Political parties and style of representation: (Maiden name: The input of politcians role-orientation: Representatives, parties and political systems) This is a work in process, please do not cite without the author s permission Eva Heida Onnudottir PhD student, political science CDSS, Mannheim University Email: eonnudot@mail.uni-mannheim.de Paper presented in the panel Roles of representation - a comparative approach at the 7th ECPR General Conference Bordeaux, Domaine Universitaire, 4-7 September Abstract This paper focuses on how party characteristics explain styles of representation emphasised within parties. Style of representation is measured at the party level as the proportion of representatives within parties who are trustees, partisans or delegates. Based on the assumption that the main difference between those three styles is the source used for decision making, where trustees use their own judgment as a source, partisans the party policy and delegates the voters, it is reasonable to assume that there are partly different incentives for each style. My findings support that one of the main explanatory factor for the proportion of trustees within parties is how often parties have been represented in government. For the proportion of partisans the parties leadership control over nomination and the socialization effect of parties seems to matter. For delegates I show that the higher the proportion of party identifiers among party voters, the higher is the proportion of delegates within parties.
Introduction The main objective of this paper is to establish how party characteristics explain a difference in emphasis on style of representation within parties. Style of representation is divided into the proportion of representatives within each party who are trustees, partisans or delegates. The main explanatory factors the proportion of candidates who are nominated by the parties leaderships, the proportion of representatives who have background as locally and/or regionally elected representatives, how often parties have been represented in government and the proportion of party identifiers among its voters. In this paper I use the term representatives interchangeably with candidates, referring to both actual and potential representatives for their respective party. In recent years there has been a renewed interest in studying both the input and output of style of representation (Blomgren and Rozenberg 2012). Input refers to what explains style of representation and output what the consequences of it are. There is a growing consensus among scholars that the Responsible Party Model (RPM) that has heavily influenced studies of representation is too simplistic (see for example Bengtsson and Wass 2011, Wessels and Giebler 2011, Valen and Narud 2007). RPM assumes that parties policies are the main factor contributing to the representational bond between the represented and parties (APSA 1950), leaving no or little room for individual representatives or voters to shape or contribute to this bond. One of the requirements of RPM to be effective is that representatives of the party should follow the party policy or in other words, they should preferably all be partisans for representation to work. This contradicts the classical notion, originally used to explain representation in the US, dividing representatives into trustees and delegates, where trustees use their own judgement to take decisions while delegates supposedly follow the voters opinion regardless of their own stand on issues (see for example Eulau et al. 1959). Examining representation in France, Converse and Pierce (1986) add the partisan role to the trustee versus delegate typology, claiming that this threefold classification is closer to capturing the practice or reality of political representation. While the trustee versus delegate role might be appropriate explaining style of representation in the US, those two are not sufficient when explaining representation in Europe where parties play a bigger role in politics compared to the US. In a recent publication Rozenberg and Blomgren (2012) argue that a threefold classification into trustees, partisans and delegates is linked to both a normative debate on representation as well as 2
a philosophical one, with the major question being how elected representatives should make decisions in modern democracies. This threefold classification is more realistic than both the requirement of RPM assuming that all representatives should be partisans and the too simple twofold classification into trustees and delegates. Those three different roles reflect different sources representatives use, or claim to use, to take decisions in their work as representatives specifically when there are conflicting opinions. The core of political representation is that there is a relational element between those who are represented and representatives (Castiglione and Warren 2006). This relation is marked with ambiguities and one could say that the political use of the term political representation has further added to its ambiguity. Defining the relationship between the electorate and voters, as acting for, gives rise to different understandings of the relationship involved reflected in representatives style of representation. One feature of political representation is that it is socially constructed and it is important to understand the context it emerges from as well as it operates within. The concept, political representation, was first attributed to legislatures as a whole in the early days of democracy and only later did it become a feature of individual representatives and parties (Pitkin 1967 via Castiglione and Warren 2006). This brings forth that representation was first, and still is, a collective (i.e. a legislature or a party) representing a collective (i.e. the nation or party voters). The representation of those collectives is carried out by individuals who operate within parties and the electorate votes for those parties and by that giving those parties a mandate to represent. Based on that political representation is about a collective representing a collective indicates that it is meaningful to analyse style of representation on the party level and how parties differ in the emphasis of its representatives (who constitute a collective) on different styles of representation. In this paper the analysis is on the party level examining how party attributes explain the difference between parties in its proportion of representatives who are trustees, partisans or delegates. Style of representation Many of the first studies on style of representation originated in the US as early as in the 1960s (see for example Dalton 1985, Eulau and Karps 1977). Eulau et al. (1959) made a distinction between the focus and style of representation. The focus refers to whom a representative 3
