1As a result of the electoral “earthquakes” that marked the beginning of the 2000s, the working classes are back at the centre of public debate, having been eclipsed by the middle classes, who seemed to have monopolized political attention for some time. With the Front National reaching the second round of the 2002 French presidential elections, and the subsequent rejection of the plan for a European Constitution in 2005, the question of working-class living standards is once again coming to the fore. There are numerous analyses seeking to gauge the political consequences of the precaritization of the situation and prospects of a growing fringe of marginalized workers and employees.  And where certain works seemed to have officially confirmed a decline in the explicative power of the “heavy variables” to explain electoral behavior,  the link between living standards and workers’ or “working class” votes now once more figures highly on the media agenda. Too often, however, the working class is presented as artificially homogeneous, whether it is a matter of describing their disaffection with politics, their divergence from the parties of the left, or their inexorable slide toward the extreme right.
2In the social sciences, however, there exist numerous works illustrating the heterogeneity of the social groups that make up the working class. For instance, in his work on the contemporary working class Olivier Schwartz suggests that, although they should certainly always be analyzed in terms of domination since certain groups are still relegated to situations of “subordination in the division of labor and in political and social relations”, they should also be understood in terms of acculturation, since “ever-growing fringes of the dominated classes now participate increasingly in the dominant culture, especially through education”. When we approach the situation of the working class, therefore, we must take into account the processes of stratification that cut across them, for “as soon as we turn to subaltern groups who are not in immediate poverty, we encounter a multitude of mixed, intermediate, and undecidable situations”. If “the differences between them may often seem slight, when viewed from the top of the social structure”, this is a “perspectival illusion”. 
3From this point of view, although numerous ethnographic studies have enabled granular description of many of these segments of the working-class population,  few recent works allow us to grasp the working class as a whole and to situate its different constituent categories in relation to one other. Over the last three decades, important work has certainly been done in proposing new representations of a social structure undergoing profound changes, and which bears the marks of new divisions; but none of this work specifically addresses working-class spaces.  As far as quantitative work is concerned, the question of sample sizes is central here: the studies that have been done, comprising, at best, a few thousand individuals, rarely permit one to make any more meaningful distinctions than the classical opposition between skilled and low-skilled non-managerial workers.
4Using surveys concerning electoral participation and respondents’ relationship to politics which, on the other hand, allow us to make far finer distinctions among employees and workers, this article aims to make two contributions: firstly, to add to our knowledge of the relationship to politics of various different segments of the working class; and secondly, to add to our knowledge of the internal processes of stratification operating within the working class. The question of the coding and classification of professions and socio-professional categories has been the subject of numerous works and controversies – whether questioning the reasons for a decline in their usage as an explanatory variable  or considering new classifications – and this article should be considered within this context. 
5Before presenting the data and the classifications used, I will attempt a brief overview of an extremely rich seam of research: that of the working class’s relationship with politics. Finally, I shall present the results of my study.
The “working class” and political participation
6In France, participation in various elections has remained a widespread practice observed by the majority of the populace, even into the 2000s. Yet since the end of the 1980s there has been a relatively steady growth in abstentionism, which saw it exceed 50% in the 1999 European elections and in the subsequent 2010 regional elections. As far as major national elections are concerned, we must make a distinction between the presidential election and legislative elections. There is a clear enough decline in participation in the latter,  whereas the rate of participation in presidential elections remains high, in spite of the record of 21 April 2002, when 28% of the registered electorate did not vote. Five years later, in 2007, only 16.2% of the electorate abstained – a proportion comparable to that seen in 1974.  Even though participation declined once again in the 2012 election, it remained around the 80% mark. Nevertheless, these variations across different elections themselves indicate a rise in intermittent participation on the part of an electorate which is becoming increasingly sensitive to the context and the stakes of elections. This is an important point, for it calls attention to the fact that “the cycle of low turnout into which France fell from the end of the 1980s therefore apparently does not, at this point, imply a massive process of electoral exit”:  only around one in ten registered voters consistently abstain from voting.
Reasons for abstention
7The sociological determinants of abstention have been known for some time. Voting in an election requires resources and skills which are social markers, and are unequally distributed. Numerous works around the notion of the “hidden census”  have analyzed the way in which those most lacking in educational capital tend to self-exclude from a political field  – a field that can keep them at a distance all the more easily given that subjective incompetence and indifference constitute an invisible barrier, and are particularly responsible for fuelling mass abstention by the dominated groups.  This sociological model has subsequently been enriched by work emphasizing the decisive nature of the processes of electoral mobilization:  membership of groups or networks is a key element of this mobilization, since it “forges an identity, collective solidarities, an interest in acting together”.  These sociological determinants are also in play when we try to explain the rise in intermittent participation, for the latter is over-represented among the working class, and more generally among the most vulnerable individuals. 
8However, these classic sociological variables are not enough to account for the rise in abstention. For abstention is also rising among far more socially well-equipped individuals and groups, for whom it takes on a more political significance. We find this “political” abstention among well-educated voters close to the middle or upper class, voters who, moreover, show a real interest in politics – for example, those of the younger generation, who are particularly sensitive to the potential effects of the electoral offer.  Beyond this, the major impact of the electoral context and the issues at stake in an election is now well documented: strong and clearly-defined issues within the framework of an electoral competition increases participation noticeably.  Here again, the participation of the working class cannot be explained by their socio-demographic characteristics alone: the material organization of the vote, and the mechanisms of “material facilitation” and of mobilization, must also be taken into account. 
