Sie sind auf Seite 1von 23

European Journal of Political Research 47: 687709, 2008 doi: 10.1111/j.1475-6765.2008.00778.

687

Explaining the salience of leftright ideology in postindustrial democracies: The role of structural economic change
TIMOTHY HELLWIG
Department of Political Science, University of Houston, USA

Abstract. Does leftright ideology structure electoral competition? While many studies show that ideology is a powerful determinant of party choice, a growing number of scholars claim that the salience of left and right has declined in recent decades. These contrasting views motivate a consideration of whether the salience of the leftright dimension varies across political parties and national contexts. Drawing on recent advances in the study of the welfare state, this article argues that just as policy appeals expressed in the language of left and right crystallised during a period of economic change, changes in the organisation of postindustrial economies should weaken the leftright bases of competition. Analyses of 87 parties in 16 parliamentary democracies show that occupational heterogeneity in postindustrial economies indicated by employment in the private service sector and in industries exposed to international competition reduces the salience of the leftright dimension for the vote. The implications of study ndings for future work on policy responsiveness and electoral change in advanced capitalist democracies are discussed.

Introduction
To what extent does leftright ideology structure policy competition in advanced industrial democracies? By assimilating the many issues that appear before the electorate, the language of left and right provides parties and voters with a shared policy space and students of politics with a powerful tool for evaluating elitemass correspondence over policy. It should come as little surprise, therefore, that studies of many electoral contexts nd that leftright ideology ranks as one of if not the largest determinants of party choice. These ndings appear inconsistent with research heralding a decline in the salience of leftright orientations in many advanced industrial societies (Inglehart 1990; Kitschelt 1994; Kitschelt & Rehm 2005; Kriesi 1998; Kriesi et al. 2006; Manza & Brooks 1999). According to the latter perspective, the utility of the leftright continuum for structuring competition in established party systems has waned for some time, so much so that traditional socioeconomic cleavages now risk becoming secondary in importance to other dimensions of policy contestation.
2008 The Author(s) Journal compilation 2008 (European Consortium for Political Research) Published by Blackwell Publishing Ltd., 9600 Garsington Road, Oxford, OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA

688

timothy hellwig

While it is commonplace to show that ideology in terms of left and right inuences the voters decision, no research has directly examined whether the salience of the leftright dimension varies across political parties or electoral contexts and, if so, whether this variance can be explained systematically. An understanding of the salience of leftright appeals, however, is important for several reasons. First, a single encompassing dimension of political contestation helps the voter sift through the many issues in election campaign and select the party that comes closest to his or her own preferences. Second, thinking in terms of left and right provides a concise means for elites to relate to one another. In this way, it serves as a coordinating device for avoiding cyclical majorities. Perhaps most important, the existence of a salient super-dimension is normatively appealing. To select parties on substantive policy grounds rather than on less tangible factors like personal appeals requires that voters evaluate the policy positions of competing elites in relation to each other. When parties become inseparable on policy grounds, substantive political representation is likely to suffer as a result (Pitkin 1987: 210213). The purpose of this article is to examine the salience of left and right in light of the now well-documented changes in advanced capitalist society. My argument is that the salience of political ideology is a function of socioeconomic heterogeneity. Just as the socio-economic leftright divide solidied in what were relatively homogeneous societies, so has leftright ideologys salience declined as electorates have become more diverse. I identify two sources of heterogeneity: occupational change and dependence on world markets. These factors increase the heterogeneity of the electorates preferences and, in turn, diminish the utility of single-dimensional policy voting. The argument is tested using representative sample surveys from 16 parliamentary democracies. Consistent with expectations, results demonstrate that the salience of the leftright dimension is weaker for those employed in services or in traded industries. Study ndings carry implications for voting behaviour, party strategies and electoral change more broadly. The next section reviews the ways past research has measured and used leftright ideology to understand political competition. Building on ideas that new politics issues pose a challenge to the dominance of traditional policy appeals, I then advance the claim that two structural factors deindustrialisation and globalisation may account for whether voter decisions continue to be made in leftright terms. The fourth section reports results from an empirical analysis of recent national-level elections from 16 democracies. I conclude with a discussion of study implications for future work on electoral change in postindustrial societies.
2008 The Author(s) Journal compilation 2008 (European Consortium for Political Research)

the salience of leftright ideology

689

The salience of leftright ideology


The salience of political ideology for structuring political competition raises at least three questions: How do we measure it; what does it mean; and how important is it for the things we care about? Regarding the rst question, scholars have measured leftright placements using three principal means: from surveys of party experts, from public opinion surveys and from party manifestos. Internal validity checks and comparisons across data sources have found these measures to do a satisfactory job of capturing policy-based political competition, particularly in the developed democracies (Benoit & Laver 2006; Budge et al. 2001; Gabel & Huber 2000; Huber 1989; Huber & Inglehart 1995; Marks 2006; Warwick 2002).1 For empirical studies of representation, this means that the leftright scale can be used as a kind of super issue dimension to assess the preferences of citizens and policy makers. The leftright dimension also has been used to examine how party strategies affect vote outcomes and in conjunction with other factors like partisan loyalties or performance evaluations to judge the relative contribution of policy considerations to party choice (e.g., Blais et al. 2004; Duch & Stevenson 2008; Lewis-Beck 1988: 84). In short, leftright ideology is widely shown to have a strong inuence on the voters decision. Less attention has been given to the content of the leftright scale. Many analysts shy away from confronting the issue of content and instead leave it to the individual respondent (either a respondent on a public opinion survey or an expert tasked to judge party positions). Nearly all empirical studies, however, conclude that party placements along the leftright scale are structured to a great extent by positions on state involvement in the economy. Huber and Inglehart (1995) nd that economic issues were cited as the most important component of the leftright dimension in all but ve of the 42 countries they examined. In a secondary analysis of Laver and Hunts (1992) expert judgments, Warwick (2002) shows that the leftright scale loads most strongly on a set of economic issue items. And analyses of Markss and Steenbergens (1999) expert survey nds that their general leftright scale correlates highly with an explicitly economic leftright scale (r = 0.92), indicating that the two scales tap the same policy orientations for most parties. All told, evidence from empirical investigations support Downs (1957) claim that party competition takes place along a leftright dimension based on disagreements over the scope of government intervention in the economy. Finally, researchers have been concerned with the third question about the importance of left and right. A key question in comparative political economy, for example, is whether and how government partisanship matters for actual policy outcomes (e.g., Rueda 2005). This research programme
2008 The Author(s) Journal compilation 2008 (European Consortium for Political Research)

