Do Party Lists Matter? Political Party Strategies in Legislative Candidate Nominations

Do Party Lists Matter? Political Party Strategies in Legislative Candidate Nominations Abstract In electoral systems with closed party lists, it is argued that the importance of central party organisation increases at the expense of individual candidates’ role in candidate nomination processes. This logic also underestimates individuals’ electoral potential and focuses on individuals’ allegiance to the leadership as the main asset for increasing their chances of being nominated. We argue that forming a party-list is a strategic decision based on the principle of furthering the interest of the party as a whole rather than rewarding individuals’ commitment to the party and is conditional on inter-party competition. We conveyed an original dataset of candidate lists for major parties in Turkey’s parliamentary elections between 1999 and 2015 and found empirical evidence for the significance of candidate lists as being used as strategic tools in inter-party electoral competition. 1. Introduction Elections constitute the primary instruments whereby political power is to some extent delegated from the populace to politicians in democratic systems. Politicians are mandated with making policies and are held accountable, at least in theory, for their actions (Strøm, 2000). However, political parties, with varying selection processes, exert considerable influence in this process via their candidate nomination decisions. Additionally, electoral institutions play an external role in shaping political parties’ behaviour. In this analysis, we focus on the impact of inter-party competition on parties’ electoral behaviour in a closed-list Proportional Representation (PR) system. 2. The puzzle Mitchell (2000) provides a typology of electoral systems based on the significance of political parties vis-à-vis the candidates. According to him, in candidate-centred electoral systems where district magnitude is low or in which voters have the chance for preference voting, a candidate’s reputation becomes more salient than when opposite conditions prevail. On the other end, there exists party-centred systems associated with closed-list PR systems. Compared with closed-list systems, in single-member district (SMD) systems and in open-list PR, candidates become more important, as personal reputation is a key factor in getting elected. However, in PR systems with closed lists, voters endorse their preferred political party’s list as a whole (e.g., Taagepera and Shugart, 1989; Sartori, 2005). According to Carey and Shugart (1995); vote pooling, type of ballot and number and type of votes cast all have a major impact on whether and the extent to which candidate reputation takes precedence over party reputation. Accordingly, closed-list PR systems—where each voter casts only one vote which contributes to the party’s list as a whole—the reputation of any given individual candidate is expected to be only marginally important. Therefore, compared with the majoritarian electoral systems and other applications of PR systems (i.e., preferential voting), in closed-list PR systems, the influence of party executives would be greater, as they have the privilege of proposing the party-lists. Thus, closed-list PR system contributes to centralisation in political parties with what may be significant consequences. Scholars have argued that this system bolstered a leader-dominated nomination process and contributes to the constant tension between the collective interest of the political party and the individual interest of a politician running for office (Fiorina, 1977; Cox, 1997). We claim that the implications of this difference between elections with closed party lists and open party lists are not always easy to determine. Electoral institutions enforce political parties and make them the ultimate decision-takers on whom to nominate, however, politics is an evolving phenomenon and impact of institutions are conditional on external social factors (Neto and Cox, 1997). In closed party list systems, as Lancaster (1986) argues, political parties become the constituency for the candidates as their re-election is critically dependent on parties’ choices rather than the voters’ preferences. Therefore, even when institutional rules restrict personal reputation among voters, political parties still have to make choices for some candidates to appear on the lists. For political parties the list of candidates is a strategic tool for fostering inter-party competition at the district level (Strøm, 1990). Parties have incentives to nominate those who are likely to mobilise the highest numbers of voters for the party list even under electoral systems with closed party lists. Therefore, even though the electoral system is party-centred, there is room for personal reputation and performance since political parties aim to maximise their electoral gains either for vote, office or policy purposes (Strøm, 1990). To investigate nomination strategies in closed-list PR systems, a case study based on a time-series panel data has some potential. In this case, potential explanatory factors of political institutions or culture and external shocks to political party system would be controlled and the impact of electoral competition between rival parties would be easier to detect. Thus, we focus on the General Elections (GE) in Turkey for the period of 2002–2015 and consider major parties’ strategies in nominating candidates. Turkey presents a unique opportunity to test Carey and Shugart’s (1995) theory, as the country’s electoral system provides all the conditions that are ordinarily expected to decrease the importance of a candidate’s personal qualities: (a) voters cannot have any impact on a party list prepared by leaders, (b) a vote for a candidate is counted first as a vote for the whole party, that is, vote pooling and (c) each voter casts only one vote. In such systems, Carey and Shugart (1995) argue the district magnitude to emerge as a factor of personal reputation whereas lower district magnitudes are associated with larger importance of personal vote. As district magnitudes in Turkey’s GE vary while parties’ nomination processes remain constant across districts, it enables us to test whether the number of seats per district play a role in the importance of personal vote. Yet, our findings provide mixed results. We found substantial evidence to suggest that in proposing party lists, party leaders refer to their parties’ district-level electoral success in the previous elections and hold candidates accountable for. In addition, the effect of the district size is rather ambiguous. Therefore, individual politicians are punished or rewarded for past electoral performance at the district level. Further, the emphasis on this factor is more pronounced when the inter-election period is short, as changes in social and economic factors are generally more limited over short than over long periods. Political parties consistently preserve continuity in their party lists in districts where they are electorally strong, which is in accord with Moral et al. (2015)’s finding of an incumbency advantage in Turkish politics. Hence, this article has two major objectives. First, by observing continuity versus change over candidate lists across five successive elections in the same political context, we will measure political party behaviour in regard to nominating candidates. In this respect, we will draw on an original dataset pertaining to the Turkish GE for the 2002–2015 period.2 Second, we will contribute to growing literature on Turkish politics by empirically investigating the determinants of the political parties’ respective choices regarding nominating candidates (Ciftci and Yildirim, 2017; Yildirim et al., 2017). Based on our results, we argue that even a centralised decision-making process incorporates district-level factors and shows the influence of inter-party competition on party lists at the district level. 3. Political parties and candidate nominations Candidate nomination is denoted as an important act of political parties (de Luca et al., 2002, p. 413) with implications for representation and electoral competition. Norris and Lovenduski (1995) answer ‘Who gets selected, and why?’ from supply and demand-side explanations. Here, supply-side explanation deals with the availability of applicants for candidacy who pursue a political career, and demand-side explanation discusses how party selectors select the candidates among these applicants by highlighting the importance of decision makers in nomination process. We concentrate on the demand side and delve into selection processes of candidates by political parties. Field and Siavelis (2008) note that in all electoral environments, whether democratic or not, the candidate selection process dramatically narrows the electorate’s choices with profound effects on representation (Mitchell, 2000). Therefore, the role of the party’s executive leadership and the strategic choices of this group have a significant impact on who will represent the citizenry in the national parliament and therefore the subsequent legislative roles in the post-election period (Koop and Bittner, 2011). More specifically, nominations also have considerable impact on individual politicians’ career paths and the mechanism of representation and significant implications for inter-party competition at the district level. Albeit regulated by same electoral rules, the size of the city also matters as in Turkey’s major cities such as İstanbul, Ankara, and İzmir, where districts are large, the connection between candidates and voters becomes uncertain, while in districts with lower magnitude such as Tunceli or Hakkari, the reputation of the candidates matters a great deal. We assume that voters in larger districts are usually under-informed about most of the candidates and mostly vote for the party with which they strongly identify. This observation can be regarded as reinforcing the conventional wisdom in the electoral politics literature: SMD systems are more likely to increase accountability between the district representative and voters. This is held to be the case because elected politicians have to make their political campaign visible to the public such that voters can demand policy choices, pork or parliamentary activity, from member states, which, in turn, protects the interests of voters and districts alike. Agency loss takes place much less in systems where candidates are more accountable to the voters via the electoral mechanism (Mitchell, 2000). However, this clear-cut division between PR and SMD plurality systems blurs with decreasing district magnitude. Therefore, in small districts, both the extent of the candidates’ popularity and the nature of their platforms and campaigns may have a significant influence on voters’ choices. This holds true for political systems, like Turkey’s, where clientelism is extensive (Güneş-Ayata, 1994; Sayarı, 2011; Kemahlıoğlu, 2012) and where exists a fierce intra-party competition for candidacy and district level inter-party competition for election. In most cases, those who capitalise on existing clientelistic ties with a constituency are more likely to be nominated compared with those who focus more on programmatic concerns over politics (Hellman, 2012). The individual candidate’s attributes, thus, increase the importance of individual candidates at the expense of the party’s leadership, as the latter is constrained in accord with the extent of a candidate’s electoral potential in the district. The recent history of Turkish politics includes instances of individual candidates who ran independently and even for another party when their own parties chose not to nominate them. Therefore, we argue that nominations in the Turkish electoral system do not fulfil expectations associated with a pure closed-list PR system wherein the traits of an individual candidate are of little consequence. Instead, we observe that individuals’ role in electoral competition is regarded as quite important. Earlier works also reveal that as district magnitude decreases, the importance of the locality increases (Moral et al., 2015).3 Hence, we argue that nominations processes provide a dynamic interaction between political parties and individual candidates and electoral systems with closed party-lists are not excluded. Norris and Lovenduski (1993)’s supply and demand framework reveals that the interaction between pool of candidates and parties’ choices matters for the nomination process. On the demand side, party selectors choose candidates based on their individual qualities, experiences or abilities. However, political parties are aware that a large proportion of the votes are received as a result of national-level appeal of political parties and this is unlikely to change conditional on the district-level candidate profiles. On the other hand, a group of individual politicians generate personal ties with the electorate and their appearance on the candidate list matters for the voters in the respective districts. On the supply side, not every party member is eager to run for candidacy. Those eager politicians usually claim that their candidacy is crucial for political parties’ success at the district level as their connection with the electorate is strong. Electoral results are good indicators for such claims, particularly in political contexts where political parties are relatively centralised and nomination decisions are made by the party leader and top cadet members. In such contexts, party lists are prepared conditional on the previous election results. In districts where previous group of candidates achieved a level of success measured in the form of number of votes or seats increased, political party leaders have less incentives to make modifications in the next electoral cycle. 3.1 The context: elections in Turkey The Turkish political system has been recognised as a parliamentary democracy since the end of military rule in 1983. Despite major problems in the country’s democratic process (Gümüşçü, 2013; Kalaycıoğlu, 2015), inter-party competition remains as an important aspect of Turkish electoral politics and the critical choices made by parties at all levels of electoral competition require further scrutiny. Elections for legislature, that is, the Türkiye Büyük Millet Meclisi (TBMM—Turkish Grand National Assembly), are held every four years by implementing closed-list PR4 with an unprecedentedly high national threshold of 10%, and D’Hondt formula is used in 85 electoral districts with varying district magnitude.5 Since the 1995 GE, the closed party list system has been used and has contributed to increased levels of party centralisation and intra-party discipline. The electoral rules in Turkey remained unchanged since 1995 GE. Yet, the party system and effective parties in the system differed tremendously between 1990s and 2000s, whereas the former is associated with an unstable multi-party system predisposed to coalitional governments while the latter is associated with a consolidated multi-party system where a single party government was more likely as two parties emerge as major competitors. Therefore, we concentrate on the latter for two reasons. First, stable party system allows us to track changes between different elections. Second data availability for 1990s is a major problem. Thus, analysing the GEs between 2002 and 2015 yields a good opportunity to focus on four major parties’ behaviour. 