9In this article, I seek to advance knowledge of the relationship between the working class and political participation, while taking into account the considerable heterogeneity of the groups that make up these classes. At the beginning of the 1980s, Guy Michelat and Michel Simon warned against the temptation to neglect the different sub-populations that make up the category of “workers”. In particular, they emphasized that it was necessary to take into account the specific nature of communist workers and union activists, whose intense political participation demands, at the very least, that we further develop the theory of the hidden census to take account of the fact that certain factions of the working class are characterized by a high degree of politicization. Three decades later, we must also take into account the effects of two successive explosions in education, which have appreciably extended the schooling of employees occupying non-managerial posts. The data from the Insee Employment Survey reveals that in 2012, 20% of manual workers were qualified to at least baccalaureate level, as opposed to less than 4% in 1987. Although this rise in educational achievement has not translated into a proportionate rise in social mobility,  it has nonetheless drastically disrupted the world of the working class, and must not be underestimated when analyzing the relationship to politics, since cultural resources are decisive in this matter. In this study I address two dimensions of this relationship to politics. First, data on participation in presidential and legislative elections during the 2012 electoral cycle provides the “objective” dimension as far as political participation is concerned. But I also aim to analyze a more subjective relationship to politics via an indicator of self-exclusion from political life, linked to the capacity of individuals to grasp political issues and to their degree of politicization.
New forms of polarization among non-managerial employees
10This article places the two socio-professional categories of “workers” and “employees” at the heart of the analysis. This definition of the working class is imperfect, however. It seems to exclude the self-employed from the outset, and yet the affiliation of certain of the self-employed with the working class is something that merits serious discussion. Conversely, among employees, there are certain categories of job which should lead us to situate their holders higher in the social sphere. The objective of this research is thus not to produce an exhaustive or exclusive cartography of the space of the working class, but to contribute toward situating the different categories of non-managerial employees in relation to one another.
11Despite common assumptions to the contrary, employees and workers still constitute half of all those in work in France today. Although the proportion of such workers in the active population diminished slightly between 1989 and 2009, dropping from 55% to 51%, their number has grown by more than a million, from 11.9 to 13 million.  On the other hand, over recent decades the working class have been profoundly affected by transformations in the employment landscape, and in particular by the tertiarization of the economy, so that today the image of the industrial worker, once archetypal of the “working class”, corresponds to only a minority of the contemporary working class. Among those working today, employees are now appreciably more numerous than workers (7.5 million against 5.5 million); and among the latter, workers in the service sector are increasingly numerous. What is more, new divisions have progressively opened up which tend to complicate, if not entirely supplant, the traditional distinction between employee and worker. This is the case, first of all, with the widening gap between the skilled and the low-skilled. Recent studies have shown that low-skilled workers and employees increasingly represent a distinct segment of the workforce, or even a “new social class”.  These 5.5 million low-skilled employees and workers are indeed objectively distinct from other employees and workers. First of all, in terms of salary and standards of living, they are noticeably less well off than the rest of the workforce.  Those low-skilled occupations in which women, the young, the uneducated, and immigrants are over-represented are also those that impose more arduous working conditions than other non-managerial jobs.
12These low-skilled jobs represent a risk of isolation and stagnation for those who hold them, since “the sphere of low-skilled posts offers few connections to the sphere of skilled work”.  This growing disconnect between skilled and low-skilled work is part of a far more widespread movement of polarization of the social structure, which itself testifies to the impact of the globalization of trade.  Indeed, the gap between “winners” and “losers” of globalization occupies a central place within recent studies, and supplies a possible framework within which to understand the success of extreme-right populist parties in Europe:  although it is difficult to identify “winners” of globalization among the working class, a certain subaltern segment of the workforce are probably more exposed to the negative consequences of a global competitive labor market.
13There is another gap that seems pertinent, one that is also linked to the globalization of trade but cannot be reduced to its effects alone: the gap between skilled and low-skilled partly coincides with that which separates insiders from outsiders in the labor market. Unlike insiders, who benefit from relatively stable and secure forms of work, outsiders, who are over-represented among the least skilled workers and employees, face working conditions and types of employment contracts that are far more precarious, as well as career paths strongly marked by periods of unemployment or by recurrent underemployment.  Hence this article tries to take into account the nature of the employment contract, and more generally the quality of a worker or employee’s position in the labor market, so as to gauge the possible effects of these factors upon political participation. In doing so, it follows the path opened up by earlier studies that sought to analyze the political consequences of this widening gap. 
Data and classification
Two dimensions of the relationship with politics
14Voting in different elections is a central practice to be taken into consideration when seeking to objectively determine an individual’s relation to politics. From this point of view, Insee’s Electoral Participation Survey offers ideal material, since it is based not on the retrospective declarations of its subjects, as are other surveys, but instead reports on actual practice, supplemented by information collected by researchers. The 2012 survey took a random sample of almost 40,000 voters listed on the electoral register. Researchers then travelled to the prefectures to consult the attendance lists and to record whether the randomly selected individuals voted or abstained. These results were cross-referenced with the data of the Permanent Demographic Sample (the first large-scale socio-demographic panel established in France) so as to discover the principal demographic and socioeconomic characteristics of each individual. With a final sampling rate of almost one in a thousand, this study made it possible to establish relatively granular categories, even when the analysis is limited to active employees and workers alone. Furthermore, the Electoral Participation Survey offers an initial basis for this research, allowing us to gauge the registration of individuals on electoral registers, this time with a sample of the more than 280,000 individuals from which the 40,000 individuals of the final sample were drawn.
15In addition to the actual practice of voting, this article also investigates a far more subjective dimension of the relationship to politics. By compiling multiple iterations of the European Social Survey (ESS), it is possible to address the way in which individuals interact with political debate, by constructing an attitude scale that measures individuals’ own feelings about their competence and their degree of politicization, or, to put it in negative terms, the degree to which they self-exclude from political debate.
16As a secondary exploitation of the Electoral Participation survey, I use the available classification, that of the two-character socio-professional categories which distinguish between five categories of employees and seven categories of workers. The type of employment contract is also included in the data, meaning that it is possible to test for the possible effects of one particular form of weakness in the subjects’ position in the labor market. The ESS data, on the other hand, allows us to look at things in a different way, and to construct a more granular classification, since the cumulative data of six successive surveys gives us a sample of almost 2,800 employees and workers. In this study, I use the first four rounds of the survey, conducted between 2002 and 2008, in which identical questions were posed to subjects concerning their relationship to politics. In these surveys, professions are coded using the ISCO classification. In its most fully-fledged form, this classification sets out ten “major groups” of occupations according to their skill level, itself judged according to the criteria of the International Standard Classification of Education (ISCED) and its four levels. This decision led the International Labor Organization to adopt a “pyramidal” structure of occupations.
