Determinants of Electoral Non-Registration and Sensitive Neighbourhoods in France Jean-Louis PAN K É S HON*

The abstention rate in French elections has been rising steadily for the past twenty years. While many commentators have inveighed against this self-disenfranchisement and the crisis it is causing in representative democracy, another group that has shunned civic commitment in a much more fundamental way — the non-registered voters — has not elicited such criticisms. One singular feature of the French electoral system is that citizens who wish to vote must actively register their eligibility on an electoral register. There is no penalty for failure to do so as there is in Belgium or Greece (Boy and Mayer, 1997a). The voting qualifications are: being 18 years of age, a French national — nationals of other European Union countries can register on supplementary lists for local elections and elections to the European Parliament —, not being under a legal disability, and proving a firm connection with the municipality of registration (i.e. either having one’s legal address or at least six months’ continuous and effective residence there, or having paid the land tax, community charge, or business use tax there for at least five years previously). Since 2001, however, town halls have had a duty to automatically register young people who will be aged 18 at the time of an election. Registration must be carried out before the year of the election, and is then permanent but subject to revision. Registers are, at least in theory, updated annually. Electors who change address, including within the same municipality, must be removed from the register if they fail to re-register at their new address. The non-registered population thus consists of eligible voters who have not registered, as well as voters removed from the register for reasons other than legal disability and who have not re-registered. In all, non-registration results from the following circumstances: accidental or intentional failure to register, and, congruently, accidental or intentional failure to re-register. Non-registrants can, therefore, be presumed to form a heterogeneous group. Non-registrants are often classed with abstentionists because of their common failure to participate in elections, but they differ in several respects. The first part of this study reviews the differences between nonregistration and abstention. It then goes on to consider the explanatory factors of non-registration using data from the 2001 Neighbourhood Life Survey ( Enquête Vie de Quartier ). With this survey, individual characteris-

* Institut National de la Statistique et des Études Économiques, Paris. Translated by Glenn D. Robertson. Population-E 2004, 59(1), 143-156

144

J.-L. PAN KÉ SHON

tics, not usually available simultaneously, can be compared in a single model that evaluates the specific ceteris paribus effects of each on nonregistration. This ad hoc survey, included as part of the continuous surveys on household living conditions (EPCV) established by INSEE, set out to study individual disparities in the various socio-economic categories of neighbourhood (1). To facilitate comparison at the extremes, the residents of high-income and low-income neighbourhoods were over-represented in the sample. Data collection took place between April and June 2001 on a sample of 11,919 people. Beyond the influence of individual characteristics, it is legitimate to wonder whether the concentration of socially disadvantaged groups in one place creates an effect specific to the neighbourhood which would increase non-registration. The premise here is that a sense of being second-class citizens makes residents of poor neighbourhoods indifferent to political discussions they see as irrelevant to their problems, and leads them to disengage from electoral life. The existence of a potential “socially disadvantaged neighbourhood effect” on electoral registration will be tested for in the third section, using individual-level data from the Neighbourhood Life Survey.

I. Non-registration is different to abstention Both abstention and non-registration concern groups who are younger and lower-income than the voting-age population as a whole (see in particular, Gaxie, 1978; Morin, 1983; Percheron et al., 1987). But among all registered voters, young people have the highest participation rates in legislative and presidential elections — local elections excite less interest at that age — which decline rapidly after the age of 20 and begin to climb again only as they approach their thirties (Héran, 1997). Young people also have the lowest registration rate (Héran and Rouault, 1995a). This contradiction in behaviour between young non-registrants and young abstentionists is only apparent and is based on a selection effect: young registrants are more highly-motivated, and so logically have a higher turnout rate. Another “explanation” for abstention is the circumstances of the vote in terms of the range of electoral options on offer, and the degree of uncertainty surrounding the outcome. Where the candidates represent a broad spectrum of political views, turnout among their supporters is higher. Just as a closely contested election between two candidates mobilizes voters (Héran, 1997), so too does the urge to block a candidate, as happened in the second round of the French presidential election in 2002. Abstention is also related to the hierarchy of elections: turnout is higher for presidential than legislative elections, and for legislative than local elections, while participation in referendums varies with the type of question asked. (1) The survey was conducted in partnership with the Délégation Interministérielle à la Ville (DIV), Plan Urbanisme Conception Architecture (PUCA) of the Ministère de l’Équipement, Observatoire National de la Pauvreté et de l’Exclusion Sociale, Direction de la Recherche, des Études et des Statistiques (DARES) and Direction de la Recherche, des Études, de l’Évaluation et des Statistiques (DREES) of the Ministère du Travail et de la Santé, Union Nationale des HLM, Institut des Hautes Études sur la Sécurité Intérieure (IHESI) of the Ministère de l’Intérieur and the Caisse Nationale d’Allocations Familiales (CNAF).

