CAIRN-INT.INFO : International Edition
“Le vote mutile.” [2]


1International studies have generally concluded that there has been a decline in electoral participation in most Western democracies. In particular, this is the conclusion reached recently by Pascal Delwit [3] based on an investigation covering thirty-five countries and 402 elections. In France, there is concern that the country is evolving into an “abstention democracy.” [4] Abstention at the individual level may signify a lack of social integration [5] or individuals’ lack of political competence, [6] while at the collective level, abstention may lead to a strengthening of social inequalities when it comes to people’s having a political voice. [7] Yet there remains a paradox. Citizens have never had more formal qualifications, [8] and in France and elsewhere, acquiring information through the media is becoming easier and easier. This ought to have led to a kind of “supercitizen,” [9] but abstention has nevertheless gained ground. Several studies point to generational renewal as a possible explanation. [10]

2The objective here is to verify and in particular to demonstrate that, behind generational renewal, there has been a decentering of voting in terms of citizens’ attitudes toward politics. This decentering has translated into a rise in intermittent abstention (or voting). Older cohorts may be characterized by a culture of deference, while the younger ones might be more characterized by a remote form of citizenship. [11] The first of these outlooks may consist in trusting and submitting oneself [12] to parties and the political class, as well as in voting based on a sense of duty. [13] High levels of participation in elections would therefore not necessarily signify a strong interest in public affairs but rather an allegiance to the political system and to representative democracy. By contrast, the second outlook is characterized by a certain criticism of politics as it is conducted and in particular of the classical actors who embody it. The act of voting has therefore apparently lost its centrality.

3I consider there to be two phenomena at work behind this decentering. In one case, there is an intermittent abstention associated with social inequalities that has caused those with few formal qualifications and members of working-class categories to head to the ballot box less and less. The more recently members of these categories were born, the more likely they are to abstain in some elections. This development can also be explained by a decline in microincentives to vote among these social sectors. The other case is a matter of citizens for whom voting has become less important than other forms of politicization and participation. For these individuals, voting is no longer enough. There are overlaps here with the political theory of acts of citizenship, [14] according to which individuals mobilize forms of actions and engagement that differ from the traditional forms of conventional political participation, which are deemed too restrictive.

4To demonstrate this evolution, in this article I propose to conduct analyses that are based on a dataset comprising the “participation” surveys of the Institut national de la statistique et des études économiques (INSEE) (French National Institute of Statistics and Economic Studies) covering the period 2002–2017, to highlight the evolution of abstention through the prism of generations and based on the DYNAMOB (Dynamics of Mobilization) panel, which makes it possible to follow the same individuals from September 2013 and therefore to cover their electoral behavior from France’s 2014 municipal elections to the 2017 legislative elections. Surveys on electoral participation allow the best possible assessment of the sociological logics behind abstention, particularly because they measure individuals’ actual behavior. However, they offer no variable that allows political factors to be brought into the equation; as will be seen, these factors are essential. The DYNAMOB panel does not feature the full sociological wealth that participation surveys do, but it is relevant here for three reasons. First, it covers all types of election, which allows intermittent abstention to be taken into account, as is the case with participation surveys. Second, it goes beyond presidential elections, which, as will be seen, constitute an exception in terms of political participation. Finally, it was designed to capture the panelists’ attitudes toward politics (and therefore it includes the political factors required for analytical purposes) and to compare different campaign dynamics, in particular by bringing the micropressures that panelists may face to the fore. It therefore allows the collective dimension of voting to be taken into account.

Abstention: A matter of elections

5Under France’s Fifth Republic, participating in the various elections has continued to be a widespread behavior and has generally involved a majority of voters, even in the twenty-first century. An exception here is European elections, which have historically been the elections that motivate voters the least. Besides them, the 2012 regional elections and the 2017 legislative elections are the only ones in which abstention has passed the 50% mark.

Figure 1[15]

Electoral abstention during the Fifth Republic

Figure 115

Electoral abstention during the Fifth Republic

Sources: official results, Ministry of the Interior (compiled by the author).

6Abstention increasingly depends on the type of election. To prove this, I have calculated the maximum variations in participation over presidential terms relative to the three elections falling within the whole period. The maximum variation in participation was 16 points between 1958 and 1965 (mainly because of the legislative elections of 1962). However, between 1965 and 1969, it stagnated at 7 points, and at 9 points between 1969 and 1974. Between 1969 and 2002, the maximum variations ranged between 13 and 15 points. From 2002, they stood at around 23 points. They reached 29 points in 2017. If we consider the same indicator but apply it this time to all elections that took place from the 1981–1986 period, variations in participation went from 24.5 points to 43 points for the period beginning in 2007.

7The electoral context plays more and more of a role in whether people vote or abstain, and it can lead to people’s returning to the ballot box, sometimes on a large scale. This is the case for presidential elections in general, but also for the regional elections of 2004, for example. It therefore cannot simply be said that abstention is making headway in France; it depends to a very large extent on the election. [16]

8Classical approaches that rely on sociopolitical inequalities or social integration to explain turnout are not sufficient to explain why an individual votes in one election and abstains in another while that person’s level of formal qualifications or occupation remains constant. By contrast, the increasing power of context seems more in line with the “civic voluntarism model” put forward by Sidney Verba, Kay Lehman Schlozman, and Henry E. Brady. [17] They consider that three conditions must be fulfilled for individuals to participate: 1) citizens must be able to fulfill their role (which raises the question of inequalities in political competence); 2) they must be willing to do so (out of interest, engagement, or taste); and 3) they must be asked to do so.

9The second condition relates to my hypothesis of a changing attitude toward voting. In the case of duty-based voting, inclination to vote remains stable from one election to another because individuals vote out of a moral obligation (toward their elders or society), something that seems to correspond to the elections of the 1960s and 1970s. But if whether a person votes is conditioned by the interest that he or she takes in this act, it can be postulated that voting for a president holds more interest than voting for members of the European Parliament does, in view of their respective weights in public decision-making. The third condition is also essential insofar as it brings into the equation the role of political and media actors as well as society in the development of incentives to participate. From above, it is known that politics and journalism professionals’ mobilization is not the same in a presidential election as it is in a European one, and this will have an impact on the call to the ballot box made to voters. From below, “electoral buzz” [18] also differs: the likelihood that a voter will be encouraged to “go out and vote” by his or her work colleagues, friends, or family will vary in strength depending on whether the election is of a “high” or “low” intensity. [19] Finally, distinguishing between conditions two and three is essential insofar as they allow the personal and collective dimensions of political participation to be linked together. Through doing so, the “atomistic fallacy” that is often found in electoral sociology research is avoided (or at least reduced). [20]

Generational variations in abstention

10In France, electoral participation has remained stable for presidential elections, but it has become more and more varied from one type of election to another. In my view, this variation is driven by generational renewal. To demonstrate this, I will turn to INSEE’s participation surveys.

Participation surveys: Presentation and assessment of the results

11These surveys have acquired a central place in measuring and explaining attitudes toward voting in France. [21] They have four major advantages. First, they are based on INSEE’s Permanent Demographic Sample and therefore cover a range of social situations that is much broader than that of conventional surveys, which struggle to cover the whole of the social spectrum, and in particular the poorest citizens. [22] Second, they identify actual participation behaviors, duly recorded by INSEE officers based on voting lists and not on statements that can be affected by social desirability or erroneous reconstructions of behaviors, quite common phenomena in many sociopolitical surveys in France and elsewhere. Third, the size of the sample is particularly large (for example, over 39,000 people were surveyed in 2007–2008). Fourth, they take the form of a panel. This is important because at a minimum it allows us to follow the behavior of individuals over four presidential and legislative ballots, but also in the case of the 2007 survey, the municipal elections of 2008, and in that of the 2002 survey, the 2004 regional and European elections. Attitudes toward voting can therefore be tested across different types of elections.

