關於「單記非讓渡投票」選制之下的有效政黨數目，曾有學者延伸杜弗傑定律，主張在應選名額為M的選區中，有效參選人數目應趨近於M+1。然而，這M+1位參選人究竟來自幾個政黨，卻仍沒有定論。本文主張，單記非讓渡投票制之下的有效政黨數目取決於選區應選名額與選民的偏好結構。當選民重視政策理念時，此一選舉制度的「離心效應」將使有效政黨數趨近M+1。當選民僅重視分配性的利益時，政黨數目會因為部份候選人必須組成穩定的分配聯盟而少於M+1；但政黨若有地域色彩則數目又會增加。本文假設理念型選民的比重和都市化程度成正比，並以農業人口比例和選區應選名額為自變數，解釋了台灣、日本與南韓在過去數十年間有效政黨數目的變化。 It has been demonstrated that, as an indicator of the fragmentation of political system, the effective number of candidates running under the single nontransferable vote (SNTV) system converges to M+1, where M is district magnitude. How these candidates are to be grouped into parties, however, remains unanswered. I propose that the effective number of parties under SNTV is determined by electoral system as well as the voters' preference profiles. When the voters are idealistic and select the candidates according to their issue positions, the centrifugal tendency of SNTV fosters M+1 effective parties. When the voters care only about absolute gains, the need to maintain a stable distributive coalition encourages one-party dominance, unless regionalism hinders the formation of a grand national party. Using the percentage of agricultural population to approximate the proportion of realist voters, I show that this formula explains well the evolution of party systems in Japan, South Korea, and Taiwan.