represents such as constituency voters, party voters or the nation as a whole. The style refers to how representatives approach their role, traditionally classified as trustees and delegates (see for example Eulau et al. 1959). In practice the focus and twofold classification into trustees and delegates is closely related, with trustees more prone to consider themselves to represent the nation as a whole and delegates to focus on specific group(s) (see for example Bengtsson and Wass 2011, Wessels and Giebler 2011, Eulau et al. 1959). Because of seemingly limited impact of style of representation on the behaviour of individual representatives, studies on role-orientation went out of fashion in the 80s, but are now appearing again on the academic agenda (see for example Blomgren and Rozenberg 2012, Bengtsson and Wass 2011, Wessels and Giebler 2011). However, it is apparent from both the early literature as well as recent publications that there is little consensus about how to define style of representation, what terms to use, what explains it and what are its consequences. While the trustee versus delegate typology, sometimes adding the partisan as a role, has been quite dominant, many authors have proposed other terms and definitions such as purposive roles (Wahlke et al. 1962), position and preference roles (Searing 1994) and policy representation and interest representation (Thomassen and Esaiasson 2006) - only to name a few examples. Rozenberg and Blomgren s (2012) distinction between legislative and representation roles is quite useful and clarifies the subject; where legislative roles are concerned with how representatives organize their work in the legislative while representational roles are about whom to represent as the trustee, delegate and partisan roles do reflect. Another important distinction is between the output and input of style of representation, the former referring to the consequences of it and the latter to what explains it. Only if style of representation explains a difference in its output it becomes meaningful to study what explains its input. While the analysis in this paper is about the input, I first discuss research and theories about the consequences of different styles of representation in order to establish the relevance of examining what explains styles of representation. The output of style of representation When analysing representatives style of representation it is important not to assume that if there are behavioural consequences, that the same or similar behaviour can be expected under all circumstances. Even if representatives are for example classified as trustees that does not mean 4
that they will never follow the party policy or voters cues in their roll-call vote or other work as a representatives. It has been established that there are numerous other factors than style of representation that explains how they make decisions and vote in the parliament, as for example party discipline (Kristinsson 2011), the limited number of issues representatives can be experts on and not all issues are important to their constituents (Andeweg 2012). Even if role-orientation lacks explanatory power when studying legislative behaviour, quite often operationalized as their roll-call vote (see for example Converse and Pierce 1986, Kuklinski and Elling 1977) it could very well explain other behaviour of MPs. Style of representation might for example explain how MPs approach their work in parliamentary committees or question times in parliament and as pointed out by Andeweg (2012) their interaction and attitudes about voters. Esaiasson and Holmberg (1996) use the terms bottom-up and top-down representation about the different mechanism parties use to approach their voters. Top-down representation, which they show to be dominant in the Swedish parliament, reflects elite driven party politics where parties seek a mandate from voters to implement the policy preferences of the party. Bottom-up representation is about incorporating the policy preferences of voters into the party policy. Using those terms Andeweg (2012) shows that bottom-up representatives in the Netherlands are in more contact with voters while top-down representatives have more frequent contact with ministers and officials to discuss the problems of individual citizens. Using the terms trustees and delegates, adding the role of politicos, which is between the trustee and the delegate role (sometimes follow voters and sometimes use own judgement), he finds that delegates are in less contact with voters and more cynical about them. These findings are counter to what is expected, if it is assumed that delegates should have a more favourable view about voters compared to trustees and politicos. Andeweg suggests that these contradictionary findings could be a problem with the typology, dividing representatives into trustees, delegates and politicos; that those roles are a gross over simplification which bears little relevance to reality. A step forward is to recognize that representatives take on different roles in different circumstances and regarding different issues. In a forthcoming paper Onnudottir establishes that style of representation on the party level does explain policy congruence between parties and its voters. She shows that parties with a high number of partisans have lower policy congruence with party voters compared to parties with a high number of trustees who have higher policy congruence with its voters. Her findings 5