12Several major sets of results from past exploitations of these surveys are worth recalling. For example, François Héran has shown that attitudes toward voting cannot be reduced to an alternative between systematic voting and systematic abstention. [23] In 1995, he identified 11% of the population as constant abstainers, but also only 54% as constant voters and 35% as intermittent voters. In 2002, the share of constant voters declined (to 47%) and that of intermittent voters rose (to 40%). In 2017, [24] intermittent voters were the majority group (51%), whereas the constant voters only accounted for 36% (down by 18 points relative to 1995). The figure for constant abstainers stagnated at around 13% of the electorate. These surveys confirm the relevance of conducting analysis according to social inequalities. [25] Systematic abstention concerns those with no higher-education qualification more so than university-educated citizens (24% in the case of the former and 8% in that of the latter), and it is less frequent among executives than it is among manual workers (in 2012, 72% of executives voted in the first round of the legislative elections, versus 61% of white-collar workers and 57% of manual workers, for example). It is also more often the case for unemployed or single people, something that brings to mind Alain Lancelot’s social integration approach. [26] Finally, the link between age and electoral participation can be confirmed. Abstention tends to peak during a person’s younger years, and it returns from the age of 75. Youth corresponds to a period of political moratorium. [27] Just as individuals learn to become parents, workers, or neighbors, they discover themselves as citizens when faced with voting and alternatives in terms of political expression. Old age, meanwhile, entails an effect of progressive distancing from the social world, though biological aging also has an impact (diseases, mobility difficulties, dependence, and so forth).

Alongside classical effects, the impact of cohorts

13If one considers the behavior of cohorts, [28] very often one’s first instinct is to focus on the effects of age or life cycles. Indeed, the distribution of abstention follows a U-shaped curve for all elections with maximums among the youngest and the oldest, while civic mobilization is highest among voters born between 1938 and 1967. In the cohort born in 1927 or before, withdrawal from the ballot box became increasingly significant between 2002 and 2017. Abstention was already significant in 2002 (35% on April 21, 2002, and 29% on May 5, 2002), but it exceeded 40% in the two rounds of the presidential election in 2012 and rose to 67% on April 23, 2017. These levels and their evolution concern the whole set of elections, in particular legislative elections (from 42% in 2002 to 70% in 2017). We can conclude that aging has had an effect on this cohort, whose youngest members were 75 in 2002 and 89 in 2017. Moreover, a similar phenomenon can be discerned in the case of the 1928–1937 generation for the 2012 and 2017 presidential elections: a third of them abstained in the 2017 presidential election, compared with, for example, 13% in 2007. At the other end of the spectrum, we find the most recent cohorts. The members of the 1978–1987 cohort were particularly likely to abstain in the last four legislative elections, the 2004 regional elections, and the 2008 municipal elections. This is even more true of the cohort born in 1988 and after; 71% of them abstained in the 2017 legislative elections, for example. Abstention is also notable among the 1968–1977 cohort in legislative, regional, and municipal elections. However, two additional observations can be made: 1) Participation among voters born after 1978 remains strong when it comes to presidential elections; 2) The behavior of the 1968–1977 cohort in presidential elections is very similar to that of individuals born between 1938 and 1967.

14Most analyses of abstention do not go beyond life cycle effects. However, such a reading is far from sufficient if we are to understand variations in abstention. For example, as the 1968–1977 cohort’s voting behavior depends on the type of election, it is difficult to understand its behavior according to the theory of a political moratorium alone. This cohort’s lack of mobilization was particularly pronounced in April 2002 (34% abstention). However, at the same time it proved to be much more assiduous in heading to the polls for subsequent presidential elections (between 14% and 16% abstention) and much less likely to participate in the other elections (between 40% and 50% abstention). Yet this variability occurred even though the individuals concerned were aged between 40 and 49 years in 2017. In other words, even though their years in study were lengthened and the boundary between youth and adulthood blurred, [29] they had reached a social age at which their career paths had largely been established and, in many cases, they had married, had children, and owned their home. In addition, the 1958–1967 cohort presents behavioral similarities to that which immediately follows it.

Figure 2

Abstention by cohorts (2002–2017)

Figure 2

Abstention by cohorts (2002–2017)

Sources: INSEE participation surveys, 2002, 2007, 2012, and 2017; author’s own calculations.
Figure 3

Variations in abstention between presidential elections and other elections (first round)

Figure 3

Variations in abstention between presidential elections and other elections (first round)

Sources: INSEE participation surveys, 2002, 2007, 2012, and 2017; author’s own calculations.

15When the data are presented in a different way, we obtain confirmation that the more recent a cohort is, the more its behaviors vary between presidential elections and other sorts of election. Members of the oldest cohort participate less and less, but when they do, they do so in a very constant manner. [30] This relative constancy in attitudes toward voting can be found in the 1928–1937 and 1938–1947 cohorts. For these three generations, the maximum variation between two types of election is 16 points.

16The variations are much more substantial for later cohorts. In the case of the 1968–1977 cohort, variations in abstention between the presidential elections and legislative elections reached 33.5 points in 2017. [31] In particular, this cohort’s behavior is not what one would expect based on the classical effects of the life cycle. There is no “stabilizing” in attitudes toward voting with age: their voting was less intermittent in 2002 between the legislative elections and the presidential elections (11%), at a time when these voters were aged between 25 and 34, than it was in 2017 (33.5%), when they were between 40 and 49. Similar phenomena can be found among the 1958–1967 cohort, although more moderately so (27% in 2017). The 1978–1987 cohort exhibits even stronger variations (39% in 2017), and its members are now beyond the moratorium period, as they were between 30 and 39 years during this election. The cohort born in 1988 or after seems to follow the same path in terms of voting intermittence, but it is necessary to be careful here, because some voters are still in the political moratorium phase. It should be recalled that there has been an increase in intermittent voting among these cohorts, but clearly no growth in systematic abstention. They are therefore much more sensitive to the context and the nature of elections than their predecessors are.

17If these differences in behavior are the result of a cohort effect, their consequences will in my view be enormous in the future, since the “stabler” generations will be progressively replaced by these “conjunctural” cohorts.

“No one was left untouched”: Interactions between cohorts, sociological variables, and abstention

18The link between cohorts and voting also redefines the classical effects of other variables that measure social inequalities. In a recent article in this journal, [32] Céline Braconnier, Baptiste Coulmont, and Jean-Yves Dormagen offer a reminder of the importance of qualifications and occupations, particularly in relation to legislative elections. However, we are confronted with an apparent paradox. The most recent cohorts are also the most highly qualified and have the most executives among them. Yet they are also the most intermittent voters. By contrast, the cohorts that participate the most in elections are the least formally qualified [33] and contain significantly fewer members of upper-level professions; [34] this ought to signify a lower level of electoral participation. The explanation for why this is relates to an interaction effect between cohorts, qualifications (or occupations), and voting. As we will see, not only is intermittent abstention progressing among the whole set of more recent cohorts, regardless of whether or not individuals are highly qualified or whether they belong to working-class or upper-level categories, but inequalities are rising within these cohorts too.

Figure 4

Abstention in presidential and legislative elections according to cohort and qualification level

Figure 4

Abstention in presidential and legislative elections according to cohort and qualification level

Figure 5

Abstention in presidential and legislative elections according to cohort and occupation

Figure 5

Abstention in presidential and legislative elections according to cohort and occupation

Sources: INSEE participation surveys, 2002, 2007, 2012, and 2017; author’s own calculations. [35] In terms of qualifications, I compared secondary-level qualifications and higher-education qualifications (denoted by “upp.”). In the case of occupations, I have grouped together executives, those in upper-level intellectual professions, and those in intermediate professions (denoted by “upp. and int.”) and compared them to manual and white-collar workers (denoted by “low.”).