17The classification then proposes three more detailed levels, recognizing 28 “sub-major groups”, 118 “minor groups”, and 390 “unit groups”. These unit groups are composed of occupations which themselves group together jobs whose tasks are seen to be similar. Employees and workers are present in six major groups as a function of a division whose principles do not quite coincide with those that determine the architecture of the French classification of socio-professional categories. “Skilled” jobs, which are considered to demand skills at level 2 (completion of the first stage of secondary education, up to 17 or 18 years old) are distinguished from “low-skilled” jobs requiring only primary (level 1) education. By cross-referencing this classification with the variables measuring the status of the activity (independent/salaried) and the status of the job (private sector/public sector), I have constructed a classification that distinguishes between fifteen categories (see Table 1). This classification allows us to drill down to the level of occupations, and in so doing following Weeden and Grusky, who showed that the analysis of political behavior was improved by such drilling down.  Since one of my hypotheses is that the variables linked to occupation allow us to better explain differences in political attitudes, I have excluded retired people from my analysis, so that the results concern only active workers and employees, whether they are employed or registered as unemployed (in which case I have access to data on their most recent employment).
Fifteen categories of employees and workers
Fifteen categories of employees and workers
18As for measuring the quality of the position within the labor market itself, the literature offers many different mechanisms to do this; some principally based on job security at a given moment, some taking a more biographical approach, and looking at the probability of a subject’s being faced with unemployment in the course of their career. Others seek to construct an indicator of exposure to different types of risks.  In this regard, the particular variables at our disposal are the type of employment contract (permanent or temporary) and, for the ESS data, the extent of unemployment in the years preceding the survey.
Different “professional spheres”
19Within the “archipelago of employees”,  administrative employees make up the world of the office. This workforce can be split into private sector administrative staff and those working for the state or for local public bodies. Beyond this important division, the most frequent occupations are those of secretary, financial and accounting staff, employees responsible for procurement and inventory management, and mailroom and other mail service employees. The sectors concerned are those of IT (computer “operators”), accounting, banks, and insurance, whose employees have been described as the aristocracy of the workforce.
20Reception employees constitute the world of public contact. This includes counter clerks, receptionists, and switchboard operators, whose job is to respond to client requests or to direct them. In my sample, two thirds of them are salaried employees in the private sector, essentially acting as receptionists, the remaining third are those counter staff in the public service and administration, who thus find themselves in a paradoxical position of “subordinate authority”, invested with an “institutional authority” but “subject to numerous constraints”. 
21Kitchen staff and serving staff constitute a world of catering, which seems to involve very particular working conditions (working hours, intensity of work, physical demands of the job, etc.). Moreover, since they work in bars or restaurants, they are also close to a very particular universe with a strong identity, that of traders, business owners, and entrepreneurs.
22The world of retail is also represented in my sample in the form of 120 salespeople, mostly in shops, a very few on markets. Supermarket cashiers are also included in this sub-group.
23Finally, we have the vast world of personal services. Following the logic of the ESS, it is split into two categories chiefly according to skill level, which for the most part coincides with different sectors of activity. Almost all of the 300 “skilled” personal service employees are care and childcare professionals, or carers for the elderly or for the ill in clinics and hospitals. Those listed as “low-skilled” principally work in the domain of cleaning and waste management (cleaning and housekeeping in private homes, business premises, or institutions). Although the ESS distinction according to skill level may be arguable, it seems that the distinction between personal care and cleaning certainly makes sense in terms of differing working conditions and the unequal prestige afforded to these two sectors. 
24The “workers” group is more difficult to divide up. Aside from “laborers”, who feature at the very bottom of the hierarchy established by the ISCO classification, this group is constructed so as to distinguish workers according to the sector in which they work. Along with agricultural workers and art and precision craft workers, three major sectors are identified: building, metalwork and mechanics, and machine operators and assembly workers.
Upstream of the vote: an exhaustive assessment of electoral non-participation
25Before shedding light on the determinants of participation and abstention, this section of the article addresses the question of electoral participation at the level of socio-professional groups, not just that of registered voters. Quantitative studies on electoral participation give us results for the voters who are listed on electoral registers – the only ones, by definition, able to choose whether or not to vote. In doing so, they obscure what is in play upstream of this decision: voter registration, obviously; but also the right to vote, a right that some foreign workers do not have. In a study that aims to describe the working class’s relation to politics, we cannot possibly neglect the nationality of the workforce. In some categories of workers and employees, more than 10% of those in employment are of foreign nationality and do not have the right to vote in national elections.
26So as to better define the outlines of electoral non-participation at the level of socio-professional groups, we begin with the question of registration to vote, then go on to introduce nationality into the analysis.
Registering on the electoral roll
27Around 6% of employees and 9% of workers are not registered to vote (see Table 2). Apart from skilled industrial workers, who have a rate of non-registration lower than that of personal services employees and comparable to that of commercial employees, all categories of workers have a higher rate of non-registration than all categories of employees.
Rate of non-registration on electoral register (in %)
Rate of non-registration on electoral register (in %)
28Let us add another observation here: there are noticeable gaps within both the employee and the worker groups. In the former, almost four percentage points separate private sector administrative employees from personal services employees, while in the workers group there is a difference of more than five percentage points between skilled industrial workers and low-skilled artisanal workers. In addition to a rather high variance, among employees these first results immediately highlight the considerably inferior situation of personal services employees as far as registration on the electoral register is concerned. However, these first observations are potentially biased because of the structural difference in the different categories as far as age is concerned. On average, workers are considerably younger than employees and, among the latter, personal services employees in particular are older. Age is one structuring variable of electoral participation. When one controls it a minima – for example by measuring the rate of non-registration among 18-34 year olds – there is still a gap of three percentage points between workers and employees taken as a whole. Moreover, one of the more remarkable results concerns personal services employees who, this time, with a rate of non-registration higher than 11%, have a score considerably higher than skilled industrial workers and comparable to most of the other categories of workers.
29As far as the type of employment contract is concerned (Table 3), it does indeed seem to have a real effect since, apart from personal services employees,  those with a permanent contract (CDI) are in every case more likely to be registered to vote than those on temporary contracts (CDD or interim contracts). Among low-skilled workers with temporary contracts, the rate of non-registration reaches particularly high levels of up to one in seven or eight.
