Aggregation Level Matters: Evidences from French Electoral Data

Russo Luana (1) , Laurent Beauguitte (2)

1 CEVIPOF - Centre de recherches politiques de Sciences Po (Sciences Po, CNRS)
2 IDEES - Identité et Différenciation de l’Espace, de l’Environnement et des Sociétés

This working paper, issued from the Cartelec project, raises a method- ological issue regarding ecological inference. Studying abstention on two recent French elections in the Parisian agglomeration, a linear regression is launched on three diff erent levels of aggregation from the nest one (polling station) until cities and electoral districts (circonscriptions). Tests on multicollinearity clearly indicate that employing the nest level of aggregation allows more variables being signi cant, enhances the explanatory power of the model, and improves collinearity amongst dependant variables.

Mots clés: Abstention; Ecological Inference; Linear Regression Model; Multi- collinearity; Parisian Agglomeration; Political Geography

Domaines: Géographie; Science politique

Russo Luana, Laurent Beauguitte. Aggregation Level Matters: Evidences from French Electoral Data. 2012. ⟨halshs-00717982⟩