parties, elections, voters

cleavages, issues and parties: acritical overview of the literaturejosep m. colomer a and riccardo puglisi baHigher Council of Scientific Research, Barcelona and Department of Economics,University Pompeu Fabra, Ramon Trias Fargas 25, Barcelona 08005, SpainE-mail: Josep.colomer@upf.edubMassachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge,MA 02139-4307, USAE-mail: r.puglisi@lse.ac.uk

doi:10.1057/palgrave.eps.2210054

Among the books reviewed in this article:

Party Systems and Voter Alignments: Cross-National PerspectivesSeymour M. Lipset and Stein Rokkan (eds.) (London and New York, Collier-Macmillan-Free Press, 1967), xviþ554pp., ISBN: 67 25332

Patterns of Democracy: Government Forms and Performance in Thirty-six CountriesArend Lijphart (New Haven, Yale University Press, 1999), xivþ351pp., ISBN:0 3000 7894 3

Agenda FormationWilliam H. Riker (ed.) (Ann Arbor, The University of Michigan Press, 1993),viiiþ285pp., ISBN: 0 4721 0381 4

Mapping Policy Preferences: Estimates for Parties, Electors, and Govern-ments 1945–1998Ian Budge, Hans-Dieter Klingemann, Andrea Volkens, Judith Bara and EricTanenbaum (Oxford and New York, Oxford University Press, 2001), 274pp.,ISBN: 0 1992 4400 6

The relation between social clea-vages, policy issues and politicalparties has been one of the most

extensively studied subjects in compara-tive political science, since at least the1960s. The subject seems to target one ofthe core areas of politics, but differentapproaches and schools of thought have

not reached a single consistent, generallyaccepted and empirically successful ana-lytical framework. We distinguish anoverview between two groups of contri-butions. The first takes political parties asthe dependent variable to be explained,by pre-existing social cleavages andissues, as in the classical works of

european political science: 4 2005

(502 – 520) & 2005 European Consortium for Political Research. 1680-4333/05 $30 www.palgrave-journals.com/eps

502

Lipset–Rokkan and Lijphart. The authorsof the second group take political partiesas an independent variable with strongexplanatory power regarding policy issuechoices, electoral campaigns and politicalparty competition, and includes differentapproaches represented by Riker and byBudge et al. Our critical review of theexisting literature concludes with anew analytical proposal integratingthe variables mentioned, in which weemphasise the initiative of party leadersas the origin of the politicisation ofissues through public policy design, aswell as the indirect formation of socialstructures.

FROM CLEAVAGES TOPARTIES

A generally recognised starting point forthe study of the subject here consists ofStein Rokkan and Seymour Lipset’s con-tributions in the late 1960s (especiallyLipset and Rokkan; Rokkan et al, 1970;see also the compilation by Flora, 1999).The usual way in which these seminalcontributions were understood impliedthat deep social cleavages, produced byremote historical events and transforma-tions, shaped the formation of politicalparties and, as a whole, the party systemin each country. National revolutions inthe French variant produced centre-per-iphery and state–church cleavages, whileindustrial revolutions in the English var-iant produced land-industry and owner–worker cleavages. Conservative and lib-eral political parties were prominent inthese processes. But new parties wereformed on the basis of the mentionedcleavages, especially ethnic–territorial,radical versus religious, agrarian, andsocialist parties. In each country, theexistence and relative strength of eachparty depended on the importance of theoriginating revolutions and the depth ofcorresponding cleavages.

Once cleavages were ‘translated’ (in atypical expression) into party systemsduring critical junctures, such as pro-cesses of democratisation involving theestablishment of mass suffrage andstable electoral systems, they were ‘fro-zen’ (also a very typical word) for a verylong term. As famously stated, ‘The partysystems of the 1960s reflect, with few butsignificant exceptions, the cleavagestructures of the 1920s’ (Lipset andRokkan, p. 50).The Lipset–Rokkan approach implied a

kind of sociological determinism. Theauthors did not even make a clear dis-tinction between social cleavages andpoliticised issues sustaining the formationof political parties. Although they men-tioned that strategic considerations re-garding organisation, elections, andcoalition formation should be taken intoaccount, they did not elaborate or evenconsider political party leaders’ moti-vations in such intermediate processes(as early noted, for instance, by Alfordand Friedland, 1974; see also Flora,1999: 46).Even more so, the bulk of empirical

literature produced by the followers ofthis approach did not take into account anumber of interesting suggestions, alter-native hypotheses and hints prudentlyintroduced by the founding authors intheir seminal works. Regarding, in parti-cular, the number of cleavages, Lipset andRokkan indeed expected voter alignmentsto be shaped ‘by such obvious socio-cultural criteria as region, class, andreligious denomination’. But they alsonoted that, in any society, there was apotential for politicisation arising from ‘agreat variety of relationships in the socialstructure’, even if many of these socialrelationships had not yet been trans-formed into occasions of political polar-isation. They even proposed to ‘considerthe possibility that the parties themselvesmight establish themselves as significantpoles of attraction and produce their own

503josep m. colomer and riccardo puglisi european political science: 4 2005

alignments independently of the geogra-phical, the social, and the cultural under-pinnings of the movements’ (Lipset andRokkan, p. 3, underlined in the original).This consideration would have been veryconsistent, in fact, with the alternativeapproach later developed in the literaturehere to be reviewed, in which politicalparties are taken as the origin ofissue salience and even cleavage forma-tion.These caveats somehow eroded the

authors’ ‘freeze’ thesis. In fact, the verylogic of the analysis suggested that new‘critical junctures’ created by unpredictedsocial transformations, as well as institu-tional changes, especially of electoralsystems, could foster significant innova-tions in party systems. When looking atcases of party systems which were shownto be ‘more fragile and open to new-comers’ than the ‘freeze’ thesis wouldhave permitted to expect, Lipset andRokkan included as exceptions France,Germany, Italy and Spain all outstandingcases in their initially small sample ofcountries. They even dared mockingsome easy, rigid implications that couldbe derived from their own vision: ‘In thefifties many observers feared the devel-opment of permanent majority parties.. .It is heartening to see how quickly theseobservers had to change their minds’.Regarding the future, they did not in factpredict stability of voter alignments topolitical parties, but rather that ‘therewill clearly be greater fluctuationsthan before’, producing not only fre-quent alternations of previously existingparties in government but also ‘newvarieties of coalition-mongering’ andparty formation (Lipset and Rokkan,pp. 50–55).Further work arising from this initial

inspiration took two main directions: (1)social cleavages as a structural explana-tion of voters’ behaviour; and (2) therelation between politicised policy issuesand the number of political parties. The

first section of the present review dealswith these two strands of the literature.

SOCIAL CLEAVAGES

Consistent with Lipset and Rokkan’s gen-eral suggestion for a ‘comparative politi-cal sociology’, the basic orientation offurther empirical work gave priority tothe discovery of social parameters andoperationalised data. They sought toconfirm and refine the ‘social cleavage’model and develop predictions about thepresence, strength, cohesion and fate ofvarious parties. Two different lines ofinvestigations were, however, developed.The first sought to prove a direct relation-ship between social cleavages and politi-cal party strength as measured by voteralignments and stability. The second ac-tually tried to prove a close relationshipbetween politicised issues (not necessa-rily related to social cleavages, althoughthis was not always clear in the analysis)and political party formation and differ-ences.

