UCLA

Democrats often live with minimal exposure to Republicans and Republicans often live with minimal exposure to Democrats. Leading accounts of partisan segregation document intense and rising segregation both across and within counties. Existing approaches for measuring partisan segregation either drop nonpartisans altogether or model their partisanship. These approaches also use single scale measures of segregation, like the index of dissimilarity. In this thesis, I measure partisan segregation with national voter registration records and a multiscale approach—Expected Kullback-Leibler Divergence—finding stable and relatively low levels of within-county segregation. I further describe patterns in segregation with a spatiotemporal conditional autoregressive model.