ELECTORAL NON-REGISTRATION AND SENSITIVE NEIGHBOURHOODS IN FRANCE

145

Unlike for abstention, election type seems to have little influence on registration levels, since the proportion of non-registered potential voters fell by only 0.2 points, from 10.0% to 9.8%, between the 2001 local elections and the 2002 presidential election (Table 1) (2). Non-registration also differs from abstention in being unaffected by the circumstances of the vote, which are not known at the time when potential voters are authorized to register: nomination papers have not yet been lodged, and not all candidates have declared. Finally, abstention rates have risen over the past twenty years, while non-registration rates have remained remarkably stable, or even declined very slightly over the same period, which would suggest that they are distinct phenomena. In fact, the proportion of non-registrants in 1983 was 11.3% against 10% in 2001, compared with a rise in the local election abstention rate from 21.6% to 32.7% over the same period… (Lehingue, 2001). The apparent stability of the non-registration rate may conceal divergent trends: it could be inferred from rising qualification levels and population aging that these factors may have been behind the higher registration levels in France between 1983 and 2002, while other factors may have pulled in the opposite direction. It would in any case be difficult to explain away the lack of variation over that twenty-year period whereas abstention levels have risen. One reason for the limited fluctuations in non-registration compared with abstention is the stock effect that dampens volatility in the former. This is because once registration has taken place, it does not have to be renewed, unless there is a change of address, and deregistration is not possible. So, a stock of registered voters is progressively built up, an unknown number of whom are captive, which is not the case for abstention. The number of registered voters can then vary only marginally as a result of the new young potential voters and those changing address who must re-register, which limits the fluctuations. TABLE 1.– TREND IN NON-REGISTRATION AND ABSTENTION RATES, 1983-2002 (%) Presidential and legislative elections

Local elections

Non-registrants(a) Abstentionists(b)

1983 (all France)

1995 (metropolitan France)

2001 (all France)

2002 (metropolitan France)

11.3 21.6

9.2 30.6

10.0 32.7

9.8 –

(a) Among

potential voters. registered voters. Reading: In the 1983 local elections, 11.3 % of the potential electorate was not registered on an electoral register and 21.6% of registered voters abstained. Sources: (a) For 1983, 1995 and 2002: INSEE, comparison of the Demographic Longitudinal Sample (Échantillon Démographique Permanent – EDP) and electoral registration records; for 2001: calculation from INSEE data of the estimated total population at 1 January and the number of ineligible voters in 2001, and Ministry of the Interior data for registrations at 1 March 2001. (b) Ministry of the Interior, election results. (b) Among

(2) And it should be noted that this difference may also result from the two different types of calculation as explained in the notes to Table 1.

146

J.-L. PAN KÉ SHON

Finally, unlike non-registration which equates to total non-participation, most abstentions are occasional. Persistent abstention (in both rounds of an election) concerned only 11% of the potential electorate for the two elections (presidential and local) held in 1995 combined (Héran, 1997). Conversely, when two elections fall in the same year, only approximately half of the registered voters take part in all four rounds. Thus the 1995 presidential and local elections combined attracted 55% of “persistent participants” (Héran and Rouault, 1995b), against 47% in the 2002 presidential and legislative elections (Clanché, 2002).

