Documents
Resources
Learning Center
Upload
Plans & pricing Sign in
Sign Out

VOTER TURNOUT IN FRANCE

VIEWS: 5 PAGES: 32

									          FIFTH FRAMEWORK RESEARCH PROGRAMME (1998-2002)




 Democratic Participation and Political Communication in Systems of
                       Multi-level Governance




Turnout and abstention at multi-level elections in France


                              Stéphanie Abrial

                                Bruno Cautrès

                              Nadine Mandran


       Centre d’Informatisation des Données Socio-Politiques
            Centre National de la Recherche Scientifique
                                     Grenoble



                                 Work in Progress


                                    March 2003



             Draft text not to be quoted without permission of the authors.
      Introduction

As in other European countries, the political debate in France is dominated since few years by
two related questions: the rising importance and structured role of extremist parties (not only
the extreme-right voting but also now the extreme-left one) and the level of abstention. In
2002, two main national elections took place (presidential and five week later legislative
elections) : on April the 21st, the first ballot of the presidential election shown that more than
half of the French electorate either abstained, voted “blank” or voted for the extremist parties.
The last two referendum organised in France, one was a key institutional question (to reduce
or not the mandate of the French president from 7 to 5 years), also shown high level of
abstention. All of that are visible signs of an increasing gap between the French electorate and
the political supply, a gap which is also visible by the emergence of new political cleavages in
the electorate, specially those linked to the question of European integration.


On the evening of 21 April 2002, the country faced up two phenomena which still occupy the
media and political life today: the presence of the Extreme Right candidate at the second
ballot and the record abstention level for this type of election. Of France’s 39,350,086
registered voters, only 28,721,939 voted at this ballot. Abstention rose to over 27% in an
election which has traditionally been the French population’s favourite and which has
consequently shown the strongest levels of mobilisation during the Fifth Republic. If this
result was quite spectacular it was not an exception or a “storm in a blue sky” : abstention to
any level of voting is increasing in France since the mid eighties. Another sign of it can be
seen if we look to the European Parliament elections : in 1999, less than half of the registered
electorate turned out, more precisely the abstention rose to 53% !


Overall, then, a significant decrease in turnout in France has thus took place in France over
the last twenty years even if differences exist according to the type of election as can be seen
on figure 1. This figure does only indicate turnout at the main elections and wiil be completed
later by indicating the same trend at the local elections.
         Figure 1 – Turnout trends in France (1978-2002)
         P = Presidential elections
         L = Legislative elections
         E = European elections
                                                                        Turnout evolution in France
                              85
                                                 P                          P
                                                                                                              P
                              80                                    L


                              75
                                                                                                                                                P
                                                 L
                                                                                                   L                     L
                              70
                                                                            L

                              65                                                                                                                L

                                      E
                Turnout (%)




                              60                             E

                                                                                                        E
                              55
                                                                                 E
                              50                                                                                                    E


                              45
                               1978                  1982          1986              1990              1994                  1998              2002
                                          1980              1984          1988              1992                  1996                  2000          2004
                                                                                       Year




What is spectacular with figure 1 is the evidence of an underlying trend and phenomena
whatever the type and level of election : presidential election is an election with high media
coverage, strong personalisation of the debates (one national constituency); legislative
elections have gained in political importance since the three experience of “cohabitation”
(1986-1988, 1993-1995 and 1997-2002) and European elections have revealed the impact of
extreme right voting in 1984 and still again since that. Whatever the specificities and
importance of these different elections, trend in declining turnout is the same.


Turnout and abstention in France have been studied mainly through the analysis of post-
electoral surveys and ecological data. We will also rely on these two types of data and will
investigate the sociological and political explanations of turnout and abstention in the light of
a body of academic tradition. Indeed, since Alain Lancelot’s seminal research published in
1970,2 quite a body of evidence has accrued in the French political science about the
explanations of turnout. At the level of individual analysis using survey data, abstention and
non-registration have been analysed using both social integration and political integration
indices. From the sociological perspective, age, education, degree of social stability and of
social integration have been identified as major factors affecting abstention. Since abstention

2
    Lancelot, Alain, L’abstentionnisme électoral en France, Paris, Presses de la FNSP, 1968
is also linked to electoral registration, geographical and social mobility have also played their
part in explaining the propensity to abstain at the individual level. From the political
perspective, interest in and knowledge of politics, and electoral politics socialisation also have
significant effects. The short-term factors – interest in the election campaign and contextual
political effects – interact to produce a political "climate" in which electoral participation
occurs.


In this chapter, we precisely aim to analyse the phenomenon of turnout and abstention in
France, using individual and aggregate data and dimensions.
I - Participation and Abstention across level of governance




      I- 1 : Trends over time in turnout


The study of abstention in France has always fed contrasting analyses regarding the future of
national and European political life. These analyses have been at times optimistic,
emphasising political issues and contextual explanations and, at other times, they have been
pessimistic, even shedding doubt on the democratic system. If we look at data on electoral
participation in the mid-60s (Table 1), we see that, beyond the real phenomenon of a decrease
in voter turnout in the last few years - especially if one looks at the 2002 election -, this
decline may not be as critical as it may appear. Indeed, in the long term, the gap in
participation levels between the different election types (presidential, legislative and
European) remains constant.
Electoral data have been collected from several data files (origins : Home Office) deposited at
the BDSP3 which is the French link to the CESSDA. The basic unit for France is the "canton",
the smallest unit possible since the census data are not available at the communes level for
every commune. In the quasi-nested hierarchy of geographical units in France, the "canton"
follows immediatly the "communes" (more that 36 000 communes) and is still small enough
to catch, for example, the urban-rural cleavage. The canton is a very old administrative unit :
its creation, in fact linked to the creation of the "départements", is as old as 1789. At the
moment there are 96 departements on the territory of France (what is called "France
métropolitaine") and 4 more departements for the overseas French territories. As a whole
there is a little bit more that 4000 cantons. When aggregate data are not available at the canton
level, we use information at the constituency or département levels.