Employment contract and non-registration on electoral register (in %)
Employment contract and non-registration on electoral register (in %)
30Nevertheless, age and type of employment contract are only two variables among many others. For example, a vast majority of personal services employees are women, few if any are graduates, and they are frequently immigrants or descendants of immigrants. So as to take these different dimensions into account, I estimated a model of linear regression bringing in age, sex, level of education, the fact of being an immigrant or not (the only available – rather frustrating and imperfect – indicator), size of urban unit, two-character socio-professional category, and type of employment contract. Because of the sample size, all coefficients appear statistically significant, so that it becomes a matter of considering those whose value seems significant. This model is reproduced below as an appendix; here I will limit myself to listing its principal findings. Apart from the effect of immigration (immigrants have a probability of not being registered which is more than 25 percentage points higher), all other things being equal it is (unsurprisingly) graduate status that has the most significant effect (a difference of nine percentage points between graduates of higher education and non-graduates) and, to a lesser extent, the size of the commune in which the individual lives (five percentage points). The effect of socio-professional category is of the order of two or three points, to the detriment of low-skilled workers and, to a lesser extent, personal services employees. If we control for all of these dimensions, however, the type of employment contract does not introduce any significant differences.
The perspective of an entire professional group: the question of nationality
31The non-registration of French citizens on electoral registers is in a certain sense bound up with self-exclusion from the electoral process. At the analytical level of socio-professional groups, we must take into account another form of exclusion, legal this time, which impacts workers of foreign nationality. For we risk misrepresenting the landscape of the working class if we do not take into account the question of nationality. More than one in ten low-skilled employees and artisanal workers is of foreign nationality, whereas this proportion is only 3% for intermediate professions and public servants. Taking this variable into account, the table below gives us an estimate of the proportion of the actively employed who do not take part in elections. It combines three types of non-participation, since when we seek to measure the number of actors who do not participate in elections, we must consider three situations: (1) being a foreigner; (2) being of French nationality but not registered on the electoral register, (3) being of French nationality and registered, but abstaining. For the latter situation, I have maintained a strict definition of abstention that includes only constant abstention over the two rounds of the presidential and legislative elections. Evidently, what is presented here is only an estimate arrived at on the basis of the proportion of foreigners by socio-professional category identified in the 2012 Employment Survey. Nonetheless it offers some enlightening orders of magnitude.
Estimation of electoral non-participation by professional group (in %)
Estimation of electoral non-participation by professional group (in %)
32The total percentage is calculated as follows, for example for commercial employees: 5.5% of them are of foreign nationality, to which we add 7.5% of the remaining 94.5% (those who are not registered), and 11.6% of the remaining 87.4% (94.5 × 0.075), giving a total of just over 22%.
33At the level of all professional groups, around 18% of subjects are unregistered, but the proportions vary widely, from 10% for higher-grade employees to 28% for workers. As far as electoral participation and political representation is concerned, this result is far from negligible. It invites us to consider the weight of different socio-professional groups in the process of electoral decision. Certainly, workers are more numerous than managers in the active population (22% against 17% in 2012), but the weight of the second group is ultimately greater: because workers are more likely to abstain, of course, but also because they are less likely to have the right to vote, and because they are less likely to register to vote. If we take into account these different reasons for non-participation, we can begin to reveal the role of social inequality in electoral participation far better than if we observe only the abstention of registered voters. If politics is considered to be a means for regulating conflicts between social groups, and if an electoral contest expresses divergent interests which in part correspond to the professional position of the individuals who live and work in a country, then it is essential to take these different elements into account in order to better gauge the weighting of different segments of society.
34Furthermore, this estimate reveals large disparities within the groups of employees and workers. Only 13% of administrative employees are non-participants – that is, only half as many as personal services employees. Similarly, among workers the proportion varies from 20% (for skilled industrial workers) to 40% (for low-skilled artisanal workers). Entire swathes of certain categories of workers and employees remain on the margins of voting, and these non-participation “prizewinners” implicitly reveal the internal stratification of the working class, with particularly marked zones of weakness among the least skilled service employees and the least skilled artisanal workers.
Voting at elections: major disparities in participation
35Let us now turn precisely to the analysis of abstention in an electoral cycle. Who, among those registered to vote, decides not to vote? Analysis of the 2012 election shows first of all that the phenomenon of abstention is very unequally distributed between employees and workers.
To vote or to abstain: marked disparities between employees and workers
36A first important result is the considerable difference in participation among employees and workers in the 2012 presidential and legislative elections. The gaps are even greater for the legislative elections, which enjoy a lower turnout than the two rounds of the presidential election, especially when they immediately follow the latter (Table 5).
Socio-professional category and participation in 2012 elections (in %)
Socio-professional category and participation in 2012 elections (in %)
37Among employees, the rate of participation in the two rounds of the presidential election varies from less than 78% for commercial employees to more than 83% for private sector administrative employees.  As for workers, ten percentage points separate low-skilled artisanal workers (71%) from skilled industrial workers (81%). When it comes to participation in the two rounds of the legislative election, the size of the gap is of the order of fifteen percentage points for employees (between commercial employees and public servants) and remains around ten percentage points for workers. These figures indicate the outlines of a well-established hierarchy of employees ranging from the least skilled fringes to public and private administrative jobs, while the results for workers, beyond the distinction between skilled and low-skilled jobs, are distributed per sector, with higher participation in industry. Thus, with 77% participation in the two rounds of the presidential election, low-skilled workers participate as much as skilled artisanal workers, an observation that remains valid for the two rounds of the legislative elections.
