The Representative Roles of MPs: A Citizen Perspective

25
The Representative Roles of MPs: A Citizen Perspective Åsa Bengtsson* and Hanna Wass From the normative point of view, there is a general agreement that representatives should act in line with the interests of those being represented. The knowledge about citizens’ preferences for representation is very limited, however.This study examines MP’s representative roles from the perspective of the citizens. It utilises a task definition approach in the Finnish institutional setting, which substantially differs from the context of earlier investigations in terms of open- list electoral systems with mandatory preferential voting. Based on the 2007 Finnish National Election Study (n = 1,422), voters’ preferences concerning four different representational roles are analysed: as representatives pursuing the interests of their electoral district, party, indi- vidual voters or being independent actors. Next, voters’ preferences are accounted for by the factors related to each type of representation: citizens’ regional electoral context, party attach- ment and electoral supply, political engagement and political competence, respectively.The results show that citizens living in electoral districts located far away from the political centre or in constituencies where it is more difficult for small parties to win political representation are most prone to prefer regional representation. Similarly, voters who have closer ties with political parties prefer party-centred representation while those who feel less politically effi- cient favour close ties with their MPs. Education in turn increases the support for a political representative to act independently from the electorate or the party. Introduction As ‘representation’ is widely defined as the relationship between MPs and their electors (Urbinati & Warren 2008, 389), the citizen perspective has to a large extent been neglected in the empirical research of representation. With the exception of recent studies by Bengtsson and Wass (2010), Carman (2006, 2007), Méndez-Lago and Martínez (2002), and the earlier ones by Davidson (1970), McMurray and Parsons (1965) and Patterson et al. (1975), empirical analyses in the field have been elite-driven, concentrating solely on MPs views of their representational roles (e.g., Converse & Pierce 1979; Esaiasson 2000; Gross 1978; Gunlicks 1969; Katz 1997; Oksanen 1972; Studlar & McAllister 1996; Wahlke et al. 1962). Carman (2007, 1) suggests * Åsa Bengtsson, Department of Political Science, Åbo Akademi University, Fänriksgatan 3a, 20500 Åbo, Finland. E-mail: asa.bengtsson@abo.fi ISSN 0080–6757 Doi: 10.1111/j.1467-9477.2011.00267.x © 2011 The Author(s) Scandinavian Political Studies © 2011 Nordic Political Science Association Scandinavian Political Studies,Vol. 34 – No. 2, 2011 143

Transcript of The Representative Roles of MPs: A Citizen Perspective

The Representative Roles of MPs:A Citizen Perspective

Åsa Bengtsson* and Hanna Wass

From the normative point of view, there is a general agreement that representatives should actin line with the interests of those being represented.The knowledge about citizens’ preferencesfor representation is very limited, however.This study examines MP’s representative roles fromthe perspective of the citizens. It utilises a task definition approach in the Finnish institutionalsetting, which substantially differs from the context of earlier investigations in terms of open-list electoral systems with mandatory preferential voting. Based on the 2007 Finnish NationalElection Study (n = 1,422), voters’ preferences concerning four different representational rolesare analysed: as representatives pursuing the interests of their electoral district, party, indi-vidual voters or being independent actors. Next, voters’ preferences are accounted for by thefactors related to each type of representation: citizens’ regional electoral context, party attach-ment and electoral supply, political engagement and political competence, respectively. Theresults show that citizens living in electoral districts located far away from the political centreor in constituencies where it is more difficult for small parties to win political representation aremost prone to prefer regional representation. Similarly, voters who have closer ties withpolitical parties prefer party-centred representation while those who feel less politically effi-cient favour close ties with their MPs. Education in turn increases the support for a politicalrepresentative to act independently from the electorate or the party.

IntroductionAs ‘representation’ is widely defined as the relationship between MPs andtheir electors (Urbinati & Warren 2008, 389), the citizen perspective has toa large extent been neglected in the empirical research of representation.With the exception of recent studies by Bengtsson and Wass (2010), Carman(2006, 2007), Méndez-Lago and Martínez (2002), and the earlier ones byDavidson (1970), McMurray and Parsons (1965) and Patterson et al. (1975),empirical analyses in the field have been elite-driven, concentrating solelyon MPs views of their representational roles (e.g., Converse & Pierce 1979;Esaiasson 2000; Gross 1978; Gunlicks 1969; Katz 1997; Oksanen 1972;Studlar & McAllister 1996; Wahlke et al. 1962). Carman (2007, 1) suggests

* Åsa Bengtsson, Department of Political Science, Åbo Akademi University, Fänriksgatan 3a,20500 Åbo, Finland. E-mail: [email protected]

ISSN 0080–6757 Doi: 10.1111/j.1467-9477.2011.00267.x© 2011 The Author(s)

Scandinavian Political Studies © 2011 Nordic Political Science Association

Scandinavian Political Studies, Vol. 34 – No. 2, 2011 143

that two types of prejudices account for this deficit. First, as both scholarsand ordinary citizens are expected to homogeneously favour the mandateconception of representation in which representatives closely follow theinstructions given by the voters (also Manin et al. 1999, 30), empiricalresearch has not seemed relevant. Second, voters have been assumed to beincapable of forming preferences for the complex process of representation.

There are, however, numerous bases for the studying of voters’ prefer-ences. The finding that voters have opinions is important as such given thecomplex nature of representation. Carman (2007, 2, referring to Zaller &Feldman 1992) points out that knowing that voters have preferences andthat there is a cross-sectional variation in them is a sufficient point ofdeparture for a citizen-based study of representation. Furthermore, a poten-tial gap between citizens’ expectations and the MPs’ actions could increasethe feeling of alienation from politics and discontent with MPs and politicalparties and more generally weaken trust in governmental institutions(Carman 2007, 104; Méndez-Lago & Martínez 2002). This might also have anegative impact on turnout. Consequently, research on voters’ views couldimprove the functioning of representation by increasing the MPs’ awarenessof the expectations of the citizens, and consequently, their ability to meetvarious needs. The final reason for the study of citizens’ expectations is inline with Castiglione and Warren (2006, 11), who suggest that even thoughthere are rules and norms that define representational roles, these roles arerelational, constituted by both the self-understanding of the representativeand by the expectations of those being represented.

Based on the Finnish National Election Study (FNES) 2007 (FSD2269,n = 1,422), we examine voters’ preferences for various representationalroles and the extent to which these preferences are accounted for by factorssuch as the electoral context, party attachment, political engagement andpolitical competence. Our study builds on a previous one on citizens’ pref-erences for the style of representation in the Finnish context (Bengtsson &Wass 2010) by adding a new element in it – namely ‘focus of representation’.While focus and style of representation are analytically separate from eachother (Eulau et al. 1959, 744), in the Finnish context of an open-list electoralsystem with mandatory preferential voting they are in practice closely inter-twined. This aspect is particularly interesting as the majority of existingstudies concerning voters’ views on representation have been conducted inthe United States and the United Kingdom – countries that use pluralityelectoral systems with single member districts.1 The main contribution ofour study is to utilise a task definition approach to representation by Esa-iasson (2000, 52) from the point of view of citizens in an institutional settingthat substantially differs from the context of the earlier investigations.

The study is structured as follows. We begin by discussing the focusand style dimensions of representation in general. Then we introduce four

© 2011 The Author(s)Scandinavian Political Studies © 2011 Nordic Political Science Association

144 Scandinavian Political Studies, Vol. 34 – No. 2, 2011

different roles of representation that can be regarded as the most relevantin the Finnish context. These include three types of interest representation– geographical, party and individual representation (Esaiasson 2000, 55;Eulau & Karps 1977, 248, for terms) – as well as an independent, trustee-type of representation. In the empirical section, we first present our researchdesign and hypotheses, and continue with the analyses of voters’ prefer-ences for various representation roles. In the second part of the study,citizens’ preferences are analysed in relation to factors relevant to each typeof representation. We finish with concluding remarks and a discussion con-cerning the implications of our findings for the future of electoral-basedresearch on representation.

The Representative Roles of MPsRepresentation is often divided into two distinct dimensions: focus andstyle. While the former concerns the question of whose interests legislatorsrepresent, the latter is related to the manner in which citizens are beingrepresented (Wessels 2007, 838). Eulau et al. (1959, 744) argued in theirseminal study of representational roles that focus and style should betreated separately in empirical investigations as style is neutral in relation tothe different interests MPs might pursue. In practise, however, the twoconcepts are closely related. In his famous speech in Bristol in 1774,Edmund Burke linked the focus of representation with a particular repre-sentational style when he suggested that the ideal representative shoulddefend the national interest as an independent actor (Eulau et al. 1959, 744;Judge 1999, 53; Thomassen 1994, 238–9).

