Contents lists available at ScienceDirect Political Geography journal homepage: www.elsevier.com/locate/polgeo Partisan and apportionment bias in creating a predominant party system Ali Çarkoğlu a,* , Deniz Aksen b a Koç University, College of Administrative Sciences and Economics, Department of International Relations, Rumelifeneri Yolu Sarıyer, 34450, Istanbul, Turkey b Koç University, College of Administrative Sciences and Economics, Department of Business Administration, Rumelifeneri Yolu Sarıyer, 34450, Istanbul, Turkey ARTICLE INFO Keywords: Apportionment Partisan bias Proportional representation systems Predominant parties Turkey ABSTRACT Moving beyond the analytical characteristics of apportionment methods or election systems, this article focuses on their outcomes in practice. We illustrate how apportionment and partisan biases working with a high threshold created an electoral environment conducive to the establishment of a predominant party system. We use the historical example of the Turkish experience. We trace the historical development of disproportionality for the entire multi-party elections for the 1950–2015 period. Focusing on the five most recent elections of this period since 2002, we demonstrate how the biases introduced by the apportionment method in use and the 10% threshold have advantaged the leading Justice and Development Party (Adalet ve Kalkınma Partisi, AKP). Our study suggests that a partisan bias favoring AKP still continues to exist at a lower level even after correcting the apportionment and the threshold biases. We underline how these biases form the foundation for a conservative over-representation and emphasize the path-dependent dynamics that keep challengers to the AKP away from the electoral scene, effectively helping to continue its hegemonic position in the system. 1. Introduction The representative capacity of political institutions is indispensable for legitimacy and effective governance in modern democracies. Parliaments bring together a number of legislators who are expected to represent a much larger number of individual citizens. A system of representation first relies on a limited number of representatives who are linked to geographic constituencies in accordance with their po- pulations via an apportionment system. For each of the geographic units, citizen support for candidates or parties is transformed into an earlier determined number of seats in each geographic constituency via an election system. The analytical basis of representation at both levels is the same one of linking many people to the much fewer number of either seats or party representatives. Given the distribution of people across geographic units or behind different political parties, seat dis- tributions in a representative assembly could be significantly different under different apportionment or election systems. Although characteristics of apportionment methods or election systems are well-known, their outcomes in practice are relatively scantly studied. 1 Our objective here is to offer an in-depth analysis of a proportional representation system and exhibit the nature and sources of the partisan bias in this system. Given the logic of a proportional representation system that aims to allocate seats in accordance with votes a party receives, one might be tempted to think that there should be no partisan bias in such systems. However, it is known that the D'Hondt method, for example, favors the larger parties at the expense of the smaller parties (Rae, Hanby, & Loosemore, 1971; Schuster, Pukelsheim, Drton, & Draper, 2003). Malapportionment effects that overlap with geographic distributional differences of parties' electoral support could also conceivably create a systematic bias favoring one party or electoral tradition over others. The Turkish party system pro- vides an excellent case to study partisan bias in a proportional re- presentation system. We know from past work of Çarkoğlu and Erdoğan (1998) that malapportionment is significant in Turkey and 10% threshold not only restricts regional parties' representation, but it ef- fectively also brings about a minimum size effect into the system. 2 Both factors are likely to create and maintain partisan biases despite pro- portional representation in use. We argue that the peculiar historical constellation of these biases in the apportionment as well as the election system in Turkey prepared a vastly advantageous ground for the Justice and Development Party (Adalet ve Kalkınma Partisi, AK Parti or AKP) capturing the executive https://doi.org/10.1016/j.polgeo.2018.11.009 Received 17 February 2018; Received in revised form 9 October 2018; Accepted 27 November 2018 * Corresponding author. E-mail addresses: acarkoglu@ku.edu.tr (A. Çarkoğlu), daksen@ku.edu.tr (D. Aksen). 1 For analytical treatments see Schwingenschlögl and Drton (2004). Balinski and Young (2010), Pukelsheim (2014), Schuster et al. (2003) and Taagepera, R., & Shugart (1989) provide empirical perspectives as well. 2 The 10% threshold was imposed in the aftermath of 1980 military coup in June 1983 before the November 1983 general elections. This threshold requires that a party receives at least 10% of the valid votes in order to gain representation in the parliament. Political Geography 69 (2019) 43–53 Available online 13 December 2018 0962-6298/ © 2018 Elsevier Ltd. All rights reserved. T