Looker Blog : Data Matters

Is our story of the 2016 election wrong? Let’s Ask the Data.

Daniel Mintz, Chief Data Evangelist

May 31, 2017

There were a lot of stories told about the 2016 Election. Working Class Whites? Hispanic voters? Unlikable candidates? But now the data’s in. Which holds up?

Earlier this month, the Census released interviews with 130,000 Americans in the wake of November’s election. Now we have the data necessary to test each narrative and see if they stand up in the light of data

Unfortunately, the Census’ tool for accessing the data is very early 2000s. (It’s a Java Applet.) So I downloaded the whole dataset, loaded it into Google BigQuery and threw Looker on top so we can quickly and easily slice and dice the dataset.

So, let’s get started.

A core story of the election was that the White Working Class (WWC) loves Donald Trump, and that they would turn out in huge numbers for him. Political pundits were sure that Trump would bring out voters who usually stayed home. Except, not really.

So then I figured maybe it was a regional thing. In the Midwest, surely the WWC turned out in droves, right? Wrong. In fact, the data shows voting rates actually dropped off from 2012 among the White Working Class voters in Wisconsin, Michigan, Iowa, and Ohio. They did rise in Pennsylvania (a key battleground state), but only by 1% point.

Alright, let’s check out the flip side. Another theory was that Hispanic voters were going to turn out in huge numbers to oppose Donald Trump. Buuuuut, that doesn’t hold up either. Hispanic turnout peaked in 2008 at 50% of eligible voters, but was back down to 48% by 2016.

Ok, so what actually happened? One thing that wasn’t much discussed, but clearly had a big impact, was the dropoff in voting rates among African Americans. This might be expected after two consecutive elections with Barack Obama on the ballot. But still, the dropoff was very significant. Participation of Black men in 2016 was even lower than it had been in 2004 — before Obama was on the ballot. Among Black women, voting dropped seven percentage points from 2012 to 2016.

When we zoom out a bit and look at the long-term trends in American elections, it turns out that 2016 wasn’t so much an aberration as another step in the same long pattern.

The data shows that White, non-Hispanic voters with no college education are making up a smaller and smaller share of the electorate. Hispanic voters are making up a larger and larger share of the electorate, but it’s because the population is growing, not because voting rates are changing. Black voters are relatively stable, with increases in 2008 and 2012, and a dropoff in 2016.

Basically, the story is that there wasn’t a huge new thing that made this election “different.” In short, we live in an evenly split country that’s changing slowly. So even small changes in the makeup of the electorate can swing an election.

Maybe that isn’t the story we were telling ourselves last October, but, well, data has a great way of knocking down alternative facts.

Next Previous

Subscribe for the Latest Posts

Recent Posts