Three million Trotskyists? Explaining extreme left voting in France in the 2002 presidential electionejpr_1901 359..392 NATHAN SPERBER Department of Politics & International Relations, University of Oxford, UK Abstract. In the first round of the 2002 French presidential election, three million voters (10.4 per cent of the national vote) supported Trotskyist candidates. This unprecedented electoral result has received little academic attention. This study aimed to identify the strongest socio-demographic and attitudinal predictors of support for the new extreme left in 2002.A multivariate framework was applied in a series of models,using data from the 2002 French Electoral Panel.The study also aimed to understand the rise of the Trotskyists in the context of broader social and political developments.The analysis was grounded in series of hypotheses constituting a model of class voting in postindustrial France. Overall, the analysis tended to confirm the predictions of the model, with younger voters at the lower end of the service sector being the most likely to support the three Trotskyist parties. With regard to attitudes, opposition to economic liberalism proved the strongest single predictor of Trotskyist voting, followed by liberal attitudes on cultural issues, political distrust and political disengagement. However, in terms of economic attitudes, Trotskyist voters still came out as surprisingly close to mainstream left voters. In conclusion, it is argued that models of class voting should reconsider the political role of social class in a postindustrial context, and pay particular attention to the trajectories of different classes over time in terms of changing employment conditions and life chances in order to understand how class is likely to shape party preferences. Introduction In the first round of the 2002 French presidential election, three candidates referring to themselves as ‘Trotskyists’ obtained 10.4 per cent of the national vote. The three candidates, Arlette Laguiller, Olivier Besancenot and Daniel Gluckstein,represented,respectively,Lutte Ouvrière (Working-class Struggle), the Ligue Communiste Révolutionnaire (Revolutionary Communist League, LCR) and the Parti des Travailleurs (Workers’ Party).These names were by no means mere labels devoid of revolutionary intent. All three candidates, for instance, declared that they agreed that capitalism should be overthrown and that the ‘permanent revolution’ – a concept dear to Trotskyists – should be put into motion. In the first years of the twenty-first century, at a time when Communist ideology,not to speak ofTrotskyism,seems to have lost its momentum both in Western politics and on the world stage, the success of Trotskyist European Journal of Political Research 49: 359–392, 2010 359 doi: 10.1111/j.1475-6765.2009.01901.x © 2009 The Author(s) Journal compilation © 2009 (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 parties at a presidential election in France may appear deeply puzzling. Nonetheless, the rise of the new French extreme left has been largely understudied. The rise of the Trotskyist parties squarely contradicts the predictions of prevailing theories of electoral change in advanced democracies.At first sight, they offer a surprising combination of so-called‘old’ and‘new’ politics.In 2002, they focused their campaigns on the socio-economic dimension of political competition, violently denouncing economic inequality, flexible labour market policies and rising corporate profits. Most of their electoral support came from younger voters to be found in the manual working class and the lower end of the service sector. In terms of organisation, these parties remain small and deprived of a political machine comparable to that which the French Communist Party (CP) still manages to maintain in many localities.Thus, in 2002, they almost exclusively relied on the media for campaigning. Their rapid rise and the seemingly transient nature of their success point to a picture where party attachment plays little role. This article will attempt to argue against certain misconceptions regarding the Trotskyists. First, they should not be seen merely as the tail of the comet of France’s postwar radical left.This old radical left is embodied in France by the Communist Party – a once-glorious mass party now in disarray (the CP obtained only 3.4 per cent of the vote in the 2002, a poor showing compared to its 21.3 per cent in 1969).Whereas CP voters in France are older than average, supporters of the new extreme left are found among younger age groups (Knapp 2004).This study provides evidence that the two types of voters in fact follow markedly different socio-demographic and attitudinal patterns.In terms of party organisation, the new radical left also contrasts sharply with the old. Yet the Trotskyist parties also fail to correspond to what is usually described as ‘new politics’ in the literature (e.g., Dalton 1988; Inglehart 1977, 1990, 1997); they cannot be defined as ‘New Left’, or ‘left-libertarian’ in Kitschelt’s (1988) sense. First, their main emphasis is on the material, and not the postmaterial, dimension of political competition. Despite their adoption of very liberal positions on issues such as immigration or women’s rights, especially the LCR, this libertarian element is secondary in their platforms and rather discrete in their campaigns. Second, most of their electoral support does not come from the middle class, whereas this is usually held to be a typical attribute of a New Left or left-libertarian party. The very notion that new political parties on the left of the political spectrum in wealthy industrialised nations will tend to be left-libertarian, and emphasise postmaterialist goals over materialist ones, as argued by Inglehart (1990, 1997), appears to be contradicted by these parties’ electoral success. Finally, it might also be tempting to emphasise only the protest character of Trotskyist voting and therefore to assimilate the new extreme left with the 360 nathan sperber © 2009 The Author(s) Journal compilation © 2009 (European Consortium for Political Research) anti-system extreme right, represented in France by the Jean-Marie Le Pen’s Front National. This third possible misconception about the new French extreme left has proven somewhat popular among French commentators (e.g., Taguieff 2002; Lévy 2007), but this view hardly seems defensible given that in terms of party platforms the extreme left and the extreme right were radically opposed on virtually all economic and cultural issues in 2002. The primary aim of this study is to provide an account of Trotskyist voting in France in 2002, paying attention to both socio-demographic and attitudinal factors. For this purpose, a quantitative analysis using multivariate logistic regressions is carried out. However, the analysis was also grounded in a more general theoretical framework allowing one to understand the changing influence of class on the vote within advanced capitalist countries. Drawing on a large literature describing how employment relations and class conditions in France were affected in the past few decades in the process of transition from an industrial to a postindustrial society, an historical interpretation of the electoral rise of the Trotskyist parties is provided, and a series of hypotheses constituting a model of class voting in France are put forward and tested. It will emerge that class voting is highly relevant to understanding the rise of the new radical left in France,although it might not be class voting as we used to know it. Indeed, the Trotskyist parties do not represent the extreme left of an industrial society,but of a postindustrial one,and their rise has to be understood in relation to the process of transition from the former to the latter.As such,the French extreme left constitutes a genuinely new political development. The setting The rise of the Trotskyist parties For a long time following the official foundation of the (Trotskyist) Fourth International in 1938, French Trotskyist organisations refused to engage in the electoral process and typically denounced elections as a sham. The argument usually put forward was that bourgeois democracy was but a formal façade to capitalist exploitation, and that the overthrowing of the capitalist order would only come through proletarian revolution and not elections. However, since the presidential election of 1969, there has always been at least one candidate representing a Trotskyist organisation in the first rounds of presidential contests. While never actually aiming to assume office or even enter a governing coalition, the French Trotskyists have mainly used elections as a way of making their positions known to the public, as well as promoting their respective organisations by attracting members and militants (Raynaud 2006; Reynié 2007). three million trotskyists? 361 © 2009 The Author(s) Journal compilation © 2009 (European Consortium for Political Research) Trotskyist presidential candidates between 1969 and 2007 were consistently affiliated to one of the three parties already mentionned: Lutte Ouvrière, the Ligue Communiste Révolutionnaire and the Parti des Travailleurs. Until the 1990s, the three parties put together never obtained more than 2.5 per cent of the vote. In 1995, however, Trotskyist voting reached 5.3 per cent (1.6 million voters), after which it peaked at 10.4 per cent (3 million voters) in 2002 before declining to 5.8 per cent (2.1 million voters) in 2007.1 The two strongest Trotskyist parties in electoral terms have been Lutte Ouvrière (LO) and the Ligue Communiste Révolutionnaire (LCR), whereas the Parti des Travailleurs (PT) never scored over 0.5 per cent. While LO and the PT still exist today, the LCR disappeared in 2009 to give way to a new entity, the NPA (Nouveau Parti Capitaliste). Until this recent development, the LCR was the official French section of the Fourth International. It was also by far the most open of the three Trotskyist parties in terms of doctrine as well as social composition. Often taking anti-nuclear or feminist positions in addition to the more classic revolutionary stance, it pursued a systematic strategy of accompanying a variety of social movements,both‘materialist’ and ‘postmaterialist’ (Raynaud 2006; Bensaïd 2002). By contrast, LO still overwhelmingly concentrates its discourse on the manual working class and this in turn is reflected in a more working-class heavy party membership.Its organisation has often been denounced as opaque and secretive (e.g., CohnBendit & Cohn-Bendit 2002). The case of the PT is more complicated. Although its candidate in 2002 declared himself to be a Trotskyist, the party does not officially refer to itself as such and welcomes non-Trotskyist militants. It is,however,the heir to the Organisation Communiste Internationaliste (OCI), a radical Trotskyist group that used to denounce the Fourth International for being too supportive of the Soviet Union (Raynaud 2006; Campinchi 2001). It might appear surprising that the threeTrotskyist parties never united into a single political force, all the more so since their electoral platforms differed quite little in practice, at least as far as LO and the LCR are concerned.A large part of the explanation for this puzzle lies in the intense rivalries, dating back to the years of immediate postwar France, that marked the groups from which these organisations emerged (Pina 2001; Raynaud 2006). These rivalries were in turn based on doctrinal differences in the context of the Cold War, mostly pertaining to the identity of the Soviet Union and various other