II. The main causes of non-registration Extensive research into electoral participation has revealed the complexity of the factors in play (Michelat and Simon, 1977; Gaxie, 1978; Percheron et al., 1987; Subileau and Toinet, 1989; Mayer and Percheron, 1990; Ysmal, 1990; Mayer and Perrineau, 1992; Mayer, 1997, etc). Using the data on a large number of individual characteristics from the Neighbourhood Life Survey, we estimated the probabilities of noninclusion on the electoral registers in 2001 as a function of sociodemographic characteristics. The results of the logistic regression analysis to assess the ceteris paribus effects are outlined in Table 2.

1. Educational level and cultural behaviour Firstly, a low educational level makes non-registration more likely. In 2001, the probability of electoral non-registration is 8.2 points higher among people with no educational qualification than among those with more than 2 years of higher education ceteris paribus (Table 2). In a general sense, the effect of educational level can be interpreted as reflecting a good understanding of democratic issues, or the internalization of a personal competence — or lack of it — necessary to engage in political expression (Bourdieu, 1979; Gaxie, 1978). Apart from educational level, certain cultural characteristics encourage involvement in the public sphere and in particular participation in elections. Previous studies have shown that differences in behaviour between public and private sector employees (de Singly and Thélot, 1988) derive as much from a pre-selection of like-minded individuals by a public institution as from the imposition on those individuals of a culture of public service and efficient government. Public sector employees have clearly defined positions in the democratic debate (Boy and Mayer, 1997b). They are also significantly distinguished from private sector employees by their greater civic involvement in the form of electoral registration: the probability of non-registration is 2.3 points lower among public than private sector employees ceteris paribus (Table 2). It should be stressed that the variables in the model include educational level and perceived religious affiliation, and that this result is therefore obtained by controlling for these variables. It can be interpreted as the result both of public employees’ involvement in the running of State business, and of the selection of individuals with a sense of civic responsibility who gravitate towards public service occupations.

ELECTORAL NON-REGISTRATION AND SENSITIVE NEIGHBOURHOODS IN FRANCE

147

2. Weaker social integration The degree of social integration significantly affects the registration versus non-registration decision. Thus membership of an organized group, especially a political party, trade union or more generally a “civic association”, tends to favour registration. While membership of this kind of association indicates the individual’s desire to influence civic life and logically results in a higher registration rate, membership of other kinds of association (parent-teacher, ex-servicemen, sports) also has a significant effect that is variable according to the type of association(3). More generally, belonging to a sociability network (group of friends or colleagues) also has a positive effect on registration. Being part of any kind of network entails a variety of interpersonal exchanges, and leads individuals to make up their minds and, ultimately, to want to have their voice heard. It is also the vector for a regimentation and moral “control” by the group over the individual, and accepting that control is certainly less costly than rejecting it (Percheron et al., 1987). The specific influence on electoral registration of belonging to a network would then derive from the logics of stimulation and control of the group. In addition, residential stability is conducive to electoral registration. The probability of non-registration falls sharply as length of occupancy in a dwelling increases. Likewise, home ownership indicates an attachment to locality that is reflected in a higher frequency of electoral participation. The frequency of registration rises with age; the lower frequency of registration among young people is what prompted French legislators to introduce automatic registration. Interpreting this concept of age is never straightforward, since what is involved here is not biological aging but rather the progressive entry of young people into the adult world. At the age of legal majority they have to assume their institutional role as adults, and not all are ready for it. In this respect, it is relevant to note that adult children still living with their parents have a probability of non-registration that is 4.2 points higher than that of other household members. An alternative and more mundane interpretation is that of registration inertia. As noted earlier, an individual who never changes address may have occasion to register once only during his or her lifetime. So the much higher frequency of registration among elderly adults would be due to an accumulation or stock effect. Likewise, the most highly mobile individuals themselves create the circumstances that necessitate their re-registration. Recent arrivals in a municipality are thus at risk of being at least temporarily non-registered, with a higher risk among tenants than homeowners who by definition have lower residential mobility. But this postulate alone cannot account for the effects of educational level, network membership, sex or country of birth. Persons born abroad are less often registered than those born in France ceteris paribus . Does this participation gap reflect the lack of a parental model, where parents settled in France are often foreigners with no right to vote (except in local and European elections for EU nationals)? It is singular to see that French nationals born in Africa have a lower (3) Membership of a commonhold management committee does not affect the probability of registration on the electoral registers, so this variable has not been included in the model presented here.