19In the case of each cohort, it can be confirmed that the greater the number of individuals who hold a secondary-level qualification rather than a university one or are white-collar or manual workers rather than executives or in an intermediate profession, the higher the proportion of abstainers. In accordance with the results obtained by Braconnier, Coulmont, and Dormagen, variations tend to increase for legislative elections relative to presidential ones. Among the 1958–1967 cohort, 15% of those with secondary-level qualifications abstained in the first round of the 2017 presidential election, versus 11% of those who held university degrees: in the first round of the legislative elections, the figures for these two segments of the cohort were respectively 46% and 31%. In 2017, just 12% of executives and those in intermediate professions within this same cohort abstained in the presidential election, and 33% did so in the legislative elections; the equivalent figures were 18% and 47% for its white-collar and manual workers.

20However, the variations between cohorts are equally, if not more, significant than those produced by qualifications or occupations. Let us take the example of the 1958–1967 and 1968–1977 cohorts. In the 2017 presidential election, 11% of those with university qualifications who were born between 1958 and 1967 abstained, whereas 18% of those who held secondary-level qualifications and were born between 1968 and 1977 did so, making for a variation of 7 points. In the 2017 legislative elections, the variations quadrupled, and the respective abstention rates of the same two groups were 31% and 58%. More interestingly still, holders of secondary-level qualifications from the 1948–1957 cohort take part in elections “as though” they were holders of university qualifications from the 1968–1977 cohort. The same pattern is reproduced when one focuses on occupation: executives and members of intermediate professions from the 1968–1977 cohort abstain as much as white-collar and manual workers from the 1958–1967 cohort do.

21The variations are much less pronounced within the older cohorts. Let us focus on the 1938–1947 and 1948–1957 cohorts, once again in the 2017 elections. In the presidential election, abstention was between 8% and 10%, depending on qualification level. In the legislative elections, abstention reached 22% for holders of higher-education qualifications from the 1938–1947 cohort and 37% for holders of secondary-level qualifications from the 1948–1957 cohort.

22Civic norms undoubtedly remain significant among the older cohorts, regardless of education level or occupation. They make individuals who do not have the “resources” [36] that qualifications or occupations provide nevertheless head to the ballot box. On the other hand, in the younger cohorts, it seems likely that civic norms play less and less of a role, which causes a disconnection from voting that becomes especially strong during legislative elections: 68% of manual and white-collar workers from the 1978–1987 cohort abstained in June 2017, for example.

23The cohort that a person belongs to therefore has a significant impact on his or her attitude toward voting. My goal is to confirm this via classical logistic regressions on a dataset that combines the four periods analyzed. Cohorts will be considered against two life-cycle effects: political moratorium and old age. This will be possible because electoral participation will simultaneously be addressed at four moments in time. [37] The construction of the moratorium variable assumes that it will have a different effect depending on whether the individual concerned was between 18 and 24 years, between 25 and 29, or over 29. In the case of old age, I have distinguished between those aged under 75, those aged between 75 and 79, those aged between 80 and 84, and those aged over 84. Participation begins to decline at the age of about 75. The other independent variables are the “usual suspects” from research on electoral participation: sex and qualifications. [38]

24I propose to confirm the impact of cohorts in relation to three behaviors: voting or abstaining in presidential elections; voting or abstaining in legislative elections; and voting intermittence between these pairs of elections. The analysis strategy follows the logic of nested models: I began by testing the explanatory logics behind participation by using the classical variables (models 1 to 3), and then I introduced birth cohorts to measure their impact (models 1b to 3b).

Table 1

The sociological logics behind electoral participation (2002–2017)***,**,*

Table 1
Presidential elections, first round Legislative elections, first round Legislative elections, first round Presidential elections, first round Legislative elections, first round Legislative elections, first round 1 2 3 1b 2b 3b Female 1.04*** .98 .97** .97* 1.03** .96*** Male Moratorium Under 25 years .42*** .33*** .38*** .79*** .74*** .85*** 25–29 years .52*** .38*** .41*** 1.16*** 1.08** 1.15*** Over 29 years Old age 75–79 years .97 1.53*** 1.70*** .87*** .71*** .96 80–84 years .66*** 1.23*** 1.56*** .69*** .47*** .88** Over 84 years .29*** .57*** .95* .33*** .21*** .54*** Under 75 years Qualification level None/primary .68*** .88*** 1.00 .73*** .62*** .82*** Baccalaureate 1.09*** 1.08*** 1.06*** 1.17*** 1.13*** 1.15*** Higher education 1.36*** 1.39*** 1.33*** 1.71*** 1.51*** 1.64*** No response/ongoing .52*** .60*** .72*** .62*** .53*** .74*** Secondary Vote in the presidential election Yes 10.71*** 10.53*** No
Table 1
Cohort 1927 and earlier 1.02 1.04 1.05 1928–1937 1.16*** 1.10*** 1.15*** 1948–1957 .76*** .87*** .76*** 1958–1967 .53*** .72*** .53*** 1968–1977 .33*** .54*** .34*** 1978–1987 .21*** .40*** .22*** 1988 and after .15*** .29*** .17*** 1938–1947 Year 2007 2.01*** .79*** .55*** .83*** 2.10*** .58*** 2012 1.68*** .72*** .53*** .91*** 1.99*** .66*** 2017 1.11*** .44*** .34*** .61*** 1.40*** .46*** 2002 Constant 3.99*** 2.42*** .41*** 3.69*** 4.89*** .64*** R2 5% 4% 16% 7% 6% 18% N 162,298 162,298 162,298 162,298 162,298 162,298

The sociological logics behind electoral participation (2002–2017)***,**,*

*** Means that the coefficient is significant at p < 0.01;
** at p < 0.05; and
* at p < 0.10.
Sources: 2002, 2007, and 2012 INSEE participation surveys.

25Models 1, 2, and 3 confirm that sociological inequalities persist when it comes to attitudes toward voting, with the exception of sex. [39] Women continue to take less of an interest in politics and feel less as though they have legitimacy in expressing political opinions than men do, but it is more likely that they will participate in presidential and legislative elections. The fewer qualifications one has, the less likely it is that one will take part. The logics behind voting in the legislative elections and those behind voting in presidential ones often intersect, even if inequalities are growing in legislative elections. [40] Finally, the age effects are precisely those that one would expect during presidential elections. In legislative elections, the impact of political moratorium is increasing, while those aged between 75 and 84 unexpectedly appear to be more likely to vote in them than are the economically active age classes. Finally, voting intermittence is clearly linked to social inequalities. Ceteris paribus, those with few formal qualifications are more likely to disengage from the election cycle, even if they had voted in the presidential election.