690

timothy hellwig

assumes that party policy preferences can be assimilated onto a single left right dimension. Likewise, studies of dyadic representation often employ left right orientations to link citizens to elites (e.g., Miller et al. 1999). As Pierce (1999: 30) puts it, the issue to which [voters] are likely to give high priority . . . is the ideological super-issue. . . . Voterparty congruence on more specic issues, even those that are traditionally linked to the ideological dimension, is much more limited. The salience of leftright ideology, however, has been shown to be in decline in Western democracies for some time. Some 25 years ago, for example, only a third of French respondents agreed that ideas of right and left are obsolete. By 2002, however, fully 60 per cent believed this to be the case (TNS Sofres 2007). Instead of old socio-economic, class-based conicts easily summarised by left and right, scholars argue that an alternative dimension of policy competition tapping non-economic, non-class-based concerns is growing in salience. This alternative dimension has been variously termed postmaterialist-materialist (Inglehart 1990), libertarian-authoritarian (Flanagan & Lee 2003; Kitschelt 1994; Kitschelt & Rehm 2005), green/alternative/ libertarian traditional/authoritarian/nationalism (Hooghe et al. 2002) and integration-demarcation (Kriesi et al. 2006). As evinced by the various labels employed, there is some disagreement over this second dimensions policy content. Yet among its proponents, there is no disagreement that these issues are distinct from the traditional socio-economic divisions associated with the traditional leftright divide. The emergence of a new set of issues carries several implications, including the fortunes of class/mass political parties, the emergence of anti-system parties, and, in general, the increase in electoral volatility and, possibly, electoral dealignment. Have changes in electorates diminished the utility of the leftright dimension as a coordinating device connecting voters to parties? And if so, what explains variations in the salience of the leftright dimension? The next section advances the claim that the utility of a single super-dimension for structuring party choice can be related to the diversity of the electorate and, moreover, that this diversity can be attributed in part to structural change in advanced capitalism.

Explaining the salience of the leftright dimension: Socio-economic heterogeneity


Political parties may be conceived as group-based organisations charged with maximising the utility of their constituencies. These constituencies have traditionally been characterised by inter-group differences in the direction of policy
2008 The Author(s) Journal compilation 2008 (European Consortium for Political Research)

the salience of leftright ideology

691

and by intra-group homogeneity over the salience (rank order) of policy priorities. If the electorate is composed of two groups (e.g., the working class and the middle class), delineated according to preferences in a single issue area (e.g., concerns for redistribution), then a single leftright dimension provides order to policy competition. However, if either the range of between-group preferences over the direction of policy declines or the diversity of withingroup preferences over issue salience rises, then masselite congruence based on a single dimension will weaken. To illustrate, consider two ideal-type scenarios. In both cases, the parties announce a set of coherent policy proposals known to all. In the rst scenario, all voters are assumed to have identical rankings in terms of the salience of the issues. The only difference among citizens rests in the direction of their positions. These directions dene the parties constituencies. Here, the voters decision problem simplies to a task of identifying the party closest to his or her preferred position on the most salient issue. All other issues are collapsed into a dominant policy dimension or are discounted such that their effect on the voters choice is indistinguishable from zero. In this scenario, a single dimension works well for structuring electoral competition. The issue environment changes, however, when the salience of preferences within the party constituencies becomes more diverse. Some voters may be concerned primarily with issues of economic equality and economic efciency, while others care mainly about issues that are not explicitly economic- or class-based, such as health care or the environment. In this second scenario, this diversity of considerations strains the capacity of parties and voters to communicate using a single policy dimension. A consideration of the role of voter heterogeneity coupled with a Downsian understanding that the content of the leftright dimension is dominated by socio-economic/class-based concerns directs our attention on factors expected to contribute to socio-economic heterogeneity. I identify two such factors: the expansion of the service sector and the exposure of national economies to world markets. Expansion of the service sector The policies that emerged in the years following the Second World War were a product of the conditions prevailing at the time a time in which economies were organised for producing and competing in the manufacturing sector. Industrialisation provided a vocabulary for mass politics as well. The working class identied with positions on the left side of the continuum, while employers and capital owners identied with the right. These positions were linked closely to policy-oriented parties. As Hibbs (1977: 1470) put it: Although the
2008 The Author(s) Journal compilation 2008 (European Consortium for Political Research)

692

timothy hellwig

importance of socioeconomic status as a basis of electoral cleavage varies substantially across party systems, the mass constituencies of political parties in most advanced industrial democracies are distinguished. . . by class, income and related socioeconomic characteristics. The implication was that party issue positions could be readily arrayed along the traditional leftright spectrum. In this way, the economic leftright dimension crystallised in lock step with the rise of industrial societies. Election outcomes were shaped by collective preferences for state involvement in industry and protecting the rights of the industrial working class, on the one hand, and preferences for free markets and limited government, on the other (Borre 2001). In recent decades, however, economies have experienced large-scale structural change. While social classes have never acted as monolithic entities, recent changes have made it particularly dubious to assume preference homogeneity within class groups. Changes include the decline in union power, decreasing employment in manufacturing, the growth of Post-Fordist production technologies and the outsourcing of product inputs. A dening feature of advanced capitalism, then, has been the decline of the industrial working class and its replacement by the white-collar service sector. As Figure 1 shows (solid line), the average share of workers employed in the private service sector relative to the public and industrial sectors of the economy has grown steadily over the past 30 years in OECD countries.2 The effects of deindustrialisation range beyond the individual workers material well-being. As Iversen (2005; see also Iversen & Cusack 2000; Iversen & Wren 1998) argues, the size of the welfare state expanded as a kind of compensation strategy to counteract the dislocating effects of service sector employment. Via labour market deregulation and the curtailment of active forms of employment protection, governments encouraged the growth of the service economy as a strategy for coping with the constraints of otherwise non-expansionary postindustrial economies. In this way, employment in the service sector comprises a distinct welfare-production regime, separable from regimes that rely on public sector employment or forms of social insurance as forms of welfare compensation.3 In sum, if the leftright basis for political competition crystallised during the period of industrialisation, its decline might be attributed to deindustrialisation.4 If true, then the leftright dimension should be less salient for voters employed in the service sector of the economy relative to the electorate overall. This brings us to our rst testable hypothesis: Hypothesis 1: Leftright ideology is less salient for voters employed in the service sector of the economy than for those not employed in the service sector.
2008 The Author(s) Journal compilation 2008 (European Consortium for Political Research)

the salience of leftright ideology


44 42 40 Trade as % of GDP Per cent 38 36 34 50 32 30 1975 1980 1985 1990 Year 1995 2000 Private Services Employment, % of Workforce 45 40 70 65 60 55 85 80 75