Communist regimes around the world, and to the exclusive role of the manual working class in the revolution. The creation of the NPA, as the successor to the LCR, has significantly altered the landscape of the radical left in France. Following a good showing at the 2007 presidential elections (4.3 per cent), the LCR’s leader, Olivier Besancenot, announced the creation of a new party. Officially established in Febru- 362 nathan sperber © 2009 The Author(s) Journal compilation © 2009 (European Consortium for Political Research) ary 2009, the Nouveau Parti Capitaliste resulted from the merger of the LCR with other like-minded political groups (Le Monde 2009). Interestingly, the NPA is no longer an officially Trotskyist organisation, and the formal link to the Fourth International has been abandoned. For their part, the LO and the PT refused to dissolve into the NPA at the moment of its creation, and seem bent on maintaining their independence from the new party. The 2002 presidential election The 2002 presidential election, which produced a shock in France as a result of the qualification for the presidential run-off of the extreme right candidate, Jean-Marie Le Pen, was marked by the lowest turnout for a presidential election in the Fifth Republic (71.6 per cent), as well as by a general climate of distrust in politics. The five previous years, from 1997 to 2002, had seen a Socialist Prime Minister, Lionel Jospin, govern with a socialist parliamentary majority but a right-wing president, Jacques Chirac. In 2002, both politicians were candidates for the presidency in a context of economic slowdown after the 1997–2000 recovery. As the campaign went on, they were both seen as increasingly disconnected from the French electorate (Courtois 2002). Jospin, trying to reach the centre ground, declared his programme was ‘not socialist’, thus infuriating many voters on the left.At the same time, the three Trotskyist candidates, while declaring themselves favourable to social revolution, chose to campaign on immediate socio-economic concerns, such as the introduction of a tax on capital to finance pensions, making public transportation free and more generally strong public services (Pina 2007).The platforms of the LO and the LCR were fairly similar, the LCR adding feminist and anti-nuclear concerns. The PT campaigned on the defence of public services and a strong anti-Maastricht Treaty stance. Eventually,Arlette Laguiller, the candidate for LO, obtained 5.7 per cent of the vote, Olivier Besancenot from the LCR 4.3 per cent and André Gluckstein from the PT 0.5 per cent. The CP reached a meagre 3.4 per cent, while the incumbent president, Jacques Chirac, came to a surprisingly low 19.9 per cent, as the extreme right leader Le Pen (16.9 per cent) ended in front of the incumbent premier Jospin (16.2 per cent). An understudied political oddity The rise of the Trotskyist parties has led to few academic studies, especially compared to the large attention received by Jean-Marie Le Pen’s extreme right. In France, scholarship on the subject has remained almost exclusively qualitative, with a substantial number of historical enquiries into the roots of three million trotskyists? 363 © 2009 The Author(s) Journal compilation © 2009 (European Consortium for Political Research) the three Trotskyist parties (Raynaud 2006; Bensaïd 2002; Campinchi 2001). Johsua (2007a, 2007b) has studied extensively the social origins, and political outlook, of succeeding generations of LCR militants, using both qualitative and quantitative methods. Research on extreme left voting based on the analysis of survey material has remained scarce, however, and has mostly taken the form of descriptive statistics (e.g., Sineau 2004; Grunberg & Schweisguth 2003). The only study to use statistical modelling on extreme left voting is Tiberj (2007), but the author focuses on a limited number of attitudinal predictors and excludes social class. This lack of scholarly attention is understandable, but not warranted. The success of the extreme right in Western Europe, especially, is considered by many academics to be more genuinely new than electoral support for the radical left, all the more so in a country like France where the CP has experienced high levels of electoral support in the past. However, equating the new extreme left with the legacy of the CP would be misleading. In effect, as will be shown, CP voters and Trotskyist supporters follow distinct socio-demographic and attitudinal patterns. Academic enquiry into the French extreme left is all the more justified as it appears, once again since 2007, to be on the rise. In particular, Besancenot’s initiative to create the NPA has received intense media coverage as the young Trotskyist leader’s personal popularity skyrocketed in the aftermath of Nicolas Sarkozy’s election. As of September 2008, a poll by the French institute Opinion Way credited him with 13 per cent of voting intentions on the hypothesis of a presidential election to be held then and there (Le Monde 2008b). Moreover, the NPA’s standing today appears to benefit from the LCR’s past strategy of activism within a wide array of social movements.The vast majority of NPA militants are also union members, and a substantial portion is affiliated to the trade union SUD-Solidaires, a radical organisation that has recently assumed a high-profile role in episodes of labour activism (Le Monde 2008a). Theoretical framework and hypotheses: A story of class voting in a postindustrial context A new framework for analysing the political role of class in the context of French postindustrial society will be discussed in this section.As noted earlier, Trotskyist parties in France emphasise the economic dimension of political competition, and therefore cannot be assimilated to Kitschelt’s (1988) ‘leftlibertarian’ parties. Neither do they fit the model of the ‘old left’, either social-democratic or communist, as set out in the academic literature (Dalton 1988; Kitschelt 1994; Inglehart 1990). Indeed, social-democratic and commu- 364 nathan sperber © 2009 The Author(s) Journal compilation © 2009 (European Consortium for Political Research) nist parties rose in Western Europe in the postwar period in the context of the advent of industrial Fordism. In the past few decades, however, these same European countries have undergone far-reaching social transformations. Whereas evidence points to the fact that social class still influences voting in advanced Western societies (e.g., Evans 1999), it can be hypothesised that as social classes have themselves changed over time, so has their role in terms of political integration and influence. Furthermore, the nature of partisan attachment in advanced democracies has also been altered, as the experience of a stable, deep-seated attachment to a given party has become less prevalent within the electorate.It is against this backdrop that the electoral success of the Trotskyists should be understood. Partisanship in question Partisan attachment has been argued to have lost its centrality in the voting process across Western democracies. Originally an American concept, the notion of ‘partisanship’ initially found only mild support within the European academic community. In Campbell et al.’s (1960) classic study,‘partisanship’ is defined as a durable, affective psychological identification with a party. As such, partisan loyalty not only has influence on the vote, it also serves to organise a voter’s relationship to the political environment more generally, providing the partisan voter with informational short-cuts and prompting someone to follow his or her party’s positions on various political issues. French survey evidence from the 1960s reveals that many manual workers had such a relationship with the CP (Michelat & Simon 2004). Today, by contrast, survey evidence points to a general pattern of decline in declared partisan attachment across the postindustrial world (Dalton 2000; Berglund et al. 2005). This trend has also affected France (Schweisguth 2002), where partisanship had often been assumed to play a comparatively modest role to start with, given the persistent instability of the party system. Various hypotheses have been put forward to account for the widespread decline in partisan attachment. While some authors have emphasised the more general ‘modernisation process’, including the increase in educational levels and the rise of the mass media as the new principal source of information in modern democracies (Dalton 1988, 2000), others have insisted that specifically political factors have played their part, especially through a process of disaffection towards poorly performing parties in government (Schmitt & Holmberg 1995; Berglund et al. 2005). This debate is not to be settled here. For the present purpose, it can be noted that the rapid rise of the Trotskyist parties in France since the 1990s seems to suggest that party attachment, defined as a stable and durable dispothree million trotskyists? 365 © 2009 The Author(s) Journal compilation © 2009 (European Consortium for Political Research) sition on the part of the voter, has played little role in the electoral success of the new extreme left.The story of their rise fits quite well with the notion that the mass media have come to short-circuit the informational function of the large postwar parties, most notably the CP. The media were crucial to the Trotskyist candidates’ success as they enabled candidates affiliated to small, little-known parties to acquire national fame. In sum, one might hypothesise that, regarding the specific issue of partisan attachment, the new extreme left finds its place in a ‘new politics’ world where partisanship has in effect declined in relevance. The debate on class voting In Lipset and Rokkan’s (1967) early study on the influence of social cleavages on party preferences in Western Europe, the social division between the working class and the middle class is reflected in the political division between left-wing social-democratic or communist parties on the one hand, and rightwing Christian-democratic or conservative parties on the other. In recent decades, this model has come under marked criticism as numerous authors have argued for a strong decline in the influence of social class on voting behaviour (Rose & McAllister 1986; Dalton 1988; Inglehart 1990; Franklin et al. 1992; Clark & Lipset 1991, 1993). Yet several authors have been keen to denounce the methodological flaws of the early decline-of-class literature (Heath et al. 1985, 1987; Manza et al. 1995; Evans 1999). First, the favourite analytical tool of those who argued for a decline in class voting, the Alford index, does not allow one to distinguish the impact of a genuine alteration in class-based voting behaviour at the individual level from the effect of compositional changes in the macro-level class struc- ture.2 Second, and more importantly for the present purpose, this literature has usually operationalised class by means of the traditional manual/non-manual dichotomy, itself a symptom of an ‘old politics’ outlook. By offering a simplistic, presumably ahistorical, interpretation of class, many authors ignored significant and evolving social variations within the non-manual category, which has dramatically increased in size and diversified since the time of Lipset and Rokkan. Drawing on the Erikson-Goldthorpe class schema instead of the manual/ non-manual dichotomy, the contributors to Evans (1999) attempted to show that class retained much of its previous influence on the vote in advanced democracies, and that what was usually interpreted as a decline in