26Cohorts clearly have a significant impact. The more recent the cohort, the more likely it is that its members will not vote, regardless of life-cycle effects. This phenomenon appears as early as the 1958–1967 cohort: compared to voters in the 1938–1947 cohort, its members are 2 times more likely to abstain in presidential elections, 1.4 times more likely to abstain in legislative ones, and 2 times more likely to disengage between the presidential elections and the legislative ones. The variations increase with subsequent cohorts: the 1968–1977 cohort is 3 times more likely to abstain in the presidential elections and 2 times more likely to do so in the legislative ones. The 1978–1987 cohort is 5 times more likely to do so in the presidential elections and 2.5 times more likely in the legislative ones. On the other hand, the older cohorts continue to have a significant attachment to voting, even though some of their members are subject to the effects of old age. This is particularly visible in relation to legislative elections or to systematic voting. [41]

Another relationship toward voting

27How can we make sense of these differences in behavior between cohorts? Sociological explanations are not enough if we want to understand why voters who head to the ballot box differ so much in number and composition between two elections. Why do qualifications have different effects depending on whether voters were born in the 1940s or in the 1970s? Why have manual and white-collar workers born in the 1960s continued to visit the voting booth more frequently, while a majority of those born twenty years later have abstained? Part of the explanation likely relates to changes in the world of work and the decline of trade unionism, but this alone is not sufficient. Sociological explanations do not explain why members of young cohorts head to the polls for some elections but not for others. It is necessary to take into account individuals’ attitudes to politics and the context of the elections. To do so, it is necessary to mobilize sociopolitical research conducted via surveys, with all their defaults, in particular in terms of sampling and declarations of abstention.

28Two avenues of reflection may be heuristic, and I postulate that they complement one another more so than they conflict. They will apply differently depending on the individuals, their cohorts, and their place in society.

29A first avenue is based on the arguments advanced by a recent current in the political sociology of abstention that brings to the fore the micropressures that individuals may face. This idea can be found in the work of Céline Braconnier and Meredith Rolfe, as well as, more recently, the work by the Sociologie politique des élections (SPEL) (Political Sociology of Elections) collective focused on the notion of electoral buzz. [42] If voters go to the polls in “high-intensity” elections such as France’s presidential elections, they do so in part because mobilization also takes place from below. Families, friends, neighbors, and work colleagues are some of the agents who contribute to people’s heading to the ballot box. If these individual networks did not carry reverberations of the campaigns, and if civic norms were not revived and strengthened through them, some individuals would likely not vote in presidential elections. These mobilizations from below do not occur at every election. The decline in participation may therefore be the result of a lack of collective mobilization and cannot be imputed to voters alone. These mobilization effects add to the weight of social inequalities or of taste for politics. An individual who does not have the resources and capacity to participate, or who does not have the appetite for it, may be mobilized by those around him or her if those people are themselves mobilized—for example, through prompts about civic norms. On the other hand, there is a likelihood that an individual who has neither the resources, nor the taste, nor the network will stay at home—and it will be a large one, as it applies above all to the least qualified and working-class categories.

30The second avenue is the idea that we are witnessing a change of attitude toward voting that has been brought about by generational renewal. Ronald Inglehart [43] considers voting to be a means of participation that is particularly in tune with an “elite-directed” political participation in which elected representatives are the drivers and citizens are confined to a peripheral role: people vote according to the timetable and the rules laid down by the constitution, with candidates chosen from above, and once the election is over, the elected representatives govern and citizens return to being spectators. Inglehart emphasizes the existence since the 1980s of a shift toward “elite-challenging participation,” in which voting has a reduced role. Russell J. Dalton extends this reflection by proposing that through generational renewal there has been a move from a culture founded on a “duty-based vote” to a culture of political expression particularly borne by new voters. [44]

31However, it is difficult to subscribe fully to Dalton’s particularly optimistic distinction. Recent generations certainly demonstrate and petition more, but they are also characterized by a remote form of citizenship. [45] Their members are not necessarily indifferent to public affairs, but they are more distrustful of political leaders and representative democracy, and their relationship with parties has been stretched very thin. In this framework, we can suppose that their attitude toward voting is decentered. For these citizens, voting is at best one means among others for political action, and not necessarily the most effective one when it comes to making oneself heard. Their relationship with this act can be characterized more by right-based voting: I’ll vote if I’m interested.

32In this framework, we are beyond the “abstaining in the game” that Anne Muxel had observed from the 1997 legislative elections. She defines this as the result of “a dissatisfaction with the electoral offering … [and] a visible symptom of the crisis of political representation. Although this crisis has persisted for at least twenty years, it is circumstantial and periodic.” [46] I would suggest that in the case of the post-baby-boomer generations, the act of voting no longer has the importance or normative weight that it had for the previous generations. Their abstention is therefore not simply a refusal of the choice that is offered to them at a given time; it has become more mundane than that. These voters have likely already abstained several times, and they likely do not attach much importance to this behavior. However, the consequences of this change in attitude toward voting will differ based on whether or not individuals remain enthusiastic about public affairs. In one case, voting will be only one of the elements in their repertoire of action, and not necessarily the most effective when it comes to making themselves heard. If they do not express themselves in the ballot box, they will do so through demonstrations, petitions, and social networks. For others, all this will lead to their no longer being present or audible, which might strengthen political inequalities.

33In summary, I make three hypotheses: 1) The voter does not vote alone, and part of the differences between cohorts relates to the effects of the people around him or her. 2) The older generations vote, even when they are not interested in politics, out of a sense of duty and submitting oneself, while more recent cohorts vote intermittently, regardless of their interest in public matters, and they do not attribute to voting the centrality that it has for older cohorts. 3) This change in attitude toward voting goes beyond the years of youth and will become a permanent feature of these generations’ attitudes toward voting, and therefore cannot be interpreted as a conjunctural effect.

34To test these hypotheses, I will use the DYNAMOB panel, in which the same individuals were questioned between September 2013 and December 2017. The panel was coordinated by Florent Gougou and Vincent Tiberj and is based on the ELIPSS Equipex. [47] This is the only survey that makes it possible to verify attitudes toward voting for the whole set of elections, unlike the French Electoral Study, which focuses only on the presidential elections, or the ENEF panel, which covers regional, presidential, and legislative elections and whose data are not available.

35In addition, use was made in the DYNAMOB panel of an indicator that is particularly important in taking into account the collective dimension of voting. For each election in the 2014–2017 period, panelists were asked if they had listened to or watched programs, read leaflets, met with candidates, and above all if they had discussed the election, as well as whether they had done so often, from time to time, or never. Certain elections understandably elicit more discussion than others. In total, 43% of the panelists discussed the presidential elections often, whereas 7% never discussed them. No more than 34% discussed the municipal elections often (versus 10% who never did so), and 20% often discussed the European or departmental elections (versus 20% who never did so). Of course, the frequency of political discussions is linked to interest in politics, but one does not perfectly predict the other. For example, among the panelists interested in politics, 11% never discussed the European elections with their close friends and relatives (versus 32% of those who said they were not interested). But 58% of those who were not interested in politics discussed them from time to time and 8% did so often, versus figures of 60% and 28% of those who were interested. In other words, discussion as a variable reflects individuals’ willingness, based on their taste, to talk about politics, but it also encompasses a dimension that relates to the networks that individuals are part of. In this respect, this variable can indirectly reflect the micropressures that panelists faced, as well as the media and political buzz surrounding elections. This makes the use of the DYNAMOB panel all the more interesting.