Also in the Finnish open-list proportional representation (PR) system(e.g., Marsh 1985, 365), which is the context of this study, focus and style ofrepresentation are difficult to separate. Whereas in many PR systems inWestern Europe voters are entitled to indicate their favoured candidatewithin their favourite party list, in Finland it is compulsory to vote for aspecific candidate (Reynolds et al. 2005, 84). The number of seats won byeach party is based on the number of total votes gained by the party’scandidate.The MPs representing each party are then selected on the basis oftheir personal votes. According to Esaiasson (2000, 53), the role played bythe party in MPs’ task definition is expected to be much smaller incandidate-centred systems compared to electoral systems in which voterschoose between party lists.2 From the voters’ point of view, candidate-centred electoral systems might raise several types of expectations on theroles pursued by MPs. On the one hand, voters could form expectations ofa close relationship with their MP. On the other, voters might prefer anindependent, trustee-like behaviour from their MP, especially if they haveprimarily voted for a candidate and only secondary the party he or she

© 2011 The Author(s)Scandinavian Political Studies © 2011 Nordic Political Science Association

Scandinavian Political Studies, Vol. 34 – No. 2, 2011 145

represents.3 Clearly, both considerations are closely related to two classicalmodels concerning the style of representation – namely the ‘delegate’ and‘trustee’ types of representation (Pitkin 1967).The delegate style model canalso be regarded as interest representation in terms of individual voters.Thetrustee model, however, belongs theoretically to category of style eventhough in practice it could be argued that it relates to the focus of repre-sentation in the Finnish context.

Perspectives of Representation

As mentioned in the previous section, the focus of representation refers tothe question of whose interests the MPs represent. Eulau and Karps (1977,248) divide the general representation focus of ‘constituency’ into threecategories. The first is formed on a geographical basis and may include focisuch as the nation, regime, state, district or any other territorial level. Thesecond category consists of various functional groups such as religious,ethnic, economic and ideological groups.The third focus refers to individualpersons. Esaiasson (2000, 55) remarks that also fourth category needs to beadded – namely the interests of the party.4 In his empirical analysis of taskdefinition among MPs from five Nordic countries, Esaiasson (2000, 58–9)includes the fifth category illustrating Burkean representation which meansthat MPs act independently from their parties and voters. In the following,we will discuss four of the aforementioned categories in more detail.5

In the classical debate, the focus of geographical representation wasmainly viewed as a struggle between the national interest and the interest ofthe constituency – that is, local interest (Narud & Valen 2007).6 Territoriallydefined constituencies originally formed the basis of representation, and therepresentatives from single-member electoral districts were regarded asspokesmen for local interests (Valen et al. 2000, 107). In the first phase ofrepresentative democracy, which Manin (1997, 202–3) labels ‘parliamenta-rism’, representatives were frequently in contact with their constituencies.They belonged to the same social community as their voters, whether in theform of a geographical area such as a constituency, a town, a city, or a county,or in the form of more general common interests.

In the contemporary context of representation, territory still has a sub-stantial role as the main basis of representation, and MPs are formallyconsidered to be representatives of a certain geographical district (Judge1999, 149). Constituency-based representational focus is particularly appar-ent in political systems such as the United States, where the organisationalstructure of political parties is highly decentralised, the level of party disci-pline is low and programmatic platforms are rather minimalistic (Marsh &Norris 1997, 156). In addition, locality seems to become more importantalongside the global level. The candidate-centred single transferrable vote

© 2011 The Author(s)Scandinavian Political Studies © 2011 Nordic Political Science Association

146 Scandinavian Political Studies, Vol. 34 – No. 2, 2011

(STV) system used in Ireland forms a relevant example where localitystands out as one of the most important motivations for a voter to choosecandidate (Marsh 2007, 509). This tendency might also be connected to thecentre-periphery division (Valen et al. 2000, referring to Rokkan 1970,Rokkan & Urwin 1982, 1983). In their study of three Scandinavian coun-tries, Valen et al. found that representatives of the periphery have higherpropensity to emphasise the importance of promoting regional interestscompared to those that represent constituencies situated in the centre.Moreover, institutional arrangements may affect the importance given togeographical representation. MPs can be expected to pay more attention tolocal constituency interests in systems based on local nomination of thecandidates compared to contexts where the nomination process is morecentralised (Esaiasson 2000, 53). From this perspective, it is to be expectedthat geographical representation is highly relevant in Finland, where localparty branches propose candidates and the final composition is decided atthe constituency party meeting (Heidar et al. 2000, 32, for review). In par-liamentary elections, Finland is divided into 15 electoral districts. Moreover,there is substantial variation in the number of representatives elected fromeach constituency.

The model of party representation describes a practise of representationrather than forms a theory by itself (Judge 1999, 71). Political parties wereoriginally considered harmful to democracy. This was partly related to thefact that major theories of political representation were formulated priorto the formation of modern parties (Thomassen 1994, 250). As suffragewas substantially extended towards the end of the nineteenth and thebeginning of the twentieth century, political parties were established tomobilise the enlarged electorate (Manin 1997, 206; Thomassen 1994, 250–1). Alongside the increasing role of parties in elections, the role of thepolitician also changed. MPs became more coordinated in their decisionmaking and parties gradually began to control the whole process of rep-resentation (Méndez-Lago & Martínez 2002, 64). Consequently, electorsno longer voted for a candidate with whom they had personal contact, butinstead started to vote for a representative of a party (Manin 1997, 206).With this development, a new role of representation was establishedwhich can be described both in terms of style and focus (Méndez-Lago &Martínez 2002, 64).

In what is called a ‘responsible party model’, representation takes placebetween voters and parties, which are viewed as unitary actors. In thecontext of strict party discipline, the role played by individual politicians isseverely limited (Thomassen 1994, 251–2). The modern interpretation ofrepresentation often refers to a party-based model, which is particularlycommon in parliamentary systems with strong, competitive and program-matic parties (Marsh & Norris 1997, 154). Even in those electoral systems

© 2011 The Author(s)Scandinavian Political Studies © 2011 Nordic Political Science Association

Scandinavian Political Studies, Vol. 34 – No. 2, 2011 147

where some individual representation is allowed, elections are mainly acompetition between various parties. In spite of the dominant role ofparties, many constitutions and laws either do not mention parties at all, orinclude only a very general reference to them that does not define therelationship between parties and individual representatives (Méndez-Lago& Martínez 2002, 65).

As mentioned in the previous section, the Finnish MPs’ perceptions oftheir role as political representatives are far less party-centred than theircolleagues in other Nordic countries. This is probably linked to the strongfocus on individual candidates in the Finnish electoral system (Esaiasson2000, 59, 61–2). Moreover, while party unity in parliamentary voting isgenerally very high in the Nordic countries, Finland also deviates slightlyfrom this pattern with a less pronounced discipline (Jensen 2000, 217–20).Based on these facts, we might expect party representation to be lesspopular among the Finnish voters.

Defining individual representation is a more challenging task than outlin-ing geographical and party representation.A potential perception is the oneof private citizens seeking help from their representatives (Esaiasson 2000,56, for terminology), which clearly belongs to the category of the focus ofrepresentation. As Esaiasson (2000, 56) remarks, such focus does not have aparticular place in representation theory. In practice, however, MPs regardrepresentation of individual voters as a part of their duties (Esaiasson 2000,60). Correspondingly, half of the electorate considers the candidate to be amore important factor when deciding upon how to cast their vote.7 Thismight suggest that they anticipate a more personal relationship with theirrepresentatives.

Interestingly, previous research among Nordic MPs points in a differentdirection. An individual type of representation is considered more impor-tant among MPs from other Nordic countries than Finland (apart fromDenmark) even though their electoral systems are less candidate-centred(Esaiasson 2000, 59).8 The finding naturally says nothing about citizens’expectations. Moreover, it might actually constitute an example of discrep-ancy between MPs’ and citizens’ views on representation.

The fourth representational role examined in this study – the trustee typeof representation – relates mainly to the style of representation. It is basedon classical trustee model of representation introduced by Burke in which arepresentative is expected to make independent political decisions based onwhat he or she thinks is best for the citizens.9 While the trustee model ofrepresentation has often been considered not to apply to modern societieswith an educated citizenry, universal suffrage and political parties, it still hasa place both in theory and practise of representation (Esaiasson 2000, 58–9).Even though parties remain major actors in parliamentary arenas, especiallyin closed-list electoral systems, some recent developments suggest that the

© 2011 The Author(s)Scandinavian Political Studies © 2011 Nordic Political Science Association

148 Scandinavian Political Studies, Vol. 34 – No. 2, 2011

role of the individual politician is once more becoming more apparent. Inthe era of electronic media, which Manin (1997, 220–1) refers to as ‘audiencedemocracy’, candidates can communicate directly with voters without themediating role of political parties. With the enlarged scope of governmentand a more complex environment, characterised by growing economic inter-dependence and multiple actors taking part in decision making, it hasbecome increasingly difficult for parties and individual candidates to take astand on various issues in advance.As voters are aware of the complexity ofthe current decision-making context, they might also be inclined to give amore open mandate to candidates whose personal qualities and capabilitiesthey value (Manin 1997, 220–1). At the same time, the level of party iden-tification has declined in most industrialised democracies (Dalton 2000).

