The article was first published by the Greek Sunday paper, Kathimerini, on March 24, 2012. For the article in Greek see:
I have been watching from a distance with incredulity the debates about the Greek educational reform. Now I see that the issue has been transposed to the highest court of the land (Symvoulio Epikrateias) to assess the constitutionality of the law. The arguments are (among others) that the STV electoral system used is a marginal and extremely complicated system fit “only for countries with high party discipline like the absolute two-party system of Malta”. These statements are incorrect.
Let me first of all explain the logic of this system in a simple example, and then I will discuss where it is used and why. There was recently a referendum in the UK where STV was proposed by the liberal party that was defeated, so the UK will continue using the current plurality (first past the post) electoral system. According to the current system, the country is divided into 646 single member constituencies (since the 2001 election) and in each constituency the candidate with most votes wins the seat. The result of this electoral system is a very strong two party system (with very few exceptions – the current government being one of them – there has always been a single party government in the UK). The reason is that voters are not likely to “waste their vote” for a third party (in this case the Liberals).
For this reason the Liberals have long wanted to alter the electoral system, and were able to hold a referendum on the electoral system as a condition to participate in government. Here is how the system would have worked. The number of constituencies would have remained the same, and each voter would have one vote, but would rank his preferences among the candidates. So, a Conservative voter would likely write 1 next to the name of the Conservative, 2 next to the Liberal, and 3 next to the Labour candidate, (because a Conservative is likely to prefer the Liberal candidate over the Labour one). A Labour voter would place the number 1 next to the Labor, 2 next to Liberal, and 3 next to Conservative candidate. A Liberal voter would place the number 1 next to his own candidate, and depending whether he was leaning to the right or the left, would place the number 2 next to the Conservative or Labour (and conversely the number 3). In order to be elected in this electoral system a plurality of votes in not sufficient as in the current system. For single member districts a majority is required (generally, it is one over the number of seats plus 1, so for single member districts in becomes 1/(1+1)). If one of the candidates has a majority of votes, he is declared the winner. If no candidate has a majority, the candidate with least votes is eliminated, and the second preference of his voters is taken into account. This process continues until one candidate has a majority.
Let me use two examples to show what happens in such a system: Suppose that the first preferences produce the following percentages: Conservative 40, Liberal 25, Labour 35. Then the Liberal candidate is eliminated, but his voters are the ones that decide whether the seat will be won by the Conservative or the Labour candidate (if more than 40% of the originally Liberal vote (i.e., 10% of the voters in the district) reports the Conservative as the second choice, the Conservative wins the seat; otherwise Labour does). Suppose now that the first votes were distributed so that the Liberal candidate would have the first or second number of votes: say Conservative 40, Liberal 35, and Labour 25. Then the Labor candidate would be eliminated and his second votes would transfer to the Liberal (remember that Labour voters prefer Liberals over Conservatives) and the Liberal would win the seat. Here is the important result of the STV system: it makes the center party voters either pick the winner or elect their own representative. This is why the Liberals wanted the STV system, why the two major parties did not, and why the Liberals lost in the referendum. The STV system produces centrist outcomes in a single dimensional policy space (a left right division of the country). This is why it was proposed by Maurice Duverger to the French Prime Minster Edgar Faure for the 1954 elections (personal communication). This is the political use and potential of STV. But the more interesting story is its use in non primary political contexts.
STV has been used for elections in other contexts besides parliamentary competitions. For example, it has been used for municipal elections in American cities like Cleveland Ohio, Sacramento California, New York City and others between the wars (after WWII it was replaced because it was leading to the election of black or communist representatives). It is used in the election of school boards. It is used by student associations in Harvard, Berkeley, Princeton, UC Davis, Cambridge, and Oxford. It is used by political parties, trade unions, and peak business associations in Australia. It is also used by the Royal Statistical Academy in the UK as well as the Academy awards for best picture in the US.
So, the argument that it is applied only in cases “strong partisans and partisan identification” is patently false. Actually the opposite is true. There is no party organization with so much discipline as to impose the whole list of preferences to the voters. A party may ask the voters to vote only for its own candidates and nobody else. It may or may not succeed in such an enterprise, because it is in the interest of voters to provide a complete list (otherwise their vote may be wasted). It may ask its supporters to rank their votes in a particular order for coalition formation purposes. However, voters may have different preferences over the available candidates than do party leaders. If a voter provides a complete list of his choices, in an environment that is not extremely politicized, chances are that his second or third vote will go for the most well known and the most appropriate individuals to fill the position regardless of their political affiliation.
Also, because the system of counting is time consuming it reduces the possibility of strategic voting (that is, misrepresentation of ordinal preferences in order to produce better results for a voter). According to the Gibbard–Satterthwaite theorem there is no strategy proof deterministic electoral system with more than three alternatives, but it is quite unlikely that voters or even parties will be able to acquire the necessary amount of information to manipulate the system.
Finally, there is one important psychological advantage of the STV system. Every vote is counted, if not as a first preference as second or third and so forth (unless the voter does not provide all his preferences). This provides profound satisfaction to all the participants in an election using the STV system.
The system is used beyond Malta (and Ireland and Australia). More importantly, it is used beyond politics and it has very desirable properties for the election of academic councils: it moves voters beyond party lines, it discourages strategic voting, and it rewards voters with a vote that counts.