38A second result: the hierarchy between employees and workers is not absolute, in so far as those belonging to the least skilled fringes of service employees are more distanced from the election than are some categories of skilled workers. I have underlined above the relatively low participation of commercial employees: in fact, at the presidential election, they abstained more than skilled industrial workers and as much as other skilled workers, whereas in the legislative elections they are far less likely – 38%, or a proportion comparable to that observed for low-skilled workers – to participate in both rounds of voting than all categories of skilled workers. The figures for systematic abstention over the two rounds of the presidential election or two rounds of the legislative vote confirm this result, and highlight the particular position of commercial employees. Finally, close to 12% of them systematically abstained throughout all four rounds of voting – a proportion almost twice as high as for skilled industrial workers. In comparison, the rather high participation of direct personal services employees (childminders, carers, cleaners, maintenance staff, etc.) poses a problem: their below-average level of qualification and their socio-demographic characteristics should place them among the most likely to abstain. The fact that they are generally older gives us a partial explanation. The figures from the 2012 Employment Survey indicate that their average age is over 42, against 35, for example, for commercial employees. Similarly, 30% of personal services employees are over 50, as opposed to only 15% of commercial employees. Since age is strongly correlated with participation, a quick test of this variable allows us to better situate them. If we limit the analysis to those younger than 40, it is indeed the personal services employees who are least likely to participate in both rounds of the presidential election (70% against a 74% average for all employees).
39A third result concerns the extent of constant abstention. Certainly, as emphasized above, constant abstention remains largely a minority practice, concerning only 11% of workers and 8% of employees. The great majority of those who abstain, then, do so intermittently. In fact, intermittent abstention could even be said to be the rule since, apart from public servants, an absolute majority of other employees and workers abstained at least once. Nevertheless, when we add to constant abstention the phenomenon of non-registration (this time without considering the question of nationality), exclusion from the electoral process among the French is seen to be greater than previously observed. Admittedly, only 8% of employees constantly abstain, but to this we must add the 6.1% of employees who are not registered to vote. In total that makes 13.7% of employees who do not vote in any round of the electoral sequence of 2012, while this proportion is close to 20% for workers. For employees it reaches nearly 17% for personal services employees and 18% for commercial employees. Among workers, it is higher than 20% for the low-skilled, and even accounts for a quarter of low-skilled artisanal workers. Generally speaking, the social gradient of this extreme distancing from the electoral contest seems far more pronounced than if we only looked at voting in the presidential election, for which eight percentage points separate workers from those in higher grade jobs. Here, they are separated by more than 13 percentage points, since this applies to less than 7% of those in senior jobs.
The effect of precarious employment contracts
40This article seeks to take into account the fragility of subjects’ position within the labor market. My results seem to indicate that the type of employment contract has an influence on electoral participation. The figures allow us to test its effect for certain categories of workers and employees (Table 6).
Employment contract and voting in the 2012 elections (in %)
Employment contract and voting in the 2012 elections (in %)
41For the presidential election – once again setting aside personal services employees, for whom the type of employment contract does not constitute a good indicator of job security – the gap between those on permanent contracts and those in temporary employment varies from five to over ten percentage points. Thus, for employees, we find almost ten percentage points difference separating public servants from commercial employees. For the legislative elections, the difference is still greater: as much as fifteen percentage points for commercial employees and twenty percentage points for the available categories of workers.
In search of the determinants of participation among the working class
42Up to this point, we have considered employees and workers as such, without taking into consideration a certain number of actually rather important structural effects. The first of these is sex, in so far as the division of labor in non-managerial occupations coincides with the gender divide. According to the figures of the Employment Survey, in 2012 77% of employees were women whereas 80% of workers were men. The effect of age is mentioned above: employees are on average older than workers, and within each group there may be important differences (for example, commercial employees are of an average age of 35, against 42 for personal services employees). Differences owing to level of education are also very noticeable: for example, 34% of administrative employees are graduates of higher education as opposed to less than 5% of low-skilled workers. Finally, ethnic origin is a variable that must be taken into account. Certainly, by definition, foreign nationals in active employment are excluded from the field of the survey, but among those surveyed, migration may well be a factor in relation to voting. So as to control for these different dimensions, which we know from elsewhere are closely correlated with electoral participation, I have estimated regression models to identify the principal determinants of voting and abstention. These models must also test for the persistence of the effects of the type of employment contract, which here means that we must once again take into account structural age differences (since we know that there is a strong correlation between age and type of employment contract), sex, and level of education. I have thus estimated regressions seeking to model the probability of abstaining in two rounds of the presidential election (N=1,061), in one of the two rounds of the presidential election (N=2,166), and in two rounds of the legislative elections (N=3,732).
43Apart from the known effects of age, sex, level of education, and size of commune, the results concerning the effect of socio-professional category merit attention, at least concerning abstention in the two rounds of legislative elections: with other characteristics controlled for, commercial and personal services employees have a probability six percentage point higher than public servants of abstaining in both rounds, which once again testifies to their very particular position among employees. Conversely, the results for some categories of workers (in particular skilled workers in logistics and industry) do not differ significantly from those for administrative employees in the private and public sectors.
44Ultimately, whatever the model considered, the type of employment contract still has a significant effect on the probability of abstaining, beyond any effects of age, sex, or education. Those in temporary employment have a significantly higher probability of abstaining than those with a permanent contract, the gap widening from three percentage points for systematic abstention in presidential elections to five percentage points for systematic abstention in legislative elections.
Probability of abstaining in…
Probability of abstaining in…*** at threshold of 1%, ** at threshold of 5%
Note: with other characteristics controlled for, and compared to unqualified individuals, holders of a BEP or a CAP have a 5 percentage point lower probability of abstaining in both rounds of the 2012 legislative elections. This difference is significant at a threshold of 1%.
A more subjective dimension: self-exclusion from political debate
45Turning out to vote is an important indicator of a relationship with politics, yet it is possible to envisage a more subjective dimension, in terms of the degree to which individuals participate in debate around political issues, or even their self-perceived capacity or entitlement to voice an opinion. For the “hidden census” does not only operate on polling day: it can be seen as the result of a deeper process of self-exclusion from political debate, resulting from subjective indifference or incompetence. Following the two explosions in education, what has happened to the subjective relation to political debate of the different segments of the working class? In 2012, although 32% of workers did not have a degree, this was 28 percentage points less than in 1987. Over this period, the number of graduates multiplied fivefold, reaching 20% of workers. Has the widespread diffusion of educational culture allowed workers and employees to better grasp political debate? Subsequent rounds of the ESS, which has been conducted every two years since 2002, do not really put us in a position to respond to this question, because they do not go back far enough. The data does however allow us to sketch out a map of the working classes’ relationship to politics.