The importance of a trustee type of representation is clearly demon-strated by the fact that several nations, including Finland, have included theprinciple of independence of the MPs in their constitutions (Wessels 2007,840). Moreover, personal opinions of the MPs is considered to be the mostimportant interest category among Nordic MPs alongside the party theyrepresent (Esaiasson 2000, 59). From the perspective of citizens especially inthe Finnish context, however, a trustee role of representation is probablynot so much associated with independence from voters as from parties.

Research DesignThe Finnish National Election Study (FNES) 2007 (FSD2269) is one ofthe few publicly available surveys that include questions concerningvoters’ preferences on political representation. The dataset thus provides arare opportunity to examine how voters view the relationship betweenthemselves and their political representatives, which constitutes the firstaim of this study. The second aim is to account for the differences invoters’ views by using four different models designed for each particulartype of representation.

FNES 2007 is a post-election survey conducted after the parliamentaryelections held in March 2007. The data was collected in two stages. The firststage involved face-to-face interviews with a total of 1,422 respondentsbased on quota sampling.10 The second stage involved data collection via aself-administrated questionnaire, to be returned by mail. The questionnairewas answered by 1,033 of the respondents interviewed in the first stage.The questions on representation analysed here were asked in the first,face-to-face, stage of the study. The survey includes an oversample of theSwedish-speaking population in Finland, which is controlled for by usingappropriate weights.

The perspectives of representation examined here can be identified as themost relevant in the Finnish political context. Consequently, we utilise a task

© 2011 The Author(s)Scandinavian Political Studies © 2011 Nordic Political Science Association

Scandinavian Political Studies, Vol. 34 – No. 2, 2011 149

definition approach by Esaiasson (2000) that takes into consideration themost appropriate representational roles from citizens’ point of view asopposed to any specific theoretical perspective. In FNES 2007, respondentswere confronted with the question ‘What kind of aims should Members ofParliament primarily promote?’ with the following exclusive response cat-egories: (1) aims set by their party, (2) aims they themselves see as the mostimportant, (3) aims important to their constituency, and (4) aims of theirvoters.11 On the one hand, these alternatives reflect the controversy over adirect relationship between MPs and voters (the delegate) and a moreindependent role of political representatives (the trustee). On the otherhand, it also includes the dimension concerning MPs’ role as representativesfor locally defined interests in terms of the interests of the constituency orindividual voters, or the more modern representation focus of parties as themain aggregator of voters’ interests.12

Before embarking on the discussion concerning voters’ preferences forvarious roles, some clarifying remarks are in order. First, it should be notedthat the different perspectives of representation are treated separately foranalytical purposes only. Respondents were simply asked to reveal their firstpreference. The preferences were listed in a very general manner. Conse-quently, the distribution of responses could have been different if the ques-tion were linked to some particular decision-making situation or policyissue. Pitkin (1967, 10–1; also Schwindt-Bayer & Mishler 2005, 407) pointsout in her seminal study that representation contains several dimensionswhich are interrelated.13 In practise, various representational roles oftenintertwine with each other. Eulau et al. (1959, 745) suggested as early as atthe end of the 1950s, that apart from voters, representatives also face expec-tations from other clienteles such as political parties, pressure groups andadministrative agencies (also Judge 1999, 149). In addition, different roles ofrepresentation might be mutually inclusive if, for instance, one party isdominating in a particular area to such an extent that MPs view the interestsof the constituency and the party are identical. Finally, empirical evidencefrom the United Kingdom demonstrates that voters are, at least to someextent, aware of the complexity of representation and that they do notregard alternative roles of representation as being separate (Carman 2006,111–12). Previous studies have, however, convincingly shown that voters dohave preferences concerning representation that are not random, but thatvary systematically (Bengtsson & Wass 2010; Carman 2006, 2007; Méndez-Lago & Martínez 2002).

In spite of its relative scarcity, previous literature in the field has intro-duced some results concerning systematic variation in voters’ preferencesthat can be utilised to deduce relatively distinct hypotheses. The primarystudies of relevance are those by Carman (2006, 2007) and by Bengtssonand Wass (2010), but utilise also earlier work by McMurray and Parsons

© 2011 The Author(s)Scandinavian Political Studies © 2011 Nordic Political Science Association

150 Scandinavian Political Studies, Vol. 34 – No. 2, 2011

(1965) and Patterson et al. (1975). Based on these previous studies as well asby introducing new lines of reasoning partly developed from the specificfeatures of the Finnish electoral context, we will outline one hypothesis foreach of the four perspectives of representation and introduce the indepen-dent variables that will be included in the empirical analyses. Hypothesesare formulated on the individual as well as on the contextual levels.

The first representational role relates to the local or constituency per-spective of representation – that is, the expectation that MPs shouldpromote the aims important to their constituency (geographical representa-tion). The related hypothesis is linked to the classic centre-periphery divide,research on political representatives’ own role definitions, as well as therecent debate about the Finnish electoral system. In parliamentary elec-tions, Finland is divided into 15 constituencies, from which a number ofrepresentatives proportional to the number of citizens residing in eachdistrict is elected (6–39 seats in 2007).14 While a fixed electoral threshold isnot in use in Finland, the effective threshold varies extensively from oneconstituency to another according to the number of seats distributed. In theparliamentary elections of 2007, the effective threshold ranged from 2.9 to14.3 percent in the various constituencies (Borg & Paloheimo 2009). Thedifferent prerequisites facing parties as well as voters from different parts ofFinland set off an intensive debate after the 2007 election.

Given the large variation in district magnitude (the number of seatsdistributed in each constituency), the level of proportionality also varies fromone constituency to another (Söderlund et al. 2011, 96–7). In practice, thisimplies that it is more difficult for small parties to gain representation in someconstituencies than in others.This is likely to affect the behaviour of voters interms of strategic considerations.15 More importantly,however,we expect thedifferent prerequisites and the higher amount of strategic voting to have animpact on citizens’ tendency to prefer their political representatives toprioritise the regional perspective in their work.16

It is also possible that factors other than strategic considerations con-nected to the different electoral context affect the expectations voters haveon MPs and their actions.The potential effect might be further strengthenedby geographical aspects since the constituencies that are expected to con-tribute to a stronger preference for geographical representation are locatedin the eastern and northern part of Finland, far from the capital Helsinkiand the centre of politics. Valen et al. (2000) has convincingly shown thatthose Nordic MPs that represent constituencies located in the peripherygive higher priority to promoting the interests of their own region comparedto MPs from constituencies in the centre. Consequently, we expect votersfrom districts where it is more difficult for smaller parties to gain representa-tion and/or constituencies far away from the capital Helsinki to be moreinclined to prioritise regional representation. To test the hypotheses we use

© 2011 The Author(s)Scandinavian Political Studies © 2011 Nordic Political Science Association

Scandinavian Political Studies, Vol. 34 – No. 2, 2011 151

several contextual variables, such as the effective threshold in the respon-dent’s constituency (Taagepera 1998), district magnitude (the number ofseats distributed), the level of disproportionality (Loosmore & Handby1971)17 and the logged distance from the capital Helsinki.18

The second of the four alternative representative roles is acting accor-dance with the aims set by the representatives’ party (party representation).In this case, it seems natural to assume that the ties that respondents have toparties, as well as the selection of alternatives that voters have to choosefrom, will influence their tendency to opt for the alternatives of party astheir first preference.Voters with a distinct party preference and a feeling ofcloseness to a specific party in the Finnish party system can be expected tobe more satisfied with the central role of the parties in the representationprocess. From a contextual perspective, we are interested in the effect of thenumber of effective parties contesting in each constituency. A high numberof effective parties suggests that more options are open to voters (Cox1999), which in turn might increase voters’ propensity to prefer party rep-resentation. We thus assume that party supporters and voters living in aconstituency where the range of party choice is wider are more inclined toprioritise representation where the agenda set by the party plays a substantialrole (e.g., Carman 2006, 109).