148

J.-L. PAN KÉ SHON TABLE 2.– ESTIMATED PROBABILITIES OF ELECTORAL NON-REGISTRATION IN THE 2001 LOCAL ELECTIONS Variables

Sex Male Female Age 18-21 22-25 26-34 35-44 45-64 65 and over Educational attainment None Primary school, lower secondary Lower-level vocational, higher-level vocational, upper secondary 2 years higher education More than 2 years higher education Individual’s country of birth France (metropolitan France) African country Other country Position in household Child of couple Other Household size Less than six people Six or more people Annual income per SU(a) Under 20,000 F 20,000-30,000 F 30,000-70,000 F 70,000-100,000 F 100,000-240,000 F Above 240,000 F Housing occupancy status Owner, free lodging Tenant Duration of housing occupancy 0-3 years 4-10 years Over 10 years Not given

Deviation from reference probability (in points) Ref. – 1.4*** 6.6*** 3.4*** 1.9** Ref. – 3.5*** – 5.8*** 5.4*** 3.5*** Ref. n.s. – 2.8*** Ref. 8.0*** 10.8*** 4.2*** Ref. Ref. 4.9*** n.s. 2.4* Ref. n.s. – 1.1* – 2.5* Ref. 5.2*** 11.4*** 4.5*** Ref. 10.5***

probability of non-registration than those born in other countries. If confirmed by future studies, this finding could reflect the ties that persist between the former colonies and France, and that find practical form in a higher registration rate.

ELECTORAL NON-REGISTRATION AND SENSITIVE NEIGHBOURHOODS IN FRANCE

149

TABLE 2 (CONT.).– ESTIMATED PROBABILITIES OF ELECTORAL NON-REGISTRATION IN THE 2001 LOCAL ELECTIONS Variables Activity status Unemployed Homemaker Employed Student Pensioner Activity sector Public Private Member of an association Yes, civic(b) No Yes, ex-servicemen No Yes, sports No Yes, parent-teacher No Religious affiliation Yes(c) No Refused to answer In a group of friends Yes No Living in a neighbourhood classified as a sensitive urban zone (ZUS) Yes No Reference value

Deviation from reference probability (in points) 2.7*** 2.0* Ref. n.s. – 2.8** – 2.3*** Ref. – 5.7*** Ref. – 5.0** Ref. – 2.3*** Ref. – 4.0*** Ref. Ref. 2.8*** 2.6* – 1.7*** Ref.

1.6** Ref. 8.6 %

(a) Income per spending unit (SU) takes account of household structure (the first person counts as 1 unit, those aged 14 and over as 0.5 unit, and those aged under 14 as 0.3 unit). Incomes are expressed in French francs (1 euro = 6.55957 francs). (b) Political party, trade union, environmental protection group, etc. (c) Covers regular worship, occasional worship, feelings of religious affiliation. Deviation significant at level of: ***: 1%, **: 5%, *: 10%; n.s. = not significant. Reading: Individuals whose characteristics correspond for each variable to the reference situation (male, aged 3544, etc) have an 8.6% probability of non-registration on the electoral registers at the time of the 2001 local elections. The probability of non-registration among people aged 18-21 is 6.6 points higher than that of individuals aged 35-44, ceteris paribus. Note: Calculated for potential voters in metropolitan France. Source: INSEE, Neighbourhood Life Survey, EPCV, April-June 2001.

3. Social status Finally, the probability of electoral registration versus non-registration varies according to the social status of individuals. Hence registration rates are lower among the unemployed than among the employed. Membership of a large family (6 people or more) — which does not signify direct experience

150

J.-L. PAN KÉ SHON

of social disadvantage but does increase the risk of belonging to a lowincome family — is also more likely to lead to higher non-registration. Apart from the people with the lowest incomes, a proportion of whom are students, individuals with low incomes are less often registered; high earners, by contrast, are more likely to be registered. Overall, unemployment, low income and large family membership are all factors that significantly affect, ceteris paribus , the probability of electoral non-registration, though are less determinant than educational level and degree of social integration (Table 2).