The huge amount of empirical andanalytical work that developed from thepurely ‘social cleavage’ assumption,especially through the sociological tradi-tion of political behavioural research re-presented by the Columbia and Michiganschools cannot all be reviewed here. Incomparison with the macro-structuralrelations between social cleavages andparty systems previously sketched, theanalysis was here transferred to micro-level relations between individual charac-teristics, such as race, language, religion,income or profession, and voter align-ments with one or another political party.It was, thus, taken for granted thatcleavages were given and fixed, accord-ing to the ‘freeze’ thesis. The hypothe-sised line of causality was from socialcleavages to individual characteristics,which were produced by the former, topolitical parties, which were conceived as‘expression’ and ‘representation’ of those

504 european political science: 4 2005 cleavages, issues and parties

social and individual treats. Logically, theexpected finding was stability of votingbehaviour, although this was not neces-sarily the only inference possible fromLipset and Rokkan’s approach.We do not review this body of literature

here, but only summarise some evalua-tions of the general results in electoralsociology after several generations ofscholarly endeavours. Regarding the Uni-ted States, Seymour Lipset himself veryearly on observed, ‘the existing politicalparties have found it difficult to linkpositions on the new issues to their tradi-tional socioeconomic bases of supportyparty loyalties have declined’ (Lipset,1981). Lipset, thus, somehow confirmedhis and Rokkan’s own prediction about afuture of fluctuations. But much morerecently, a general survey of the accumu-lated research still saw ‘social cleavagesas a necessary condition’, although ‘not asufficient one for the emergence of poli-tical cleavages’ (Manza and Brooks,1999). (For the sake of clarity, in thepresent review we are calling politicalcleavages ‘issues’).According to multi-country comparative

studies, also very early on since the1970s, ‘change became the normal pat-tern in many countries’. After revising aseries of contributions during the 1980s,prominent authors in the field sum-marised that ‘social structure has longbeen irrelevant to party choice’, while thedecline of cleavage politics was presentedas the removal of a straightjacket openingthe gates to initiatives for politicising avariety of issues producing massive elec-toral change. ‘Indeed, much recent scho-larship has emphasised the apparentlyincreasing independence of issue-basedvoting choice from conventional cleavagestructures’ (Franklin, Mackie and Valen,1992: 55; after revising Nie et al, 1981;Franklin, 1985; Rose and McAllister,1986; and Knutsen, 1988). More expedi-tiously, Mark Franklin concluded that‘social cleavages had (finallyy) become

irrelevant to partisanship’ (Franklin,1992: 404).

POLITICISED ISSUES

A second group of contributions focusedon the relation not between social clea-vages and political parties but betweenpoliticised issues and parties. The mainreference in a comparative perspective isthe work of Arend Lijphart (1984, 1999:Chapter 5). He still used some ‘socialcleavage’ vocabulary, stating, for in-stance, ‘a relatively large number ofparties are needed to express all thesedimensions’ (p. 89). But in his empiricalrecollection of relevant issues in 21 or 36countries (respectively, in the two edi-tions mentioned), Lijphart grouped to-gether the four basic social cleavagespreviously identified by Lipset and Rokkan(ethnic, religious, rural–urban, and socio-economic dimensions, roughly corre-sponding to the previously mentionedcentre–periphery, state–church, land–in-dustry, and owner–worker cleavages, re-spectively) plus another three dimensionsnot derived from social structures butmore directly introduced and politicisedby political entrepreneurs: regime sup-port, foreign policy, and materialist ver-sus post-materialist (also based onprevious work by Taylor and Laver,1973; Dodd, 1976; Sartori, 1976; andInglehart, 1977). Lijphart found an aver-age of more than two relevant issuedimensions per country, with values be-tween 0.5 and 3.5.The crucial point was the relation be-

tween these issue dimensions, whetherderived or not from social cleavages, andthe party system. Initially, Lijphart notedthat ‘some important issues in a countrymay not constitute issue dimensions of itsparty system, they may divide partiesinternally instead of from each other’. Butwhen proceeding to the applied analysis,issue dimensions were ‘defined in termsof differences between instead of within

505josep m. colomer and riccardo puglisi european political science: 4 2005

parties’ (Lijphart, pp. 78, 87, emphasisadded). Logically, he found a very strongcorrelation between the number ofissue dimensions in each country, asdefined in terms of differences betweenparties, and the number of differentpolitical parties.A more precise relation between the

number of issues and the number ofparties was established, on the basis ofLijphart’s first collection of data, by ReinTaagepera and Bernard Grofman (1985).They apparently still confused social clea-vages and politicised issues, as revealedin their interpretation that both ‘Lipsetand Rokkan (1967) and Lijphart (1984)may be interpreted as standing for theproposition that ‘the more axes of clea-vage there are within a society, thegreater will be the number of politicalparties’’. But, by taking Lijphart’s list ofissue dimensions for what they were –politicised issues, derived or not fromsocial cleavages – Taagepera andGrofman found a simple relation thatmade much sense: ‘parties minus issuesequals one’, or Iþ1¼P (where I standsfor issues and P for parties). Thebasic logic behind this finding is thatwhen there is a single politicised issuedimension two parties proposing alterna-tive policies define the issue as a con-troversial one. Thus, the minimum valuesin democracy would be I¼1, P¼2.A second issue can be politicised if anew, third, party takes the initiative ofintroducing a new policy proposal on thatissue alternatively to the status quo. Thepoliticisation of a third issue wouldimply that a fourth party enters thescene, and so on.Of course, it may also happen that a

new policy issue is raised by an alreadyexisting party. Conversely, there can bemore than two policy proposals (and thecorresponding parties) on an issue, or anew issue can give place to the creation ofnot one but two new parties with oppositestands. In fact, Taagepera and Grofman

found the best empirical fit for theirrelation as Iþ1¼P71. But this somehowdeviated from Lijphart’s operative defini-tion of issues as differences between, notwithin, parties. In further work, Taage-pera (1999) himself clarified the ques-tion. First, he referred to issues as ‘thenumber of social cleavages that arepoliticised’, noting that ‘social heteroge-neity is not the same as political hetero-geneity. The former deals with potentialcleavages, the latter with the actuallypoliticised ones’. Although he still men-tioned that ‘low heterogeneity puts a lidon the number of partiesy because therewill be no demand for many parties’(emphasis added), he also noted – fromwhat could be rather called, in contrast, a‘supply-side’ approach – that ‘some poli-tical issues do not reflect pre-existingsocial heterogeneity’. The number ofissues can, thus, be higher or lower thanthe number of social cleavages. But it willcertainly be related to the number ofparties. Taagepera concluded that if ‘po-liticised issues mean issues on whichsome parties disagreey the connection(between the number of issues and thenumber of parties) may be tautological’(Taagepera, 1999: 545).The substantive findings of the social

cleavage literature can be summarised inthe following way. First, causality fromsocial cleavages to the creation andstrength of political parties has not beenproved, but rather dismissed and foundirrelevant, probably because the structu-ral-deterministic hypothesis did not ser-iously consider the intermediate strategicstage of politicising cleavages into issuesand building the corresponding organisa-tions or coalitions. Second, the correla-tion between the number of politicisedissues and the number of political partiesis strong (of the type Iþ1¼P71), butsomewhat tautological since, in the em-pirical data used, issues had beendefined precisely as differences betweenparties.