3.2 Candidate nomination processes in Turkey The analysis will be based on four consistent and major parties in the Turkish political party system: Adalet ve Kalkınma Partisi (AKP—Justice and Development Party), Cumhuriyet Halk Partisi (CHP—Republican People’s Party), Milliyetçi Hareket Partisi (MHP—Nationalist Movement Party) and Halkların Demokratik Partisi (HDP—People’s Democratic Party). Throughout the time period of this study, district magnitude in Turkey varied between 1 (e.g., Bayburt in 2011) and 31 (e.g., İstanbul district No. 1 and No. 3 in 2015), averaging 6.5 in 85 electoral units.6 In terms of candidacy, from the 2002 to the 2015 GE, the MHP nominated 2065 different candidates, whereas the corresponding figure was 2044 for the CHP and 1800 for the AKP. Among those who were candidates in the November 2015 GE for one of the four focal parties, only 27 (less than 1%) were nominated in all the GE from 2002 to November 2015, including prominent party members such as Kemal Kılıçdaroğlu (the CHP leader since 2011), Devlet Bahçeli (the MHP leader) and Pervin Buldan from the HDP. The number of the AKP candidates who were candidates in the five successive elections is very small, as it implemented a self-imposed three-term limit for the candidates who ran successively. In addition, 70 candidates could not be nominated in the June 2015 GE due to this term limit, yet 23 of those 70 were candidates in the November 2015 GE. 3.3 Party structure in Turkey As mentioned earlier, electoral laws alone do not explain the level of recruitment and nomination centralisation (Siavelis and Morgenstern, 2008). Parties exhibit different types of nomination processes. While for some implement inclusive mechanisms including local party primaries or national level conventions, for others, nomination is an exclusive procedure where only top party cadets have the authority to make decisions (Hazan and Rahat, 2010). Candidate selection processes both affect the characteristics and acts of elected representatives in their offices (Carey and Shugart, 1995; de Luca et al., 2002). More inclusive candidate selection processes would create better incentives for legislators or candidates to well develop their proposed policy agendas as individuals to better compete with other candidates. On the other hand, if the legislators’ re-nomination chances are under the control of local governors or local party bosses rather than the constituencies, these legislators professional career path development and specialisation in certain policy areas can be limited (Jones et al., 2002). Norris and Lovenduski (1995) classifies different recruitment process on two dimensions: having formal rules and the levels of recruitment process. We assume that the more centralised the structures are the more important priority lies with winning maximum number of seats to take part in the government formation or to become a significant party in the parliament after the elections. The influence of electoral competition over leaders is expected to increase in centralised parties as leaders have less information about the lower ranks of the party politics and their main tool for evaluation would be retrospective electoral results. Turkish case provides us an ample opportunity to control for different party structures. Because, it is well established among scholars of Turkey’s party system that parties are centralised structures in which hierarchy is institutionalised in making policy choices and nominations since electoral democracy was initiated in 1950s (Ayan, 2010; Wuthrich, 2015) and that many parties, regardless of ideological left–right positioning, employ a central control mechanism (Kınıklıoğlu, 2002). Although there are no legal regulations regarding holding primaries or fulfilling a gender quota, several left-wing parties employ such practices to some extent: The CHP held primaries in several elections, and the HDP specifies a gender quota in its by-laws, but they are as binding as party leadership is committed to these mechanisms. Given the level of party centralisation and the role of party executives in candidate nominations in Turkish politics, it is a critical choice for political parties to present a list of candidates in each district for two major reasons. First, as a tenet of closed party lists, nominations have a significant role in determining who is elected to parliament. Second, it is a way to maximise party votes, as candidates with strong social ties are more likely to successfully mobilise the electorate for parties that have nominated them. For this reason, it is important to understand the factors that shape parties’ political behaviour and analyse the extent to which parties take inter-party competition, local factors and the attributes of the candidates into consideration—all of which are affected by both socio-economic factors and competition. To understand these relationships, we now turn to a multivariate analysis of the candidate nomination strategies of Turkey’s four major political parties for the GE from 2002 to November 2015. 4. Hypotheses and data Based on the argument explained above we have derived the following hypotheses: H1: Parties are more likely to show continuity in party lists where they are electorally stronger. H2: Parties are more likely to change their lists in districts where they have faced vote loss in the previous elections. We collected data from official resources (see Supplementary Material). Our unit of analysis is the party-district. Each variable is measured at the district level for four major parties in Turkey’s party system. 4.1 Dependent variable(s): Continuity in nominations We differentiated between the candidates based on their probability of gaining office, and we created two categories of candidates. The first dependent variable is the rate of continuity of all candidates in the party list from previous elections. We compared two party lists at t−1 and t and divided the total number of recurring candidates by the district magnitude as follows: rate of continuity =Number of recurring candidatesdistrict magnitude. The second dependent variable is referred to as the rate of continuity of favourites. As parties must provide a list of candidates that is in line with the district magnitude and the electoral system is a closed party list, those in the upper part of the list are more likely to get elected than are those who appear lower in the list.7 We referred to those whose names appear in the upper part of the list as favourites. However, it was necessary to make an additional decision in order to ascertain empirically what the upper part of the list refers to. We decided that if party A won three seats in the previous elections (t−1), the candidates ranked in the top 3 spots of the list in the current elections (t) would constitute favourites. That is, candidates in the top 3 have a much better chance of re-election than do those placed lower down on the list. Therefore, we created our second dependent variable, the rate of continuity of favourites, based on this information by dividing the actual number of incumbents re-nominated as favourites by the number of seats won in the previous election: rate of continuity of favourites=Number of incumbents as favouritesnumber of seats won at (t-1). These two dependent variable ranges between 0 and 1 where 0 means all favourites/candidates changed and 1 means all favourites/candidates continued to be so in the next elections. 4.2 Independent variables: Electoral competition T−1 vote shares (%): This variable is calculated by the percentage of votes won by each political party at the district level. We expect those who were successful in the previous elections to be more likely to keep their candidates on the list because we expect parties to take electoral success into consideration when making nominations. Vote loss: This variable is measured by subtracting the percentage of the vote won by a political party at elections at t−1 from t−2. Seat loss: This variable is calculated by dividing the difference between the number of seats won at t−2 and t−1 by the number of seats won at t−2. This variable is also used to determine electoral success because parties may consider the actual number of seats gained (or lost) rather than mere percentages and re-consider their nomination strategies in response to an overall loss of seats. We expect to find that vote or seat loss would drive political parties to change party lists in the next election. Both for electoral loss and seat loss variables, positive numbers are associated with losing votes and seats, respectively, as we calculate the difference between the elections at t−2 and t−1. 4.2.1 Control variables Economic development: This variable is calculated by dividing the gross-domestic added value of a district by the total gross-domestic added value at the national level. Thus, we were able to find the weight of the district in the economic production of the entire country. We controlled for this variable, as variations in the extent of economic interaction might have a role in the relative strength of the linkage between voters and politicians. In highly developed districts, clientelistic linkages between candidates and voters are less important than in districts where pre-industrial economic ties are more common and poverty levels in constituencies are relatively high (see e.g., Weitz-Shapiro, 2009; Stokes et al., 2013). Nevertheless, we approach this control variable with great caution because electoral districts are shaped by administrative functions rather than socioeconomic factors; therefore, in highly developed districts such as large urban areas (e.g., Istanbul or Ankara), we still observe clientelistic practices in Turkish context. However, given the absence of a comparative index of clientelism across different districts, we suggest that economic development variable is still the best available proxy to control for the effects of clientelism as well as other potential mechanisms in accordance with previous literature. Large districts is measured on the basis of district magnitude. We expected that in small districts continuity in candidates would be higher as politicians’ level of interaction with constituencies increase in small districts. Urban/rural share: We adopted this indicator from the Türkiye İstatistik Kurumu (TÜİK—Turkish Statistical Institute), where the level of urbanisation is measured by dividing the urban population by the total population of a district. 5. Analysis and findings 5.1 Candidate nominations in Turkey 1999–2015 We have conducted the analysis for past elections between 1999 and 2015 to test our main hypotheses stated above.8 To investigate the determinants of the strategies pursued by political parties in making up party lists for electoral districts, we ran the multivariate generalised linear model (GLM) between 2007 and June 2015 GE. We selected GLM because our dependent variable is a rate of the continuity, which varies between 0 and 1, and running a linear regression would also predict values below 0 and above 1. The descriptive results reported in Table 1 show that political parties indicate less continuity in both candidates and favourite candidates in previous elections in comparison to 2015 November GE. Table 1 Continuity in nominations in Turkey’s GE (2007–2015) 2007 2011 2015 June 2015 November Party names Continuity in No. % No. % No. % No. % AKP Candidates 189 0.34 149 0.26 156 0.28 297 0.54 Favourites 157 0.43 114 0.33 122 0.37 183 0.70 Elected 145 0.43 108 0.33 98 0.37 188 0.59 CHP Candidates 117 0.21 74 0.13 112 0.20 321 0.58 Favourites 65 0.36 27 0.24 41 0.30 125 0.95 Elected 47 0.42 26 0.19 36 0.28 123 0.92 MHP Candidates 97 0.18 122 0.22 82 0.15 243 0.44 Favourites NA 29 0.40 25 0.47 69 0.86 Elected NA 23 0.43 23 0.29 37 0.93 HDP11 Candidates NA NA NA 236 0.43 Favourites NA NA NA 67 0.83 Elected NA NA NA 51 0.86 2007 2011 2015 June 2015 November Party names Continuity in No. % No. % No. % No. % AKP Candidates 189 0.34 149 0.26 156 0.28 297 0.54 Favourites 157 0.43 114 0.33 122 0.37 183 0.70 Elected 145 0.43 108 0.33 98 0.37 188 0.59 CHP Candidates 117 0.21 74 0.13 112 0.20 321 0.58 Favourites 65 0.36 27 0.24 41 0.30 125 0.95 Elected 47 0.42 26 0.19 36 0.28 123 0.92 MHP Candidates 97 0.18 122 0.22 82 0.15 243 0.44 Favourites NA 29 0.40 25 0.47 69 0.86 Elected NA 23 0.43 23 0.29 37 0.93 HDP11 Candidates NA NA NA 236 0.43 Favourites NA NA NA 67 0.83 Elected NA NA NA 51 0.86 Table 1 Continuity in nominations in Turkey’s GE (2007–2015) 2007 2011 2015 June 2015 November Party names Continuity in No. % No. % No. % No. % AKP Candidates 189 0.34 149 0.26 156 0.28 297 0.54 Favourites 157 0.43 114 0.33 122 0.37 183 0.70 Elected 145 0.43 108 0.33 98 0.37 188 0.59 CHP Candidates 117 0.21 74 0.13 112 0.20 321 0.58 Favourites 65 0.36 27 0.24 41 0.30 125 0.95 Elected 47 0.42 26 0.19 36 0.28 123 0.92 MHP Candidates 97 0.18 122 0.22 82 0.15 243 0.44 Favourites NA 29 0.40 25 0.47 69 0.86 Elected NA 23 0.43 23 0.29 37 0.93 HDP11 Candidates NA NA NA 236 0.43 Favourites NA NA NA 67 0.83 Elected NA NA NA 51 0.86 2007 2011 2015 June 2015 November Party names Continuity in No. % No. % No. % No. % AKP Candidates 189 0.34 149 0.26 156 0.28 297 0.54 Favourites 157 0.43 114 0.33 122 0.37 183 0.70 Elected 145 0.43 108 0.33 98 0.37 188 0.59 CHP Candidates 117 0.21 74 0.13 112 0.20 321 0.58 Favourites 65 0.36 27 0.24 41 0.30 125 0.95 Elected 47 0.42 26 0.19 36 0.28 123 0.92 MHP Candidates 97 0.18 122 0.22 82 0.15 243 0.44 Favourites NA 29 0.40 25 0.47 69 0.86 Elected NA 23 0.43 23 0.29 37 0.93 HDP11 Candidates NA NA NA 236 0.43 Favourites NA NA NA 67 0.83 Elected NA NA NA 51 0.86 This is not surprising given the fact that the inter-election period between regular elections were four years and many intra-party political factors as well as individual and voluntary factors contribute into this rate of change. However, for all parties, rate of continuity among favourite candidates was higher than rate of continuity in all candidates. Another important finding is that scores of continuity in favourites are very close to scores of continuity in elected candidates. This also points out that political parties have a very important role in deciding whom to be elected as nominating someone at the top of a party list significantly increase their chances of being elected. The results from CHP also point to the importance of party leaders in nominations. Leadership change in major Turkish political parties is uncommon. Between 2002 and 2015, MHP experienced no leadership change while change in CHP in 2011 was due to an unexpected scandalous case that overthrew the previous leader. AKP also changed its leader before 2015 elections as a