693

Figure 1. Employment in business services and trade openness: select countries, 19752002. Notes: Services Employment (solid line, measured on the left-hand axis) is the number of workers in the business service sector as a per cent of all workers for 13 OECD countries (Australia, Canada, Denmark, Finland, France, Great Britain, Ireland, Netherlands, Norway, Portugal, Spain, Sweden, and the United States). Trade Openness (dashed line, measured on the right-hand axis) is measured as exports plus imports as a per cent of GDP for 21 OECD countries (Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Great Britain, Greece, Ireland, Italy, Japan, Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, Switzerland, and the United States). Sources: OECDs Industry Structural Analysis (STAN) database and World Banks World Development Indicators.

Economic globalisation Along with occupational change, the growth of the international economy ranks as the most fundamental change to domestic economies since the postwar era. In 1970, the average level of trade (exports plus imports) as a share of gross domestic product (GDP) averaged less than 50 per cent across 21 OECD countries. By 2000 this share was nearly 75 per cent (Figure 1, dashed line). Exposure to the world economy may affect the salience of the leftright dimension in two ways. First, like deindustrialisation, economic integration affects the heterogeneity of constituency preferences. As classic studies by Gourevitch (1986) and Rogowski (1989) have shown, exposure to international trade shapes political alignments within countries, creating tensions between industry- and class-based coalitions. While domestic cleavages due to preferences toward the international economy are not unique to postindustrial economies, global markets affect larger shares of electorates now than at any time in the past 60 years. Resulting tensions between left versus right and
2008 The Author(s) Journal compilation 2008 (European Consortium for Political Research)

694

timothy hellwig

open versus closed makes the task of representation through political parties more difcult. In open economies, parties must respond not only to their constituencies in the domestic market, but also to the overlapping and cross-cutting interests of capital and labour in the import-competing sectors of the economy. This increases the chances that social cleavages will be cast in terms of preferences for free trade versus protectionism rather than for or against market regulation in general. Globalisation might also affect the salience of the leftright dimension through a separate channel: the policy capacity of national ofcials. The utility of leftright policy voting or any sort of voting over policy, for that matter should depend on the capacity of politicians to control policy levers. Rational voters should make decisions based on preferences over the type and direction of government action only if they believe that public intervention has the capacity to affect citizen welfare. Dependence on global markets, however, may compromise this capacity. If globalisation reduces parties ability to pursue their preferred economic policy positions (in terms of lowering interest rates or raising corporate taxes, for example) and if the public knows this to be the case (as previous work indicates: Hellwig 2001; Hellwig et al. forthcoming), then voters have little incentive to maximise utility in terms of leftright policy congruence.5 This argument translates directly into a second expectation of how occupational alignments should affect policy voting: If opening up the economy makes for more heterogeneous electorates, then the leftright dimension should be less salient for voters employed in the exposed sector. This is our second hypothesis: Hypothesis 2: Leftright ideology is less salient for voters employed in traded industries than for those not in industries exposed to competition from abroad.

Analysis
In order to assess the inuence of leftright ideology on the vote, I employ a simple spatial model of voter utility. Spatial models express individual utility in terms of the distance between the issue position of the party and the position of the voter. This can be written in one dimension using a quadratic proximate utility model as:

Uij = ( vi pj ) + j zi,
2
2008 The Author(s) Journal compilation 2008 (European Consortium for Political Research)

(1)

the salience of leftright ideology

695

where Uij is the utility of voter i for party j, vi is the ideal point of voter i on the leftright dimension, pj is the corresponding position of party j and b is a salience parameter greater than zero. The minus sign in front of the expression denotes that utility is negatively related to the spatial distance between the voter and the party. The vector dj represents the effect of non-policy considerations zi on voter utility. Using data from public opinion surveys, I estimate a statistical model for multinomial choice where pij is the probability of individual i (i = 1, 2, . . . , n) voting for party j (j = 1, 2, . . . , m). The model predicts that the probability that i selects j is a function of the voters utility for j taking into consideration his or her utility for all other parties such that:

(votei = j ) ij =

k =1 exp ( vi pk ) + k zi
m 2

exp ( vi pj ) + j zi
2

(2)

Equation (2) can be estimated using conditional logit.6 Models akin to Equation (2) found in the literature generally assume that all voters place the same weight, or salience, on elitemass policy congruence (vi - pj). This assumption, however, may be inappropriate in some, if not most, election contests, as work by Rivers (1988) and Glasgow (2001) suggests. In order to test research claims about the bases of leftright salience, I allow the magnitude of b to vary according to the voters occupational sector. According to the research hypotheses, the utility gained for voter i from proximity to party j on the leftright dimension should be less (in absolute magnitude) if i is employed in the private service sector or in an import-competing industry. These expectations pertaining to heterogeneity in the electorate can be incorporated into the model by interacting the leftright distance measure(vi - pj)2 with indicator variables for respondent occupation. Data and measures The empirical approach entails estimating party choice models using individual-level data from 16 parliamentary democracies between 1999 and 2003. I use post-election surveys for Finland, Germany, Ireland, the Netherlands, New Zealand, Norway, Sweden and Switzerland provided by the second module of the Comparative Study of Electoral Systems project (CSES 2006). Due to missing or incomplete data on key variables, these studies are supplemented with data from the original national election studies surveys for Australia (Bean et al. 2002), Denmark (Goul-Andersen et al. 2003) and Great Britain (Sanders et al. 2002), and with data from the European Social Survey
2008 The Author(s) Journal compilation 2008 (European Consortium for Political Research)