In the following analysis, we use four different variables to test thehypothesis: three individual-level variables and one contextual variable.Thefirst variable is a degree of party identification, which relates to the strengthin the emotional ties between citizens and specific parties. The second is theextent to which voters have a distinct political orientation separate from themiddle on an 11-point left–right scale.19 The third variable measures thesubjective weight given to parties versus candidates in voter’s vote choice.Since Finland uses a mandatory preferential voting system, voters are notable to vote without casting a preferential vote. Thus, the question aboutwhich mattered more, the candidate or the party that the candidate repre-sents, is a valid indicator of the relative importance that parties have forvoters. The contextual variable tests the importance of the electoral context– in this case, the range of alternatives that are available to voters, opera-tionalised as the effective number of parties in each constituency.

The third perspective of representation available to the respondents is theclose relationship between voters and MPs. This perspective is very closelyrelated to the delegate model of representation and is here labelled ‘indi-vidual representation’. Our hypothesis concerning individual representationis connected to the views voters have about the political system. Morespecifically, we expect voters who feel alienated from the political system orare dissatisfied with how it works to be more inclined to prefer a closerelationship with their political representatives.The reason behind this expec-tation is that voters who feel disconnected from the political system might

© 2011 The Author(s)Scandinavian Political Studies © 2011 Nordic Political Science Association

152 Scandinavian Political Studies, Vol. 34 – No. 2, 2011

have a lower level of trust in current actors and institutions in relation totaking into account their interests and therefore prefer to maximise controlover the political decision making (Carman 2006, 108). The hypothesis istested by using three individual-level variables: external efficacy, politicalinterest and turnout in the 2007 elections. External efficacy is an index ofthree different statements measuring how responsive, or open to input,voters consider the political system to be. The level of political interest andvoting in turn indicate the level of political engagement.

The fourth hypothesis is related to the preference for the alternativewhere MPs act independently from parties and voters (trustee type of rep-resentation). As the label reveals, this kind of representational role closelyresembles the classic trustee model. Based on previous research, whichshows that citizens with a higher level of education are more inclined tofavour an independent political representative (Bengtsson & Wass 2010;Carman 2006, 110), we expect political competence to play an importantrole. Furthermore, voters with low awareness about political reality areexpected to feel less comfortable with an abstract relationship with theirMP. Hence, the more knowledge and understanding of how politics worksand of the complexity of the political process voters have, the more we expectthem to favour an independent political representative.

To measure the political competence of voters we use three differentindividual-level variables: education, political knowledge and internal effi-cacy. The level of political knowledge is measured by an index based on theanswers to five questions concerning actors and institutions in the Finnishpolitical system and international politics. Internal efficacy illustrates thesubjective political competence – that is, how well respondents feel thatthey understand what is going on in the political arena. The level of edu-cation functions, as a contrast to the two previous measures, as an indirectcompetence estimate.

Finally, we also introduce two control variables in all of the analysesperformed. Age and gender are significant predictors of political attitudesand behaviour and have been found to influence views of representation inprevious research (Bengtsson & Wass 2010; Carman 2006; McMurray &Parsons 1965; Patterson et al. 1975). Men and older voters appear to bemore inclined to prefer a representational role of an independent trustee(Bengtsson & Wass 2010, 70; Carman 2006, 113). More detailed informationabout the variables and coding is available in the Appendix.

The following analyses are based on multinomial logistic regressionmodels, which are utilised to analyse the relationships of dependent vari-ables with more than two unordered categories. We proceed with a block-wise inclusion of the four sets of independent variables designed to test thehypotheses for each representational role, and conclude with a modelincluding the variables that have turned out significant in the previous

© 2011 The Author(s)Scandinavian Political Studies © 2011 Nordic Political Science Association

Scandinavian Political Studies, Vol. 34 – No. 2, 2011 153

analyses. Due to a high level of correlation between the five contextualvariables, only one macro variable at a time will be included in the analyses.Where more than one contextual variable could be used to test the afore-mentioned hypotheses, several separate analyses were conducted and thevariable with the strongest effect is presented. To control for the nestedstructure of our data when including contextual-level data (n = 14), we usea robust-clustered standard errors approach that allows for variance withinclusters.20 All of the five analyses presented include two control variables:age and gender.

Empirical Analyses

Voters’ Preferences for Representation

Political representatives have to deal with conflicting interests. In parlia-mentary systems, the interests of the parties have a strong impact onindividual MPs. Without coherent parties and party discipline, stable gov-ernments would be difficult to form as well as maintain. In a candidate-centred electoral system such as the Finnish one, the actions of individualMPs often attract attention, particularly from the local perspective. Conse-quently, voters’ expectations might reflect these various conflicting interests.

Regardless of the different and inconsistent expectations, Table 1 showsthat voters have rather straightforward opinions on the matter of represen-tation. Despite the visible role of parties and speculations about a complexpolitical agenda bringing more attention to individual MPs and their voters,the most popular perspective among voters is geographical representation.From the four options available, almost 50 percent of all respondents con-sider the interest of the constituency to be the most important considerationfor MPs in their task as political representatives. Given the development ofmodern representative democracy from defending local interest towardsgeneral policy making at the national level, the result seems quite surprising.Moreover, it is interesting that locality has such as strong influence in thecurrent political context that is characterised by globalisation and increasing

Table 1. Voters’ Preferences for MP’s Role of Representation

% (n)

Aims important to their constituency 49 (692)Aims of their voters 22 (306)Aims set by their party 18 (248)Aims they themselves see as the most important 11 (156)Total 100 (1,403)

© 2011 The Author(s)Scandinavian Political Studies © 2011 Nordic Political Science Association

154 Scandinavian Political Studies, Vol. 34 – No. 2, 2011

interdependence.21 One plausible interpretation might be related to the factthat politics as a whole seems vague and diffuse, while the interests of thelocal community or the constituency are considered to be more concreteand easy to define.

The second most popular representation role – individual representation– is supported by 22 percent of the respondents. This type of close relation-ship between MPs and voters is often considered difficult to carry out inpractice. The development of electronic media along with an increase in theassistance that MPs’ have has made representatives more easily availablefor voters, however.We also know that Finnish voters are relatively eager tocontact their politicians (Bengtsson & Christensen 2009) and that the man-datory preferential voting system increases the importance on MPs beingavailable and visible to their potential voters (Bowler & Farrell 1993).

Voters’ preference for a close relationship with their MP reflects thestrong candidate-centeredness in the Finnish system. Candidate-cent-eredness might also be connected to the importance given to local perspec-tives since individual MPs are well known and often have a visible role intheir electoral district with the local media closely following their actions.This interpretation is supported by the findings from the British first-past-the-post context22 where the interests of the constituency or voters areranked higher than the trustee or party perspective (Carman 2006, 111). Itis also supported by results by Marsh (2007, 509) in the candidate-centredIrish system, where locality stands out as an important motivation behindthe choice of a candidate.23

The perspective of representation in which the parties formulate thecentral aims is ranked third among the Finnish electorate. This role ofrepresentation is viewed as the most important by only 18 percent of therespondents. Given the important role that parties play in political decisionmaking even in the candidate-centred Finnish electoral system, this figurecan be regarded as low.The low support for the party perspective can eitherbe interpreted as a genuine opinion about how politics should be designed,or as an expression of discontent with the way politics currently works.According to the latter option, the important role of political parties isrecognised, but more input from the side of individual MPs is seen asdesirable. The other possibility is that voters think that parties have largelyplayed out their role as aggregators of voters’ interests and as organisers ofgroup interests in society.

The least popular representative approach is the aims defined individuallyby the MPs themselves. This view is supported by only 11 percent, eventhough the mandatory preferential voting system to some extent encour-ages this type of relationship. Low support might reflect a low level of trustin the ability of MPs to take into account ‘the common good’ instead of theirpersonal interests.

© 2011 The Author(s)Scandinavian Political Studies © 2011 Nordic Political Science Association

Scandinavian Political Studies, Vol. 34 – No. 2, 2011 155

Multivariate Models

In this section of the empirical analyses, voters’ different priorities concern-ing representational roles are accounted for by their socio-economic andpolitical backgrounds. In Table 2 we present five different analyses. The firstfour test our hypotheses concerning each perspective of representationtogether with the control variables of age and gender. The fifth modelincludes all the variables with a statistically significant effect in the previousanalyses. When interpreting the results in Table 2, it is important to bear inmind that the group of comparison varies from one model to anotherdepending on which of the perspectives is of central interest in the analysis.In the final model, we have chosen the most popular alternative as a refer-ence category: geographical representation.