3
    Banque de Données Socio-Politique. http://www.cidsp.com
      Table 1 – Turnout in France (1965-2002) : descriptive statistics



                                                                 Standard
                                Minimum Median   Maximum Mean    deviation            Units (a)
Présidential   1965, 1er tour    53.29   85.28    91.14  84.95     3.63      N=474    Circonscriptions
               1969, 1er tour    58.54   78.35    96.73  78.25     3.72      N=474    Circonscriptions
               1974, 1er tour    65.25   85.15    91.46  84.84     2.81      N=474    Circonscriptions
               1981, 1er tour    43.66   82.55    96.90  81.97     4.28      N=3696   Cantons
               1988, 1er tour    54.08   83.01    98.17  82.47     3.86      N=3829   Cantons
               1995, 1er tour    52.06   80.86    90.38  80.23     4.02      N=3927   Cantons
               2002, 1er tour    56.71   73.17    81.64  72.54     3.86      N=555    Circonscriptions
Législative    1973, 1er tour    62.65   81.65    90.61  81.30     3.75      N=474    Circonscriptions
               1978, 1er tour    68.75   83.63    88.32  83.39     2.80      N=96     Département
               1981, 1er tour    62.07   71.72    78.96  71.52     3.49      N=96     Département
               1986, 1er tour    71.74   79.15    86.23  79.24     2.73      N=96     Département
               1988, 1er tour    47.19   67.85    91.65  67.76     5.70      N=3810   Cantons
               1993, 1er tour    54.25   70.52    97.98  70.34     4.70      N=3860   Cantons
               1997, 1er tour    51.63   69.96    87.46  69.71     4.87      N=3861   Cantons
               2002, 1er tour    29.52   64.97    78.02  64.29     6.10      N=577    Circonscriptions
European       1979, 1er tour    46.96   60.97    72.03  61.16     4.41      N=96     Département
               1984, 1er tour    38.06   58.99    80.59  58.79     6.12      N=3696   Cantons
               1989, 1er tour    21.79   50.75    100.00 51.02     5.91      N=3828   Cantons
               1994, 1er tour     4.84   54.60    97.87  54.74     5.89      N=3861   Cantons
               1999, 1er tour    21.43   48.99    70.44  48.95     5.99      N=3943   Cantons
(a) : “circonscriptions” means legislative electoral constituencies (N=577 in 2002); “cantons”
means the electoral constituencies on the basis of which the departemental assemblies are
elected (what is called the “conseils généraux”). Cantons are the smallest geographical units
used in this chapter.

Presidential elections have always succeeded in mobilising the higher percentage of voters.
Since 1965 (the first direct universal suffrage election), such elections have shown to be
highly legitimate and expected periods of political mobilisation : the election of the French
President represents what can be called the “key election” since the President nominates the
Prime minister and the Constitution organizes a distribution of power highly favourable to the
President. Since 1965, the Presidential election also is a highly “personalised” election : the
candidates have to dominate their camp at the first round and then to enlarge their audience at
the second round. With a two-rounds majority vote, this election structures political cleavages
around national issues4, in addition to the personality phenomenon related to the candidates.
One can be surprised by the 969 election low turnout : it can be explained by the period’s
political context and by the succession of several political events (a referendum in April on
the creation of regions, the senate’s reform and General de Gaulle’s departure). This, in
addition to the presidential election, probably gave voters the feeling that they had already
voiced their opinion during the referendum (only 19.4% turnout)5.


For the 2002 election, the explanation for the relatively low voter turnout is much more
complicated as it involves the whole electoral process. As put forward by a number of
researches6, we have to take into account the juxtaposition of two phenomena related to
political supply and demand. While voters are more demanding when it comes to the
transparency of political issues, candidates have an increasing difficulty in distinguishing the
content and the hierarchy of the political projects they offer. What is fascinating about the
2002 sequence of elections if the following. Abstention declined strongly between the first
and second ballot of the presidential election (from 27% to 20.3%) in an exceptional context :
Jean-Marie Le Pen, the leader of the Front national, was opposed to Jacques Chirac, the
incumbent President, during the second round. The presence of the extreme-right leader at this
round created such a reaction in the electorate (huge demonstrations in the main cities, call
from the left parties to their electorates to join their vote to the moderate right candidate
Jacques Chirac) than turnout jumped. But five week later; for the first round the legislative
elections, abstention went back to 35.6% for first round, and 39.7% for the second ! Which
means that there is a room for specifically highly mobilizing contexts even in a trend of rising
abstention : an interesting perspective which indicates that the sociological explanation of
abstention (more individualistic European societies, less social attachments and loyalties) is
not the whole story.


European elections since 1979, in France like in several European countries7, are showing
decreasing turnout rates. Here too, contextual factors - such as the Kosovo crisis of 1999 -
may help explain this low rate on European issues. However, it seems also necessary to
analyse - through the framework of ‘second-order’ elections8 - the impact of the political
spectrum's explosion, the primacy of national issues over European ones and the emergence


4
  Cf. PERRINEAU, Pascal, "Elections présidentielles", Dictionnaire du vote, PUF, Paris, 2001.
5
  Cf. BRECHON, Pierre, (sous dir.), Les élections présidentielles en France. Quarante ans d'histoire politique,
La Documentation Française, Paris, 2002.
6
  BRECHON, Pierre, "Comprendre les logiques de l'abstention", Revue Politique et Parlementaire, n°1020-
1021, Paris sept-oct./ nov-déc. 2002.
7
  Cf. ABRIAL, Stéphanie, PINA, Christine, "Les élections européennes de juin 1999 dans les quinze pays de
l’Union : une consultation de “second ordre” ?", Revue Française de Science Politique, Paris, volume 49, n°4,
août-octobre 1999.
8
  Cf. REIF, Karlheinz, SCHMITT, Hermann, "Nine Second Order National Elections. A Conceptual Framework
for the Analysis of European Election Results", European Journal of Political Research, 8, 1980.
of sanction voting. What is interesting to note about France is that turnout for the Maastricht
Treaty referendum in 1992 was high


I- 2 : Regional variations in turnout


Figure 1 gives only national trends in levels of turnout. But regional variations are significant
and sociologically decisive to an understanding of abstention in France. If we look at the
geography of voter turnout, we see great differences between zones of high and low electoral
participation. These differences between regions last through time: a look at the map (figure
2) shows the existence of high participation and abstentionist strongholds, beyond election
types. Participation levels represented on the map (darker shades show high participation
regions) show strong contrasts between regions. Thus, France’s southwest regions (Aquitaine,
Midi-Pyrénées), Nord-Pas-de-Calais, Picardie and other relatively isolated departments such
as Côtes d'Armores or Vendée, are high voter turnout regions. On the other hand, the eastern
part of the country (Alsace, Champagne-Ardennes), Ile-de- France, the Provence-Alpes-Côte-
d'Azur region, as well as some isolated departments such as Pyrénées Orientales are high
abstention areas.



      Figure 2 – Regional variations of turnout in France
I- 3 : The spatial cleavages basis of turnout : agreggated data analysis


To understand the patterns of regional variations and their stability through time, one need to
conduce more sophisticated statistical analysis able to identify if there are latent dimensions
underlying the spatial distribution of abstention in France. To do this, we decided to realise a
principal component analysis (so called PCA analysis) using data from the more recent census
available, the 1990 one. What we did was using 17 socio-demographic variables as “active”
one : it means that we have considered spatial variations of France as the phenomena to be
shown. We’ve then projected on this 15 electoral variables (percentages of abstention) and 8
socio-professional categories (1990 census) as supplementary variables. Basically it means to
analyze the latent socio-demographic structure of the French cantons and to observe how
abstention, on different elections, fits on these new structures. The appendix 1 gives the
complete list of variables used in this analysis.

