Ecological Inference for the characterization of electoral turnout: The Portuguese Case

Ecological Inference (IE) is a set of statistical methods that estimate the cells of a contingency table when only the marginal totals are known. Based on King’s model (1997) and considering the legislative elections in Portugal between the years 2002 and 2005, we try to find the stability coefficients (the citizens who, keep the same attitude towards voting in both elections, i.e., they opt for vote or abstention in two consecutive elections) and electoral instability (the citizens who vote in one election and opt for abstention on the other, regardless of the order) for every and each of the municipalities in Portugal. In the Portuguese case, King’s method did not give good estimations. Therefore, in order to find spatial homogeneity in terms of the main political tendencies on the elections under study, we propose territorial “reorganization” based on an abstention pattern arising from the HJ-Biplot method (Galindo, 1986). The territorial “reorganisation” has provided 6 groups of provinces to which King’s model was applied in order to find the percentage of electors who voted or chose abstention in both elections, as well as the percentage of floating electors, i.e., the electors who voted in one election and not on the other.