The first model studies the importance of the electoral context and thecentre-periphery division for voters’ expectations on representation. Hereseveral variables (district magnitude, disproportionality, effective electoralthreshold and the distance to the capital Helsinki) are tested separately toavoid multicollinearity.24 The first part of our hypothesis suggests that votersliving in electoral districts where the possibility of casting a wasted vote ishigher should be more inclined to prefer regional representation. Thesecond part, in turn, deals with the potential effect of being alienated fromthe central political arena and suggests a similar effect. The results are, tosome extent, in line with expectations. The two variables that turn out tohave the strongest effects are the effective electoral threshold and thedistance variable, where the latter yields a slightly higher pseudo R2 and isthus presented in Table 2.

We therefore find that the propensity for voters to regard the party’s, thevoters’ or the MPs’ aims as the most important task for political represen-tatives to pursue declines among voters that are resident in constituencieswhere it is more difficult for small parties to win political representation,where strategic considerations may play a more enhanced role for voters’electoral choices and where the election outcome is less proportional. Inter-estingly, an even stronger effect in the same direction is found among votersliving far away from the centre. The fact that Finland is very unevenlypopulated, with the largest concentration of people living in the area aroundthe capital Helsinki in the south, influences voters expectations for repre-sentation. Voters living in the less populated areas consider the regionalfocus of representation as more important than others. The result corre-sponds to the finding by Valen et al. (2000) concerning MPs’ own views ontheir representational role.

The second model investigates the rather self-evident hypothesis that thelevel of party attachment has an effect on voters’ propensity to prefer aparty-centred role of representation. In addition, we expect that a wider

© 2011 The Author(s)Scandinavian Political Studies © 2011 Nordic Political Science Association

156 Scandinavian Political Studies, Vol. 34 – No. 2, 2011

Tabl

e2.

Mul

tino

mia

lReg

ress

ion

Ana

lysi

sfo

rth

eP

refe

rred

Rol

esof

Rep

rese

ntat

ion

(Non

-sta

ndar

dise

dR

egre

ssio

nC

oeffi

cien

ts(b

)w

ith

Stan

dard

Err

orin

Par

enth

esis

)

Mod

el1:

geog

raph

yas

refe

renc

eca

tego

ryM

odel

2:pa

rty

asa

refe

renc

eca

tego

ryM

odel

3:in

divi

dual

asa

refe

renc

eca

tego

ry

Par

tyTr

uste

eIn

divi

dual

Trus

tee

Geo

grap

hyIn

divi

dual

Par

tyG

eogr

aphy

Trus

tee

Dis

tanc

efr

omH

elsi

nki(

lg)

-1.5

0***

-1.4

9***

-0.8

6***

(0.2

0)(0

.12)

(0.1

7)P

arty

iden

tific

atio

n-1

.06

-0.8

9*-0

.52

(0.4

4)(0

.36)

(0.2

9)Po

litic

alor

ient

atio

n-0

.22

-0.6

5*-0

.29

(0.5

6)(0

.32)

(0.2

3)P

arty

(ver

sus

cand

idat

e)-1

.04*

*-0

.82*

*-0

.87*

**(0

.37)

(0.2

6)(0

.25)

Eff

ecti

venu

mbe

rof

part

ies

-0.3

9-2

.50*

*-0

.84

(0.5

5)(0

.72)

(0.8

7)E

xter

nale

ffica

cy2.

26**

*1.

27**

1.58

*(0

.52)

(0.4

5)(0

.66)

Polit

ical

inte

rest

0.96

*-0

.24

0.17

(0.4

6)(0

.36)

(0.5

1)V

oted

0.60

0.27

0.62

(0.3

3)(0

.23)

(0.3

5)E

duca

tion

Polit

ical

know

ledg

eIn

tern

alef

ficac

yG

ende

r(m

ale)

-0.1

70.

37*

-0.0

90.

65**

0.14

0.03

-0.1

30.

07-0

.48*

(0.1

8)(0

.16)

(0.1

0)(0

.22)

(0.1

9)(0

.17)

(0.2

0)(0

.16)

(0.2

3)A

ge/1

00-0

.60

-2.8

4***

-0.1

8-1

.53

1.56

***

1.72

***

-0.9

40.

23-3

.21*

**(0

.37)

(0.5

9)(0

.38)

(0.8

8)(0

.39)

(0.5

9)(0

.63)

(0.5

0)(0

.63)

Inte

rcep

t0.

330.

61-0

.10

1.52

*3.

59**

*0.

71-1

.78*

**0.

07-0

.70

(0.2

3)(0

.32)

(0.1

8)(0

.68)

(0.7

0)(0

.69)

(0.4

8)(0

.35)

(0.5

0)

Log

pseu

do-l

ikel

ihoo

d-1

685.

87-1

382.

98-1

531.

07P

seud

oR

20.

027

0.04

60.

032

N1,

407

1,18

41,

280

© 2011 The Author(s)Scandinavian Political Studies © 2011 Nordic Political Science Association

Scandinavian Political Studies, Vol. 34 – No. 2, 2011 157

Tabl

e2.

Con

tinue

d

Mod

el4:

trus

tee

asre

fere

nce

cate

gory

Fina

lmod

el:

geog

raph

yas

refe

renc

eca

tego

ry

Par

tyG

eogr

aphy

Indi

vidu

alP

arty

Indi

vidu

alTr

uste

e

Dis

tanc

efr

omH

elsi

nki(

lg)

-1.2

4***

-1.0

8***

-1.4

4***

(0.1

6)(0

.20)

(0.1

6)P

arty

iden

tific

atio

n0.

600.

73*

-0.3

5(0

.34)

(0.3

4)(0

.57)

Polit

ical

orie

ntat

ion

0.61

0.48

0.38

(0.3

2)(0

.30)

(0.5

0)P

arty

(ver

sus

cand

idat

e)0.

74**

-0.6

2-0

.35

(0.2

2)(0

.21)

(0.2

9)E

ffec

tive

num

ber

ofpa

rtie

sE

xter

nale

ffica

cy0.

69-1

.09*

*0.

12(0

.54)

(0.4

1)(0

.26)

Polit

ical

inte

rest

0.98

*0.

280.

16(0

.46)

(0.3

6)(0

.48)

Vot

edE

duca

tion

-0.3

7-0

.61

-1.5

9***

-0.0

5-0

.93*

0.43

(0.4

4)(0

.39)

(0.4

4)(0

.55)

(0.4

0)(0

.47)

Polit

ical

know

ledg

e-0

.29

-0.5

4-1

.20*

-0.1

6-0

.34

0.74

(0.5

2)(0

.44)

(0.2

9)(0

.48)

(0.4

0)(0

.66)

Inte

rnal

effic

acy

-0.0

6-0

.68

-0.6

2(0

.34)

(0.2

9)(0

.34)

Gen

der

(mal

e)0.

52*

-0.2

7-0

.32

0.11

-0.0

60.

45*

(0.2

3)(0

.21)

(0.2

3)(0

.22)

(0.1

1)(0

.20)

Age

/100

2.49

**3.

18**

*3.

04**

*-1

.58*

**-0

.41

-3.5

5***

(0.7

3)(0

.62)

(0.6

7)(0

.43)

(0.6

1)(0

.84)

Inte

rcep

t0.

011.

10**

1.20

**-1

.22*

*0.

290.

21(0

.45)

(0.3

7)(0

.42)

(0.4

3)(0

.36)

(0.4

1)

Log

pseu

do-l

ikel

ihoo

d-1

675.

27-1

283.

53P

seud

oR

20.

029

0.06

9N

1,40

11,

121

Not

es:

***p

<0.

001;

**p

<0.

01;

*p<

0.05

.D

epen

dent

vari

able

:a

vari

able

wit

hfo

urva

lues

:pa

rty,

trus

tee,

loca

lan

dvo

ters

.E

ntri

esar

eno

n-st

anda

rdis

edre

gres

sion

coef

ficie

nts.

Rob

ustc

lust

ered

stan

dard

erro

rsar

eus

edw

hen

cont

extu

alva

riab

les

are

incl

uded

inth

ean

alys

es(m

odel

s1,

3an

dth

efin

alm

odel

).A

llva

riab

les

are

mea

sure

don

asc

ale

betw

een

0–1.

Ade

taile

dde

scri

ptio

nof

the

vari

able

sis

avai

labl

ein

the

App

endi

x.M

ulti

colli

near

ity

diag

nost

icst

atis

tics

show

noca

use

for

conc

ern.

