The method of Single Transferable Voting (STV) underpins elections in several countries. The advantages claimed for STV are that, firstly, it allows each voter to express his/her preferences over all the candidates and, secondly, it takes account of each voter's range of preferences in determing the electoral outcome. A disquieting feature of STV -and one that has hardly been commented upon -is that the second point is not true: some voters have more than just their first preference taken

more »

... t of; for other voters, it is only their first preference votes which are counted, their remaining preferences being ignored. This creates two classes of voters -termed in this paper as 'further-preference' and 'first-preference only' voters. Applying these concepts to the (STV based) Northern Ireland Assembly elections of 2003, this paper shows that over half of all voters were 'first-preference only' voters. Moreover, the different parties had different endowments of voters from these groups: in particular, the Unionist parties had a disproprtionately larger share of 'further-preference' voters compared to the Nationalist parties. This might go some way to explaining why, even though the vote share of the Democratic Unionist Party was only slightly higher -and the vote share of the Ulster Unionist Party was actually lower -than that of Sinn Féin, both parties had disproportionately more seats in the Assembly. The paper proceeds to argue that, if society is averse to inter-voter inequality, it might prefer a voting method which treated all voters equally -even though it allowed them a more limited expression of preferences over candidates -to the STV method. * I am grateful to John Fitzgerald and George Tridimas for comments though, needless to say, I am solely responsible for any errors.