class voting had merely been a trendless fluctuation. Among the contributors to Evans’s collection, however, De Graaf and Nieuwbeerta (1999) struck a different note, 366 nathan sperber © 2009 The Author(s) Journal compilation © 2009 (European Consortium for Political Research) arguing for an overall decline, if not disappearance, of class voting over the postwar period in advanced democracies, even when using the EriksonGoldthorpe schema. This study fully embraces the argument that the simple dichotomy between manual and non-manual activities is unlikely to capture the various possible influences of class position on the vote in postindustrial societies. It adopts the neo-Weberian approach of class devised by Erikson and Goldthorpe (1992) (see infra), and rejects the manual/non-manual dichotomy.3 If class voting may well have declined, it cannot be seriously said to have disappeared (De Graaf & Nieuwbeerta 1999).4 In particular, it remains susceptible to party strategies of class mobilisation (Evans et al. 1999; De Graaf et al. 2001).This last point is likely to be especially relevant with regard to the three Trotskyist parties in France, whose rhetoric is all but classist. A new type of proletarian worker in a service economy A new framework for understanding class voting in France has to take into account the dramatic changes that have affected the country’s economic and social structures since the 1970s.As industrial employment started to decline in the mid-1970s, the Fordist production model was progressively called into question and, as in most other advanced capitalist countries, new industrial patterns emerged (Piore & Sabel 1984; Castells 1996; Boltanski & Chiapello 1999; Gershuny 2005).As French firms increasingly pursued strategies of ‘flexible specialisation’ (Piore & Sabel 1984), the practice of subcontracting diffused throughout French industry to reach a share of 21 per cent of industrial output in the 1990s (Hannoun & Guerrier 1996). In addition, French firms became, on average, smaller in size, reversing a secular trend (Marchand & Thélot 1997). Subcontracting production to smaller specialised units allows business to avoid agreements with trade unions, and work conditions are usually worse in such units (Gershuny 2005). Non-standard job contracts, in the form of part-time or temporary work, have also become common practice since the 1970s, affecting in particular the lower end of the service sector and the manual working class (Boltanski & Chiapello 1999). For each of these two categories, corresponding respectively to class III and classes VI and VII in the Goldthorpe seven-class schema (see infra), unemployment averaged 20 per cent in the 1990s (Boltanski & Chiapello 1999; Chenu 2005). In addition, there seems to have been in France a generational trend concerning all social classes, whereby those who entered the labour market in the 1980s or after experienced higher unemployment and lower income than the previous generation, despite superior educational credentials (Chauvel 1998). three million trotskyists? 367 © 2009 The Author(s) Journal compilation © 2009 (European Consortium for Political Research) As the manual working class was both declining in number and drawn away from a Fordist production model, the lower end of the service sector (Goldthorpe’s class III) expanded. Mostly composed of women, this class is characterised by a high diversity in work situations, from routine administrative work in the public sector to supermarket employee or cleaning woman (Chenu 1990). However, as shown by Chenu (2005), the working conditions and standards of living of this class as a whole have steadily deteriorated in France since the 1960s. Whereas routine non-manual employees had an average income 20 per cent higher than the average manual working-class income in 1954, it was only 2 per cent higher in 2002, and the increase in job insecurity and unemployment has affected both classes alike. In a comparative study, Esping-Andersen (1993: 229) has called the low-skilled service class a ‘service proletariat’. The political integration of manual workers and lower-grade employees From the end of the Second World War to the 1970s, the French manual working class experienced a high degree of political integration. This integration was closely associated with the presence of a very strong CP, which obtained on average one-quarter of the national vote from 1944 to 1972, and with the Communist-led Confédération Générale du Travail (CGT), the largest trade union. A plethora of local associations were dependent on the CP for functioning and funds, and, in addition, the party had active militant cells in many French industrial firms (Courtois & Lazar 2000; Lavabre & Platone 2003). Surveys conducted in the 1960s show that a majority of French manual workers expressed a strong feeling of belonging to the working class, believed in the reality of the class struggle and voted for the left, especially for the CP (Michelat & Simon 2004). Yet, by the 1990s little was left of this working-class political culture. Industrial restructuring simply erased from the map the workplaces where the CP used to be present (Lavabre & Platone 2003), and strike action, measured in terms of the number of days of work lost to strikes, was down to less than one-tenth of what it had been in the 1950s and retained a strong presence only in transportation and the public sector (Groux 1998, 2007).5 Finally, the CGT’s membership plummeted to 10 per cent of the workforce and only 5 per cent of the private sector, while aging substantially (Rey 2004; Labée 1996). In the early 2000s, trade unions in France, and especially the CGT, were often criticised for having only marginally modified their theoretical outlook since the 1960s, and for paying little attention to the issues of temporary work contracts and unemployment, both of which have affected the younger generation much more than the older one (Groux 1998; Beaud & Pialoux 1999). In 2002, only 6 368 nathan sperber © 2009 The Author(s) Journal compilation © 2009 (European Consortium for Political Research) per cent of the manual working class cast a CP vote, many of them belonging to the older generation of workers who had entered the labour market during the postwar Long Boom rather than during the crisis years that followed (Lavabre & Platone 2003; Knapp 2004).The manual workers from the younger generation have been described as politically distrustful, hostile to the political system and opposed to economic liberalism (Michelat & Simon 2004). Under such conditions, it can be advanced that younger manual workers are likely to be attracted by the new extreme left. The story for routine non-manual employees is quite a different one since this class never reached the level of political integration experienced by the manual working class in the 1960s. Trade unions have always had lower membership in this class than in the manual working class, and their presence has declined since the 1970s. Rather, routine employees have often been described as a comparatively apolitical class, more prone to contemplating the hope of upward social mobility for their children than to joining the working class in any radical political struggle (Crozier 1965; Bourdieu 1977). However, this view has come under increasing criticism over time (Schweisguth 1983; Chenu 2005), and in effect voting patterns of routine employees in France are closer to those of the manual working class than ever before (Chenu 2005). Given a clearly disadvantaged position in French society in terms of income and employment security, many routine employees are likely to be attracted to the left.At the same time, a centrist campaign on the part of the Socialist Party, as well as the characteristic workerist tradition of the CP, would tend to bring left-wing members of the ‘service proletariat’ towards the Trotskyist parties. In addition, for the same reasons as for the manual working class, it might be expected that the Trotskyist parties’ appeal would be stronger for the younger generation of lower-grade employees. A model of Trotskyist voting in France The preceding theoretical and historical considerations suggest that Trotskyist voting is likely to take place in a context where: partisan ties have declined in relevance; social class may still influence voter preference for leftist economic policies; and distrust towards established political parties on the part of younger voters from less advantaged social classes calls for an alternative to the parties of the postwar left. In turn,it is argued that distrust on the part of the working class and the lower end of the service sector vis-à-vis the established left-wing parties in France – namely the Socialist Party and the Communist Party – can be understood primarily in relation to the dramatic changes that have affected these two social classes over the past few decades in terms of labour market conditions,employthree million trotskyists? 369 © 2009 The Author(s) Journal compilation © 2009 (European Consortium for Political Research) ment and living standards in the process of transition from an industrial to a postindustrial society. This model advances that labour market entrants from the 1980s onwards occupying manual working-class or lower-grade service occupations will be the most likely to vote for the new extreme left.By contrast, older left-leaning voters having experienced the postwar Long Boom during most of their working lives will be less sensitive to its appeal. The role of attitudes This study is also concerned with identifying attitudinal predictors of Trotskyist voting in France.Whereas social class is among the traditional determinants for analysing voting behaviour, it has tended to play a smaller part in France in accounting for electoral outcomes than in most other advanced capitalist countries (Dalton 1988). Furthermore, the instability of the party system and the high level of electoral volatility recorded throughout the twentieth century (Bartolini & Mair 1990) have made the prevalence of ‘issue voting’ a longlasting characteristic of the French political system (Weakliem & Heath 1999; Fleury & Lewis-Beck 1993). As far as the 2002 presidential contest is concerned, several sets of attitudes can be associated with a likely tendency to cast a Trotskyist vote. Political distrust and disengagement As already noted above, support for the new extreme left is most likely to affect voters with low levels of partisan attachment and who feel dissatisfied with the two established parties of the French left. This prediction was made specifically for the members of the manual working class and for lower-grade employees, but there are grounds to believe that distrust in politics, on the one hand, and political disengagement, on the other, might predict extreme left voting across all social classes. France has been affected by a severe drop in public confidence in parliament and politicians since the 1970s (Dalton 1999; Mayer 2002), in a way rather similar to most other advanced democracies (Norris 1999). This mounting public distrust led to the rise of a strong extreme right party in the 1980s, the Front National, whose success has been strongly related to its anti-establishment appeal (Mayer & Perrineau 1992; Perrineau 2003). The three Trotskyist parties’ strong attacks on the political system are also likely to attract voters who feel they are no longer represented by mainstream politicians. 