696

timothy hellwig

(ESS) for Austria, Belgium, France, Greece, and Italy (Jowell et al. 2003). I use these data to estimate separate party choice models for each country.7 Models have the following variables in common. The dependent variable, Party Choice, is coded with as many categories as there are parties standing in the election. Those parties that lack measures on the leftright scale in the survey or obtain less than 5 per cent of the vote are omitted from the analysis, per standard conventions (e.g., Alvarez et al. 2000; Blais et al. 2004; Duch & Stevenson 2008).8 For the election study surveys, the item asks for respondent vote choice in the current election. For surveys from the ESS, respondents are asked which party they voted for in their countrys most recent national election. The Appendix lists elections and parties included in the analysis. The main explanatory variable, LeftRight Distance, is measured as the squared distance between the voters position on the leftright scale, vi, and the partys position, pj. For the voters position, I use the standard 11-point question: In politics people sometimes talk of left and right. Where would you place yourself on a scale from 0 to 10 where 0 means the left and 10 means the right? For the CSES and election study surveys, the partys position is taken from the mean of the respondents placement of the party on the identical scale.9 For the ESS studies, I use leftright position as identied by experts from Benoit and Laver (2006).10 Finally, all country surveys include measures of respondent occupation. For all studies, I make use of occupational codes from the International Standard Classication of Occupations (ISCO-88).11 I use these data to classify respondents according to whether or not they are employed in the service sector of the economy or in a traded sector of the economy. Codes for service industry employment are informed by Iversen and Wren (1998) and approximate those in Kitschelt and Rehm (2005). Respondents are classied as belonging to the service sector if they identify themselves as clerks (ISCO-88 codes 411-422), service workers and market sales workers (511-523), or sales and services elementary occupations (911-916). Following Hays et al. (2005), I identify traded sectors using the OECDs Industry Structural Analysis (STAN) database. Respondents employed in traded industries are those who identify themselves as skilled agricultural and shery workers (ISCO-88 codes 611-21), as craft and related trade workers (711-44), as plant and machine operators and assemblers (811-34), or as working in elementary occupations in agriculture, mining, manufacturing and transportation (921-33). In addition to measures for party choice, ideological distance and occupation, each country model includes a battery of individual-specic covariates as informed by prior voting behaviour research.
2008 The Author(s) Journal compilation 2008 (European Consortium for Political Research)

the salience of leftright ideology Results

697

Before reporting results across the 16 countries, I use the 2001 Danish Election Study data to illustrate the analytic approach. The Danish party system includes representatives from most major party families, making it an appropriate case for illustrative purposes. I begin with estimating the conditional logit model. Model estimates are reported in Table 1. The dependent variable is self-reported vote choice for one of the six largest parties in the Danish Folketing (parliament). The second through fth columns of the table report the effects of individual-specic predictors on the vote, with the incumbent Liberal Party (V) arbitrarily selected as the base category. Our primary interest is with the rst column, which reports results of the choice-specic policy distance measure. As expected, the coefcient on LeftRight Distance is negative and precisely estimated. This means that the greater the distance between the voter and the party, the less likely the voter will be to select the party. The magnitude of this effect, however, is contingent on occupational sector. Coefcients on the two interaction terms, LeftRight Distance Service Sector and LeftRight Distance Traded Industry are both in the expected positive direction. These results support the argument that leftright ideology is less salient for voters who earn their income from the private services and traded sectors of the Danish economy. Leftright placement, however, may be more critical for some parties than others. Specically, older mainstream parties such as the Danish Social Democrats or Liberals have more invested in appealing to class-based constituencies using a language of left and right than do parties less than a decade old, like the Danish Peoples Party.12 However, since LeftRight Distance is modeled as choice-specic, results reported in Table 1 do not allow us examine the differences in the salience of political ideology across the party choices (Train 2003). Post-estimation analyses are in order. I rst obtain a set of predicted vote probabilities by drawing a vector of coefcients from a distribution with mean equal to the vector of parameter estimates and variance estimated from the variance-covariance matrix. I then calculate the predicted probability that a hypothetical voter, i, selects party j using the formula for the conditional logit model (see Equation 2).13 In order to provide a common benchmark for gauging leftright salience, for each party I set vi equal to pj. This enables me to estimate the probability that i chooses j if is policy preferences are identical to js position on the leftright scale.14 I perform this exercise three times corresponding to three occupational proles. Results are reported in Table 2. The rst column of predicted probabilities displays the probability that the voter selects the party when he or she is employed neither in the service sector nor by an industry susceptible to
2008 The Author(s) Journal compilation 2008 (European Consortium for Political Research)

698

timothy hellwig

Table 1. Conditional logit estimates of the 2001 Danish election. Dependent variable: Respondent Vote Choice (Liberal Party coefcients normalized to zero) SD/V Choice-specic LeftRight Distance (LRD) LRD Service Sector LRD Traded Industry Individual-specic Service Sector Traded Industry Age Female Income Education Union Rural Constant -2 log likelihood N 2500.37 1,044 0.063 (0.272) 0.539* (0.279) (0.009) 0.086 (0.225) -0.055 (0.048) 0.023 (0.122) 0.677** (0.277) -0.193 (0.220) (0.781) -0.056 (0.339) 0.993** (0.308) (0.011) -0.240 (0.278) (0.053) 0.037 (0.151) (0.320) -0.238 (0.257) (0.953) 0.194 (0.324) 0.188 (0.355) 0.039** (0.011) -0.407 (0.281) (0.048) 0.505** (0.158) (0.268) -0.411 (0.291) (0.935) -0.128 (0.451) 0.631 (0.466) 0.055** (0.016) 0.321 (0.366) -0.113 (0.079) 0.673** (0.217) 0.908 (0.586) -0.208 (0.395) (1.348) -0.080 (0.411) -0.366 (0.552) 0.040** (0.014) 0.082 (0.325) 0.027 (0.063) 1.108** (0.240) -0.312 (0.358) 0.141 (0.347) (1.246) -0.232** (0.018) 0.137** (0.023) 0.144** (0.022) DF/V KF/V SF/V RV/V

0.026** -0.012

-0.153** -0.057

0.695** -0.192

-2.080** -0.244

-3.925** -6.694** -7.423**

Note: Cells report conditional logit regression coefcients with standard errors in parentheses. The Liberal Party is the reference category for the individual-specic coefcients. V = Liberals, SD = Social Democrats, DF = Danish Peoples Party, KF = Conservatives, SF = Socialist Peoples Party, RV = Radical Liberals. ** p < 0.05, * p < 0.10, two-tailed test. Source: Data are from the 2001 Danish Election Study.

