Michael Laver, "Analysing Structures of Party Preference in Electronic Voting Data," Party Politics, 10 (September, 2004), 521-541.

First Paragraph:
In the 2002 Irish general election, 138,011 voters in three of the 42 constituencies cast their votes electronically, as part of a trial that will lead to electronic voting throughout Ireland by the next general election. Ireland uses the single transferable vote (STV) electoral system, each vote registered requiring the voter to rank the candidates contesting the constituency.1 Voters may rank any number of candidates, from just one to every candidate in the race. The publication of the full set of anonymized electronic ballots has generated very rich data on the candidate rankings of a large number of voters in a real general election. This article presents a preliminary analysis of these data. The intention is both to reveal systematically for the first time a range of features of Irish voting behaviour and to draw more general inferences about the structure of voter preferences.

Figures and Tables:
Table 1. Maximum number of preferences expressed by voters
Table 2. Percentages of voters registering N rankings
Table 3. Voting for multi-candidate parties
Figure 1. Number of preferences registered by voters by first preference party
Table 4. Mean proportions of ëbustí ballots, by first preference party
Table 5. Mean preference rankings of candidate
Table 6. Rank order correlations (Spearman's rho) between candidate rankings

Last Paragraph:
This article has merely scratched the surface of a rich new data set. For those interested in the general structure of party preferences, much more detailed analysis may reveal the extent to which voters have detailed preferences that extend across all of the options on offer, or in contrast use more general categorizations of options into what they do, and do not, like. Furthermore, the low levels of consistent party as opposed to candidate voting revealed in these data invite additional comparative research. For those interested in estimating spatial representations of party competition, there are many possible ways to analyse estimates of similarities and dissimilarities between parties that the electronic voting data can generate. For those interested in electoral systems, it will be interesting to explore how STV generates broad proportional representation of parties, despite the relatively low levels of party voting demonstrated here. And for those who are indeed interested in the red meat of Irish party politics, electronic voting data are clearly going to provide an entirely new intellectual armoury.