Congrès AFSP 2009

Section thématique 48 : Maurice Duverger aujourd’hui

Session 6

From volatility towards strategy:tTwo measures of the dynamic of electoral behaviours

in presidential elections from 1988 up to 2007

Jean Chiche, Sciences Po, Centre de recherches politiques de Sciences Po (CEVIPOF), CCNRS.

jean.chiche@sciences-po.frElisabeth Dupoirier,

Centre de recherches politiques de Sciences Po (CEVIPOF). elisabeth.dupoirier@sciences-po.fr,

From volatility to strategy: two measures of the dynamic of electoral behaviour in presidential elections from 1988 to 2007

Abstract : Like most other European democracies, France has had to face an increasing tendency towards voting mobility since the middle of the eighties. This study confirms that the multiparty system and the electoral formula based on a two-round majority system have encouraged this tendency. The study also deals with strategic voting based on the same conditions that André Blais used to describe the 2002 presidential strategic phenomenon. Comparing the levels and structures of strategic presidential voting in 2002 and 2007, we point out that even if a two-round system encourages more candidates to run, (thus causing a much more fractioned vote than the one-round system), these two conditions in favour of strategic voting can be neutralized by the specificity of the political context within a given election. Unlike in 2002, the overall level of strategic voting in 2007 presidential election was weak and was much more structured by the desertion of weak than of viable candidates. The majority of 2007 strategic voting was done by individuals who had voted outside the political mainstream during the 2002 election and who cast a “vote utile” in favour of one of the viable candidates - Ségolène Royal or Nicolas Sarkozy.

Keywords :Voting mobility, volatility, strategic voting, single-member dual ballot system, French presidential election, political context.

Version definitive CB 30 janvier17 mars

Ever Ssince the eighties, France has been involved in anexperiencing increasing movement of volatility of thein electoral behaviours (Grunberg, 1984; Dupoirier, 1986). As elsewhere, this movement has speeded up (accelerated?) the electoral change and therefore has therefore provided brought about an increasing turn over of the governments. Studying several Western democracies, Ivor Crew and David Denver analysed this volatility in the behaviour ofof citizens’ behaviours in several European countries as being the the result of athe rise of in individualistic values and attitudes. According to them, this encouraged that leaded the voters off to feel more and more freer vis- a- vis the political parties and, therefore, more opened to casting a vote based on the specific issues of each election instead of a on partisan alignment (Crew, Denver 1985). At the same time, a French analysise of the voters’ mobility from the 1986 ’general elections to the first round of the 1988 ’presidential election concluded that the huge partmajority of the electoral volatility provided stemmed from “unreasoning voters” as Philipp Converse had described them a few years earlierbefore in a seminal paper (Converse and Pierce, 1986). At the end of the eighties, only a tiny part amount of unsteady electoral behaviours from onean election to the following onenext might could be considered as a strategic voting based on the political issues of the election in question and the wishwill to give strategic support strategically to a candidate who appeared more likely to win rather than the preferred candidate (Boy and Dupoirier, 1993; Swyngedouw, Boy ,and Mayer, Swyngedouw 2000). An other study dealing with the 1995 presidential election concluded that the strategic vote wasis still weak in France inat the middle of the nineties (Chiche and Jaffré, 1997).With Adopting a different approach, severalome Northd American researchers studied the relationship between the weightnumber of unstable voters and the rules governing the electionelectoral rule itself. They sought out studied the relationship between one or two- rounds electoral systems and strategic voting (Cox, 1997; Blais and Massicot, 2002). At the opposite ofUnlike Duverger’s assertion that in the first round of a two- rounds system, voters cast their vote to express their preferences, they argued that even in France where this electoral system exists has existed for years, strategic voting is also exists and is increasingexisted and growing up. Based on thean 2002 Panel Electoral Français electoral survey(PEF) dealing withof the 2002 presidential election, a Blais’ study set up precise items to describe what French strategic voting was. At the end of the empirical research, and he concluded at the end of the empirical research that strategic voting reached 9.5% of the presidential voters (Blais, 2006).This Our study will deal withextends those this previous researches. A The first stage, based on all the presidential post-electoral surveys conducted by CEVIPOF since 1988, will be dedicated toconcentrates on the study ofstudying the process of mobility and strategic voting from the end of the eighties to nowadaysuntil the present time asusing the same methods for assessment as they were assessed those used in the first 1988 study. Do Has the weightnumber of unstable voters increased all alongthroughout the last twenty years more or less in accordance with the global movement that is observed in all the Western democracies? And among changing those French changing voters, how many might be considered as strategic voters? Does Have their sociological and political profiles changed since the eighties?

2

A The second stage will deals with the only 2007 presidential election only and isbe based on the presidential panel conducted by CEVIPOF (about 4,500 interviews). We will aim to experiment with the same conception of strategic voting that André Blais defended for the 2002 presidential election. Does a new empirical study on a this currentrecent presidential election confirm that a significant and increasing part number of French voters citizens have becomeare strategic voters?

I Mesuring French mobility from onea national election to the next one since 1988.

I.1. Mobility: an increasing political phenomena. As suggested above, a lot ofmany different methods are used in order to assess voting mobility. This study deals withoperates under the same conditions that as those chosen by Boy and Dupoirier to evaluate the votering mobility from the 1986 general elections to the 1988 presidential election. Firstly, this method considers as aa “mobile voter” to be the voter and only voter an individual who cast his/her vote from an election to the next one for parties belonging to two different sides of the left/right borderfrontier from one election to the next. Secondly, we have only selected the only voters who cast a vote at both elections: turn out voters or voters who decide to cast a white blank ballot were excluded from the 1988 study and are still also excluded from this new one.Therefore, Wwe can therefore observeassess that between 1988 and 2007, the voting mobility from onea general election to to the next one increased from 23.2% between the 1986 and 1988 elections up to 46.,1% between the 2002 and 2007 elections (Table 1). MThe mobility increased regularly and the rise seems to have been accentuated recently: the movement has reached 12% from between the 2002 presidential election to and the 2007 one, instead ofas opposed to 5% during the former previous period (1988-2002). As the a result of this tendency, the amountlevels of the Cramer V coefficient1 have been decreasing irregularly since 2002: from 0..57580 for the beginning of the period down to 0.517 atfor the end of the period.