506 european political science: 4 2005 cleavages, issues and parties

Electoral sociology, even after a long-term sustained effort with sophisticatedanalytical techniques and abundant em-pirical data, seems to have abandonedthe hypothesis that social cleavages canexplain individual votes for political par-ties. In some of the other cases, theweakness of positive results can be dueto the number of variables considered,the data available or the operationalisingmethods used. But none of the disparatescholarly contributions gathered and re-viewed above, all of them trying toexplain the formation, survival andstrength of political parties as dependenton some social structural variables, hasachieved conclusive results.

FROM PARTIES TO ISSUES

A completely different approach has triedto explain the number and selection ofpoliticised issues in elections, as well asits role in explaining electoral competi-tion, post-electoral coalition formationand government performance, not asderived from social cleavages or anyother similar structural variable, but frompolitical parties’ strategies. In the follow-ing, we will refer to ‘political parties’ in thelimited sense of organisations driven byelectorally oriented leaders. This implies,a vision of political parties as organisa-

tions developing an activity directly ad-dressed to voters in search of their votes,but also having a crucial role in thedesign, implementation and evaluationof public policy. We do not consider hereother important aspects of political par-ties, such as their different forms of massor activist organisation or other institu-tional features.Four different kinds of contributions are

considered, coming from the rationalchoice school in political science, the insti-tutionalist branch of political economy,electoral campaign and media studies,and the party manifestos project.Initially, the ‘agenda formation’ model

(Riker, 1983, 1986, 1993, 1996) emergedas an alternative to Downsian spatialmodels of electoral competition (Downs,1957). Roughly speaking, in the Down-sian models it is assumed that bothvoters’ preferences and the issue policyspace (whether one-dimensional or multi-dimensional) are exogenously given, whileparties or candidates choose policy-ideological ‘positions’ in the availablespace. The Rikerian agenda model, incontrast, assumes that voters’ prefer-ences and party positions are basicallygiven (as constrained by an ideological‘argument’) and then party leaderschoose issue dimensions to be givensalience in order to shape the policyspace.Some elements in this approach had

already been sketched out by Stokes(1963) in an early critique of Downs’theory. Based on his and other colleagues’empirical findings in the field of electoralsociology mentioned above (especiallyCampbell et al, 1960), Stokes remarkedthat the electoral policy space tends to bemulti-dimensional and it does not have astable structure. In order to understandelectoral competition, he recommendedthat ‘different weights should be given(to) different dimensions at differenttimes’. Stokes even sketched the strate-gic argument: ‘The skills of political

‘Electoral sociology,even after a long-termsustained effort with

sophisticated analyticaltechniques and abundantempirical data, seems to

have abandoned thehypothesis that socialcleavages can explainindividual votes forpolitical parties.’

507josep m. colomer and riccardo puglisi european political science: 4 2005

leaders who must maneuver for publicsupport in a democracy consist partly inknowing what issue dimensions are sali-ent to the electorate or can be madesalient by suitable propaganda’ (Stokes,1963: 372).More specifically, Stokes made an ana-

lytical distinction between position- andvalence-issues, which would clarify someof the further discussion. For position-issues, there is a set of policy alternativesthe parties can take and for which adistribution of voters’ preferences can bedefined – for instance, on trade or schoolissues. Stokes suggested that, on theseissues, the Downsian analytical frame-work, in which parties choose policypositions, would be appropriate. For va-lence-issues, in contrast, there can onlybe a dichotomous positive–negative evalu-ation. An ‘overwhelming consensus’ isassumed to be the goal of governmentaction on these issues: all parties andelectorate want it – for instance, peace orprosperity. Party competition on thoseissues consists in claiming credit or get-ting blamed for it (based on the party’sprevious government record and the like-lihood of its performing well in the future).Interestingly, Stokes acknowledged thata position-issue can become a valence-issue (for instance, a consensus for lesstaxes can be created) and vice versa(peace at any cost, for instance, cansuddenly become a divisive issue). Also,voters can change their minds regardingwhich parties can be positively or nega-tively associated with each issue.To summarise according to the Down-

sian model of two-party competition,we should expect that parties will con-verge in their positions on position-issues. But, according to Stokes’contribution, parties will coincide onwanting more of consensual valence-issues. So, in general, no confrontationbetween highly different policy positionscould be expected in electoral competi-tion in the long term.

A number of crucial points in thisapproach which have been further devel-oped are discussed in the followingpages: (1) the origins of policy issueproposals and preference formation; (2)the role of parties and ideologies in policydesign; (3) the implications for electoralcampaigns; and (4) the evaluation ofpolicy performance and the subsequentattachment of certain issues to specificparties.

THE ORIGINS OF POLICYPROPOSALS

An issue may become salient in voters’perception if it is known that problemsrelated to it have occurred which deservesome policy action (note that ‘agenda’,which literally means ‘things to be done’,has the same Latin root of ‘action’ and‘agent’). The status quo policy might havebecome unsatisfactory, even if it has beenstable for a long period. Durable dissatis-faction can be the result of the fact that asingle vote on multiple policy issues mayimpede the implementation – on lesssalient issues – of policy positions thatare preferred by a majority of voters (aDownsian insight noted, for instance, byBesley and Coate, 2000, 2003). Dissatis-faction may also derive from changingcircumstances, such as technology ormigration, causing a traditional policy toproduce new unintended, undesirableeffects. But all of this also implies thatvoters’ preferences may not be exogen-ously given but formed in the process ofchallenging the status quo and giving newsalience to an issue.For a political party or entrepreneurial

politician, giving salience to an issue(which essentially means talking about itand making it news) implies taking a new‘position’ on the issue itself in contrast tothe presumably unsatisfactory status quopolicy, as well as framing the new positionor policy proposal with some value or argu-ment. Only the presentation of alternative

508 european political science: 4 2005 cleavages, issues and parties

policy proposals induces the ‘activation’of relevant voters’ preferences. Voterscan then form their preferences overdifferent policy alternatives on an issueby comparing the status quo policy on theissue and some new policy proposal.Thus, voters’ clearly defined preferencescannot exist in the absence of alternativepolicy proposals and the correspondingissue salience. If no new policy proposal ispresented, the issue will not be salient,but then voters will not have preferenceson the issue either. This line of reasoningis consistent with the idea that voters asdecision makers have a limited amount ofattention to devote to the formation ofpreferences regarding policy issues. Oncean issue becomes salient thanks to a newpolicy proposal, voters will dedicate sometime and attention to defining their pre-ference.William H. Riker once made a distinc-

tion between ‘heresthetics’ as the art ofselecting issues and ‘rhetoric’ as the art ofarguing about the issue by means ofpersuasive values. Regarding the latter,it has been frequently remarked that twotypes of arguments exist: ‘negative’ ar-guments oriented to rejecting the status(quo) and which, according to standardpsychology, are likely to be given rela-tively high weight by voters; and ‘posi-tive’ arguments for choosing newproposals, which may be accepted bydefault. (cf. in Bailey, 1969; Davis andFerrantino, 1996; Riker, 1996; Sartori,1998).It should be possible to develop similar

spatial models for each of the two manip-ulative strategies just mentioned. In amulti-dimensional issue space, for eachissue-dimension there can be a multi-dimensional value space. For instance,governmental social spending and abor-tion can be two salient issues in anelection forming a two-dimensional issuespace. But for the issue of social spendingone party may want to emphasise itscharacteristic of social investment in

favour of the poor while the other partymay give salience to the implication that itrequires new taxes, thus creating a two-value dimension for the issue. On theabortion issue, one party will give sali-ence to the value of ‘freedom’ of choice,while another party will try to make thevalue of ‘life’ more salient. Giving a newissue salience may make the space one-issue dimensional, but this might be onlythe first step in a process in which theparty disadvantaged on that issue willreact by giving salience to an alternativevalue on the issue, by this way creatingnew multi-dimensionality in the space. Insuch a framework, rhetoric can be con-sidered as a further development ofheresthetics, within a single issue thatcan be framed in different ways.In this approach, the formation of the

public agenda is explained as the result ofendogenous salience structures based oninformation and messages, without hav-ing to include previously formed voters’preferences or exogenous preferencechanges. The selection of salient issuesis not determined by pre-existing socialcleavages, but the result of politicalparty’s or entrepreneurial politician’s in-itiative to provide information, news andvalues on some potentially politicisedissues.

