Up: Home page Previous: Issue 14 Next: Issue 16

Voting matters - for the technical issues of STV

Issue 15, June 2002

(Complete issue in PDF, 188Kb.)


On the 17th May 2002, the Dáil constituencies of Meath, Dublin North and Dublin West used an experimental system for electronic voting. It is expected that this system will be used exclusively for local and national elections in the Irish Republic in the near future.

Of course, the three constituencies used the same electoral rules as in the other 40 — essentially a hand-counting system which has at least one 'problem' in that the result can depend upon the order in which papers are transferred on a surplus. Examination of such issues has traditionally always been hampered by the lack of complete information of all the preferences expressed by the voters.

We now have a significant step forward for electoral studies since the Irish electronic voting results includes the complete data input to the electronic counting software. One can reasonably expect future issues of Voting matters to analyse this data.

The first paper in this issue is indeed an analysis of Irish election data, but only uses the result sheets. Philip Kestelman shows statistically significant bias according to the alphabetic position (on the ballot paper). I might add that even a casual inspection of the full data mentioned above shows a tendency for the final few preferences to be in strictly ascending or descending order.

In the second paper, Eivind Stensholt considers the problem when additional support for a candidate results in that otherwise elected candidate not being elected. This property of non-monotonicity applies even to the case of electing a single candidate, as shown in this paper. On the other hand, the paper indicates that it is relatively rare.

In the third paper, Markus Schulze considers an algorithm for electing candidates with preference voting proposed by Professor Sir Michael Dummett. Sir Michael has chosen not to respond to the criticisms made.

In the last paper, David Hill and Simon Gazeley produce a new STV-like algorithm which merges the ideas of Condorcet and STV. The advantage of this algorithm is to avoid the property of all conventional STV algorithms of premature exclusion, such as for a universal second-choice candidate. On the other hand, this method has the disadvantage of later preferences could possibly upset earlier ones in rare cases.

McDougall Trust: STV Resources CD

A proof copy was prepared in February, but the publication date has not yet been agreed.

Brian Wichmann

Papers with citations

  1. Philip Kestelman: Positional Voting Bias Revisited [16, 1].
  2. Eivind Stensholt: Nonmonotonicity in AV
  3. M Schulze: On Dummett's 'Quota Borda System'
  4. I.D. Hill and Simon Gazeley: Sequential STV - a new version [22, 1-6, PDF].

Main index      Author index