One of the defining features of America's political divide throughout the post-Cold War era was a growing relationship between a voters' income and their voting behavior, as rich voters started to increasingly lean Republican at the presidential level, and poor voters Democratic. Simultaneously, however, a seemingly paradoxical pattern emerged when the income divide in voting behavior was gauged at the state level, as rich states were increasingly voting for Democratic presidential candidates, and poor ones for Republicans. Gelman, et al., (Quart. J. Pol. Sci. 2007), seeking an explanation for the so-called red-blue paradox, showed through multilevel linear modeling approaches that income had become far less predictive of voting behavior in richer states and more predictive in poorer states by the mid-2000s.
Since then, however, the American political landscape (and, with it, the major parties' electoral coalitions) has undergone a seismic transformation in the wake of the Great Recession. Exit polling data from the 2008, 2012 and 2016 elections suggested that income's influence on voting behavior might have started to wane. However, many scholars approach exit polling data with caution, given widespread concerns about their ability to gain a representative picture of the American electorate. Overall, in recent years, little research into whether the red-blue paradox persists in the wake of the shifting political landscape has been done. In order to provide more insight into that lingering question, this study seeks to build upon Gelman, et al.'s seminal work in the wake of 2008, 2012 and 2016 presidential elections. A combination of linear modeling approaches very similar to theirs was employed.
Thus far, initial attempts to replicate key regression models in the project have succeeded. These models involve assessing the state-level per-capita income's statistical association with the Republican presidential candidate's vote share in each presidential election going back to the 1950s. Then, the slope of each regression line (which highlights the strength of the relationship between income and GOP vote share for that year) was itself plotted on a new graph. An exact replica of Figure 1 in Gelman's paper, illustrating that in recent elections that rich states indeed are voting for Democrats, was produced. However, additional work remains to conduct the necessary modeling on individual-level income-voting patterns, as well as link the state and individual-level models together with multilevel logistic-regression model.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent