This paper examines the correlations between basic voter characteristics, economic indicators, migration flows and the change in Republican presidential vote share in the 2016 election. Using county-level data from 2,784 counties and county-equivalents in the United States, the paper develops a regression analysis that finds these correlations do in fact exist. The correlations reveal that 1) basic voter characteristics have strong explanatory power, 2) poverty and unemployment have a statistically significant near-zero effect on voting behavior, and that 3) different immigrant characteristics have conflicting impacts on the Republican presidential vote. I conclude that an inexplicable “Trump effect” does not appear to exist, and that the 2016 election appears to be significantly understandable through regression analysis.

Read the full paper here.