#8 Education and the electoral geography of the radical right in the Netherlands

On Wednesday, November 22nd, the elections for the Dutch lower chamber delivered a surprising victory for veteran far-right leader Geert Wilders. His Party for Freedom (Partij voor de Vrijheid) doubled its seat share, emerging as the largest party in Dutch history for the first time. Wilders will now endeavor to form a coalition government—an arduous task in the Dutch fragmented party system. As a reminder: the Netherlands has an extremely proportional electoral system with only one electoral district, which means that a party only needs 0,67% of the vote to obtain a seat in the Lower House.

How can we explain why Wilders did better in some places rather than others? I got hold of data on the electoral score of each party in each municipality in this week’s election and correlated it with data I had used in the past from the Dutch statistics office, encompassing various social and demographic indicators. This allows for a glimpse into the factors that explain why Wilders garnered more support in some areas.

Beginning with the straightforward visualization above —a map depicting the electoral scores akin to a thermometer: areas shaded in blue (cold) indicate voting below the national average for the PVV, while “warm” red areas denote greater support than average. Dark blue areas encompass larger cities, particularly those with universities (such as Amsterdam, Rotterdam, Utrecht, Leiden, and Groningen), where the PVV performed well below average. Conversely, it gained substantial traction in Limburg in the South-East (Wilders’s home region) and the peripheral North-East.

Let's now examine the correlation between the PVV vote share and some demographic and socio-economic indicators. As we have previously mentioned education and universities, the Dutch statistical bureau (CBS) provides data on the educational composition of the population in each municipality. To be direct: this factor is the most significant predictor of the PVV vote share, as illustrated in the scatterplot below.

The horizontal axis represents the percentage of the population in each municipality with what the CBS defines as a low education level (primary schooling, vmbo, the first 3 years of havo/vwo, or mbo 1 in the Dutch system). This relationship exhibits an R-square value of .54, indicating that 54% of the variation in the PVV vote share across municipalities can be accounted for by this variable. This aligns with what we knew of the PVV electorate from previous research, but the correlation here is ...