有關選舉無效票（即廢票）的研究很少，主要原因是無效票比率通常不高，對選舉結果的影響不大；而且無效票一般被視為是「疏失」造成，並非選民「意志」的表現，因此很少人探討其發生的原因。作者考察我國 1992至 2008年， 37個不同類型的選舉，結果發現無效票率的空間分布並非隨機，且有一定的模式。不同類型的選舉，無效票率的高低有跡可尋，代表無效票率其實是可以被解釋的，其背後有社會與制度因素存在。經由無效票率的追蹤分析 (panel data analysis)後，發現選舉法規、選舉複雜程度、選舉受重視程度、選舉年度及經社背景變數（教育年數、老年人口、人口密度比及原住民人口比例），約可解釋 34%之無效票率變異。作者進一步以空間迴歸模型探討，發現無效票率存在正向之空間自相關情形，部分地區則有明顯空間異質現象，極有可能是該地區民眾長期對政治冷漠使然，如果我們要提昇民主的內涵，則這些無效票率特別高的地方值得關注。

The issue of invalid votes receives few scholarly attentions. Many consider invalid votes as the result of unintended behavior and do not explore the factors that may influence the incidence of invalid votes. We examine 37 elections of different levels between 1992 and 2008 in Taiwan by using panel data analysis. We find that the distribution of invalid votes are not random but are influenced by institutional and socioeconomic factors. Regarding institutional factors, the elections codes, the complexity of elections, the importance of elections, and years after the democratic transition are the main influencing factors. Years of education, the percentage of elders, population density, and percentage of indigenous citizens are the main socioeconomic variables that affect the incidence of invalid votes. The two strings of factors can explain about 34% of the variance in invalid votes. In addition to the pooled OLS model, we also employ spatial lag model and spatial error model. The two models show that the distribution of invalid votes exhibit positive spatial autocorrelation. In addition, some areas also exhibit spatial heterogeneity, which is likely to be attributed to the alienation of voters in the districts. We might need to pay special attention to these areas to enhance the quality of democracy.