Working on a data set and then posting it on this blog (or putting it in a presentation for work) takes a long time. In The Nation Magazine I read an article that must have taken a long time to put together on white voters fear of diversity and their likelihood to vote for Trump. Jason McDaniel and Sean McElwee analyzed individual level data from the Cooperative Congressional Analysis Project (CCAP) where the same individuals were surveyed annually on attitudes regarding the election. This is called a panel survey.
They conducted an analysis of the data from 2016 and from 2012 and 2008 for white voters attitudes towards ethnic diversity and their likelihood to vote for Trump. This is probably a logistic generalized estimating equation model looking at changes over time in voter preference. The model controlled for other potential confounding variables of age, race, education, income, gender, party affiliation, and concerns about their financial well being as well as the predictor variable of interest, concern about immigration and racial resentment.
The results presented in the graph above show that the likelihood of supporting Trump increased with increasingly negative views of ethnic diversity. These views were not as strongly predictive of support for Romney and McCain among white voters when they ran for President. This suggests that Trump was better at tapping into their concerns than the previous two GOP nominees.
Their analysis does coincide with mine of hate group concentration and Trump's % of the vote at the state level. This effect was still significant after controlling for variables like poverty and the % of the population with a bachelor's degree or higher in that state. McElwee and McDaniel (happy St. Patty's day guys) provide validation for my method. It also provides validation for the Southern Poverty Law Center's method for tracking hate groups by showing that the number of hate groups in a state are a reflection of negative attitudes toward diversity in that state.