France: an example of changing party systems in Western Europe
1The emergence and expansion of both green (environmental) parties and extreme right-wing, anti-immigration parties since the end of the 1960s has profoundly restructured national political party systems in Western Europe. Regardless of whether they are analysed as a product of the cultural changes brought about by the progress of post-materialist values,  or as a product of the transformations engendered by globalisation and the post-industrial revolution,  these new parties have helped to redefine the main axis of partisan conflict. Although traditionally governed by class divisions and religious cleavages, the left/right opposition has been refashioned by the emergence of a new fault line between a culturally liberal left and an authoritarian and ethnocentric right. 
2In this context, since the 1980s France has proven to be an exemplary illustration of the general reconfiguration of a party system due to the rise of a new player: the radical right.  The eruption on the political scene of issues related to immigration and financial insecurity has transformed the cultural dimension of political competition, hitherto essentially shaped by religious cleavages, and has resulted in the creation of a new party system characterised by a tripartition between the left, the moderate right and the radical right.
The right: driving changes in the French party system since 2007
3Despite reinforcing its existence during the presidencies of François Mitterrand and Jacques Chirac, the tripartite nature of the French party system was abruptly called into question during 2007’s presidential and legislative elections, following the collapse of the Front National. Resulting from Nicolas Sarkozy’s strategic attempts to win over Le Pen’s electorate, a large electoral bloc was constituted that rallied the majority of both moderate and radical right-wing voters around a single candidate and party; this broke with over two decades of the Front National’s growth and consolidation, while simultaneously opening the door to the hitherto unseen empowerment of centrists led by François Bayrou.
4Beyond those analyses underlining the success of Sarkozy’s endeavour,  which allowed those in government to maintain power and avoid a repetition of the protest vote which had dominated 2004’s mid-term elections,  two main interpretations have been proposed to explain the right’s electoral and partisan evolution in France. The first interpretation is in line with Grunberg and Haegel’s hypothesis that the French political system leans towards a bipartite division,  thus suggesting that 2007’s elections strengthened this tendency thanks to the UMP’s increasing hold over the right as a whole.  The second falls under the theory of realignment  and argues that the 2007 elections provoked the collapse of the electoral system in place since 1984’s European elections,  with the right as the epicentre of the global reconfiguration of the structures of political competition.
5Since 2007’s elections, the remarkable attraction wielded by Nicolas Sarkozy and the UMP over the supporters of the radical right has progressively dwindled, due to the intensification of the financial and economic crisis and the growing unpopularity of those in power. Consequently, the recovery of the Front National, evidenced by Marine Le Pen’s high scores in 2012’s first round of the presidential elections, invites us to examine the type of reconfigurations occurring at the heart of the right in France. More specifically, the results of 2012’s elections suggest the possible reinstatement of the tripartition that existed prior to 2007, a structure that simultaneously revealed the presence of a strong but radical and isolated FN, and the unique character of the radical right’s electorate, highly polarised on immigration issues and largely uninterested in voting for the moderate right.
6The goal of this article is to demonstrate that the FN’s return to power did not coincide with the reinstatement of the electoral fault line between the moderate right and the radical right that had existed prior to 2007. In opposition to theories seeing a “return” to the old system, we suggest that the UMP’s ideological radicalisation and Nicolas Sarkozy’s presidential campaign in 2012 – even more focused on immigration and national identity than in 2007 – strengthened the process which was bringing together the UMP and the FN’s voting blocs. Consequently, far from reinvigorating the tripartition, 2012’s springtime elections have perhaps prolonged the phase of realignment and electoral confusion inaugurated in 2007: a phase whose still uncertain outcome could well involve a sort of “cultural war” waged by a right-wing bloc radicalised and united by its fear of Islam and immigration, its attachment to national identity and its belief in repressive action against delinquency.
Data and methods
7From an empirical point of view, our argument relies on the combined use of actual election results and data obtained via polls and surveys. The evidence thus collected indicates the existence of pluralism. The aggregate election results come from the Ministry of the Interior’s files for the years after 1993, and from the former Banque des données sociopolitiques in Grenoble (Bdsp – Socio-political Database) for the period prior to 1993. Among these files, some were substantially reshaped in order to better reflect the electoral offering; for specific references concerning the data used and a discussion of the main methodological choices made, please see Annex 1. With regard to polling data, our analysis relied on the post-electoral surveys conducted by university establishments during each presidential election cycle since 1988: the Cevipof polls for the period between 1988 and 2007, and the French electoral survey conducted by N. Sauger and the TriÉlec network for the year 2012. For a complete description of this data, please see Annex 2.
8Our argument is structured into four parts. The first part presents the origins, basis and main characteristics of tripartition in French political life. The second part focuses on changes in power relations within the right-wing political world and analyses evolutions in the strength of various political forces from 2007 to 2012. The third section studies the transformation of right-wing and centrist electoral geography and links these structural changes to the macro-geography of immigration. Finally, the fourth section examines the added level of competition between the moderate and radical right vis-à-vis a voting public that is increasingly hostile towards immigration, thus showing the growing similarity between the voting rationales of UMP and FN supporters.
Tripartition in French political life
An ideological dimension: voters’ clashing values
9The concept of tripartition is a recent one, and refers to the lasting implantation of the Front National within the French political party system. Coined by Grunberg and Schweisguth in the middle of the 1990s to describe a new political landscape, this term originally referred to the ideological structuring of partisan voters, illustrating the existence of a clear-cut opposition of values between moderate right-wing voters and radical right-wing voters, in particular on matters of authority and immigration. Within this framework, tripartition is essentially the product of the Front National’s electorate’s inherently different nature. While moderate right-wing voters are primarily characterised by a high level of support for economic liberalism and the defence of religious morality, radical right-wing voters are instead distinguished by their massive rejection of universal values and their attachment to xenophobic and authoritarian principles. Accordingly, the FN’s electorate cannot be considered as merely being the right-most fringe of the right, which would be more economically liberal and more conservative in terms of values than moderate right-wing parties.  The vigour of Front National voters’ anti-universalism and their relatively moderate position with regard to economic liberalism and religious issues create an entirely new axis in the political landscape, differentiated from both the left and the moderate right.