Actually, we suppose that the 8 “socio-professional categories” express a sort of "ideal"
socio-demographic combinaison that is quite redundant with other variables like age,
education or employment situation. There is, for example, a strong statistical association
between cantons that are characterised by a great number of higher grade managers or
teachers and cantons of highly education people (R2 = .95). In the same way, cantons having
a population with a lot of retired people vary in the same direction than cantons characterized
by 65-69 years old people (R2 = .94). Thus, we decide to exclude the socio-professional
categories from the active variables list in order not to have the PCA solution dominated by
the socio-professional categories variables.


Basically, a principal component analysis is a factor analysis which indicates if there are
latent dimensions (ie factors) explaining the variation (the variance in statistical terms) of the
units (here the cantons) according to the “active” variables (here the 17 socio-demographics
census data). One can then look at the correlation between this underlying latent dimensions
and the “supplementary” variables (here the 15 rates of abstention observed at the cantons
levels for the presidential, legislative and European elections since 1988).


The first factor can be described as an opposition between “young-precarious-urban” and
“old-well-established-rural” populations in the different cantons. This factor captures 38.13%
of the variance or inertia. It looks like the “urban/rural” cleavage that we previously
discovered but it is, in reality, much more “precise” than that. This first factor presents indeed
two supplementary elements of the territorial units : their structure by age and property
status. It means than in cantons where lots of urban people can be counted, there is also a
great number of 18-24 years young people, a great number of HLM lodging renters and a
great number of non permanent employees. Conversely, in cantons where a lot of people live
in rural communes, there is a great number of self-employed and employers, a great number
of owners and a great number of 65-69 years old people. The sociological interpretation of the
first axis of our PCA is thus more than just an “urban/rural” cleavage : we propose to label
and name it a “social roots” factor. It involves that the content of this opposition goes beyond
the general cleavage between “towns and countryside” : it emphasizes that beyond the urban
or rural composition of the cantons discriminated by the first axis, there is something related
to a social integration dimension : to be, or not to be, socially integrated into a precised place,
for a long period of time (social stability through space and time).


Observing supplementary socio-professional category variables, we can confirm the
coherence of the interpretation : in “young-precarious-urban” cantons there are also lower
grades (CSP4) + employees (CSP5) + higher grade (CSP3) ; in “old-well-established-rural”
areas we find a lot of farmers (CSP1) and a lot of retired people (CSP7).



The second factor expresses the opposition between actives who have a job and unemployed
people. This factor captures 14.90% of the variance. The distinction between cantons
characterised by a great number of active people and the cantons with numerous inactive ans
unemployed individuals is straight forward. This distinction is well structured around
employment (active people, 40-44 years old, owners) and unemployement (no matter the
reasons for the lack of activity : unemployment = lack of activity ; without diploma =
difficulties being active ; retirement = end of activity). There is strong coherence in the
interpretation of that second axis with the supplementary variables : in one hand, into the
"inactive" cantons, we find retired people (CSP7) ; in the other hand there are higher grade
teachers, professors, administrative and business managers.


The third factor is a little bit more difficult to understand but is worth keeping in the analysis:
it structures a cleavage cleavage between variables related to “intellectual professions-non-
owners-females” and “males-HLM tenants” people. This factor captures 11.87% of the
variance, not so far from the proportion of inertia explained by the second factor. It is
sociologically very interesting as it is characterised by the opposition between rather urban
cantons that include rather well educated individuals who do not own but rent without being
renters of HLM ; and rather urban cantons characterised by males having to live in social
housing. Its interpretation is thus like a second version of the “urban/rural” cleavage that was
underlying the first factorial dimension of our analysis. We propose to label this third
dimension as an opposition between cantons under more or less “social constraint”. In “less
social constraint” cantons, we find high level managers, retired people as well as artisans and
small business owners. In the “more social constraint" cantons, we find inactive workers and
students-others.

We can resume our principal component analysis by the way of table 2.
Table 2 : Principal component analysis of the spatial socio-demographic cleavages of the
French cantons (1990 census data) : main findings.
                         Actives variables       Supplementary variables :
                                                 socio-professional category
1st axis   +   •18-24 years old                •primary schools teachers,
                                               technicians and business lower
               •urban                          level (CSP4)
               •renter of HLM lodging          •Employees (CSP5)
               •precarious                     •higher grade teachers,
                                               professors, administrative and
                                               business managers (CSP3)
           -   •rural                          •farmers (CSP1)
               •self-employers/employers       •retired (CSP7)
               •owners
               •65-69 years old
 nd
2 axis     +   •unemployed                     •retired (CSP7)
               •no diploma
           -   •active                         •primary schools teachers,
                                               technicians and business lower
               •40-44 years old                level (CSP4)
               •owners                         •higher grade teachers,
                                               professors, administrative and
                                               business managers (CSP3)
3rd axis   +   •renter of non-HLM lodging      •retired (CSP7)
               •end of secondary education +   •artisans and small business
               university graduation           owners (CSP2)
               •65-69 years old                •higher grade teachers,
                                               professors, administrative and
               •women                          business managers (CSP3)
           -   •men                            •workers (CSP6)
               •renter of HLM lodging          •students and other non-active
                                               (CSP8)




I – 4 : Matching the spatial sociodemographic cleavages and turnout


Abstention rates as supplementary variables


We now would like to use our turnout electoral data as supplementary variables in order to
analyze the relation between the spatial socio-demographic cleavages revealed by the PCA
and the territorial logics of turnout in France.
1st factor : social roots / social instability


Generally, no matter the type of election, we find the highest rate of abstention in the cantons
characterised by a large number of young precarious urbanites. This means the abstention rate
of a canton varies positively with the rate of “social rooting” of individuals living in those
cantons. From the variables used, the lower the rate of “social roots” in a canton, the higher
the abstention rate is.
The elections that best fit with this schema are the presidential (2ndr), the legislative (2ndr) and
the 1998 regional elections. On the first dimension of the PCA, abstention rate is thus not
specifically related to a type of election. This is not surprising as the territories defined by this
first axis are characterised by a strong left-right partisan opposition, which are at the heart of
national and local political issues. On the other hand, we see the absence of European
elections among those that mobilise the least: as these elections stigmatise less the left-right
cleavage, they do not give rise to an ideological type of opposition in those cantons
characterised by the urban-rural opposition.


2nd factor : active / non-active


We can observe that the abstention rate is highest in the cantons characterised by the most
inactive population. Indeed, abstention and inactivity are positively correlated: the higher the
inactivity rate of a canton, the higher its abstention rate as well. The elections that best
illustrate this are the presidentials.


3rd axis: urban opposition between cantons exposed to more or less “social constraint”


The abstention rate is highest in the cantons where individuals seem most preoccupied by
their material situation. Opposingly, the abstention rate is lowest in cantons characterised by a
“less social constraint" population (intellectuals, high level of education and tenants). We
observe that the abstention rate is there especially high in European elections.


It is on this third axis that we see emerging the distinction between abstention in national
elections, from the abstention in European elections. The territories of national election
abstentions are more closely related to the rural-urban cleavage which itself is underlined by a
well known political effect arising from the left-right dichotomy. As for the phenomenon of
high abstention in European elections, we do not find the same territorial dimension. Indeed,
it is in those areas where social and material preoccupations are strongest that European
elections attract least voters, especially compared with national elections. Thus, we can
formulate the following hypothesis: the areas with a high abstention rate in European
elections, more than in the case of national elections, are territories where social exclusion is
higher.