result of Erdoğan’s ascendance to presidential office, however, his de jure leadership on AKP continued albeit the absence of legal ties and 2017 constitutional referendum allowed him to be back at the helm of AKP while keeping presidential post. Therefore, it is fair to state that the only leadership change in political parties in the period we cover here took place at CHP and it explains the anomalous findings for continuity in nominations in 2011 GE. While Turkish case presents an overall anomaly in lower levels of leadership turnover in comparative perspective, it allows to gauge trends that would be confounded in party systems with a lot of central leadership turnover.9 We ran the multivariate GLM for all the elections from 2002 to June 2015, the results of which are reported in Table 2. We left November 2015 elections as it was the only snap elections throughout the period covered in this paper and it generates some irregularities that will be discussed below. Table 2 reveals that the percentage of votes realised by the parties remains an important predictor for rate of continuity of candidates on the lists which supports our first hypothesis indicated earlier. Parties keep their candidate list, to a greater extent, when they are performing well at the district. On factor that strengthens this relationship is that political parties nominate members affiliated with national leadership of the party at stronghold districts and the candidate is locally known so that their re-election is ensured.10 Thus, it is even more important for smaller parties to use this strategy as they have higher proportion of prominent members when compared with potential seats. However, for our second hypothesis which emphasises the relationship between the vote loss and continuity of candidates, we have ambiguous results. Vote loss, in the whole sample turns out to be insignificant while the coefficient is negative as expected, and only significant for the MHP. We argue that this is due to the MHP’s insecure electoral insecurity to exceed 10% electoral threshold. Table 2 GLMs: determinants of continuity in party lists: from 1999 to June 2015 GE All parties AKP CHP MHP MHP (t−1) vote share 2.70*** 2.46*** 1.67*** 1.66*** 2.38*** 1.40 −0.96 4.84*** (0.26) (0.28) (0.34) (0.33) (0.6) (1.04) (1.95) (1.57) Vote loss −1.23 −6.48*** (0.78) (2.21) Seat loss −0.81*** −0.81*** −0.21 −1.45*** −0.80 (0.23) (0.23) (0.40) (0.36) (−0.95) Economic development 0.045 0.04 0.05 0.06 0.07 0.07 −0.017 (0.05) (0.05) (0.05) (0.07) (0.11) (0.17 (0.08) Large districts −0.026 −0.03 −0.04*** −0.04*** −0.04 −0.06 −0.07 0.001 (0.017) (0.01) (0.02) (0.01) (0.025) (0.04) (0.06) (0.027) Urban/rural 0.0005 0.001 −0.002 −0.001 −0.007 0.005 0.007 (0.0045) (0.005) (0.004) (0.007) (0.007) (0.011) (0.008) Constant −1.98*** −2.02*** −1.15 −7.74*** −1.70 −0.56 −0.99 −2.86*** (0.28) (0.29) (0.33) (−1.30) (0.49) (0.56) (0.94) (0.53) Number of observation 510 510 329 329 167 116 46 170 All parties AKP CHP MHP MHP (t−1) vote share 2.70*** 2.46*** 1.67*** 1.66*** 2.38*** 1.40 −0.96 4.84*** (0.26) (0.28) (0.34) (0.33) (0.6) (1.04) (1.95) (1.57) Vote loss −1.23 −6.48*** (0.78) (2.21) Seat loss −0.81*** −0.81*** −0.21 −1.45*** −0.80 (0.23) (0.23) (0.40) (0.36) (−0.95) Economic development 0.045 0.04 0.05 0.06 0.07 0.07 −0.017 (0.05) (0.05) (0.05) (0.07) (0.11) (0.17 (0.08) Large districts −0.026 −0.03 −0.04*** −0.04*** −0.04 −0.06 −0.07 0.001 (0.017) (0.01) (0.02) (0.01) (0.025) (0.04) (0.06) (0.027) Urban/rural 0.0005 0.001 −0.002 −0.001 −0.007 0.005 0.007 (0.0045) (0.005) (0.004) (0.007) (0.007) (0.011) (0.008) Constant −1.98*** −2.02*** −1.15 −7.74*** −1.70 −0.56 −0.99 −2.86*** (0.28) (0.29) (0.33) (−1.30) (0.49) (0.56) (0.94) (0.53) Number of observation 510 510 329 329 167 116 46 170 Note: Standard error in paranthesis. Table 2 GLMs: determinants of continuity in party lists: from 1999 to June 2015 GE All parties AKP CHP MHP MHP (t−1) vote share 2.70*** 2.46*** 1.67*** 1.66*** 2.38*** 1.40 −0.96 4.84*** (0.26) (0.28) (0.34) (0.33) (0.6) (1.04) (1.95) (1.57) Vote loss −1.23 −6.48*** (0.78) (2.21) Seat loss −0.81*** −0.81*** −0.21 −1.45*** −0.80 (0.23) (0.23) (0.40) (0.36) (−0.95) Economic development 0.045 0.04 0.05 0.06 0.07 0.07 −0.017 (0.05) (0.05) (0.05) (0.07) (0.11) (0.17 (0.08) Large districts −0.026 −0.03 −0.04*** −0.04*** −0.04 −0.06 −0.07 0.001 (0.017) (0.01) (0.02) (0.01) (0.025) (0.04) (0.06) (0.027) Urban/rural 0.0005 0.001 −0.002 −0.001 −0.007 0.005 0.007 (0.0045) (0.005) (0.004) (0.007) (0.007) (0.011) (0.008) Constant −1.98*** −2.02*** −1.15 −7.74*** −1.70 −0.56 −0.99 −2.86*** (0.28) (0.29) (0.33) (−1.30) (0.49) (0.56) (0.94) (0.53) Number of observation 510 510 329 329 167 116 46 170 All parties AKP CHP MHP MHP (t−1) vote share 2.70*** 2.46*** 1.67*** 1.66*** 2.38*** 1.40 −0.96 4.84*** (0.26) (0.28) (0.34) (0.33) (0.6) (1.04) (1.95) (1.57) Vote loss −1.23 −6.48*** (0.78) (2.21) Seat loss −0.81*** −0.81*** −0.21 −1.45*** −0.80 (0.23) (0.23) (0.40) (0.36) (−0.95) Economic development 0.045 0.04 0.05 0.06 0.07 0.07 −0.017 (0.05) (0.05) (0.05) (0.07) (0.11) (0.17 (0.08) Large districts −0.026 −0.03 −0.04*** −0.04*** −0.04 −0.06 −0.07 0.001 (0.017) (0.01) (0.02) (0.01) (0.025) (0.04) (0.06) (0.027) Urban/rural 0.0005 0.001 −0.002 −0.001 −0.007 0.005 0.007 (0.0045) (0.005) (0.004) (0.007) (0.007) (0.011) (0.008) Constant −1.98*** −2.02*** −1.15 −7.74*** −1.70 −0.56 −0.99 −2.86*** (0.28) (0.29) (0.33) (−1.30) (0.49) (0.56) (0.94) (0.53) Number of observation 510 510 329 329 167 116 46 170 Note: Standard error in paranthesis. On the other hand, parties’ responsiveness to the seat loss is more evident and negatively associated with the outcome of interest, as we expected. Therefore, when parties lose seats, they are more willing to change their candidates in the upcoming elections. District magnitude has negative impact on continuity of party lists while it is only significant in models when seat loss is concerned for the whole sample. Therefore, even though results are somewhat mixed, findings reveal that the larger the districts are the more likely the lists will be changed in accordance with the expectations from the pre-existing literature which reveals that incumbents have more advantage in smaller districts. All remaining control variables turn out to have insignificant effects. From these results, we conclude that inter-party competition at district level explains the continuity in party lists at different degrees. However, mostly the relationship is reverse, in other terms, the more they lose electorally, the more likely they will change their candidate lists. One noticeable finding is that when we ran the analysis separately for political parties we find finer results on how parties react differently to electoral outcomes. For example, MHP is more responsive to electoral loss as while CHP is to seat loss and AKP is more considerate of the previous electoral strength at the district. We claim that this is due to the fact that MHP risks national threshold and therefore is more focused on its vote share while CHP is rather concentrated in competing with AKP at districts to maximise its parliamentary seats. Thus, while all parties are found to be considerate of their previous electoral results, their concerns might differ. 5.2 2015: an exceptional year Turkey lived through an exceptional electoral year in 2015 as the country held two GE on 7 June and 1 November 2015. The main reason for this unprecedented electoral competition was the failure of the parties to form a governing coalition following the 7 June elections in which no single party won a parliamentary majority and subsequent coalition talks broke down. As a result, snap elections were called for 1 November. The inter-election period witnessed an increase in civil strife, coalition talks and mutual blame between the political parties accompanied by economic instability (Sayarı, 2016). For the political parties, there were few opportunities to make major changes in regard to implementing a policy position or furthering a political agenda. However, the political parties were required to form new lists of candidates in the electoral districts. Therefore, that two GEs were held in such a short period of time provided an ideal opportunity for analysis to control for temporal and socio-economic factors. As a result of electoral considerations, the political parties made several changes prior to the second GE in 2015, which were held on 1 November. Table 3 shows that compared with the other major parties, the AKP reacted very differently to the 7 June election results and kept only 54% of those on the candidate lists and 70% of their favourite candidates. While AKP does not differ on the overall continuity of the candidates from rival parties, they evaluated their June 2015 election performance as below-par and made sever changes at the upper echelons of the party list. On the other hand, the CHP, MHP and HDP made relatively smaller changes in the upper part of their respective lists even though each changed the lower part of its list to an even greater extent than did the AKP. Table 3 Continuity in nominations in the November 2015 GE Party name MPs elected in June 2015 Continuity of candidates % of continuity Continuity of favourite candidates % of elected candidates MVs elected in November 2015 MP change AKP 258 297 54.00 183 70.9 317 +59 CHP 132 320 58.19 125 94.7 134 +2 MHP 80 243 44.18 69 86.25 40 −40 HDP 80 235 42.72 67 83.75 59 −21 Party name MPs elected in June 2015 Continuity of candidates % of continuity Continuity of favourite candidates % of elected candidates MVs elected in November 2015 MP change AKP 258 297 54.00 183 70.9 317 +59 CHP 132 320 58.19 125 94.7 134 +2 MHP 80 243 44.18 69 86.25 40 −40 HDP 80 235 42.72 67 83.75 59 −21 Table 3 Continuity in nominations in the November 2015 GE Party name MPs elected in June 2015 Continuity of candidates % of continuity Continuity of favourite candidates % of elected candidates MVs elected in November 2015 MP change AKP 258 297 54.00 183 70.9 317 +59 CHP 132 320 58.19 125 94.7 134 +2 MHP 80 243 44.18 69 86.25 40 −40 HDP 80 235 42.72 67 83.75 59 −21 Party name MPs elected in June 2015 Continuity of candidates % of continuity Continuity of favourite candidates % of elected candidates MVs elected in November 2015 MP change AKP 258 297 54.00 183 70.9 317 +59 CHP 132 320 58.19 125 94.7 134 +2 MHP 80 243 44.18 69 86.25 40 −40 HDP 80 235 42.72 67 83.75 59 −21 Recall that the June 2015 elections were considered as a major failure for the AKP given that the party lost its parliamentary majority for the first time since its entry into electoral competition in the 2002 elections. One factor behind this change was the AKP’s implementation of the three-term limit specified in its by-laws, according to which no candidate can run for more than three consecutive terms on the AKP ticket. Some of the politicians who were subject to this rule in the June 2015 elections were later reinstated to the party lists in accordance with the preferences of party leaders. According to our descriptive findings relating to the November 2015 elections, it can be argued that the AKP chose to shuffle its candidates in order to recover from the losses it had experienced in the June 2015 elections. For example, in the November 2015 elections, the AKP replaced all of its June 2015 candidates with new party members in six districts (Gümüşhane, Ağrı, Iğdır, Karabük, Edirne and Hakkari). Further, if we compare the AKP’s results in the June 2015 elections with its results in the 2011 elections, we observe that the percentage of party votes decreased in all these districts by 9% in Gümüşhane, 29% in Ağrı, 17% in Iğdır, 10% in Karabük, 5% in Edirne and 7% in Hakkari. However, a decline in the AKP’s votes prevailed across multiple electoral units. In some of the electoral units, however, the AKP responded differently. For example, despite an electoral decline in constituencies including Bilecik, Düzce and Sinop, the AKP did not change its list for the November 2015 elections. Therefore, further scrutiny is required if we are to understand why parties change their candidates in some districts but not in others. Given that 2015 was an exceptional year in regard to inter-election time period of GE, it is necessary to compare the findings for the elections in this year with those for other elections. In this way, it will be possible to determine whether there is a consistent trend across previous elections. Our major motivation is to find out the reasons for this variation among political party behaviour. To investigate, we ran the multivariate GLM only for two GEs in 2015 and reached to findings below. Models on the left side of the table (I, II and III) predict continuity in candidate lists as a whole, while those on the right (IV, V and VI) predict candidates that are nominated at the upper echelons of the party list. The results from the GLM model above show that all the parties have more continuity in their party lists in districts where they have a large vote share. The t−1 vote share variable is significantly associated with continuity in regard to nominating all and favourite candidates. When we test our main hypothesis of party competition we find strong evidence that political parties considered their electoral loss, measured by vote and seat loss and altered their party lists accordingly. Given that the AKP considered June 2015 elections as an electoral failure, we added a dummy variable for the AKP and found that the AKP modified its candidate list more frequently compared with other parties in November 2015 elections. When we considered seat loss as an indicator of electoral loss, we found that parties that had lost an incumbent seat in the last elections changed their list of candidates. From the results reported in Table 4, we found that when the number of seats a party won in the election declined when compared with the seats won in the last elections, those parties were more likely to make major changes in their party lists in the next elections. This indicates that the parties consider lost seats to be an electoral failure on the part of party members on the list and nominate different members in the next election to mobilise support in an effort to regain seats lost in the previous election. Losing electoral votes also has a similar effect for political parties and they form their party lists in next elections based on previous success or failures. On the other hand, we found all the socio-economic variables used herein as a proxy for other relevant mechanisms including clientelism and political culture to be insignificant in explaining the level of change in nomination lists except district magnitude in only Model 1. In larger districts, continuity is higher and we claim that this might occur due to an effort by parties to guarantee prominent members’ re-election instead of local electoral concerns while further investigation is required. Table 4 GLM: determinants of continuity in party lists—November 2015 GE Continuity in candidate lists Continuity in favourite candidates I II III IV V VI (t−1) vote share 4.35*** 2.66*** 