Table 1 . Mobile voters in French presidential elections : 1988-2007

Mobilityé dfrome G19L86 toà P1988*

fromde G19L93 toà P1995

from de G19L97 to àP2002

de from P2002 à to P2007

votants stablesStable voters

76.8 71.8 65.7 53.9

votants mobilesMobile voters

23.2 28.2 34.3 46.1

Value of Cramer’s V Valeur du V de Cramer

.575 .580 .520 .517

Durée de la trajectoireLength of the trajectory

2 ans 2 ans 5 ans 5 ans

candidats à la présidentiellePresidenti

9 9 16 12

1 The Cramer V coefficient based on the Chi-square test-which assesses the significance of the relationship between the general and presidential elections -

3

al candidatesNTotal 2602 2466 1753 1309Sources : CEVIPOF/ enquêtes post électoralespost electoral surveys* G= general elections; P= presidential election.ADD IN NOTE

Two significant changes happened occuredoccurred during this time. The first one wais the rise in the number of presidential candidates. Even if the 16 candidates inof the first round of of the 2002 election appeared to beas a “historical accident” and even if there were “only” 12 candidates in 2007 number of candidates was limited to 12, , it may be assessedremains true that the more variedous and numerous the candidates are, the highestr is the probability is that the voters will cast their presidential vote differently from the last way they voted in the previous general election vote. The second disruption, which modifies theour analyses, is the length of the period of time between general and presidential elections which has increased from 2 years to 5 years since 2002. It may be assessed that tThe longer the time between two national elections is, the higherst is the probability is that voters will reconsidered their last vote. According to the theories of “retrospective voting” and “reasoning voting” an increasing number of voters who had supported a party belonging to the majority coalition, might well could decide to cast a “punishment vote” which leading them up to a volatile behaviour with respect toregarding their former vote.

At During the 1988 presidential election, volatile voting concerned both left-wing and right- wing candidates were concerned in the same and weak proportions by volatile voting: around 4%. But However, the increasing mobility which had started withfrom the 1995 presidential election concerned neither all the left-wing nor all the right-wing voters in the same proportions (Table 2.). The Lleft- wing voters used to be much more mobile thant the right- wing voters whose amongst whom mobility has started more recently during: for the 2007 presidential election. For on the opposite of the leftist voters, whose mobility had begun earlier and had never stopped all overcontinued throughout the period.

Table 2.Cross-border Mobile Vvoters coming from the left wing and from the right wing parties

familiesTrends 1988-2007

Taux de mobilité enRate of mobility in %

G1986/P/1988

G1993/P1995

G1997/P2002

G2002/P2007

Mobilityé2007/1988

Mobile left-wing and ecologist voters Electeurs de gauche et écologistes mobileshors de leur campoutside their camp

4.3 7.3 13.5 14.6* + 10.3

Mobile right-wing and ExR voters Electeurs de la droite et EXD mobileshors de leur campoutside their camp

4.4 4.1 3.8 12.0 + 7.6

A vote pour for François Bayrou is considered as a non-left voteconsidéré comme un vote hors gauche.

Source : CEVIPOF/ enquêtes post eélectoral surveyses

4

Inside the left-wing and the right-wing party families, the voters who were Mmobile voters, both from the right and the left, above alltend to cast e their vote for a majority party at from the former previous general election: According to the theory of retrospective voting, the people who were not satisfied with the outcomes of the government’s policies changed their vote at the next following presidential election in favour of a candidate belonging to an oppositional party. More often than not, the party which caughtthey gave their vote to the party which was the one most likelywhose probability to win the election was the highest. At the oppositeAt the other end of the spectrum, voters who supported radical or extremist movements used to changed their vote for a party belonging to the same political block rather then to cast a vote for a party belonging to the other side of the left-right border. For instance , from 1988 to 2002 at each new national election, 99% of the FN voters cast the same vote in favour of Le Pen’s party. But However, duringat the 2007 presidential election, the reproduction of FN votes were not reproduced in the same numbers. was stoppedLess than half of the 2002 FN voters remained faithful to Le Pen. . Faithful Le Pen’s voters were limited to less of the half of 2002 FN voters. The second half cast their voted for the UMP, the majority party of the right-wing family of parties parties family. Fewer than 2% of FN voters who cchanged their vote for a left-wing candidate were less then 2%.

I.2. Who was the mobile voter?To At the endconclu ofde their 1988 study, Daniel Boy and Elisabeth Dupoirier wrote concluded that their outcomes did not confirm the model of a voter whose mobile behaviour might be described as “strategic”: the strategic voters did were not be numerous enough to be observed by statistical methods. At the opposite of a tinyRather than the numerically weak relationship between mobility and strategic voting, they highlighted that mobility was linked with solid socio- economic al pressures and/or/and weak low levels of education al level or/and/or low levels of political interest in politics. They therefore they concluded that the Michigan Model and Converse’sthe seminal Converse’ paper (Converse, 1986) which describeding a voter who felt bewildered by the complexity of political choice, was still the best onethe most credible way to explain mobility voting.Basically and despite the fact that mobility had beenwas increasing, our analysise concerning the 1988-2007 period of time showsed that the former conclusions arewere still up to daterelevant to explain the majority part of voting mobility phenomena nowadays. However, some significant differences have appeared. The first one was aboutconcerns the relationship between the socio-economic profile of the voters and their electoral behaviours – which is more or less volatile. In the 1988 study, the probability to belikelihood of being a mobile voter was strictly linked to the socio economic pressures which the weighed on the voter had to assume. Based on the a measurement of inequalities of inproperty ownership, the socio economic pressures model of socio economic pressures showeded that the less fewer items of property a voter owned, the more she/he was likely the probability to be a mobile voter he had. Used ten years after later to assess the mobility from between the general elections in 1997 up toand 2002, general elections the outcomes of the model pointed out the weakness of the socio economic explanation concerning of volatility: the decrease of thein value of the Gamma coefficient2 over in all the owner ’s categories highlighted the decrease of in the relationship between economic pressures and mobility. On the other hand, the “poorest” voters are still the mostre mobile, a fact that which suggested suggests that Converse’s conclusions are still operated in operation (Converse, 1986). As the 1988 study showed, the

2 The Gamma coefficient measures the strongness strength of the link between two variables. In thaist case, these arey the are level of property ownership and the stability/instability of electoral behaviour.

5

mobile voters were not voters free enough from socio- economic pressures to base their vote on imunmaterialistic considerations only ( Table 3).