Multiple linear regression analysis


We now are using the coordinates of the cantons on the three dimensions of the PCA as
explanatory factors in regression analysis models. The aim is to understand the effect of our
three variables, i.e. weak social rooting, inactivity and urban social constraint, on abstention
in cantons. The dependent variables are the abstention rates at the different level of
governance elections from 1988 to 1999.
Table 3 : Regression analysis : abstention explained by socio-demographic cleavages
(aggregated data analysis, 1988-1999).

1988 General         1st Ballot                                   2nd Ballot
elections
                     Coeff.         Standard deviation            Coeff.               Standard deviation
Intercept            32.238         0.07**                        27.646               0.08**
Factor1              1.276          0.03**                        1.271                0.03**
Factor 2             0.540          0.04**                        0.756                0.05**
Factor 3             -0.389         0.05**                        -0.571               0.05**
                     R2=.37, R=.61, Model significant 0.0001      R2=.346 R=.60, Model significant 0.0001


1993 General elections 1st ballot                                       2nd ballot
                       Coeff.                Standard deviation         Coeff.            Standard deviation
Intercept              29.661                0.06**                     30.865            0.10**
Factor1                0.906                 0.02**                     0.979             0.04**
Factor 2               0.767                 0.04**                     0.485             0.06**
Factor 3               -0.01                 0.04 (ns)                  0.001             0.07 (ns)
                       R2=.31, R=.56, Model significant 0.0001          R2=.14, R=.38, Model significant
                                                                        0.0001
1997 General elections 1st ballot                                   2nd ballot
                       Coeff.                Standard deviation     Coeff.            Standard deviation
Intercept              30.288                0.06**                 26.955            0.07**
Factor1                1.141                 0.02**                 1.206             0.03**
Factor 2               0.507                 0.04                   0.652             0.04**
Factor 3               0.274                 0.04**                 -0.039            0.05**
                       R2=.40, R=.63, Model significant 0.0001      R2=.35, R=.59, Model significant
                                                                    0.0001

1988 Presidential        1st ballot                                 2nd ballot
election
                         Coeff.                Standard deviation   Coeff.            Standard
                                                                                      deviation
Intercept                17.526               0.05**                14.475            0.04**
Factor1                  0.528                 0.02**               0.712             0.02**
Factor 2                 0.783                 0.03**               0.793             0.03**
Factor 3                 0.506                 0.04**               0.159             0.03**
                         R2=.27, R=.52, Model significant 0.0001    R2=.41, R=.64, Model significant
                                                                    0.0001


1995 Presidential        1st ballot                                 2nd ballot
election
                         Coeff.                Standard deviation   Coeff.           Standard
                                                                                     deviation
Intercept                19.773               0.05**                18.462           0.05**
Factor 1                 0.714                 0.02**               0.898            0.02**
Factor 2                 0.745                 0.03**               0.851            0.03**
Factor 3                 0.51                 0.04**                -0.056           0.03 (ns)
                         R2=.32, R=.56, Model significant 0.0001    R2=.43 R=.66, Model significant
                                                                    0.0001

1989 European election
                                     Coeff.                          Standard deviation
Intercept                            48.978                          0.08**
Factor 1                             0.937                           0.03**
Factor 2                             0.807                           0.05**
Factor 3                             -1.068                          0.06**
 R2=.29, R=.53, Model significant 0.0001

1994 European election
                                     Coeff.                          Standard deviation
Intercept                            45.257                          0.08**
Factor 1                             0.864                           0.03**
Factor 2                             0.692                           0.05**
Factor 3                             -0.541                          0.06**
 R2=.20, R=.44, Model significant 0.0001

1999 European election
                                     Coeff.                          Standard deviation
Intercept                            51.048                          0.08**
Factor1                              0.980                           0.03**
Factor 2                             0.823                           0.05**
Factor 3                             -0.978                          0.06**
 R2=.28, R=.53, Model significant 0.0001
The first element we will deal with here is the model’s significance level: no matter the type
of election, the regression model fitted to the data is significant. This means that there is a
relationship between the socio-demographic variables chosen and the abstention rate in
cantons : three axis have a significant influence on the abstention rate in all of the elections
surveyed. For any level of governance, getting a low score on the first factor (which means
being characterized as a canton as low “social root”) increase the level of abstention in a
significant way; getting a high score on inactivity factor (our second dimension of PCA) as
also a positive effect on the increase of abstention. A remarkable result of the regression
analysis is that scores of cantons on the third dimension have a different effect on national and
European elections levels of abstention: for national elections (either legislative or
presidential) the effect is positive in about every case; for European election, the effect is a
negative one. It will be necessary in our final version to go into more detailed explanation.



II/ Institutional facilitation and mobilisation


II – 1 : Institutional facilitation


In the French electoral system, vote is not an obligation. The best source to observe the
impact of electoral institutions on voting behavior is the electoral code that currently governs
elections. For some years, it can be noticed a sort of political will to further ease the electoral
process, at the same time maintaining the tradition and solemnity of voting.


French citizens, older than 18 years old and not deprived of their civil and political rights can
vote in general elections. In May 1997, there were 39 361 138 registered electors. Concerning
European and local elections, the citizens of the European Union residing durably in France
profit from the voting rights. For the 1999 European election, they were about 55 000
« European foreign citizens » registered. Because of the non-registration of certain voters, the
electorate has been cut down on average, for 15 years, about 9% of the citizens who could
claim to form part of it. According to official figures (INSEE surveys), the French electorate
would also be on the way to be dominated by rather old citizens. We know, for example, that
the average age of the registered voters passed from 46,2 to 47,4 years between 1988 and
1995.
In order to be registered on the electoral list, people have to sign up at their local town hall.
But since 1997, this measure has become easier with people reaching the age of 18 being
automatically registered. This first political step targeted at young voters has thus been
considerably simplified. This is interesting as it shows how political authorities are attempting
to face the problem of young voters’ registration - and turnout - by integrating them directly
into their electoral list. However, this could have an unexpected and negative effect: while the
number of unregistered voters is mechanically decreased, it necessarily implies a higher
abstention level9.


Voting by post does not exist in France, but by proxy does. Until 1975, it was possible to vote
by post but the number of frauds recorded involve a cancellation of this procedure. Today, the
postal way has to be used only for the professional and the university elections. The vote by
proxy thus became the only possible process for those who cannot vote personally. The right
to vote by proxy only applies to those who can produce evidence of their unavailability
(professional obligations, physical disability…). But, we can notice that for the last 2002
general elections these conditions were relaxed : people who could prove they were away due
to the regional variation of the school holidays had been able to apply for proxy rights. Once
again, we can observe that this "punctual" enlargement of proxy conditions emphasizes the
political determination to support the voter turnout. Without modifying the text the procedure
was actually facilitated. Depending on the choice of the mandant, the proxy can be given for
one ballot or for one year. The proxy representative must be registered in the same commune
as the mandant. The step does not comprise real difficulty. Even if it requires to move
preliminary in a place of public authority (like the police station), the proxy constitutes a
practice well used.