4.44*** 1.84 0.50*** 4.39*** (0.51) (0.59) (0.74) (1.30) (1.19) (1.48) Vote loss −2.95*** −12.36*** (1.04) (2.82) Seat loss −0.83*** −0.55** −2.13** −0.46 (0.28) (0.3) (0.80) (0.65) AKP −0.79*** −2.10*** (0.22) (0.48) Economic development −0.05 0.05 0.04 −0.08 −0.08 0.05 (0.07) (0.07) (0.07) (0.11) (0.12) (0.12) Large districts 0.06*** (−0.002) (0.004) 0.03 0.008 0.01 (0.02) (0.02) (0.02) (0.04) (0.044) (0.04) Urban/rural −0.001 −0.002 −0.003 −0.005 0.005 0.002 (0.006) (0.007) (0.007) (0.015) (0.014) (0.01) Constant −1.47*** −0.43*** −0.65 1.44 0.97 0.91 (0.40) (0.50) (0.52) (1.18) (1.08) (1.25) Number of observation 255 173 173 179 163 163 Continuity in candidate lists Continuity in favourite candidates I II III IV V VI (t−1) vote share 4.35*** 2.66*** 4.44*** 1.84 0.50*** 4.39*** (0.51) (0.59) (0.74) (1.30) (1.19) (1.48) Vote loss −2.95*** −12.36*** (1.04) (2.82) Seat loss −0.83*** −0.55** −2.13** −0.46 (0.28) (0.3) (0.80) (0.65) AKP −0.79*** −2.10*** (0.22) (0.48) Economic development −0.05 0.05 0.04 −0.08 −0.08 0.05 (0.07) (0.07) (0.07) (0.11) (0.12) (0.12) Large districts 0.06*** (−0.002) (0.004) 0.03 0.008 0.01 (0.02) (0.02) (0.02) (0.04) (0.044) (0.04) Urban/rural −0.001 −0.002 −0.003 −0.005 0.005 0.002 (0.006) (0.007) (0.007) (0.015) (0.014) (0.01) Constant −1.47*** −0.43*** −0.65 1.44 0.97 0.91 (0.40) (0.50) (0.52) (1.18) (1.08) (1.25) Number of observation 255 173 173 179 163 163 Note: Standard errors in parentheses. Table 4 GLM: determinants of continuity in party lists—November 2015 GE Continuity in candidate lists Continuity in favourite candidates I II III IV V VI (t−1) vote share 4.35*** 2.66*** 4.44*** 1.84 0.50*** 4.39*** (0.51) (0.59) (0.74) (1.30) (1.19) (1.48) Vote loss −2.95*** −12.36*** (1.04) (2.82) Seat loss −0.83*** −0.55** −2.13** −0.46 (0.28) (0.3) (0.80) (0.65) AKP −0.79*** −2.10*** (0.22) (0.48) Economic development −0.05 0.05 0.04 −0.08 −0.08 0.05 (0.07) (0.07) (0.07) (0.11) (0.12) (0.12) Large districts 0.06*** (−0.002) (0.004) 0.03 0.008 0.01 (0.02) (0.02) (0.02) (0.04) (0.044) (0.04) Urban/rural −0.001 −0.002 −0.003 −0.005 0.005 0.002 (0.006) (0.007) (0.007) (0.015) (0.014) (0.01) Constant −1.47*** −0.43*** −0.65 1.44 0.97 0.91 (0.40) (0.50) (0.52) (1.18) (1.08) (1.25) Number of observation 255 173 173 179 163 163 Continuity in candidate lists Continuity in favourite candidates I II III IV V VI (t−1) vote share 4.35*** 2.66*** 4.44*** 1.84 0.50*** 4.39*** (0.51) (0.59) (0.74) (1.30) (1.19) (1.48) Vote loss −2.95*** −12.36*** (1.04) (2.82) Seat loss −0.83*** −0.55** −2.13** −0.46 (0.28) (0.3) (0.80) (0.65) AKP −0.79*** −2.10*** (0.22) (0.48) Economic development −0.05 0.05 0.04 −0.08 −0.08 0.05 (0.07) (0.07) (0.07) (0.11) (0.12) (0.12) Large districts 0.06*** (−0.002) (0.004) 0.03 0.008 0.01 (0.02) (0.02) (0.02) (0.04) (0.044) (0.04) Urban/rural −0.001 −0.002 −0.003 −0.005 0.005 0.002 (0.006) (0.007) (0.007) (0.015) (0.014) (0.01) Constant −1.47*** −0.43*** −0.65 1.44 0.97 0.91 (0.40) (0.50) (0.52) (1.18) (1.08) (1.25) Number of observation 255 173 173 179 163 163 Note: Standard errors in parentheses. The results from the November 2015 elections show that the parties view inter-party competition as an important consideration in determining their party lists. However, as we have explained, the November 2015 elections constitute an exceptional case given that they took place only five months after the previous election. Therefore, inter-party competition might be the most important factor given that changes in other factors including urbanisation, economic development and patron–client relationships remained stable between June and November 2015 and this is clearly reflected in our hypotheses tests. 6. Conclusion In this paper, we investigated political parties’ nomination in Turkey’s parliamentary elections since 1999 by using an original dataset of individual candidates from the 1999 to the 2015 elections. We focused on the lists of candidates from each district as an indicator of the political parties’ strategic decision. Our findings support the hypotheses listed above and indicate that political parties modify their candidate lists in districts where they previously lose electoral support, whereas candidates in districts where the party is electorally stronger are continued. We found ambiguous relationship between party nominations and institutional factors (i.e., district magnitude) and found no support for socio-economic factors such as urbanisation and economic development. We argue that inter-party competition plays an important role in political parties’ decisions to propose party lists in each district. We claim that this study possesses several challenges to studies on electoral systems and party politics. First, it reveals that institutional impact on the importance of cultivation of personal vote is conditional on the degree of electoral competition. Contrary to conventional wisdom on closed party list systems which states that voters make their choices at the ballot box in accord with their party preferences, such that candidates matter less in such systems, we found that candidates do matter. We are not denying that political parties are the ultimate decision takers in making up the lists; however, we argue that candidates’ popularity at the district level is an important factor being played into party leaders’ decisions when candidate lists are being made. Political parties strategically set up candidate lists to appeal to voters particularly even when nomination processes are centralised around the party leadership. Secondly, we also claim that this mechanism also operates in countries with extensive clientelistic networks. It is generally argued that clientelism is an important factor in Turkish politics in establishing the linkage between voters and parties. Even though it is argued that in such systems parties prefer candidates who have clientelistic ties rather than those who perform better in political campaigning or policy making, we found significant results that candidates’ electoral potential matters a great deal for parties in Turkish politics. Parties prefer to keep candidates with electoral potential, whether that comes from clientelistic networks or political activity. Supplementary Data Supplementary data are available at Parliamentary Affairs online. Conflict of interest The author reports no conflict of interest. Footnotes 1 A previous version of this article was presented at the second Annual Workshop of Empirical Studies in Political Analysis in Antalya, ECPR Standing Group in Parliaments 2016 Annual Meeting in Munich and Sabancı University Brown Bag Seminars. We would like to thank H. Ege Özen, Özge Kemahlıoğlu, Ersin Kalaycıoğlu, Umman Mahir Yıldırım, Şebnem Gümüşçü, Kerim Can Kavaklı and the ESPA 2016 panel participants for their comments on an earlier version of this article. 2 We analyse the GE between 2002 and November 2015 (five elections). However, we also include the 1999 GE in order to calculate the extent to which continuity was important in 2002. 3 The most well-known cases are those of Mesut Yılmaz (Rize), Muhsin Yazıcıoğlu (Sivas), Ufuk Uras (İstanbul) and Fadıl Akgündüz (Siirt) whose individual attributes were important. 4 The ordered list of respective candidates is written on ballots, yet the voters cannot change the order, as they can vote for the party as a whole. 5 These elections were held every five years prior to constitutional change in 2007. 6 See Supplementary Material for a summary of the profile of candidates. 7 We ran a bivariate regression of the number of seats for each party over the previous elections. The relationship between two variables is statistically significant as expected, R-square equals 0.89, and these two variables are highly correlated at 94%. Therefore, we assume that parties are informed about this relationship between consecutive election results and set their expectations for the next election accordingly. 8 In order to have t−2 GE data for 2007 GE, we collected 1999 GE data as well. 9 We would like to thank anonymous reviewer for this point. 10 We would like to thank anonymous reviewer for this comment. 11 HDP only ran as a political party in two GEs in 2015. Its predecessors preferred to nominate independent candidates to avoid 10% threshold. Therefore, we do not have a reference to compare its results to other parties and also it disables us to use HDP observations in the regression models below. A more descriptive analysis of HDP’s electoral behaviour reveals that they are more likely to nominate independent candidates which have been stronger in previous elections. A research note on this topic will be available at author’s website. References Ayan P. ( 2010 ) ‘ Authoritarian Party Structures in Turkey: A Comparison of the Republican People’s Party and the Justice and Development Party ’, Turkish Studies , 11 , 197 – 215 . Google Scholar CrossRef Search ADS Carey J. M. , Shugart M. S. ( 1995 ) ‘ Incentives to Cultivate a Personal Vote: A Rank Ordering of Electoral Formulas ’, Electoral Studies , 14 , 417 – 439 . Google Scholar CrossRef Search ADS Ciftci S. , Yildirim T. M. ( 2017 ) ‘ Hiding Behind the Party Brand or Currying Favor with Constituents: Why Do Representatives Engage in Different Types of Constituency-Oriented Behavior? ’, Party Politics , DOI: 1354068817720438. Cox G. W. ( 1997 ) Making votes count: strategic coordination in the world’s electoral systems , Cambridge, Cambridge University Press. De Luca M. , Jones M. P. , Tula M. I. ( 2002 ) ‘ Back rooms or ballot boxes? Candidate nomination in Argentina ’, Comparative Political Studies , 35 , 413 – 436 . Google Scholar CrossRef Search ADS Field B. N. , Siavelis P. M. ( 2008 ) ‘ Candidate selection procedures in transitional polities: A research note ’, Party Politics , 14 , 620 – 639 . Google Scholar CrossRef Search ADS Fiorina M. P. ( 1977 ) ‘ An outline for a model of party choice. American Journal of Political Science ’, 21 , 601 – 625 . Gümüşçü Ş. ( 2013 , April) ‘ The Emerging Predominant Party System in Turkey ’, Government and Opposition , 48 , 223 – 244 . Google Scholar CrossRef Search ADS Güneş-Ayata A. ( 1994 ) ‘Roots and Trends of Clientelism in Turkey’. In Roniger L. , Güneş-Ayata A. (eds) Democracy, Clientelism and Civil Society , London , Lynne Rienner Publishers , pp. 49 – 63 . Hazan R. Y. , Rahat G. ( 2010 ) Democracy within parties: Candidate selection methods and their political consequences , Oxford, Oxford University Press. Hellman O. ( 2012 ) ‘ Outsourcing Candidate Selection: The Fight against Clientelism in East Asian Parties ’, Party Politics , 20 , 1 – 11 . Jones M. P. , Saiegh S. , Spiller P. T. , Tommasi M. ( 2002 ) ‘ Amateur legislators–professional politicians: The consequences of party-centered electoral rules in a federal system ’, American Journal of Political Science , 656 – 669 . Kalaycıoğlu E. ( 2015 ) ‘ Turkish Popular Presidential Elections: Deepening Legitimacy Issues and Looming Regime Change ’, South European Society and Politics , 20 , 1 – 22 . Google Scholar CrossRef Search ADS Kemahlıoğlu Ö. ( 2012 ) Agents or Bosses? Patronage and Intra-Party Politics in Argentina and Turkey , Colchester , ECPR Press . Kınıklıoğlu S. ( 2002 ) ‘The Democratic Left Party: Kapıkulu Politics Par Excellence’. In Rubin B. , Heper M (eds) Political Parties in Turkey , London, Portland, OR , Frank Cass , pp. 4 – 24 . Koop R. , Bittner A. ( 2011 ) ‘ Parachuted into Parliament: Candidate Nomination, Appointed Candidates, and Legislative Roles in Canada ’, Journal of Elections, Public Opinion & Parties , 21 , 431 – 452 . Google Scholar CrossRef Search ADS Lancaster Thomas D. ( 1986 ) ‘ Electoral structures and pork barrel politics ’, International Political Science Review , 7 , 67 – 81 . Google Scholar CrossRef Search ADS Mitchell P. ( 2000 ) ‘ Voters and Their Representatives: Electoral Institutions and Delegation in Parliamentary Democracies ’, European Journal of Political Research , 37 , 335 – 351 . Moral M. , Ozen H. , Tokdemir E. ( 2015 ) ‘ Bringing the Incumbency Advantage into Question for Proportional Representation ’, Electoral Studies , 40 , 56 – 65 . Google Scholar CrossRef Search ADS Neto O. A. , Cox G. W. ( 1997 ) ‘ Electoral Institutions, Cleavage Structures, and the Number of Parties ’, American Journal of Political Science , 41 , 149 – 174 . Google Scholar CrossRef Search ADS Norris P. , Lovenduski J. ( 1993 ) ‘ If Only More Candidates Came Forward: Supply-Side Explanations of Candidate Selection in Britain ’, British Journal of Political Science , 23 , 373 – 408 . Google Scholar CrossRef Search ADS Norris P. , Lovenduski J. ( 1995 ) Political recruitment: Gender, race and class in the British Parliament , Cambridge, Cambridge University Press. Sartori G. ( 1976 ) Parties and Party Systems: A Framework for Analysis , Cambridge Univerity Press . Sartori G. ( 2005 ) Parties and Party Systems: A Framework for Analysis , Colchester, ECPR Press . Sayarı S. ( 2011 ) ‘Clientelism and Patronage in Turkish Politics and Society’. In Birtek F. , Toprak B. (eds) In the Post-Modern Abyss of the New Politics of Islam: Assabiyah Revisited—Essays in Honor of Şerif Mardin , İstanbul , İstanbul Bilgi University Press , pp. 81–94. Sayarı S. ( 2016 ) ‘ Back to a Predominant Party System: The November 2015 Snap Election in Turkey ’, South European Society and Politics , 21 , 1 – 18 . Google Scholar CrossRef Search ADS Siavelis P. M. , Morgenstern S. ( 2008 ) ‘ Candidate recruitment and selection in Latin America: a framework for analysis ’, Latin American Politics and Society , 50 , 27 – 58 . Google Scholar CrossRef Search ADS Stokes S. C. , Dunning T. , Nazareno M. , Brusco V. ( 2013 ) Brokers, Voters, and Clientelism: The Puzzle of Distributive Politics , New York , Cambridge University Press . Google Scholar CrossRef Search ADS Strøm K. ( 1990 ) ‘ A Behavioral Theory of Competitive Political Parties ’, American Journal of Political Science , 34 , 565 – 598 . Google Scholar CrossRef Search ADS Strøm K. ( 2000 ) ‘ Delegation and Accountability in Parliamentary Democracies ’, European Journal of Political Research , 37 , 261 – 290 . Taagepera R. , Shugart M. S. ( 1989 ) Seats and Votes: The Effects and Determinants of Electoral Systems , New Haven, Yale University Press . Yildirim T. M. , Kocapınar G. , Ecevit Y. A. ( 2017 ) ‘ Staying Active and Focused? The Effect of Parliamentary Performance on Candidate Renomination and Promotion ’, Party Politics , DOI:1354068817740338. Weitz-Shapiro R. ( 2009 ) ‘Choosing Clientelism: Political Competition, Poverty, and Social Welfare Policy in Argentina’. APSA 2009 Toronto Meeting Paper. Wuthrich F. M. ( 2015 ) National Elections in Turkey: People, Politics, and the Party System , Syracuse, New York , Syracuse University Press . © The Author(s) 2018. Published by Oxford University Press on behalf of the Hansard Society; all rights reserved. For Permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Parliamentary Affairs Oxford University Press

Do Party Lists Matter? Political Party Strategies in Legislative Candidate Nominations

Parliamentary Affairs , Volume Advance Article (3) – Jan 19, 2018

Loading next page...
