跨黨投票的現象在國內外都頗受矚目，然而文獻中對此一重要議題，從概念界定、測量指數，到數據資料、分析方法等，卻眾說紛紜。本文是從方法論的角度，觀照「一致與分裂投票」的基本概念及其使用的個體和集體資料、分析方法等層面，一方面比較各種研究途徑的優缺點；另一方面則提出台灣本土硏究應有的反思。本文認爲：研究者應該把個體和集體數據視爲是同一個現象在不同觀察層次所呈現出來的訊息 (information)，而學界的目標，就是要整合不同形態的訊息，使之交相爲用丶相輔相成，並建立一套能跨越並貫穿微觀丶中觀、宏觀各層次的一致與分裂投票理論架構，使不同分析層次的切入點非但能關照同一現象的不同層面，而且各點均能推導出邏輯一致丶上下連貫的論證，前後呼應。

The practice of endorsing candidates of different parties while marking a ballot, called split-ticket voting, has long attracted the attention of scholars interested in electoral studies. However, despite its prominence, the literature of this subject suffers from a lack of agreement on the appropriate research methodology. This paper first provides a comprehensive review of how straight-and split-ticket voting is conceptualized and practised, and then examines various analytical approaches for individual and aggregated data.  We argue that micro-level surveys and macro-level statistics should be treated as information observed at different levels of the same studied phenomenon, i.e., vote choice. The long-term aim is to construct a unified theory for ticket splitting on both the micro and macro level, that integrates the motives of individual voters and the voting patterns of the electorate as a whole. Such a micro-macro model can only be developed, however, if research­ers think carefully about the underlying mechanism of voting behavior as well as the data generation process of the levels of information at hand.