Marked self-exclusion by personal services employees
46I have constructed a scale of participation in politics which measures the degree of politicization (or, to put it in negative terms, of self-exclusion from the political arena) of those surveyed on the basis of their responses to the three following questions:
An individual’s score is boiled down to a scale from 0 to 10, and Cronbach’s alpha shows a good correlation between the three dimensions (0.66). The higher the score, the more it testifies to active participation in political debate. If we consider the average score of the fifteen categories of employees and workers presented above, two hubs seem to emerge (Table 8), even if the comparatively small numbers in certain cases suggests that we should be, to say the least, prudent in drawing any conclusions.“Are you interested in politics a lot, somewhat, a little, or not at all?”
“Do you ever end up thinking that politics is so complicated that you cannot really understand what is going on (never, rarely, sometimes, often, very often)?”
“Do you find it easy or difficult to form an opinion on political problems (very difficult, difficult, neither easy nor difficult, easy, very easy)?”
Score for political integration
Score for political integration
47Where self-exclusion is relatively marked, we find personal services employees obtaining the lowest average score, along with laborers in industrial and construction sectors and, to a lesser degree, catering and retail employees. The ISCO classification allows us to distinguish two types of personal services employees, in theory according to their level of qualification. In reality, it is more a matter of a distinction between employees working in personal care (childminders, carers for the elderly) and cleaning and maintenance employees. Considered together, these two groups correspond closely enough to the PCS 56 (personal services workers) category used in the section concerning electoral participation. Here, those occupied in these two types of personal services jobs figure among the least politically integrated, with average scores of less than 4.
48Conversely,  a more integrated hub seems to group together private and public administrative employees, metalworkers, and, rather unexpectedly, drivers, who are the only ones to obtain an average score greater than five.  These few indications underscore the porosity of socio-professional categories: apart from the least skilled fringe, workers are distinguished by lower scores for self-exclusion than most service employees, and in particular personal services employees.
49To verify whether this hierarchy subsists beyond the major structural differences in terms of age, sex, and education, I estimated a regression model bringing in three types of variable: the socio-demographic characteristics of age, sex, level of education, size of the commune of residence, nationality,  and revenue; and variables linked to the nature of the work being done (type of employment contract, size of the employing establishment, profession); two variables bearing on the professional trajectory of the surveyed subjects (past or present membership of a union, experience of a period of unemployment of at least three months in the years preceding the survey). This model was estimated three times, for employees only, for workers only, and for all employees and workers (Table 9).
Score for political integration
Score for political integration*** at threshold of 1%, ** at threshold of 1%, * at threshold of 5%
Other variables controlled for: religion, type of employment contract, experience of unemployment
50An initial and unsurprising result is the strong effect of educational level. With other characteristics controlled for, two percentage points separate workers who have attended higher education from those who have at most a diploma of primary education, and the difference is 1.5 points for employees. We also find that sex has an important effect: beyond the sexual division of labor that reigns among non-managerial employees, women employees and workers are distinguished by their rate of self-exclusion, significantly higher than men in the same professions. Apart from the effect of age (the integration score increases with age), the effect of nationality also merits attention: although the fact of having at least one parent born outside France does not seem to significantly change attitudes, nationality per se does have an effect.
51In effect, foreign workers and employees self-exclude significantly more than the French. Finally, the significant effect of the level of revenue (declared at the household level) should be emphasized: self-exclusion seems to decrease as revenue increases. These results confirm the classic and still close link between the different types of resources that individuals have at their disposal and their degree of politicization.
52A second result is that the variables linked to the nature of the work and professional trajectory of the surveyed subjects do not allow us to assume any significant effect from the type of employment contract, the size of the establishment, or the experience of unemployment. On the other hand, membership (past or present) of a union increases the politicization score significantly (with a particularly marked effect among employees, where the difference is of the order of one percentage point, as opposed to 0.5 points among workers).
53Third, and finally: even when controlling for the effect of all the dimensions mentioned above, there remain significant differences between the fifteen types of professions. For employees, there is still a gap between the worlds of the office, public contact, and retail on one hand, and that of personal services on the other. Whether it is a matter of caring or cleaning, personal services employees are at the lowest level, along with security employees. Among workers, apart from the high score of drivers mentioned above, we find a significantly higher score for metalworkers and for art and precision craft workers. When the model is estimated for all employees and workers, it is industrial laborers and transport workers who, all other things being equal, score the lowest. And yet the score of personal services employees and security employees is not significantly higher, whereas metalworkers are situated at levels comparable to those of administrative employees.
54How are we to interpret these results? Personal services employees, like security employees, carry out their profession in a solitary setting, without the benefit of any work collective in the strong sense of the term. Certainly they may enjoy a relative autonomy at work in so far as there is often no hierarchy in the workplace (this is the case with most personal services employees, who tend to work in private homes); but at the same time, they work alone, isolated, and often without colleagues. Conversely, metalworkers, and industrial workers more generally, are integrated into interpersonal relationships with colleagues, and benefit from the presence of unions in the workplace. The examination of our sample of workers according to the size of workplace tends to confirm this hypothesis. Construction workers include masons, painters, roofers, electricians or plumbers. Just under half of them work in companies of fewer than ten employees. Conversely, 55% of metalworkers are employed in companies of more than 100 employees.
55The opposition between isolation at work and integration into a collective is probably not sufficient on its own to explain these non-negligible differences, resistant to the classic determinants, between different professional universes. Nevertheless, in so far as the relation to politics is partly bound up with professional context, interactions with colleagues and with the hierarchy, we should not underestimate its importance. 
