為什麼我國單一選區的村里長選舉，並未如杜佛傑法則所預測地出現全面性兩大黨競爭，而是大量的無黨籍當選者？政黨標籤對村里長選舉是否毫無幫助？本文認為其關鍵在於選區內選民數多寡，選民少時候選人可靠己身社會網絡及資源交換掌握足夠票數、選民多時則需透過政黨標籤來爭取無法親自接觸的選民並降低資訊成本。換言之，選擇披黨籍參選與否是候選人追求勝選的策略選擇，而選區選民數越多時政黨標籤對勝選幫助越大，本文稱之為政黨標籤的規模效應。由於歷史沿革及後續發展，我國各村里的選民數從數十到上萬，差異極大，故可在控制政府層級的情況下，驗證選民人數與政黨標籤交互作用下對候選人競選策略及得票的影響。作為第一篇全國性的村里長選舉量化分析研究，2014年村里長選舉結果顯示，選區選民數少時較容易出現無黨籍、同額競選、與現任者連任。相較之下，選民數較多的選區，有較多代表兩大黨參選的候選人、且更容易變成兩大黨對決。然而，迴歸模型僅發現民進黨籍在選民數越多時吸引越多票，同樣的狀況卻未出現在國民黨籍候選人上，這可能來自兩大黨招牌背後代表的動員模式造成標籤規模效應的不同。此外，統計結果中亦發現村里長選舉裡中年、男性、現任者之得票優勢。 Contrary to Duverger's Law, most of the winners in Taiwan's village and li Single-Member-District elections are independent. Why? Is partisanship not effective on attracting ballot in this election? This article emphasizes the amount of electorates in districts as the key moderating variable. When personal resource is not enough in exchange of winning because of the large number of electorates, party brand can be a cost-saving heuristic cue on attracting partisan voters; party brand has ＂scale effect＂ in district with more electorates. Due to the special colonial background followed by economic development, the size of village and li in Taiwan varies tremendously-some are small enough that every electorate knows each other, the others are so large that it is impossible for candidates to meet everyone. This special setting enables researchers to test the interaction between partisanship and number of electorates on voteshare, controlling for the level of government and electoral system. Being the first quantitative paper on Taiwan's village and li elections, this article uses 2014 election results to reveal that (1) number of electorates negatively correlates to singlecandidate district and independent incumbent reelection. (2) Number of electorate positively correlates to larger proportion of candidates from two major parties. (3) DPP candidates received more votes when the number of electorates in district is larger, which is consistent with the scale effect assumption. However, the same effect does not appear on KMT candidates, which implies the meanings of the two major party brands are different. Empirical results suggest that vote-seeking candidates choose campaign strategy conditioning on the number of electorates in district. Regression result also suggests that candidates benefit from middle age, male, and incumbency.