Single Transferable Voting

The Preference-Based Voting Method known as the Single Transferable Vote

The voting system recommended uses a widely understood version of Preference Voting (‘PV’) known as the ‘Single Transferable Vote’, or ‘STV’. Taken together, the different approaches to preference voting (the mechanism used to give effect to proportional representation) are used by more nations than the plurality voting system. For example, all of the members of the European Parliament are elected this way.

The main benefits of this system are:

1. No vote is wasted. For example if your first preference loses then your first preference vote is transferred to your second preference and so on. If your first preference wins with an excess of votes (over the quota) then the unneeded portion of your vote is transferred to your second preference and so on.
2. Preference voting reduces the need to consider strategic voting – not voting on merit in the hope of getting a better outcome. For example, you can safely give your first preference to a candidate unlikely to win knowing that the vote will not be wasted as explained above.
3. With first-past-the-post voting, on a two candidate preferred basis, the least popular candidate can win. This risk is eliminated in preferential voting.

How Does It Work?

In STV candidates are automatically elected if they achieve a minimum number of votes called the quota. The quota is defined as (total number of formal votes) divided by (the number of candidates to be elected +1) or a minor variation of this. This formula ensures there are never more winners than vacancies.

The process of determining winners and losers is achieved in various counting rounds called counts. The first count finds all candidates who reach or exceed the quota from the number of first preferences received. Commonly, one or more candidates receive more first preference votes than the required quota. If the required number of vacancies is not filled in the first count, the excess of votes of quota is not wasted because the excess fraction of each vote for each winner is transferred to the remaining undecided candidate(s) according to the second preferences of those voters. This results in fractions of votes being transferred to the remaining candidates.

Another count then follows, and the candidate with the smallest number of votes, including transferred votes, is then excluded, and again so that no votes are wasted all votes attached the excluded candidate are transferred in full according to the voters’ preferences. The process then repeats with another count to find the next winner(s) and loser until the required number of vacancies is filled.

What Happens if the Result is a Tie?

Sometimes there is a tie for lowest number of votes in a count when attempting to exclude candidates. There are two common tiebreaker rules that cover this situation:

>> 1. lookback to previous counts, and
>> 2. random draw

Voting systems can be configured to use each method in series to resolve a tie.

A lookback first refers to the previous count and excludes the candidate with the fewest number of votes in that count. If that also results in a tie, then the system keeps looking back to earlier counts until the tie is broken.

If there is no earlier count, or if the lookback method otherwise results in a tie, the common fallback is random selection. In BigPulse, the commercial voting system in use in ICANN’s At-Large community, each poll is assigned a ‘seed’ for the random number generator at the time the poll is opened. Each poll has a different seed.

At the start of each count a unique random number, generated from the seed, is assigned to each candidate. Because the seed is constant, the random numbers don't change for a given count, which is the desired behaviour.

For Further Information

Wikipedia has some very useful information on STV, including examples that walk the reader through how the system works. You may find that information here: http://en.wikipedia.org/wiki/Single_transferable_vote.