PARTY AND IDEOLOGY

It is mainly the parties that choose newissues in their platforms and public de-bate. Each party, apart from decidingwhich position to hold on each issue, hasa fixed endowment of ‘effort’ (measurablein terms of time, money, personnel,organisation) that can be distributedacross issues. The share of effort devotedto an issue can determine the weightgiven by voters in evaluating the party’sposition on that issue (Cantillon, 2001).In addition, parties may also want toselect candidates who are more compe-tent on the issue they want to highlight.

509josep m. colomer and riccardo puglisi european political science: 4 2005

This could be a way to commit credibly todevote effort to a given issue.As remarked above, voters’ preferences

and choices are also determined by thelimits of their resources, in this caseinvolving memory and capacity of atten-tion span. They are likely to give moreweight to repetitive, intense messages,as is often remarked by cognitive psychol-ogy (see comments by Riker, 1983). Formany voters, a reasonable hypothesis isthat salience suggests the pre-commitmentof the party, if it arrives in government,to dedicate the corresponding fraction oftime and resources to the issue. Analternative hypothesis more in accordwith traditional Downsian assumptions,would be that a voter can choose a partyon the basis of the voter’s concern for anissue and the party’s position on the issueeven if it is not raised or giving salience bythe party. However, it would be highlyrisky for a voter to vote for a party on thebasis of non-salient issues because it isobviously likely that the party, once ingovernment, will not pay much attentionto them and will spend more effort onissues in which it has committed itselfduring the electoral campaign.Parties’ innovative proposals are, thus,

limited, especially by the party systemitself and the communication role ofpolitical ideologies. First of all, policyinnovation is limited because a party sellspackages on several issues at the sametime. This implies that a system of policychoice based on political parties’ initiativemay be unsatisfactory for many citizenson many policy issues. New issues can bedeveloped inside old parties or createthe occasion and motive for the formationof new parties – as already suggestedby Lijphart’s comments. Which of thetwo alternatives will occur strongly de-pends on the electoral and institutionalsystem.Also, simple institutional frameworks –

such as a single-chamber parliamentaryregime with single-member districts – in

which a single-seat election is decisive forthe composition of both the parliamentand the cabinet, as well as for all the set ofcorresponding policies, is likely to bepotentially highly multi-dimensional andopen to issue innovation. In contrast,within institutional frameworks involvingdivision of powers and decentralisation,each election for a separate institution(say, the presidency, each of the cham-bers of parliament, regional and localgovernments) is likely to deal withrelatively low numbers of potentialissues and make innovation more diffi-cult. Ultimately, if there were a singleissue for each election because thedivision of powers among differentinstitutions made policy decisions highlyfragmented, no choice of issues would bepossible. The standard Downsian modelof a single-issue space with only aposition-taking strategy would applyvery well.The second limitation on policy innova-

tion derives from the role of ideology. Therelation between policy issues and politi-cal ideology has been systematically ex-plored in extensive empirical analyses bythe Manifesto Research Group led by IanBudge (Robertson, 1976; Budge et al,1987, 2001; Laver and Budge, 1992).Even in an internally highly flexible partyor in a multi-party system with low entrycosts, policy-issue innovation is limited bythe encompassing, rigid role of politicalideologies. Although ideologies may notgive detailed guidance of which positionto take on the issue in policy space, theydo indicate the general policy ‘area’ (in aspatial sense) that a party should occupy.On the basis of general ideologies, partiesrecognise each other’s spatial boundariesand, then, ‘parties cannot move muchbeyond the centre, nor change theirrelative positions to left or to right,because of the confusing effects thiswould have on electors and the lack ofcredibility of a party which repudiated itspast commitments, not to mention the

510 european political science: 4 2005 cleavages, issues and parties

policy beliefs of leaders themselves’(Budge, 1994: 451).In fact, this is a very Downsian argu-

ment. According to Downs, the relevanceof the left-right (or a similar) dimensionfor a high number of economic, social,moral and cultural issues is largely due tocommunicative restrictions imposed bythe existence of mass electorates andmedia. It would not be rational for a voterwith very small influence on the electoralresult to pay high costs for obtainingdetailed information about each party’sstands on each issue. General ideologicaland symbolic messages may providesufficiently good hints and cues to makea voter’s choice probably correct.But a party’s ideological consistency

can produce cognitive dissonance amongvoters. Certain voters can find it hard tomanage instances in which their agree-ment with a party is partial, in the sensethat they agree on some issues and theydefinitely disagree on some other issues.Instead of weighting pros and cons ondifferent issues when making theirchoice, some voters may prefer to em-brace a given political party, which mayentail dissonance on some issues.At the same time, communicational

requirements prevent parties from adopt-ing disparate or apparently contradictorypositions on different issues and forcethem to take rather predictable positionswhen a new issue emerges in order to beunderstood by the electors as beingconsistent with the party’s previous posi-

tions on other issues. The need to main-tain ‘ideological consistency’ in order tokeep their members together and com-municate in simple ways with voterslimits parties’ capabilities to innovate orfight successfully on certain issues be-cause they cannot take the most popularposition on them.Ideological consistency, thus, con-

demns some parties to appearing asdisadvantaged on certain issues andtherefore not interested in emphasisingor giving them salience. However, assome authors remark, the specific policyposition contents of ‘left’ and ‘right’ or of‘progressive’, ‘liberal’ and ‘conservative’global ideological positions are acciden-tal: ‘There is after all no logical or inherentreason why support for peace (for in-stance) should be associated with govern-ment interventionism (also for instance)’(Budge et al). What is less accidental isthe convenience to maintain predictableand relatively stable positions on eachissue and on the encompassing ideologi-cal dimension, in order to be able to offerunderstandable ‘packages’ to the votersover time.The analysis of party electoral manifes-

tos in twenty-five countries during theperiod 1945–1998, which relies heavilyon ‘intensive reading of the texts them-selves’, distinguishes fifty-seven policyissues and a limited number of ideologicaldimensions. These are the politicisedissues that take salience in electoralcampaigns, in a similar vein to Lijphart’spreviously mentioned issue dimensions,which claimed to capture all relevant or‘salient’ issues in elections. Budge and themembers of the Manifesto ResearchGroup distinguished seven groups ofissues (close to but not exactly corre-sponding to Lijphart’s seven dimensions),roughly characterised as: minority groups,morals, the economy (including agrarianprotectionism), social issues (includingenvironment), political regime, foreignpolicy, and government effectiveness

‘Even in an internallyhighly flexible party orin a multi-party systemwith low entry costs,

policy-issue innovationis limited by the

encompassing, rigid roleof political ideologies.’