Table 3Measurement of stability/instability of vote

considering a scale of property owned by the voters Relationship between Number of Items Owned and Voter Trajectories

(as measured by the Gamma Coefficient)

Elémentsde patrimoineNumber of items owned

coefficient gammaVote trajectoriestrajectoires de

votes1986/1988

Coefficient gammatVote trajectoiries

1997/2002

0-1 0.79 0.682 0.86 0.823 0.86 0.744 0.88 0.765 0.93 0.85

6 et plusand more

0.94 1.00*

Source : CEVIPOF/ enquêtes post électoralespost electoral surveys

Nowadays, the socio-economic pressure model of pressure is better highlighted by the variable which describes the voter’s greater or lesser security of more or less safety of the voter’ employment. Voters who have awith precarious situation ofinsecure employment (employment for a limited time) are basically much more mobile then those who benefit of a much more safety situation ofhave secure employment. Between 1997 and 2002, tThe gamma coefficient of which measures stability of intheir votes from 1997 up to 2002 reached 0.78 (if they forare civil servants with safety secure employment) andor 0 .73 for ‘if they worked in private sector employees with with an permanent work contractjob. By contrast, the gamma coefficient failed downfell to 0.46 for the workersemployees who hadwith a fixed-term contract only.The second point is that, nowadays, mobility more often affects more often young voters – under 40 years old - than their elders. Among the young voters less educated and educated seemed to be equally affected by electoral mobility. Whatever their level of education might be, young voters are all equally affected by electoral mobility.For both categories t For these voters, the values of the gamma coefficient scored reached 0.63 and 0.65 in a 2007 study when whereas they had scored 0.79 and 0.90 in the 1988 study (Table 4). Therefore, and as a

6

result of its increasing tendency to spread and spreading tendencies throughout the all the population, the volatility was turning into mixed phenomena. A part The majority ofof mobile voters had a are relatively sophisticatedrelevant profile of sophisticated voters: young, educated, interested in politics but without partisan identification. They may be considered as “strategic voters” and differentiated from the voters described in the 1988 study who alwaystended to looked asseem somewhat disoriented voters as described in the 1988 study. Among young voters, there are clearly fewer of the second category is clearly less numerous then than of the first one.

Table 4Relationship between age, level of education and vote trajectories(as measured by the Gamma coefficient) Corrélations (gamma) selon l’âge et le niveau de diplôme en 1988 et en 2007.

Age et niveau scolaireAge and level of education

Reproduction des voVote

trajectoriestes1986-1988

Reproduction des voVote

trajectoriestes2002-2007

Young people with low level of educationJeunes faiblement diplômés

0.79 (633)° 0.63 (162)

Seniors Older people with low level of education faiblement diplômés

0.86 (827) 0.73 (180)

SeniorsOlder diplômésGraduates 0.89 (678) 0.80 (576)Jeunes diplômésYounger Graduates 0.90 (464) 0.65 (392)Entre parenthèsesBetween brackets, numbers within the groups used to calculate the gamma coefficients effectifs des groupes sur lesquels sont calculés les coefficients gamma. Source : CEVIPOF/ post electoral surveysenquêtes post électorales

7

II Strategic voting and the first round of the 2007 presidential election.

How Blais’ had taped and the Cox’s notion of strategic voting.The second approach seeks to discusses Blais’ and Cox’ and Blais’ studies concerning relative to the existence and the amount level of strategic voting in a two- rounds majoritariany election (Cox, 1997; Blais, 2002 and 2006). Basically, both researchers aimed to questiondisputed Duverger’s assumptionquestioning of the existence of writing that strategic voting could not have existed within such an electoral formula. Duverger argued that because the voters were not ablecould not to obtain the information that strategic voting requires before casting their vote the information that strategic voting required (Duverger 1951). To emphazise However, given that today’s voters have such a sophisticated level of knowledge (Chiche, Haegel, 2002) about politics, the importance of disputing Duverger’s conclusions seems particularly relevant. with Cox and Blais the importance of disputing Duverger’s conclusions specially on French case, we could argue the presidential election did not have existed in France when Duverger had written his seminal comparative book on political parties and the political effects of electoral formula (Duverger, 1950). In addition, we could also point out the increasing number of electoral surveys that had been published by the French newspapers all alongthroughout the various presidential campaigns should also be pointed out. During the 2007 campaign, 293 surveys had been were published (Dupoirier, 2008). MoreMany of them had beenwere analysed by newsmen newscasters before the first round of the election relating the hypothetical results of several second round’s scenarios. We argue that, nowadays, surveys and the media provide enough significant additional information to help outcome-oriented voters to make form antheir opinion about the main candidates’ chances of getting elected before the first round takes placeabout the chances the main candidates have of getting elected before the first round. With In agreement with Cox, we argue that some voters don’t care about the candidate they feel closestthe most closed to themselves in terms of values or interests or party identification:. But “practically speaking… (they) care only about the outcome of the current election and have rational expectations”3 .Considering With reference to the 2002 French presidential election, Blais had dealt withused this Cox’s concept of “strategic voting in single-member dual-ballot systems”4 as a starting point. Blais defined Cox’s concept stated as “a vote for a candidate that is not the preferred one, motivated by the intention to affect the outcome of the election”5. Blais had based his study on the first wave of the 2002 Panel Electoral FrenchFrançais Electoral Panel which had beenwas realized during the run up of to the first round. He only had taken took into account the only people who had decided to cast a vote into account6. He had used four questions specially designed to tap describe the concept of strategic voting. Two of the questions are designed in order to collect data concerning on the names of the top two preferred candidates foras president (first and second choices). The other two questions are designed to and the forecasts about the chances of those candidates to beof being elected president7.

3 Cox, 1997, p124.4 Cox, 1997,p123-138.5 Blais, 2006, p 98. 6 The French Electoral PanelPanel Electoral Français was a three-wave survey jointly conducted by CEVIPOF, CIDSP and CECOP. For the first wave 4,017 at home interviews were conducted and 3,123 respondents indicated a vote intentionhow they intended to vote at in the first round. The data had beenwas weighted “so as to reflect the socio-demographic profile of the fFrench electorate as well as the outcome of the election” (Blais, 2006). 7 See the questions in Blais, 2006, p96.

