期刊內容 Issue content

量化研究的比較問題邏輯:因果異質性與縣市長選舉投票模型的建立
The Logic of Quantitative Research in Comparative Inquiries: On Casual Heterogeneity and Model Building of Voting Behaviors in Magisterial and Mayoral Elections in Taiwan
黃信豪(Hsin-Hao Huang)
46卷3期(2007/09/01)

在調查資料的累積下,研究者可運用計量方法來進行以「比較問題」為目的之實證研究。在問題取向上除了針對特定主題進行跨年度的比較外,學者也關注不同國家或區域的比較研究,也可依理論設定,對特定標的從事比較問題設定。

本文的目的在於從方法論出發,探討從事計量研究時所應關切的比較問題邏輯。首先,作者說明比較研究的問題本質,並提出比較單元與分析單位異同時的不同研究架構;接著進一步討論在比較的架構下,如何運用適當計量模型設定方式,方能對於比較問題提出直接的因果關係驗證,也說明不適當之方法所可能產生的問題。在應用上,作者採「台灣選舉與民主化調查: 2005 年縣市長選舉」之實證資料,以縣市發展程度為區隔,針對台灣民眾的投票行為進行地區效應之比較研究。實證分析結果顯示,在適當計量方法的運用下,我們方可進一步探索各比較單元內可能存在的因果異質性。在理論意涵上,筆者也討論選民所處的縣市發展程度差異如何影響其投票行為與思考模式。

 

In recent years, it is possible that researchers apply quantitative method to comparative inquiries by the accumulation of cross-sectional survey data. Accordingly, some political scientists start to explore and study comparative inquiries tested by pooled cross sectional data, sorted by the units such as time, countries and some specific subjects guided through theoretical postulates.

This paper is concerned about the logic and application of quantitative research methodologically, especially in comparative inquiries. First of all, the author addresses the essence of comparative inquiries, and advances herewith a corresponding research framework. Secondly, this paper argues that the proper use of model specification may contribute to clarify sophisticated causal relationships, and demonstrates the results of models incorrectly specified as well. Sequentially, by analyzing survey data, the author identifies the association between regional contexts and voting behaviors by comparing different levels of economic development among various counties in Taiwan.

The empirical result shows that the causal heterogeneity of comparative

units would be examined only in the compatible model specification. Moreover, the author further interprets how voting behavior is influenced by various developmental levels of counties where the voters dwell, and provides a theoretical linkage.

 

top