2008 The Author(s) Journal compilation 2008 (European Consortium for Political Research)

Table 2. The effects of service sector and tradable industry on leftright policy voting 2001 Danish election

Party j equals 2.8 4.4 4.6 7.0 7.3 8.1 0.21 0.61 0.18 0.18 0.51 0.16 0.10 0.06 0.60 0.41 0.26 0.13

LeftRight position of party j

Probability of choosing j if voter is not in service or traded sector

Probability of choosing j if voter is in service sector

Probability of choosing j if voter is in traded sector 0.18 0.45 0.03 0.13 0.37 0.33

Socialist Peoples Party (SF)

Social Democratic Party (SD)

Radical Liberals (RV)

Conservative Party (KF)

Liberal Party (V)

the salience of leftright ideology

Danish Peoples Party (DF)

Note: Calculations performed using estimates reported in Table 1. Cells report predicted vote probabilities for a hypothetical voter who is male, belongs to a union, does not reside in a rural area, and has mean values on age, education and income. For all parties the voters position placed at the identical position as the party, such that vi = pj. Source: Data are from the 2001 Danish Election Study.

699

2008 The Author(s) Journal compilation 2008 (European Consortium for Political Research)

700

timothy hellwig

international competition. Here we see that if the voter places himself or herself at the same location as the Social Democrats on the leftright scale (4.4 on the 010 scale), then he or she will vote for that party with probability of 0.60. Likewise, if the voter locates themselves further to the right, at 7.3, then he or she will select the Liberals with a similarly high probability estimated at 0.61. Leftright placement has less of an effect on the remaining four parties, as we might expect. More to the point, however, is that in all but two of 12 cells, party vote probabilities decline when we change the voters occupational status. For example, if the respondent is employed in the service sector, the probability of selecting the Social Democrats falls from 0.60 to 0.41; if he or she is employed in a traded industry, then the probability declines to 0.45.15 Having illustrated the statistical approach using the Danish data, Table 3 reports conditional logit estimates for all 16 countries. Model specications differ only in terms of the number of party choices (ranging from 3 to 7) and in the selection of model controls. Given these differences, estimates for the individual-specic covariates are omitted from Table 3 to facilitate presentation.16 Results show that in all cases, the coefcient on LeftRight Distance is negative and estimated with high precision. Again, however, our main interest pertains to voter heterogeneity due to occupation. Results across the cases show that ideological congruence tends to be less important for voters in positions in services and in industries exposed to international competition than for voters whose livelihood does not depend on these sectors. The coefcient on the LeftRight Distance Service Sector interaction term is in the expected positive direction in 12 of the 16 models and is statistically signicant in eight of these. Coefcients on LeftRight Distance Traded Industry also are positively signed in 12 of the models, and nine of these estimates attain acceptable levels of statistical signicance. Equally impressive, in no instance is an interactive coefcient large in magnitude and statistically signicant with a negative (unexpected) sign. And where results are weakest, this conforms with what we know about these political systems. Belgium and Italy are the only cases where coefcients on both interaction terms are in the unexpected negative direction, albeit with very large standard errors. Given the characteristics of these democracies, these results should not be surprising. The Belgian party system is structured based on linguistic cleavages as much as class-based ones, and, given weak parties and high electoral volatility, we would not expect strong ndings for Italy. Analysis of the Danish data showed the conditional effects of occupation were greatest for the largest and oldest party on the centre-left, the Social Democrats, and for the largest and oldest party on the centre-right, the Liberals.To ascertain whether this nding generalises across the range of our cases (i.e., that parties that have traditionally relied on leftright policy appeals have
2008 The Author(s) Journal compilation 2008 (European Consortium for Political Research)

Table 3. The effects of service sector and tradable industry on leftright policy voting conditional logit estimates for 16 countries AT 2002 -0.097** (0.008) 0.041** (0.017) 0.016 (0.025) 2329.09 1,217 IE 2002 -0.105** (0.019) 0.045* (0.023) 0.048** (0.024) 2663.44 1,102 506 829 1048.59 2274.74 (0.042) (0.037) -0.035 0.010 (0.040) (0.031) -0.023 0.007 (0.019) 0.053** (0.017) 1641.45 715 -0.035 (0.012) (0.013) (0.011) -0.123** -0.161** -0.117** IT 2001 NL 2002 NZ 2002 NO 2001 -0.193** (0.014) 0.019 (0.024) 0.061** (0.024) 2713.23 1,023 840 1,044 757 661 2202.59 2500.37 1988.69 1418.76 (0.016) (0.022) (0.022) (0.023) -0.017 0.144** 0.039* -0.004 0.057** (0.014) 3857.24 1,663 SE 2001 -0.136** (0.013) 0.026 (0.018) 0.003 (0.022) 1737.33 1,044 (0.016) (0.023) (0.020) (0.020) (0.014) -0.009 0.137** 0.052** 0.040** 0.056** (0.006) (0.018) (0.014) (0.012) (0.009) -0.047** -0.232** -0.137** -0.121** -0.122** -0.191** (0.016) 0.016 (0.030) 0.115** (0.025) 2804.73 1,633 CH 2003 -0.148** (0.013) 0.069** (0.018) 0.074** (0.019) 1632.31 783 BE 1999 DK 2001 FI 2003 FR 2002 DE 2002 GB 2001

AU 2001

LeftRight Distance

-0.085**

(0.007)

LRD Service Sector

0.027**

(0.013)

LRD Traded Industry

-0.005

(0.017)

-2 log likelihood

2314.27

1,112

GR 2000

LeftRight Distance

-0.333**

(0.018)

LRD Service Sector

-0.135

(0.124)

LRD Traded Industry

0.070*

the salience of leftright ideology

(0.039)

-2 log likelihood

805.93

1,208

701

2008 The Author(s) Journal compilation 2008 (European Consortium for Political Research)

Notes: Results report only the choice-specic coefcients with standard errors in parentheses from estimating a k-choice conditional logit model where k is the number of major parties competing in the election. Results of the individual-specic parameters are available from the author. ** p < 0.05, * p < 0.10, two-tailed test. AU = Australia, AT = Austria, BE = Belgium, DK = Denmark, FI = Finland, FR = France, DE = Germany, GB = Great Britain, GR = Greece, IE = Ireland, IT = Italy, NL = The Netherlands, NZ = New Zealand, NO = Norway, SE = Sweden, CH = Switzerland. Sources: Data are from CSES (2006), Jowell (2003), and select national election studies, as noted in the text.