8

In order toTo have be en considered as “strategic”, a vote wouldmust have beenbe caste for a candidate who had is not been the second preferred favourite one. and tThe voter’s decision in favour of this “second choice” would have is been related to his/her perceptions of the outcome of the race. As a result of those two conditions, Blais had pointed out two sorts of strategic voting. The first one occurs when the voter gives up had leaded to give up a favourite candidate who is “out of the running”. This is the well known “vote utile” (useful vote) in favour of one of the most “viable” candidates (utility voting). The second sort of strategic voting is more sophisticated. The voter would usesd the first round as an opportunity to deliver a message to his/her preferred candidate using his/her ballot for as a “vote d’avertissement” (warning voteing). This sort of vote needs two conditions. Firstly, the voter wouldmust have a strong insurancebe convinced that his/herthe preferred candidate would has geta very strong chance of making it s to access to the second round of the election and therefore doesn’t need a vote in the first round.. Deliberately, Tthe voter would deliberately chooses to cast his/her vote in favour of a weaker candidate he had perceived as to be out of the running. But, this vote for a weak candidate can only be considered as “strategic” only at theon condition that theis weak candidate had beenis closed to the issue he the voter is the most interested in. As Cox had writtenwrote “A candidate who has more votes than he neededs to get into the runoff will, in other words, be relieved of those votes by strategic voters”8 The Cox’s proposition seemounds well appropriate to describe what had happened to the leftist candidate, Lionel Jospin, ion the 2002 presidential election: the a second round race with Jospin and /Chirac was the only one expected one according to the surveys and the media results. According to this framework, In this context, Blais had estimated that 9.,5% of the whole electorate had caste such a vote duringon the first round of the 2002 presidential election.

Assessing 2007 strategic voting.

In oOur study of the 2007 presidential election, we sought to emulate the conditions set up by Blais in 2002. recover conditions of experimentation as closed as possible to the 2002 Blais’s conditions. LikeAs him,self we used the first wave of the CEVIPOF 2007 French Electoral Panel Electoral Français and likeas him,self we only selected the only people who indicated a votthat they intended to votee: 3,238 respondents9. We also used a data file weighted to be closereflect to the socio-demographic profile of the 2007 French electorate as well asand to reflect the outcome of the presidential election10. Five of the survey questions of the survey helped us to tap describe the concept of strategic voting. Two questions has been chosenwere designed to collectfind out who the voters’ preferred presidential candidate was as President and the forecast of his/her chances to beof being elected11. Two others questions concerned the intended votes foron each round. The last question collected the name of the candidate who - according to the voter’s opinion - proposed the best solution to what the voter considered to be the main electoral issue.Before assessing the amount of strategic voting with using Blais’s method, we had to frame the electoral context of the 2007 presidential election and to point out the differences of voting conditions from onean election to the another. For a lot ofmany reasons, and compared to the former presidential elections, 2007 appeared as to break with“a breaking off” previous elections (Perrineau, 2008). Three of those breaks may well have influenced ing off could have hung over the behaviour of outcome- oriented 8 Cox, 1997, p129.9 Blais had worked with 3,123 respondents from the 2002 French presidential French panel.10 We used the French Electoral PanelPanel Electoral Français conducted by the CEVIPOF with the financial grants fromof the Fondation Nationale des Sciences Politiques and the Department of the Interior.11 The same questions as those Blais used in his 2002 study. See Blais, 2006, p.96.

9

voters. ’s behaviours. Without a doubt the first of theseone was the end of the formerprevious political apathy on the part of French voters. All overThroughout the campaign, the intendedmotivated voters cared about the media information. As a result, the turnout has beenwas very high even at during the first round (85.2%). The second breaking off was aboutconcerned the undecided outcome of the first round between the 3 three candidates selected by the surveys and the media for the final horse race: Ségolène Royal (PS), François Bayrou (Centere) and Nicolas Sarkozy (UMP) were all introduced presented as potential winners until the end of the run up. The third breaking off was the electoral volatility of the issues called upevoked during the campaign. At the opposite ofUnlike the 2002 campaign, none of them succeededs in focusing the voters’ interest for a long time. The more or less attractive political images of the candidates ( Boy, Chiche, 2008) played the role the issues had played during the 2002 campaign to help the voters to make their choice. The fourth break ing off was aboutconcerned the main candidates: at the opposite ofunlike Lionel Jospin and Jacques Chirac in the 2002 race, Ségolène Royal and Nicolas Sarkozy were “new “presidential candidates”: among the viable candidates François Bayrou was the only one who had an electoral past as presidential candidate. All ofNone of theose breaksing off were not in favoured of strategic voting but all of them had played a role in shaping the voters’ judgments about on the electoral competition. Table 5 shows how the 2002 and 2007 elections were different in terms ofthat they revealed a hiatus between the voter’s’ preferences and their estimation of the ed chances their preferred candidate to had of winning win the race.

Table 5

Preferred candidate and pessimistic forecast of his/her chance to of winning the racePreferrrred candidate Pessimistic forecast

for preferred candidate

2002%

Pessimistic forecast for preferred

candidate2007

%Un A petit weak left-wing candidate candidat de gauche (100%)

51 94

Un A weak petit candidat de droiteright-wing candidate (100%)

37 33

Le Pen (100%) 37 33Bayrou (100%) 24A PS candidateUn candidat du PS* (100%)

34 9

Un candidat de la droite modéréeA right-wing candidate** (100%)

40 1

Ensemble de l’électoratTotal (100%) 37 21 *Lionel Jospin at the 2002 election; Ségolène Royal at the 2007 election; ** Jacques Chirac at the 2002 election; Nicolas Sarkozy at the 2007 election. Source CEVIPOF Panels 2002 and 2007, wavess number No 1.

10

At During the 2002 presidential election, the pessimistic attitude regarding the chances of the preferred candidate had of winning, to win had been were widely spread all overthroughout the whole electorate (37%). This was almost equally true , among the supporters of viable candidates as the and supporters of the weaker ones. Even As many as 37% of Jacques Chirac’s voterssupporters as and 40% of Lionel Jospin’s voterssupporters, were indicated pessimistic forecasts about their candidate’s chances of winningabout the chances they had to win t the race. The 2007 electoral context was quiete different. Only tiny minorities of the two viable candidates – 9% of Royal’ voters as and 1% of Sarkozy voters – had doubts about whether their most preferred candidate could win or not. Numerous The many pessimistic forecasts were limited to thesupporters of the weaker candidates supporters: from 33% for weaker right-wing candidates up to o…94% in the case offor supporters of the weaker leftist candidates supporters (instead ofas opposed to 51% in the 2002 election). With 24% of pessimistic forecasts, the Bayrou ’s case was perceived by his supporters to be halfway between those abovethe weak and strong candidates. ThisIt was the first of the indications of the multidimensional perceptions of Bayrou’s chances to be inof making it to the second round of the race. Basically, the pessimism of the whole electorate as a whole had decreased from 37% to 21% from between the 2002 to and 2007 presidential elections. The second difference between the two elections was the amounts size of the hiatus between preferred candidate and intended vote. Table 6 shows that at during the 2002 election, more thaen a quarter of the whole electorate (27%) had intended to vote to for a candidate who had was not been the most preferred one. Desertion was strongest It had been among the supporters of the two viable candidates - Lionel Jospin and Jacques Chirac – that the desertion had been the strongest: 40% for the first oneformer ; 33% for the second onelatter. On the other hand, the supporters of the weakest candidates had were only been a tiny minority ( 5% of Le Pen supporters) or slight small minorityies ( 18% and 17% of supporters of the weakthe left wing and right- wing candidates respectively weak candidates) to express an intention tod a vote for one of the two viable candidates. Regarding In comparison to this former context, the 2007 presidential election appeared as very different. Firstly, the hiatus between preference and intended vote deeply greatly decreased: it reached 17% only which was 10 points under below the 2002 levelamount. Secondly, and at with the exception of the case of the weak leftist candidates, the hiatus was always slight whaotever the preferred candidate might bewas.

