Chia-Hung Tsai, "Policy-making, Local Factions and Candidate Coordination in Single Non-Transferable Voting: A Case Study of Taiwan," Party Politics, 11 (January, 2005), 59-77.

First Paragraph:
Elections are widely perceived as zero-sum games, with electoral rules dictating the method of seat allocation, such as 'proportional representation' (PR), 'single-member simple-plurality' (SMSP) and 'single nontransferable voting' (SNTV). As players in the game, political parties follow electoral rules and choose the best strategy for their share of seats. Electoral rules remain the same unless the parties agree to amend them, and this process may require a great deal of time and effort if consensus is to be reached among the competing parties. Every election is a new game in which parties and candidates are faced with new competitors, issues and voters. The uncertainty of competitors makes candidate coordination crucial to electoral victory, because nomination errors - nominating too many candidates or putting quality candidates in less hopeful districts - will cost party seats. With unchanged electoral rules, however, parties can devise the strategy which will achieve the best outcome.

Figures and Tables:
Figure 1. Percentages of seats of political parties, 1992-2001
Table 1. Seats and candidates, 1989-2001 legislative elections
Table 2. Votes, seats and candidates for the KMT and the DPP, 1989-2001 legislative elections
Table 3. KMT chairpersons and their rankings in the previous election by year
Table 4. Logistic estimate of chairmanship on the probability of being re-elected
Table 5. Determinants of the number of seats won by the KMT

First paragraph in the conclusion:
The main objective of this article was to explain how political parties, through legislative politics, coordinate candidates in multi-member districts. Light is shed on the success of the KMT and DPP's nominations, and how the KMT gained advantage by using local factions and chairmanship is explained. Chairmanship is one of the particularistic interests that political parties can distribute among party members to maximize seat gains. I also found that chairpersons are more likely to be re-elected than non-chair incumbents. Moreover, the KMT has done an excellent job in equalizing chairmanship, with the exception of the period between 1998 and 2001. Lastly, this study confirms that the KMT's seat share directly increases with the number of chairmanships and local factions. For unexplained reasons, the KMT failed to distribute chairmanship effectively in 2001 and consequently suffered great seat losses in the 2001 election.