1897 Tasmania used STV for election of members of state chamber for cities of Hobart and Launceston.
First use of STV (with both transfers from eliminated and elected candidates) in the world, and it was the first use of STV of any kind in a government election in the British Empire (British Commonwealth).
Partial PR - STV used to elect ten members while 28 elected through FPTP elsewhere.
Hobart elected six members; Launceston elected four, in city-wide districts.
The form of STV was a variant called Hare-Spence. It used the Hare quota, surplus transfers were done in whole votes by a random method; voter had to mark as many preferences as half the number of seats in the district.
the system used a random whole-vote method to transfer surplus votes. The electon rules tell me that the method used was to couint down the candidate's stack until quota was arrived, then to transfers all subsequent votes as surplus votes. The surplus votes thus chosen might be gropedl by a particluar preceinct or polling place and might have been predominatntly for one of the other remaining candidates or for another.
(Even where the next usable marked prefence on each ovet is mathematically reduced to thits due porton ofhe suplus, random effects might creep in - when the votew are trasferred subsequently -. the lower prefences that are piggybacked blindly with the next usable preference might be skewed.
However an analysis of the two STV contests published shortly after the election say there was no fear of that in Launceston and very little fear of it in Hobart.
This writer (R.M. Johnston) divides up the surplus transfers into two types - the surplus votes of those who are elected in the first count, and those that occur later after the successful candidate have received incoming transfers.
(The difference is about the depth of attraction of the voter to the candidate and whether or not it can be measured accurately - observers know that the "transfers of the first order" (those of first-count winners) are in line with second preferences (or prefernces very close the secondary ones); while "transfers of the second order" - surplus votes of candidates elected after incoming transfers - may be set by preferences lower than second preferences but it is not clear if that is the case and if so, how much lower are the relevant preferences.)
"The feebleness of the influence [of transfers of surplus votes] in altering the final determination of the all-powerful influences, viz., the first count and the votes (two or higher preference) of the previously excluded candidates, is seen by the results. For, although the lowest candidate (Sutton, 283) before distribution was only nine votes behind the next lowest in order (Fowler, 292), yet the inclusion of quota-excess distribution of the second order, although differing in force only by six votes, to the advantage of the lowest, the same order of importance was undisturbed, Fowler still keeping the lead by three votes! [Even if transfers go to a less-popular candidate in larger numbers than to a higher-ranked candidate, it is rare for the difference to allow a lower-ranked candidate to pass a higher-ranked candidate. for a change in the ordering of the candidates to happen, the difference between the rate of transfers must be sufficient to close the gap between the candidates but meanwhile the higher-placed candidate is in most cases not standing still - his or her vote tally is increasing as well, just not as quickly.]
This inevitable result is beyond any shade of dispute, as, in the distribution of the quota-surplus 16, there entered no element of chance selection. Such an element could only enter where there was a possibility of a portion of this 16 being afterwards redistributed. [this is in refernce ot the piggybacked lower prefernces mentioned above]
In Launceston such a possibility could not occur. It is proved, therefore, that in the Launceston election, the possible influence of the element of chance was positively nil.
In Hobart the final results, although affected by four quota excesses (one of the first order and three of the second), were, even in the aggregate, too feeble to exercise any disturbing influence upon the true relative positions which, as in Launceston, were altogether dominated and determined by votes of first counts and by next in order preferences of lowest excluded candidates.
The total force of the transfer votes of quota-excesses of the first and second order in Hobart only amounted to 3.54 percent of all effective votes, as shown by the following analysis:
Analysis.
Number of votes Percent of total effective
transferred votes
Quota-excess Votes of First Order (1) Fysh 44 1.25 percent
Quota-excess Votes of Second order
(3) Bradley, Clark, Mulcahy 81 2.29 percent
Quota-excess votes of Both types (4) ... 125 3.54 percent
All other effective votes 3411 96.46 percent
Total effective votes 3536 100.00 percent
Difference between lowest candidate elected and
the highest of the candidates excluded 129 3.39 percent
Highest number of votes originally transferred to any one candidate of the quota-excess of the first order (Fysh 44) which in case of re-transfer still involves an infinitesimal element of chance 27 0.76 percent
Actual number of votes re-transferred by quotasurpluses of second order and by transferred votes of lowest excluded candidates in which any element of chance selection was involved... 39 1.10 percent
Average number of such votes for each candidate 3.25 0.09 percent.
The above analysis is interesting and instructive. It shows that among the 3536 total effective direct and next in order of preference votes, only 125, or 3.54 percent were derived from all quota-excesses;
that of these, only 39 were redistributed in which any element of chance entered under the method provided by Mr. Clark, Clause 115, Sect. VI., for the determination of the proportion by which the 39 papers were actually distributed ;
and that this, in the aggregate, only represents 1.10 percent of all effective votes, or a mean of 325 votes per candidate.