Introduction to the ELIPSS panel

This survey was conducted over the Internet, which removes the effects of social desirability produced by the interviewer’s presence. ELIPSS is an instrument that reduces drop-out in panelists: after individuals are randomly selected by INSEE, they are invited to participate in the instrument, and they are provided with a tablet and an Internet subscription; all that they have to do in return is spend thirty minutes each month responding to questions. The DYNAMOB team has therefore been able to requestion the same individuals between September 2013 and December 2017 over nineteen waves of surveys. From the outset, the panel had as its specific goal to better understand voting by covering the whole set of elections that were held over its period of focus (especially through preelection and postelection waves).
Table 2

Characteristics of the DYNAMOB panel’s electoral waves

WaveDateNumber of panelistsElectionsGross declared participation
4March/April 2014723Municipal77%
6June 2014696European62%
9March/April 2015655Departmental60%
12December 2015610Regional71%
16May 2017525Presidential91%
17June/July 2017563Legislative70%

Characteristics of the DYNAMOB panel’s electoral waves

Despite the instrument’s advantages, panelists’ declared participation is substantially higher than that of the electorate. This may be due to several phenomena. A selection bias that produces a sample of panelists who are characterized in particular by a high qualification level can be observed. This would be a problem if these data were used within a logic focused on the sample’s representativeness of the population, but what we are interested in here is comparing how voters behave during different elections and the explanatory logics behind these behaviors. Participation can also be affected by “panel conditioning.” [48] Individuals who are regularly exposed to political questions modify their behaviors. They pay more attention to politics, current political events, and its actors, than they did before they became involved in the research instrument. Thus, panelists become more likely to be mobilizable and mobilized by the different campaigns. It should also be noted that loss of panelists is still a significant phenomenon. This makes it necessary to adopt the least costly analytical strategies in terms of participant numbers. In order to limit “statistical noise” that might arise from problems with participant numbers, the decision was made to radically recode the independent variables. This meant measuring only two levels of interest or qualification and taking year of birth as a continuous variable. The choice was also made to only present here the predicted probabilities originating from the logistic regression models and not the raw data.

The DYNAMOB panel’s voters during elections

36The aim of the regression models is to make it possible to understand the logics behind abstention in the first rounds of municipal, departmental, regional, presidential, legislative, and European elections. The independent variables are as follows: the variable birth year, a moratorium variable (which distinguishes the panelists aged under 25 in 2013); sex; qualification; interest in politics (two modalities); and frequency of discussions about the election during the campaign (three modalities). [49]

Table 3

Modeling of abstention declared in the DYNAMOB panel***,**,*

Table 3
Municipal elections European elections Departmental elections Regional elections Presidential elections Legislative elections Year of birth 1.01 1.02*** 1.03*** 1.02* 1.04*** 1.04*** Moratorium Over 25 years25 years and under .41(ref) .36*(ref) .62(ref) .40(ref) .64(ref) .71(ref) Sex Male Female 1.26(ref) .90(ref) .79(ref) 1.02(ref) 1.3(ref) .93(ref) Discussions about the election From time to time Never Often 1.97**5.49***(ref) 2.34***3.49***(ref) 1.69*3.28***(ref) 1.60*2.85**(ref) 3.7310.90***(ref) 1.68*3.61***(ref) Qualification level Baccalaureate and above Lower than baccalaureate .92(ref) .57**(ref) .64*(ref) .52***(ref) .49*(ref) .49***(ref) Interest in politics Interested Not interested .76(ref) .68(ref) 1.35(ref) .88(ref) .93(ref) 1.18(ref) Constant .00 .00 .00 .00 .00 .00 R2 7% 8.5% 5.5% 5.5% 13% 8.5% N 445 441 427 422 481 427

Modeling of abstention declared in the DYNAMOB panel***,**,*

*** Means that the coefficient is significant at p < 0.01;
** at p < 0.05; and
* at p < 0.10.
Sources: DYNAMOB panel, waves 4, 6, 9, 12, 16, and 17. To maintain a certain homogeneity between models, political interest was declared in Wave 1, and only panelists who participated in the presidential election and at least one other election were retained.

37The low number of participants means that conventional variables do not reach the usual significance thresholds. Nevertheless, we find the usual logics behind abstention. Those who hold a baccalaureate or higher level of qualification are significantly less likely to abstain in most elections, except municipal ones. The moratorium of youth has its expected influence of lowering participation, but there are not enough individuals for the coefficient to be significant. The DYNAMOB panel overestimates participation, but the logics behind voting or abstaining converge with those of INSEE’s participation surveys and of research conducted in France and elsewhere on the subject. It could be said that the DYNAMOB panel is flawed in terms of participation levels, but not in terms of structures.

38Three lessons from these models are particularly important. First of all, it can be seen that political interest is not significant. This can be explained in two ways. The cause may be related to statistics. The dimension of political taste that covers the variable of interest in politics is also covered by the variable of discussions about the election; the two variables may thus be too correlated. The cause seems substantial and very illuminating: it can be assumed that the role of interest differs according to the cohorts and the attitudes to voting that are predominant within them. Members of older cohorts vote out of duty and not interest, while members of younger cohorts who are interested in politics have moved away from voting as a practice. Their taste is therefore less and less predictive of their behavior when faced with the prospect of an election. I will return to this issue later.

39There is also a strong link between frequency of discussions with close friends or relatives and likelihood of voting or abstaining. If individuals have never discussed the election, they are usually between 2.8 and 5.5 times more likely to abstain than are the panelists who talked about it often. [50] For each election, this effect of interpersonal networks and the buzz surrounding the election can be seen. This interpretation is all the more plausible given that the dimension of taste for politics is taken into account by the interest in politics variable. These significant odds ratios are the product of discussions and incentives to talk politics originating from close friends and family and not of a voluntary and affinity-based disposition on the part of panelists to do so.

40Finally, year of birth clearly has an effect that goes in the expected direction: the more recent the year of birth, the more likely abstention becomes. This effect endures even when one takes the moratorium into account, and this confirms that generational renewal clearly accompanies a change in attitude toward voting that cannot be reduced to, for example, a matter of qualifications or interest.

Figure 6

Mean predicted probabilities of abstaining according to election and year of birth

Figure 6

Mean predicted probabilities of abstaining according to election and year of birth

Sources: DYNAMOB panel, waves 4, 6, 9, 12, 16, and 17; calculations by the author.

41An examination of the probabilities predicted by the regression models confirms this. Let us begin by considering panelists born in the second half of the 1980s and after. Among them, the likelihood of abstaining has grown strongly. Care must be taken, however: it is necessary to be wary of the impact of the moratorium variable, because it does not reach the significance thresholds. [51] A panelist’s likelihood of abstaining varies according to whether he or she is voting for a president, a municipal list, a deputy or a list of regional councilors, a tandem of departmental councilors, or a list of members of the European Parliament. In this respect, the DYNAMOB panelists come closer to exhibiting the participation behaviors found on the scale of the electorate. Finally, the likelihood of abstention depends increasingly on the election at hand the more recently an individual was born. The likelihood of abstention among panelists born in the 1940s ranges between 5% and less for the 2017 presidential election and about 25% for the 2015 departmental elections. For panelists born in the 1960s, the likelihood goes from 6% to 10% for the presidential elections and from 35% to 39% for departmental elections, and for those born in the 1970s, it goes from 10% to 15% for the presidential election and from 39% to 45% for departmental elections. The variations therefore range from fewer than 20 points for panelists born in the 1940s to more than 30 points for those born in the 1970s.

42Here I do not take into account differences between panelists in terms of qualification levels or attitudes toward politics. When one integrates these into the analysis, developments in attitudes toward voting can be measured even more effectively. Respondent numbers make it necessary to consolidate the panelists into three cohorts (1940–1960, 1961–1975, and 1976–1987), and those still in the moratorium period have been left out of the analysis.

Figure 7

Average predicted probabilities of abstaining according to election type

Figure 7

Average predicted probabilities of abstaining according to election type

Sources: DYNAMOB panel, waves 4, 6, 9, 12, 16, and 17; calculations by the author.

43The joint effects of qualifications and cohorts are included for the sake of consistency with the first part of this research. The DYNAMOB panel presents the same explanatory logics. The more qualifications an individual has, the less likely he or she is to abstain, and the more recently an individual was born, the stronger the likelihood that he or she will abstain, regardless of his or her qualification level.

