On the night of November 8th 2016 Donald Trump won the US presidential election with 306 electoral votes (vs. 232 for Hilary Clinton). Most notably, all of the numerous election forecasts failed to predict Trump's victory. It was preceded by Trump's unforeseen achievement in the primaries. The main question arises "Who exactly voted for him?". In this regard we analyze the primaries of the Republican Party of 2016. Given the total failure of survey-based polls we base our empirical analysis on socio-demographic factors of the electoral constituency at county-level (2764 counties) to predict Trump's actual voting shares. The regression analyses show that a larger proportion of White Americans leads to an increasing share of votes for Trump. But there is no statistically significant impact of the share of Evangelical Protestants. By contrast, we cannot reject the hypothesis that a large proportion of veterans in the population goes hand in hand with Trump's success. The study also outlines that low education, low income and a high unemployment rate have a positive impact on votes cast for Trump. However, the population density has no influence. Thus, beside the aforementioned socio-demographic variables the rural versus urban difference per se has no explanatory power. Yet, these variables together explain only 13 percent of the variance of the vote shares of Trump. The momentum effect that is, the time a primary took place in a county increases this explained variance. But in fact, state specific differences of Trump's votes are by far the most relevant factor. Thus, the regression analyses prove once more the fact that the United States of America are no homogenous country. Beside differences in income, education, employment and population density, cultural or traditional values and other deep-rooted regional disparities matter as well.
voting US primaries 2016 election Trump Socio-demographic influences bandwagon states