered in a bye-election has often not been maintained when the same constituency is polled at a General Election. A considerable proportion of bye-elections are consequent upon the taking of office by members of Parliament, and it is generally agreed that such bye-elections are not necessary. Further, the House of Commons has already resolved that it is desirable to reduce the length of parliaments to five years, which in practice would mean a working life of four years. The shortening of parliaments would destroy what little value bye-elections possess.