Geographic polarization and clustering of partisan voting: A local-level analysis of Stockholm Municipality

Sammanfattning: In the last decade, the topic of political polarization has become a growing concern within academic and public debate. Many highlight the linkage between increased political polarization and democratic problems such as political tribalism and uncivil activism. It is also argued to be a global phenomenon. Despite this, research on political polarization - and its outcome in space (i.e. geographic polarization) - has predominately been conducted in the United States where an increased tendency for the partisan vote to spatially cluster at the local-most level of neighborhoods has been observed .However, this also means that no studies have been conducted in political systems that is not characterized by the binary two-party-system. This study investigates the prevalence of geographic polarization within the Municipality of Stockholm, Sweden between 1998 and 2018. As the topic of geographic polarization has been largely neglected in Sweden, the aim of this study is to gain insight on the prevalence, magnitude and longitudinal development of polarization in the Swedish multi-party system context. The methodological approach of this study is to measure the degree of geographical polarization, at electoral districts, utilizing Global and Local Moran’s I, a common spatial regression statistic within geographic information systems (GIS) which has been used in several studies to examine the degree in which the partisan vote tends to cluster. This is done at [1] the Global level, i.e. the overall tendency for the partisan vote to cluster, and [2] the Local level, the tendency for the partisan vote to cluster at different magnitudes across neighborhoods within the municipality. As results show, within the Municipality of Stockholm there was a decreasing trend of clustering of the partisan vote between the elections of 1998 and 2010. Between 2010 and 2018 there was an increasing trend of the partisan vote to cluster. This is a much more ambiguous result compared to similar studies in the United States, where similar studies show a much clearer linear trend. At the local level, the magnitude in which the partisan vote tend to cluster at different parts of the city is clear. The right-wing vote is mostly clustered at the central part of the city, mainly around the neighborhoods of Norrmalm and Östermalm. The left-wing vote is mostly clustered in the Sub-urbs of Rinkeby-Kista and Spånga-Tensta. This pattern is repeated for all elections. In conclusion, between 1998 and 2018, the Municipality of Stockholm showed a varying degree of geographic polarization, but no unambiguous evidence of an increase. The municipality was just as geographically polarized as 2018 as of 1998. At the local level, the tendency for the partisan vote to cluster at different parts of the municipality is repeated at each election, indicating a stable electoral geography . Hence, claims of an increased geographic polarization within the municipality cannot be supported.