contradict the Responsible Party Model (RPM) which assumes that representatives sticking to the party policy is the most optimal way to represent the voters of the party; who should have according to RPM, have voted for the party that is closest to their own policy preferences. She suggests that there is possible a bidirectional relationship between style of representation and policy congruence. Parties with a high number of competent trustees could encourage voters to trust them and for that reason party voters are closer to them in their policy preferences. On the voters side it could be that if their policy preferences are already close to the parties that could encourage representatives to take on the trustee role because they are already close to the parties potential voters and have to establish that they are sufficiently skilled to solve the policy problems they are faced with. Onnudottir claims that in this two way relationship parties weigh more heavily as they are more often the leading actor in the partnership between parties and voters (Holmberg 2011). The input of style of representation An important feature of style of representation using the distinction between trustees, partisans and delegates is that those are about how representatives think about their representational role and what source(s) they use (or claim to use) to take decisions. Under the trustee role the source for decision making is the representative himself, under the partisan role it is the party policy and under the delegate role the source is the voters. Considering that the main difference between those roles is the source used it is reasonable to assume that there are different factors linked to each source that encourage different roles. If individual representatives style of representation could only be explained by their personal background, such as their education or personality that would mean that it only reflects their personal characteristics and could not be used as a dynamic analytic tool to study how political context explains style of representation (Jewell 1970 via Wessels 1999). If, as has been shown, that institutional factors, with features of the electoral system as the most common explanatory factors (see for example Esaiasson and Heidar 2000, Wessels 1999) explains style of representation, the importance of understanding how the political context affects or encourages certain style becomes meaningful to better understand the representational process. Even if the effect of the institutional setting of the electoral system on style of representation has received considerable attention, those studies quite often either include one 6
country or only a handful of countries, leading to that the effect of the electoral system can only be speculated on. However, from this there are a few exceptions. Using data from the European Election Studies (EES) Wessels (1999) finds that the smaller the district magnitude the more both European and national MPs focus on their constituency. Farrell and Scully (2010) show that the more open the electoral systems are, in the sense of a greater flexibility for voters to influence the placement of candidates on the party list (ballot-structure), the more prone elected representatives in the European Parliament (EP) are to focus on the constituency. More recently, Wessels and Giebler (2011) find that the higher the chance candidates for the EP consider themselves to have of getting elected the less likely they are to be partisans, indicating that the more secure the candidates election is, the less stronghold do the parties hold over them. Under the rational approach, Strøm (2012) casts style of representation as strategies representatives use to reach their political goals, whether that is (re-)selection on the party list, (re-)election or to acquire or maintain a party office. These strategies are conditioned by contextual factors. The main contextual factors discussed by Strøm are the electoral system and the parties representatives operate within. The most obvious party characteristics as a contextual effect are its control over the nomination of its candidates and the allocation of party and parliamentary offices. The greater the control of a party over those processes, the more likely it would be for representatives to adopt the partisan style - if it increases their chances of advancement for and within the party. Taking this up to the party level, the more control the parties leaderships have over who is nominated, the higher the proportion of partisans should be within the parties. This leads to my first hypothesis: H1: The higher the proportion of party candidates who are nominated by the parties leaderships, the higher is the proportion partisans within parties. Zittel (2012) adopts Strøm s rational approach explaining style of representation. Using the German Candidate Survey from 2005, his main contextual factors are party competition, party socialisation and the mode of the election (elected via party list or in a single seat district). Zittel finds no support for that party socialisation encourages representatives to take on the partisan role, but he finds that representatives who consider themselves to have a fair chance of winning and are elected in single member districts are more inclined to consider themselves to represent 7
their district, while representatives who consider it unlikely to win are more inclined to be partisans. This indicates that representatives style of representation is motivated by strategic factors that enhance their political career, such as their chance of a (re-) selection and (re-) election, and questions the role of parties in socializing their members. Zittel operationalizes party socialisation on the individual level as years of party membership, party employment in years and for how many years representatives have held a local party office or a regional party office. Finding no support for the effect of party socialisation through these measurements, but finding that younger representatives are more leaning towards the partisan role could signal in Zittel s view increasing professionalization of politics, with young politicians subscribing to the partisan role as the best strategic choice for them to advance their careers. This can be taken a step further as implied by Zittel, that once they have established themselves as politicians, they have more flexibility from their party and security to adopt other representational roles, such as the trustee role. Holding a party office is far from the only way to enhance a political career. Parties socialisation