As the total redistributed quota-excess votes of the first order (39) only represent 3.25 percent of the final difference between the lowest candidate elected and the next in order — the highest candidate who was last excluded from the poll — it is clearly demonstrated that the remaining element of chance selection in practice is infinitesimal in its influence, and did not in the slightest degree affect the relative order of candidates as mainly determined by the combined influence of (No. 1 preference) votes of the first count, and Nos. 2 and 3 preferences of transfer votes of the lowest excluded candidates.
These latter together (3411) represent, as already shown, 98.46 of the total effective voting force; and this fact alone should show that too much importance, by far, has been commonly attached to all rival modes for dealing with the distribution of quota-surpluses and their possible but small element of chance.
The reduction of the original small element of chance from 1.25 percent of all effective to O.09 percent for each candidate should surely satisfy anyone that the ideal elimination of elements of chance, so far as the true order of final results are concerned, have been practically and successfully achieved by the Clark-Hare method introduced at the last general election in Hobart and Launceston.
...
[Hare's concept had the surplus votes trasferred just what came next to hand.]
If the excess papers were taken, however indiscriminately, from either top, bottom, or middle of the whole parcel of first counts, it is almost certain that the second and higher preferences would vary with each chance selection, and the voters whose papers were selected for transfer to next in order of preference would thus by mere chance have an undue advantage in the determination of the candidates next in order of choice. This is accomplished by redistributing the whole of the successful candidates' voting papers among the candidates not yet excluded from the poll on the basis of the next in order of preference i.e., No. 2 — and afterwards allotting to each candidate such a proportion of papers, so distributed, to each candidate as is equivalent to the proportion which the quola-excess bears to the total parcel of first counts of the successful candidate. [this is the mathematical reduction method also used in most of the old-time Canadian STV elections. The math reduciton method is explained on page 78 of Johnston's essay.]
This is a just distribution and entirely removes the element of chance, so far as the second preference is concerned.
A similar provision is made for removing, or rather minimising, the very trifling element of chance in quota-excesses of the second order i.e., where a former transfer paper may again be transferred to the third or next in order of preference —the determinants in the latter case being the whole of the transferred papers, only, which may have helped to complete a candidate's quota. The process is extremely simple and effective. The only objection to the method is that it may add about 20 percent to the work of handling the papers, as in the Hobart election.
[it is not clear to me whether "the whole of the transferred papers" means all the marked preferences such as used in Gregory method or if transfers are dictated just by the next marked preference.]
Does the Clark-Hare method entirely eliminate the element of chance in the transfer of quota-excesses ?
Answer. —Yes, entirely, as regards quota-excesses of the first order.
As regards transfers of the second order, I estimate that the element of chance for each candidate only represents 0.09 percent of all effective votes. This is so trifling an influence that it may be safelv ignored in practice.
(R.M. Johnston, F.L.S., OBSERVATIONS ON THE WORKING RESULTS OF THE HARE SYSTEM OF ELECTION IN TASMANIA (Papers and Proceedings of the Royal Society of Tasmania), p. 74--78)
https://eprints.utas.edu.au/16182/1/johnston-observations-hare-system-election-1897.pdf
[I have always heard that the whole-vote method of transfer of surplus votes was at least somewhat random so Johnston's statment is interesting - it may be that Johnston is missing some random-ness. more study needed...]
-------------------
Hare-Spence used again in the 1900 election, then Tasmania went back to using FPTP in single-member districts. (Until 1907, STV was not used anywhere in world for government election. But from 1907 unitl the present, there has not been any time that members elected hrough STV have not been in office somewhere in the world.)
1907 STV extended to all members in Tasmania. (Farrell and McAllister, The Australian Electoral System, p. 27)
More on the 1897 election
Already by 1897 the significance of the First Count (the "primary vote") was being noted.
An article published the day before the election states how the "primary" vote will pretty much dictate who is elected.
recommended the election official announce the First Count results as they are very important and will relieve the anxiety of the crowd and then work at leisure to produce final results.
This point - how most of the winners are already set with the announcement of the primary vote - is emphasized in an essay on the election
OBSERVATIONS ON THE WORKING RESULTS OF THE HARE SYSTEM OF ELECTION IN TASMANIA. BY R. M. Johnston, F.L.S.
This essay states quite clearly that looking just at transfers, the quota and rankings on the votes is overlooking "the keystone of the Hare system" and "the chief merit of the Hare System" - the use of single voting in multi-member districts.
HOBART
6 members to be elected
12 candidates Each voter has to mark three preferences to elect six.
total valid votes: 2745
quota 457
one member elected on First Count
the most-popular six in the First Count are certain to be elected..., says newspaper article on election day.
Komentáře