 
/lp/ou_press/do-party-lists-matter-political-party-strategies-in-legislative-nwP2NqcPbH
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of the Hansard Society; all rights reserved. For Permissions, please e-mail: journals.permissions@oup.com
ISSN
0031-2290
eISSN
1460-2482
D.O.I.
10.1093/pa/gsx083
Publisher site
See Article on Publisher Site

Abstract

Abstract In electoral systems with closed party lists, it is argued that the importance of central party organisation increases at the expense of individual candidates’ role in candidate nomination processes. This logic also underestimates individuals’ electoral potential and focuses on individuals’ allegiance to the leadership as the main asset for increasing their chances of being nominated. We argue that forming a party-list is a strategic decision based on the principle of furthering the interest of the party as a whole rather than rewarding individuals’ commitment to the party and is conditional on inter-party competition. We conveyed an original dataset of candidate lists for major parties in Turkey’s parliamentary elections between 1999 and 2015 and found empirical evidence for the significance of candidate lists as being used as strategic tools in inter-party electoral competition. 1. Introduction Elections constitute the primary instruments whereby political power is to some extent delegated from the populace to politicians in democratic systems. Politicians are mandated with making policies and are held accountable, at least in theory, for their actions (Strøm, 2000). However, political parties, with varying selection processes, exert considerable influence in this process via their candidate nomination decisions. Additionally, electoral institutions play an external role in shaping political parties’ behaviour. In this analysis, we focus on the impact of inter-party competition on parties’ electoral behaviour in a closed-list Proportional Representation (PR) system. 2. The puzzle Mitchell (2000) provides a typology of electoral systems based on the significance of political parties vis-à-vis the candidates. According to him, in candidate-centred electoral systems where district magnitude is low or in which voters have the chance for preference voting, a candidate’s reputation becomes more salient than when opposite conditions prevail. On the other end, there exists party-centred systems associated with closed-list PR systems. Compared with closed-list systems, in single-member district (SMD) systems and in open-list PR, candidates become more important, as personal reputation is a key factor in getting elected. However, in PR systems with closed lists, voters endorse their preferred political party’s list as a whole (e.g., Taagepera and Shugart, 1989; Sartori, 2005). According to Carey and Shugart (1995); vote pooling, type of ballot and number and type of votes cast all have a major impact on whether and the extent to which candidate reputation takes precedence over party reputation. Accordingly, closed-list PR systems—where each voter casts only one vote which contributes to the party’s list as a whole—the reputation of any given individual candidate is expected to be only marginally important. Therefore, compared with the majoritarian electoral systems and other applications of PR systems (i.e., preferential voting), in closed-list PR systems, the influence of party executives would be greater, as they have the privilege of proposing the party-lists. Thus, closed-list PR system contributes to centralisation in political parties with what may be significant consequences. Scholars have argued that this system bolstered a leader-dominated nomination process and contributes to the constant tension between the collective interest of the political party and the individual interest of a politician running for office (Fiorina, 1977; Cox, 1997). We claim that the implications of this difference between elections with closed party lists and open party lists are not always easy to determine. Electoral institutions enforce political parties and make them the ultimate decision-takers on whom to nominate, however, politics is an evolving phenomenon and impact of institutions are conditional on external social factors (Neto and Cox, 1997). In closed party list systems, as Lancaster (1986) argues, political parties become the constituency for the candidates as their re-election is critically dependent on parties’ choices rather than the voters’ preferences. Therefore, even when institutional rules restrict personal reputation among voters, political parties still have to make choices for some candidates to appear on the lists. For political parties the list of candidates is a strategic tool for fostering inter-party competition at the district level (Strøm, 1990). Parties have incentives to nominate those who are likely to mobilise the highest numbers of voters for the party list even under electoral systems with closed party lists. Therefore, even though the electoral system is party-centred, there is room for personal reputation and performance since political parties aim to maximise their electoral gains either for vote, office or policy purposes (Strøm, 1990). To investigate nomination strategies in closed-list PR systems, a case study based on a time-series panel data has some potential. In this case, potential explanatory factors of political institutions or culture and external shocks to political party system would be controlled and the impact of electoral competition between rival parties would be easier to detect. Thus, we focus on the General Elections (GE) in Turkey for the period of 2002–2015 and consider major parties’ strategies in nominating candidates. Turkey presents a unique opportunity to test Carey and Shugart’s (1995) theory, as the country’s electoral system provides all the conditions that are ordinarily expected to decrease the importance of a candidate’s personal qualities: (a) voters cannot have any impact on a party list prepared by leaders, (b) a vote for a candidate is counted first as a vote for the whole party, that is, vote pooling and (c) each voter casts only one vote. In such systems, Carey and Shugart (1995) argue the district magnitude to emerge as a factor of personal reputation whereas lower district magnitudes are associated with larger importance of personal vote. As district magnitudes in Turkey’s GE vary while parties’ nomination processes remain constant across districts, it enables us to test whether the number of seats per district play a role in the importance of personal vote. Yet, our findings provide mixed results. We found substantial evidence to suggest that in proposing party lists, party leaders refer to their parties’ district-level electoral success in the previous elections and hold candidates accountable for. In addition, the effect of the district size is rather ambiguous. Therefore, individual politicians are punished or rewarded for past electoral performance at the district level. Further, the emphasis on this factor is more pronounced when the inter-election period is short, as changes in social and economic factors are generally more limited over short than over long periods. Political parties consistently preserve continuity in their party lists in districts where they are electorally strong, which is in accord with Moral et al. (2015)’s finding of an incumbency advantage in Turkish politics. Hence, this article has two major objectives. First, by observing continuity versus change over candidate lists across five successive elections in the same political context, we will measure political party behaviour in regard to nominating candidates. In this respect, we will draw on an original dataset pertaining to the Turkish GE for the 2002–2015 period.2 Second, we will contribute to growing literature on Turkish politics by empirically investigating the determinants of the political parties’ respective choices regarding nominating candidates (Ciftci and Yildirim, 2017; Yildirim et al., 2017). Based on our results, we argue that even a centralised decision-making process incorporates district-level factors and shows the influence of inter-party competition on party lists at the district level. 3. Political parties and candidate nominations Candidate nomination is denoted as an important act of political parties (de Luca et al., 2002, p. 413) with implications for representation and electoral competition. Norris and Lovenduski (1995) answer ‘Who gets selected, and why?’ from supply and demand-side explanations. Here, supply-side explanation deals with the availability of applicants for candidacy who pursue a political career, and demand-side explanation discusses how party selectors select the candidates among these applicants by highlighting the importance of decision makers in nomination process. We concentrate on the demand side and delve into selection processes of candidates by political parties. Field and Siavelis (2008) note that in all electoral environments, whether democratic or not, the candidate selection process dramatically narrows the electorate’s choices with profound effects on representation (Mitchell, 2000). Therefore, the role of the party’s executive leadership and the strategic choices of this group have a significant impact on who will represent the citizenry in the national parliament and therefore the subsequent legislative roles in the post-election period (Koop and Bittner, 2011). More specifically, nominations also have considerable impact on individual politicians’ career paths and the mechanism of representation and significant implications for inter-party competition at the district level. Albeit regulated by same electoral rules, the size of the city also matters as in Turkey’s major cities such as İstanbul, Ankara, and İzmir, where districts are large, the connection between candidates and voters becomes uncertain, while in districts with lower magnitude such as Tunceli or Hakkari, the reputation of the candidates matters a great deal. We assume that voters in larger districts are usually under-informed about most of the candidates and mostly vote for the party with which they strongly identify. This observation can be regarded as reinforcing the conventional wisdom in the electoral politics literature: SMD systems are more likely to increase accountability between the district representative and voters. This is held to be the case because elected politicians have to make their political campaign visible to the public such that voters can demand policy choices, pork or parliamentary activity, from member states, which, in turn, protects the interests of voters and districts alike. Agency loss takes place much less in systems where candidates are more accountable to the voters via the electoral mechanism (Mitchell, 2000). However, this clear-cut division between PR and SMD plurality systems blurs with decreasing district magnitude. Therefore, in small districts, both the extent of the candidates’ popularity and the nature of their platforms and campaigns may have a significant influence on voters’ choices. This holds true for political systems, like Turkey’s, where clientelism is extensive (Güneş-Ayata, 1994; Sayarı, 2011; Kemahlıoğlu, 2012) and where exists a fierce intra-party competition for candidacy and district level inter-party competition for election. In most cases, those who capitalise on existing clientelistic ties with a constituency are more likely to be nominated compared with those who focus more on programmatic concerns over politics (Hellman, 2012). The individual candidate’s attributes, thus, increase the importance of individual candidates at the expense of the party’s leadership, as the latter is constrained in accord with the extent of a candidate’s electoral potential in the district. The recent history of Turkish politics includes instances of individual candidates who ran independently and even for another party when their own parties chose not to nominate them. Therefore, we argue that nominations in the Turkish electoral system do not fulfil expectations associated with a pure closed-list PR system wherein the traits of an individual candidate are of little consequence. Instead, we observe that individuals’ role in electoral competition is regarded as quite important. Earlier works also reveal that as district magnitude decreases, the importance of the locality increases (Moral et al., 2015).3 Hence, we argue that nominations processes provide a dynamic interaction between political parties and individual candidates and electoral systems with closed party-lists are not excluded. Norris and Lovenduski (1993)’s supply and demand framework reveals that the interaction between pool of candidates and parties’ choices matters for the nomination process. On the demand side, party selectors choose candidates based on their individual qualities, experiences or abilities. However, political parties are aware that a large proportion of the votes are received as a result of national-level appeal of political parties and this is unlikely to change conditional on the district-level candidate profiles. On the other hand, a group of individual politicians generate personal ties with the electorate and their appearance on the candidate list matters for the voters in the respective districts. On the supply side, not every party member is eager to run for candidacy. Those eager politicians usually claim that their candidacy is crucial for political parties’ success at the district level as their connection with the electorate is strong. Electoral results are good indicators for such claims, particularly in political contexts where political parties are relatively centralised and nomination decisions are made by the party leader and top cadet members. In such contexts, party lists are prepared conditional on the previous election results. In districts where previous group of candidates achieved a level of success measured in the form of number of votes or seats increased, political party leaders have less incentives to make modifications in the next electoral cycle. 3.1 The context: elections in Turkey The Turkish political system has been recognised as a parliamentary democracy since the end of military rule in 1983. Despite major problems in the country’s democratic process (Gümüşçü, 2013; Kalaycıoğlu, 2015), inter-party competition remains as an important aspect of Turkish electoral politics and the critical choices made by parties at all levels of electoral competition require further scrutiny. Elections for legislature, that is, the Türkiye Büyük Millet Meclisi (TBMM—Turkish Grand National Assembly), are held every four years by implementing closed-list PR4 with an unprecedentedly high national threshold of 10%, and D’Hondt formula is used in 85 electoral districts with varying district magnitude.5 Since the 1995 GE, the closed party list system has been used and has contributed to increased levels of party centralisation and intra-party discipline. The electoral rules in Turkey remained unchanged since 1995 GE. Yet, the party system and effective parties in the system differed tremendously between 1990s and 2000s, whereas the former is associated with an unstable multi-party system predisposed to coalitional governments while the latter is associated with a consolidated multi-party system where a single party government was more likely as two parties emerge as major competitors. Therefore, we concentrate on the latter for two reasons. First, stable party system allows us to track changes between different elections. Second data availability for 1990s is a major problem. Thus, analysing the GEs between 2002 and 2015 yields a good opportunity to focus on four major parties’ behaviour. 