過去對於台灣黑金問題的研究，多著重於社會制度或歷史結構上的分析，本文則從選民的角度出發，以Barry S. Rundquist等人（1977）所提出的「隱含交易」（implicit trading）理論為基礎，以1996年國大代表的選舉為對象，應用條件式勝算對數模型（Condi-tional Logit），來解構台灣選民投票支持黑金政黨的原因。實證的結果發現，政黨涉及黑金的形象的確不利其吸納選票，但是政黨如果可以提出吸引選民的政見，黑金的負面影響可以獲得大幅改善。換句話說，選民會因為政黨提出他所偏好的政策理念，降低對政黨黑金的苛求，並進一步支持他所認為的黑金政黨。此外本文也發現，選民這種對黑金的刻意忽視，在考量他們本身的政黨認同等其他相關因素之後，愈形明顯。就各別政黨而言，相較於民進黨及新黨，國民黨的政策理念在當年選舉時對其支持度的影響最小，不過對國民黨而言，黑金形象對其選票的殺傷力也遠低於其他政黨。

Previous studies on the corruption have been focused on specific set-tings of the institutions and social structures. This paper, con the con-trary, emphasizes on the individual voters and is aimed to explain why do voters support for the corrupted parties. Based on the “implicit trading theory” proposed by Barry S. Rundquist et al. (1977), Conditional Logit analysis is estimated to understand the Taiwanese voting behavior. The result shows that the party’s corruption imagines has a negative effect to draw votes. While this negative effect can also be minimized if the party proposes the attractive policy platforms. That is to say, voters will com-promise to the corrupted parties for their policy preferences. As for the specific effect to each party, it is the KMT that the corruption imagines has the minimum influence, comparing to DPP and NP.