III. Is there a specific neighbourhood effect on electoral non-registration in sensitive urban zones? Social problems tend to be concentrated in poorer neighbourhoods. “Why specifically inquire into citizenship in relation to the least socially developed neighbourhoods?” asked Henri Rey (1999), going on to list three possible points of inquiry: is there an observable lack of civic responsibility? forms of dissent? is there greater civic involvement in these areas? But where then could that lack or dissent originate from? The question often arises of whether people’s social characteristics alone “explain” individual behaviour — in this case, registration — or whether living in a socially disadvantaged neighbourhood increases the disparities (for an introduction and review of the literature on “neighbourhood effects”, see Marpsat and Laurent, 1997; Marpsat, 1999). Supporters of the theory argue that the effect stems from a counter-culture or counter-norm produced by an adaptation to local models, transmitted through social emulation and which, in this particular case, would cause lower participation in democratic institutions.

1. Method and formalization The Neighbourhood Life Survey 2001 collected information on residence in a district classified as a sensitive urban zone (Zone urbaine sensible – ZUS). ZUS classification is mainly for administrative purposes and is intended to implement economic and administrative measures in these neighbourhoods as part of “urban policy” ( “la politique de la ville” ). The list is decided by the Interdepartmental Commission for Urban Policy working with prefectures and local authorities (751 ZUS have been defined to date). Selection is mainly of large outlying neighbourhoods based on their poorer socio-economic profiles relative to the city or conurbation they belong to, and on the priorities set in “urban development contracts”. It is therefore a relative classification from which no absolute national scale of neighbourhood deprivation can be derived; this means that a sensitive urban zone in a prosperous region could be declassified were it to be in a poorer region. Nonetheless, it is invariably disadvantaged neighbourhoods that are concerned by the ZUS classification. Table 2 shows that living in a neighbourhood classified as a sensitive urban zone increases the probability of voter non-registration ceteris pari-

ELECTORAL NON-REGISTRATION AND SENSITIVE NEIGHBOURHOODS IN FRANCE

151

bus . The effect is limited but significant at the 5% level. If the analysis went no further, we could therefore conclude for the presence of a neighbourhood effect. But this result may be related to characteristics not included in the model, or to the effect of a hidden variable picked up by the ZUS variable. If so, to attribute the effect to the ZUS would be incorrect. To overcome this problem, a second logistic model can be used to test, for each of an individual’s characteristics, the significance of the observed differentials by whether or not they live in a ZUS-classified districts so as to reveal any characteristics specific to the inhabitants of ZUSs (the method is based on that of Olivier Godechot, 2000). For this we calculate: j=n j=n   P ( y i = 1 ) = 1 + exp –  a + a′1 z + ∑ b jz x jiz + ∑ b jz x jiz + u i 2 1 1 2 2   j=1 j=1

–1

where — P ( y i ) is the probability of electoral non-registration; — b jz is the estimated parameter associated with the variable x j in 1

the population z 1 (non-ZUS); — x jiz is the value of the variable xj for individual i in the popula1

tion z 1; — 1 z is an indicator for the population z 2 (in ZUS); 2

— a is the value of the constant for population z 1 and a + a′ is the value of the constant for population z 2; — u i is the residual of the model. A large number of significant differences between ZUS and non-ZUS residents would suggest the existence of a neighbourhood effect. But it will be hard to prove for certain, since it could be argued that ZUS neighbourhoods contain a higher concentration of characteristics not included in the model, in which case the variables crossed with the ZUS variable would pick up that information instead of a specific ZUS-residence effect (see Marpsat, 1999). It is unlikely, however, that one or more hidden variables could affect all the results of the regression analysis. Conversely, a finding of nonsignificant differentials between the logistic regression parameters would demonstrate the absence of a neighbourhood effect for the ZUSs based on the data from the Neighbourhood Life Survey. What is the real answer?

2. Rare, weak and doubtful effects The only significant differences between residents in the two types of neighbourhood are observed for women, persons who are disabled and beneficiaries of a survivor’s pension (significant at the 5% level) on the one hand, and for young people aged 18-24, and people reporting a perceived religious affiliation (at the 10% level) on the other hand (Table 3).