702

timothy hellwig

the most to lose in the changing organisation of national economies), I again examine parties separately using post-estimation analyses. As above, I calculate predicted vote probabilities for a hypothetical voter who shares the partys ideological position and who is: not employed in either occupational sector considered here; employed in the service sector; and employed in a traded industry. Results of these analyses are displayed in Figure 2. To facilitate presentation, mean vote probabilities are grouped by party family.17 A comparison of the height of the bars across party groups shows that, ceteris paribus, citizens are more likely to vote for members of the established parties (i.e., the social democratic, conservative and Christian democratic party families) than for members of other families. This result could mean one of two things: that these are the parties with the highest degrees of support in these democracies, and/or that these are the parties for which leftright policy congruence matters most. More interesting, however, are the comparisons within each party family. These comparisons provide the leverage necessary to

0 communist soc dem

Probability of Voting for Party .1 .2 .3 .4

.5

green

liberal

chr dem

conserv

other

baseline vote probability - not services, not traded employed in traded industry employed in services

Figure 2. The effects of service sector and tradable industry on leftright policy voting, grouped by party family. Notes: Bars report mean vote probabilities calculated from estimates in Table 3 for a hypothetical individual who possesses the same values on the leftright scale as the party (vi = pj) and mean or modal values on all other. Baseline vote probabilities (black bars) are calculated by setting Service Sector and Traded Industry variables to 0. Employed in services vote probabilities (white bars) are calculated by setting Service Sector to 1. Employed in traded industry vote probabilities (gray bars) are calculated by setting Traded Industry to 1. Mean probabilities grouped according to party family, as per categories in Budge et al. (2001). The number of parties in each category is: communist 9, social democrat 17, green 10, liberal 14, Christian democrat 10, conservative 13, other 14.
2008 The Author(s) Journal compilation 2008 (European Consortium for Political Research)

the salience of leftright ideology

703

demonstrate that occupation sector matters for how voters weigh the many considerations involved in selecting their preferred party. The residual other category (which is mostly composed of new parties) notwithstanding, the gure shows that leftright congruence is less salient for individuals employed in the service sectors of their national economies across all party types. Similarly, voters in traded industries also are less likely to base their decision on ideological congruence to political parties, save for the Christian democrats (and, again, the residual other group). The Danish results reported in Table 2 apply more broadly: voter support based on leftright congruence declines with voter occupational status. Across a range of parliamentary democracies, structural economic change in postindustrial society depresses the salience of party policy appeals in terms of left and right.

Conclusion
This study has addressed questions fundamental to social cleavages and voter alignments questions with a rich tradition in political science and political sociology (Lipset & Rokkan 1967). How much does the summary leftright dimension of political competition matter for voter decisions in advanced capitalist societies today? Put differently, how much does party success hinge on appealing to the electorates preferences on this single super-dimension? Although leftright scales frequently are invoked in popular discourse and contemporary empirical political science alike, little research explicitly takes on these questions. This lack of attention appears surprising given sustained disagreement as to whether the concepts of left and right are still relevant for politics in established democracies today. Study ndings show that for many voters, left and right remain an important means for sifting through the many issues encountered during a campaign. However, this is not the case for those for whom the standard socioeconomic, class-based divide should have less appeal. Specically, those employed in the burgeoning service sector and those in industries exposed to international competition should have little interest-based motivation to cast their votes based on these traditional policy appeals. Thus, both deindustrialisation and globalisation reduce the salience of the leftright dimension of political contestation. Study results have implications for several areas in political science. Students of electoral behaviour have demonstrated that contextual factors condition the determinants of party choice, including retrospective economic perceptions (Duch & Stevenson 2008), issues (Kedar 2005) and partisan attachments (Huber et al. 2005). This article contributes to this research programme by considering how changes in the structure of national economies
2008 The Author(s) Journal compilation 2008 (European Consortium for Political Research)

704

timothy hellwig

shape the salience of elitemass policy congruence in terms of left and right. In arguing that the salience of leftright ideology still matters but is conditional on the voters occupational status in the postindustrial economy, this study provides a micro-mechanism to help move theoretical and empirical work forward, both for those who assert a continued relevance for the language of left and right as well as for those who question the utility of such concepts for postindustrial societies. Study ndings also are relevant for research on party strategies. According to spatial models with policy-seeking parties, politicians face a dilemma. On the one hand they wish to announce policies that reect their sincere, unwavering and (typically) non-centrist policy beliefs. At the same time, however, they face incentives to adjust their stated positions so as to improve their chances of winning the election. Changes in the structure of postindustrial economies may bring something of a solution to this dilemma albeit one of dubious distinction. By reducing the salience of policy-based appeals, deindustrialisation and globalisation reduce politicians costs of pursing purely ofceseeking strategies and, therefore, should produce parties that are keener to relate to voters in terms of non-policy factors like partisanship or reputation. Indeed, this view nds support in recent work on the growing inuence of valence issues in studies of voter choice (e.g., Green & Hobolt forthcoming). Finally, this study has implications for the future of democratic representation in those democracies where it ought to work best. Regardless of content, the existence of a salient dimension of policy contestation has desirable properties insomuch as it provides voters with information about the preferences of competing political parties, both in absolute terms and relative to one another. If the salience of this organising dimension is undermined, then substantive political representation will suffer as well. This would appear to be especially true for supporters of traditional mainstream parties, such as social democrats. Future work both in terms of data collection and data analysis should extend this research to examine the extent to which parties that solidied under industrialisation can avoid becoming irrelevant in postindustrial society through elitemass coordination on some alternative dimension of policy competition.

Acknowledgements
A previous draft of this article was presented at the 2006 meeting of the American Political Science Association in Philadelphia. I thank Anna Mikulska and Steven Tran for research assistance, and Ernesto Calvo, Ray Duch, Matt Gabel, John Huber, Noah Kaplan, Randy Stevenson and the anonymous
2008 The Author(s) Journal compilation 2008 (European Consortium for Political Research)

the salience of leftright ideology

705

reviewers for helpful comments and discussions. Support from a University of Houston New Faculty Grant is gratefully acknowledged.

