Fig 1.
Counties with more than 20,000 residents in 2010 are used in the analysis. The percentage of votes for GOP are mapped for years 2004 (A), 2008 (B), 2012 (C), and 2016 (D), showing regional trends.
Fig 2.
Heapmaps based on raw migration data.
The following heatmaps depict the average number of migrants from counties with certain GOP presidential voting rates (x axis) to counties with certain GOP rates (y axis).
Fig 3.
Heatmaps of migrants normalized by population of origin counties.
The following heatmaps depict the average number of migrants from counties with certain GOP presidential voting rates (x axis) to counties with certain GOP rates (y axis). Values are normalized by the population of origin counties.
Fig 4.
Heatmaps of migrants normalized by population of destination counties.
The following heatmaps depict the average number of migrants from counties with certain GOP presidential voting rates (x axis) to counties with certain GOP rates (y axis). Values are normalized by the population of destination counties.
Fig 5.
Heatmaps of migrants normalized by the gravity model.
The following heatmaps depict the log-scale ratio between the volume of actual migration flows and those estimated using a gravity model.
Table 1.
Explanation of variable terms used in the models.
Table 2.
Variables in the two groups of models.
Table 3.
Summary of independent variables included in each model.
Table 4.
Adjusted R2 values of different models.
Table 5.
Result summary of model groups A and B.
Fig 6.
Dyadic effects for model group A.
The integrated dyadic terms’ effects based on GOP_diff, GOP_shared_bias, and GOP_prod between counties with certain proportion votes supporting the GOP candidate for model group A.
Fig 7.
Dyadic effects for model group B.
The integrated dyadic terms’ effects based on GOP_diff, GOP_shared_bias, and GOP_prod between counties with certain proportion votes supporting the GOP candidate for model group B.