10This ideological fault line between moderate and radical right-wing voters does not exist on the left. Although it is true that extreme left-wing parties possess some of their own particular characteristics – in particular their propensity for protesting how the political system functions – their voters are not fundamentally different from the electorate of other parties on the left. Much like the governmental left, Trotskyist parties can first and foremost be characterised by their stance towards economic liberalism.  In other words, the radical left only differs by degree, which thus excludes the possibility of a quadripartite political landscape in France. 
An electoral and partisan dimension: the realignment perspective
11The concept of tripartition cannot be reduced to the ideological dimension alone. In addition to how voters’ values are structured, it can also be used to describe the partisan system and the electoral power relations which were crystallised following the realignment phase of 1981-1984,  with the consolidation of three different axes: a left-wing axis dominated by the PS (Parti socialiste), a moderate right-wing axis led by the RPR (Rassemblement pour la République, which became the UMP (Union pour un mouvement populaire) from 2002) and an anti-system extreme right-wing axis embodied by the Front National. Our analysis thus stems from this perspective, associated with the theory of realignment but likewise perfectly compatible with the studies conducted by Grunberg and Schweisguth.
12Functioning as a pillar of the electoral order until 2007’s elections, France’s political tripartition firstly illustrates the party system’s configuration and the isolation of the FN on the far right. Despite its substantial electoral results, the FN was ostracised by all the other political parties, in particular by moderate right-wing parties. This exclusion was expressed during every election,  whether voting was uninominal in two rounds or proportionally representative. The Front National never brokers alliances with other parties, never receives support from another party when it makes it to the second round (with the exception of extreme right-wing parties, such as the MNR (Mouvement national républicain) in 2002), and most of the time refuses to choose between left-wing and moderate right-wing parties if it is eliminated during the first round. In a symmetrical manner, the left and the right have increasingly worked together to oppose the spread of the far right, following a more or less explicit strategy termed the “Republican front”.  In the end, the opposition between the radical right and other parties seems to be as structurally important as the opposition between the left and the right. Consequently, Jacques Chirac’s victory against Jean-Marie Le Pen in 2002’s presidential election – aided by the support of a majority of the left – can be seen as a natural outcome of France’s tripartite political system.
13Moreover, this tripartition has a specifically electoral dimension, which is embodied by the split between moderate right-wing voters and extreme right-wing voters, and evidenced by the inability of the UDF, the RPR, and in turn the UMP, to attract a significant percentage of Front National voters to their camp. This division on the right engenders different voting blocs and has led to the empowerment of the Front National, which gained strength throughout the 1990s by making strong inroads in working-class environments,  as well as by promoting a growing rejection of parties “in the system”. This electoral aspect of the tripartition has mainly affected the first rounds of elections, but is also sometimes expressed in one-on-one conflicts between the left and the moderate right. In the latter case, the situation usually involves the wide dispersal of FN voters, a large number of whom prefer to abstain, cast a blank ballot or even vote for the left rather than vote for the moderate right (as was the case during the second round of 1995’s presidential election). 
Tripartition and level shifts in right-wing and centrist parties
2007’s elections and the collapse of the Front National
14The current reconfiguration of right-wing and centrist parties in fact began with the elections that took place in the spring of 2007.  Significant level shifts occurred among the various political players during the first round of the presidential election: winning only 10.7% of the vote in metropolitan France, Jean-Marie Le Pen suffered an important drop (- 8.9 points) compared to the far right’s composite score in 2002 (19.6%). This drop benefited Nicolas Sarkozy, who obtained 31% of the vote, thus performing almost as well as the moderate right as a whole in 2002 (31.5%), and this despite the intrusion of François Bayrou, who once again ran for president in 2007 but this time on a new centrist platform, and who scored 18.8% in 2007 against 6.9% in 2002.
15During the first round of legislative elections, this shift of FN voters towards the governmental right only increased. Front National candidates faltered and failed, falling for the first time since 1984’s European elections below the 5% threshold, whereas UMP candidates, Nouveau Centre (NC) candidates and various other parties on the right experienced a significant boost, obtaining a total 44.5% of the vote. Likewise, the MoDem, which replaced the UDF in order to better represent the new independent, centrist vision proposed by François Bayrou, suffered from the high degree of bipolarisation within the election (for or against the new government) and had to content itself with only three members of parliament. With 7.7% of the vote,  it nevertheless managed to establish itself as the country’s third electoral power.
Crossovers: mid-term elections during Sarkozy’s five-year tenure
16These shifts occurring within the balance of power in the right-wing and centrist camps were followed by new developments brought about by mid-term elections during Sarkozy’s presidency. From this perspective, three dynamics are particularly important: 1) the Front National’s recovery; 2) the overall weakness of the right; and 3) the flagging performance of the MoDem, which largely helped the left.
17The resurgence of the radical right began with 2009’s European elections, with a change index of 0.70 compared to 2004’s European elections and an odds ratio of 0.68 (see Table 1). As a whole, the radical right was still behind compared to the results that it obtained prior to 2007, but proportionally much less so. This recovery continued apace during 2010’s regional elections (change index of 0.77 compared to 2004 and an odds ratio of 0.74) and especially during 2011’s cantonal elections (change index of 1.23 compared to 2001 and an odds ratio of 1.27), when the far right surpassed – for the first time since 2007 – the score that it had obtained in the preceding electoral contest of the same type. The percentage obtained by the FN in 2011’s cantonal elections (15.6%) – despite the tendency of cantonal elections to handicap parties lacking local roots – was even more significant, as it was garnered with only 1,441 candidates (compared to 1,850 in 2004).
The Front National’s recovery since 2007
The Front National’s recovery since 2007N.B. The results are expressed in percentage of votes cast. Each election contest was automatically compared to the preceding election of the same type: 2007’s presidential election with 2002’s presidential election, 2007’s legislative elections with 2002’s legislative elections, 2008’s cantonal elections with 2001’s cantonal elections, 2009’s European elections with 2004’s European elections, 2010’s regional elections with 2004’s regional elections, 2011’s cantonal elections with 2004’s cantonal elections, 2012’s presidential election with 2007’s presidential election and 2012’s legislative elections with 2007’s legislative elections. The change indices were calculated by simple division; the odds ratio was obtained by dividing between the probability ratio of voting for the FN.
18The faltering of the right, including the FN, represents another remarkable development: as a whole, the right obtained 41.9% of the vote in 2009’s European elections and only 40.1% in the first round of 2010’s regional elections. The right had never been so weak, across the board, during the history of the Fifth Republic. This general weakness was also observed, albeit to a lesser degree, in 2008’s cantonal elections (46.1% compared to 52% in 2001) and 2011’s cantonal elections (47.5% compared to 49.9% in 2004), where the levels recorded were lower than any witnessed since 1979.  Although this historical drop was linked to the president’s high disapproval ratings at the time, which consequently influenced the UMP’s results, it can more aptly be explained by the emancipation of voters who had previously cast a ballot for the right in 2004’s mid-term elections, despite the government’s high disapproval rating then as well. This was thus a deeper change than a mere protest vote: a gradual development that traces its origins back to the first round of 2007’s presidential election, when the total score for Sarkozy, Villiers and Le Pen was already substantially below 50% (44% compared to 51% for the right as a whole in 2002 and 59.1% in 1995).
19The third important change was the MoDem’s flagging performance. Dipping below the 10% threshold as early as 2007’s legislative elections, the party then stabilised its results until 2009’s European elections (8.4%), continuing its attempt to empower a whole swath of the right-leaning electorate (12% for the UDF in 2004). The 2010 regional elections then signalled the beginning of a pronounced decline, with only 4.3% of the vote. This drop almost exclusively benefited the left, which surpassed the 50% threshold during the first round of a race (53.7%) for the first time since 1981. The MoDem continued to falter during 2011’s cantonal elections: hindered by a very small number of candidates (231 for 1,940 cantons), the party obtained a disastrously low score of 1.2%.
2012’s elections: the return of tripartition?
20The elections held in 2012 confirmed the evolving power relations in the right and centre camps seen previously during mid-term elections: the Front National’s recovery and the MoDem’s failure. By obtaining 18.3% of the vote, Marine Le Pen compensated for her father’s 2007 failure (+ 7.6 points). Nevertheless, she did not match the totals obtained by Le Pen and Mégret in 2002 (19.6%), and the increase in her vote compared to far right candidates in 2011’s cantonal elections was relatively modest in the 1,450 cantons where the latter were present (19.6% compared to 19.3%). Although the level of participation varied significantly from canton to canton in the cantonal elections (43.6%) compared to the presidential election (81.1%), the Front National’s weak dynamic between the two elections indicated that its electoral potential for the presidential contest was probably greater than the score obtained by Marine Le Pen: unlike in the cantonal elections, the FN was not penalised for a lack of local implantation and it was even able to capitalise on the personal charisma of its presidential candidate. 
21With 27% of the vote, Nicolas Sarkozy only lost 4 percentage points (5.1 if we take into account the support for Frédéric Nihous, a candidate in 2007). Despite his lack of popularity, the incumbent president attained a score higher than the total obtained by Chirac, Boutin and Madelin in 2002 (24.6%) and scarcely lower than this total combined with Jean Saint-Josse (28.9%), whereas François Bayrou (9.2%) experienced a significant drop from 2007 (- 9.6 points) but maintained a score higher than the one he obtained in 2002 (6.9%). With 47.2% of the vote, all the parties on the right performed much better than in 2009’s European elections and 2010’s regional elections; they even fared better than in the first round of 2007’s presidential election (+ 3.2 points). Nevertheless, this recovery was still insufficient to reach the 50% threshold, which had previously been crossed in 1988, 1995 and 2002. Consequently, it was primarily the left that seemed to benefit from François Bayrou’s diminished performance (+ 7.3 points), even if, with 43.4% of the vote, it did not obtain scores as high as during mid-term elections.
22The legislative elections that took place immediately afterwards magnified the MoDem’s faltering performance: outside of the context of the presidential election and of François Bayrou as a candidate, independent centrists no longer succeeded in securing a significant portion of the electorate (1.7%). This disappointing score essentially benefited the left (+ 9.1 points compared to 2007), already on a high from its presidential victory, whereas the right as a whole fell behind slightly (- 1.6 points). Within the right’s political landscape, the FN confirmed its recovery, to the detriment of the UMP and its allies: the Rassemblement Bleu Marine obtained 14% of the vote (14.2% for the far right as a whole), compared to 33.9% for the UMP, the NC and the DVD (Divers droite) combined.
Tripartition and structural changes on the right and in the centre
The changing electoral geography of right-wing and centrist parties
23Although the Front National’s return to centre stage and the MoDem’s decline could suggest that the 2007 elections were merely a parenthesis in the development of French political life, the changing electoral geography of right-wing and centrist parties indicates on the contrary that they prompted reconfigurations that are still evolving in 2012.
24Whereas the moderate right-wing electorate remained stable during the first rounds of presidential elections from 1988 to 2002, with correlation coefficients at the departmental level hovering near 0.9 (see Table 2), 2007 heralded a break with the past: the combined Sarkozy-Villiers correlation coefficient fell to 0.42 compared to Chirac-Barre in 1988, 0.44 compared to Chirac-Balladur-Villiers in 1995 and 0.49 compared to Chirac-Bayrou-Madelin-Boutin in 2002 (0.46 compared to the Chirac-Madelin-Boutin total).  The correlation coefficients calculated for legislative districts during 1988 to 2007 show the same changes, albeit less dramatically than for departmental data, including a high level of stability between 1988 and 2002, followed by a break in 2007. 
Changes in the geographical structure of voting for the right in presidential elections
Changes in the geographical structure of voting for the right in presidential electionsN.B. The correlation grids were calculated at the departmental level for metropolitan France (Pearson’s r). For the first round of the presidential election, the moderate right chose Jacques Chirac and Raymond Barre in 1988; Jacques Chirac, Édouard Balladur and Philippe de Villiers in 1995; Jacques Chirac, François Bayrou, Alain Madelin and Christine Boutin in 2002; Nicolas Sarkozy and Philippe de Villiers in 2007; Nicolas Dupont-Aignan and Nicolas Sarkozy in 2012.
25The dramatic transformation of the geographical voting structure in favour of moderate rightwing candidates in 2007 was confirmed during the first round of 2012’s presidential election: the combined geography for Sarkozy and Dupont-Aignan was almost identical to 2007’s Sarkozy-Villiers distribution, with a correlation coefficient of 0.94 at the departmental level (0.95 at the cantonal level).  In other words, despite a drop in levels, the moderate right and Sarkozy both maintained in 2012 the outlines of their electoral coalition, confirming 2007’s break.
26It is significant that this structural transformation of moderate right-wing voting between 2007 to 2012 and 1988 to 2002 can also be observed if we consider François Bayrou as a right-wing candidate in 2007 and 2012. In 2007, as in 2012, the electoral geography of the right-wing and centrist parties was substantially different from that presented by the Chirac-Bayrou-Boutin-Madelin total in 2002 (coefficients of 0.76 and 0.75 at the departmental level) or by 1995’s Chirac-Balladur-Villiers total (coefficients of 0.63 and 0.65) or by 1988’s Chirac-Barre total (coefficients of 0.68 and 0.70). As a result, no longer considering François Bayrou as a rightwing candidate from 2007 is not sufficient to explain the important changes in electoral geography for the moderate right. The transformation of voting patterns for the moderate right cannot be reduced to the empowerment of a segment of the centre right’s electorate: it is first and foremost the fact that the moderate right captured a segment of the far right’s electorate that accounts for the geographical changes in voting patterns in its favour.
27The stability of Nicolas Sarkozy’s electorate between 2007 and 2012 echoes the even greater consistency of the FN’s voting geography during the same period, with a correlation coefficient of 0.97 at the departmental level.  This extremely high coefficient – even though Marine Le Pen took over from her father and the voting levels obtained were different – contrasts with the coefficients registered in 2007 compared to 2002 (0.84) and especially compared to 1995 (0.74) and 1988 (0.56). Although Marine Le Pen’s success erased 2007’s failure, it also confirmed the limited but non-negligible structural changes which occurred in that year, linked to the new competition presented by Nicolas Sarkozy (see Table 3).
Changes in the geographical structure of voting for the far right in presidential elections
Changes in the geographical structure of voting for the far right in presidential electionsN.B. The correlation grids were calculated at the departmental level for metropolitan France (Pearson’s r). For the first round of the presidential election, the far right included Jean-Marie Le Pen in 1988 and 1995; Jean-Marie Le Pen and Bruno Mégret in 2002; Jean-Marie Le Pen in 2007; and Marine Le Pen in 2012.
The influence of immigration’s macro-geography
28These important changes in the geographical structure of voting for the right reflect the moderate right’s shift onto the far right’s territory. This evolution is particularly evident when we compared the electoral shifts on the right with regard to three zones of foreign – specifically North African or Turkish – immigration.  Whereas since 1984’s European elections the moderate right had consistently obtained inferior scores in zone 1 (the zone where the number of immigrants was highest) when compared to zone 3 (where the number of immigrants was lowest), the inverse occurred during 2007’s presidential and legislative elections (see Table 4). The reversal was spectacular: with a positive deviation of more than 4 points between zones 1 and 3, the map of the moderate right was more influenced by the macro-geography of immigration than that of the far right, whose interzone deviation dropped below 3 points. And yet, the far right had until then been alone in having a positive interzone deviation, generally greater than 7 points (see Table 5).
Results for the moderate right by immigration zone (1984 –2012)
Results for the moderate right by immigration zone (1984 –2012)N.B. Each zone is composed of 32 departments, according to the proportion of North-African or Turkish foreigners compared to the population as a whole, as recorded by the 1982 census. Zone 1 has the highest proportion of foreigners; zone 3 has the lowest. The results are expressed in percentage of votes.
Results for the far right by immigration zone (1984 –2012)
Results for the far right by immigration zone (1984 –2012)N.B. Each zone is composed of 32 departments, according to the proportion of North-African or Turkish foreigners compared to the population as a whole, as recorded by the 1982 census. Zone 1 has the highest proportion of foreigners; zone 3 has the lowest. The results are expressed in percentage of votes.
29In the mid-term elections held during Nicolas Sarkozy’s tenure as president, these interzone deviations evolved in opposite directions. The interzone deviation for moderate right-wing voting decreased during 2008’s cantonal elections (+ 1.2), during 2009’s European elections (- 0.7) and during 2010’s regional elections (+ 1.8), before returning to a clearly negative score during 2011’s cantonal elections (- 3.1). On the other hand, the interzone deviation for the far right electorate increased in 2008 (+ 4.3) and then again in 2010 (+ 6.8), before hitting a record high in 2011 (+ 12.4).  Nevertheless, 2007’s structure was reinstated during the first round of 2012’s presidential election, then during the following legislative elections, once again demonstrating a clearly positive interzone deviation for the moderate right (+ 2.7 and + 2.4) and an interzone deviation for the far right that was lower than observed prior to 2007 (+ 3.2 and + 5.9). 
30The convergence between the geographical structures of moderate right-wing and far rightwing voting patterns is not only exhibited during the first round of elections. It can also be observed during the second round of presidential contests: Nicolas Sarkozy’s electoral map during the second rounds of 2007 and 2012’s presidential elections was manifestly more influenced by the macro-geography of immigration than Chirac’s was in 1988 or 1995 (see Table 6). In other words, Sarkozy fared better in harnessing the core Front National voting bloc when pitted against the left than Chirac did. This difference in power of attraction largely explains the weakness of the correlation coefficients between Chirac’s results in 1988 and 1995 and Sarkozy’s in 2007 and 2012 (Table 2). 
Results for the right during the second round of presidential elections by immigration zone
Results for the right during the second round of presidential elections by immigration zoneN.B. Each zone is composed of 32 departments, according to the proportion of North-African or Turkish foreigners compared to the population as a whole, as recorded by the 1982 census. Zone 1 has the highest proportion of foreigners; zone 3 has the lowest. The results are expressed in percentage of votes.
France’s tripartition challenged by the emergence of a new competitive structure on the right
The changing patterns of the vote for the moderate right
31The post-electoral surveys conducted after each presidential election since 1988 allow us to examine even more closely this growing permeability between UMP voters and FN voters. In particular, they allow us to model the impact of the two value axes that most significantly structure the ideological landscape of French voters since 1981 –1984’s realignment period: ethno-authoritarianism and a belief in economic liberalism.  Consequently, we can thus analyse the logic behind the formation of different electorates.
32Establishing variables to express these two value axes presented a certain number of methodological problems. Due to the small number of shared questions between the surveys conducted between 1988 and 2012, it was impossible to have access to robust indicators that were strictly comparable over time. Following research done by Vincent Tiberj,  an alternative measure based on the hypothesis of latent variables was implemented. For each survey, thanks to multiple correspondence analysis, the two axes of ethno-authoritarianism and economic liberalism were constructed.  Models of multinomial regression were accordingly estimated for each first round of a presidential election, predicting votes in favour of each candidate. The principal coefficients of these models are presented in Table 7.
Voting logics: a multinomial model of the impact of values on voting patterns during the first round of presidential elections
Voting logics: a multinomial model of the impact of values on voting patterns during the first round of presidential electionsN.B. The mode of reference was the vote for the RPR candidate, then the UMP candidate: Jacques Chirac in 1988, 1995 and 2002; Nicolas Sarkozy in 2007 and 2012. The significant levels retained were: * < 0.05; ** < 0.01.
33These models confirm that votes cast for the Front National during presidential elections are highly influenced by hostility towards immigration.  From this point of view, the graphical representation of the probabilities of voting for the FN predicted by the models, according to the level of support for economic liberalism or the level of ethno-authoritarianism, is very revealing: votes for Le Pen essentially depended on the cultural axis, almost systematically hitting maximum probability among voters belonging to the most ethno-authoritarian decile, and this regardless of their position on the economic axis (see Figure 1). 
Patterns of voting for UMP and FN candidates
Patterns of voting for UMP and FN candidatesN.B. These graphs illustrate the voting probabilities predicted by the regression models whose coefficients can be found in Table 7. The value variables were converted into deciles.
34The logic behind voting for Chirac in 2002 – but also for Sarkozy in 2007 and 2012 – was more complex, in the sense that there was a close relationship between the economic and cultural dimensions. In general, the probability of voting for the UMP candidate increased with higher positions along the economic liberalism and ethno-authoritarianism axes.
35However, 2012’s presidential election illustrated a significant change: a clear rise in the influence of the cultural axis. Although the difference in voting probability for Chirac for the most ethno-authoritarian voters and the least ethno-authoritarian voters never went above 0.2 in 2002, regardless of voters’ position on the economic liberalism axis, it suddenly reached 0.4 for Sarkozy among the most economically liberal voters; this was the case up to the fifth decile of economic liberalism. This split was confirmed during 2012’s presidential elections: the difference between the probability of voting for Sarkozy among the most and least ethno-authoritarian voters was once again higher than 0.4 in the two most economically liberal deciles. In general, Sarkozy maintained his position on both the economic and cultural axes,  but he fell sharply among voters less hostile to immigration.
Competition between the moderate right and the far right for ethno-authoritarian voters
36This change in voting rationale for Sarkozy compared to Chirac had important consequences for the reconfiguration of French right-wing parties: the UMP candidate was now in more direct competition with the FN candidate among the electorate that was the most hostile to immigration (see Figure 2).  By insisting on the themes of immigration and national identity in 2007, Sarkozy was not merely content with stealing away a portion of Jean-Marie Le Pen’s electorate – he sought to compete with him for his core voting bloc. During the first round of 2007’s presidential race, and for the first time since the emergence of the Front National as a serious electoral threat, a moderate right-wing candidate succeeded in significantly surpassing Jean-Marie Le Pen among voters in the most ethno-authoritarian decile. The probability of voting for Sarkozy was 0.47 among this decile, whereas the probability of voting for Le Pen was 0.25.
Voters’ relationships to ethno-authoritarianism, and the vote for right-wing and centrist candidates
Voters’ relationships to ethno-authoritarianism, and the vote for right-wing and centrist candidatesN.B. These graphs represent the predicted probabilities of voting for the main right-wing and centrist candidates, by decile of ethno-authoritarianism. These probabilities were calculated based on regression models whose coefficients can be found in Table 7. They are shown in the form of cumulated areas in order to facilitate comparison between the structure of the electorates.
37Nicolas Sarkozy was not able to repeat this performance during the first round of 2012’s presidential race against Marine Le Pen; the latter dominated the incumbent president among voters belonging to the three most ethno-authoritarian deciles (see Figure 2), but her campaign confirmed the growing polarisation of her electorate regarding matters of authority and immigration. Like Marine Le Pen, Sarkozy obtained very high scores among the most ethno-authoritarian voters, and very low scores among the least ethno-authoritarian voters, whereas prior to 2007 the vote for the moderate right was significantly less polarised by the cultural dimension. Consequently, Sarkozy profoundly changed the face of the moderate right, which is now experiencing serious difficulties in appealing to voters that are less hostile to immigration.  In the end, the outgoing president continued to work on bridging the gap between voting for the UMP and the FN, and contributed to weakening the electoral split between the moderate right and the far right.
40The tripartition between the left, the moderate right and the far right constituted one of the fundamental elements of the electoral structure that crystallised in France following 1984’s European elections. Intrinsically linked to the Front National’s meteoric rise and its segregation from other political players, this tripartite structure associated electoral, ideological and partisan dimensions and reflected a clear split between the moderate right and the radical right.