56* * *
57The primary objective of this study was to contribute to the debate around the working classes’ relationship to politics. The results suggest that an analysis in terms of resources remains entirely valid: the various types of cultural and economic capital are still decisive in explaining the politicization of employees and workers. The working class is thus indeed a heterogeneous whole, shot through with divisions which should not be underestimated. The data on electoral participation have demonstrated that there are important variations in participation between different categories of employees and workers. It is important to emphasize that these differences cannot be explained solely by the influence of socio-demographic characteristics, since they persist even when these dimensions are controlled for. In reality, my results underscore the necessity of taking into account the importance of different professional universes. Integration into a stable and established work collective favors political participation. The world of industry, despite its rapid numerical decline, remains in part a world of big companies in which unions are still present, a fact that contributes toward the politicization of its employees. Conversely, the professional isolation of most personal services employees deprives them of this important element of political socialization. Similarly, the persistent deficit in participation among those in temporary employment implicitly highlights the importance of durable inclusion in stable work collectives.  If educational level, age, sex, and revenue are decisive resources, so too is professional integration, for its absence distances the individual from politics. This helps us to understand why, when all the evidence suggests we should consider certain segments of the non-managerial workforce in terms of domination – metalworkers, for example, work in a sector exposed to heightened international competition – these segments are actually the most politicized within the space of the working class. These results, then, underscore the importance of continuing to reflect on the coding and classification of professions, especially as far as subaltern workers are concerned. Although this article deals only with abstention, the links it posits between different professional universes and the relationship to politics are probably relevant for an understanding of voting patterns, especially in a context where the rise of populist extreme-right parties in Europe is continually connected to the destabilization of the working class.
58The second objective of this work was to contribute to the understanding of the process of internal social stratification within the working class. The use of large sample sizes allowed us to bring to light the magnitude of the differences and divisions that run through the working class. As far as electoral participation and self-exclusion from the political field are concerned, the boundary between employees and workers seems quite porous, in so far as those employees on the fringe of the least skilled personal services seem to be in the worst position. As well as their socioeconomic vulnerability (precarity of employment, difficult working conditions, fragmentation of working hours, multiple employers, lack of any real access to training and thus to any real career advancement), as emphasized above their isolation at work renders them almost politically invisible. Now, among both employees and workers, there are certain categories which have seen their share of the active population grow most rapidly. Between 1987 and 2012, the number of employees classed by Insee as “personal services employees” grew from 1.2 million to 2.2 million. During the same period, the number of industrial workers’ jobs, skilled or low-skilled, fell from 3.5 to 2.5 million. Paradoxically, even as the educational level of the population in active employment has risen considerably, the dynamics of employment may have counteracted the expected benefits in terms of citizenship and participation in public debate, in any case among non-managerial workers.
Determinants of non-registration on electoral registers
Probability of not being registered to vote
Probability of not being registered to vote
“Workers” and “employees” have specific meanings in this article, reflecting their usage.in the French classification of socio-professional categories. “Employees” means white-collar workers, both skilled (administrative assistants, commercial secretaries, office workers, social workers, executive assistant, special educational needs teacher, etc.) and low-skilled (childcare assistants, auxiliary nurses, those working in retail, etc.). “Workers” is essentially used to describe those in skilled and semi-skilled (“blue-collar”) manual jobs in industry, building trades etc.
This point has fuelled debate among specialists in electoral sociology since Mark Franklin’s prophetic work The Decline of Class Voting in Britain (Oxford: Clarendon Press, 1985) and the extending of his analysis to other European democracies: Mark Franklin, Thomas Mackie, Henry Valen (eds), Electoral Change: Responses to Evolving Social and Attitudinal Structures in Western Countries (Cambridge: Cambridge University Press, 1992). Taking into account other indicators, and rethinking class and socio-professional divisions, other works draw the contrary conclusion that these divisions are permanent. See in particular Jeroen Van Der Waal, Peter Achterberg, Dick Houtman, “Class is not dead – it has been buried alive: class voting and cultural voting in postwar Western societies (1956-1990)”, Politics and Society, 35(3), 2007, 403-26. For an overview of these debates, see, for example, Nonna Meyer, “Que reste-t-il du vote de classe?”, in Pascal Perrineau, Luc Rouban (eds), La politique en France et en Europe (Paris: Presses de Sciences Po, 2007), 287-310.
Olivier Schwartz, “La notion de ‘classes populaires’”, postdoctoral thesis, Université de Versailles, Saint-Quentin-en-Yvelines, 1998. [All translations from the French are by Cadenza Academic Translations, unless an alternative published English-language source is given.]
For example, on the situation of personal services employees, whose numbers are rapidly growing, see Christelle Avril, Les aides à domicile. Un autre monde populaire (Paris: La Dispute, 2014).
See, for example, Erik Olin Wright, Classes (London: Verso, 1985); David Grusky, Social Stratification: Class, Race, and Gender in Sociological Perspective (Boulder, CO: Westview Press, 1994); and, more recently, Daniel Oesch, “Coming to grips with a changing class structure: an analysis of employment stratification in Britain, Germany, Sweden and Switzerland”, International Sociology, 21(2), 2006, 263-88. Similarly, the whole debate around the classification of socio-professional categories in France, like European classification projects, revolves around the divergent visions of the social structure as a whole.
See in particular Emannuel Pierru, Alexis Spire, “Le crépuscule des catégories socio-professionnelles”, Revue française de science politique, 58(3), 2008, 457-81.
The debates between French and British statisticians provoked by the European project for classification (ESeC) afford us a glimpse into what is at stake in any enterprise of classification. See, for example, Cécile Brousse, “ESeC, projet européen de classification socio-économique”, Courrier des statistiques, 125, 2008, 27-36, Alexandra Filhon et al., “Un projet de nomenclature socio-professionnelle européenne: une construction savante face aux variations nationales de représentation du monde social”, Sociologie, 4(4), 2012, 373–93.
Referring only to the legislative elections following a presidential election: 70% in the first round in 1981, 65% in 1988, 64% in 2002, 60% in 2007, and 58% in 2012.
Anne Muxel, “La mobilisation électorale: l’envers de 2002 et un sursaut généralisé”, Revue française de science politique, 57(3-4), 2007, 315-28.
Céline Braconnier, Jean-Yves Dormagen, “Logiques de mobilisation et inégalités sociales de participation électorale en France, 2002-2012”, French Politics, 30(3), 2012, 20-44.
Daniel Gaxie, Le cens caché. Inégalités culturelles et ségrégation politique (Paris: Seuil, 1978).