511josep m. colomer and riccardo puglisi european political science: 4 2005

(the latter not included in Lijphart’s) (see,for instance, Budge et al, Table 3.2).As can be seen, these distinctions be-tween groups of politicised issues werenot necessarily related to pre-existingsocial cleavages, since several of themwere strictly related to government per-formance.By using factor analysis, the Manifesto

Research Group tried to identify howmany issue dimensions, that is, groupsof issues on which parties could bedistinguished, were relevant in eachcountry. Budge and his co-authors haverepeatedly emphasised that the left-rightideological dimension is the most com-mon one across countries and they haveprovided the corresponding relative partypositions over time. On all issues partiestend to locate themselves on the samerelative part of the spectrum with respectto other parties (no ‘leap-frogging’),although they do not follow any patternof convergence or divergence in the longterm (Budge et al; see also Laver andHunt, 1992; Laver, Benoit and Garry,2003). ‘In most countries, there is nosteady movement to convergence ordivergence; parties come together andmove apart presumably in response toimperatives of party competition, not tosecular trends towards deideologisation’(Budge, 1993).In the short term, during election cam-

paigns, parties can fight on several di-mensions at once by choosing differentissues to emphasise. No single number ofissue dimensions has been found appro-priate to make generalisations acrosscountries. Initially, a major finding wasthat ‘the optimal spaces for each countrywere never less than three-dimensional,and sometimes four- or five-dimensional’(Budge et al, 1987; Chapter 18; Budge,1993: p. 58). Further research has pre-sented results in two (Robertson, 1976;Schofield and Laver, 1985; Miller andSchofield, 2003), three (Warwick, 2000),or five issue dimensions (Budge et al,

1987; see survey by Budge et al, 2001).According to these analyses, the inde-pendence of the number of issues politi-cised by political parties in electoralcampaigns from social cleavages or otherpre-existing structural variables seems tobe complete.

ELECTORAL CAMPAIGNS

The literature reviewed here supports thehypothesis that, in electoral campaigns,parties devote more effort to persuadingvoters that some issue should be ‘salient’in their decision rather than to confrontdifferent policy proposals on any issue,that is, party ‘positions’ in the typicalspatial approach. In the public debate,using campaign advertisements andmedia messages, each party seeks togive salience to those issues on which itis more credible and expects to obtainvoters’ attention and votes. It follows thatparties do not debate policies but simplytry to give salience to different issues,even using ambiguous statements re-garding their policy proposals.No party is, of course, in full control of

the environment, in which other parties,pressure groups, unexpected events andthe media contribute to shaping the publicagenda. From this point of view, twotypes of campaign agendas could bedistinguished, depending on whetherthey are endogenous to party competitionor exogenous if imposed by externalevents, pressure groups or independentmedia.For endogenous formation of campaign

agendas, Riker distinguished two strate-gic principles. By the ‘dominance princi-ple’, the party insists on an issue wherethe party proposal proves to be success-ful; by the ‘dispersion principle’, the partyabandons an issue where the party pro-posal fails in attracting voters’ attentionor support. These party strategies pro-duce electoral campaigns in which fail to

512 european political science: 4 2005 cleavages, issues and parties

discuss with one other – so they talkabout different issues and change thesubject when are explicitly challenged(Riker).Several electoral tactics can be identi-

fied in the continuing effort to attractattention to some preferred issue. Onemay include personal attacks of provoca-tion on the rival party or candidate inorder to divert their attention or put themon the defensive and then fill the corre-sponding vacuum with the preferredmessage. Also, a party can create oppor-tune events, which may go from the usualtown visits and press conferences to bookpresentations and artfully provoked inter-national conflicts.Similarly, Budge underlined that elec-

toral campaigns are mostly about sal-ience, not confrontation, therefore lea-ding to no real debates (see also Simon,2002 for empirical support). Somewhatmore strongly, Budge even held thatparties ‘rarely take specific policy standsat all’, but, at the same time, emphasisesome policy areas because ‘their credibil-ity on that position is strong enough topick up votes’ or the party is ‘committedand hence most trusted by electors’ (seeBudge et al, 1987, 2001).In ‘exogenous’ campaign agendas,

however, a party may face itself forcedto deal with an issue on which it has adisadvantaged position – say, for in-stance, after a big scandal, massivepopular protests, a terrorist attack or anexternal war. Then, given that the topicmay have received overwhelming weightin news and voters’ perception, the bestresponse for the disadvantaged party is togive salience to the least unfavourablevalue attributes within the topic itself. Asalready mentioned, the correspondingdiscussion on different values for thesame issue will create a new multi-dimensional value space within the one-dimensional issue space. Parties will talknow on the same issue, but still withoutdiscussing it with one another.

Electoral campaigns are also charac-terised by a prominent role of massmedia in forming the agenda. As wasalready remarked by Cohen (1963), themedia ‘may not be successful much ofthe time in telling people what to think,but it’s stunningly successful in tellingits people what to think about’. McCombsand Shaw (1972) is the seminal empiricalcontribution in which such concept ofagenda-setting effects of news coveragehas been put to test. Further empiricalstudies have adopted a wide range ofresearch designs, from cross-sectionalsurveys to aggregate time series ana-lysis, from repeated cross-sections tocontrolled experiments. The broadmessage stemming from this literature,even if with some internal variation, isthat agenda-setting effects on publicopinion are indeed sizeable. However,there are several causal links connectingvoters’ real-world experience, themedia agenda, and the voters’ concerns.Real-world cues and experience influenceboth the media and the voters’ concerns,while the media, if the theory of agenda-setting is correct, has a strong andseparate influence on the saliencestructure of the voters. Additionally,political parties can have clear incentivesto alter the set of news that votersreceive. Political leaders may try andmanipulate media outlets by buying theirsilence on the bad news; a more subtleway of obtaining the same result is tomake the story on the preferred issuemore palatable to the taste of mediaeditors or journalists (Besley and Prat,2004; Puglisi 2004a). From an empiricalpoint of view, Puglisi (2004b) in factshows that the New York Times, overthe time period spanning from 1946 to1994, systematically gives more cover-age during presidential campaigns to theDemocratic topics of health care, civilrights, labour and social welfare, but onlywhen the incumbent president is aRepublican.