41Nicolas Sarkozy’s 2007 presidential campaign radically challenged the principle governing the rules of political competition. His strategy to conquer some of Jean-Marie Le Pen’s voters and his self-avowed radicalisation regarding immigration, national identity and insecurity provoked a rupture in French political life: a breakdown that manifested itself by the dramatic collapse of the Front National and the creation of a large conservative and ethnocentric voting bloc. Despite the FN’s subsequent recovery and the high scores obtained by Marine Le Pen, 2012’s presidential election strengthened the move towards a merger of the UMP and FN electorates. Consequently, this election also confirmed a fundamental element of 2007’s break with tradition: polarisation on immigration, which defined the FN vote, now became an important characteristic of the UMP’s electorate. The presidential race of 2012 likewise confirmed the UMP’s increased capacity to attract traditional FN voters in the second round, if not during the first. In this regard, the tripartite structure has been weakened: the electoral fault line between the moderate right and the far right is no longer as obvious as it was in the past.
42Nevertheless, the merging of UMP and FN voting patterns does not mean that all aspects of tripartition have disappeared. In terms of the party system, the UMP’s national leadership has maintained the boundary between the moderate right and the far right. Although the idea of a “Republican front” waged against the FN has largely been abandoned,  the refusal to form any alliances with the radical right was reiterated during 2012’s legislative elections, and the FN consequently remained isolated on the right. With regard to the ideological dimension, this article has not addresed changes in the values of UMP and FN voters, even if we have suggested that their inherent differences may have been called into question by recent developments.
43It is still too early to assert that the weakening of the electoral fault line between the moderate right and the far right observed in 2007 and 2012 will lead to the emergence of a new electoral landscape devoid of any tripartite structures. Ongoing economic difficulties and the fragility of the European Monetary Union will likely prolong the realignment phase begun in 2007. In such a chaotic context, Marine Le Pen’s dangling of initiatives designed to attract the cultural and economic losers of globalisation (withdrawing from the Euro, protectionism, etc.) could once again strengthen the differences between the FN and the UMP. In any case, the reconfiguration of the French right should be analysed as a crucial element of the redefinition of the nation’s party system, and more generally of the ongoing transformation of political structures in Western Europe. 
Annex 1 - Results of French elections
44This study used results from presidential elections (P), legislative elections (L), regional elections (R), cantonal elections (C) and European elections (E) with different levels of aggregation. The data came from the digital files produced by the Banque de données sociopolitiques (Bdsp – Socio-political Database) in Grenoble for elections prior to 1993, and from the digital files of the Ministry of the Interior for elections later than 1993. These files are outlined in the table below.
45For the results of 2009’s European elections, 2010’s regional elections and 2012’s legislative elections, the labels attributed by the Ministry of the Interior were systematically verified and modified in order to better represent the electoral offer.  These files are available upon demand.
The context of metropolitan France
46These analyses were limited to metropolitan France. Although this choice could be seen as debatable, especially with regard to presidential and legislative elections, it was necessary in order to compare different types of elections, as citizens in overseas departments and abroad cannot vote in every election. In addition, the disconnect between overseas party systems and the metropolitan party system justifies not taking the former into account when investigating changes to the balance of power that structures the national party system. 
Results of national, regional and European elections in metropolitan France (2000 –2012, in %)
Results of national, regional and European elections in metropolitan France (2000 –2012, in %)NB. The results are expressed by percentage of registered voters and ballots cast, and by percentage of votes for the results of each political group. For the first round of the presidential election, Jean-Pierre Chevènement (MRC) and Jean Saint-Josse (CPNT) in 2002, François Bayrou (UDF) and Frédéric Nihous (CPNT) in 2007 and François Bayrou (MoDem) in 2012 are counted in Other due to their rejection of the left/right opposition. Two numbers per cell indicate the results of both rounds.
Annex 2 - French political surveys
47This article draws on the major French post-electoral surveys conducted by university establishments. Despite the absence of a tradition similar to the American National Elections Studies (ANES) or the British Election Studies (BES) in the repetition of indicators and variables, these constitute a rather long series, which covers all the presidential elections since 1988.
48The main characteristics of the surveys used are outlined in the table below. With the exception of the 2012 survey, which utilised random sampling, these surveys were conducted using the quota method. For the French Electoral Panel in 2002, only the post-presidential wave (2) was used.
Establishing the attitude axes
49The limited number of questions repeated throughout the entire survey corpus immediately excluded the possibility of using measures that were strictly comparable over time. To overcome this difficulty, an alternative measurement strategy was implemented. It relied on the notion of latent dimensions: for each poll, we hypothesised that voters could be distinguished according to their belief in economic liberalism and their level of ethno-authoritarianism, and that these two axes could be approached via associated opinion questions asked. 
50The statistical method used here to establish these latent dimensions was specific multiple correspondence analysis (MCA). This technique is perfectly suited to explore the response modalities to opinion questions contained in post-electoral polls: it presupposes nominal variables and allows for the inclusion of individuals who did not respond to a question, without affecting the definition of the axes (only terms of interest were used to calculate inter-individual distance and scatter plots).  Moreover, it does not give equal weight to each response modality when establishing the axes, which allows for greater nuance regarding each attitude dimension.
51The two attitude axes were constructed separately: different analyses were conducted for each survey with regard to economic liberalism and ethno-authoritarianism. This choice allowed us to refrain from making hypotheses on the relationship between the two axes: in their construction, they could be partially correlated. In each analysis, the dimension being looked at was represented by the first axis.
52Each axis is ultimately presented as a continuous variable, where individuals are defined by their factorial coordinate on the axis in question. However, both axes do not use the same metrics: in order to resolve this question and hopefully compare the impact of variables across several surveys, the axes were standardised using units of standard deviation. 
Establishing the variables of ethno-authoritarianism
53The following table summarises the questions which allowed us to establish variables for ethno-authoritarianism and stipulates the rate of modified inertia of the first axis of the specific MCA (the axis retained). The contribution of each response possibility is available upon demand.