Appendix
Table A1. Elections and parties included in the analyses Country (election year) Australia (2001) Parties Liberal Party, Australian Labor Party, National Party, Australian Democrats, Australian Greens, One Nation Party Social Democratic Party of Austria, Austrian Peoples Party, Freedom Party, Greens Greens, Christian-Socialists, Socialist Party, Liberal Party, Vlaams Bloc Liberal Party, Social Democrats, Danish Peoples Party, Conservative Peoples Party, Socialist Peoples Party, Radical Left Social Democrats, Center Party, National Coalition Party, Left Alliance, Swedish Peoples Party, Greens, Christian Democrats Socialist Party, RPR, Greens, UDF, National Front, French Communist Party Social Democrats, CDU/CSU, Greens, Free Democrats, PDS Conservative Party, Labour Party, Liberal Democrats PASOK, New Democracy, Communists Fianna Fail, Fine Gael, Labour, Greens, Progressive Democrats, Sinn Fein Left Democrats, La Margherita, Communists, Forza Italia, National Alliance PvDA, CDA, VVD, Democrats 66, Green Left, List Pim Fortuyn, Socialist Party Labour Party, National Party, New Zealand First, Act New Zealand, Green Party, United Future Socialist Left Party, Labour Party, Liberal Party, Christian Peoples Party, Center Party, Conservative Party, ProgressParty Left Party, Social Democrats, Center Party, Peoples Party Liberals, Conservative Party, Christian Democrats FDP/PRD, CVP/PDC, SP/PS, SVP/UDC, GPS/PES(Greens)

Austria (2002) Belgium (1999) Denmark (2001)

Finland (2003)

France (2002) Germany (2002) Great Britain (2001) Greece (2000) Ireland (2002) Italy (2001) Netherlands (2002) New Zealand (2002) Norway (2001)

Sweden (2002)* Switzerland (2003)

Note: * Green party not used due to lack of leftright and party identication measure.

2008 The Author(s) Journal compilation 2008 (European Consortium for Political Research)

706 Notes

timothy hellwig

1. Nearly all of these studies contend that leftright placement measures policy preferences to some extent, rather than some other afnity toward politics, such as partisanship (see especially Huber 1989). 2. Figures are from the OECDs Industry Structural Analysis (STAN) database from those countries for which comparable data are available. 3. For more on the structural changes facing welfare state economies and their electoral consequences, see Kitschelt (1994); Pontusson (1995); Rueda (2005). 4. This claim parallels arguments about the growth and retrenchment of the welfare state in general (Pierson 1996; Clayton & Pontusson 1998). 5. The issue is whether the voter believes his or her governments policy options are limited by engagement with world markets. Whether or not market integration actually reduces policy efcacy is a separate matter. 6. The conditional logit model species that the probability that voter i selects party j is proportional to the exponential of the deterministic component of utility and that probabilities over m parties sum to one. For discussions on estimating models of multiparty elections using survey data, see Adams et al. (2005); Dow & Endersby (2004); Duch & Stevenson (2008). 7. Canada and Japan are not included due to missing requisite measures for respondent occupation and leftright placement, respectively. Since ideology is more familiar in terms of liberalconservative than leftright, and since it is a candidate-based pure presidential democracy, I omit the United States from the analysis for reasons of questionable comparability on the dependent and key independent variables (see Dalton 2006: 217219; Pierce 1999). 8. In nearly all cases, parties lacking measures on the leftright scale are those that fail to obtain 5 per cent of the vote. Nonetheless, excluding parties lacking measures on the leftright scale may bias results. Yet by eliminating many of the newer parties that tend to attract the sorts of voters of interest here (those in the private services and exposed sectors of the economy), the effect of this bias should be to yield conservative results and therefore strengthen our condence in the statistical analyses reported below. 9. Using sample means for pj rather than the respondents own perception of js location is preferable on grounds that the latter would be subject to concerns over biased measures (i.e., the respondent may locate the party for whom he or she casts a ballot closer to their own position than they otherwise would, and vice-versa for parties he or she did not choose on election day). 10. Benoit and Lavers (2006) expert data were collected in or very close to the same year as the elections under consideration. For the 70 parties for which we have leftright placements from both public opinion and expert surveys, the bivariate correlation coefcient is 0.94. 11. The exception is France, where Eurostats Statistical Classication of Economic Activities is used instead. 12. This claim is consistent with recent research on the behaviour of niche versus mainstream parties (see Adams et al. 2006, Ezrow 2008, and especially Meguid forthcoming). 13. The hypothetical voter has mean or modal values on the individual-specic covariates. For the Danish data, this makes the hypothetical voter a male union member who does not reside in a rural area and has mean values on age, education and income.
2008 The Author(s) Journal compilation 2008 (European Consortium for Political Research)

the salience of leftright ideology

707

14. I also calculated predicted probabilities by setting the most extreme left- and right-wing parties for respondents located at the endpoints of the scales (0 and 10, respectively). For Denmark in 2001, these are the Socialist Peoples Party and the Danish Peoples Party. Results of these analyses, while increasing predicted probabilities for extreme parties, do not produce different results in terms of support for research hypotheses. I thank an anonymous reviewer for calling this to my attention. 15. Likewise, the probability that this individual selects the Liberals falls from 0.61 to 0.51 if employed in services, and 0.37 if employed in a traded industry. 16. Full model specications for all 16 country cases are available online at: http:// www.polsci.uh.edu/hellwig. 17. Party family classications are from Budge et al. (2001). The other category includes parties classied as agricultural (5 parties), national (5), ethnic (2) and not otherwise classied (2).

References
Adams, J. et al. (2006). Are niche parties fundamentally different from mainstream parties? The causes and the electoral consequences of Western European parties policy shifts, 19761998. American Journal of Political Science 50(3): 513529. Adams, J.F., Merrill III, S. & Grofman, B. (2005). A unied theory of party competition. Cambridge: Cambridge University Press. Alvarez, R.M., Nagler, J. & Willette, J. (2000). Measuring the relative impact of issues and the economy in democratic elections. Electoral Studies 19: 237253. Bean, C. et al. (2002). Australian election study, 2001. Canberra: Social Science Data Archives, Australian National University. Benoit, K. & Laver, M. (2006). Party policy in modern democracies. London: Routledge. Blais, A. et al. (2004). Which matters most? Comparing the impact of issues and the economy in American, British and Canadian elections. British Journal of Political Science 34: 555564. Borre, O. (2001). Issue voting. Aarhus: University of Aarhus Press. Budge, I. et al. (2001). Mapping policy preferences. Oxford: Oxford University Press. Clayton, R. & Pontusson, J. (1998). Welfare-state retrenchment revisited. World Politics 51: 6798. Comparative Study of Electoral Systems (CSES). (2006). CSES Module 2, 4th advance release [dataset]. Ann Arbor, MI: University of Michigan, Center for Political Studies [producer and distributor] (http://www.cses.org), 10 April 10. Dalton, R.J. (2006). Citizen politics, 4th edn. Washington, DC: Congressional Quarterly Press. Dow, J. & Endersby, J. (2004). A comparison of conditional logit and multinomial probit models in multiparty elections. Electoral Studies 23: 107122. Downs, A. (1957). An economic theory of democracy. New York: Addison Wesley. Duch, R.M. & Stevenson, R. (2008). The economic vote: How political and economic institutions condition election results. Cambridge: Cambridge University Press. Ezrow, L. (2008). On the inverse relationship between votes and proximity for niche parties. European Journal of Political Research 47: 206220. Flanagan, S.C. & Lee, A. (2003). The new politics, culture wars and the authoritarianlibertarian value change in advanced industrial democracies. Comparative Political Studies 36(April): 235270.
2008 The Author(s) Journal compilation 2008 (European Consortium for Political Research)