effect might manifest itself through the support needed when competing for a public office on the local and/or regional level. Political careers in public office generally take off at the local and/or regional level. It is rational for representatives on the local and/or regional level to gain and maintain the support of party elites within their districts (Zittel 2012) to enhance their political career. For that reason it can be assumed that those who have been elected on those lower political levels are more likely to be partisans. Based on this I test whether parties who have a high proportion of representatives who have been elected on the local and/or regional level are more likely to have a higher proportion of partisans: H2: The higher the proportion of party candidates who have been elected officials on the local/and or regional level, the higher is the proportion of partisans within parties. As already mentioned, not many studies have been published using party characteristics as explanatory factors for style of representation. Apart from the abovementioned hypothesis about parties leaderships control over nomination and party socialisation based on recent publications, I suggest that the frequency of parties representation in government explains the proportion of trustees within parties. I assume that representatives of government parties are more likely to 8
consider themselves to represent the nation as a whole because presumable the government works in the interest of all citizens. It has been established that trustees are more likely to have a nation-wide focus about whom they represent (see for example Wessels and Giebler 2011). If the nation-wide focus goes together with the trustee role then: H3: The more often parties have been represented in government the higher is the proportion of trustees within the parties. The last hypothesis considers the effect of party voters on the proportion of delegates. Assuming that the source for decision making is the voters in the delegate case, it can be argued that one of the main incentives for that role originates stems from the voters side. Strøm (2012) argues that the more a representative is dependent on his constituents for a (re)-election, as feature of the electoral system and party competition, the more likely it is he will adopt a role that conforms to the expectations of his constituents. Which roles he would adapt to under those circumstances could depend on the homogeneity of the constituency. The more heterogeneous the policy preferences of constituency voters are, the trustee role might be more fruitful, as it will be harder to advocate the delegate role in a constituency that is highly diverse. Similar, a delegate role could be more useful the more homogeneity there is within the constituency. While I cannot test for the homogeneity of the policy preferences within constituencies, I can test whether the proportion of party identifiers for each party does explain style of representation, assuming that party identification reflects homogeneity of party identifiers for a given party. H4: The higher the proportion of party voters who identify with a party, the higher the proportion of delegates within parties. Research design Data and selection of countries Data on style of representation, the proportion of candidates nominated by parties leaderships and the proportion of candidates who have background as elected representatives on the regional and/or local level is from the Comparative Candidate Survey (CCS). I kindly remind the reader 9
that in I use the term representatives interchangeably with candidates. The number of countries and national elections in the CCS data who include the variables of interest are the following nine: Belgium 2007, Germany 2009, Greece 2007, Hungary 2010, Iceland 2009, Ireland 2007, Netherlands 2006, Portugal 2009 and Sweden 2010, covering in total 58 parties. In countries where legislatives are bicameral only representatives running for the lower house are included in the CCS and the research is therefore limited to those 1. Information about parties representation in government is from the Parliament and Government Composition Database (ParlGov). Three sources are used for the proportion of party identifiers for each party; the Comparative Study of Electoral Systems (CSES), the European Election Study (EES) and the Irish National Election Study (INES). Response variables: Styles of representation In the CCS data there are three items that reflect representatives style of representation as trustees, partisans and delegates. Those three are questions are about how an MP should vote in parliament if there are different opinions between: 1. The party position and his/her party voters opinion. 2. Constituency voters opinion and MP s opinion. 3. Party position and MP s opinion. I refer both to constituency voters and his/her party voters as voters 2. To classify representatives style of representation into trustees, partisans and delegates I use two out of the three questions for each group. Those who say that the MP should vote according to his own 1 This applies to Belgium 2007. 2 There is a difference in terminology between questions 1 and 2, when asking voters opinions. The response category to the first question refers to the MP s own party voters, and the second to constituency voters. However both are contrasted with either the party position or the MP s own opinion. Even if I cannot exclude the possibility that representatives might have replied differently if the contrast had been the same on the voters side on both questions, I assume that the responses reflect a difference between a bottom-up process as the role of delegate assumes, and a top-down process inherent in the role of partisans and trustees. Based on that I take those together who choose voters in question 1 and 2 as delegates and contrast them with partisans and trustees. 10
