ABSTRACT
What influences the timing of vote choices in mixed-member systems and to what extent does this influence split-ticketing voting? Although voters cast both ballots in effect simultaneously, they may make their decisions on which candidate and party to support sequentially. Using Japan’s 2012 election as a case study, empirical analyses find that the district nominations influence the timing of one’s district vote intention. Meanwhile the timing of vote choices corresponds with ticket-splitting, even after controlling for partisan and socio-economic factors.
Notes
1. Under the less common one-vote mixed-member system, voters cast a district ballot and these votes are aggregated to the regional or national level to determine party list seats. See Ferrara (Citation2006).
2. A mail survey sent the day prior to the 16 December election was collected through the end of January. Data used in this article can be acquired at: http://www.masaki.j.u-tokyo.ac.jp/utas/utasv.html .
3. McKean and Scheiner (Citation2000) contend that dual listing, by incentivising party list candidates to build local roots for district competition reduces the number of viable district entrants, although Rich and Banerjee (Citation2016) find that districts with more than two dual-listed candidates corresponded with a higher number of effective parties.
4. See Q010300 (for party list) and Q010500 (district vote) in the Asahi survey codebook.
5. Party identification measures here and later in the paper come from Q013700, which asks about long-term party identification.
6. The parties were (the Liberal Democratic Party (LDP), Democratic Party of Japan (DPJ), the Japan Future Party (JFP), Komeito, Japan Restoration Party (JRP), Japanese Communist Party (JCP), Your Party (YP), and the Social Democratic Party (SDP).
7. This variable was created by matching the prefecture and district data in the survey with Asahi election coverage on the election results and the number of candidates in each district.
8. Age was measured in six cohort ranges (20s, 30s, 40s, 50s, 60s, and 70s and older).
9. We recognise the limits in interpretation of these ordinal measures as if they were integers as the distance between measurements are not fixed.
10. Unless otherwise stated, all predicted probabilities hold the additional independent variables at their means.
Additional information
Notes on contributors
Timothy S. Rich
Timothy S. Rich is an Associate Professor of Political Science at Western Kentucky University. His research focuses on the electoral politics of East Asia.
Vasabjit Banerjee
Vasabjit Banerjee is an Assitant Professor of Political Science at Mississippi State University and Research Associate in Sociology at the University of Pretoria. His research focuses on contentious politics, electoral competition, and foreign policies of developing societies.