Fig. 2: Party identification of socioeconomic groups.
From: Charting multidimensional ideological polarization across demographic groups in the USA
Gradient vectors for each attribute computed from their opinion distribution shown in Fig. 1. We discarded cells with less than ten respondents. We slightly smoothed data by applying a different Gaussian filter with a low standard deviation for each attribute. The blue–red colour map obtained for the party attribute is used, with Republicans in dark red and Democrats in dark blue. The vectors are coloured according to the Republicans–Democrats axis, that is, the colour of each point in the vectors is proportional to the radius multiplied by the cosine of their angle with the Republicans–Democrats axis.