opinion when contrasted with voters and party in items 2 and 3 are trustees, those who name the party in items 1 and 3 are partisans and those who say that the MP should follow the voters view in items 1 and 2 are delegates. The representatives who cannot be categorized according to this rule are coded as non-classifiable. Table 1 lists the proportion of representatives classified under each role for the 58 parties included in this study. It is notable that in 51 out of the 58 parties less than 15% of the candidates are non-classifiable and 53 of them less than 20%. The systematic low proportion of nonclassifiables indicates the there is a systematic component in representatives replies that can be used to categorize them as trustees, partisans or delegates. At the bottom of the table, the total proportions for each style are shown firstly for the pooled un-weighted data, secondly weighted by country and thirdly by party. In the analysis representatives replies are not weighted since the unit of analysis is at the party level. For party measures on style of representation, I use the proportion of representatives within parties who are; 1) trustees, 2) partisans or 3) delegates. Representatives that are non-classifiable are included in the calculation for the proportion of each style group within parties, but not analysed any further. An example of party measures for the Socialist Party in Belgium is 34.9 for the proportion of trustees, 37.2 for the proportion of partisans and 14.0 for the proportion of delegates. 11
Table 1. Proportion of partisans, delegates, trustees and non-classifiable within parties Belgium Trustees Partisans Delegates Nonclassifables Socialist Party (PS) 34.9% 37.2% 14.0% 14.0% 43 Reformist Movement - (MR) 32.6% 28.3% 13.0% 26.1% 46 Humanist Democratic Center (CDH) 33.3% 44.4% 8.3% 13.9% 36 Ecolo 27.3% 47.7% 2.3% 22.7% 44 National Front (FN) 20.8% 25.0% 50.0% 4.2% 24 Christian Democratic and Flemish (CD&V) 33.3% 41.7% 14.6% 10.4% 48 Soicalist Party. Different (SP.A) 28.6% 57.1% 14.3% 0.0% 35 Spirit 25.0% 62.5% 12.5% 0.0% 8 Open VLD (Flemish Liberals and Democrats) 30.0% 30.0% 30.0% 10.0% 50 Flemist Interest (Vlaams Belang - VB) 29.4% 41.2% 20.6% 8.8% 34 Green! (Groen) 48.2% 32.1% 12.5% 7.1% 56 Germany Social Democratic Party (SPD) 56.3% 19.9% 9.3% 14.6% 151 Christian Democratic Union (CDU) 72.7% 11.5% 9.4% 6.5% 139 Christian Social Union (CSU) 87.5% 0.0% 4.2% 8.3% 24 Free Democratic Party (FDP) 77.6% 6.3% 5.6% 10.5% 143 Alliance 90 / Greens 79.5% 4.6% 7.9% 7.9% 151 Left Party 51.8% 13.1% 23.4% 11.7% 137 Greece New Democracy 23.1% 47.4% 20.5% 9.0% 78 Panhellenic Socialist Movement (PASOK) 40.2% 27.1% 24.3% 8.4% 107 Hungary Fidesz / KDNP (Hungarian Civic Union / Christian Democratic Peoples Party) 14.3% 27.2% 40.8% 17.7% 147 MSZP (Hungarian Socialist Party) 15.4% 46.2% 26.9% 11.5% 52 Jobbik 14.5% 9.1% 73.6% 2.7% 110 LMP (Politics Can be Different) 38.8% 7.5% 41.8% 11.9% 67 Iceland Social Democratic Alliance (SDA) 58.1% 10.8% 16.2% 14.9% 74 Progressive Party (PP) 61.5% 16.9% 7.7% 13.8% 65 Independence Party (IP) 73.2% 8.9% 7.1% 10.7% 56 Left Green Movement (LG) 52.9% 13.2% 17.6% 16.2% 68 Civic Movement (CM) 63.5% 6.3% 25.4% 4.8% 63 N 12
Table 1 continued Ireland Trustees Partisans Delegates Nonclassifables Fianna Fáil (FF) 14.3% 57.1% 7.1% 21.4% 42 Fine Gael (FG) 30.6% 44.4% 8.3% 16.7% 36 Labour 17.6% 47.1% 23.5% 11.8% 17 Green Party 28.6% 53.6% 14.3% 3.6% 28 Progressive Democrats (PD) 11.1% 44.4% 11.1% 33.3% 9 Sinn Féin (SF) 0.0% 66.7% 16.7% 16.7% 12 Netherlands Christian Democratic Appeal (CDA) 30.0% 56.7% 0.0% 13.3% 30 PvdA Labour Party 47.6% 33.3% 9.5% 9.5% 21 Peoples Party for Freedom and Democracry (VVD) 46.2% 30.8% 15.4% 7.7% 26 Green Left (GroenLinks) 72.7% 27.3% 0.0% 0.0% 11 Socialist Party (Socialistische Partij - SP) 28.0% 64.0% 0.0% 8.0% 25 Democrats 66 (Democraten 66) 72.2% 22.2% 0.0% 5.6% 18 Christian Union (ChristenUnie - CU) (GPV, RPF) 33.3% 55.6% 0.0% 11.1% 9 Political Reformed Party (SGP) 27.3% 63.6% 9.1% 0.0% 11 Partij vd Dieren 27.3% 36.4% 9.1% 27.3% 11 Portugal Left Bloc (BE) 39.0% 12.2% 39.0% 9.8% 41 Peoples Party (CDS-PP) 33.3% 18.5% 37.0% 11.1% 54 Democratic Unity Coalition (CDU) 20.0% 52.0% 28.0% 0.0% 25 Social Democratic (PSD) 42.9% 26.5% 22.4% 8.2% 49 Socialist (PS) 52.0% 24.0% 16.0% 8.0% 25 Sweden Center Party 37.8% 38.8% 16.3% 7.1% 196 Liberal Party 38.8% 26.5% 21.8% 12.9% 147 Christian Democrats 38.5% 36.5% 15.1% 9.9% 192 Green Party 42.1% 21.8% 27.4% 8.6% 197 Moderate party 24.6% 46.9% 17.6% 10.9% 256 Social Democrats 24.6% 53.1% 12.5% 9.8% 256 Sweden Democrats 24.0% 60.0% 12.0% 4.0% 25 Left Party 34.2% 39.0% 16.6% 10.2% 187 Total 40.6% 29.4% 19.5% 10.5% 4065 Weighted by country 38.1% 30.8% 18.1% 9.7% Weighted by party 37.9% 33.2% 18.3% 10.6% N 13
Explanatory variables: Party characteristics The study makes use of four explanatory variables: the proportion of party candidates who are nominated by the parties leaderships, the proportion of candidates who have background as elected members of local and/or regional MPs, how often the parties have been represented in government and the proportion of party identifiers among their voters. For a measure on parties leaderships control over who is nominated I use data from the CCS. Respondents were asked who made the decision about their nomination as candidates and the response categories are; a) Voters at large b) Voters of my party. c) Members of my party. d) A party delegate conference. e) Party leadership 3. For a measure on the party leaderships control over who is nominated, I contrast parties leaderships control with other selectors, whether those are a party delegate conference, party members, party voters, voters at large or other selectors. For parties socialisation effect I use the proportion of representatives of a given party who have been elected for local and/or regional office. Data from ParlGov is used to construct a measure on parties representation in government. This measure is the proportion of days the parties have been represented in government in the past four electoral terms preceding the election under study in each country. Information about the proportion of party identifiers among the voters of each party are from the third wave of the Comparative Study of Political Systems (CSES) for Iceland, Germany, Netherlands and Portugal, from European Election Study 2009 (EES) study for Belgium, Greece, Hungary and Sweden, from and from the INES (Irish National Election Study) for Ireland. For this measure I use the proportion of party identifiers among actual (did vote for the party) or 3 Not all nine countries used all the five categories. For example in the German data the options were between a delegate conference either within constituencies or nation-wide both resulting in the same code for all parties (a party delegate conference). 14