TABLE 3.– ESTIMATED PROBABILITIES OF ELECTORAL NON-REGISTRATION FOR THE 2001 LOCAL ELECTIONS BY RESIDENCE IN A ZUS OR NON-ZUS NEIGHBOURHOOD, AND SIGNIFICANCE LEVEL OF THE PROBABILITY DIFFERENCES

Variables(a)

Ref. – 1.8

ZUS

Significance level

***

Ref. 0.1

Significance level

Significance of difference in probabilities

n.s.

**

18.1 8.2 4.7 Ref. – 3.1

*** *** ***

* n.s. n.s.

**

n.s.

*** *

n.s. n.s.

** ***

5.7 2.2 Ref. – 2.6 – 2.5

* *

n.s. n.s.

*** *

Ref. – 0.3 1.0

n.s. n.s.

n.s. n.s.

6.2 Ref. –

***

n.s.

n.s.

–

2.3 Ref. – 2.3

**

n.s.

***

n.s.

n.s. *** ** n.s.

n.s. ** n.s. n.s.

11.0 7.3 3.5 Ref. – 2.9

*** *** ***

6.0 3.3 Ref. – 1.5 – 3.0

*** ***

Ref. – 1.8 3.4

Deviation from reference (in points)

***

4.2 Ref. 23.5

***

3.9 Ref. – 3.3

***

3.4 3.4 8.8 7.6 Ref.

*** ** *** ***

***

***

0.6 11.9 4.7 3.0 Ref.

J.-L. PAN KÉ SHON

Sex Male Female Age 18-24 25-34 35-44 45-64 65 and over Educational attainment None Primary school, lower secondary Lower-level vocational, higher-level vocational, upper secondary 2 years higher education More than 2 years higher education Annual income per SU Under 20,000 F 240,000 F or more Not stated Housing occupancy status Tenant Owner Other Duration of housing occupancy 0-3 years 4-10 years Over 10 years Activity status Employed Disabled, beneficiary of survivor’s pension Unemployed Homemaker Pensioner, student

Deviation from reference (in points)

152

Non-ZUS

Father’s country of birth Metropolitan France Dom-Tom Spain, Italy, Portugal, Greece, Turkey, Malta Other European country Africa Other country Household type or position in household Isolated individual Childless couple Adult of a couple with children Child of couple Single-parent family Other household type Household size Less than six people Six people or more Religious affiliation Yes No Member of a civic association Yes No Iris indicator of insufficiently occupied dwellings in the neighbourhood(b) Yes No Reference value

Non-ZUS ZUS Significance of difference Deviation from Deviation from Significance Significance in probabilities reference reference level level (in points) (in points) – 1.8 6.4 7.3 Ref. 3.8 21.7

*** * *** ** ***

– 0.8 9.5 9.3 Ref. 2.3 13.9

1.2 Ref. 1.4 6.9 3.3 4.6

* *** *** **

1.3 Ref. 0.3 1.9 2.6 1.0

Ref. 3.6

**

Ref. 6.2

n.s. ** **

n.s. n.s. n.s.

n.s. **

n.s. n.s.

n.s.

n.s.

n.s. n.s. n.s. n.s.

n.s. n.s. n.s. n.s.

***

n.s.

– 2.4 Ref.

***

– 1.1 Ref.

n.s.

*

– 4.9 Ref.

***

– 3.8 Ref.

***

n.s.

– 1.5 Ref. 8.0 %

***

– 1.3 Ref. 7.4 %

n.s.

n.s.

ELECTORAL NON-REGISTRATION AND SENSITIVE NEIGHBOURHOODS IN FRANCE

Variables(a)

(a)

153

For the definition of certain variables, see Table 2. (b) Indicator based on a sub-municipal zoning used by INSEE. Dwellings in the district are regarded as insufficiently occupied where the ratio of the number of occupants to number of rooms is below 0.8. Deviation significant at the level of: ***: 1 %, **: 5 %, *:10 %; n.s. = not significant. Reading: ceteris paribus, the probability of electoral non-registration among young people aged 25-34 is higher than that of people aged 45-64 (+7.3 points in non-ZUS neighbourhoods and +8.2 points in ZUSs, the deviations being significant at the 1% level). But the difference in the probabilities of non-registration among young people aged 25-34 according to whether or not they live in a ZUS is not statistically significant (last column). Note: Calculated for potential voters in metropolitan France. Source: INSEE, Neighbourhood Life Survey, EPCV, April-June 2001, plus regional extension of the survey in Brittany.

