Strategic Voting in the 1996 Japanese General Election

Japan's new electoral system may or may not be good for Japanese democracy, but it is almost ideally designed for the study of strategic voting. No other existing electoral system provides political scientists with data on such neatly separated but simultaneous single-member districts (SMD) and proportional representation (PR) votes. Using candidate-level aggregate data, I find strong evidence for strategic voting: Candidates with a good chance of winning receive more SMD than PR votes, and those with little chance of winning receive more PR votes. Strategic voting explains not only the basic pattern but also several exceptions to the rule. In addition, I find that a realistic but relatively inaccurate ex ante prediction of a candidate's probability of victory produces a superior explanatory model than does a perfect ex post prediction calculated from the final electoral results. These results support the hypothesis that voters base their strategic voting calculations on simple cues available to them before the election.