近年來制度至關重要的論點，一直是學界關注各種政治行為形塑的重要變項，許多選舉經驗研究指出，包括政黨提名候選人策略與政黨間合作協調、國會議員立法行為與政策主張、候選人選區經營與競選活動、選民投票決策、甚或是政黨體系的變動等，均深受不同選制結構的影響。準此，探索政治場域上相關行為者的行為或動機時，不同選制架構內涵的影響效果，深值剖析。本文主軸鎖定於日本自1994年推行四大政改法案後，改採小選舉區比例代表並立制架構下之候選人重複提名制度，對於日本政壇主要政黨（包括自民黨、民主黨、社民黨）提名參選策略與其實質成效的影響。有別既有文獻分析兩票連動效果經常未顧及變項間內因性與非隨機選組機制因素，導致參數值估算偏誤的因果推論問題，本文爰引James J. Heckman 建構之效應模型（treatment-effects model）進行準確校估。實證結果顯示：日本各主要政黨在歷屆眾議員選舉對單一選區候選人採重複提名參選的策略，其對政黨比例代表得票率的效應會因黨、因年而異；當中包括自民黨在2009年選舉、民主黨在2003年選舉，社民黨在2000年、2005年、2009年選舉，均以重複提名策略有效提升該黨的政黨比例代表得票率。此一分析結果，不僅適度佐證既有文獻對於日本政黨重複提名參選成效的預期，也吻合混合選制學理兩票連動效果的論點。

In recent years, scholars have paid a great deal of attention to the influence of institutions on political behavior. The claim that “institutions matter” in politics is widely accepted. Empirical studies show that electoral systems have a great impact on parties, candidates, and voters with respect to nomination, coordination strategies, legislative behavior, voting behavior, electoral campaigns, and the development of party systems. Accordingly, when we try to understand the behavior and motivations of political actors, it is necessary to examine institutional rules.After the electoral reforms of 1994, the single nontransferable vote (SNTV) was replaced with a mixed-member majoritarian (MMM) system. This paper seeks to evaluate the impact of this new electoral system based on district-level data. We are especially interested in the interaction effects (or the so-called “contamination effects”) on the two-vote structure resulting from dual candidacy strategies adopted by political parties. To avoid causal biased estimation and capture the contamination effect more precisely, we apply Heckman’s treatment-effects model to remove the endogeneity and self-selection problems. The result not only shows the dual candidacy strategies influence the interaction between two ballots, it also indicates the effect varies between different parties and different periods of time. More specifically, the dual candidacy strategy was especially beneficial to the Liberal Democratic Party’s (LDP) PR vote in 2009, the Democratic Party of Japan’s PR vote in 2003, and the Social Democratic Party’s (SDP) PR vote in 2000, 2005, and 2009. In conclusion, this study supports the existing literature on dual candidacy and contamination effects in mixed-member electoral systems.