Questions used to establish ethno-authoritarian variables
Questions used to establish ethno-authoritarian variables
Establishing the variables of support for economic liberalism
54The following table summarises the questions which allowed us to establish the variables of support for economic liberalism and stipulates the rate of modified inertia of the first axis of the specific MCA. The contribution of each response possibility is available upon demand.
Questions used to establish economic liberalism variables
Questions used to establish economic liberalism variables
Ronald Inglehart, The Silent Revolution. Changing Values and Political Styles among Western Publics (Princeton: Princeton University Press, 1977); Piero Ignazi, “The silent counter-revolution. Hypotheses on the emergence of extreme right-wing parties in Europe”, European Journal of Political Research, 22(1), 1992, 3-34.Online
Hanspeter Kriesi, “The transformation of cleavage politics. The 1997 Stein Rokkan Lecture”, European Journal of Political Research, 33(2), 1998, 165-85; Hanspeter Kriesi et al., West European Politics in the Age of Globalization (Cambridge: Cambridge University Press, 2008).Online
Herbert Kitschelt, in collaboration with Anthony J. McGann, The Radical Right in Western Europe. A Comparative Analysis (Ann Arbor: University of Michigan Press, 1995).
Simon Bornschier, “France: the model case of party system transformation”, in H. Kriesi et al., West European Politics in the Age of Globalization, 77 –104; Simon Bornschier, Romain Lachat, “The evolution of the French political space and party system”, West European Politics, 32(2), 2009, 360-83.Online
Nonna Mayer, “Comment Nicolas Sarkozy a rétréci l’électorat Le Pen”, Revue française de science politique, 57(3-4), 2007, 429-45. Online
Jean-Luc Parodi, “Les élections ‘intermédiaires’ du printemps 2004: entre structure et événement”, Revue française de science politique, 54(4), 2004, 533-43. Online
Florence Haegel, Gérard Grunberg, La France vers le bipartisme? La présidentialisation du PS et de l’UMP (Paris: Presses de Sciences Po, 2007).
Gérard Grunberg, “Vers un espace politique bipartisan?”, in Pascal Perrineau (ed.), Le vote de rupture. Les élections présidentielle et législatives d’avril-juin 2007 (Paris: Presses de Sciences Po, 2008), 253-70; Florence Haegel, Gérard Grunberg, “Le bipartisme imparfait en France et en Europe”, Revue internationale de politique comparée, 14(2), 2007, 325-39. Online
Pierre Martin, Comprendre les évolutions électorales. La théorie des réalignements revisitée (Paris: Presses de Sciences Po, 2000).
Pierre Martin, “Les scrutins de 2007 comme ‘moment de rupture’ dans la vie politique française”, Revue politique et parlementaire, 1044, 2007, 167-75.
Within this framework, the theory of tripartition has been heavily criticised. Robert Andersen, Jocelyn Evans, “Values, cleavages and party choice in France, 1988 –1995”, French Politics, 1(1), 2003, 83-115. This analysis of a simple bipartition of French political life is confirmed in Robert Andersen, Jocelyn Evans, “The stability of French political space, 1988 –2002”, French Politics, 3(3), 2005, 282-301. Our article does not enter directly into this debate, but sees itself as more in sympathy with the conclusions of Gérard Grunberg and Étienne Schweisguth. Online
Vincent Tiberj, “L’électorat trotskyste: votes extrêmes ou vote de gauche? Une analyse par les systèmes de valeurs et leurs recompositions”, in Dominique Reynié (ed.), L’extrême gauche, moribonde ou renaissante? (Paris: PUF, 2007), 129-51.
Gérard Grunberg, Étienne Schweisguth, “French political space: two, three or four blocs?”, French Politics, 1(3), 2003, 331-47.
P. Martin, Comprendre les évolutions électorales.
Local alliances with the FN were tolerated by the RPR and the UDF (Union pour la démocratie française) until 1988 (agreements were reached in seven regions during regional elections in 1986, reciprocal drop-outs happened in the Bouches-du-Rhône department during legislative elections in 1988).
This dynamic accelerated after the regional elections crisis in 1998, when the moderate right-wing party was shown to have benefited from the FN’s support in order to conserve presidencies in five different regions.
Florent Gougou, “La droitisation du vote des ouvriers en France. Désalignement, réalignement et renouvellement des générations”, in Jean-Michel De Waele, Mathieu Vieira (eds), Une droitisation de la classe ouvrière en Europe? (Paris: Economica, 2012), 142-72.
Pascal Perrineau, “La dynamique du vote Le Pen. Le poids du gaucho-lepénisme”, in Pascal Perrineau, Colette Ysmal (eds), Le vote de crise. L’élection présidentielle de 1995 (Paris: Presses de Sciences Po, 1995), 243-61.
A summary of election results from 2002 to 2012 can be found in Table 9, Annex 1.
This percentage becomes even more significant when we consider that it was obtained despite the fact that the majority of UDF elected officials broke with Bayrou after the first round of the election and rallied around Nicolas Sarkozy by forming the Nouveau Centre.
Florent Gougou, Simon Labouret, “The 2011 French cantonal elections: the last voter sanction before the 2012 presidential poll”, French Politics, 9(4), 2011, 381-403.
What’s more, it is likely that the FN was hindered during the cantonal elections by low levels of participation, its working-class electorate being more likely to abstain than other segments of the population.
The linear correlation coefficients for the departmental level for Nicolas Sarkozy alone were 0.34 compared to 2002’s Chirac-Barre total, 0.37 compared to 1995’s Chirac-Balladur total and 0.43 compared to 2002’s Chirac-Bayrou-Madelin-Boutin total (0.40 for Chirac-Madelin-Boutin alone).