The elections take part in every district, where several polling stations can be provided. Each
polling station refers to a specific area and number of voters. The geographical and "material"
organization of the vote (place of the polling station, place of polling boothes, the panelling
round the room…) is always the same one. Whatever the type of elections, voters always find
the same spatial diagram of voting.



9
 In his 2002 election analysis, Pierre Bréchon estimates the increase in abstention related to this new measure at
1%. Cf. BRECHON, Pierre, "Comprendre les logiques de l'abstention", Revue Politique et Parlementaire,
n°1020-1021, Paris sept-oct./ nov-déc. 2002.
Ballot takes place during one day, always Sunday. One could disgree with the fact that
everybody must vote on the same day. But at the same time, we should bear in mind that this
electoral method belongs in French history and is deeply rooted in the political habits of the
population.


II-2 : Institutional mobilisation


If we now look at the socio-political characteristics present in the electoral competition,
several aspects of the French political system could also be integrated in an analysis of
electoral mobilisation.


Described up until the 1980s as a “bipolar four-part” system (i.e., two poles and four parties
with the same political weight - UDF/RPR versus PS/PC), the French party system has
considerably changed. 1981 was indeed a fundamental rupture point in the balance created by
the hegemony of the RPR and the Socialist party10. The appearance of the far right and the
gradual disappearance of the communist party (only 3.4% in the 2002 presidential elections)
have also modified the make-up of an increasingly diverse party system. In the long run,
presidential French elections are characterised by an increasing number of candidates: six in
1965, seven in 1969, 12 in 1974. Then, in 1976, the minimum number of signatures necessary
for running changed, increasing from 100 to 500, however, this seems to have only a limited
impact: in 1981, there were10 candidates in 1998; nine in 1995 and; 16 in 2002.
The 2002 legislative elections also beat all records in the number of candidates, with 8446, or
approximately 15 for each electoral county. This multiplication can be seen both between and
within partisan blocs11. However, this dynamic political life is probably accompanied by a
greater difficulty in clearly seeing issues and traditional political partisan differences. This
probably contributes to the electoral volatility and abstention.
Another change in the French political system is related to party campaign and funding
procedures. Due to several corruption and dirty financing cases, several laws that did not exist
prior to 1988 were implemented, aiming at regulating campaign funding. In 1990, a national
commission on political and campaign financing was created. Three years later, funding from
private firms and more generally ‘moral’ persons was forbidden. These measures aimed at

10
   Cf. MARTIN, Pierre, Comprendre les évolutions électorales. La théorie des réalignements revisitée, Presses
de sciences Po, Paris, 2000.
11
   Cf. ABRIAL, Stéphanie, CAUTRES, Bruno, EVANS, Jocelyn, "Stabilité et recomposition du système de
partis français", Revue politique et parlementaire, Paris, n°1020-1021, sept.-oct./nov.-déc. 2002.
improving not only political transparency but also politicians’ image. But, with the continued
decline in electoral participation, one is bound to wonder about the impact of these measures
on voter turnout.
The change that took place within the 2002 electoral calendar - the legislative and presidential
elections were reversed - also contributed to further complicate our attempts at explaining the
decrease in voter turnout. Indeed, one can raise the question of the goal behind this change:
what other reason can there be for the change in the electoral calendar than the elite’s -on each
side of he left-right spectrum - convenience?
In a context in which electoral campaigns do not allow debates on programmes nor on issues
(demonstrated by the universal conviction that Lionel Jospin and Jacques Chirac would be the
final two candidates in 2002), it would thus appear that the very nature of competition has
been modified.


III - Individual facilitation and mobilisation


In now look at the individual level of analysis using mainly Eurobarometers surveys (the post-
European election studies conduced in 1994 and 1999), but also the ASES study and the
national elections studies.


III-1 : Facilitation


By personal facilitation of turnout we mean the main characteristics of individuals that make
voting more easy, or let’s say less “costly”. Voting requires electoral registration, at least
some interest into the elections and taking time going to the polling station. One of the main
indicator of the facilitation characteristics of individuals is education. Does education matters
on voting per se? Or do we have interactive effects of education and age, two individual
characteristics that are closely linked? Does other socio-economic resources of individuals
(such as income, social status and social class, religious integration) facilitates or not turnout
[this section has to be completed]?


Age and education
The observation that age and turnout are correlated in true in France as in other European and
Western countries. In France, the 18-24 years group knows an abstention rate which is much
higher than the other age groups. This difference corresponds to what Anne Muxel has called
a “political moratoire”, a period of entrance into the civic life, still marked by the wish of
experiencing and keeping some distances to the “adult” world of politics. Our data also
suggests that there is in France, as in other Western countries, an increase of that abstention
rate among young French citizens in more recent cohorts. Age per se is not a “pure” age effect
: age is in fact an aggregated indicator, an index of both social and political integration.
Young people are less politically and socially integrated in that they still experience
geographical, professional and familial mobility and instability. Their lack of political
experience and political knowledge could explain their lack of participation to elections, but
their recent or still not started entrance on the job market also. Values associated to politics
could (lack of trust and confidence into the efficacy of politics and elections) also be part of
the story. An interesting result coming from table 4 is that age differences are stronger for EP
elections than for national elections.


Table 4 : Voter turnout and reasons for abstention by age in the 1994 and 1999 European elections in
           France

EP94 (EB41.1)

                                        18-24 yrs 25-34 yrs 35-54 yrs      55+ yrs    Total     N
  Voted                                     39.8      51.0      71.2          83.9     67.5 590
  Voluntary abstainer                       29.5      23.7      18.1           9.7     17.8 156
  Circumstantial                            13.6      15.2        7.8          3.9      8.8 77
  Other reasons                             17.0      10.1        2.9          2.5      5.8 51
  Total                                      100       100       100          100      100 874
  N                                           88       198       309          279      874
EP99 (EB52.0)

                                      18-24 yrs 25-34 yrs 35-54 yrs 55+ yrs      Total      N
   Voted                                     38.2      49.8      61.5      68.3      56.9       551
   Voluntary abstainer                       26.5      28.6      24.3      14.9      23.5       228
   Circumstantial                            21.3      14.5      10.8      13.1      13.7       133
   Other reasons                             14.0        7.1      3.5        3.6        5.9      57
   Total                                      100       100      100        100        100      969
   N                                          136       241      371        221        969
1995 Presidential election (CEVIPOF
1995 election study)

                                      18-24 yrs 25-34 yrs 35-54 yrs 55+ yrs     Total    N
   Did not vote                              22.1      22.8      16.5      13.1     17.4 3364
   Voted                                     77.9      77.2      83.5      86.9     82.6 709
   Total                                      100       100       100       100       100 4073
   N                                          561      833      1323      1356      4073
Education is another important element that may facilitates turnout at the individual level. It is
also an element of social status which is related to age and we will have to control age effect
for education levels differences among age groups.