Table 6The hiatus between preference and intended vote atduring the 2002 and 2007 presidential

electionscandidate preference Intention toV vote for a

candidate

different fromotother than the preferred candidate 2002

%

Vote Intention to vote for a candidate

different fromother than the preferred candidate

2007%

Un A petit weak left-wing candidatecandidat de gauche (100%)

17 30

Un A weakpetit right-wing candidate candidat de droite (100%)

18 18

11

Le Pen (100%) 5 16Bayrou (100%) 17A PS candidateUn candidat du PS* (100%)

40 16

Un candidat de la droite modéréeA right-wing candidate** (100%)

33 14

Ensemble de l’électoratTotal (100%) 27 17 *Lionel Jospin duringat the 2002 election; Ségolène Royal duringat the 2007 election; **Jacques Chirac duringat the 2002 election; Nicolas Sarkozy duringat the 2007 election. Source CEVIPOF/  Electoral Panels 2002 and 2007, waves Nonumber 1.

Regarding With regard to theose differences inof electoral context and in order to assess the level of strategic voting in 2007, amount of strategic voting we built two scenarios. The first one took into account the same conditions as those Blais had studied into account: Ségolène Royal and Nicolas Sarkozy were the only “viable” candidates as Lionel Jospin and Jacques Chirac were in 2002. As a result, François Bayrou was considered as a weak candidate. The second scenario dealt with the undecided rank occupied by Bayrou candidate in the last surveys published before the first roundun. We also tested the voters’ choices in this new case ofscenario with three viable candidates.

Table 7Two scenarios assessing strategic voting in the 2007 presidential election

Second round Sscenario du 2ème tourRoyal Sarkozy

%

Second round Sscenario du 2ème tourRoyal Bayrou Sarkozy

%

« vote utilevote utile » vers for Ségolène Royal

2.1 2.0

«  vote utilevote utile » vers for François Bayrou

- 0.6%

« vote utilevote utile » vers for Nicolas Sarkozy

0.7 0.5

Total « Vote utilevote utile » 2.8(1.0 ein 2002)

3.1

« vote d’avertissement » against « vote d’avertissement » contre Royal

0.5 0.5

« vote d’avertissement » against «  vote d’avertissement » contre Bayrou

- 0.2

« « votevote d’avertissement » d’avertissement  » contre against Sarkozy

0.5 0.4

Total « votevote d’avertissement d’avertissement »

1.0(8.5 ein 2002)

1.1

Total strategic vote stratégique 3.8(9.5 ien 2002)

4.2

Source: 2007 CEVIPOF-Panel wave 1.

Basically and regarding both scenariosi the amounts levels of strategic voting collapsed ion the 2007 election: from 9,5% of the whole electorate down to 3.8% oin the first round in

12

2007, first round if we took into account the same scenario then as Blais’ (with only two viable candidates)s into account;) down to 4.2% if considering François Bayrou was considered as a third viable candidate. The collapse of the strategic voting led to a 2007 new balance in 2007 between the “vote d’avertissement” -– which decreaseding straight right down from 8.5% to 1% -and the “vote utile”- which increased slightly increasing from 1% to 2.8%. Ségolène Royal took a better advantagebenefited more from of the “vote utile” (2%) thaen Nicolas Sarkozy (0.5%). However, the both the viable candidates managed to escaped fromavoid the dangerous “vote d’avertissement” which had caused so much damage tod the 2002 Jospin’s candidacy in 2002 duringfrom the first round.The conclusion of study on the 2002 and 2007 experimentations elections is that the specific electoral context of each election must be taken into account to assess strategic voting,: to point out its amountthe level of strategic voting and as to investigate upon the origins of this kind of rational behaviour.

From strategic voting to strategic voters: voters’ profiles in 2007 voters’profiles.

Both Blais ands Cox were more interested in assessing the origins and the amounts of strategic voting then in identifying the profile of those of the voters who caste their voted strategically. This third part of our study deals with strategic voters. The question is: in 2007, towards what extentd 2007 were strategic voters different from others categories of voters? In order tTo answer to this question, we built a new variable which classified the respondenters into three categories12. The first one gathered grouped together the strategic voters as they had been assessed into the second scenario of strategic voting (see part II). With In this scenario, François Bayrou was regarded as if he was one of the “viable” candidates and strategic voters weighted formade up 5% of the wholethe 2,652 respondents voters. TRegarding their socio- economic profiles they were not marked bydidn’t have any salient feature. The second category bound grouped together the respondentrs whose preferred candidate was a weak candidate and yetwho nonetheless intended to cast a vote for him/her although they thought s/he had no chance of beings to be elected as President. On the opposite ofUnlike the strategic voters who were outcome-oriented voters, this category ofthese respondentrs were opinion-oriented voters. Theyy wereare much far more numerous numerous then the first categorystrategic voters: 17% of the total with some a slight tendencyies to be towards younger (22% among the 18-24 years olds) individuals) and bemore highlyst educated voters than in the electorate as a whole voters. The third category gathered includedthe voters who indicated Royal, or Bayrou or Sarkozy as their preferred candidate for whom they intended to cast their vote believthinking by the way that their candidate had a serious chances to beof being elected. We considered theose voters as “conformist” voters in the sense that they didn’t haved to face any hiatus between preference, optimistic forecast and vote13. On the opposite ofUnlike the others categories of voters, the number of conformist voters grewincreased up among the older individuals voters and among those who indicated a regular religiousn practice.In order tTo explain the variations of in electoral behaviours among the voters, we used multinomial regression analyses. The dependant variable was the variable thatone which described the three categories of voters: “strategic-oriented”; “opinion-oriented” and “conformist-oriented”. We chose the conformist-oriented voters category as the reference 12 We only took into account the responderntss who indicated an intended vote and opinions on the two others questions used to build the three categories of voters into account. A total of 2,652 individuals had beenwere selected. 13 See Annexe 1 to see how the three categories of the variable the building ofwere built the three categories of the variable. The variable only took only into account the voters who answered to the three3 questions used to build the variable. into account.