Pierre Bourdieu, La distinction. Critique sociale de jugement (Paris: Minuit, 1979).
Cf. Frédéric Sawicki, Annie Collovald, “Le populaire et le politique: quelques pistes de recherche en guise d’introduction”, Politix, 13(4), 1991, 7-20; Olivier Schwartz, “Sur le rapport des ouvriers du Nord à la politique: matériaux lacunaires”, Politix, 13(4), 1991, 79-86.
Sidney Verba, Kay Lehman Schlozman, Henry Brady, Voice and Equality: Civic Voluntarism in American Politics (Cambridge, MA: Harvard University Press, 1995); Céline Braconnier, Jean-Yves Dormagen, La démocratie de l’abstention. Aux origines de la démobilisation électorale en milieux populaires (Paris: Gallimard, 2007).
Nonna Meyer, Sociologie des comportements politiques (Paris: Armand Colin, 2010), 181.
François Héran, “Les intermittences du vote: un bilan de la participation de 1995 à 1997”, Insee Première, 546, 1997; and more recently Braconnier and Dormagen, “Logique de mobilisation”.
Anne Muxel, “La participation politique des jeunes: soubresauts, fractures et ajustements”, Revue française de science politique, 52(5), 2002, 521-44.
Mark Franklin, Voter Turnout and the Dynamics of Electoral Competition in Established Democracies Since 1945 (Cambridge: Cambridge University Press, 2004).
For a case concerning a poor Brazilian district, see Céline Braconnier, Jean-Yves Dormagen, Daniella Rocha, “Quand les milieux populaires se rendent aux urnes: mobilisation électorale dans un quartier pauvre de Brasilia”, Revue française de science politique, 63(3), 2013, 487-518.
Camille Peugny, Le destin au berceau. Inégalités et reproduction sociale (Paris: Seuil, 2013).
Viviane Le Hay, Camille Peugny, “Fragilités et fractures de la société française”, in Vincent Tiberj (ed.), Des votes et des voix. De Mitterrand à Hollande (Paris: Champ Social Éditions, 2013), 14-24.
Thomas Amossé, Olivier Chardon, “Les travailleurs non qualifiés: une nouvelle classe sociale?”, Économie et statistique, 393(1), 2006, 203-29.
Yves Jauneau, “Les employés et ouvriers non qualifiés: un niveau de vie inférieur d’un quart à la moyenne des salariés”, Insee Première, 1250, 2009.
Philippe Alonzo, Olivier Chardon, “Quelle carrière professionnelle pour les salariés non qualifiés?”, Données sociales, 2006, 265-72.
Cf. Anne-Catherine Wagner, Les classes sociales dans la mondialisation (Paris: La Découverte, 2007).
Cf. Hanspeter Kriesi, Edgar Grande, Romain Lachat et al., West European Politics in the Age of Globalization (Cambridge: Cambridge University Press, 2008).
Patrick Emmenegger, Silja Haüsermann, Bruno Palier, Martin Seeleib-Kaiser (eds), The Age of Dualization: The Changing Face of Inequality in Deindustrializing Societies (New York: Oxford University Press, 2012).
David Rueda, “Insider-outsider politics in industrialized democracies: the challenge to social democratic parties”, American Political Science Review, 99(1), 2005, 61-74.
Kim Weeden, David Grusky, “The case for a new class map”, American Journal of Sociology, 111(1), 2005, 141-212.
Maria Oskarson, “Social risk, policy dissatisfaction and political alienation: a comparison of six European countries”, in Stefan Svallors (ed.), The Political Sociology of the Welfare State: Institutions, Social Cleavages and Orientations (Stanford, CA: Stanford University Press, 2007), 117-48.
Alain Chenu, L’archipel des employés (Paris: Insee, 1990).
Yasmine Siblot, Faire valoir ses droits au quotidien. Les services publics dans les quartiers populaires (Paris: Presses de Sciences Po, 2006), 131.
François-Xavier Devetter, Sandrine Rousseau, Du balai. Essai sur le ménage à domicile et le retour de la domesticité (Paris: Liber-Raisons d’agir, 2011).
The type of employment contract is a poor indicator of job quality or integration within the labor market for these categories of occupation, where individuals are very often employed on a permanent contract (as is the case for kindergarten staff and childminders), even when the working week, as well as comprising fewer hours than the average working week, is parceled out between a number of employees (as is the case for cleaning staff).
The case of police and military employees should be mentioned here. Even though military staff were granted the right to vote in 1945, the low rate of turnout among them is probably connected to the history and specific nature of military employment.
The difference between these two hubs is statistically significant (test T for difference of averages).
How is this result to be interpreted? Beyond the fact of the relatively low numbers in certain cases included in the sample, which should urge us to prudence in analysis, one hypothesis is the strong exposure of these professions to information of a political nature as a result of their working conditions. In particular, listening to the radio probably constitutes an important channel for drivers to inform themselves on political matters. Recent studies have shown how exposure to information, far from being solely the result of active and intentional attempts to inform oneself, was largely dependent upon the effects of the media. See Charlotte Dolez’s doctoral thesis, “L’écume des news: sociologie politique des usages des informations”, Sciences Po Paris, 2013.
The ESS gives us the nationality of the subject and the country of birth of both of their parents. We have therefore constructed this as a three-value variable: foreign nationality, French nationality with both parents born in France, and French nationality with at least one parent born elsewhere. This last category allows us to take into account, albeit imperfectly, the situation of French citizens who are “children of immigrants”.
Guy Michelat, Michel Simon, “Déterminations socio-économiques, organisations symboliques et comportement électoral”, Revue française de sociologie, 26(1), 1985, 32-69. The role of work as an authority within political socialization and a site of politicization is highlighted in many studies. For a recent insight see, for example, Ivan Sainsaulieu, Muriel Surdez (eds), Sens politiques du travail (Paris: Armand Colin, 2012).
This link between modes of professional integration and political attitudes is made in many works. See, for example, Serge Paugam, “Formes d’intégration professionnelle et attitudes syndicales et politiques”, Revue française de sociologie, 40(4), 1999, 715-51.