The 2016 election is, and will likely remain for a while, the most closely analyzed election in American history.
(It's also the last presidential election I can find a good county-level dataset for.)
Donald Trump overturned decades of conventional wisdom, taking down the Democratic "blue wall" in the Midwest and
flipping many areas Republicans had feared lost after two straight losses in 2008 and 2012. By looking at where now-President
Trump outperformed previous Republicans, and what demographic characteristics correlated with his success,
we might be able to figure out why.
All data is from the MIT Election Data and Science Lab,
and can be found here. As to why data is missing
in Kansas, Mississippi, Alaska, and a bit of Iowa... you'll have to take it up with them.
National Demographics
Use the following buttons to look at the corresponding variables by county in the US!
Remember, counties don't have the same amounts of people in them -- you can barely see New
York City on the map, but it has a greater population than all but 11 states. If you click on
a county, you can see its percentile rank nationally in the demographic categories on the spider chart below.
Click a county on the map or a bubble on the scatterplot to find out more about the county!
What demographic characteristics are correlated with Donald Trump's success?
OK, that spider chart was cool, but it doesn't tell us anything about support for Trump or Clinton as a whole.
Let's take a look at the demographic factors most correlated with Trump's margin of victory. Remember, there are thousands
of counties in the US, so these might get a little messy. And once again, click a bubble to find out more about the county.
It seems from these scatterplots that Trump performed better in counties
with lower populations, more voters who were white, old, or rural, with lower
median incomes, and with fewer foreign-born or college-educated residents.
Remember, by the way, that we're grouping this data by county. For example, you might
have been surprised that the richest counties in the country voted Democrat, but it's likely
that the countries with the highest median income are urban and suburban, and therefore lean blue;
I'm going out on a limb here, but it's also possible that the richest Republicans are more likely to
live in heavily Democratic urban areas where our metric might not weight them heavily (remember, median,
not average.) Also, if you're wondering, the vertical lines on either end of the chart for "2016 House Republican Vote Share"
are there because the major parties sometimes choose not to contest very safe congressional seats, which might even contain
counties that vote the other way.
But we also see that the spread of these variables is relatively large, and there are plenty of outliers.
There are heavily white Democratic counties, Republican counties with many immigrants, very rural blue counties,
and very educated red counties. Demographic analysis of voting can be dangerous, because it allows us to place
each other into neat little boxes. This approach is simple; it is also deeply flawed. The above graphs tell us a lot
about how Americans vote as a whole, but they tell us nothing about how any single individual votes.