3.2 Candidate nomination processes in Turkey The analysis will be based on four consistent and major parties in the Turkish political party system: Adalet ve Kalkınma Partisi (AKP—Justice and Development Party), Cumhuriyet Halk Partisi (CHP—Republican People’s Party), Milliyetçi Hareket Partisi (MHP—Nationalist Movement Party) and Halkların Demokratik Partisi (HDP—People’s Democratic Party). Throughout the time period of this study, district magnitude in Turkey varied between 1 (e.g., Bayburt in 2011) and 31 (e.g., İstanbul district No. 1 and No. 3 in 2015), averaging 6.5 in 85 electoral units.6 In terms of candidacy, from the 2002 to the 2015 GE, the MHP nominated 2065 different candidates, whereas the corresponding figure was 2044 for the CHP and 1800 for the AKP. Among those who were candidates in the November 2015 GE for one of the four focal parties, only 27 (less than 1%) were nominated in all the GE from 2002 to November 2015, including prominent party members such as Kemal Kılıçdaroğlu (the CHP leader since 2011), Devlet Bahçeli (the MHP leader) and Pervin Buldan from the HDP. The number of the AKP candidates who were candidates in the five successive elections is very small, as it implemented a self-imposed three-term limit for the candidates who ran successively. In addition, 70 candidates could not be nominated in the June 2015 GE due to this term limit, yet 23 of those 70 were candidates in the November 2015 GE. 3.3 Party structure in Turkey As mentioned earlier, electoral laws alone do not explain the level of recruitment and nomination centralisation (Siavelis and Morgenstern, 2008). Parties exhibit different types of nomination processes. While for some implement inclusive mechanisms including local party primaries or national level conventions, for others, nomination is an exclusive procedure where only top party cadets have the authority to make decisions (Hazan and Rahat, 2010). Candidate selection processes both affect the characteristics and acts of elected representatives in their offices (Carey and Shugart, 1995; de Luca et al., 2002). More inclusive candidate selection processes would create better incentives for legislators or candidates to well develop their proposed policy agendas as individuals to better compete with other candidates. On the other hand, if the legislators’ re-nomination chances are under the control of local governors or local party bosses rather than the constituencies, these legislators professional career path development and specialisation in certain policy areas can be limited (Jones et al., 2002). Norris and Lovenduski (1995) classifies different recruitment process on two dimensions: having formal rules and the levels of recruitment process. We assume that the more centralised the structures are the more important priority lies with winning maximum number of seats to take part in the government formation or to become a significant party in the parliament after the elections. The influence of electoral competition over leaders is expected to increase in centralised parties as leaders have less information about the lower ranks of the party politics and their main tool for evaluation would be retrospective electoral results. Turkish case provides us an ample opportunity to control for different party structures. Because, it is well established among scholars of Turkey’s party system that parties are centralised structures in which hierarchy is institutionalised in making policy choices and nominations since electoral democracy was initiated in 1950s (Ayan, 2010; Wuthrich, 2015) and that many parties, regardless of ideological left–right positioning, employ a central control mechanism (Kınıklıoğlu, 2002). Although there are no legal regulations regarding holding primaries or fulfilling a gender quota, several left-wing parties employ such practices to some extent: The CHP held primaries in several elections, and the HDP specifies a gender quota in its by-laws, but they are as binding as party leadership is committed to these mechanisms. Given the level of party centralisation and the role of party executives in candidate nominations in Turkish politics, it is a critical choice for political parties to present a list of candidates in each district for two major reasons. First, as a tenet of closed party lists, nominations have a significant role in determining who is elected to parliament. Second, it is a way to maximise party votes, as candidates with strong social ties are more likely to successfully mobilise the electorate for parties that have nominated them. For this reason, it is important to understand the factors that shape parties’ political behaviour and analyse the extent to which parties take inter-party competition, local factors and the attributes of the candidates into consideration—all of which are affected by both socio-economic factors and competition. To understand these relationships, we now turn to a multivariate analysis of the candidate nomination strategies of Turkey’s four major political parties for the GE from 2002 to November 2015. 4. Hypotheses and data Based on the argument explained above we have derived the following hypotheses: H1: Parties are more likely to show continuity in party lists where they are electorally stronger. H2: Parties are more likely to change their lists in districts where they have faced vote loss in the previous elections. We collected data from official resources (see Supplementary Material). Our unit of analysis is the party-district. Each variable is measured at the district level for four major parties in Turkey’s party system. 4.1 Dependent variable(s): Continuity in nominations We differentiated between the candidates based on their probability of gaining office, and we created two categories of candidates. The first dependent variable is the rate of continuity of all candidates in the party list from previous elections. We compared two party lists at t−1 and t and divided the total number of recurring candidates by the district magnitude as follows: rate of continuity =Number of recurring candidatesdistrict magnitude. The second dependent variable is referred to as the rate of continuity of favourites. As parties must provide a list of candidates that is in line with the district magnitude and the electoral system is a closed party list, those in the upper part of the list are more likely to get elected than are those who appear lower in the list.7 We referred to those whose names appear in the upper part of the list as favourites. However, it was necessary to make an additional decision in order to ascertain empirically what the upper part of the list refers to. We decided that if party A won three seats in the previous elections (t−1), the candidates ranked in the top 3 spots of the list in the current elections (t) would constitute favourites. That is, candidates in the top 3 have a much better chance of re-election than do those placed lower down on the list. Therefore, we created our second dependent variable, the rate of continuity of favourites, based on this information by dividing the actual number of incumbents re-nominated as favourites by the number of seats won in the previous election: rate of continuity of favourites=Number of incumbents as favouritesnumber of seats won at (t-1). These two dependent variable ranges between 0 and 1 where 0 means all favourites/candidates changed and 1 means all favourites/candidates continued to be so in the next elections. 4.2 Independent variables: Electoral competition T−1 vote shares (%): This variable is calculated by the percentage of votes won by each political party at the district level. We expect those who were successful in the previous elections to be more likely to keep their candidates on the list because we expect parties to take electoral success into consideration when making nominations. Vote loss: This variable is measured by subtracting the percentage of the vote won by a political party at elections at t−1 from t−2. Seat loss: This variable is calculated by dividing the difference between the number of seats won at t−2 and t−1 by the number of seats won at t−2. This variable is also used to determine electoral success because parties may consider the actual number of seats gained (or lost) rather than mere percentages and re-consider their nomination strategies in response to an overall loss of seats. We expect to find that vote or seat loss would drive political parties to change party lists in the next election. Both for electoral loss and seat loss variables, positive numbers are associated with losing votes and seats, respectively, as we calculate the difference between the elections at t−2 and t−1. 4.2.1 Control variables Economic development: This variable is calculated by dividing the gross-domestic added value of a district by the total gross-domestic added value at the national level. Thus, we were able to find the weight of the district in the economic production of the entire country. We controlled for this variable, as variations in the extent of economic interaction might have a role in the relative strength of the linkage between voters and politicians. In highly developed districts, clientelistic linkages between candidates and voters are less important than in districts where pre-industrial economic ties are more common and poverty levels in constituencies are relatively high (see e.g., Weitz-Shapiro, 2009; Stokes et al., 2013). Nevertheless, we approach this control variable with great caution because electoral districts are shaped by administrative functions rather than socioeconomic factors; therefore, in highly developed districts such as large urban areas (e.g., Istanbul or Ankara), we still observe clientelistic practices in Turkish context. However, given the absence of a comparative index of clientelism across different districts, we suggest that economic development variable is still the best available proxy to control for the effects of clientelism as well as other potential mechanisms in accordance with previous literature. Large districts is measured on the basis of district magnitude. We expected that in small districts continuity in candidates would be higher as politicians’ level of interaction with constituencies increase in small districts. Urban/rural share: We adopted this indicator from the Türkiye İstatistik Kurumu (TÜİK—Turkish Statistical Institute), where the level of urbanisation is measured by dividing the urban population by the total population of a district. 5. Analysis and findings 5.1 Candidate nominations in Turkey 1999–2015 We have conducted the analysis for past elections between 1999 and 2015 to test our main hypotheses stated above.8 To investigate the determinants of the strategies pursued by political parties in making up party lists for electoral districts, we ran the multivariate generalised linear model (GLM) between 2007 and June 2015 GE. We selected GLM because our dependent variable is a rate of the continuity, which varies between 0 and 1, and running a linear regression would also predict values below 0 and above 1. The descriptive results reported in Table 1 show that political parties indicate less continuity in both candidates and favourite candidates in previous elections in comparison to 2015 November GE. Table 1 Continuity in nominations in Turkey’s GE (2007–2015) 2007 2011 2015 June 2015 November Party names Continuity in No. % No. % No. % No. % AKP Candidates 189 0.34 149 0.26 156 0.28 297 0.54 Favourites 157 0.43 114 0.33 122 0.37 183 0.70 Elected 145 0.43 108 0.33 98 0.37 188 0.59 CHP Candidates 117 0.21 74 0.13 112 0.20 321 0.58 Favourites 65 0.36 27 0.24 41 0.30 125 0.95 Elected 47 0.42 26 0.19 36 0.28 123 0.92 MHP Candidates 97 0.18 122 0.22 82 0.15 243 0.44 Favourites NA 29 0.40 25 0.47 69 0.86 Elected NA 23 0.43 23 0.29 37 0.93 HDP11 Candidates NA NA NA 236 0.43 Favourites NA NA NA 67 0.83 Elected NA NA NA 51 0.86 2007 2011 2015 June 2015 November Party names Continuity in No. % No. % No. % No. % AKP Candidates 189 0.34 149 0.26 156 0.28 297 0.54 Favourites 157 0.43 114 0.33 122 0.37 183 0.70 Elected 145 0.43 108 0.33 98 0.37 188 0.59 CHP Candidates 117 0.21 74 0.13 112 0.20 321 0.58 Favourites 65 0.36 27 0.24 41 0.30 125 0.95 Elected 47 0.42 26 0.19 36 0.28 123 0.92 MHP Candidates 97 0.18 122 0.22 82 0.15 243 0.44 Favourites NA 29 0.40 25 0.47 69 0.86 Elected NA 23 0.43 23 0.29 37 0.93 HDP11 Candidates NA NA NA 236 0.43 Favourites NA NA NA 67 0.83 Elected NA NA NA 51 0.86 Table 1 Continuity in nominations in Turkey’s GE (2007–2015) 2007 2011 2015 June 2015 November Party names Continuity in No. % No. % No. % No. % AKP Candidates 189 0.34 149 0.26 156 0.28 297 0.54 Favourites 157 0.43 114 0.33 122 0.37 183 0.70 Elected 145 0.43 108 0.33 98 0.37 188 0.59 CHP Candidates 117 0.21 74 0.13 112 0.20 321 0.58 Favourites 65 0.36 27 0.24 41 0.30 125 0.95 Elected 47 0.42 26 0.19 36 0.28 123 0.92 MHP Candidates 97 0.18 122 0.22 82 0.15 243 0.44 Favourites NA 29 0.40 25 0.47 69 0.86 Elected NA 23 0.43 23 0.29 37 0.93 HDP11 Candidates NA NA NA 236 0.43 Favourites NA NA NA 67 0.83 Elected NA NA NA 51 0.86 2007 2011 2015 June 2015 November Party names Continuity in No. % No. % No. % No. % AKP Candidates 189 0.34 149 0.26 156 0.28 297 0.54 Favourites 157 0.43 114 0.33 122 0.37 183 0.70 Elected 145 0.43 108 0.33 98 0.37 188 0.59 CHP Candidates 117 0.21 74 0.13 112 0.20 321 0.58 Favourites 65 0.36 27 0.24 41 0.30 125 0.95 Elected 47 0.42 26 0.19 36 0.28 123 0.92 MHP Candidates 97 0.18 122 0.22 82 0.15 243 0.44 Favourites NA 29 0.40 25 0.47 69 0.86 Elected NA 23 0.43 23 0.29 37 0.93 HDP11 Candidates NA NA NA 236 0.43 Favourites NA NA NA 67 0.83 Elected NA NA NA 51 0.86 This is not surprising given the fact that the inter-election period between regular elections were four years and many intra-party political factors as well as individual and voluntary factors contribute into this rate of change. However, for all parties, rate of continuity among favourite candidates was higher than rate of continuity in all candidates. Another important finding is that scores of continuity in favourites are very close to scores of continuity in elected candidates. This also points out that political parties have a very important role in deciding whom to be elected as nominating someone at the top of a party list significantly increase their chances of being elected. The results from CHP also point to the importance of party leaders in nominations. Leadership change in major Turkish political parties is uncommon. Between 2002 and 2015, MHP experienced no leadership change while change in CHP