154

J.-L. PAN KÉ SHON

If in general perceived religious affiliation is associated with a higher registration rate, this is even more the case in non-ZUS than ZUS neighbourhoods. All that can be put forward here are hypotheses, especially since the differences are significant only at the 10% level, which make for less than robust results. It must first be stressed that this variable does not pick up the father’s birth country effect, since this information is present as such in the regression. The difference may come from the relative shares of the various religions, which probably varies with the social composition of the neighbourhoods, and from their differential effects. For example, Durkheim (1995) noted in Le suicide that propensity to suicide varied between Catholics, Protestants and Jews. Weber (1964) attributed the earlier emergence of capitalism in the English-speaking countries to the Protestant ethic. In a similar way, it is conceivable that the relative share of the different religions by type of neighbourhood influences the probability of voter registration. This hypothesis would obviously require closer examination using more detailed data. The frequency of non-registration among young people under 25 is higher in ZUS than in non-ZUS neighbourhoods. But this finding is hard to interpret, especially since the difference has a low level of significance. It is perhaps due entirely to characteristics that are unobserved or statistically unobservable and that bias the evaluation. As regards non-economically active persons in receipt of a survivor’s pension and persons who are disabled, the deviation probably stems from different relative shares of these two groups by type of neighbourhood. To the question “Do you have health problems or disabilities that prevent you from working or studying?”, 4.8% of people living in non-ZUS neighbourhoods responded “yes, all the time” or “yes, often”, as against 8.3% in ZUS neighbourhoods. Reduced mobility can lead to a disengagement from electoral life. Finally, there is a higher rate of female than male registration in nonZUS neighbourhoods, compared with undifferentiated female/male behaviour in ZUS neighbourhoods. In this case, too, we can do no more than put forward hypotheses that are uncertain and open to criticism. The result may, for instance, be due to disparities in cultural capital not captured by the educational attainment variable and that depend on whether women live in ZUS or non-ZUS neighbourhoods. Another hypothesis is that women who live in ZUS neighbourhoods have a more traditional pattern of behaviour than those living elsewhere. In this case, the difference observed would stem from a selection effect rather than from an effect specific to the neighbourhood. The foregoing remarks, combined with the non-significance of the many other variables in the model, lead to concluding for the lack of a specific “ZUS neighbourhood” effect on non-registration, at least as this can be determined using data from the Neighbourhood Life Survey. Where electoral registration is concerned, therefore, the hypothesis of dissenting behaviour or impaired citizenship specifically attributable to living in ZUSs cannot be confirmed. Differences in behaviour over electoral registration would thus appear to originate in the individual characteristics of the inhabitants.