These coefficients were, respectively: 0.96 for Chirac-Barre 1988/Chirac-Balladur-Villiers 1995; 0.95 for Chirac-Balladur-Villiers 1995/Chirac-Bayrou-Madelin-Boutin 2002; 0.89 for Chirac-Barre 1988/Chirac-Bayrou-Madelin-Boutin 2002. They fell sharply to 0.68 for Sarkozy-Villiers 2007/Chirac-Barre 1988; 0.67 for Sarkozy-Villiers 2007/Chirac-Balladur-Villiers 1995; 0.72 for Sarkozy-Villiers 2007/Chirac-Bayrou-Madelin-Boutin 2002.Online
This coefficient was 0.91 at the departmental level and 0.94 at the cantonal level, if we consider only Nicolas Sarkozy’s scores.
At the cantonal level, the correlation coefficient was likewise very high: 0.95.
The construction of these three immigration zones is outlined in P. Martin, Comprendre les évolutions électorales, 270-8. Each zone is composed of 32 departments, according to the proportion of foreigners of North African or Turkish origins present during 1982’s general census. For a discussion of the pertinence and topicality of this indicator, see, by the same author: “L’immigration, un piège pour la droite?”, Commentaire, 132, 2010, 1027-36.
This record deviation can in part be explained by the smaller number of candidates in zone 3’s departments, hence a score of only 7.5% (the far right was present in 90% of cantons in zone 1, 85% of cantons in zone 2 and only 48% of cantons in zone 3).
The decrease in the interzone deviation with regard to FN votes in the presidential election went hand in hand with better results in zone 2 compared to zone 1 (as in 2007). From this perspective as well, 2007’s elections marked a break with the past: until then, the far right always obtained its best scores in zone 1.
The transformation of the right’s electoral map between Chirac and Sarkozy also reflects, albeit more marginally, differences in personal involvement and implantation, revealing a “friends and neighbours” effect in the Corrèze region and its environs for Jacques Chirac.
These two value axes refer directly to the issues at the heart of 1984’s electoral landscape: immigration and insecurity on one hand, and unemployment and the role of the state with regard to the economy on the other. For an empirical analysis of their decisive role in structuring the French ideological landscape, see Jean Chiche, Brigitte Le Roux, Pascal Perrineau, Henry Rouanet, “L’espace politique des électeurs français à la fin des années 1990. Nouveaux et anciens clivages, hétérogénéité des électorats”, Revue française de science politique, 50(3), 2000, 463-88.
Vincent Tiberj, “La politique des deux axes. Variables sociologiques, valeurs et votes en France (1988 –2007)”, Revue française de science politique, 62(1), 2012, 71-106 (published online in the RFSP (English) as “Two-axis politics: values, votes and sociological cleavages in France (1988-2007)”, 62 (1), 2013, 67-103); and “Values and the votes from Mitterrand to Hollande: the rise of the two-axis politics”, Parliamentary Affairs, 66(1), 2013, 69-86.
For more details concerning the construction of these axes, and more specifically concerning the questions used in the analyses, please see Annex 2.
For more on this issue, already well covered in the existing literature, see Nonna Mayer, Ces Français qui votent Le Pen (Paris: Flammarion, 2002). In general, the results of our analysis confirm and expand upon the results presented in Tiberj, “La politique des deux axes”.Online
A slight dip can be seen on the economic axis starting in 2007: the tendency to vote for the FN among the most ethno-authoritarian voters increased alongside hostility towards economic liberalism (except for the most anti-liberal deciles), whereas it had a tendency to decrease in 2002. The variations produced remain minor, however. Online
The presidential election of 2007 also marked an important change with regard to the economic dimension, with Sarkozy voters clearly more strongly polarised than Chirac voters in 2002. It is likely that this situation was partially influenced by the absence of Alain Madelin in 2007. Nevertheless, the Sarkozy vote reflects the two-pronged radicalisation, both economic and cultural, of the moderate right-wing voter in general.
Jean-Marie Le Pen never managed to secure hegemonic influence in this segment of the electorate: in 1988, 1995 and 2002, the moderate right was already attracting a significant part of the most xenophobic electorate. The moderate right’s ability to attract an ethno-authoritarian population should be likened to the influence exerted on FN votes by the presence (or absence) of immigrants. As Pierre Martin and then Nonna Mayer have shown, those hostile to immigrants only overwhelmingly vote for the FN when they live in a department with a high concentration of foreigners. Cf. P. Martin, Comprendre les évolutions électorales, 277-8; N. Mayer, Ces Français qui votent Le Pen, 274-5. Online
From this point of view, the basis of 2007’s Bayrou vote sheds light on the rationale behind the empowerment of a portion of the centre-right electorate: unlike the Sarkozy vote and the Le Pen vote, the Bayrou vote decreased with the level of ethno-authoritarianism (which was much less the case in 2002, and not at all the case in 1995 for Balladur or in 1988 for Barre).
The refusal to call for the defeat of the Front National was still provoking heated disputes during 2011’s cantonal elections, when the Prime Minister François Fillon called on people to “vote against the FN” in the case of a duel between the left and the FN. In 2012, the “neither-nor” (neither voting for or against the FN) strategy adopted by Sarkozy and Jean-François Copé was accepted and adopted by all UMP political players.
Hanspeter Kriesi et al., Political Conflict in Western Europe (Cambridge: Cambridge University Press, 2012).
Florent Gougou, Simon Labouret, “The 2010 French regional elections: transitional elections in a realignment era”, French Politics, 8(3), 2010, 321-41. Florent Gougou, Simon Labouret, “Revisiting data on the 2012 French legislative elections: political supply, party competition and territorial divisions”, French Politics, 11(1), 2013, 73-97.
For a detailed discussion of this point, see F. Gougou, S. Labouret, “Revisiting data on the 2012 French legislative elections”.
For a similar approach, and a more nuanced discussion of the methodological assumptions, see V. Tiberj, “La politique des deux axes”.Online
For an exploratory use of specific MCA, but also for a presentation of its statistical principles, see J. Chiche, B. Le Roux, P. Perrineau, H. Rouanet, “L’espace politique des électeurs français à la fin des années 1990”.Online
This process does not guarantee the substantial comparability of the axes, in the sense that they were not established using the same questions. Comparability is thus hypothesised.