44The most significant results concern the interaction between cohorts on the one hand and interest in politics or frequency of political discussions on the other. Regarding interest in politics, its nonsignificance can be better understood in the light of the probabilities predicted by the models, in particular in relation to European elections. Panelists born in the 1940s and 1950s are the most likely to vote, but those within these cohorts who show little interest in politics take part in elections more than their counterparts who were born in the 1976–1987 period, including those who say that they are interested in politics. [52] The former fall within the logic of duty-based voting: they declare that they go out and vote even though they have little taste for politics, something that should in fact encourage them to abstain. The latter have a remote relationship with voting: they have the competencies for voting (particularly in terms of qualifications), and some even have the motivation (they are interested in politics), and yet they abstain. Undoubtedly among them there are people who abstain out of indifference and prefer not to go out and vote because they have assessed the issues involved in the election and concluded that they are not important.

Votes, taste, and micropressures: Further differences among cohorts

45Ultimately, the influence of mobilization from below can be confirmed, since the impact of political discussion (controlled by the level of political interest) is such that it influences electoral mobilization more so than education level does. Among the 1961–1975 cohort, whether or not an individual holds a baccalaureate changes the likelihood of abstaining by 13 points in the European elections and by 7 points in departmental elections. On the other hand, whether an individual has engaged in discussions about the elections often or never creates, within the same cohort, variations of 29 points for European elections and 23 points for departmental ones. Discussions and the social networks that are involved in them therefore seem particularly important in whether or not voters are mobilized.

46Moreover, there is a link between generations and engaging in discussions about elections, and this link also sheds new light on the mechanisms that lead to the decentering of voting. These mechanisms come in several types. First, discussions about elections are less and less frequent among the younger cohorts. Only 7% of panelists born before 1960 never talked about the legislative elections, versus 15% of members of the 1961–1975 cohort and 27% of those in the 1976–1987 cohort. For European elections, the respective proportions are 13%, 20%, and 30%, and for departmental elections they are 14%, 16%, and 30%. The differences are smaller in the presidential and municipal elections, but they are not completely erased: from 4% to 11% for presidential elections and from 7% to 14% for municipal ones. In other words, recent cohorts are immersed in an environment and in networks in which elections are clearly not as worthy of interest. Therefore, the younger a cohort is, the fewer chances its members will have to be informed about and made aware of upcoming elections, or to be reminded of the civic duty of voting.

47Furthermore, this greater or lesser frequency of discussions across cohorts is not only a question of taste for politics. Of course, in each cohort, the more a person is interested in politics, the more likely it is that he or she will be involved in an electoral discussion, either because he or she begins one or because his or her close friends and relatives are also more frequently politicized. But the results vary depending on the cohort and the enthusiasm for politics. Even when individuals born before 1960 are not interested in politics, they are confronted with political discussions more frequently than are the cohorts that came after them. In this cohort of baby boomers, 22% of those with little or no interest never discussed the European elections; the equivalent figures are 18% for departmental elections, 16% for regional elections, 10% for legislative elections, and 9% for presidential elections. In the case of their counterparts born in the 1976–1987 period with little or no interest, 39% never talked about the European elections, 38% never talked about the departmental elections, 33% never talked about the legislative elections, 30% never talked about the regional elections, and 15% never talked about the presidential elections. In other words, ordinary citizens are immersed in networks of discussion that are more or less conducive to the transmission of information about elections, and the situation is clearly very different for the old cohort than it is for the younger ones. Among the former, even if they have no taste for politics, they are taken to the polls by a sense of duty on the one hand and by micropressures from close family and friends on the other. For the latter, an accumulation of factors impedes some citizens’ voting: a low enthusiasm for politics, increasingly weak binds from civic norms, and lower and lower taste for hearing people talk about elections. It is easy to understand why the likelihood of abstaining predicted by the models approaches 60% for departmental and European elections in the case of people in the 1976–1987 cohort who do not engage in discussions about politics.

48Ultimately, citizens interested in politics have different probabilities of discussing elections depending on the cohort to which they belong. Unsurprisingly, baby boomers who are interested in politics talk about it with their close relatives and friends: only 12% at most did not discuss the European elections and 8% did not discuss the departmental elections; for the other elections, the figures are between 0% and 4%. But in the 1976–1987 cohort, individuals who are interested in politics but do not discuss it stand at 4% for the presidential election, at 17% for European and departmental elections, and at 21% for legislative elections; these proportions are similar to the figures for citizens who have little or no interest in politics and were born in 1960 or earlier. This result is particularly surprising. Either these interested post-boomers most often belong to social networks that are not particularly politicized most of the time, except in the case of presidential elections, or the kind of politics that elicits their taste has less and less to do with the electoral sphere.

49* * *

50Having reached the end of this analysis, it can undoubtedly be said that generational renewal does alter voting and electoral participation. It contributes a share of the explanation alongside conventional effects related to social inequalities and the life cycle. We are not heading toward an increase in systematic abstention but toward a rise in intermittent voting. The presidential election still brings people to the ballot box, unlike other elections. In particular, it can be seen that intermittent voting is not a matter of life cycle and that it affects cohorts whose members have now reached their forties or, in the case of the older members, are approaching their fifties. In other words, variations from one election to another will become more and more significant.

51The consequences in terms of the electoral influence of different cohorts could prove particularly important in the long run. In the 2017 legislative elections, the 1938–1947 and 1948–1957 cohorts weighed 1.3 times more than their demographic weight among those registered to vote, while voters born after 1977 had a weighting of 0.7 relative to their weight among registered voters. And these imbalances may well become more accentuated in the future. Moreover, the 1938–1947 and 1948–1957 cohorts saw their weight increase between the 2002 and 2017 legislative elections. This development would not be of concern if the cohorts were indistinguishable in terms of interests or values, but we know full well that this is not the case. For example, Les Républicains is a party that polls best among older cohorts. [53]

Figure 8

Generational imbalance in legislative elections (2002–2017)

Figure 8

Generational imbalance in legislative elections (2002–2017)

Sources: INSEE participation surveys; calculations by the author. Reading note: a score of 1 means that the cohort weighs as much at the ballot box as it does among the electorate; a score of 0.5 means that it weighs two times less at the ballot box than it does among the electorate; and a score of 1.5 means that it weighs one and a half times more at the ballot box than it does among the electorate.

52The progress of intermittent voting that can be found among the post-boomer generations can be explained in two ways. On the one hand, a change in attitudes toward voting between cohorts can be observed. The oldest continue to be characterized by their duty-based voting, which leads them to participate regardless of their enthusiasm for politics. Among the youngest, taste for politics and voting are less and less linked, which means that even interested individuals no longer go out and vote. The second explanation is collective, and it also relates to the diminished buzz that elections arouse among younger cohorts. In general, the younger a cohort, the less its members discuss elections. But this is now also true of cohort members who are interested in politics, surely because their definition of politics is less bound up with ballot boxes and elected representatives and because their repertoire of actions now incorporates other means of action.

53However, the accumulation of these effects is particularly significant among ordinary citizens—that is, the least politically-interested citizens, as well as those with the fewest qualifications. For example, during the 2017 legislative elections, manual and white-collar workers from the cohort born in 1978 or after weighed two times more among the population than they did at the ballot box. This is a particularly worrying finding, and it calls into question the ability of France’s democratic political system to hear these voters. Post-boomers with qualifications and an interest in politics may choose not to vote and consider other means to be more effective, in particular because they have the resources to make themselves heard in other ways. They become involved in petitions, they join associations, and they mobilize in the street or on the internet.