513josep m. colomer and riccardo puglisi european political science: 4 2005

POLICY PERFORMANCE ANDPARTY ADVANTAGE

When a new policy issue is given saliencein order to attract voters’ attention andvotes, three alternative outcomes mayhappen. First, the party may fail in itsendeavour, never arrive in governmentand not be able to implement the policy.Second, the party may win sufficientsupport to enter government and imple-ment the policy, but this may produceunexpected or undesirable effects, caus-ing voters to prefer again the previousstatus quo or a similar position. Finally,the new policy may be successful in thesense of being satisfactory for the citizensand this may reinforce the party’s elec-toral support. In the first and the secondoccurrences, which both imply a policyfailure, it is likely that the party, havingpromoted salience for the new issue, willbe either electorally weakened in futureelections or withdraw its policy proposal.The previous status quo policy will pre-vail. In the third occurrence, as the newpolicy will be implemented with widepopular acceptance, it is likely that theother party or parties will lower thesalience of the issue or even adapt theirpositions on the issue to the new statusquo. A new policy consensus may becreated and as a consequence, the issue

will also lose salience in future contests.In the long term, thus, whether newpolicy proposals fail or succeed, weshould expect increasing policy consen-sus among political parties. The numberof issues potentially to be politicised andgiven salience by offering new policyproposals will always be very large, butan increasing number of them will besuccessively discarded from the electoralcontest.At this stage, we can compare again the

implications of this model of agendaformation with some basic elements inclassical Downsian spatial models of elec-toral competition. In the latter, in whichthe basic strategy is the party’s choice ofpolicy positions, the main results areeither policy convergence on a singleissue-dimension or chaotic trajectoriesor unpredictably changing policies in amulti-dimensional issue space (see Grof-man, 2004 for a revision). In the model ofagenda formation, in contrast, a multi-dimensional issue space does not nece-ssarily produce policy instability. It isexpected that parties will select someissues to be given salience, while voterswill vote on the basis of the vector ofweights they attach to the differentissues. In spatial terms, the sequenceimplicitly assumed in the agenda modelcan be presented with the followingsteps:

(1) There is a status quo policy on anissue which is non-salient, on whichvoters have no real preferences (be-cause they have no alternatives tocompare), and on which there is noparty competition.

(2) An innovative party chooses a newissue ‘position’, that is, a new policyproposal in contrast to the status quoand give it salience and value.

(3) Voters form their preferences bycomparing the status quo with thenew policy proposal. It is in thesubsequent distribution of voters’

‘Political leaders may tryand manipulate mediaoutlets by buying their

silence on the bad news;a more subtle way of

obtaining the same resultis to make the story on

the preferred issue morepalatable to the tasteof media editors or

journalists’

514 european political science: 4 2005 cleavages, issues and parties

preferences that a ‘median voter’emerges. All the previous stages arenot contemplated by the classicalspatial models, which assume thatthe issue space and the voters’ pre-ferences are given.

(4) It is only at this stage that basicassumptions of Downsian models be-gins to apply. On a new salient singleissue there is party distance andperhaps polarisation. One party –typically the closest one to the med-ian voter – has electoral advantageand wins.

(5) The other party, whether a defeatedprevious incumbent or a failed chal-lenger, may unilaterally converge tothe new winning position. A newconsensus can be formed, as alsopostulated by Downs. But note thatin this approach convergence andconsensus may to be formed arounda vague political ‘centre’ defined by apreviously given median voter’s posi-tion. If the challenge of the innovativepolicy proposal turns out to be suc-cessful, the new consensus will beformed around the new winning posi-tion which will have proven to be ableto attract popular support and definea new median voter.

(6) The issue in question will loss sal-ience. A new issue space is likely to beshaped for the next election.

This dynamic sequence suggests that,in a static analysis, different policy issuesmay be ‘owned’ or ‘leased’ by someparties (as in the classical classificationby Petrocik, 1996). But – as we willdiscuss below – they may also be eithercontroversial or consensual.In most of the literature reviewed here,

it is implicitly assumed that the partypolicy stand is sufficiently well known bythe voters to give the party ‘credibility’and ‘trust’ for its enforcement. In otherwords, it is usually taken for granted thata policy proposal already exists, that it

has proven successful or has gained avalue-argumentative effort of persuasion,and, as a consequence, has the support ofall or at least a majority of voters. It isalso assumed that the successful policyproposal is attached in the eyes of thepublic to one of the parties, presumablythe one having originally introduced theproposal and/or having transformed itinto successful government policy.Hence, leftist parties like the socialists

(but also the Christian Democrats) tendto favour social welfare policy issues,while the liberals emphasise free-marketeconomic efficiency, and the conserva-tives prefer to give salience to defence,foreign and interior policy issues (seeBudge and Keman, 1990). In the UnitedStates, ‘Democrats have an electoraladvantage when problems and issuesassociated with social welfare and inter-group relationships are salient. Republi-cans have an advantage when issuesrelated to taxes, spending, and the sizeof government are high on the publicagenda,’ (Petrocik et al, 2002). Otherempirical analyses have also postulatedthat different issues are ‘owned’ by dif-ferent parties, as developed, for instance,by Carmines and Stimson (1989) regard-ing the issue of race in the United States.In a very long term perspective, theabsence of direct rivalry between partieson each issue (although nothing thatparties can change their emphasis ondifferent issues over time) has also beenassumed by Miller and Schofield (2003);see also Schofield et al, 2003).By analysing news content, answers to

open-ended questions about issue sal-ience, and the vote itself for US presiden-tial elections in a long-term period,Petrocik shows that candidates indeedtend to emphasise ‘owned’ issues in theirpolitical speeches. If issues owned, forinstance, by the Democratic party aresalient, Republican voters are less willingto go to the polls and vote for theircandidate, independents are more willing

515josep m. colomer and riccardo puglisi european political science: 4 2005

to vote for the Democratic candidate, andfinally democratic turnout increases, to-gether with their vote for their partisancandidate. Of course, the opposite partyadvantage also applies.While ‘owned’ issues are those in which

a party is reckoned as more capable on along-term basis, ‘leasing’ may exist onissues in which, for instance, the incum-bent has performed acceptably well on ashort-term basis. Performance issues,such as the conduct of government offi-cials, the state of the economy, or thecountry’s status and security amongother nations, are not automaticallyowned by a single party, but can providean advantage to a candidate whenevents, official behaviour, and policy fail-ures allow the candidate to claim creditfor good times or blame the opposition forbad times (Petrocik et al, 2002).In this approach, ‘party ownership’ of

issues results from the accumulation ofpositive policy performances over time.This analytical framework may evokesome rationalist revision of the old theoryof ‘party identification’ by voters, in whichit is conceived as deriving from theaccumulation of positive retrospectivevoting (see Fiorina, 1977). This approachmay explain that certain issues mayappear as temporarily ‘owned’ by a party,but, after a perhaps unexpected big fail-ure – which could be produced by tech-nological or population changes or otherfactors – issue ownership can alsochange. As already noted in Stokes’seminal contribution mentioned above,even ‘valence-issues’ enjoying broad con-sensus may become controversial posi-tion-issues. Also, new governmentalperformances may modify a party’s tradi-tional advantage on an issue.Just as happened with ‘party identifica-

tion’ in the above-mentioned interpreta-tion, so ‘party ownership’ can come toimply an incumbent’s advantage. Theincumbent party or candidate may obtainadvantage in electoral competition from a

higher capability for providing favourableinformation and good news on the pre-ferred issues, as well as the possibility ofhaving performed temporarily well onsome unowned issues. However, the in-cumbent may also have some disadvan-tages. For certain issues, values implyingcriticism of governmental interventionismor of rising taxes can backfire against theincumbent (Jacoby, 2000). In general –as also noted in cognitive psychology – aparty in government can be rewardedmuch less for good performance than itis punished for bad performance (Ansola-behere and Iyengar, 1994).In contrast to issue ‘ownership’ and

‘leasing’, no comparable attention hasbeen devoted to the permanence of con-troversial policy proposals that can stillsplit the electorate after their initialemergence from an inconclusive govern-mental performances. This, however,could explain party polarisation and thecorresponding alternation of winning par-ties and policies on a single issue, which iscertainly a rather common occurrence.On some issues not yet settled, ambiguityregarding policy effects, controversy inpublic debate, uncertainty about voters’support, and party alternation in govern-ment to implement alternative policiesmay subsist. Political competition mayinflame conflicts and generate polarisa-tion on those issues. Precisely becausemany differences between party plat-forms tend to disappear, a few potentiallycontroversial differences are exagger-ated. Although it could be assumed thatthese features would correspond to ‘pri-mitive’ stages in the formulation of in-novative policies, they also seem to existin reality even for very ‘old’ issues such asfamily values and sex issues, for instance(Fiorina et al, 2004).