370 nathan sperber © 2009 The Author(s) Journal compilation © 2009 (European Consortium for Political Research) ‘Political disengagement’, defined as a lack of interest and participation in politics, is also a potential predictor of Trotskyist voting, provided that the disengaged citizens do not simply abstain from the vote. For instance, the extremist discourse of Trotskyist parties, like that of the Front National, might attract unsophisticated and ill-informed voters.Alternatively,one might expect an association of Trotskyist voting with political disengagement as a result of a class effect or a generational effect since, on the one hand, manual workers and lower-grade employees in France have been singled out for their tendency to abstain and to declare little or no interest in politics (Michelat & Simon 2004; Chenu 2005) and, on the other hand, the younger generation also has a clear tendency to be politically disengaged (Schweisguth 2002). Economic interventionism and cultural liberalism Although it has been asserted that advanced capitalist societies are moving away from material concerns towards giving a larger place to issues of quality of life and the environment (Inglehart 1990), political competition in France is still very much informed by the debate between economic interventionism on one side, and market liberalism on the other. As already pointed out, the refusal of economic liberalism and the denunciation of the iniquities of capitalism are the main components of the Trotskyist parties’ rhetoric. Accordingly, there are strong reasons to believe that voter attitudes on economic issues are among the main predictors of Trotskyist voting. It is also likely that electoral support for the new extreme left is associated with liberal attitudes on issues such as immigration, minority rights, the death penalty and so on. Although it has been suggested by several commentators that Trotskyist voters tend towards antisemitism (Taguieff 2002), xenophobia (Perrineau & Ysmal 2003) and ‘social-nationalism’ (Reynié 2005), Tiberj (2007) refutes such claims in his analysis of the attitudes of extreme left voters. In addition, given the electoral platforms of the Trotskyist parties, extreme left voters are unlikely to be culturally conservative, at least as far as the LCR and LO are concerned.The LCR, in particular, gave a substantial place to feminist and environmental concerns in its campaign, and LO, though usually very discrete on cultural matters, declared itself against all barriers to immigration. Data and methods The 2002 French Electoral Panel At the time it was conducted, the Panel Electoral Français (PEF) 2002 was the largest election survey ever done in France. Not actually a ‘panel’ in the usual three million trotskyists? 371 © 2009 The Author(s) Journal compilation © 2009 (European Consortium for Political Research) sense, it was composed of three waves administered at a few weeks’ interval each, in the context of a two-month election period with four successive consultations (the first and second rounds of the presidential election,followed by the first and second rounds of the legislative elections). The PEF 2002 was designed and conducted by two research teams – the CEVIPOF (Centre for Studies in French Political Life) in Paris and the CIDSP (Centre for Computerised Socio-Political Databases) in Grenoble – and one private political expert agency – the CECOP (Centre for Studies of Public Opinion).6 Only the second wave of the PEF 2002 is used in this study. The second wave, conducted in May 2002 after the presidential election and before the legislative election, was designed to be used as a post-election survey for the presidential election. The interviews of the second wave were conducted by telephone with a sample of 4,017 individuals, out of which 3,179 said that they had voted for a candidate in the first round. The dependent variable The analysis uses multivariate logistic regressions in three different kinds of models: with socio-demographic predictors, with attitudinal predictors and with both types of predictors in a mixed model. The dependent variable is Trotskyist voting, defined as casting a ballot for any one of the three Trotskyist parties (LO, the LCR and the PT). Only the respondents who reported having actually voted for a candidate in the first round of the election are included in the analysis. Furthermore, since the interest of this study is the electoral success of the new extreme left as a whole, aggregating votes for the three parties into one single variable seems warranted. Regression models were also run separately for LO and LCR voting (see Appendix Table A1). It emerges from these models that the predictors of LO and LCR voting differ little from those of Trotskyist voting as a whole. The independent variables Three age groups are distinguished: respondents born before 1940, respondents born between 1940 and 1960, inclusive, and respondents born after 1960. The first age group is composed mainly of retired individuals, who experienced the postwar Long Boom during most of their working lives. The third age group, by contrast, is mainly composed of respondents who entered the labour market during the economic slowdown of the 1980s and 1990s. Social class is operationalised by means of the Goldthorpe class schema (Erikson & Goldthorpe 1992). Based on an analysis of employment relations, 372 nathan sperber © 2009 The Author(s) Journal compilation © 2009 (European Consortium for Political Research) the Goldthorpe seven-class schema distinguishes whether an individual’s employment situation is regulated by the ‘service relationship’, which includes a long-term exchange with the employer and guarantees of job security and career opportunities, or the ‘labour contract’, which involves by contrast ‘a short-term specific exchange of money for effort’ (Erikson & Goldthorpe 1992: 41–42), and thus reduced opportunities and job security. For this study, Goldthorpe’s seven-class schema was collapsed into a fiveclass schema, also known as the ‘Goldthorpe-Heath class schema’, as advised by Evans (1992). Five social classes will therefore be considered: the salariat, corresponding to classes I and II in Goldthorpe’s seven-class schema, which is the group benefiting from the advantageous ‘service relationship’; the routine non-manual employees, class III, a highly feminised class composed of lowergrade workers in the service sector; the self-employed and independent, class IV; foremen and technicians, class V; the working class, corresponding to classes VI and VII in the seven-class schema, and composed of skilled and unskilled manual workers. For the unemployed and the retired, class position was determined by the last occupation held. Since a working woman’s economic interests might be heavily influenced by her husband’s class position, especially if the husband has a much higher-status occupation than his wife, a dummy variable for female routine non-manual employees with husbands members of the salariat was included in the models. The other socio-demographic predictors are gender, higher education, home ownership, marital status and employment in the public sector (all binary variables). A variable for income was also included, divided into nine increments and treated as continuous,7 as well as a binary variable for being employed under a non-standard work contract (typically entailing part-time or temporary work). Finally, religious practice, often argued to be one of the main socio-demographic predictors of voting behaviour in France (Michelat & Simon 1977, Cautrès 2004), was operationalised as a binary variable equal to 1 for respondents who reported attending mass at least once a month. Regarding attitudes, a strategy of constructing a number of theoretically grounded scales with high internal reliability was pursued (see Heath et al. 1994). This was seen as the best way to capture the various attitudinal dimensions identified as potentially relevant to Trotskyist voting, as opposed to a simple indicator of left-right self-placement. Although identification with the left or the right of the political spectrum is a particularly strong determinant of voting behaviour in France (Michelat & Tiberj 2007), left-right self-placement typically draws on both attitudes towards economic policy and attitudes on socio-cultural issues simultaneously (Van der Eijk et al. 2005). As such, the very ambiguity of this measurement makes it unsuitable to the purpose of this three million trotskyists? 373 © 2009 The Author(s) Journal compilation © 2009 (European Consortium for Political Research) study, which is to disentangle the effects of attitudes held on different policy dimensions. It proved possible to build suitable scales for all the sets of attitudes identified as potentially favouring Trotskyist voting, with the notable exception of attitudes towards the economy, which failed to combine in a scale with an acceptable Cronbach’s alpha. The failure of economic preferences to form a coherent cluster of attitudes in France has already been noted by Grunberg and Schweisguth (2003).Accordingly, three questions were used to form three separate variables: one on economic inequality (‘Do you consider it important to reduce the inequality between the rich and the poor?’); one on state dirigisme (‘In the face of economic difficulties, do you think the state should give more freedom to firms or exercise more control over them?’); and one on the labour market (‘Do you agree that layoffs on the part of private companies should be banned?’). For cultural liberalism, distrust in politics and political disengagement, three scales were constructed. The ‘cultural liberalism’ score (Cronbach’s alpha 0.79) is based on responses to eight separate questions pertaining to immigration, the death penalty, perceptions of tradition and so on. The ‘political distrust’ scale (Cronbach’s alpha 0.74) draws on five questions asking respondents whether they think politicians are honest and whether they trust political parties and various democratic institutions. The third scale, ‘political disengagement’, was constructed from four questions with an alpha of 0.75, and brings together a question about interest in politics and questions about the frequency of political discussions with family and friends. Finally, an index of political knowledge was built by adding up the number of correct answers to five right-or-wrong questions about the French political system. All the questions used are presented in the Appendix. Analysis and results Partisanship and the Trotskyist parties The PEF data reveal that only 23 per cent of Trotskyist voters declared in 2002 that they identified with any of the Trotskyist parties. Although strikingly low, this figure is not surprising. As noted earlier, the Trotskyist parties’ electoral rise began in the 1990s. Therefore they were unlikely to generate substantial partisan attachment within the electorate by 2002. Being small and relatively unknown to the public from the outset, these parties chose during the 2002 campaign to rely almost exclusively on the media appearances of their presidential candidates to foster support. Furthermore, the previous discussion also 374 nathan sperber © 2009 The Author(s) Journal compilation © 2009 (European Consortium for Political Research) suggested that they were likely to attract relatively disenfranchised voters distrustful of the political system. Such voters would in general tend to experience low levels of partisan attachment. Socio-demographic patterns As an introduction, Table 1 provides the distribution of extreme left voters for the socio-demographic variables retained for the analysis. Separate proportions are also given for LCR and LO voting, although not for the PT given the small