Is there such a thing as extreme-right voters? : The case of the French Front national

Since its electoral take-off in 1983-1984, the Front national (FN), a French extreme-right party, has been puzzling scholars, whether political scientists, sociologists, geographers, historians or anthropologists. Among numerous and debated questions, one may wonder how to understand the vote for the Front national. Are individual factors, socioeconomic or psychosocial, sufficient to account for this electoral behavior? Doesn't its strong spacialization suggest that non-individual factors may play an important role? Don't the difficulties that opinion polls encounter when trying to account for this vote plea for an ecological approach of this phenomenon? This paper presents an analysis of the FN vote at 1995, 2002 and 2007 presidential elections (that is four ballot rounds, since Jean-Marie Le Pen, the FN's leader and candidate, qualified for the second round of ballot in 2002). The methodology is based on both ecological analysis and multilevel modeling – I use multilevel models of ecological, i.e. aggregated, data. This makes it possible to take into account the fact that the FN electorates are strongly spatialized, while showing their social bases. The use of the plural for “the FN electorates” is voluntary, since our methodology emphasizes the heterogeneity of these electorates, in both time and space. To put it differently, the hypothesis here is that the social processes behind the FN votes will be all the better shown that one allows them to vary across time and space. The data used are, besides election results, socioeconomic data from the 1999 census: proportion of each socioeconomic status (in a 42-categories detailed nomenclature), summarized through a principal component analysis (PCA); socioeconomic evolution of the population since 1969; proportion of social housing; (log of) mean income; type of commune (urban, peri-urban, rural); distance of the commune to the nearest town over 200.000 inhabitants. All these data are provided at the commune level, the commune being the thinnest administrative and political level of the French system. There are over 36.000 of them in the country, which allows for very powerful statistical treatments of the data. The level-2 unit of analysis is the departement, an important unit of public action and a long-established political arena. There are 90 of them. Six different multilevel regressions have been conducted with these data for every round of ballots; that makes 24 different regressions overall (plus 4 “empty models”). Some detailed results will be presented at the workshop; only the main conclusions can be provided here: 1. From a methodological standpoint, the results show the relevancy and the power of using multilevel modeling on aggregated data. Their explanatory power becomes much greater. 2. The relationships between explaining and explained variables vary across time and space, in some instances greatly. For example, the FN vote is linked to urban communes in some départements and to rural communes in others; and this changes strongly across time. Similarly, the relationship between the FN vote and the mean income level is both very strong and very variable across time and space. 3. In each case, the variations across space tends to be spatialized rather than random. 4. All the variations of the relationships between explaining variables and the FN vote are linked to each other; they are systematic rather than random or specific. This makes it possible to distinguish four different causal configurations – drawn from a K-means clustering procedure led on all the coefficients of all regressions –, which are very spatialized. The map below shows how they stand on a map. a. In the first configuration, the FN vote is rather linked to rural or peri-urban communes, quite working-class, having experienced the decline of peasantry, where the mean income is low, with important social housing. In these areas the FN vote is linked to the decline of the Industrial Revolution France. b. The second configuration is an intermediate type between the three other configurations. c. The third configuration is the one in which the FN vote is, in average, the lowest. It is the Western France, rural and traditionally conservative, of Catholic culture. In these départements, the Fn vote is the highest where the contemporary socioeconomic reality is the most different from this image, where the “productive” economy is located, where the changes have been the greatest. Here, the FN vote is rather linked to modernization and change, rather than to decline. d. Finally, the fourth configuration gathers seven départements, all from the Mediterranean South – the former Midi rouge (Red South). Here, the FN vote is high, rather urban or peri-urban, and rather linked to working-class voters living in deindustrialized areas, that have become high-standing and expensive. In this configuration, the higher the economic inequalities in a commune, the higher the vote for the FN