54But what about manual workers, white-collar workers, relatively unqualified individuals, and precarious members of these new generations? If one considers the use of demonstrations, petitions, or boycotts among younger cohorts, it can be seen that these activities have widespread acceptance, but also that there are widening inequalities between those who turn to them and those who do not, and in particular between those with qualifications and those without and between executives and white-collar or manual workers. [54] In other words, some may become proficient participants for whom voting is only one means among others (and not necessarily the most effective) for expressing themselves, while others not only no longer vote but also no longer protest. How will they be heard?


Annex 1: An alternative model for sociological analysis of voting (2002–2012)

55Not having variables that measure social inequalities in general or only being able to look at qualifications is naturally frustrating. One can only lament the impossibility of distinguishing retirees based on their former occupation in 2017, which means that a former manual worker is treated in the same way as a former executive. The fact that it was possible to do so in previous surveys only increases this frustration. Clearly, occupation and former occupation have an impact on voting when the analysis is confined to the 2002–2012 period.

tableau im12
Presidential elections, first round Legislative elections, first round Legislative elections, first round Presidential elections, first round Legislative elections, first round Legislative elections, first round 1 2 3 1b 2b 3b Female 1.03** 1.02* 1.01 1.02 .99 .98 Male Moratorium Under 25 years .50*** .38*** .42*** .80*** .81*** .85*** 25–29 years .59*** .40*** .41*** 1.10** 1.08** 1.06 Over 29 years Old age 75–79 years .91** 1.39*** 1.54*** .72*** .84*** .92 80–84 years .64*** 1.11*** 1.40*** .50*** .70*** .85** Over 84 years .27*** .50*** .84*** .21*** .32*** .51*** Under 75 years Qualification level None/primary .77*** .95*** 1.04** .69*** .77*** .85*** Baccalaureate 1.03 1.04** 1.04* 1.08*** 1.13*** 1.13*** Upper level 1.11*** 1.11*** 1.09*** 1.28*** 1.43*** 1.40*** No response/ongoing .53*** .59*** .70*** .54*** .60*** .72*** Secondary Occupation Self-employed .95 .85*** .84*** .98 .90*** .88*** Intermediate professions .96 .83*** .82*** 1.00 .90*** .88*** Manual workers/white-collar workers .77*** .62*** .62*** .85*** .74*** .74*** Others .76*** .68*** .70*** .82*** .79*** .81*** Executives and upper-level intellectual professions Presidential vote Yes 10.65*** 10.50*** No
tableau im13
Cohort 1927 and earlier 1.00 1.05 1.11* 1928–1937 1.00 1.14*** 1.17*** 1948–1957 .88*** .77*** .77*** 1958–1967 .69*** .53*** .54*** 1968–1977 .52*** .34*** .36*** 1978 and after .39*** .22*** .23*** 1938–1947 Year 2007 2.11*** .82*** .57*** 2.18*** .86*** .59*** 2012 1.70*** .73*** .53*** 1.91*** .88*** .64*** 2002 Constant 4.68*** 3.39*** .58*** 5.60*** 4.56*** .78*** R2 5.5% 3.5% 16% 6% 5.5% 17.5% N 118,317 118,317 118,317 118,317 118,317 118,317

56The results are clear. First, taking occupation into account does not call into question the results for the cohorts. The youngest cohorts participate much less in legislative elections and presidential elections and tend significantly more to disengage between these two elections. But the relevance of this model also stems from its emphasizing the impact of occupation or former occupation, an impact that can be added to that stemming from qualification level, for example. Manual workers and white-collar workers are 1.4 times more likely than executives to abstain in legislative elections or to disengage after presidential elections.


