After the 2016 election, Ryne Rohla embarked on a massive project for the political site Decision Desk HQ: Compiling precinct-level results for the presidential election from every state in the country. Not county — precinct. The smallest units of electoral division. Neighborhoods, essentially, in which the votes of hundreds of people are recorded, as opposed to tens of thousands. The result? A remarkably detailed picture of how the election turned out.

One of the interesting things about Rohla’s data is precisely its granularity. We often talk about the bubbles in which Americans live, surrounding themselves online and in real life with people who generally share their political views. Rohla’s data allows us to evaluate that with much more specificity, drilling down into counties to see the extent to which people in neighborhoods agreed with one another on who should be president. Do you live in a place in which most of your neighbors agreed with your presidential pick? Rohla’s data allows us to see.

Rohla and Decision Desk provided The Washington Post with his data. We used it to figure out how many voters in each state live in places where most of their neighbors agreed with them. As it turns out, this was much more common in states that tended to prefer Donald Trump.

We’ll begin with an interactive. The graphs below show the percentage of voters who live in precincts voting for Trump or Hillary Clinton by various margins. At the left, precincts in which Clinton won by a 91-to-100-percent margin. At the right, precincts in which Trump won by that same differential. Taller columns further to the sides indicate more voters in that state living in more politically homogeneous neighborhoods. Taller columns near the center of the graph shows more voters living in places where the results were mixed.

You’ll notice that the distributions vary widely by state. So let’s consider the outliers.

There’s no formula for defining where a regional bubble begins or ends, so we chose an arbitrary definition: Precincts where Trump or Clinton earned more than 50 percentage points more support than their main opponent. In other words, if in your precinct 77 percent of voters backed Trump and 26 percent backed Clinton — that’s a bubble.

Here are the percentage of voters in each state who live in either a pro-Trump or pro-Clinton bubble.

(Notice that we’re only talking about voters here, not population.)

You’ll notice something that may have been apparent in the interactive: Red states often had much higher percentages of Trump-bubble voters than blue states had Clinton-bubble voters.

The state with the most voters who live in pro-Trump bubbles was Wyoming, where more than half of the population lived in precincts where Trump beat Clinton by more than 50 points. On the Clinton side, the most extreme state was California, where about a third of voters did.

But voters in red states lived in Clinton bubbles, and voters in blue states lived in Trump ones. If we look at the number of people living in bubbles overall, the state with the most voters that live near people who voted like them was Louisiana, in which more than 6-in-10 voters lived in precincts that met that description. That cluster of states in the Deep South and Appalachia — plus Wyoming — were the states in which residents were most likely to live in such bubbles.

For the South, this isn’t actually unusual. In the early 20th century, those states used to routinely prefer Democratic presidential candidates by massive margins — until the civil rights movement flipped the region’s political polarity.

This data should, however, prompt some reconsideration of the nature of our political homogeneity.

Bubbles don’t just occur in major Democratic cities. They happen everywhere.