ELECTORAL NON-REGISTRATION AND SENSITIVE NEIGHBOURHOODS IN FRANCE

155

However, ZUS neighbourhoods are considered here as a generality, although by definition they differ clearly in terms both of the difficulties they face and their socio-demographic make-up. Aggregating the ZUSs smoothes out their differences and “averages” the results. The very few studies conducted on electoral abstention among the inhabitants of ZUS neighbourhoods have highlighted sharp differences in behaviour between the residents of ZUSs in the same region, and even in the same town (Oger, 1995). In this perspective, what emerges from this study is more the lack of validity of the ZUSs as a relevant statistical category for determining the effects of socially disadvantaged neighbourhoods, rather than evidence of the social organization within these neighbourhoods as detected through electoral registration. REFERENCES BOURDIEU Pierre, 1979, La distinction : critique sociale du jugement (Le sens commun), Paris, Éditions de Minuit, pp. 463-485. B OY Daniel, MAYER Nonna (dir.), 1997a,“Les formes de la participation”, in L’électeur a ses raisons (Références inédites), Presses de la Fondation Nationale des Sciences Politiques, pp. 25-65. BOY Daniel, MAYER Nonna (dir.), 1997b, “Secteur public contre secteur privé : un nouveau conflit de classe ?”, in Les modèles explicatifs du vote : un bilan des études électorales en France (Logiques Politiques), Éditions de L’Harmattan, pp. 111-131. C LANCHÉ François, 2002, “La participation électorale au printemps 2002 – De plus en plus de votants intermittents”, Insee première, No. 877, INSEE. D URKHEIM Émile, 1995, Le suicide : une étude de sociologie, Puf (8th edition, Quadrige). G AXIE Daniel, 1978, Le cens caché : inégalités culturelles et ségrégation politique, Paris, Seuil. G ODECHOT Olivier, 2000, Plus d’amis, plus proches. Essai de comparaison de 2 enquêtes peu comparables, Working document, No. 0004, INSEE. H ÉRAN François, 1997, “Les intermittences du vote – Un bilan de la participation de 1995 à 1997”, Insee première, No. 546, INSEE. H ÉRAN François, ROUAULT Dominique, 1995a, “La présidentielle à contre-jour, abstentionnistes et non-inscrits”, Insee première, No. 397, INSEE. H ÉRAN François, ROUAULT Dominique, 1995b, “La double élection de 1995 : exclusion sociale et stratégie d’abstention”, Insee première, No. 414. LEHINGUE Patrick, 2001, “Faire parler d’une seule voix ? Les scrutins municipaux des 1118 mars 2001”, Regards sur l’actualité, La Documentation française, pp. 3-18. M ARPSAT Maryse, 1999, “La modélisation des “effets de quartier” aux États-Unis”, Population, 54(2), pp. 303-330. M ARPSAT Maryse, LAURENT Raphaël, 1997, “Le chômage des jeunes est-il aggravé par l’appartenance à un quartier en difficulté ?”, in En marge de la ville, au cœur de la société : ces quartiers dont on parle, Éditions de l’aube, pp. 321-348. M ARTIN Pierre, 2001, “Les élections municipales en France”, Notes et études documentaires, No. 5, pp. 126-127. M AYER Nonna (dir.), 1997, Les modèles explicatifs du vote : un bilan des études électorales en France (Logiques Politiques), Éditions de L’Harmattan. M AYER Nonna, P ERCHERON Annick, 1990, “Les absents du jeu électoral”, Données sociales, INSEE, pp. 398-401. M AYER Nonna, P ERRINEAU Pascal, 1992, Les comportements politiques, (Cursus, série « Science politique »), Armand Colin. M ICHELAT Guy, Simon Michel, 1977, Classe, religion et comportement politique, Paris, Presses de la Fondation nationale des sciences politiques. M ORIN Jean, 1983, “Un Français sur dix ne s’inscrit pas sur les listes électorales”, Économie & statistique, INSEE, No. 152, pp. 31-37. O GER Pascal, 1995,“L’abstention aux élections dans les quartiers prioritaires des contrats de ville”, Indicateurs de l’économie du Centre, No. 8, INSEE, pp. 17-20.

156

J.-L. PAN KÉ SHON

P ERCHERON Annick, SUBILEAU Françoise, TOINET Marie-France, 1987, “Non-inscription, abstention et vote blanc et nul en France”, Espace Population Sociétés, 3, pp. 511-521. REY Henri, 1999, “Développement social et citoyenneté”, in Mozère Liane, Peraldi Michel, Rey Henri (dir.), Intelligence des banlieues, (L’aube territoires), Éditions de l’aube, pp. 47-62. S INGLY François de, THÉLOT Claude, 1989, Gens du privé, gens du public, la grande différence, Paris, Dunod. S UBILEAU Françoise, TOINET Marie-France, 1989, “L’abstentionnisme en France et aux ÉtatsUnis : méthodes et interprétations”, in Daniel Gaxie (dir.), Explication du vote : un bilan des études électorales en France, Paris, Presses de la Fondation Nationale des Sciences Politiques (2nd ed.), pp. 175-198. WEBER Max, 1964, L’éthique protestante et l’esprit du capitalisme, Paris, Plon. Y SMAL Colette, 1990, Le comportement électoral des Français, La Découverte.

Jean-Louis PAN K É S HON , Institut National de la Statistique et des Études Économiques, 18 bd Adolphe Pinard, 75675 Paris Cedex 14, e-mail: [email protected]