  • [1]
    Translator’s note: Unless otherwise stated, all translations of cited foreign language material in this article are our own.
  • [2]
    A tag spotted at the Marché des Capucins in Bordeaux, April 3, 2017. Literally meaning “voting mutilates.”
  • [3]
    Pascal Delwit, “The End of Voters in Europe? Electoral Turnout in Europe since WWII,” Open Journal of Political Science 3, no. 1 (2013): 44–52.
  • [4]
    Céline Braconnier and Jean-Yves Dormagen, La démocratie de l’abstention. Aux origines de la démobilisation en milieu populaire (Paris: Folio, 2007).
  • [5]
    Alain Lancelot, L’abstentionnisme électoral en France (Paris: Armand Colin, 1968).
  • [6]
    Robert E. Lane, Political Man (New York: The Free Press, 1972).
  • [7]
    Kay Lehman Schlozman, Sidney Verba, and Henry E. Brady, The Unheavenly Chorus: Unequal Political Voice and the Broken Promise of American Democracy (Princeton, NJ: Princeton University Press, 2012).
  • [8]
    In 1968 in France, 8% of the population held a baccalaureate or a higher-education qualification, while 45% had no formal qualifications; in 2006, 37% held a baccalaureate, while 20% had no formal qualifications.
  • [9]
    Russell Dalton, Citizen Politics in Western Democracies: Public Opinion and Political Parties in Advanced Industrial Democracies (Chatham, NJ: Chatham House, 1988).
  • [10]
    Russell Dalton, The Good Citizen: How a Younger Generation Is Reshaping American Politics (Washington, DC: CQ Press, 2007); André Blais and Daniel Rubenson, “The Source of Turnout Decline: New Values or New Contexts?” Comparative Political Studies 46, no. 1 (2013): 95–117; Vincent Tiberj, Les citoyens qui viennent (Paris: PUF, 2017).
  • [11]
    Tiberj, Les citoyens.
  • [12]
    Daniel Gaxie, Le cens caché. Inégalités culturelles et ségrégation politique (Paris: Seuil, 1978).
  • [13]
    Sophie Duchesne, Citoyenneté à la française (Paris: Presses de Sciences Po, 1997); Dalton, The Good Citizen.
  • [14]
    Engin F. Isin and Greg M. Nielsen (eds.), Acts of Citizenship (London: Zed Books, 2008).
  • [15]
    Color versions of this article’s figures are available at the following address:
  • [16]
    This is a point that separates me from the works of André Blais or Daniel Rubinson, for whom a decline in participation is beyond question. In France, systematic participation is what is in decline.
  • [17]
    Sidney Verba, Kay Lehman Schlozman, and Henry E. Brady, Voice and Equality: Civic Voluntarism in American Politics (Cambridge, MA: Harvard University Press, 1995).
  • [18]
    Éric Agrikoliansky et al., Les sens du vote. Une enquête sociologique (France, 2011-2014) (Rennes: Presses Universitaires de Rennes, 2016).
  • [19]
    Braconnier and Dormagen, La démocratie.
  • [20]
    Céline Braconnier, Une autre sociologie du vote. Les électeurs dans leurs contextes: bilan critique et perspectives (Paris: Lextenso/LEJEP, 2010).
  • [21]
    Guillemette Buisson and Sandrine Penant, “Élections présidentielles et législatives de 2002 à 2017: une participation atypique en 2017,” Insee première 1671 (October 2017); Sébastien Durier and Guillaume Touré, “Élections de 2017: 6,5 % des citoyens ont fait une démarche volontaire pour s’inscrire,” Insee Focus 80 (March 2017); Sébastien Durier and Guillaume Touré, “En 2014, 85 % des jeunes Français de 18 à 24 ans étaient inscrits sur les listes électorales,” Insee Focus 22 (March 2015); Xavier Niel and Liliane Lincot, “L’inscription et la participation électorales en 2012: qui est inscrit et qui vote?” Insee Première 1411 (September 2012); Stéphane Jugnot, “La participation électorale en 2007: la mémoire de 2002,” Insee Première 1169 (January 2008).
  • [22]
    Patrick Lehingue, Le vote. Approches sociologiques de l’institution et des comportements électoraux (Paris: La Découverte, 2011), 129.
  • [23]
    François Héran, “Voter toujours, parfois … ou jamais,” in Le nouveau désordre électoral. Les leçons du 21 avril 2002, edited by Bruno Cautrès and Nonna Mayer (Paris: Presses de Sciences Po, 2004): 351-367.
  • [24]
    Buisson and Pénant, “Élections présidentielles.”
  • [25]
    Jan E. Leighley and Jonathan Nagler, “Socioeconomic Class Bias in Turnout, 1964–1988: The Voters Remain the Same,” American Political Science Review 86, no. 3 (1992): 725–736; Marvin E. Olsen, “Social Participation and Voting Turnout: A Multivariate Analysis,” American Sociological Review 37, no. 3 (1972): 317–333.
  • [26]
    Lancelot, L’abstention.
  • [27]
    Anne Muxel, L’expérience politique des jeunes (Paris: Presses de Sciences Po, 2001).
  • [28]
    Readers may wonder about the cutoff points in the dividing up of the cohorts. The approach taken is necessary because of the 2007 participation survey, which only breaks down age into five-year segments.
  • [29]
    Cécile Van de Velde, Devenir adulte. Sociologie comparée de la jeunesse en Europe (Paris: PUF, 2004).
  • [30]
    Out of voters born before 1928 who voted in the 2002 presidential election, 86% also voted in the legislative elections. For 2012 and 2017, these proportions are 72% and 70%.
  • [31]
    At the individual level, this means that during this year only 60% of voters from this cohort who voted in the presidential election also participated in the legislative elections.
  • [32]
    Céline Braconnier, Baptiste Coulmont, and Jean-Yves Dormagen, “Toujours pas de chrysanthèmes pour les variables lourdes de la participation électorale: chute de la participation et augmentation des inégalités électorales au printemps 2017,” Revue française de science politique 67, no. 6 (December 2017): 1023–1040.
  • [33]
    The proportion of individuals who hold a qualification level of baccalaureate or higher is 60% in the 1968–1977 cohort, whereas it is 26.5% in the 1938–1947 cohort, which votes in elections more.
  • [34]
    13% of the members of the 1968–1977 cohort are executives, and 26% of it occupy an intermediate profession, whereas 6% of the 1938–1947 cohort are executives and 19.5% are in intermediate-level professions.
  • [35]
    It is not possible to distinguish between retirees according to their former occupations in the 2017 participation survey. The chart on cohorts, occupations, and abstention therefore only includes cohorts that were economically active in 2017, but it covers the whole set of cohorts for the previous years.
  • [36]
    To use Verba, Lehman Schlozman, and Brady’s terms of “voice” and “equality.”
  • [37]
    For example, there will be no multicollinearity among members of the 1968–1977 or 1978–1987 cohorts and being in the moratorium period. On the other hand, it is necessary to be careful when it comes to the coefficients for the youngest and the oldest cohorts.
  • [38]
    When one works on a cumulative dataset, one faces problems of comparability between variables. Qualifications and occupations do not have the same coding from one year to the next. These restrictive groupings prevent full advantage being taken of the sociological richness of these data. It has also been necessary to remove respondents’ occupation or former occupation because of the available coding from 2017. Nevertheless, the models with occupations for the 2002–2012 period are presented in the Appendix.
  • [39]
    For the 2002–2012 period, a significant occupation effect can also be observed: manual and white-collar workers tend to abstain more than executives do, taking account of the other independent variables.
  • [40]
    This is a phenomenon noted by Braconnier, Coulmont, and Dormagen, as is mentioned above.
  • [41]
    Particularly substantial variations in coefficients in terms of qualification level can be noted once cohorts are brought into the analysis. This clearly confirms the interaction effect between qualification and cohort that the descriptive analysis had brought to light.
  • [42]
    Braconnier, Une autre sociologie du vote; Meredith Rolfe, Voter Turnout: A Social Theory of Political Participation (Cambridge, MA: Cambridge University Press, 2012); Agrikoliansky et al., Les sens du vote.
  • [43]
    Ronald Inglehart, Culture Shift in Advanced Industrial Society (Princeton, NJ: Princeton University Press, 1990).
  • [44]
    Dalton, The Good Citizen.
  • [45]
    Tiberj, Les citoyens. Online
  • [46]
    Anne Muxel, “L’abstention: déficit démocratique ou vitalité politique?” Pouvoirs 120 (2007): 43–55, 49.
  • [47]
  • [48]
    John Robert Warren and Andrew Halpern-Manners, “Panel Conditioning in Longitudinal Social Science Surveys,” Sociological Methods & Research 41, no. 4 (2012): 491–534.
  • [49]
    The exact wording is: “During the [type of] election campaign, did you talk about the elections with your friends or your family? Often, from time to time, never.” Note that this is the activity most frequently engaged in by the panelists, far ahead of, for example, viewing television programs, reading newspapers or leaflets, or looking at websites.
  • [50]
    For the presidential election, the odds ratio rises to a likelihood of abstaining that is ten times greater. This can be explained by the low number of panelists who have never talked about the presidential election: 6% versus up to 20% for the European elections.
  • [51]
    It is nevertheless interesting to note that the potential extent of the moratorium varies according to the type of election. It is very low for presidential elections, but particularly strong for municipal, European, and regional ones.
  • [52]
    This finding was already made via the European Electoral Studies (Tiberj, Les citoyens, 142).
  • [53]
    Tiberj, Les citoyens.
  • [54]
    Tiberj, Les citoyens.

Intermittent voting is on the rise in France. In this article, generational renewal is highlighted as a key factor behind this phenomenon. In addition to cohort effects, however, the importance attributed to the act of voting is also currently shifting. Older cohorts are marked by a culture of deference and voting is a regular occurrence based on a sense of duty, whereas for younger cohorts, voting is more episodic. For the latter, casting one’s ballot is seen as less important than other forms of political action. This development tends to entrench social inequalities through voting. To show this, our analysis uses data from the 2002 to 2017 INSEE electoral participation surveys and the DYNAMOB panel.

  • election
  • turnout
  • abstention
  • generations
  • citizenship
  • sociopolitical inequalities
Vincent Tiberj
Vincent Tiberj is a full professor, a researcher at the Emile Durkheim Center, and dean of research at Sciences Po Bordeaux. Between 2002 and 2015, he was a research fellow at Fondation nationale des sciences politiques (FNSP) (National Foundation of Political Science), first within CEVIPOF, and then within the Centre d’études européennes (CEE) (Center for European Studies). His publications include: Les citoyens qui viennent (Paris: PUF, 2017); Sociologie plurielle des comportements politiques (Paris: Presses de Sciences Po, 2017), as editor with Olivier Filleule, Florence Haegel, and Camille Hamidi; Des votes et des voix. La France des urnes de Mitterrand à Hollande (Paris: Champ social éditions, 2013); La crispation hexagonale. France fermée contre France plurielle, 2001-2007 (Paris: Plon/FJJ, 2008); Français comme les autres? Enquête sur les citoyens d’origine maghrébine, africaine et turque (Paris: Presses de Sciences Po, 2005), coauthored with Sylvain Brouard. A specialist in political psychology and in analysis of electoral and political behaviors in France, Europe, and the United States, his work focuses on “ordinary” citizens’ reasoning and the political sociology of social and ethnic inequalities, as well as on xenophobic prejudices and value systems.
Uploaded on on 31/07/2019
Distribution électronique pour Presses de Sciences Po © Presses de Sciences Po. Tous droits réservés pour tous pays. Il est interdit, sauf accord préalable et écrit de l’éditeur, de reproduire (notamment par photocopie) partiellement ou totalement le présent article, de le stocker dans une banque de données ou de le communiquer au public sous quelque forme et de quelque manière que ce soit.
Loading... Please wait