13

category and explored the other two others. Three models were investigated. Without Unsurprisingly,e a socio- demographic model did no’t reveal show any salient results to explain strategic-voting or opinion-voting14. The pseudo R2-deux was only 0.038 and at with the exception of the age, the others variables were not significant (Clarke and Lipset, 2001). Another model which stressed classic items used to describe a sophisticated voter - basically interested in politics, informed about the electoral campaign, feeling self efficientautonomous enough to understand politics but without party identification – failed to explain more then 0.011 (value of the pseudo R2-deux). The variable of self efficiency variable was the only one which worked significantly. The third model dealt with items that described a traditional relationship between individuals and the political system: party identification; position of the individual on a left wing/right wing scale that described the political space and former previous votes duringon the 2002 presidential election. Among the three models, it this was the only one which explained strategic voting or opinion voting on the one hand compared to conformist voting on the other hand. The value of the pseudo R2-deux was 0 .192 and all the variables were significant. TAmong them the former 2002 presidential vote was the most potent variable. All things being equal, anyone who voted the 2002 votes in favour of one of the weak non- socialist candidates in 200215, was three times more likely to vote strategically strengthened the probability of voting strategically ion the 2007 presidential election up to 30% of chances. . At the opposite end of the political landscapespace, a same similar relationship existed between 2002 Le Pen’s voters in 2002 and strategic voting in 2007 strategic voting: anyone who voteding for the FN leader ion the 2002 election was gave a rise of 20% twice as likelyof the chances to vote strategically ion the 2007 election. With In others words, the most partmajority of 2007 strategic voting resulted from individuals who had voted outside the political mainstream at during the 2002 election and who decided to come into theside political mainstream during the 2007 election political mainstream: they caste a “vote utile” in favour of S. Royal or N. Sarkozy (Table 8). On the other hand, 2007 opinion-oriented voting appeared to be much more better explained by the the partyisan identification variable than byt the 2002 vote variable. In 2007, to be a radical-left wing party supporter increased wasup to 2.9 times% the chancesmore likely to be an opinion-voter. ExR supporters were only 2.2 times more likely to become opinion votersThose chances grew up to 20% in the case of FN supporters (Table 8).

Tableau 8Modeèl ofe de Political Alignment l’alignement politique : Estimations des paramètresof Parameters

STRATEGIC VOTERS

B Exp(BConstante -3,076

14 The model was based on three variables: age, gender, level of education. AThe age was the only significant variable which was significant. AsT a whole, the capacity of the model to explain strategic voting ands opinion voting was near to 0 ( Nagelkerke’s coefficient=.038). 15 Daniel Gluckstein, Arlette Laguiller, Olivier Besancenot, Marie Georges Buffet, Jean Pierre Chevènement, Christiane Taubira, Noël Mamère

14

Partisan proximityroximité partisaneExXLG] 1.,198** 3,313

GaucheLeft .,281 1,325DroiteRight .,314 1,369EXDExR ,849 2,338

Ni Gauche ni droiteNeither Left nor Right 0(b) .

VPresidential voteote présidentiel 2002 1st er tourround

Petit candidat de gaucheWeak left-wing candidate 1,131*** 3,100

Jospin -,164 ,848Chirac -,971 ,379Le Pen ,724 2,063

Petit candidat de droiteWeak right-wing candidate 0(b) .

Echelle/gauche droiteLeft/Right scale

LeftGauche] ,452 1,571DroiteRight -1,076*** ,341

Neither Left not RightNi gauche ni droite 0(b) .

OPINION VOTERS

B Exp(B)Constante -1,045

Partisan proximity roximité partisane

ExXLG 1,071*** 2,917GaucheLeft -,237 ,789DroiteRight -,679*** ,507

ExXRD ,789*** 2,201Ni gauche ni droiteNeither Left not

Right 0(b) .

Vote présidentielPresidential vote 2002 1st er tourround

Petit Weak left-wing candidate de gauche ,794*** 2,213

Jospin -,601*** ,548Chirac -,700*** ,497Le Pen -,273 ,761Weak right-wing candidatePetit

candidat de droite 0(b) .

Left/right scale Echelle/gauche droite

gaucheLeft [ ,140 1,150droiteRight -,954*** ,385Ni gauche ni droiteNeither Left not

Right 0(b) .

15

a base category= conformist votersb Ce This parameter was put back to zero as it is superfluousparamètre est remis à zéro parce qu'il est superflu.** probabilityotent < 0.05 ; *** potent probabilty < 0.01. Source: 2007 CEVIPOF Panel wave 1.

In order to increase the model’s explanatory the capacity, of explanation of this political model we added two others significant variables. One of these, age, was which were the more significant from the socio-cultural model – the age - and the other, the feeling of self efficiency, from the efficiency model.: the feeling of self efficiency16. Basically tThe new model based on five variables is not much significantly more usefulpotent thant the political model only based on three traditional political variables only: the pseudo R2-deux reached 0 .204 instead ofas opposed to 0 .192 in the former modelcase. By the way insideWithin the model, the power ability of political variables to explain the act of voting was strengthened.