As can be seen in table 5, the turnout rates differences among levels of education are not so
big for national elections than for EP elections, a result which recalls the one obtained on age
effects. It looks like if participating to EP elections was a more “difficult” task, or one could
say that EP elections participation revealed stronger social cleavages than participating to
national elections, a result which will have to be confirmed by using more data on national
elections.

Table 5, Voter turnout and abstention by level of education in the 1994 and 1999
European elections and 1995 presidential election


EP94 (EB41.1)
                                  Lower       Higher
                 Primary      secondary    secondary   Third level   Total     N
Did not vote         25.7          27.9         39.4         32.4      34
Voted                74.3          72.1         60.6         67.6      66
Total                 100           100          100          100     100
N                      35           204          360          293     892
ASES (1995 presidential
election)
                                  Lower       Higher
                   Primary    secondary    secondary   Third level   Total     N
Did not vote           3.3         13.4         14.0         12.0     11.5   113
Voted                 96.7         86.6         86.0         88.0     88.5   870
Total                  100          100          100          100     100    100
N                      153          336          178          316     983
CEVIPOF 1995
presidential election study
                                  Lower       Higher
                   Primary    secondary    secondary   Third level   Total      N
Did not vote          17.5          17.1        15.6         19.1     17.5    714
Voted                 82.5          82.9        84.4         80.9     82.5   3364
Total                100.0         100.0       100.0        100.0    100.0   4078
N                     1239         1438          486          915    4078
EP99 (EB52.0)
                                  Lower       Higher
                   Primary    secondary    secondary   Third level   Total     N
Did not vote          63.9          36.9        46.3         42.7     63.9   440
Voted                 36.1          63.1        53.8         57.3     36.1   528
Total                100.0         100.0       100.0        100.0    100.0   968
N                      108           179         400          281     968
Socio-economic status : social class and income.

Income differences also matters (table 6) : turnout is not only associated with age and
educational differences but also with socio-economics characteristics. Turnout is also
associated with social class differences : working class or employees occupation people are
les likely to vote to either national or European elections. Unemployed people also.



Table 6 :Voter turnout and abstention by level of income in the 1994 and 1999 European
elections and the 1973 and 2002 general elections (per cent) in France

EP94 (EB41.1)
                         Lowest     Low              High    Highest   Total     N
Did not vote               37.4     35.0             34.2       25.9    33.6   268
Voted                      62.6     65.0             65.8       74.1    66.4   530
Total                       100      100              100        100     100   798
N                           195      214              231        158     798

EP99 (EB52.0)
                         Lowest     Low              High    Highest   Total     N
Did not vote               43.2     26.5             25.7       38.3    32.5   114
Voted                      56.8     73.5             74.3       61.7    67.5   237
Total                       100      100              100        100     100   351
N                            74       83              113         81    351

ASES (1995 presidential election)
                        Lowest      Low    Middle    High    Highest   Total     N
Did not vote               47.6     16.3       9.6     8.4       7.9    11.1   101
Voted                      82.4     83.7     90.4    91.6       92.4    88.9   810
Total                       100      100      100     100       100     100    911
N                           102      166      230     274       139     911
CEVIPOF 1995 presidential
election study
                        Lowest      Low    Middle    High    Highest   Total      N
Did not vote               25.2     21.7     15.7    14.5       15.1    17.6    667
Voted                      74.8     78.3     84.3    85.5       84.9    82.4   3122
Total                       100      100      100     100       100     100    3789
N                           540      627      720     958       944    3789




III – 2 : Mobilisation


We now would like to turn our attention to factors that may concern the mobilization process
that conduce to turnout.


Media exposure and political knowledge
Media use is among the mobilisation factor a good indicator of interest into political matters
and exposure to information. As table 7 shows, exposure to the media has a significant effect
on turnout either for national or European elections. Regular readers of newspapers specially
are more likely to participate in European election and the readership of newspaper is, not
surprisingly, more significant than exposure to TV channels. Again we can see from the
ASES study that exposure to the media is more significantly related to turnout in EP elections
than in national elections.


Table 7 : Voter turnout and abstention by media use in the 1999 European elections in
France

Newspaper use:

EP99 (EB52.0)
                                       Once or twice a Several times
                 Never   Less often               week       a week Everyday      Total     N
Did not vote      56.2        58.2                49.4          49.4    30.4       45.6   441
Voted             43.8        41.8                50.6          50.6    69.6       54.4   527
Total            100.0       100.0               100.0        100.0    100.0      100.0   968
N                  153         165                 164           160     326       968


Television use:

EP99 (EB52.0)
                                       Once or twice a Several times a
                 Never    Less often             week             week Everyday   Total     N
Did not vote      57.1         56.1               64.1            45.9     40.3    45.6   442
Voted             42.9         43.9               35.9            54.1     59.7    54.4   527
Total            100.0        100.0             100.0            100.0    100.0   100.0   969
N                   35            41               117             196      580    969
Table 7 (continued) : Voter turnout and abstention by media use in the national election
in France

Level of media usage:

ASES : 1995 presidential election - Local media use
                               Never From time to time     Regularly   Total     N
Did not vote                    15.4                  15         7.8    11.5   113
Voted                           84.6                  85       92.2     88.5   870
Total                            100                 100        100      100   983
N                                143                 353        487     983
ASES : 1995 presidential election - National media use
                               Never From time to time     Regularly   Total     N
Did not vote                    19.8                14.4         8.9    11.5   111
Voted                           80.2                85.6       91.1     88.5   870
Total                            100                 100        100      100   983
N                                 91                 284        608     983
ASES : 1995 presidential election - Foreign media use
                               Never From time to time     Regularly   Total     N
Did not vote                    11.2                13.4         8.5    11.5   113
Voted                           88.2                86.6       91.5     88.5   870
Total                            100                 100        100      100   983
N                                615                 262        106     983

ASES : 1999 EP election - Local media use
                              Never From time to time      Regularly   Total     N
Did not vote                    39.3             29.0          17.8     25.0   247
Voted                           60.7             71.0          82.2     75.0   741
Total                            100              100           100     100    988
N                                145              355           488     988
ASES : 1999 EP election - National media use
                              Never From time to time      Regularly   Total     N
Did not vote                    44.6             31.5           19.0    25.0   247
Voted                           55.4             68.5           81.0    75.0   741
Total                            100              100           100     100    988
N                                 92              286           610     988
ASES : 1999 EP election - Foreign media use
                              Never From time to time      Regularly   Total     N
Did not vote                    24.4             29.8           16.7    25.0   247
Voted                           75.6             70.2           83.3    75.0   741
Total                            100              100           100     100    988
N                                618              262           108     988


Exposure to the media has obviously to see with the capacity of voters to get into an interest
to politics process (even if it is a necessary but not sufficient condition of interest into
politics) and has to see with political knowledge. Competence and political knowledge is a
“double face” indicator, both a facilitating and a mobilisating one. Table 8 shows that
political knowledge as it is measured by Eurobarometers as very a strong impact on turnout,
perhaps one of the most significant impact seen in our analysis. We have to question this
result that may have some “tautology” component. It is worth interesting to look at the
differences between Eurobarometers and ASES indicator of political knowledge, the former
being much more related to turnout than the latter.