© 2011 The Author(s)Scandinavian Political Studies © 2011 Nordic Political Science Association

158 Scandinavian Political Studies, Vol. 34 – No. 2, 2011

range of choices available to voters will increase the propensity to emphasiseparty-oriented representation. Here we find that all the variables, at least tosome extent, show an effect that corresponds to our expectations. The maineffect, however, appears to focus on the choice between local and partyinterests. Voters with a strong party identification, who have a politicalorientation that deviates from the centre, who consider the choice of partymore important than a candidate when casting their vote, or who live inconstituencies with a wide selection of parties to choose from,are more proneto favour a role of representation that concentrates on party-central interests,rather than local interests. It also becomes clear that factors related to partyattachment have no impact on preferences for other representational roles.Afactor that goes beyond distinguishing between the party and the constitu-ency perspective is the weight given to party versus candidate by voters.Herewe find that voters who prioritise the party have substantially higher propen-sity to prefer the party perspective of representation compared to the otherthree options.

The third model includes factors related to political engagement. Themodel tests the hypothesis that voters who are less satisfied with the respon-siveness of the political system and have a lower level of interest in politics,and who do not vote, are more inclined to prefer a close relation with theirpolitical representative compared to their fellow citizens. In sum, voters whoprefer an individual role of representation are expected to be found on whatmight be described as ‘the margins of politics’.

The results in Table 2 partly support our main expectations: voters with alow sense of external efficacy (i.e.,who do not consider the political system tobe very responsive) are more prone to opt for the voter alternative ratherthan any of the other three options. Political interest, in turn, has a positiveeffect on the party-centred perspective. Voters who prefer a political repre-sentative that concentrates on the interests of the party that he or sherepresents are therefore more interested in politics than voters who wouldlike MPs to act directly on the wishes of his or her voters. Political activity interms of voting does not have an effect on the perspective of representationthat voters prefer.

The fourth model examines the role of political competence focusing onthe trustee role of representation, which is the group of comparison.According to our hypothesis, citizens with a high level of political compe-tence and knowledge are more disposed to prefer a trustee model of rep-resentation due to a wider understanding of the complexity of the politicalworld.As it turns out, two of the three variables included in the model showan effect that corresponds to our hypothesis. Once again it appears that twoof the four roles are negatively related to each other – namely the indepen-dent trustee role and the voter-centred role.A higher level of education andpolitical knowledge increases the propensity of voters to prefer that MPs act

© 2011 The Author(s)Scandinavian Political Studies © 2011 Nordic Political Science Association

Scandinavian Political Studies, Vol. 34 – No. 2, 2011 159

following their own judgement rather than the wishes of their voters. Inother words, voters considered to be politically sophisticated or competentare more inclined to prefer a political representative that acts indepen-dently as opposed to having a close bond to his or her voters, which is in linewith the outlined hypothesis.

To summarise, the profiles for each of the alternative role of represen-tation are relatively distinct and logical. Despite the fact that all theresults did not correspond to our hypotheses, many of the tendenciesfound pointed in the expected direction. In the final model, in which onlythe variables with a statistically significant effect in the prior analyses areincluded, it is apparent that the centre-periphery distinction is the mostimportant factor in distinguishing between voters who prioritise the aimsof the constituency over other interests. We also find that voters who givehigher priority to the party instead of the candidate are more inclined toprefer a form of representation that concentrates on the central aims ofparties. The same effect is apparent on the level of political interest. Yetwe also find that voters who have a negative sense of the responsivenessof the political system and voters with lower levels of education have atendency to opt for a close and individualised relationship between votersand MPs.

Although we did not present any clearly spelled out expectations con-cerning the two control variables included in the analyses, some interestingresults are still worth discussing. In line with previous research (Bengtsson& Wass 2010; Carman 2006, 113), we find that men have a higher tendencyto give ‘carte blanche’ to political representatives (i.e., to prefer an indepen-dent role for MPs). The strongest effect, however, concerns age, which has anegative effect in most of the analyses. This clearly suggests that the ten-dency to view the aims of the constituency as the most important perspec-tive for MPs to take into account is more common among older voters. Thelargest difference in relation to age is found among voters who give highestpriority to the trustee perspective. Moreover, voters who prioritise a party-centred role of representation tend to be significantly younger than thosewho prefer a constituency-based role of representation.

ConclusionsIn most established democracies, the process of representation is largelycontrolled by political parties. Yet in the theoretical discussion of represen-tation, party representation usually refers to the practise of representationrather than the normative model by itself (Judge 1999, 71). While somecontexts, such as the American presidential system, certainly have a moreindividualistic approach on representation and MPs therefore have betteropportunities to define their tasks, even in Western Europe with strong role

© 2011 The Author(s)Scandinavian Political Studies © 2011 Nordic Political Science Association

160 Scandinavian Political Studies, Vol. 34 – No. 2, 2011

of political parties MPs can stress different interests and strategies in theirwork (Esaiasson 2000, 51).

The main aim of this study was to examine the perspectives of represen-tational roles from the point of view of the citizens in the Finnish proportionalelectoral context with multimember districts and a strong candidate-centredvoting system. Due to its institutional setting, Finland is expected to haveseveral special characteristics with regards to representation, such as supportfor geographical representation and preference for a close link betweenvoters and MPs,on the one hand,and independent action of MPs,not the leastfrom their parties, on the other, as well as a relatively low preference forparty-dominated representation.

Our results correspond to a large extent with the hypotheses. First, ourresults further stress the finding from previous research that voters havedifferent expectations on how their political representatives should act andthat these expectations are not random. On the contrary, they vary accord-ing to patterns that are relatively logical and distinct. Citizens living inelectoral districts that are not in the centre are the most prone to regionalrepresentation.As expected, voters with closer ties to political parties preferparty-centred representation. Our results are also in accordance with theprevious interesting finding by Carman (2006, 2007) which suggests thatthose who feel less politically efficient prefer a closer relationship with theirMP. Similarly, political sophistication increases the support for an indepen-dent MP that takes on the role of a trustee.

As regards to implications on the electorate-based study of representa-tion, our results stress the need for two types of comparisons. First, the cleardeviation in support for various representative roles shows that none ofthem is totally dominant among the electorate.As representation in practiceis a multifaceted task and MPs have multiple roles depending on the issueat hand, it is to be expected that at least a certain portion of the electorateshould be satisfied with the manner they are represented. In order toperform empirical tests on the level of congruence between voters’ andMPs’ views on representation, comparable information on MPs own roledefinitions is required. Second, our study shows that voters’ preferences forpolitical representation differ substantially according to the politicalcontext. Consequently, comparative studies from other political systemswould be particularly fruitful.

ACKNOWLEDGEMENTSThis article is a part of the research projects ‘Generations and Political Behaviour: Genera-tional Effect in Electoral and Other Forms of Political Participation’ and ‘Preferred PoliticalProcesses: Voters’ Preferences for Political Decision Making’, funded by the Academy ofFinland (projects number 131701 and 136563, respectively). We would like to thank EkremKarakoc, Lauri Karvonen and the anonymous referees for their helpful comments, and PeterSöderlund for providing the database on district-level variables.

© 2011 The Author(s)Scandinavian Political Studies © 2011 Nordic Political Science Association

Scandinavian Political Studies, Vol. 34 – No. 2, 2011 161

NOTES1. The only studies of voters’ views of representation in a proportional electoral system

with multimember districts that we are aware of are those by Bengtsson and Wass(2010) based on Finnish data, and by Méndez-Lago and Martínez (2002) based onSpanish data.

2. An empirical analysis of MPs’ own definition of their task in five Nordic countriesdated in the 1990s shows that this expectation holds. In Finland, only 9 percent of theMPs considered promoting the policies of one’s own party very important. The corre-sponding figures in Sweden and Norway, which both employ party lists, were 77 and 68percent, respectively (Esaiasson 2000, 59).

3. In the Finnish parliamentary elections of 2007, 51 percent of the voters reported thatthe candidate was a more important factor in their choice than the party he or sherepresented (FNES 2007, data, FSD2269).

4. According to Esaiasson (2000, 80), it is unclear why ‘party’ is left out from Eulau’s andKarps’ (1977) categorisation of the representational foci. It might be included in thesecond category in terms of ideologically defined functional interest. Given both thetheory formation and the contemporary context in representation, ‘party’ should beincluded as a category of its own.

5. Unfortunately, as ‘functional representation’ is not included as a representational focusin our data, we are not able to test its role empirically. On the other hand, Esaiassons’(2000, 59) empirical analysis shows that at least for MPs, functional representation ingeneral plays a smaller role in Finland compared to other Nordic countries. In any case,the lack of this category is admittedly a pitfall.

6. The division between national and local interest is also closely connected to the liberaltheory of representation, which suggest that MPs should give higher priority to theinterest of the whole nation instead of local interest, first formulated by Burke (seeEsaiasson 2000, 55).As this notion of representation in terms of geographical focus hasbecome so widely accepted, it is difficult for MPs to promote more specified localinterests openly.