number of PT supporters in the sample.The two social classes that were the most likely to support Trotskyist parties in 2002 were the lower end of the service sector and the manual working class (14 and 13 per cent, respectively). The very strong emphasis on the plight of the manual working class that marked LO’s campaign did not prevent lower-grade employees from giving a comparatively large number of votes to this party. The smallest proportion of extreme left supporters is to be found among the self-employed and independent (3 per cent), which is by no means surprising since this group is known to be a strong conservative force in France (Cautrès 2004). Individuals subject to non-standard work contracts, usually entailing less job security and a lower income, tend to support Trotskyist parties quite strongly (15 per cent). The distribution of Trotskyist voters as a function of age reveals that support for the new extreme left among the older generation is dismal (3 per cent),whereas it is the highest for individuals born after 1960 (14 per cent).Not surprisingly, home owners and individuals who attend church are less likely to give their support to parties of the new extreme left. The same can be said of married individuals, although an age effect is likely to be at play. Similarly, the negative association between higher education and Trotskyist voting is likely to be mediated by social class and income. At first sight, comparatively high support for the new extreme left among public sector employees (12 per cent) would seem to run counter to the previous discussion, which suggested that Trotskyist support was rather to be found among people with disadvantaged employment conditions. This association, however, is not surprising given the traditional leftist tendencies of the public sector in France. Moreover, a growing share of public sector employees does not benefit from the traditional civil service employment guarantees, and is instead employed on the basis of non-standard job contracts. As for the apparent support of women for the extreme left, it might be related to the propensity of French women to be distrustful of the political class (Sineau 2004), but could also be due to the prevalence in France of worse working conditions and lower earnings for women than for men, particularly within the routine non-manual class (Chenu 2005). Finally, the comparatively three million trotskyists? 375 © 2009 The Author(s) Journal compilation © 2009 (European Consortium for Political Research) high proportion of female LO voters relative to female LCR voters might be related to the fact that LO’s candidate in 2002,Arlette Laguiller, was a woman, although little hard evidence can be put forward to support this idea. Table 2 provides logit estimates for socio-demographic predictors of extreme left voting in 2002. In keeping with the hypotheses, Model 1 reveals that the associations of Trotskyist voting with both the lower-grade nonmanual class and the manual working class resist well multivariate analysis Table 1. Distribution of Trotskyist voters by socio-demographic attributes (percentages; N = 3,179) Trotskyist support LO support LCR support Gender Female 12 7 5 Male 9 4 4 Age Born before 1940 3 2 1 Born between 1940 and 1960 11 7 4 Born after 1960 14 7 6 Social class Salariat (I+II) 8 4 4 Routine non-manual (III) 14 8 6 Self-employed and independents (IV) 3 1 1 Foremen and technicians (V) 9 5 3 Manual working class (V+VI) 13 7 5 Employment sector Public 12 6 5 Private 9 5 4 Marital status Married 10 6 4 Not married 12 5 6 Home ownership Home owner 9 5 4 Not home owner 12 6 6 Higher education With higher education 9 5 4 Without higher education 11 6 4 Non-standard work contract 15 9 6 Church attendee 5 3 2 376 nathan sperber © 2009 The Author(s) Journal compilation © 2009 (European Consortium for Political Research) since both classes are significant at a 0.01 level. The dummy variable added to control for female routine employees with husbands in the salariat is also significant, suggesting the introduction of this variable was justified.8 Not surprisingly, income has a significant effect and is negatively associated with Trotskyist support. Being a woman appears as a significant positive predictor of extreme left support, and so does the fact of being from the younger generation, although there is less discrepancy than had been hypothesised between the size of the effect for the younger generation, born after 1960, and the size of the effect for the intermediate generation, born between 1940 and Table 2. Socio-demographic predictors of Trotskyist voting (multivariate logistic regressions; N = 2,651) Model 1 Model 2 coefficient coefficient Female 0.36*** 0.36*** Age Born before 1940 (reference) Born between 1940 and 1960 1.56*** 1.56*** Born after 1960 1.80*** 1.65*** Social class Salariat (I+II) (reference) Routine non-manual (III) 0.47*** 0.30 Self-employed and independents (IV) -0.41 -0.41 Foremen and technicians (V) 0.15 0.15 Manual working class (V+VI) 0.46** 0.48 Wife (III) husband (I+II) dummy variable -0.93** -0.95** Income -0.11** -0.11** Public sector 0.42*** 0.42*** Non-standard work contract 0.25 0.26 Home owner -0.04 -0.05 Higher education -0.07 -0.07 Married 0.03 0.03 Church attendee -0.29 -0.29 Class III ¥ born after 1960 0.44 Class VI+VII ¥ born after 1960 -0.07 Constant -3.42*** -3.37*** Pseudo R2 0.074 0.076 Notes: ** p < 0.05; *** p < 0.01. three million trotskyists? 377 © 2009 The Author(s) Journal compilation © 2009 (European Consortium for Political Research) 1960. This might be explained by the fact that both generations have experienced the economic crisis of the past few decades and its repercussions on working conditions and living standards, albeit the younger more acutely. Whereas being employed in the public sector appears as a positive predictor of extreme left voting, marital status, home ownership and higher education come out as insignificant, as do church attendance and the fact of being subject to a non-standard work contract. To give an illustration by predicted probabilities, an unmarried female lower-grade employee from the younger generation, employed in the private sector with a relatively low income, had a 24.9 per cent chance of voting Trotskyist in 2002. Moreover, the socio-demographic predictors for Trotskyist voting differ widely from those related to Communist Party,Socialist Party and Green Party voting (see Appendix Table A2 for models for these three parties). For the CP, being part of the younger generation has a negative effect on support, and no significant effect is associated with the lower-grade non-manual class.The same applies to the Socialist Party. As for support for the French Greens, it is positively predicted by higher education and a middle-class occupation. In Model 2, interaction effects have been added to test further the hypotheses of the model of class voting presented previously. It has been suggested that the strongest Trotskyist supporters were likely to be found among the younger generation of lower-grade employees and manual workers. At first sight, the estimates for the interaction effects appear disappointing since only the interaction effect for the lower-grade non-manual class is significant at a 0.1 level. However, the absence of a strong interaction effects is not surprising, given that Model 1 already shows a strong age effect operating across all social class. At any rate, the model of class voting presented seems supported by the fact the most likely Trotskyist voters are the younger members of the less advantaged classes. Attitudinal patterns In Model 3,Trotskyist voting is regressed on the attitudinal predictors identified previously (Table 3). All the variables are significant at a 0.01 level in the predicted direction. Refusal of laissez-faire, liberal views on cultural issues, distrust in politics,political disengagement and low levels of political knowledge all are strongly related to support for the new extreme left. Since some of the variables are constructed scales while others are simply derived from survey questions with a dichotomous answer choice, it is not easy to assess the relative importance of the predictors.Using odds ratios,it is found that the odds ratio of voting Trotskyist for an individual declaring opposition to economic liberalism on all three indicators compared to an individual taking the opposite stand on 378 nathan sperber © 2009 The Author(s) Journal compilation © 2009 (European Consortium for Political Research) the three indicators is 4.6.For the‘cultural liberalism’ scale,the odds ratio of the most culturally liberal compared to the least culturally liberal is 4.2, which indicates the large importance of liberal values for Trotskyist voting. However, since very few individuals occupy the two extremes for the scale for cultural liberalism,whereas there are large numbers at both ends for economic attitudes, it can advanced that attitudes on the economy have more explanatory power.It should be noted, nonetheless, that the odds ratios for economic attitudes calculated in the same way for Socialist Party and Communist Party support are, respectively, 3.0 and 13.4, suggesting that in fact the ‘Trotskyists’ are much less radical on the economy than one may be led to think. Mixed model Model 4 is obtained by bringing together into one multivariate framework the significant socio-demographic variables with the attitudinal predictors of Trotskyist voting (Table 4). The public sector variable ceases to be significant at a 0.05 level, confirming that the significant relationship between employment in the public sector and extreme left support found earlier was partly mediated by the specific leftist political leanings of the public sector in France. Income ceases to be significant, as well as gender, which is by no means surprising since, as has already been pointed out, the positive relationship between being a woman and support for the new extreme left was likely to be mediated by attitudes such as cultural liberalism and distrust in politics, which Table 3. Attitudinal predictors of Trotskyist voting (multivariate logistic regression; N = 2,815) Model 3 coefficient Economic inequality 0.29*** Control over firms 0.70*** Ban on layoffs 0.22*** Cultural liberalism 0.08*** Political distrust 0.09*** Political disengagement 0.11*** Political knowledge -0.16*** Constant -6.18*** Pseudo R2 0.101 Note: *** p < 0.01. three million trotskyists? 379 © 2009 The Author(s) Journal compilation © 2009 (European Consortium for Political Research) are found more often among women than among men (Sineau 2004). The positive effect of being a member of the manual working class on Trotskyist support is divided by two and ceases to be significant, which might be attributed the prevalence of left-wing attitudes on the economy among manual workers (Michelat & Simon 2004), which are in turn conducive to Trotskyist support. By contrast, the positive effect of being a lower-grade employee not only resists the analysis, but is in fact strengthened by the introduction of the attitudinal variables. This important result suggests that, although less driven towards Trotskyist voting by their political attitudes than other social classes, Trotskyist supporters from the lower service sector will still tend to vote for the new extreme left for other reasons, associated in one way or another with their Table 4. Mixed model of Trotskyist voting (multivariate logistic regression; N = 2,397) Model 4 coefficient Female -0.01 Age Born before 1940 (reference) Born between 1940 and 1960 1.34*** Born after 1960 1.44*** Social class Salariat (I+II) (reference) Routine non-manual (III) 0.51*** Self-employed and independents (IV) -0.01 Foremen and technicians (V) 0.22 Manual working-class (V+VI) 0.24 Wife (III) husband (I+II) dummy variable -0.73 Income -0.04 Public sector 0.27* Economic inequality 0.29*** Control over firms 0.56*** Ban on layoffs 0.24*** Cultural liberalism 0.08*** Political distrust 0.08*** Political disengagement 0.07*** Political knowledge -0.19*** Constant -6.48*** Pseudo R2 0.139 Notes: * p < 0.1; *** p < 0.01. 