Hillary Clinton won 84% of the Chicago vote in Tuesday's election. There has been a lot of conversation in the media about whether and how Latinos and Blacks voted.
I will try to shed light on the "how" question with some data. Chicago is remarkably evenly divided among Whites, Blacks, and Latinos. In the
2010 census, out of a total of 2,695,598 residents, there were 872,286 Whites, 854,717 Blacks, and 778,862 Latinos. The next largest group were Asians at 144,903.
To get a rough idea of voting tendencies by race, I looked at the vote counts for all of
Chicago's 77 communities. For example, in my community of Hyde Park, 91.94% of residents voted for Clinton and 3.88% of residents for Trump. Hyde Park is also 46.7% White, 30.4% Black, and 6.3% Latino. Hillary Clinton, in contrast, garnered 49.4% of the vote in Edison Park, which has the highest percentage of whites (88.4%) of any community in Chicago and only 0.3% Blacks and 7.8% Latinos. At the other extreme, 97.3% of the vote in Auburn Gresham, which has the highest percentage of Blacks at 97.8% of any Chicago community (and 0.3% Whites and 0.9% Latinos).
I ran a regression to estimate voting by race in Chicago. I ran separate regressions to predict Clinton votes, Trump votes, and other votes, regressing share against the percentages of Whites, Black, and Latinos in each community. (For the geeks, I included squared terms for racial shares.) The obvious shortcoming of this analysis is that it assumes that groups vote equally. They do not for many reasons.
The regression says that
- 51.4% of Whites voted for Clinton, compared to 96.7% of Blacks, and 90.0% of Latinos.
- 43.5% of Whites voted for Trump, compared to 2.0% of Blacks, and 6.3% of Latinos.
- 5.2% of Whites voted for another candidate, compared to 1.3% of Blacks and 3.7% of Latinos.
It is also interesting to identify the communities which over- and under-voted Clinton according to the model. The biggest pro-Clinton outlier is Lakeview, which came in at 82.8%. With a population of 80.4% White, 3.8% Black, and 7.6% Latino, my regression preidcted Lakeview to vote 62.9% Clinton. On the other side, the biggest pro-Trump outlier is Mount Greenwood, which went 36.2% Clinton. The model predicts Mount Greenwood to have voted 60.5% Clinton, due to its composition of 80.4% White, 3.9% Black, and 7.6% Latino.
The graph below shows predictions and actual voting for Clinton for all 77 communities. To keep the graph from being too cluttered, I've only labelled the extreme residuals on one side or the other.