Table 8 : Voter turnout and reasons for abstention by level of political knowledge for the
1994 and 1999 European elections and the 1995 presidential election.

EP94 (EB41.1)
                              0                 1    2            3         4       Total            N
Voted                      49.1             59.3 74.6         80.3       89.0        67.5          591
Voluntary abstention       29.5             24.1 12.4           9.0       4.9        17.7          155
Circumstantial abstention 13.2                9.3  8.6          5.3       4.9          8.9          78
Other reasons               8.1              7.4   4.3         5.3        1.2         5.8           51
Total                     100.0            100.0 100.0       100.0      100.0       100.0          875
N                           234              162 209           188        82          875

EP99 (EB52.0)
                                      0         1       2          3         4       Total           N
Voted                             45.1      55.8    62.1       69.4      80.0         58.4         566
Voluntary abstention              33.8      24.6    15.0       13.2      12.7         22.6         219
Circumstantial abstention         14.0      14.6    15.0       13.2        5.5        13.3         129
Other reasons                       7.2       5.0     7.8        4.1       1.8          5.8         56
Total                              100       100     100        100       100          100         970
N                                  293       240     206        121       110          970

ASES : 1995 presidential
election
                                     0         1       2          3         4        Total           N
Did not vote                      11.9       5.3    13.1       13.9      10.2         11.5         113
Voted                             88.1      94.7    86.9       86.1      89.8         88.5         870
Total                              100       100     100        100       100          100         983
N                                  285        38      99        187       374          983

ASES : 1999 EP election
                                     0         1       2          3         4        Total           N
Did not vote                      27.6      13.2    30.0       29.1      20.8         25.0         247
Voted                             72.4      86.8    70.0       70.9      79.2         75.0         741
Total                              100       100     100        100       100         100          988
N                                  286        38     100        189       375         988


Note: the numbers 0-5 refer to the number of correct answers to a series of political knowledge questions. These questions
generally refer to the identification of the holders of political office or leadership positions within parties.
Opinions and attitudes towards EU and turnout


Attitudes towards EU integration is a factor that may increase or not the political mobilization
process when it comes to the EP elections. Table 9 compares the attitudes toward the power of
European and national parliaments : in both cases, a positive perception of the role of EP or
national parliament increase turnout at EP elections, showing that turnout is linked with a
kind of ‘civic’ political culture. It is worth interesting to note that there anyway still high
abstention rates among people scoring high the role of EP. One has to be careful in that
interpretation since it is based on small n. Table 10 presents the effects on turnout of the
difference between the actual perceived speed of EU integration (on a 7 point scale) and the
desired speed of it (on the same scale): people getting a negative difference would like to see
EU integration going slower than it is, people having a 0 score are jus satisfied with the actual
speed of EU integration and people having a positive score would like to EU making faster
progress. The scores obtained are good indicators of a general attitude towards EU and are
associated with other indicator of attitudes towards it. The gap between turnout rates of the
different populations discriminated by this indicator have changed from 1994 to 1999 : EU
integration attitudes seems to be more associated with turnout in 1999, a result that would
have to be checked by further analysis.

Table 9 : Voter turnout and perceptions of the power of the European Parliament in
1994 and 1999 in France

Perceived power of the European Parliament
                    No power at all                                                                 A great deal of
EB41.1 (1994)           1           2         3        4         5          6         7         8           9
Did not vote         44.4       33.3       30.1     27.4      37.3       30.3      36.3      30.9         9.5
Voted                55.6       66.7       69.9     72.6      62.7       69.7      63.7      69.1        90.5
Total               100.0      100.0      100.0    100.0     100.0      100.0     100.0     100.0       100.0
N                      27          42       103      117      228        119        91        55          21
                    No power at all                                                                 A great deal of
EB52 (1999)             1           2         3        4         5          6         7         8           9
Did not vote         63.6       58.1       59.0     48.6      43.2       34.8      41.7      41.8        42.1
Voted                36.4       41.9       41.0     51.4      56.8       65.2      58.3      58.2        57.9
Total               100.0      100.0      100.0    100.0     100.0      100.0     100.0     100.0       100.0
N                      11          31       78       138      236        158       127        79          19
Perceived power of the French Parliament (Assemblée nationale)
                    No power at all                                                                       A great deal of
EB41.1 (1994)           1           2           3               4             5       6       7       8           9
Did not vote         43.5       29.6         35.1            33.9          36.2    31.0    32.4    28.2        26.2
Voted                56.5       70.4         64.9            66.1          63.8    69.0    67.6    71.8        73.8
Total               100.0      100.0        100.0           100.0         100.0   100.0   100.0   100.0       100.0
892N                   23          27         37              59           163     126     136     149          65
                    No power at all                                                                       A great deal of


EB52 (1999)             1            2          3               4             5       6       7       8          9
Did not vote         71.4         55.2       49.2            42.7          48.0    39.4    48.6    37.5       38.1
Voted                28.6         44.8       50.8            57.3          52.0    60.6    51.4    62.5       61.9
Total               100.0        100.0      100.0           100.0         100.0   100.0   100.0   100.0      100.0
N                       7          29         61              96            225     188     140     112        42



Table 10 : Voter turnout and perceptions of the current and desired speed of the
European integration (difference between desired and perceived speed of integration)

Perceived speed of European integration
                            -     =       +         Total            N
EB41.1 (1994)
Did not vote              36.6     40.0    32.2      33.9           303
Voted                     63.4     60.0    67.8      66.1           590
Total                    100.0    100.0   100.0     100.0           893
N                          82       150     661       893

                            -     =       +         Total            N
EB52 (1999)
Did not vote              50.8     50.6    41.1      45.5           441
Voted                     49.2     49.4    58.9      54.5           529
Total                    100.0    100.0   100.0     100.0           970
N                          199      245     526       970




A multivariate analysis could obviously be done to resume all the conclusions about
mobilisation and facilitation individuals factors on electoral turnout in EP elections in France.
This will be done in full details in the coming book.