potential (intends to vote for the party) 4 voters of the party assuming that the effect of party identification depends on the base of secure party voters. Control variables Earlier research has shown that candidate centred politics in comparison with party centred politics seem to encourage the delegate role (see for example Farrell and Scully 2010). This is quite often operationalized as a distinction between majoritarian (candidate centred) and proportional (party centred) systems. As the countries included in this study all make use of different versions of a proportional electoral system I cannot use this traditional distinction to control for setting of the electoral system. Moreover, this twofold classification into majoritarian and proportional systems is probably too simplistic as it does not capture the different levels of proportion systems, some of them being more proportional compared to others. A different way would be to use the district magnitude of constituencies, where a higher magnitude reflects in general a more proportional system. Unfortunately there is not information about candidates constituency for all nine countries in the CCS data and for that reason I do not use constituencies district magnitude as a measure. Instead I use the number of effective parties within each country, with lower number indicating a less proportional system compared to a higher number. For the number of effective parties, I use the relative seat share of parties in the parliament. 5 At the end I control for Hungary as a former member of the Eastern European communist regime. The modern party system in Hungary is younger compared to the other countries in included in this study, and it is still debated whether it has stabilized itself or not (see for example Róbert and Papp 2012, Zsolt 2006). For that reason it is possible that the party indicators used in this study have less or different impact in Hungary on style of representation compared to other countries included in this study. 4 In the data from the CSES and INES (Iceland, Ireland, Germany, Netherlands and Portugal) the proportion of party identifiers is based on the reported vote (actual voters). The EES data (Belgium, Greece and Hungary) is from 2009, two years after the general elections made use of in the CCS in Belgium and Greece (2007) and one year before the elections in Hungary and Sweden (2010). For Belgium and Greece I use the reported vote from the last national election (in both instances the 2007 elections) and for Hungary and Sweden I use the vote intention in the next national election (in both instances the 2010 elections). 5 Formula used is: Number of effective parliamentary parties=1/ ((proportion of seats in the parliament)2). 15
Data analysis In order to examine how parties leaderships control over nomination, the proportion of party representatives who have experience as elected officials, how often the parties have been represented in government and the proportion of party voters who identify with the party explains style of representation I use Ordinary Least Square (OLS) regression. The response variables are the proportion of representatives within parties who are trustees, partisans or delegates. Apparently, there is a potential small-n problem, with only 58 parties and the three proportions for style of representation are not independent of each other, as a higher proportion in one style group is followed by lower proportions in the other two groups. For those reasons, I run three OLS regressions in six steps, one for each of the three style groups adding one variable (altogether six variables) at the time. In my discussion I focus on significant parameters. It should be kept in mind that the parties included are not chosen randomly and for that reason the significance levels are only meaningful as indicators about the strength of the relationship between the explanatory variables and the response variable among the 58 parties included. My first hypothesis (H1) that the higher the proportion of candidates who are nominated by the parties leaderships, the higher is the proportion of partisans is supported. However including only parties leadership control over nomination as a single explanatory variable in the first step with the proportion of partisans as the response variable, the effect seems weak and it is non-significant. What weighs more heavily is the proportion of their candidates who have experience as elected MPs at the local and /or regional level, while other party characteristics do not explain the proportion of partisans within parties. It is notable that the effect of parties leaderships control over nomination is positive for both the proportion of partisans and delegates once all explanatory have been entered into the calculation, while it is negative for the proportion of trustees. This indicates that the less control parties leaderships have over the nomination, the greater flexibility its representatives, as a collective, have to adopt the trustee role. The relationship between party socialisation and the proportion of partisans is as expected (H2); the higher the proportion of representatives that have a background in local and/or regional 16
Table 2. Determinants of style of representation. % of trustees within parties % of partisans within parties % of delegates within parties % of party representatives who: Are nominated by the party leadership Have been represented in local and or/regional politics Standardized Betas Standardized Betas Standardized Betas -.31* -.33* -.35** -.35** -.47** -.44**.19.24*.25*.28*.27+.28+.21.17.20.13.30*.26* -.16 -.21 -.21 -.21 -.28*.45***.48***.52***.52***.51*** -.38** -.31* -.39** -.39** -.30** Party variables Representation in government.24+.24+.24+.24+ -.12 -.10 -.10 -.10 -.28* -.31** -.30** -.30** Proportion of party identifiers (among party voters) -.01 -.03.06 -.14 -.15 -.13.31*.34**.22+ Country variables Number of effective parties.27+.13.02 -.01 -.37** -.18 Hungary -.27+ -.04.37** Intercept, p value:.000.000.000.000.000.000.000.149.128.060.134.136.000.000.000.000.000.000 R 2:.10.12.18.18.23.28.04.24.25.27.27.27.04.18.26.34.45.54 Change in R2 significant, p value:.017.029.063.950.053.067.145.001.336.256.898.770.119.004.026.012.003.002 Note: Response variables are three; the proportion of candidates within each party who are 1) trustees, 2) partisans and 3) delegates. Significance levels: +p<0.1; *p<0.05; **p<0.01; ***p<0.001. N=58 parties 17