Conclusion

Reaching At the end of this study, we whish would emphasise on three main conclusions concerningon the French voters facing their and presidential elections.The first conclusion is aboutconcerns the increasing mobility of the voters from onea national election to the next one since the middle of the eighties. Basically, this increasing tendency is spread present all overthroughout the European democracies mainly as a result of the weakness of the partisan alignments ( Dalton and Wattenberg, 2000; Le Duc, Niemi and Norris, 2002). In France, the multiparty system and the electoral formula based on a two- rounds majoritary system has encouraged a fractioningalization of the votes. As Like manya lot of former studies, this one confirms that the mobility of the voters mobility from onea side to anthe other of the left wing/right wing border divide –had grown up regularly increased to reach a level during the 2007 presidential election that stands at a little over three times the level in 1988at the 2007 presidential election an amount that is a little higher than three time its 1988 amount. However, another sort of electoral mobility has proved to be particularly damaging to hurt especially the government party candidates of government parties: : the increasing growing level of volatility inside within each coalition. This is as a a result of the increasing difficulties faced by the strong parties and their candidates meet nowadays to convince all their supporters that of the efficiency of the policies they proposed to in answer to the citizen’s voters’ claimsdemands are effective. TThe left- wing block has been especially injured damaged by this attitude of mistrust (Dupoirier 2008)17. The socialist party has found it increasingly had to face increasing difficulties to keep maintain its leadership all over all the left- wing parties as soonfrom as the end of the Mitterrand’s years. On the other hand, the right- wing government parties had less suffered less of from the volatility of their supporters until the 2007 presidential election when François Bayrou broke off with his the party he was formerly allied toformer partner (the RPR which then became part of the UMP with the UDF) and moved with his supporters further towards the centre ( neither left nor right- wings) of the political spacelandscape.

16 See the outcomes of the model in Annexe II.17(17) ? On the difficulties socialist candidates has met for years to force their legitimacy all over the left wing block see Dupoirier E., “Le parti socialiste et la gauche l’implacable spirale de l’échec”, in Perrineau P., Le vote de rupture, op cit.pp 145-174. More especially see the table 2 p148.

16

It would is to be expected that the new electoral law which links presidential and general elections for the same five- years term will incentive encourage mobile behaviours of theamong PS and UMP voters for two reasons. Firstly, five years without “mid-term elections”18 to listen to voters’ claims demands is a time long enough period to strengthen disappointment against with the coalition of parties that governs and to lead a significant part number of its supporters to electoral desertion at the end of the term. The second reason is that the rise of weak candidates makes these desertions very easy to practise. Even if Blais may have overestimated the visibility of all the weak candidates as a result of the French law that regulates the presidential campaigns and triesy to maintain equity among all the candidates19

with regard toof TV messages. However, between all the candidates20 it is true that a lot ofmany other favourable conditions could afford facilitate the desertion offrom the stronger candidates on during the first rounds of presidential elections.

Our second conclusion is aboutconcerns strategic voting. In additionTogether with with Blais’s study on the 2002 presidential election, the outcomes of the 2007 study highlighted that the amount level and the nature of strategic voting were strongly related to the specific electoral context of each election (Marsh, 2002). Blais pointed out the widespread forecast of a foregone conclusion of the first round (in favour of Jospin and Chirac) as the large choice between the first round candidates Blais cited two factors to explain the huge amount level (9.5%) of 2002 strategic voting in 2002 and its main orientationtendency in favour oftowards desertion of the two strongest candidates. The first of these was the widespread forecast of a foregone conclusion to the first round in favour of Jospin and Chirac. The second was the wide choice of first round candidates. However,But the results of the 2007 study do not seem to confirmwould suggest that Blais’ hypothesis Blais failed when he wrote that “the scenario that should induce the greatest amount of such strategic voting is a very close contest between three candidates in which none of the top three contenders is certain to advance to the second round”21. The 2007 presidential election was a good example of such a context without a forgone forecast conclusion as to who the top two candidates would be after the first round to know who offrom among Ségolène Royal, François Bayrou and Nicolas Sarkozy would be the two top candidates of the first round. On the other hand ofUnlike in Blais’ hypothesis, the global overall amount level of strategic voting was weakerst than thein former 2002 amount (4.2%) and was much more structured by desertion of weak candidates than desertion of the viable candidates. It would appeared that the significant variables which bestto frame the 2007 context would much more bewere rather the memory of Jospin’s defeat in 2002 Jospin defeat (for the left- wing electorate) and the attractiveness of Sarkozy’s campaign (for the rtight and extreme right- wings electorates). Both variables were at the core heart of the 2007 decrease of in strategic voting. Even if basically a two- rounds system would encourages much more candidates to run and would allowthus causing a much more fractionedalization of the vote than in a one- round system, those two conditions in favour of strategic voting could be neutralized bywhen the global political context of the election. encourage the voters to be 18 Understood in the American sense of the wordMid term elections as American practiced that kind of elections.. 19 Inside Within the time dedicated to the official TV and radio campaign, all the candidates benefit of the had the same amount of time to deliver their message. However,But those “official campaigns” are less and less watched by the public and it is difficult to maintain a balance in the broadcasting of campaign the abstracts of the campaign are more difficult to mautain well balanced between all the candidates despite the work of the “Conseil Supérieur de l’Audiovisuel” (CSA) which is the watchdog of the campaign equity of the campaign. 20 Inside the time dedicated to the official TV and radio campaign all the candidates benefit of the same time to deliver their message.But those “official campaigns” are less and less watched and the broadcasting of the abstracts of the campaign are more difficult to mautain well balanced between all the candidates despite the work of the “Conseil Supérieur de l’Audiovisuel” (CSA) which is the watchdog of the equity of the campaign. 21 Blais, 2002, p106.

17

outcome-oriented. ThisIt was the case during the on 2007 election when the outgoing President was out of the competition.

Our last conclusion is aboutconcerns the global overall influence of strategic voting on the outcomes of the French presidential elections. Concerning With regard to the 2002 election, Blais was obviously was right when he wrote that without such an amounta level of strategic voting (9.5% of the voters), Chirac and Jospin would have been the two top candidates of the first round and French voters would never have had to face the embarrassment of Le Pen on the second ballot. However, it seems just as obvious that strategic voting didn’t have anyhad no influence on the outcome of the 2007 presidential election. At the opposite ofUnlike strategic voting in 2002, strategic voting in strategic voting 2007 strategic voting was above all a “ vote utile” which meant the desertion of weak candidates in favour of viable ones as soon fromas the first round. Therefore, it cwould be argued that the voters had used the first round as the deciding round. As a result the second round would have beenwas no more than a confirmation round ( Dupoirier, 2008). Table 9 shows that the ddesertions from both the left and right-wing as from the right wing were equal had the same weak scale (13% and 14%) and thus neutralizinged each other. In addition and compared to former the 1988 and 1995 presidential elections, only a minority (25%) of Le Pen voters didn’t not transfer their vote towards Nicolas Sarkozy. Even if Bayrou’s voters were divided between the two challengers with a clear preference in favour of Ségolène Royal, (55% instead ofas against 45%) they were not numerous enough to help Royal to win the race.