To resume it anyway, one can say that abstention in France is highly determined by
sociological and political individual factors which affect the mobilisation of French voters.
We also know from our ecological data analysis that individual voters are not actors without
contexts : the geographical context of voters do affect their propensities to vote or abstain :
the urban/rural cleavage, the active/non-active one also strongly impact on abstention. The
understanding of turnout in France is a mix between this two main explanations : individual
factors of mobilisation and geographical contexts. All theses factors apply to the different
type of elections, EP elections being specially affected by the level of cognitive mobilisation
of voters and their interest in European affairs.
                                                       Appendix 1

     A- The list of variables used for the Principal component analysis

       Active socio-demographics : : sh90 : men counts ; ag90.1 : 18-24 years old people ; ag90.5 : 40-44
years old people ; ag90.7 : 50-54 years old people ; ag90.10 : 65-69 years old people ; dip90.12 : education, no
diploma ; dip90.17 : education, baccalauréat (end of secondary education) or professional, technical equivalent ; dip90.19 :
education, university graduation (baccalaréat + 3 or 4 years), postgraduation (baccalauréat + 5 years and over), engineer high
schools diploma ; emp901 : employment status, non permanent employment contract (sandwich course between job and
vocational training + contract for a limited period of time + contract under a temping agency + firm trainee ; emp90.21,
employment status, self-employed, employers ; tac90.11 : employement status, active having a job ; tac90.12 : employement
status, unemployed ; stoc90.1 : property status, owner, householder ; stoc90.2 : property status, tenant, renting of a non
furnished and non HLM lodging ; stoc90.3 : property status, tenant, renting of a non furnished and HLM lodging ; tu_rur,
rural communes, -2500 inhab. ; tu_urb : urban units, + 200 000 inhab.

        Supplementary electoral variables1 (rates of abstention) : L881abp : legislative elections, 1988, 1st round,
abstention ; L882abp: legislative elections, 1988, 2nd round, abstention ; L931abp: legislative elections, 1993, 1st round,
abstention ; L932abp: legislative elections, 1993, 2nd round, abstention ; L971abp: legislative elections, 1997, 1st round,
abstention ; L972abp: legislative elections, 2nd round, abstention ; P881abp : presidential elections, 1988, 1st round,
abstention ; P882abp: presidential elections, 1988, 2nd round, abstention ; P951abp: presidential elections, 1995, 1st round,
abstention ; P952abp: presidential elections, 1995, 2nd round, abstention ; E89abp : european elections, 1989, abstention ;
E94abp: european elections, 1994, abstention ; E99abp: european elections, 1999, abstention ; R92abp : regional elections,
1992, abstention ; R98abp : regional elections, 1998, abstention

       Supplementary socio-professional category : cs901: farmers ; cs902 : artisans and business owners ;
cs903 : higher grade teachers, professors, administrative and business managers ; cs904 : primary schools teachers,
technicians and business lower level ; cs905 : employees ; cs906 : skilled and unskilled workers ; cs907 : retired ; cs908 :
unemployed never worked, students and other non active



B – Statistical results of the Principal component analysis

Table1 : Eigen value and associated variance
                         Eigen value               % variance               Sum of eigen value         Sum % variance
1                        6.481785                  38.12815                 6.48179                    38.1281
2                        2.532348                  14.89616                 9.01413                    53.0243
3                        2.017670                  11.86865                 11.03180                   64.8930
4                        1.059270                  6.23100                  12.09107                   71.1240
5                        0.945401                  5.56118                  13.03647                   76.6851
6                        0.658799                  3.87529                  13.69527                   80.5604
7                        0.601510                  3.53829                  14.29678                   84.0987
8                        0.547905                  3.22297                  14.84469                   87.3217
9                        0.491172                  2.88924                  15.33586                   90.2109
10                       0.370552                  2.17972                  15.70641                   92.3907
11                       0.298137                  1.75375                  16.00455                   94.1444
12                       0.247969                  1.45864                  16.25252                   95.6030
13                       0.225333                  1.32549                  16.47785                   96.9285
14                       0.186591                  1.09759                  16.66444                   98.0261
15                       0.178014                  1.04714                  16.84246                   99.0733
16                       0.141968                  0.83511                  16.98442                   99.9084
17                       0.015576                  0.09162                  17.00000                   100.0000


Table 2 : contribution actives variables
                               Fact. 1                          Fact. 2                          Fact. 3

SH90_P                         0.057508                         0.035686                         0.088979
A901P                          0.074051                         0.004792                         0.017134
A905P                          0.023964                         0.139191                         0.058433
A907P                          0.000021                         0.019802                         0.010417
A910P                          0.072038                         0.026814                         0.139999
DIP9012P             0.063065                0.121364               0.000236
DIP9017P             0.047043                0.069029               0.127285
DIP9019P             0.066244                0.027976               0.148774
TAC9011P             0.035106                0.174972               0.005681
TAC9012P             0.018276                0.201671               0.006260
STOC901P             0.094066                0.084009               0.006081
STOC902P             0.039920                0.022490               0.195508
STOC903P             0.065481                0.053814               0.090397
EMP9012P             0.072121                0.002805               0.017760
EMP9021P             0.083155                0.008713               0.072774
tu_rurp              0.115818                0.006553               0.000388
tu_urbp              0.072125                0.000319               0.013893


Table 3 : coordinates active and supplementary variables
                                 Fact. 1                Fact. 2          Fact. 3
                SH90_P           -0.610537              -0.300617        -0.423709
Active
                A901P            0.692809               0.110155         -0.185934
                A905P            0.394117               -0.593701        -0.343364
                A907P            -0.011573              -0.223934        0.144977
                A910P            -0.683328              0.260582         0.531481
                DIP9012P         -0.639356              0.554379         -0.021836
                DIP9017P         0.552197               -0.418098        0.506774
                DIP9019P         0.655271               -0.266168        0.547884
                TAC9011P         0.477019               -0.665650        -0.107065
                TAC9012P         0.344178               0.714634         -0.112387
                STOC901P         -0.780841              -0.461237        -0.110767
                STOC902P         0.508677               0.238646         0.628069
                STOC903P         0.651485               0.369154         -0.427072
                EMP9012P         0.683717               0.084274         -0.189296
                EMP9021P         -0.734162              -0.148545        0.383189
                tu_rurp          -0.866433              -0.128817        0.027964
                tu_urbp          0.683737               -0.028419        0.167429
                *L881ABP         0.580149               0.153475         -0.098562
Supplementary
                *L882ABP         0.545951               0.202738         -0.136813
                *L931ABP         0.496415               0.262723         -0.005040
                *L932ABP         0.362711               0.112457         0.000206
                *L971ABP         0.603287               0.167518         0.080937
                *L972ABP         0.557556               0.188424         -0.010079
                *P881ABP         0.353615               0.328031         0.189022
                *P882ABP         0.524157               0.365257         0.065369
                *P951ABP         0.453728               0.295794         0.159788
                *P952ABP         0.563859               0.333813         -0.019707
                *E89ABP          0.410146               0.220764         -0.260950
                *E94ABP          0.378093               0.189385         -0.131993
                *E99ABP          0.417038               0.218935         -0.232120
                *R92ABP          0.548833               0.177238         -0.072918
                *R98ABP          0.607311               0.068012         -0.148796
                *CS901P          -0.776674              -0.068714        0.166541
                *CS902P          -0.349372              -0.161442        0.414293
                *CS903P          0.719799               -0.375822        0.408063
                *CS904P          0.802267               -0.427773        0.021940
                *CS905P          0.795176               -0.070044        -0.109931
                *CS906P          -0.091187              0.190747         -0.694219
                *CS907P          -0.668982              0.248365         0.516979
                *CS908P          0.254913               0.149302         -0.500753

								
To top