Split-Ticket Voting: Art or Science?

Evidence suggests a gap exists between expectations and election outcomes, with those backing winning candidates more likely to say they understood the system
Photo: AP / Kin Cheung
Timothy Rich

In Taiwan, as in most mixed member legislative systems, voters have separate ballots for a district candidate and the party list, thus allowing for split-ticket voting. But what influences splitting votes across parties?

According to Duverger’s Law, we should expect some voters to engage in strategic voting in district elections as there can only be one winner. If one prefers a candidate unlikely to win, Duverger expects voters to opt to vote for another candidate to prevent their least preferred outcome. As a result, district elections should be dominated by two candidates with broad support. Despite Taiwan only using a mixed system for three elections, district competition largely confirms to Duverger: on average over 90 percent of the district vote since 2008 went to the winner or first runner up (Kinmen County remains a consistent outlier). In contrast, party list seats can be obtained with only a sliver of the vote — at least 5 percent to be exact — presenting little reason to consider defecting to another party. Thus, supporters of smaller parties are more likely to split-ticket vote.

Recently released post-election survey data from National Chengchi University’s Taiwan’s Election and Democratization Study (TEDS) provide insight into ticket-splitting from January’s election. Seeing that the Democratic Progressive Party (DPP) and Chinese Nationalist Party (KMT) ran viable candidates in the majority of districts, it is not surprising than only around a third of these partisans split their votes. Meanwhile supermajorities of those identifying with the People First Party (PFP), New Power Party (NPP), New Party (NP) and Taiwan Solidarity Union (TSU) split their vote. Admittedly part of this is not motivated by strategic voting but by the lack of a district candidate from one’s preferred party. For example, the NP and TSU only ran district candidates in two of the seventy-three districts, whereas the PFP and NPP ran 6 and 12 respectively.

Screen Shot 2016-04-02 at 11.51.58 AM


However, another aspect influencing ticket-splitting is often ignored: do voters understand the mechanics of the mixed system? For example, a year after Taiwan’s first election under this system nearly 60 percent of respondents claimed to not understand it. Additional evidence suggested a gap between expectations and election outcomes, with those backing winning candidates more likely to say they understood the system.

A clearer measure of understanding is whether voters know that a party must receive at least 5 percent of the party list vote to receive any party list seats. The 2016 TEDS data suggests that knowing the 5 percent threshold corresponds with ticket-splitting. My previous research suggested that supporters of smaller parties were more likely to be aware of the 5 percent electoral threshold, in all likelihood because smaller parties emphasize this point. This 2016 data is consistent with this earlier finding: less than half of DPP or KMT identifiers but over 60 percent of those identifying with the other four parties correctly identified the electoral threshold at 5 percent. More broadly, 38.69 percent of those who did not correctly identify the threshold split their votes, compared to 49.38 percent who knew the threshold.


Timothy S. Rich is an assistant professor in political science at Western Kentucky University. His main research focuses on the impact of electoral reforms in Taiwan, South Korea, and Japan compared to similar legislative systems. His broader research interests include electoral politics, domestic and international politics of East Asia, and qualitative and quantitative methods.

Comments are welcome, but will be moderated. Remarks containing abusive language, personal attacks or self-promotion will not be published. We encourage healthy discussion and, above all, tolerance of other's views.