7. See Note 3 above.8. Denmark deviates from the other Nordic countries with a system of weak preferential

voting. Moreover, since the study by Esaiasson (2000) was carried out, Sweden hasintroduced a voting system that allows voters to cast a preferential vote. The Finnishelectoral system, however, still stands out as the most candidate-centred in the Nordiccontext, and Finland is the only country classified to utilise a strong preferential votingsystem (Karvonen 2004).

9. To understand the origins of the trustee model, it is important to place it in the rightcontext (Judge 1999, 47–54). Judge (1999, 48–9) suggests that Burke’s argument hasimplications on two levels: the relationship between an individual constituency andan individual representative, and the collective relationship of elected representativeswithin parliament. As regards to the former, Burke considers constituency interestsas objective. Consequently, representatives can pursue them without any instructionsfrom the side of the constituency. His view is also related to the conception ofpublic opinion as un-informed and unstable (Judge 1999, 49–51). At the level ofparliament, Burke argues that national interest can be formulated merely by aggre-gation of the objective, economic interests. As these interests are, however, also addi-tive, national interest can only be discovered through deliberation in which allconstituency interests are presented. To be able to reach a consensus on the interestsof the whole, representatives cannot be tied by their constituencies’ opinions (Judge1999, 53).

10. The quotas were based on age, gender and province of residence of the respondents.11. Respondents were asked to reveal their secondary preferences as well. They are,

however, not examined here. It should also be emphasised that the translation used inthis study deviates slightly from the official translation available at the 2007 FinnishNational Election Study (FSD2269). We use the term ‘aim’ instead of ‘goal’ since itseems to correspond better to the Finnish and Swedish original wording.

© 2011 The Author(s)Scandinavian Political Studies © 2011 Nordic Political Science Association

162 Scandinavian Political Studies, Vol. 34 – No. 2, 2011

12. Unfortunately, the choices given to the respondents do not include national goals orinterests. We are thus unable to study the classical controversy between national andlocal interest presented by Burke.

13. While Pitkin refers mainly to various styles of representation, the same argumentapplies also in regard to focus.

14. This excludes the electoral district of Åland, from which one representative is alwayselected.

15. Voters tend to direct their votes to parties that are likely to get at least one of theircandidates elected in order to avoid casting a wasted vote.

16. Naturally, the expectations presented here are also interconnected with the contextualexpectations presented for the party perspective, where the range of party choice ineach constituency is assumed to stimulate a party-oriented preference for representa-tion. ‘Effective number of parties’, ‘district magnitude’, ‘effective electoral threshold’and ‘proportionality’ are all different measures dealing with the electoral system. Thecausal links between these variables are not easy to unfold with the exception of‘district magnitude’, which can be placed farthest back in the chain (Söderlund et al.,2011, 96).

17. We would like to thank Peter Söderlund for generously providing us the access to hisdatabase on district-level variables for Finland. This data is used for the index ofdisproportionality and the effective number of parties.

18. The logged distance is operationalised as the logged distance in kilometres between thelargest city in each constituency and the capital Helsinki. The distance is logged sincethe importance of every new kilometre can be expected to decrease with the increasein distance.

19. Naturally, it could be argued that voters who place themselves in the middle of theleft–right continuum have a distinct ‘centre’ preference that is as real or much worth asa left or right preference. In practice, however, we know that a large share of the voterswho have difficulties differentiating between the different parties often place them-selves in the middle.

20. The other possible method – multi-level analysis – would not have been appropriategiven the small number of cases at the upper level (n = 14).

21. As the choices given to voters did not include national or supranational focus ofrepresentation, we are unable to make more nuanced interpretations concerning thenature of preferred geographical representation.

22. The first-past-the-post system is indeed candidate-centred, even if the system does notallow for voters to choose from different candidates running for the same parties(Karvonen 2004, 207).

23. Obviously, motivations behind the voter’s choice are a different aspect than the pre-ferred role of representation. It can, however, be assumed that if voters consider thegeographical area represented by the candidate an important motivation behind thechoice of candidate, they also expect representatives to emphasise it in their work.

24. The correlation between the four different contextual variables ranges from very low(0.05 for the logged distance variable and disproportionality) to very high (0.87 fordistrict magnitude and effective electoral threshold). In most cases it is higher than 0.6.

Appendix

Dependent Variable

The question ‘What kind of aims should Members of Parliament primarilypromote?’

1. Aims set by their party

2. Aims they themselves see as the most important

© 2011 The Author(s)Scandinavian Political Studies © 2011 Nordic Political Science Association

Scandinavian Political Studies, Vol. 34 – No. 2, 2011 163

3. Aims important to their constituency

4. Aims of their voters

Respondents who answered ‘don’t know’ are treated as missing (total of 18respondents).

Independent Variables

All independent variables are coded on a scale from 0 to 1.

Individual-level Variables

Gender: 1 = man, 0 = woman.Age: age in years divided by 100.Education: 1 = lower and higher academic degree, 0.8 = polytechnic degree,

0.6 = unfinished polytechnic or academic degree, 0.4 = lowest level tertiaryeducation, 0.2 = upper secondary level education, 0 = lower secondarylevel education.

Political knowledge: five questions about political matters, which were com-bined into an index counting the number of correct answers, divided byfive. The questions are as follows: ‘Who (of the following) was the FinnishMinister of Foreign Affairs during 2006?’, ‘Which party (of the following)is the second largest party in the new parliament measured in terms ofmandates?’, ‘Which country (of the following) is a permanent member ofthe United Nation’s Security Council?’, ‘Who are entitled to vote in Finnishparliamentary elections?’ (four alternatives were offered) and ‘What ismeant by a parliamentary form of government?’ (four alternatives wereoffered).

Internal efficacy: a statement: ‘Sometimes it feels like politics is so compli-cated that I don’t understand what is going on’ (1 = strongly agree, 0.66 =partly agree, 0.33 = partly disagree, 0 = strongly disagree).

Political interest: a question: ‘How interested are you in politics?’ (1 = veryinterested, 0.66 = quite interested, 0.33 = not very interested, 0 = notinterested at all).

External efficacy: an index including four different categorical variableswith four different values each. The statements are as follows: ‘MPs arebecoming out of touch with ordinary people’, ‘Politicians don’t care aboutwhat ordinary people think’, ‘Parties are only interested in people’s votes,not their opinions’ and ‘It doesn’t matter who is in power, politics won’tchange anyway’ (Cronbach’s alpha = 0.75).

Voting: voting in the Finnish parliamentary elections of 2007 (0 = did notvote, 1 = voted).

Party identification: a question: ‘How stable is your party choice?’ (1 = verystable, 0.66 = fairly stable, 0.33 = not very stable, 0 = not at all stable).

© 2011 The Author(s)Scandinavian Political Studies © 2011 Nordic Political Science Association

164 Scandinavian Political Studies, Vol. 34 – No. 2, 2011

Political orientation: respondent’s self-placement on a 11-point left-rightdimension (0 = five (mean), 1 = extreme right or left).

Party or candidate: a question about which was the more important factorfor the respondent when deciding upon how to cast his or her vote (1 = theparty of the chosen candidate, 0 = the candidate, don’t know, did not vote).

Contextual Variables

Logged distance from the capital Helsinki: the distance in kilometersbetween the largest city in each constituency and the capital Helsinki.Logged and divided by the largest number (2.91) (minimum: 0, maximum:1, mean: 0.64).

Effective electoral threshold: the effective, or exclusion threshold in each ofthe 15 constituencies in the parliamentary election 2007. Divided by thelargest number (minimum: 0.02, maxium: 1, mean: 0.42). T = (100%[M+1])/14.3, where T = effective threshold and M = the number of seatsdistributed in the constituency.

District magnitude: the number of seats distributed in each constituency.Divided by the largest number (minimum: 0.18, maximum: 1, mean: 0.57).

Index of disproportionality: the deviations of seat percentages from votepercentages in each constituency. Divided by the largest number.(minimum: 0.21, maximum: 1, mean: 0.43). D = (1/2S|vi - si|)/24.15, where D= disproportionality, v = vote shares, s = seat shares.

Effective number of parties: the number of parties in each constituency inrelation to their relative strength based on their vote shares. Based on theelection results (vote shares) of all of the contesting parties. Divided bythe largest number (minimum: 0.61, maximum: 1, mean: 0.87). N = (1/SPi

2)/5.91, where N= effective number of parties, P = each party’s proportion ofall votes.

REFERENCESBengtsson, Å. & Christensen, H. S. 2009. ‘Politiskt deltagande i Finland. Spridning och

drivkrafter’, Politiikka, 51(2), 77–95.Bengtsson, Å. & Wass, H. 2010. ‘Styles of Political Representation: What Do Voters Expect?’