SUMMARY

In the second group of contributions herereviewed, politicisation of issues appears

516 european political science: 4 2005 cleavages, issues and parties

as the work of entrepreneurial politiciansin their task of launching policy proposals,building parties or candidacies, persuad-ing voters and receiving their votes. Thepotential for such activity seems to beimmense, since it does not depend onprominent structural characteristics ofthe society. It can encompass literallyany topic, subject or aspect of human lifeable to be regulated by public enforceabledecisions, on which some alternative tothe status quo can be proposed. This mayinclude both issues that have not been inthe public sector before and ‘privatisation’of issues traditionally submitted to publicregulation. As Ian Budge and Judith Bararemark: ‘Most investigators in this areawould probably agree that the ‘true’ policyspace is composed of as many dimen-sions as there are political actors andpublic preferences held by them – form-ing an underlying space of almost infinitedimensions therefore (especially if wetake private preferences into account,whose translation into public onesis difficult and chancy)’ (Budge et al,p.59).

CONCLUSION

A number of contributions to the concep-tualisation and analysis of the relationsbetween social cleavages, policy issuesand political parties have been reviewed.Two bodies of literature have been dis-tinguished, according to whether theytake political parties as dependent orindependent variable in their analyses.From both schools, we can clarify theconceptual difference between socialcleavages and policy (‘politicised’) issues,as well as the connection between thelatter and political parties.The first group emphasised a line of

causality from social cleavages to politicalparties. By contrast, the findings of theliterature reviewed in the second part canbe summarised the following way:

(1) Social cleavages do not appear in theanalytical framework. It could be in-ferred from the analysis that they area product rather than the origins ofpublic policy decisions and enforce-ment.

(2) The potential of issues to be politi-cised in a complex society with aworking government is immense, notrestricted to a handful of controver-sies nor determined by deep clea-vages associated to long-term socialstructures.

(3) Policy innovation and the correspond-ing party advantage are limited bythe role of political parties in providing‘packages’ of policies and the require-ment of developing not very costlycommunication with voters throughideological consistency. Partiesplace themselves on a policy areaalong an ideological dimensionencompassing multiple issues, suchas the left-right axis, and have to staywithin.

(4) In electoral campaigns and, moregenerally, in shaping the public agen-da, parties choose to give salience tothose issues on which they have anadvantage before the electorate.Such advantage is typically based onthe trust and credibility their previousperformance has gained them. In anunstable multi-dimensional policyspace parties can give salience todifferent issues. A single-issue dis-cussion is also likely to become amulti-dimensional value space.

(5) On the basis of some previous pro-cess and experience, different policyproposals are submitted to the proofof public evaluation. If successful,they are associated with specific poli-tical parties, augmenting their trustand credibility.

This approach might enlighten the suc-cesses, failures and general evolution ofpublic policies and related political parties

517josep m. colomer and riccardo puglisi european political science: 4 2005

and ideologies over time. While policyconsensus may emerge on a number ofissues, there will be polarisation on otherissues. In any case, there will alwaysremain a number of non-salient but latentissues, potentially to be politicised, andthus open to further policy innovation andchange.In this perspective, the general relation

between cleavages, issues and partiescan be re-stated the following way. It ispolicy issue design and implementationthat can explain the formation of socialcleavages and structures, rather than theother way round. In contrast to thetraditional sociological approach, the hypo-thetical line of causality can be reversed,since it can be expected that politicisedissues in the form of policy proposals, ifthey become effective public policy, willcreate incentives for citizens, groups andcompanies to react that will eventuallytransform some social structures. Or, toparaphrase some statements in the ‘so-cial cleavage’ approach, new ‘criticaljunctures’ can be created by partiesunpredicted policy issue proposals andsalience.

The number of politicised issues cannotdirectly explain the number of politicalparties, since policy innovation may de-velop either within or between parties.Which of these alternative developmentswill be chosen highly depends on theelectoral system – a topic that has notbeen reviewed here – but it is likely thatelectoral systems will also be chosen bypolitical parties in order to permit orprevent the creation of new parties. Therole of party leaders’ initiative in bothpoliticising issues and creating parties hasbeen emphasised in what could be la-belled a ‘supply-side’ approach to politicalprocess.

ReferencesAlford, R. and Friedland, R. (1974) ‘Nations, parties and participation: a critique of political sociology’,Theory and Society 1: 307–328.

Ansolabehere, S. and Iyengar, S. (1994) ‘Riding the wave and claiming ownership over sigues. the jointeffects of advertising and news coverage in campaigns’, Public Opinion Quarterly 58: 335–357.

Bailey, F.G. (1969) Stratagems and Spoils. A Social Anthropology of Politics, Oxford: Blackwell.Besley, T. and Coate, S. (2000) ‘Issue unbundling via citizens’ initiatives’, London School of Economics,mimeo, available at http://econ.lse.ac.uk/staff/tbesley/papers/iniative.pdf.

Besley, T. and Coate, S. (2003) ‘Elected versus Appointed Regulators’, Journal of the European EconomicAssociation 1(5): 1176–1206.

Besley, T. and Prat, A. (2004) ‘Handcuffs for the grabbing hand? Media capture and governmentaccountability’, London School of Economics, mimeo, available at http://econ.lse.ac.uk/staff/prat/papers/media.pdf.

Budge, I. (1993) ‘Issues, Dimensions, and Agenda Change in Postwar Democracies: Long-term Trends inParty Election Programs and Newspaper Reports in Twenty-three Democracies’, in W.H. Riker (ed.)Agenda Formation, Ann Arbor: University of Michigan Press, pp. 41–80.

Budge, I. (1994) ‘A new spatial theory of party competition: uncertainty, ideology and policy equilibriaviewed comparatively and temporally’, British Journal of Political Science 24(4): 443–467.

Budge, I. and Keman, H. (1990) Parties and Democracy: Coalition Formation and GovernmentFunctioning in Twenty States, Oxford and New York: Oxford University Press.

Budge, I., Robertson, D. and Hearl, D.J. (eds.) (1987) Ideology, Strategy and Party Change: SpatialAnalyses of Post-War Election Programmes in 19 Democracies, Cambridge: Cambridge UniversityPress.

‘The number ofpoliticised issues cannot

directly explain thenumber of politicalparties, since policy

innovation may developeither within or between

parties.’

518 european political science: 4 2005 cleavages, issues and parties

Campbell, A., Converse, P., Miller, W. and Stokes, D. (1960) The American Voter, Chicago: Universityof Chicago Press.

Cantillon, E. (2001) ‘Electoral rules and the emergence of new issue dimensions’, Yale University: CowlesFoundation Discussion Papers, No. 1291.

Carmines, E.G. and Stimson, J.A. (1989) Issue Evolution: Race and the Transformation of AmericanPolitics, Princeton: Princeton University Press.