politics the higher is the proportion of partisans. Examining this for the other two style groups, the direction is the opposite for the proportion of delegates (the higher the proportion of representatives who have background in local/and or regional politics, the lower is the proportion of delegates) but seems to have no or a weak connection with the proportion of trustees as it only becomes significant after all explanatory variables have been entered into the calculation. Third hypothesis (H3) is supported; the more often a party has been represented in government the higher is the proportion of trustees within parties. This supports the assumption that representatives of government parties are more likely to consider themselves to represent the nation as a whole as is inherent in the trustee role. The effect of representation in government has an opposite effect on the proportion of delegates; the more often parties have been represented in government the lower is the proportion of delegates. For the proportion of partisans, the relation is non-significant. The fourth and final hypothesis (H4), that the higher the proportion of party identifiers among party voters the higher is the proportion of delegates, is supported as well. It could indicate that representatives of parties who have a high base of party identifiers among its voters are more inclined, as a collective, to promote the delegate role. This is possible due to the homogeneity of party voters assuming party identifiers of the same party are similar among them; and that this homogeneity encourages representatives to promote the delegate role as a strategic way of maximize the parties vote share. For the other two style groups, trustees and partisans, the proportion of party identifiers is non-significant. The explanatory variables used here explain the highest variance when examining the proportion of delegates with R 2 as high as.54 in the full model, compared to R 2.28 and.27 for the proportion of trustees and partisans. Discussion In this research I have established that the context of political parties does matter for style of representation emphasised within parties. I show that there are partly different incentives for the partisan, trustee and delegate roles. The subject of my first two hypotheses, that the higher the proportions of candidates who are nominated by the parties leaderships and who have background in local and/or regional politics, the higher is the proportion of partisans within parties, are both supported. Out of those two, the background in local and/or regional politics 18
seems to weigh more heavily compared to parties leaderships control over nomination at least as parties leaderships control over nomination is measured in this study. For the proportion of trustees within parties my third hypothesis, that the more often parties have been represented in government the higher is the proportion of trustees, is supported as well. This could indicate that office seeking parties (those who are more often represented in government) provide incentives to adopt the trustee role. This is perhaps so because government status encourages a nation-wide focus of representation which is inherited in the trustee role and that representatives of those parties aim for establishing themselves as sufficiently skilled trustees to deal with the policy problems they are faced with. For parties with a high proportion of delegates, the proportion of party identifiers among its voters seems to matter. Supporting my fourth hypothesis I show that the higher the proportion of party identifiers parties have, the higher is the proportion of delegates, while this does not matter for the two other style groups. It is notable when comparing the final models (who include all explanatory variables) that all four explanatory factors, parties leaderships control over nomination, background in local and/or regional politics, representation in government and party identification have a significant relation to the proportion of delegates while three two out of four explain the proportion of trustees and two out of four the proportion of partisans. As already discussed I assume that there are partly different incentives for each style, depending on the source for decision making inherent in each style. My results confirm that. Parties leaderships control over nomination and party socialisation (background in local and/or regional politics) seem to have an effect on all three style groups, representation in government matters for the proportion of trustees and delegates and the proportion of party identifiers only connects to the proportion of delegates. The assumption that the features of the source used for decision making provides different explanatory factors for each role is easily argued for partisans and delegates. Using the party as a source for decision making as partisans, it is reasonable to argue that the attributes of the parties are the main explanatory factors. Using the voters as a source in the case of delegates, it is reasonable to argue that the features of party voters are important. This assumption is harder to apply to the role of the trustee because the source is representatives themselves. However, it becomes meaningful when comparing what encourages the trustee roles with the other two roles. Contrasting trustees with partisans, the less control the parties 19