380 nathan sperber © 2009 The Author(s) Journal compilation © 2009 (European Consortium for Political Research) experience as a class. Previously, it was advanced that the decisive experience for lower-grade employees in this case would be the worsening labour market and employment conditions that have affected their class as a whole since the 1970s. Although nothing conclusive on this precise point can be drawn from the analysis, this view would tend to be confirmed by the results. Regarding age, the fact of being born after 1940 retains a strong predictive power when controlling for attitudes, although its effect is a little reduced. Although the coefficient remains higher for the younger generation than for the intermediate generation, both are very close. As already suggested, the small difference found between these two generations in terms of extreme left support might also provide evidence that it is not so much the fact of having entered the labour market in the 1980s or later that is decisive, but the fact of having experienced during most of one’s working life the years of economic crisis. As for the attitudinal predictors of Trotskyist voting, they all resist the introduction of socio-demographic variables, and all remain significant at a 0.01 level. Discussion and conclusion That three million voters in an advanced capitalist nation like France would support Trotskyist parties in an election held thirteen years after the fall of the Berlin wall can be seen as deeply puzzling.This study aimed first and foremost to identify who the ‘Trotskyists’ were, and what were the strongest predictors of support for the Trotskyist parties in 2002. It also aimed to provide a test for a series of hypotheses put forward in the form of a model of class voting suited to a postindustrial society like France. On the socio-demographic side, the strongest single predictor of Trotskyist voting in 2002 is age – more specifically the fact of belonging either to an intermediate generation, born between 1940 and 1960, or to a younger generation, born after 1960, as opposed to belonging to an older generation born before 1940. The effect is the strongest for the post-1960 generation, thus tending to support the generational hypotheses presented, although the difference between the younger and the intermediate generations was not as wide as predicted. It remains impossible to assert with certainty whether the reluctance of individuals older than 60 to vote for the new extreme left is attributable to a generational effect or an age effect, although one might suspect both are at play. As far as class is concerned, the individuals who were the most likely to support Trotskyist parties in 2002 were to be found in the lower end of the service sector – a highly feminised social category that has been called a three million trotskyists? 381 © 2009 The Author(s) Journal compilation © 2009 (European Consortium for Political Research) ‘service proletariat’ (Esping-Andersen 1993). This finding is an important one since it reveals that, in a postindustrial society, members of less advantaged social classes may still have a privileged relationship with political forces on the left running on a radical economic platform. Put differently, the strong links that existed forty years ago between the French working class and the CP may have found here a kind of avatar, but with different actors and in a social and economic context dramatically altered. On the side of political preferences, the attitudes that retained the strongest effect after a multivariate analysis were, unsurprisingly, attitudes towards the economy. In addition, three other strong attitudinal predictors of Trotskyist voting were identified – namely cultural liberalism, distrust in politics and political disengagement. In respect of the last two, extreme left voters share a common attitudinal trait with the supporters of the French extreme right, whereas in respect of cultural liberalism these two groups of voters are poles apart.As far as economic preferences are concerned, the extreme left supporters came out as less radical on this dimension than the older CP voters, and closer to the supporters of the mainstream Socialist Party than might have been expected. This would suggest, in turn, that many of these voters are very far from being actual ‘Trotskyists’. A closer look at the PEF data, however, leads one to qualify this view to some extent. What emerges is not so much the moderation of Trotskyist supporters in 2002, as a surprising radicalism on the part of Socialist Party supporters. For instance, according to the PEF, as much as 51 per cent of Socialist Party voters either ‘totally agreed’ or ‘rather agreed’ that ‘layoffs on the part of private companies should be banned’ in 2002 (for the Trotskyist voters, the proportion was 61 per cent). This puzzle notwithstanding, the ideological proximity between extreme left voters and Socialist Party voters on the economic dimension of political competition has prompted Tiberj (2007: 147) to argue that the Trotskyist parties are ‘redundant’ vis-à-vis the rest of the French left. According to this author, the essential explanation for the Trotskyists’ rise lies in a bid on the part of Trotskyist voters to influence the Socialist Party to take a leftist policy turn. This should be accepted as part of the story, and indeed one of the hypotheses proposed here has been that younger voters of less advantaged social classes were susceptible to be drawn towards the Trotskyist parties in 2002 precisely because they perceived the Socialist Party as too moderate in terms of economic policy. However,Tiberj’s attempt to explain away the rise of the Trotskyist parties in strictly political-electoral terms is not only theoretically one-sided, but misses out on the originality of these parties as bearers of a new brand of ideological radicalism within a postindustrial society. By contrast with Tiberj’s approach, this study has tried to provide an account of the 382 nathan sperber © 2009 The Author(s) Journal compilation © 2009 (European Consortium for Political Research) Trotskyists’ rise by putting it in the perspective of certain dramatic societal transformations that have affected France in the past few decades, particularly regarding employment relations. The Trotskyists’ electoral success seems to contradict the widely held notion that the collapse of Soviet Communism eliminated the potential appeal of the radical socialist left in advanced democracies, and thus ‘shortened the axis of competition over alternative modes for allocating resources’ (Kitschelt 1994: 31). In addition, this electoral success is also in apparent contradiction with the idea, defended by Inglehart (1990, 1997), that new political parties appearing on the left of the political spectrum in wealthy capitalist countries will tend to put the emphasis on ‘postmaterial’ concerns such as the environment or quality of life issues, as well as to attract middle-class voters. Although not ‘new politics’ in Inglehart’s sense, the parties that have been considered here are definitely new politics.Their crucial reliance on the media, their rather slim organisational structures, and in particular their appeal to younger and relatively disaffected voters, qualifies them as such. In order to account for the new extreme left’s success, this study has put forward a generational argument of a widely different character from Inglehart’s (1977, 1990) generational model of political change. Instead of being driven towards postmaterialism, it was hypothesised that voters from post-1960 cohorts might instead be driven towards radical ‘materialist’ politics.As younger lower-grade employees or manual workers in particular were likely to experience unemployment, job insecurity and a relative drop in living standards upon their entry in the labour market from the 1980s onwards, it was proposed that such a formative experience of economic insecurity, together with a process of political disaffection towards the established parties, would be conducive to Trotskyist voting. Moreover, this study also seems to contradict Kitschelt’s ‘realignment hypothesis’ (Kitschelt 1994; Kitschelt & McGann 1995). According to the realignment hypothesis, blue-collar workers and low-skilled employees in the private sector, becoming aware that their employment conditions and opportunities are now highly dependent on the ability of their own employers to cope with foreign competition in a globalised economy, have undergone a process of political realignment whereby they have ceased to support economic interventionism and socialism to favour neoliberal economic policies. For Kitschelt, then, with the exception of the public sector, the privileged association between the less advantaged classes and the political left is something of the past. Although it is true that manual workers and lower-grade employees are also comparatively strong supporters of the French extreme right (Mayer & Perrineau 1992; Perrineau 2003), the fact that the strongest supporters of the new extreme left are to be found among these two classes three million trotskyists? 383 © 2009 The Author(s) Journal compilation © 2009 (European Consortium for Political Research) suggests that the realignment hypothesis has been overstated. Moreover, these results converge with those of another study contesting Kitschelt’s realignment hypothesis in the cases of France and Denmark (Ivarsflaten 2006). The findings of this study on the electoral rise of the new French extreme left call for further research, both idiographic and comparative. In France, after a relative electoral setback in 2007, the extreme left appears to be once again on the rise. It has assumed, however, a different shape since the transformation of the LCR into the NPA in February 2009. The most dynamic political force on the left of the Socialist Party today, the NPA is not an officially Trotskyist organisation. Furthermore, it remains to be seen whether the French extreme left has analogues in Europe. Although it might be tempting to dismiss it as a Gallic oddity explainable by a long, highly idiosyncratic tradition of political activism, there might be less and less ground to do so. In particular, in the last German federal election in 2005, Oskar Lafontaine’s Left Party created a surprise by obtaining more than 8 per cent of the vote. This result was an advance compared with Germany’s far left result at the 2002 federal election. In sum, class voting might still matter, even if it is of a kind quite removed from class voting as it used to prevail in Western Europe in the 1950s or 1960s. Although the support of the working class for the political left in the past was never complete (Weakliem & Heath 1999), manual workers represented at the time the primary source of votes for most social-democratic or communist parties in Europe. The findings of this study point to a very different picture. Not only does party attachment play little role in Trotskyist voting in France, but the social class the most likely to support the Trotskyist parties in 2002 was not the manual working class, but the lower end of the service sector. In turn, it was argued that the reasons for this support could be understood in relation to the changes that have affected this social class in terms of labour market conditions, employment and living standards in the past decades, whereby it was progressively transformed into a ‘service proletariat’. Overall, the findings of this study confirm the need to build new models of class voting suited to a postindustrial context. When constructing such models, particular attention should be given to the trajectories of different social classes over time in terms of changing life chances. Acknowledgements The author would like to thank Anne Cornilleau, Jonathan Gershuny, Maria Grasso, George Hoare, Shreya Sarawgi, Marlon Seror, Dan Sperber, Vincent Tiberj and Radu Umbres, as well as three anonymous reviewers, for their useful comments and advice. 