Journal of Elections, Public Opinion and Parties, 20(1), 55–81.Borg, S. & Paloheimo, H. 2009. ‘Vaalipiirin koon vaikutus vaaliosallistumiseen’, in Borg, S. &

Paloheimo, H., eds, Vaalit yleisödemokratiassa. Eduskuntavaalitutkimus 2007. Tampere:Tampere University Press.

Bowler, S. & Farrell, D. F. 1993. ‘Legislator Shirking and Voter Monitoring: Impacts of Euro-pean Parliament Electoral Systems upon Legislator–Voter Relationships’, Journal ofCommon Market Studies, 31(1), 45–69.

Carman, C. J. 2006. ‘Public Preferences for Parliamentary Representation in the UK: AnOverlooked Link?’, Political Studies, 54(1), 103–22.

Carman, C. J. 2007. ‘Assessing Preferences for Political Representation in the US’, Journal ofElections, Public Opinions and Parties, 17(1), 1–19.

Castiglione, D. & Warren, M. E. 2006. ‘Rethinking Democratic Representation: Eight Theo-retical Issues’. Paper presented to ‘Rethinking Democratic Representation’, Centre for theStudy of Democratic Institutions, University of British Columbia, 18–19 May 2008.

© 2011 The Author(s)Scandinavian Political Studies © 2011 Nordic Political Science Association

Scandinavian Political Studies, Vol. 34 – No. 2, 2011 165

Converse, P. E. & Pierce, R. 1979. ‘Representative Roles and Legislative Behavior in France’,Legislative Studies Quarterly, 4(4), 525–62.

Cox, G. W. 1999. ‘Electoral Rules and the Calculus of Mobilization’, Legislative Studies Quar-terly, 24(3), 387–419.

Dalton, R. J. 2000, ‘The Decline of Party Identifications’, in Dalton, R. J. & Wattenberg, M. P.,eds, Parties without Partisans: Political Change in Advanced Industrial Democracies. Oxford:Oxford University Press.

Davidson, R. H. 1970, ‘Public Prescriptions for the Job of Congressman’, Midwest Journal ofPolitical Science, 14(4), 648–66.

Esaiasson, P. 2000. ‘How Members of Parliament Define Their Task’, in Esaiasson, P. & Heidar,K., eds, Beyond Westminster and Congress: The Nordic Experience. Cleveland, OH: OhioState University Press.

Eulau, H. & Karps, P. D. 1977, ‘The Puzzle of Representation: Specifying Components ofResponsiveness’, Legislative Studies Quarterly, 2(3), 233–54.

Eulau, H., Wahlke, W. B. & Ferguson, L. C. 1959. ‘The Role of the Representative: SomeEmpirical Observations on the Theory of Edmund Burke’, American Political ScienceReview, 53(3), 742–56.

Finnish National Election Study (FSD2269). 2007. Available online at: http://www.fsd.uta.fiGross, D. A. 1978. ‘Representative Styles and Legislative Behavior’, Western Political Quar-

terly, 31(3), 359–71.Gunlicks, A. B. 1969. ‘Representative Role Perceptions among Local Councilors in Western

Germany’, Journal of Politics, 31(2), 443–64.Heidar, K., Damgaard, E., Esaisasson, P., Hardarson, Ó. Th., & Raunio, T. 2000. ‘Five Most

Similar Systems?’, in Esaiasson, P. & Heidar, K., eds, Beyond Westminster and Congress: TheNordic Experience. Cleveland, OH: Ohio State University Press.

Jensen, T. K. 2000. ‘Party Cohesion’, in Esaiasson, P. & Heidar, K., eds, Beyond Westminster andCongress: The Nordic Experience. Cleveland, OH: Ohio State University Press.

Judge, D. 1999. Representation: Theory and Practice in Britain. London: Routledge.Karvonen, L. 2004. ‘Preferential Voting: Incidence and Effects’, International Political Science

Review, 25(2), 203–26.Katz, R. S. 1997. ‘Representational Roles’, European Journal of Political Research, 32(2),

211–26.Loosmore, J. & Handby, V. J. 1971. ‘The Theoretical Limits of Maximum Distortion: Some

Analytic Expressions for Electoral Systems’, British Journal of Political Science, 18(4), 467–77.

Manin, B. 1997. The Principles of Representative Government. Cambridge: Cambridge Univer-sity Press.

Manin, B., Przeworski, A. & Stokes S. 1999. ‘Elections and Representation’, in Przeworski, A.,Stokes, S. & Manin, B., eds, Democracy, Accountability and Representation. Cambridge:Cambridge University Press.

Marsh, M. 1985. ‘The Voters Decide: Preferential Voting in European List Systems’, EuropeanJournal of Political Research, 13(4), 365–78.

Marsh, M. 2007. ‘Candidates or Parties? Objects of Electoral Choice in Ireland’, Party Politics,13(4), 500–27.

Marsh, M. & Norris, P. 1997. ‘Political Representation in the European Parliament’, EuropeanJournal of Political Research, 32(2), 153–64.

McMurray, C. D. & Parsons, M. B. 1965. ‘Public Attitudes toward the Representational Rolesof Legislators and Judges’, Midwest Journal of Political Science, 9(2), 167–85.

Méndez-Lago, M. & Martínez, A. 2002. ‘Political Representation in Spain: An EmpiricalAnalysis of the Perception of Citizens and MPs’, Journal of Legislative Studies, 8(1), 63–90.

Narud, H. M. & Valen, M. 2007. Demokrati og ansvar. Politisk representation i et flerpartisystem.Oslo: Damm.

Oksanen, M. 1972. Kansanedustajan rooli. Tutkimus kansanedustajien suhtautumisesta edusta-jantoimeensa vuoden 1969 valtiopäivillä. Helsinki: Gaudeamus.

Patterson, S. C., Hedlund, R. D. & Boynton, G. R. 1975. Representatives and Represented. NewYork: Wiley.

Pitkin, H. F. 1967. The Concept of Representation. Berkeley, CA: University of California Press.

© 2011 The Author(s)Scandinavian Political Studies © 2011 Nordic Political Science Association

166 Scandinavian Political Studies, Vol. 34 – No. 2, 2011

Reynolds, A., Reilly, B. & Ellis, A. 2005. Electoral System Design:The New International IDEAHandbook. Stockholm: International Institute for Democracy and Electoral Assistance.

Rokkan, S. 1970. Citizens, Elections, Parties. Oslo: Universitetforlaget.Rokkan, S. & Urwin, D., eds. 1982. The Politics of Territorial Identity. London: Sage.Rokkan, S. & Urwin, D. 1983. Economy, Territory, Identity: Politics of West European Periph-

eries. London: Sage.Schwindt-Bayer, L.A. & Mishler, W. 2005. ‘An Integrated Model of Women’s Representation’,

Journal of Politics, 67(2), 407–28.Söderlund, P., Wass, H. & Grofman, B. 2011. ‘The Effect of Institutional and Party System

Factors on Turnout in Finnish Parliamentary Elections, 1962–2007: A District-level Analysis.Homo Oeconomicus, 28(1–2), 91–109.

Studlar, D. T. & McAllister, I. 1996. ‘Constituency Activity and Representational Roles amongAustralian Legislators’, Journal of Politics, 58(1), 69–90.

Taagepera, R. 1998, ‘Nationwide Inclusion and Exclusion Thresholds of Representation’, Elec-toral Studies, 17(4), 405–17.

Thomassen, J. 1994. ‘Empirical Research into Political Representation: Failing Democracy orFailing Models?’, in Jennings, M. K. & Mann, T. E., eds, Elections at Home and Abroad. AnnArbor, MI: University of Michigan Press.

Urbinati, N. & Warren M. E. 2008. ‘The Concept of Representation in Contemporary Demo-cratic Theory’, Annual Review of Political Science, 11, 387–412.

Valen, H, Narud, H. M. & Hardarson, O.Th. 2000. ‘Geography and Political Representation’, inEsaiasson, P. & Heidar, K., eds, Beyond Westminster and Congress: The Nordic Experience.Cleveland, OH: Ohio State University Press.

Wahlke, J., Eulau, H., Buchanan, W. & Ferguson, L. C. 1962. The Legislative System. New York:Wiley.

Wessels, B. 2007. ‘Political Representation and Democracy’, in Dalton, R. J. & Klingemann,H.-D., eds, The Oxford Handbook of Political Behavior. Oxford: Oxford University Press.

Zaller, J. & Feldman, S. 1992. ‘A Simple Theory of the Survey Response’, American Journal ofPolitical Science, 36, 579–616.

© 2011 The Author(s)Scandinavian Political Studies © 2011 Nordic Political Science Association

Scandinavian Political Studies, Vol. 34 – No. 2, 2011 167