708

timothy hellwig

Gabel, M.J. & Huber, J.D. (2000). Putting parties in their place: Inferring party leftright ideological positions from party manifestos data. American Journal of Political Science 44: 94103. Glasgow, G. (2001). Mixed logit models for multiparty elections. Political Analysis 9(2): 116136. Goul-Andersen, J. et al. (2003). Danish election study 2001 [DDA-12516]. Odense: Dansk Data Arkiv. Gourevitch, P.A. (1986). Politics in hard times. Ithaca, NY: Cornell University Press. Green, J. & Hobolt, S.B. (forthcoming). Owning the issue agenda: party strategies and vote choices in British elections. Electoral Studies. Hays, J.C., Ehrlich, S.D. & Peinhardt, C. (2005). Government spending and public support for trade in the OECD: An empirical test of the embedded liberalism thesis. International Organization 59: 473494. Hellwig, T. (2001). Interdependence, government constraints and economic voting. Journal of Politics 63(4): 11411162. Hellwig, T., Ringsmuth, E. & Freeman, J.R. (forthcoming). The American public and the room to maneuver: Responsibility attributions and policy efcacy in an era of globalization. International Studies Quarterly. Hibbs, D.A., Jr. (1977). Political parties and macroeconomic policy. American Political Science Review 71: 14671487. Hooghe, L., Marks, G. & Wilson, C.J. (2002). Does left/right structure party positions on European integration? Comparative Political Studies 35: 965989. Huber, J.D. (1989). Values and partisanship in leftright orientations: Measuring ideology. European Journal of Political Research 17: 599621. Huber, J.D. & Inglehart, R. (1995). Expert interpretations of party space and party locations in 42 societies. Party Politics 1: 73111. Huber, J.D., Kernell, G. & Leoni, E.L. (2005). Institutional context, cognitive resources and party attachments across democracies. Political Analysis 13: 365386. Inglehart, R. (1990). Culture shift. Princeton, NJ: Princeton University Press. Iversen, T. (2005). Capitalism, democracy and welfare. Cambridge: Cambridge University Press. Iversen, T. & Cusack, T. (2000). The causes of welfare state expansion: Deindustrialization or globalization? World Politics 52: 313349. Iversen, T. & Wren, A. (1998). Equality, employment and budgetary restraint: The trilemma of the service economy. World Politics 50: 507546. Jowell, R. et al. (2003). European social survey, 2002/2003. London: Centre for Comparative Social Surveys, City University. Kedar, O. (2005). How diffusion of power in parliaments affects voter choice. Political Analysis 13: 410429. Kitschelt, H. (1994). The transformation of European social democracy. Cambridge: Cambridge University Press. Kitschelt, H. & Rehm, P. (2005). Work, Family and Politics: Foundations of Electoral Partisan Alignments in Postindustrial Democracies. Paper presented at the annual meeting of the American Political Science Association, Washington, DC. Kriesi, H.-P. (1998). The transformation of cleavage politics. European Journal of Political Research 33: 165185. Kriesi, H.-P. et al. (2006). Globalization and the transformation of the national political space: Six European countries compared. European Journal of Political Research 45: 921956.
2008 The Author(s) Journal compilation 2008 (European Consortium for Political Research)

the salience of leftright ideology

709

Laver, M. & Hunt, W.B. (1992). Policy and party competition. London: Routledge. Lewis-Beck, M.S. (1988). Economics and elections. Ann Arbor, MI: University of Michigan Press. Lipset, S.M. & Rokkan, S. (1967). Party systems and voter alignments. New York: Free Press. Manza, J. & Brooks, C. (1999). Social cleavages and political change: Voter alignments and US party politics. Oxford: Oxford University Press. Marks, G. (ed.) (2006). Comparing Measures of Party Positioning: Expert, Manifesto and Survey Data. Special issue of Electoral Studies 26(1). Marks, G. & Steenbergen, M. (1999). Expert survey on national parties and the European Union. Available online at: http://www.unc.edu/~gmarks (accessed 5 December 2006). Meguid, B.M. (forthcoming). Party competition between unequals: Strategies and electoral fortunes in Western Europe. Cambridge: Cambridge University Press. Miller, W.E. et al. (1999). Policy representation in Western democracies. Oxford: Oxford University Press. Pierce, R. (1999). Mass-elite issue linkages and the responsible party model of representation. In W. Miller et al. (eds), Policy representation in Western democracies. Oxford: Oxford University Press. Pierson, P. (1996). The new politics of the welfare state. World Politics 48:143179. Pitkin, H. (1987). The concept of representation. Cambridge, MA: Harvard University Press. Pontusson, J. (1995). Explaining the decline of European social democracy: The role of structural economic change. World Politics 47: 495533. Rivers, D. (1988). Heterogeneity in models of electoral choice. American Journal of Political Science 32(3): 737757. Rogowski, R. (1989). Commerce and coalitions. Princeton, NJ: Princeton University Press. Rueda, D. (2005). Insider-outsider politics in industrialized democracies: The challenge to social democratic parties. American Political Science Review 99: 6174. Sanders, D. et al. (2002). British election study, 2001/02 [dataset]. Colchester: University of Essex. TNS Sofres. (2007). Nouveaux clivages/clivages anciens. Poll conducted for La Fondation Jean Jaurs and Le Nouvel Observateur, 29 March. Available online at: http://www.tnssofres.com/etudes/pol/290307_clivage.pdf (accessed July 2007). Train, K. (2003). Discrete choice methods with simulation. Cambridge: Cambridge University Press. Warwick, P.V. (2002). Toward a common dimensionality in West European policy spaces. Party Politics 8: 101122.

Address for correspondence:Timothy Hellwig, Department of Political Science, University of Houston, 447 Philip G. Hoffman Hall, Houston, TX 77204, USA.

2008 The Author(s) Journal compilation 2008 (European Consortium for Political Research)

Das könnte Ihnen auch gefallen