in 2011 was due to an unexpected scandalous case that overthrew the previous leader. AKP also changed its leader before 2015 elections as a result of Erdoğan’s ascendance to presidential office, however, his de jure leadership on AKP continued albeit the absence of legal ties and 2017 constitutional referendum allowed him to be back at the helm of AKP while keeping presidential post. Therefore, it is fair to state that the only leadership change in political parties in the period we cover here took place at CHP and it explains the anomalous findings for continuity in nominations in 2011 GE. While Turkish case presents an overall anomaly in lower levels of leadership turnover in comparative perspective, it allows to gauge trends that would be confounded in party systems with a lot of central leadership turnover.9 We ran the multivariate GLM for all the elections from 2002 to June 2015, the results of which are reported in Table 2. We left November 2015 elections as it was the only snap elections throughout the period covered in this paper and it generates some irregularities that will be discussed below. Table 2 reveals that the percentage of votes realised by the parties remains an important predictor for rate of continuity of candidates on the lists which supports our first hypothesis indicated earlier. Parties keep their candidate list, to a greater extent, when they are performing well at the district. On factor that strengthens this relationship is that political parties nominate members affiliated with national leadership of the party at stronghold districts and the candidate is locally known so that their re-election is ensured.10 Thus, it is even more important for smaller parties to use this strategy as they have higher proportion of prominent members when compared with potential seats. However, for our second hypothesis which emphasises the relationship between the vote loss and continuity of candidates, we have ambiguous results. Vote loss, in the whole sample turns out to be insignificant while the coefficient is negative as expected, and only significant for the MHP. We argue that this is due to the MHP’s insecure electoral insecurity to exceed 10% electoral threshold. Table 2 GLMs: determinants of continuity in party lists: from 1999 to June 2015 GE All parties AKP CHP MHP MHP (t−1) vote share 2.70*** 2.46*** 1.67*** 1.66*** 2.38*** 1.40 −0.96 4.84*** (0.26) (0.28) (0.34) (0.33) (0.6) (1.04) (1.95) (1.57) Vote loss −1.23 −6.48*** (0.78) (2.21) Seat loss −0.81*** −0.81*** −0.21 −1.45*** −0.80 (0.23) (0.23) (0.40) (0.36) (−0.95) Economic development 0.045 0.04 0.05 0.06 0.07 0.07 −0.017 (0.05) (0.05) (0.05) (0.07) (0.11) (0.17 (0.08) Large districts −0.026 −0.03 −0.04*** −0.04*** −0.04 −0.06 −0.07 0.001 (0.017) (0.01) (0.02) (0.01) (0.025) (0.04) (0.06) (0.027) Urban/rural 0.0005 0.001 −0.002 −0.001 −0.007 0.005 0.007 (0.0045) (0.005) (0.004) (0.007) (0.007) (0.011) (0.008) Constant −1.98*** −2.02*** −1.15 −7.74*** −1.70 −0.56 −0.99 −2.86*** (0.28) (0.29) (0.33) (−1.30) (0.49) (0.56) (0.94) (0.53) Number of observation 510 510 329 329 167 116 46 170 All parties AKP CHP MHP MHP (t−1) vote share 2.70*** 2.46*** 1.67*** 1.66*** 2.38*** 1.40 −0.96 4.84*** (0.26) (0.28) (0.34) (0.33) (0.6) (1.04) (1.95) (1.57) Vote loss −1.23 −6.48*** (0.78) (2.21) Seat loss −0.81*** −0.81*** −0.21 −1.45*** −0.80 (0.23) (0.23) (0.40) (0.36) (−0.95) Economic development 0.045 0.04 0.05 0.06 0.07 0.07 −0.017 (0.05) (0.05) (0.05) (0.07) (0.11) (0.17 (0.08) Large districts −0.026 −0.03 −0.04*** −0.04*** −0.04 −0.06 −0.07 0.001 (0.017) (0.01) (0.02) (0.01) (0.025) (0.04) (0.06) (0.027) Urban/rural 0.0005 0.001 −0.002 −0.001 −0.007 0.005 0.007 (0.0045) (0.005) (0.004) (0.007) (0.007) (0.011) (0.008) Constant −1.98*** −2.02*** −1.15 −7.74*** −1.70 −0.56 −0.99 −2.86*** (0.28) (0.29) (0.33) (−1.30) (0.49) (0.56) (0.94) (0.53) Number of observation 510 510 329 329 167 116 46 170 Note: Standard error in paranthesis. Table 2 GLMs: determinants of continuity in party lists: from 1999 to June 2015 GE All parties AKP CHP MHP MHP (t−1) vote share 2.70*** 2.46*** 1.67*** 1.66*** 2.38*** 1.40 −0.96 4.84*** (0.26) (0.28) (0.34) (0.33) (0.6) (1.04) (1.95) (1.57) Vote loss −1.23 −6.48*** (0.78) (2.21) Seat loss −0.81*** −0.81*** −0.21 −1.45*** −0.80 (0.23) (0.23) (0.40) (0.36) (−0.95) Economic development 0.045 0.04 0.05 0.06 0.07 0.07 −0.017 (0.05) (0.05) (0.05) (0.07) (0.11) (0.17 (0.08) Large districts −0.026 −0.03 −0.04*** −0.04*** −0.04 −0.06 −0.07 0.001 (0.017) (0.01) (0.02) (0.01) (0.025) (0.04) (0.06) (0.027) Urban/rural 0.0005 0.001 −0.002 −0.001 −0.007 0.005 0.007 (0.0045) (0.005) (0.004) (0.007) (0.007) (0.011) (0.008) Constant −1.98*** −2.02*** −1.15 −7.74*** −1.70 −0.56 −0.99 −2.86*** (0.28) (0.29) (0.33) (−1.30) (0.49) (0.56) (0.94) (0.53) Number of observation 510 510 329 329 167 116 46 170 All parties AKP CHP MHP MHP (t−1) vote share 2.70*** 2.46*** 1.67*** 1.66*** 2.38*** 1.40 −0.96 4.84*** (0.26) (0.28) (0.34) (0.33) (0.6) (1.04) (1.95) (1.57) Vote loss −1.23 −6.48*** (0.78) (2.21) Seat loss −0.81*** −0.81*** −0.21 −1.45*** −0.80 (0.23) (0.23) (0.40) (0.36) (−0.95) Economic development 0.045 0.04 0.05 0.06 0.07 0.07 −0.017 (0.05) (0.05) (0.05) (0.07) (0.11) (0.17 (0.08) Large districts −0.026 −0.03 −0.04*** −0.04*** −0.04 −0.06 −0.07 0.001 (0.017) (0.01) (0.02) (0.01) (0.025) (0.04) (0.06) (0.027) Urban/rural 0.0005 0.001 −0.002 −0.001 −0.007 0.005 0.007 (0.0045) (0.005) (0.004) (0.007) (0.007) (0.011) (0.008) Constant −1.98*** −2.02*** −1.15 −7.74*** −1.70 −0.56 −0.99 −2.86*** (0.28) (0.29) (0.33) (−1.30) (0.49) (0.56) (0.94) (0.53) Number of observation 510 510 329 329 167 116 46 170 Note: Standard error in paranthesis. On the other hand, parties’ responsiveness to the seat loss is more evident and negatively associated with the outcome of interest, as we expected. Therefore, when parties lose seats, they are more willing to change their candidates in the upcoming elections. District magnitude has negative impact on continuity of party lists while it is only significant in models when seat loss is concerned for the whole sample. Therefore, even though results are somewhat mixed, findings reveal that the larger the districts are the more likely the lists will be changed in accordance with the expectations from the pre-existing literature which reveals that incumbents have more advantage in smaller districts. All remaining control variables turn out to have insignificant effects. From these results, we conclude that inter-party competition at district level explains the continuity in party lists at different degrees. However, mostly the relationship is reverse, in other terms, the more they lose electorally, the more likely they will change their candidate lists. One noticeable finding is that when we ran the analysis separately for political parties we find finer results on how parties react differently to electoral outcomes. For example, MHP is more responsive to electoral loss as while CHP is to seat loss and AKP is more considerate of the previous electoral strength at the district. We claim that this is due to the fact that MHP risks national threshold and therefore is more focused on its vote share while CHP is rather concentrated in competing with AKP at districts to maximise its parliamentary seats. Thus, while all parties are found to be considerate of their previous electoral results, their concerns might differ. 5.2 2015: an exceptional year Turkey lived through an exceptional electoral year in 2015 as the country held two GE on 7 June and 1 November 2015. The main reason for this unprecedented electoral competition was the failure of the parties to form a governing coalition following the 7 June elections in which no single party won a parliamentary majority and subsequent coalition talks broke down. As a result, snap elections were called for 1 November. The inter-election period witnessed an increase in civil strife, coalition talks and mutual blame between the political parties accompanied by economic instability (Sayarı, 2016). For the political parties, there were few opportunities to make major changes in regard to implementing a policy position or furthering a political agenda. However, the political parties were required to form new lists of candidates in the electoral districts. Therefore, that two GEs were held in such a short period of time provided an ideal opportunity for analysis to control for temporal and socio-economic factors. As a result of electoral considerations, the political parties made several changes prior to the second GE in 2015, which were held on 1 November. Table 3 shows that compared with the other major parties, the AKP reacted very differently to the 7 June election results and kept only 54% of those on the candidate lists and 70% of their favourite candidates. While AKP does not differ on the overall continuity of the candidates from rival parties, they evaluated their June 2015 election performance as below-par and made sever changes at the upper echelons of the party list. On the other hand, the CHP, MHP and HDP made relatively smaller changes in the upper part of their respective lists even though each changed the lower part of its list to an even greater extent than did the AKP. Table 3 Continuity in nominations in the November 2015 GE Party name MPs elected in June 2015 Continuity of candidates % of continuity Continuity of favourite candidates % of elected candidates MVs elected in November 2015 MP change AKP 258 297 54.00 183 70.9 317 +59 CHP 132 320 58.19 125 94.7 134 +2 MHP 80 243 44.18 69 86.25 40 −40 HDP 80 235 42.72 67 83.75 59 −21 Party name MPs elected in June 2015 Continuity of candidates % of continuity Continuity of favourite candidates % of elected candidates MVs elected in November 2015 MP change AKP 258 297 54.00 183 70.9 317 +59 CHP 132 320 58.19 125 94.7 134 +2 MHP 80 243 44.18 69 86.25 40 −40 HDP 80 235 42.72 67 83.75 59 −21 Table 3 Continuity in nominations in the November 2015 GE Party name MPs elected in June 2015 Continuity of candidates % of continuity Continuity of favourite candidates % of elected candidates MVs elected in November 2015 MP change AKP 258 297 54.00 183 70.9 317 +59 CHP 132 320 58.19 125 94.7 134 +2 MHP 80 243 44.18 69 86.25 40 −40 HDP 80 235 42.72 67 83.75 59 −21 Party name MPs elected in June 2015 Continuity of candidates % of continuity Continuity of favourite candidates % of elected candidates MVs elected in November 2015 MP change AKP 258 297 54.00 183 70.9 317 +59 CHP 132 320 58.19 125 94.7 134 +2 MHP 80 243 44.18 69 86.25 40 −40 HDP 80 235 42.72 67 83.75 59 −21 Recall that the June 2015 elections were considered as a major failure for the AKP given that the party lost its parliamentary majority for the first time since its entry into electoral competition in the 2002 elections. One factor behind this change was the AKP’s implementation of the three-term limit specified in its by-laws, according to which no candidate can run for more than three consecutive terms on the AKP ticket. Some of the politicians who were subject to this rule in the June 2015 elections were later reinstated to the party lists in accordance with the preferences of party leaders. According to our descriptive findings relating to the November 2015 elections, it can be argued that the AKP chose to shuffle its candidates in order to recover from the losses it had experienced in the June 2015 elections. For example, in the November 2015 elections, the AKP replaced all of its June 2015 candidates with new party members in six districts (Gümüşhane, Ağrı, Iğdır, Karabük, Edirne and Hakkari). Further, if we compare the AKP’s results in the June 2015 elections with its results in the 2011 elections, we observe that the percentage of party votes decreased in all these districts by 9% in Gümüşhane, 29% in Ağrı, 17% in Iğdır, 10% in Karabük, 5% in Edirne and 7% in Hakkari. However, a decline in the AKP’s votes prevailed across multiple electoral units. In some of the electoral units, however, the AKP responded differently. For example, despite an electoral decline in constituencies including Bilecik, Düzce and Sinop, the AKP did not change its list for the November 2015 elections. Therefore, further scrutiny is required if we are to understand why parties change their candidates in some districts but not in others. Given that 2015 was an exceptional year in regard to inter-election time period of GE, it is necessary to compare the findings for the elections in this year with those for other elections. In this way, it will be possible to determine whether there is a consistent trend across previous elections. Our major motivation is to find out the reasons for this variation among political party behaviour. To investigate, we ran the multivariate GLM only for two GEs in 2015 and reached to findings below. Models on the left side of the table (I, II and III) predict continuity in candidate lists as a whole, while those on the right (IV, V and VI) predict candidates that are nominated at the upper echelons of the party list. The results from the GLM model above show that all the parties have more continuity in their party lists in districts where they have a large vote share. The t−1 vote share variable is significantly associated with continuity in regard to nominating all and favourite candidates. When we test our main hypothesis of party competition we find strong evidence that political parties considered their electoral loss, measured by vote and seat loss and altered their party lists accordingly. Given that the AKP considered June 2015 elections as an electoral failure, we added a dummy variable for the AKP and found that the AKP modified its candidate list more frequently compared with other parties in November 2015 elections. When we considered seat loss as an indicator of electoral loss, we found that parties that had lost an incumbent seat in the last elections changed their list of candidates. From the results reported in Table 4, we found that when the number of seats a party won in the election declined when compared with the seats won in the last elections, those parties were more likely to make major changes in their party lists in the next elections. This indicates that the parties consider lost seats to be an electoral failure on the part of party members on the list and nominate different members in the next election to mobilise support in an effort to regain seats lost in the previous election. Losing electoral votes also has a similar effect for political parties and they form their party lists in next elections based on previous success or failures. On the other hand, we found all the socio-economic variables used herein as a proxy for other relevant mechanisms including clientelism and political culture to be insignificant in explaining the level of change in nomination lists except district magnitude in only Model 1. In larger districts, continuity is higher and we claim that this might occur due to an effort by parties to guarantee prominent members’ re-election instead of local electoral concerns while further investigation is required. Table 4 GLM: determinants of continuity in party lists—November 