Cohen, B. (1963) The Press and Foreign Policy, Princeton: Princeton University Press.Davis, M.L. and Ferrantino, M. (1996) ‘Towards a positive theory of political rhetoric: why do politicianslie?’ Public Choice 88: 1–13.

Dodd, L.C. (1976) Coalitions in Parliamentary Government,, Princeton: Princeton University Press.Downs, A. (1957) An Economic Theory of Democracy, New York: Harper.Fiorina, M.P. (1977) ‘An outline for a model of party choice’, American Journal of Political Science 21(3):601–625.

Fiorina, M.P., Abrams, S.J. and Pope, J.C. (2004) Culture War? The Myth of a Polarized America, NewYork: Pearson.

Flora, P. (ed.) (1999) State Formation, Nation-Building, and Mass Politics in Europe: The Theory of SteinRokkan, Oxford and New York: Oxford University Press.

Franklin, M. (1985) The Decline of Class Voting in Britain: Changes in the Bias of Electoral Choice,1964–1983, Oxford: Clarendon Press.

Franklin, M.N. (1992) ‘The Decline of Cleavage Politics’, in M.N. Franklin, T.T. Mackie and H. Valen (eds.)Electoral Change: Responses to Evolving Social and Attitudinal Structures in Western Countries,Cambridge and New York: Cambridge University Press, pp. 385–405.

Franklin, M.N., Mackie, T.T. and Valen, H. (eds.) (1992) Electoral Change: Responses to Evolving Socialand Attitudinal Structures in Western Countries, Cambridge and New York: Cambridge UniversityPress.

Grofman, B. (2004) ‘Downs and Two-Party Convergence’, Annual Review of Political Science 7: 25–46.Inglehart, R. (1977) The Silent Revolution: Changing Values and Political Styles Among Western Publics,Princeton: Princeton University Press.

Jacoby, W.G. (2000) ‘Issue framing and public opinion on government spending’, American Journal ofPolitical Science 44(4): 750–767.

Knutsen, O. (1988) ‘The impact of structural and ideological party cleavages in WesternEuropean democracies: a comparative empirical analysis’, British Journal of Political Science 18:323–352.

Laver, M. and Budge, I. (eds.) (1992) Party Policy and Government Coalitions, London and New York:Macmillan/St Martin’s.

Laver, M. and Hunt, W.B. (1992) Policy and Party Competition, New York and London: Routledge.Laver, M., Benoit, K. and Garry, J. (2003) ‘Extracting policy positions from political texts using words asdata’, American Political Science Review 97(2): 311–331.

Lijphart, A. (1984) Democracies: Patterns of Majoritarian and Consensus Government in 21 Countries,New Haven: Yale University Press.

Lipset, S.M. (ed.) (1981) Party Coalitions in the 1980s, San Francisco: Institute for ContemporaryStudies.

Manza, J. and Brooks, C. (1999) Social Cleavages and Political Change: Voter Alignments and US PartyCoalitions, New York and Oxford: Oxford University Press.

McCombs, M.E. and Shaw, D.L. (1972) ‘The agenda-setting function of mass media’, Public OpinionQuarterly 36(2): 176–187.

Miller, G. and Schofield, N. (2003) ‘Activists and partisan realignment in the United States’, AmericanPolitical Science Review 97(2): 245–260.

Nie, N.H., Verba, S. and Petrocik, J.R. (1981) The Changing American Voter, Cambridge, MA: HarvardUniversity Press.

Petrocik, J.R. (1996) ‘Issue ownership in presidential elections, with a 1980 case study’, American Journalof Political Science 40(3): 825–850.

Petrocik, J.R., Benoit, W.L. and Hansen, G. (2002) ‘Issue ownership and presidential campaigning,1952–2000’, Political Science Quarterly 118(4): 599–626.

Puglisi, R. (2004a) ‘The spin doctor meets the rational voter: electoral competition with agenda-settingeffects’, Paper presented at the Public Choice Society Annual Meeting; 12–14 March 2004;Baltimore, MD.

Puglisi, R. (2004b) ‘Being the New York times: the political behaviour of a newspaper’, Paper presented atthe American Political Science Association Annual meeting; 1–4 August 2004; Chicago, IL.

519josep m. colomer and riccardo puglisi european political science: 4 2005

Riker, W.H. (1983) ‘Political Theory and the Art of Heresthetics’, in A. Finifter (ed.) Political Science: TheState of the Discipline, Washington DC: American Political Science Association, pp. 47–67.

Riker, W.H. (1986) The Art of Political Manipulation, New Haven: Yale University Press.Riker, W.H. (1996) (Posthumous edition: R.L. Calvert, J. Mueller and R.K. Wilson, eds.) The Strategy ofRhetoric: Campaigning for the American Constitution, New Haven: Yale University Press.

Robertson, D. (1976) A Theory of Party Competition, London and New York: Wiley.Rokkan, S., Campbell, A., Torsvik, P. and Valen, H. (1970) Citizens, Elections, Parties: Approaches to theComparative Study of the Processes of Development, Oslo: Universitetsforlaget.

Rose, R. and McAllister, I. (1986) Voters Begin to Choose, London: Sage.Sartori, G. (1976) Parties and Party Systems: A Framework for Analysis, Cambridge: CambridgeUniversity Press.

Sartori, G. (1998) Homo Videns: La sociedad teledirigida, Madrid: Taurus.Schofield, N. and Laver, M. (1985) ‘Bargaining theory and portfolio payoffs in European coalitiongovernments, 1945–1983’, British Journal of Political Science 15: 143–164.

Schofield, N., Miller, G. and Martin, A. (2003) ‘Critical elections and political realignments in the USA:1860–2000’, Political Studies 41(2): 217–243.

Simon, A.F. (2002) The Winning Message: Candidate Behavior, Campaign Discourse and Democracy,New York: Cambridge University Press.

Stokes, D.E. (1963) ‘Spatial models of party competition’, American Political Science Review 57(2):368–377.

Taagepera, R. (1999) ‘The number of parties as a function of heterogeneity and electoral system’,Comparative Political Studies 32(5): 531–548.

Taagepera, R. and Grofman, B. (1985) ‘Rethinking Duverger’s law: predicting the effective number ofparties in plurality and PR systems: parties minus issues equals one’, European Journal of PoliticalResearch 13: 341–352.

Taylor, M. and Laver, M. (1973) ‘Government coalitions in Western Europe’, European Journal of PoliticalResearch 1: 205–248.

Warwick, P.V. (2000) ‘Policy horizons in West European parliamentary systems’, European Journal ofPolitical Research 38: 37–61.

About the Authors

Josep M. Colomer is Research Professor in Political Science and the author of more than 120academic articles and book chapters, as well as author or editor of 28 books in six languages,including Handbook of Electoral System Choice (Palgrave Macmillan, 2004), winner of theLeon Weaver Award of the American Political Science Association; Political Institutions inEurope (Routledge, 2002); Political Institutions (Oxford University Press, 2001); StrategicTransitions (Johns Hopkins University Press, 2000); and Game Theory and the Transition toDemocracy: The Spanish Model (Edward Elgar, 1995). Josep M. Colomer acknowledgespartial support of grant HPSE-CT-2002-00146 from the European Commission.

Riccardo Puglisi holds a PhD from the London School of Economics and Political Science and isAssistant Professor at the department of Political Science of the Massachusetts Institute ofTechnology.

520 european political science: 4 2005 cleavages, issues and parties