Table 9Vote Transfers from First to SThe second- round Vvotes according to first round votes

( reconstruction of votes)

Vote au premier tour pourFirst round vote for:… Vote au second tour pourSecond round vote for :…Ségolène

RoyalNicolas Sarkozy

Total N

Weak left-wing candidates Petits candidats de gauche

87 13 100% 159

Ségolène Royal 96 4 100% 407François Bayrou 55 45 100% 275Nicolas Sarkozy 2 98 100% 511 Jean Marie Le Pen 25 75 100% 145OtherAutre weaks petits right-wing candidatescandidats de droite

14 86 100% 51

Ensemble des votantsAll voters 47 53 100% 15482007 CEVIPOF Panel 4 waves.

Focusing on the analysise on theof transferred votes towards Ségolène Royal, we argue that strategic voting could not have saved the left wing candidathere from the defeat. Firstly, because she had pocketed received her most significant portion of strategic voting as soon asin the first round. She clearly benefited from, taking advantage of the “vote utile” cast of theby supporters of the weak left- wing candidates and of by some Bayrou’s voters. Secondly, because as we have just written the “vote d’avertissement” that Royal could have expected for in the second round was too minuteweak to have any chance toof modifying the outcome of the race.: Aas a whole it weighted reached 7.5% of the whole second round electorate. Among

18

those 7.5% of voters who had voted strategically ion the first round, only 6% cast a vote in favour of Ségolène Royal ion the second round. Most of them were Bayrou’s voters supporters who had Royal as preferred candidate beforeat the first round. They weighted accounted for only 10% of all the strategic voters.s…

***

Obviously the presidential elections in 2002 and 2007 presidential elections gaveprovided two opposite examples of how the strategic voting mightay influence the outcome of a French presidential election. There is no doubt that further experimentationtions would be useful before drawing up a global pattern describing the influence of strategic voting on the outcomes of presidential elections.

Bibliography: - Blais A., Nadeau R., Gidengil E., Nevitte N., « Measuring Strategic Voting in Multiparty Plurality Elections », Electoral Studies, n°20 (2001), pp 343-352. “- Blais A. and Massicotte L., “Electoral systems”, in Le Duc L., Richard G., Niemi and Norris P. (eds) Comparing Democraties2. New Challenges in the Study of Elections and Voting, London, Sage, 2002.- Blais A., “Srategic Voting in the 2002 French Presidential Election”, in Lewis Beck M., The French Voter before and after the 2002 elections, Palgrave, 2006.- Boy D. and Dupoirier E., « Is the Voter Strategist? », Daniel Boy et Nonna Mayer (eds), The French Voter Decides, Ann Arbor, University of Michigan Press, 1993.- Boy D., Mayer N. Swyngedouw M., “Mesure de la volatilité électorale en France : 1993-1997 », Revue Française de Science Politique, (200-06), vol 50, n°3, p489-514. - Boy (Daniel),Chiche (Jean),"Image des candidats et trajectoires électorales" dans Bruno Cautres et Anne Muxel : /Comment les électeurs font leur choix ? Le panel électoral Français de 2007/ , Paris, Presses de Sc Po, à paraître avril 2009

-Chiche J., Haegel F., « Les connaissances politiques », Grunberg G., Mayer N. , Sniderman P.(dir.), La Démocratie à l’épreuve, Paris, Presses de Sciences Po, 2002.

- Chiche J., Jaffré J., “Mobilité, volatilité perplexité » in Boy D. and Mayer N.(dir.), L’électeur a ses raisons, Paris, Presses de la FNSP, 1997.- Clark T. and Lipset S.(eds), 2001, The breakdown of class politics A debate on post industrial stratification, John Hopkins University Press - Converse P., Pierce R., Political representation in France, Cambridge, Harvard University Press, 1986. - Cox G.W. Making Votes Count; Strategic coordination in the World’s Electoral Systems, Cambridge, Cambridge University Press, 1997.- Crew I. Denver D., Electoral Change in Western Democracies, London, Croom Helm, 1985.- Dalton R.J. and Wattenberg, M.P. (Eds.), Parties without Partisans Political Change in Advanced Industrial Democracies, Oxford University Press, 2000.- Dupoirier E. « Chassés croisés électoraux », in Dupoirier E., Grunberg G. (eddir), La drôle de défaite de la gauche les élections législatives de mars 1986, Paris, PUF, p157-188.- Dupoirier E., «  La commission des sondages et la campagne électorale de 2007 »,

19

- Dupoirier E., “Le parti socialiste et la gauche l’implacable spirale de l’échec”, in Perrineau P., Le vote de rupture, les élections présidentielle et législatives d’avril-juin 2007, Paris, Presses de Sciences Po, Chroniques électorales.- Grunberg G.”L’instabilité du comportement électoral” in Gaxie D.(dir), L’explication du vote, Paris, Presses de la Fondation Nationale de Sciences Politiques, 1984.- Le Duc L., Niemi R., Norris P. (ed), Comparing Democracies 2 New challenges in the study of elections and voting, London, Sage Publications, 2002.- Marsh M., “Electoral context”, Electoral Studies, n°21, 2002, pp. 207-217. - Perrineau P. (eddir), Le vote de rupture les élections présidentielle et législatives d’avril-juin 2007, Paris, Presses de Sciences Po, Chroniques électorales.

20

ANNEXE

Annexe 1Typology of 2007 first round voters

ConformistVoting

OpinionVoting

StrategicVoting 1

“vote utile”

Strategic voting 2Vote “d’avertissement”

Voting Intention vote First round

SarkozyRoyal

Bayrou

Schivardi, Besancenot, Laguiller,

Buffet, Bové, Voinet, Nihous,

Bayrou,de Villiers, Le Pen

SarkozyRoyal

Schivardi, Besancenot, Laguiller,

Buffet, Bové, Voinet, Nihous, Bayrou,

de Villiers, Le Pen

Who would you like as the winnerto win ?

Sarkozy Royal

Bayrou

Schivardi, Besancenot, Laguiller,

Buffet, Bové, Voinet, Nihous,

Bayrou,de Villiers, Le Pen

Schivardi, Besancenot, Laguiller,

Buffet, Bové, Voinet, Nihous,

Bayrou,de Villiers, Le Pen

Sarkozy Royal

Probability of success for the likely winner

high weak Weak high

Voting IIntention vote second round

SarkozyRoyal

The candidate who has the best proposals to solve the main issue

Schivardi, Besancenot, Laguiller,

Buffet, Bové, Voinet, Nihous, Bayrou,

de Villiers, Le PenRespondents 2007

3.1 % of the whole electorate

1.1% of the whole electorate

21