2015 GE Continuity in candidate lists Continuity in favourite candidates I II III IV V VI (t−1) vote share 4.35*** 2.66*** 4.44*** 1.84 0.50*** 4.39*** (0.51) (0.59) (0.74) (1.30) (1.19) (1.48) Vote loss −2.95*** −12.36*** (1.04) (2.82) Seat loss −0.83*** −0.55** −2.13** −0.46 (0.28) (0.3) (0.80) (0.65) AKP −0.79*** −2.10*** (0.22) (0.48) Economic development −0.05 0.05 0.04 −0.08 −0.08 0.05 (0.07) (0.07) (0.07) (0.11) (0.12) (0.12) Large districts 0.06*** (−0.002) (0.004) 0.03 0.008 0.01 (0.02) (0.02) (0.02) (0.04) (0.044) (0.04) Urban/rural −0.001 −0.002 −0.003 −0.005 0.005 0.002 (0.006) (0.007) (0.007) (0.015) (0.014) (0.01) Constant −1.47*** −0.43*** −0.65 1.44 0.97 0.91 (0.40) (0.50) (0.52) (1.18) (1.08) (1.25) Number of observation 255 173 173 179 163 163 Continuity in candidate lists Continuity in favourite candidates I II III IV V VI (t−1) vote share 4.35*** 2.66*** 4.44*** 1.84 0.50*** 4.39*** (0.51) (0.59) (0.74) (1.30) (1.19) (1.48) Vote loss −2.95*** −12.36*** (1.04) (2.82) Seat loss −0.83*** −0.55** −2.13** −0.46 (0.28) (0.3) (0.80) (0.65) AKP −0.79*** −2.10*** (0.22) (0.48) Economic development −0.05 0.05 0.04 −0.08 −0.08 0.05 (0.07) (0.07) (0.07) (0.11) (0.12) (0.12) Large districts 0.06*** (−0.002) (0.004) 0.03 0.008 0.01 (0.02) (0.02) (0.02) (0.04) (0.044) (0.04) Urban/rural −0.001 −0.002 −0.003 −0.005 0.005 0.002 (0.006) (0.007) (0.007) (0.015) (0.014) (0.01) Constant −1.47*** −0.43*** −0.65 1.44 0.97 0.91 (0.40) (0.50) (0.52) (1.18) (1.08) (1.25) Number of observation 255 173 173 179 163 163 Note: Standard errors in parentheses. Table 4 GLM: determinants of continuity in party lists—November 2015 GE Continuity in candidate lists Continuity in favourite candidates I II III IV V VI (t−1) vote share 4.35*** 2.66*** 4.44*** 1.84 0.50*** 4.39*** (0.51) (0.59) (0.74) (1.30) (1.19) (1.48) Vote loss −2.95*** −12.36*** (1.04) (2.82) Seat loss −0.83*** −0.55** −2.13** −0.46 (0.28) (0.3) (0.80) (0.65) AKP −0.79*** −2.10*** (0.22) (0.48) Economic development −0.05 0.05 0.04 −0.08 −0.08 0.05 (0.07) (0.07) (0.07) (0.11) (0.12) (0.12) Large districts 0.06*** (−0.002) (0.004) 0.03 0.008 0.01 (0.02) (0.02) (0.02) (0.04) (0.044) (0.04) Urban/rural −0.001 −0.002 −0.003 −0.005 0.005 0.002 (0.006) (0.007) (0.007) (0.015) (0.014) (0.01) Constant −1.47*** −0.43*** −0.65 1.44 0.97 0.91 (0.40) (0.50) (0.52) (1.18) (1.08) (1.25) Number of observation 255 173 173 179 163 163 Continuity in candidate lists Continuity in favourite candidates I II III IV V VI (t−1) vote share 4.35*** 2.66*** 4.44*** 1.84 0.50*** 4.39*** (0.51) (0.59) (0.74) (1.30) (1.19) (1.48) Vote loss −2.95*** −12.36*** (1.04) (2.82) Seat loss −0.83*** −0.55** −2.13** −0.46 (0.28) (0.3) (0.80) (0.65) AKP −0.79*** −2.10*** (0.22) (0.48) Economic development −0.05 0.05 0.04 −0.08 −0.08 0.05 (0.07) (0.07) (0.07) (0.11) (0.12) (0.12) Large districts 0.06*** (−0.002) (0.004) 0.03 0.008 0.01 (0.02) (0.02) (0.02) (0.04) (0.044) (0.04) Urban/rural −0.001 −0.002 −0.003 −0.005 0.005 0.002 (0.006) (0.007) (0.007) (0.015) (0.014) (0.01) Constant −1.47*** −0.43*** −0.65 1.44 0.97 0.91 (0.40) (0.50) (0.52) (1.18) (1.08) (1.25) Number of observation 255 173 173 179 163 163 Note: Standard errors in parentheses. The results from the November 2015 elections show that the parties view inter-party competition as an important consideration in determining their party lists. However, as we have explained, the November 2015 elections constitute an exceptional case given that they took place only five months after the previous election. Therefore, inter-party competition might be the most important factor given that changes in other factors including urbanisation, economic development and patron–client relationships remained stable between June and November 2015 and this is clearly reflected in our hypotheses tests. 6. Conclusion In this paper, we investigated political parties’ nomination in Turkey’s parliamentary elections since 1999 by using an original dataset of individual candidates from the 1999 to the 2015 elections. We focused on the lists of candidates from each district as an indicator of the political parties’ strategic decision. Our findings support the hypotheses listed above and indicate that political parties modify their candidate lists in districts where they previously lose electoral support, whereas candidates in districts where the party is electorally stronger are continued. We found ambiguous relationship between party nominations and institutional factors (i.e., district magnitude) and found no support for socio-economic factors such as urbanisation and economic development. We argue that inter-party competition plays an important role in political parties’ decisions to propose party lists in each district. We claim that this study possesses several challenges to studies on electoral systems and party politics. First, it reveals that institutional impact on the importance of cultivation of personal vote is conditional on the degree of electoral competition. Contrary to conventional wisdom on closed party list systems which states that voters make their choices at the ballot box in accord with their party preferences, such that candidates matter less in such systems, we found that candidates do matter. We are not denying that political parties are the ultimate decision takers in making up the lists; however, we argue that candidates’ popularity at the district level is an important factor being played into party leaders’ decisions when candidate lists are being made. Political parties strategically set up candidate lists to appeal to voters particularly even when nomination processes are centralised around the party leadership. Secondly, we also claim that this mechanism also operates in countries with extensive clientelistic networks. It is generally argued that clientelism is an important factor in Turkish politics in establishing the linkage between voters and parties. Even though it is argued that in such systems parties prefer candidates who have clientelistic ties rather than those who perform better in political campaigning or policy making, we found significant results that candidates’ electoral potential matters a great deal for parties in Turkish politics. Parties prefer to keep candidates with electoral potential, whether that comes from clientelistic networks or political activity. Supplementary Data Supplementary data are available at Parliamentary Affairs online. Conflict of interest The author reports no conflict of interest. Footnotes 1 A previous version of this article was presented at the second Annual Workshop of Empirical Studies in Political Analysis in Antalya, ECPR Standing Group in Parliaments 2016 Annual Meeting in Munich and Sabancı University Brown Bag Seminars. We would like to thank H. Ege Özen, Özge Kemahlıoğlu, Ersin Kalaycıoğlu, Umman Mahir Yıldırım, Şebnem Gümüşçü, Kerim Can Kavaklı and the ESPA 2016 panel participants for their comments on an earlier version of this article. 2 We analyse the GE between 2002 and November 2015 (five elections). However, we also include the 1999 GE in order to calculate the extent to which continuity was important in 2002. 3 The most well-known cases are those of Mesut Yılmaz (Rize), Muhsin Yazıcıoğlu (Sivas), Ufuk Uras (İstanbul) and Fadıl Akgündüz (Siirt) whose individual attributes were important. 4 The ordered list of respective candidates is written on ballots, yet the voters cannot change the order, as they can vote for the party as a whole. 5 These elections were held every five years prior to constitutional change in 2007. 6 See Supplementary Material for a summary of the profile of candidates. 7 We ran a bivariate regression of the number of seats for each party over the previous elections. The relationship between two variables is statistically significant as expected, R-square equals 0.89, and these two variables are highly correlated at 94%. Therefore, we assume that parties are informed about this relationship between consecutive election results and set their expectations for the next election accordingly. 8 In order to have t−2 GE data for 2007 GE, we collected 1999 GE data as well. 9 We would like to thank anonymous reviewer for this point. 10 We would like to thank anonymous reviewer for this comment. 11 HDP only ran as a political party in two GEs in 2015. Its predecessors preferred to nominate independent candidates to avoid 10% threshold. Therefore, we do not have a reference to compare its results to other parties and also it disables us to use HDP observations in the regression models below. A more descriptive analysis of HDP’s electoral behaviour reveals that they are more likely to nominate independent candidates which have been stronger in previous elections. A research note on this topic will be available at author’s website. References Ayan P. ( 2010 ) ‘ Authoritarian Party Structures in Turkey: A Comparison of the Republican People’s Party and the Justice and Development Party ’, Turkish Studies , 11 , 197 – 215 . Google Scholar CrossRef Search ADS Carey J. M. , Shugart M. S. ( 1995 ) ‘ Incentives to Cultivate a Personal Vote: A Rank Ordering of Electoral Formulas ’, Electoral Studies , 14 , 417 – 439 . Google Scholar CrossRef Search ADS Ciftci S. , Yildirim T. M. ( 2017 ) ‘ Hiding Behind the Party Brand or Currying Favor with Constituents: Why Do Representatives Engage in Different Types of Constituency-Oriented Behavior? ’, Party Politics , DOI: 1354068817720438. Cox G. W. ( 1997 ) Making votes count: strategic coordination in the world’s electoral systems , Cambridge, Cambridge University Press. De Luca M. , Jones M. P. , Tula M. I. ( 2002 ) ‘ Back rooms or ballot boxes? Candidate nomination in Argentina ’, Comparative Political Studies , 35 , 413 – 436 . Google Scholar CrossRef Search ADS Field B. N. , Siavelis P. M. ( 2008 ) ‘ Candidate selection procedures in transitional polities: A research note ’, Party Politics , 14 , 620 – 639 . Google Scholar CrossRef Search ADS Fiorina M. P. ( 1977 ) ‘ An outline for a model of party choice. American Journal of Political Science ’, 21 , 601 – 625 . Gümüşçü Ş. ( 2013 , April) ‘ The Emerging Predominant Party System in Turkey ’, Government and Opposition , 48 , 223 – 244 . Google Scholar CrossRef Search ADS Güneş-Ayata A. ( 1994 ) ‘Roots and Trends of Clientelism in Turkey’. In Roniger L. , Güneş-Ayata A. (eds) Democracy, Clientelism and Civil Society , London , Lynne Rienner Publishers , pp. 49 – 63 . Hazan R. Y. , Rahat G. ( 2010 ) Democracy within parties: Candidate selection methods and their political consequences , Oxford, Oxford University Press. Hellman O. ( 2012 ) ‘ Outsourcing Candidate Selection: The Fight against Clientelism in East Asian Parties ’, Party Politics , 20 , 1 – 11 . Jones M. P. , Saiegh S. , Spiller P. T. , Tommasi M. ( 2002 ) ‘ Amateur legislators–professional politicians: The consequences of party-centered electoral rules in a federal system ’, American Journal of Political Science , 656 – 669 . Kalaycıoğlu E. ( 2015 ) ‘ Turkish Popular Presidential Elections: Deepening Legitimacy Issues and Looming Regime Change ’, South European Society and Politics , 20 , 1 – 22 . Google Scholar CrossRef Search ADS Kemahlıoğlu Ö. ( 2012 ) Agents or Bosses? Patronage and Intra-Party Politics in Argentina and Turkey , Colchester , ECPR Press . Kınıklıoğlu S. ( 2002 ) ‘The Democratic Left Party: Kapıkulu Politics Par Excellence’. In Rubin B. , Heper M (eds) Political Parties in Turkey , London, Portland, OR , Frank Cass , pp. 4 – 24 . Koop R. , Bittner A. ( 2011 ) ‘ Parachuted into Parliament: Candidate Nomination, Appointed Candidates, and Legislative Roles in Canada ’, Journal of Elections, Public Opinion & Parties , 21 , 431 – 452 . Google Scholar CrossRef Search ADS Lancaster Thomas D. ( 1986 ) ‘ Electoral structures and pork barrel politics ’, International Political Science Review , 7 , 67 – 81 . Google Scholar CrossRef Search ADS Mitchell P. ( 2000 ) ‘ Voters and Their Representatives: Electoral Institutions and Delegation in Parliamentary Democracies ’, European Journal of Political Research , 37 , 335 – 351 . Moral M. , Ozen H. , Tokdemir E. ( 2015 ) ‘ Bringing the Incumbency Advantage into Question for Proportional Representation ’, Electoral Studies , 40 , 56 – 65 . Google Scholar CrossRef Search ADS Neto O. A. , Cox G. W. ( 1997 ) ‘ Electoral Institutions, Cleavage Structures, and the Number of Parties ’, American Journal of Political Science , 41 , 149 – 174 . Google Scholar CrossRef Search ADS Norris P. , Lovenduski J. ( 1993 ) ‘ If Only More Candidates Came Forward: Supply-Side Explanations of Candidate Selection in Britain ’, British Journal of Political Science , 23 , 373 – 408 . Google Scholar CrossRef Search ADS Norris P. , Lovenduski J. ( 1995 ) Political recruitment: Gender, race and class in the British Parliament , Cambridge, Cambridge University Press. Sartori G. ( 1976 ) Parties and Party Systems: A Framework for Analysis , Cambridge Univerity Press . Sartori G. ( 2005 ) Parties and Party Systems: A Framework for Analysis , Colchester, ECPR Press . Sayarı S. ( 2011 ) ‘Clientelism and Patronage in Turkish Politics and Society’. In Birtek F. , Toprak B. (eds) In the Post-Modern Abyss of the New Politics of Islam: Assabiyah Revisited—Essays in Honor of Şerif Mardin , İstanbul , İstanbul Bilgi University Press , pp. 81–94. Sayarı S. ( 2016 ) ‘ Back to a Predominant Party System: The November 2015 Snap Election in Turkey ’, South European Society and Politics , 21 , 1 – 18 . Google Scholar CrossRef Search ADS Siavelis P. M. , Morgenstern S. ( 2008 ) ‘ Candidate recruitment and selection in Latin America: a framework for analysis ’, Latin American Politics and Society , 50 , 27 – 58 . Google Scholar CrossRef Search ADS Stokes S. C. , Dunning T. , Nazareno M. , Brusco V. ( 2013 ) Brokers, Voters, and Clientelism: The Puzzle of Distributive Politics , New York , Cambridge University Press . Google Scholar CrossRef Search ADS Strøm K. ( 1990 ) ‘ A Behavioral Theory of Competitive Political Parties ’, American Journal of Political Science , 34 , 565 – 598 . Google Scholar CrossRef Search ADS Strøm K. ( 2000 ) ‘ Delegation and Accountability in Parliamentary Democracies ’, European Journal of Political Research , 37 , 261 – 290 . Taagepera R. , Shugart M. S. ( 1989 ) Seats and Votes: The Effects and Determinants of Electoral Systems , New Haven, Yale University Press . Yildirim T. M. , Kocapınar G. , Ecevit Y. A. ( 2017 ) ‘ Staying Active and Focused? The Effect of Parliamentary Performance on Candidate Renomination and Promotion ’, Party Politics , DOI:1354068817740338. Weitz-Shapiro R. ( 2009 ) ‘Choosing Clientelism: Political Competition, Poverty, and Social Welfare Policy in Argentina’. APSA 2009 Toronto Meeting Paper. Wuthrich F. M. ( 2015 ) National Elections in Turkey: People, Politics, and the Party System , Syracuse, New York , Syracuse University Press . © The Author(s) 2018. Published by Oxford University Press on behalf of the Hansard Society; all rights reserved. For Permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

Parliamentary AffairsOxford University Press

Published: Jan 19, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off