leaderships have over the nomination the greater is the flexibility for representatives of a given party to adopt the trustee role. In this contrast the source is important in the sense that it both explains the role of the partisan and the trustee, but with opposite effects. A similar argument can be made when contrasting trustees with delegates. If the voters as a source are important for delegates and not for trustees it is reasonable to assume the proportion of party identifiers has a positive relation with the proportion of delegates, but non-significant with the proportion of trustees. Again the source delegates use (or claim to use) is of importance in this contrast. While I do recognize the tentative character of my argument, I do believe that it is worthwhile to investigate further how the source for decision making does explain different styles of representation emphasised within each party. This I believe has to be done within the context of party characteristics as is done here. Moreover, it is worthwhile to take the research further and examine how the effect of different party characteristics adding for example party discipline, interact with the institutional setting of the electoral system such as the district magnitude and ballot structure, in their effect on styles of representation emphasised within parties. 20
Bibliography Andeweg, R. B. (2012). The consequences of representatives' role orientations: attitudes, behaviour, perceptions. In Blomgren, M. and Rozenberg, O., Parliamentary Roles in Modern Legislatures (p. 66-84). Routledge: Oxon. APSA. (1950). Toward a More Responsible Two-Party System: A Report of the Committee on Political Parties. 44(3). Bengtsson, Å. and Wass, H. (2011). The Representatives Roles of MPs: A Citizen Perspective. Scandinavian Political Studies (2), 143-167. Blomgreen, M. and Rozenberg, O. (2012). Parliamentary Roles in Modern Legislatures. New York: Routledge. Blomgren, M. and Rozenberg, O. (2012a). Bringing parliamentary roles back in. In Blomgren, M. and Rozenberg, O., Parliamentary Roles in Modern Legislatures (p. 211-230). Oxon: Routledge. Blomgren, M. and Rozenberg, O. (2012b). Introduction. In Blomgren, M. and Rozenberg, O., Parliamentary Roles in Modern Legislatures (p. 1-7). Oxon: Routledge. Blomgren, M. and Rozenberg, O. (2012c). Legislative roles and legislative studies: The neoinstitutional turning point? In Blomgren, M. and Rozenberg, O., Parliamentary Roles in Modern Legislatures (p. 8-36). Oxon: Routledge. Castiglione, D. and Warren, M. E. (2006). Rethinking democratic representation: Eight theoretical issues. Paper presented at the conference: Rethinking Democratic Representation" Centre for the Study of Democratic Institutions at the University of British Columbia, 18-19 May. Comparative Candidates Survey (CCS). (n.d.). Retrieved 2013, 20 August from http://www.comparativecandidates.org/ Comparative Study of Electoral Systems (CSES). (n.d.). Retrieved 2013, 1 May from www.cses.org Converse, P. E. and Pierce, R. (1986). Political Representation in France. Cambridge: The Belknap Press of Harvard University. Dalton, R. J. (1985). Political parties and political representation: Party supporters and party elites in nine nations. Comparative Political Studies, 18(3), 267-299. 21
Enyedi, Z. (2006). Stability in the shadow of chaos: The Hungarian Party system in 2006. Paper presented at the conference: Parliamentary Elections and Party Landscape in the Visegrád Group Countries, 27 October 2006. Brno: Institute for Comparative Political Research (IPSO). Esaiasson, P. and Heidar, K. (2000). Beyond Westminster and Congress: The Nordic Experience. Columbus: Ohio State University Press. Esaiasson, P. and Holmberg, S. (1996). Representation From Above: Members of Parliament and Representative Democracy in Sweden. Aldershot: Darthmouth. Eulau, H. and Karps, P. D. (1977). The puzzle of representation: Specifying the components of responsiveness. Comparative Legislative Research Center, 2(3), 233-254. Eulau, H., Wahlke, J. C., Buchanan, W. and Ferguson, L. C. (1959). The role of the representative: Some empirical observations on the theory of Edmund Burke. American Political Science Review, 53(3), 742-756. European Election Studies. (n.d.). Retrieved 2013, 5 May from: http://www.ees-homepage.net/ Farrell, D. and Scully, R. (2010). The European Parliament: One parliament, several modes of political representation on the ground. Journal of European Public Policy, 17(1), 36-54. Holmberg, S. (2011). Dynamic Representation from Above. In Rosema, M., Denters, B. and Aarts, K., How Democracy Works: Political Representation and Policy Congruence in Modern Societies (p. 53-76). Amsterdam: Pallas Publications. Irish National Election Study (INES). (n.d.). Retrieved 2013, 1 August from http://www.tcd.ie/ines/ Kristinsson, G. H. (2011). Party cohesion in the Icelandic Althingi. Stjórnmál og stjórnsýsla, 7(2), 229-251. Kuklinski, J. H. and Elling, R. C. (1977). Representational role, constituency opinion, and legislative roll-call behavior. American Journal of Political Science, 21(1), 135-147. Onnudottir, E. H. (Forthcoming). Policy congruence and style of representation: Party voters and political parties. West European Politics. Parliament and Government Composition Database. (n.d.). Retrieved 2012, 1 March from: www.parlgov.org Róbert, P. and Papp, Z. (2012). Party identification and party preference at the 2010 Hungarian general elections. Paper presented at the conference: Elections, Public Opinion and 22
Parties (EPOP) conference, University of Oxford, 7th to 9th September 2012. Oxford (UK). Searing, D. D. (1994). Westminster s World: Understanding Political Roles. Cambridge (US): Harvard University Press. Strøm, K. (2012). Roles as strategies: towards a logic of legislative behavior. In Blomgren, M. and Rozenberg, O. Parliamentary Roles in Modern Legislatures (p. 85-100). Oxon: Routledge. Thomassen, J. and Esaiasson, P. (2006). Role orientations of members of parliament. Acta Politica, 41(3), 217-231. Valen, H. and Narud, H. M. (2007). The Conditional Party Mandate: A Model for the Study of Mass and Elite Opinion Patterns. European Journal of Political Research, 46, 296-316. Wahlke, J. C., Eulau, H., Buchanon, W. and Ferguson, L. C. (1962). The legislative system: Explorations in legislative behavior. New York: John Wiley and Sons. Wessels, B. (1999). Whom to represent? Role-orientation of legislators in Europe. In Schmitt, H. and Thomassen, J., Political Representation and Legitimacy in the European Union (p. 209-234). Oxford: Oxford University Press. Wessels, B. and Giebler, H. (2011). Choosing a Style of Representation: The Role of Institutional and Organizational Incentives. Paper presented at: The 6th ECPR General Conference. Reykjavik, 25-27 August. Zittel, T. (2012). Legislators and their representational roles. In Blomgren, M. and Rozenberg, O., Parliamentary Roles in Modern Legislatures (p. 101-120). Oxon: Routledge. 23