384 nathan sperber © 2009 The Author(s) Journal compilation © 2009 (European Consortium for Political Research) Appendix Cultural liberalism (8 questions, Cronbach’s alpha 0.79): ‘Is the respect of tradition important to you?’ (Not very important; Somewhat important; Very important; Extremely important) ‘With which of these two opinions on schools do you agree the most?’ (They should provide a sense of discipline and effort;They should allow pupils to develop a critical mind) ‘Do you agree that there are too many immigrants in France?’(Strongly agree; Rather agree; Do not agree; Do not agree at all) ‘Is the presence of immigrants in France a source of cultural enrichment?’ (Strongly agree; Rather agree; Do not agree; Do not agree at all) ‘Are you afraid that France might lose its national identity and culture because of European integration?’ (Yes, I feel afraid; No, I don’t feel afraid) ‘Are you afraid that European integration might lead to a new influx of immigrants?’ (Yes, I feel afraid; No, I don’t feel afraid) ‘Do you agree that the death penalty should be reintroduced?’ (Strongly agree; Rather agree; Do not agree; Do not agree at all) ‘Do you agree that in order to fight delinquency, when a minor commits a crime his family should receive less social benefits?’ (Strongly agree; Rather agree; Do not agree; Do not agree at all) Political distrust (7 questions, Cronbach’s alpha 0.74): ‘How do you consider democracy works in France?’ (Very well; Quite well; Not very well; Not well at all) ‘Do you think that politicians are more often honest or corrupted?’ (Honest; Corrupted) ‘Do you think that politicians are in general preoccupied with what people like you think?’ (Yes, a lot; Quite a lot; A little; Not at all) ‘Do you feel political parties can be trusted?’ (I rather do; I rather don’t) ‘Do you feel Parliament can be trusted?’ (I rather do; I rather don’t) ‘Do you feel the judiciary can be trusted?’ (I rather do; I rather don’t) ‘Do you feel the state can be trusted’ (I rather do; I rather don’t) Political disengagement (4 questions, Cronbach’s alpha 0.75): ‘Do you take interest in politics?’ (Yes,a lot;Yes,quite a lot;Yes,a little;No,not at all) ‘How often do you talk about politics with your family?’ (Often; From time to time; Rarely; Never) three million trotskyists? 385 © 2009 The Author(s) Journal compilation © 2009 (European Consortium for Political Research) ‘How often do you talk about politics with your friends?’ (Often; From time to time; Rarely; Never) ‘Do you feel close to a political party today?’ (Yes, very close; Yes, somewhat close; Yes, a little; No, not at all) Table A1. Socio-demographic and attitudinal predictors of LO and LCR voting in 2002 (multivariate logistic regressions) LO support (coefficient) LCR support (coefficient) Female 0.36** -0.19 0.30 0.03 Age Born before 1940 (reference) Born between 1940 and 1960 1.39*** 1.23*** 1.51*** 1.25*** Born after 1960 1.48*** 1.02*** 2.04*** 1.75*** Social class Salariat (I+II) (reference) Routine non-manual (III) 0.49** 0.42* 0.38 0.57** Self-employed and independents (IV) -0.72 -0.48 -0.03 0.54 Foremen and technicians (V) 0.37 0.33 -0.32 0.11 Manual working-class (V+VI) 0.50* 0.15 0.32 0.33 Wife (III) husband (I+II) dummy variable -0.70 -0.49 -0.92 Income -0.13** -0.02 -0.04 -0.04 Public sector 0.32* 0.22 0.40** 0.26 Non-standard work contract 0.38 0.05 Home owner 0.03 -0.02 Higher education -0.08 -014 Married 0.28 -0.28 Church attendee -0.15 -0.36 Economic inequality 0.36*** 0.33*** 0.13 0.15 Control over firms 0.42*** 0.32* 0.89*** 0.76*** Ban on layoffs 0.17** 0.14 0.18** 0.27*** Cultural liberalism 0.04** 0.05*** 0.12*** 0.10*** Political distrust 0.07*** 0.07*** 0.11*** 0.08*** Political disengagement 0.14*** 0.11*** 0.07* 0.01 Political knowledge -0.17** -0.22*** -0.12 -0.12 Constant -3.88*** -6.24*** -6.71*** -4.49*** -7.26*** -7.42*** Pseudo R2 0.057 0.077 0.104 0.067 0.090 0.127 N 2,651 2,815 2,397 2,651 2,815 2,397 Notes: * p <0.1; ** p < 0.05; *** p < 0.01. 386 nathan sperber © 2009 The Author(s) Journal compilation © 2009 (European Consortium for Political Research) TableA2.Socio-demographicandattitudinalpredictorsofSocialistParty,CommunistPartyandGreenPartyvotingin2002(multivariate logisticregressions) SocialistPartysupport (coefficient) CommunistPartysupport (coefficient) GreenPartysupport (coefficient) Female0.15-0.49***-0.43**0.14-0.050.07 Age Bornbefore1940(reference) Bornbetween1940and1960-0.48***1.60***1.16***-0.61***-0.09-0.06 Bornafter1960-0.81***2.14***1.64***-0.88***-0.63**-0.48 Socialclass Salariat(I+II)(reference) Routinenon-manual(III)0.05-0.35-0.250.22-0.12-0.01 Self-employedandindependents(IV)-0.35-1.07*-0.92-0.04-1.14-0.82 Foremenandtechnicians(V)0.38*-0.13-0.100.55**-0.10-0.11 Manualworking-class(V+VI)0.23-0.67**-0.390.34*0.450.74* Wife(III)husband(I+II)dummyvariable-0.52-1.41-1.31-0.58-0.90-0.36 Income0.03-0.08-0.13*-0.020.01-0.11 Publicsector0.51***0.06-0.070.37***0.35-0.08 Non-standardworkcontract0.070.55*0.40-0.010.670.58 Homeowner-0.130.35*0.57***-0.150.030.14 Highereducation0.140.63***0.220.08-0.91***-1.40*** Married-0.17-0.43**0.19-0.050.040.19 Churchattendee-0.81***-1.56***-1.36**-0.70***-1.87***-1.51** Economicinequality0.24***0.16*0.160.20***0.34**0.39** Controloverfirms0.63***0.08-0.110.64***1.47***1.60*** Banonlayoffs0.080.020.000.040.36***0.22* Culturalliberalism0.06***0.22***0.20***0.08***0.06**0.11*** Politicaldistrust-0.06***0.07***0.05**-0.05***-0.01-0.01 Politicaldisengagement-0.06***0.030.02-0.06**-0.18***-0.21*** Politicalknowledge-0.08*-0.010.01-0.09*-0.03-0.07 Constant-1.16***-2.47***-3.62***-6.02***-5.67***-2.12***-2.97***-6.03***-5.99*** PseudoR2 0.0310.0680.1140.1401.1930.0980.0580.1510.235 N2,6512,8152,6512,8152,3712,3712,6512,8152,371 Notes:*p<0.1;**p<0.05;***p<0.01. three million trotskyists? 387 © 2009 The Author(s) Journal compilation © 2009 (European Consortium for Political Research) Notes 1. The Trotskyist parties also started, in the 1970s, to present candidates and lists at other elections – most notably at parliamentary, municipal and European elections (Pina 2007). Although they never fared well in parliamentary races, disadvantaged by the strict plurality system at the district level, the LO lists and the LCR lists each obtained more than 4 per cent of the vote in the municipalities in which they were present at the 2001 municipal election.At the 1999 and 2004 elections for the European Parliament, LO and the LCR presented joint lists, thus overlooking for the time of two campaigns their longstanding animosity.The joint list made 5.2 per cent in 1999, and France was therefore the only country to send deputies of the extreme left to the European parliament after that election (Knapp 2004). However, the list did not reach the 5 per cent threshold to send deputies in 2004. 2. The Alford index is simply the percentage of left voting within the manual working class minus the percentage of left voting in the rest of the population. If non-manual labour as a proportion of the workforce declines, the Alford index will decrease even if the strength of the relationship between class position and voting behaviour at the individual level remains constant. 3. Kitschelt (1994) claims to reject class analysis as a whole, but his critique of the concept of class is almost exclusively concerned with debunking the property rights approach.This is rather misleading since not all class analysis hinges on such an approach (this study is yet another example). To consider, on the other hand, that Erikson and Goldthorpe (1992) are not studying ‘class’ but something else would be to squabble over semantics and little more. 4. A recent study concludes that even when using the Alford index class voting can hardly be said to have disappeared in Western Europe (Oskarson 2005). 5. According to data from the French Ministry of Work and Employment (cited in Groux 2007), from 1947 to 1958 the average amount per year of days of work lost to striking activity in France was six million. Throughout the 1990s, the average amount of days lost was 500,000 per year, although this figure is obtained by not taking into account the year 1995 (in December 1995, France experienced a month of social unrest and strikes affected many sectors of activity). 6. The PEF 2002 was conducted using quota sampling, a practise solidly established in France (to the knowledge of the author, no large-scale post-election survey has ever been conducted in this country using probability sampling). A multi-stage survey protocol was followed, whereby a number of geographical units were selected by means of probability sampling, after which specific quotas were filled in by interviewers in respondent selection. The quota controls used were age, gender and occupation of the head of household. Statistical modelling is much less reliable when the data has been collected through quota sampling as opposed to probability sampling (Gschwend 2005). Two strategies were carried out to bolster the robustness of the results. First, as advised in Gschwend (2005), the marginal distributions obtained in the PEF for some socio-demographic variables that were not used as controls were checked against census data from 1999 and data derived from large-scale surveys conducted around 2002 by the INSEE (National Institute for Statistics and Economic Studies). Second, each model presented was run a second time applying the socio-demographic weights supplied with the PEF data. In none of the models did the use of weights affect the results. 388 nathan sperber © 2009 The Author(s) Journal compilation © 2009 (European Consortium for Political Research) 7. The PEF asks respondents about their monthly household income, including social transfers.The nine increments are the following: less than 3,000 francs, between 3,001 and 5,000 francs, between 5,001 and 7,500 francs, between 7,501 and 10,000 francs, between 10,001 and 15,000 francs, between 15,001 and 20,000 francs, between 20,001 and 30,000 francs, between 30,001 and 50,000 francs, and over 50,001 francs. 8. In none of the models does the routine non-manual class lose its significance as a positive predictor of Trotskyist voting when this dummy variable is removed. References Bartolini, S. & Mair, P. (1990). Identity, competition and electoral availability:The stabilisation of European electorates, 1885–1985. Cambridge: Cambridge University Press. Beaud, S. & Pialoux, M. (1999). Retour sur la condition ouvrière: Enquête aux usines Peugeot de Sochaux-Montbéliard. Paris: Fayard. Bensaïd, D. (2002). Les trotskistes. Paris: Presses Universitaires de France. Berglund, F. et al. (2005). Party identification and party choice. In J. Thomassen (ed.), The European voter:A comparative study of modern democracies. Oxford: Oxford University Press. Boltanski, L. & Chiapello, E. (1999). Le nouvel esprit du capitalisme. Paris: Gallimard. Bourdieu, P. (1977). Une classe objet. Actes de la Recherche en Sciences Sociales 17–18: 2. Campbell, A. et al. (1960). The American voter. New York: Wiley. Campinchi, P.-J. (2001). Les lambertistes: Un courant trotskiste français. Paris: Broché. Castells, M. (1996). The information age: Economy, society and culture. Vol. 1 The rise of the network society. Oxford: Blackwell. Cautrès, B. (2004). Old wine in new bottles? New wine in old bottles? Class, religion and vote in the French electorate:The 2002 elections in time perspective. In M. Lewis-Beck (ed.), The French voter before and after the 2002 elections. New York: Palgrave. Chauvel, L. (1998). Le destin des générations: Structure sociale et cohortes en France au XXe siècle. Paris: Presses Universitaires de France. Chenu, A. (1990). L’archipel des employés. Paris: Institut National de Statistique et d’Etudes Economiques. Chenu, A. (2005). Sociologie des employés. Paris: La Découverte. Clark, T.N. & Lipset, S.M. (1991). Are social classes dying? International Sociology 6(3): 397–410. Clark, T.N. & Lipset, S.M. (1993). The declining political significance of social class. International Sociology 8(3): 293–316. Cohn-Bendit, D. & Cohn-Bendit, G. (2002). Arlette n’est pas une sainte. Libération, 4 April. Courtois, G. (2002). La campagne floue. Le Monde, 23 March. Courtois, S. & Lazar, M. (2000). Histoire du PCF. Paris: Presses Universitaires de France. Crozier, M. (1965). Le monde des employés de bureau. Paris: Le Seuil. Dalton, R.J. (1988). Citizen politics in Western democracies: Public opinion and political parties in the United States, Great Britain,West Germany and France. Chatham: Chatham House. Dalton, R.J. (1999). Political support in advanced industrial democracies. In P. Norris (ed.), Critical citizens: Global support for democratic governance. Oxford: Oxford University Press. three million trotskyists? 389 © 2009 The Author(s) Journal compilation © 2009 (European Consortium for Political Research) Dalton, R.J. (2000). The decline of party identification. In R.J. Dalton & M. Wattenberg (eds), Parties without partisans: Political change in advanced industrial democracies. Oxford: Oxford University Press. De Graaf, N.D. & Nieuwbeerta, P. (1999). Traditional class voting in twenty postwar societies. In G. Evans (ed.), The end of class politics? Class voting in comparative context. Oxford: Oxford University Press. De Graaf, N.D. et al. (2001). Declining cleavages and political choices:The interplay of social and political factors in the Netherlands. Electoral Studies 20(1): 1–15. Erikson, R. & Goldthorpe, J.H. (1992). The constant flux: A study of class mobility in industrial societies. Oxford: Clarendon Press. Esping-Andersen, G. (ed.) (1993). Changing classes: Stratification and mobility in postindustrial societies. Beverly Hills, CA: Sage. Evans, G. (1992). Testing the validity of the Goldthorpe class schema. European Sociological Review 8(3): 211–232. Evans, G. (ed.) (1999). The end of class politics? Class voting in comparative context. Oxford: Oxford University Press. Evans, G. et al. (1999). Class: Labour as a catch-all party? In G. Evans & P. Norris (eds), Critical elections: British parties and voters in long-term perspective. London: Sage. Fleury, C. & Lewis-Beck, M. (1993). Anchoring the French voter: Ideology vs. party. Journal of Politics 55(4): 1100–1109. Franklin, M. et al. (1992). Electoral change: Responses to evolving social and attitudinal structures in Western countries. Cambridge: Cambridge University Press. Gershuny, J. (2005). What do we do in a post-industrial society? The nature of work and leisure in the 21st century. Working paper, Institute for Social and Economic Research, University of Essex. Groux, G. (1998). Vers un renouveau du conflit social? Paris: Bayard. Groux, G. (2007). L’extrême gauche politique face à la question sociale: Hégémonie ou marginalisation? In D. Reynié (ed.), L’extrême gauche: Moribonde ou renaissante? Paris: Presses Universitaires de France. Grunberg, G. & Schweisguth, E. (2003). La tripartition de l’espace politique. In P. Perrineau & C. Ysmal (eds), Le vote de tous les refus: Les élections présidentielles et législatives de 2002. Paris: Presses de Sciences Po. Gschwend, T. (2005). Analyzing quota sampling and the peer-review process. French Politics 3(1): 88–91. Hannoun, M. & Guerrier, G. (1996). Le partenariat industriel. Paris: Service des Etudes et Statistiques Industrielles. Heath, A. et al. (1985). How Britain votes. Oxford: Pergamon Press. Heath, A. et al. (1987). Trendless fluctuation: A reply to Crewe. Political Studies 35: 256– 277. Heath, A. et al. (1994). The measurement of core beliefs and values: The development of balanced socialist/laissez-faire and libertarian/authoritarian scales. British Journal of Political Science 24(1): 115–132. Inglehart,R. (1977).The silent revolution:Changing values and political styles amongWestern publics. Princeton, NJ: Princeton University Press. Inglehart, R. (1990). Culture shift in advanced industrial society. Princeton, NJ: Princeton University Press. Inglehart, R. (1997). Modernization and postmodernization: Cultural, economic and political change in 43 societies. Princeton, NJ: Princeton University Press. 390 nathan sperber © 2009 The Author(s) Journal compilation © 2009 (European Consortium for Political Research) Ivarsflaten, E. (2006). The vulnerable populist right parties: No economic realignment fuelling their electoral success. European Journal of Political Research 44(3): 465–492. Johsua, F. (2007a). Les conditions de (re)production de la LCR: l’approche par les trajectoires militantes. In F. Haegel (ed.), Partis politiques et système partisan en France. Paris: Presses de Sciences Po. Johsua, F. (2007b). S’engager, se désengager, se réengager: les trajectoires militantes à la LCR. ContreTemps. Kitschelt, H. (1988). Left-libertarian parties: Explaining innovation in competitive party systems. World Politics 40(2): 194–234. Kitschelt, H. (1994). The transformation of European social democracy. Cambridge: Cambridge University Press. Kitschelt, H. & McGann, A.J. (1995). The radical right in Western Europe: A comparative analysis. Ann Arbor, MI: University of Michigan Press. Knapp, A. (2004). Parties and the party system in France: A disconnected democracy? Basingstoke: Palgrave Macmillan. Labée, D. (1996). Syndicats et syndiqués en France. Paris: L’Harmattan. Lavabre, M.-C. & Platone, F. (2003). Que reste-t-il du PCF? Paris: Autrement. Le Monde (2008a). Les militants du NPA de Besancenot s’invitent dans le jeu syndical. Le Monde, 8 October. Le Monde (2008b). Olivier Besancenot a gagné en crédibilité auprès des sympathisants de gauche. Le Monde, 31 October. Le Monde (2009). La LCR se dissout pour donner naissance au NPA. Le Monde, 4 February. Lévy, B.-H. (2007). Ce grand cadavre à la renverse. Paris: Grasset. Lipset, S.M. & Rokkan, S. (1967). Cleavage structures, party systems and voter alignments: An introduction. In Party systems and voter alignments: Cross-national perspectives. New York: Free Press. Manza, J. et al. (1995). Class voting in capitalist democracies since World War II: Dealignment, realignment or trendless fluctuation? Annual Review of Sociology 21: 137–162. Marchand, O. & Thélot, C. (1997). Le travail en France, 1800–2000. Paris: Nathan. Mayer, N. (2002). Les dimensions de la confiance. In G. Grunberg et al. (eds), La démocratie à l’épreuve:Une nouvelle approche de l’opinion des Français.Paris:Presses de Sciences Po. Mayer, N. & Perrineau, P. (1992). Why do they vote Le Pen? European Journal of Political Research 22(1): 123–141. Michelat, G. & Simon, M. (1977). Religion, class and politics. Comparative Politics 10(1): 159–186. Michelat, G. & Simon, M. (2004). Les ouvriers et la politique: Permanences, ruptures, réalignements, 1962–2002. Paris: Presses de Sciences Po. Michelat, G. & Tiberj, V. (2007). Gauche, centre, droite et vote. Permanence et mutation d’une opposition. Revue Française de Science Politique 57(3): 369–392. Norris, P. (ed.) (1999). Critical citizens: Global support for democratic governance. Oxford: Oxford University Press. Oskarson, M. (2005). Social structure and party choice. In J. Thomassen (ed.), The European voter: A comparative study of modern democracies. Oxford: Oxford University Press. Perrineau, P. (2003). La surprise lepéniste et sa suite législative. In P. Perrineau & C. Ysmal (eds), Le vote de tous les refus: Les élections présidentielles et législatives de 2002. Paris: Presses de Sciences Po. Perrineau, P. & Ysmal, C. (eds) (2003). Le vote de tous les refus: Les élections présidentielles et législatives de 2002. Paris: Presses de Sciences Po. three million trotskyists? 391 © 2009 The Author(s) Journal compilation © 2009 (European Consortium for Political Research) Pina, C. (2001). L’extrême gauche en France: Entre permanence et évolutions. In P. Bréchon (ed.), Les partis politiques français. Paris: La Documentation Française. Pina, C. (2007). L’extrême gauche dans les élections présidentielles. In D. Reynié (ed), L’extrême gauche: Moribonde ou renaissante? Paris: Presses Universitaires de France. Piore, M. & Sabel, C. (1984). The second industrial divide: Possibilities for prosperity. New York: Basic Books. Raynaud, P. (2006). L’extrême gauche plurielle. Paris: Autrement. Rey, H. (2004). La gauche et les classes populaires: Histoire et actualité d’une mésentente. Paris: La Découverte. Reynié, D. (2005). Le vertige social-nationaliste: La gauche du Non et le référundum de 2005. Paris: Editions de la Table Ronde. Reynié, D. (2007). La gauche pure. In L’extrême gauche: Moribonde ou renaissante? Paris: Presses Universitaires de France. Rose, R. & McAllister, I. (1986). Voters begin to choose: From closed-class to open elections in Britain. London: Sage. Schmitt, H. & Holmberg, S. (1995). Political parties in decline? In H.-D. Klingemann & D. Fuchs (eds), Citizens and the state. Oxford: Oxford University Press. Schweisguth, E. (1983). Les salariés moyens sont-ils de petits bourgeois? Revue Française de Sociologie 24(4): 679–704. Schweisguth, E. (2002). La dépolitisation en question. In G. Grunberg et al. (eds), La démocratie à l’épreuve: Une nouvelle approche de l’opinion des Français. Paris: Presses de Sciences Po. Sineau, M. (2004). Les paradoxes du ‘gender gap’ à la française. In B. Cautrès (ed.), Le nouveau désordre électoral: Les leçons du 21 avril. Paris: Presses de Sciences Po. Taguieff, P.-A. (2002). La nouvelle judéophobie. Paris: Mille et Une Nuits. Tiberj, V. (2007). Les votes trotskistes: Votes extrêmes ou votes de gauche? In D. Reynié (ed.), L’extrême gauche: Moribonde ou renaissante? Paris: Presses Universitaires de France. Van der Eijk, C. et al. (2005). Left-right orientations and party choice. In J. Thomassen (ed.), The European voter: A comparative study of modern democracies. Oxford: Oxford University Press. Weakliem, D. & Heath, A. (1999). The secret life of class voting: Britain, France and the United States since the 1930s. In G. Evans (ed.), The end of class politics? Class voting in comparative context. Oxford: Oxford University Press. Address for correspondence: Nathan Sperber, Nuffield College, 1 New Road, Oxford OX1 1NF, UK. E-mail: nathan.sperber@gmail.com 392 nathan sperber © 2009 The Author(s) Journal compilation © 2009 (European Consortium for Political Research) Copyright of European Journal of Political Research is the property of Wiley-Blackwell and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.