Data visualization by Frankie Dintino and Caitlyn Hampton

This November, the Republican Party’s control of Congress is on the line. Democrats need to net 23 seats in the House to flip that chamber for the first time in eight years, and two seats in the Senate to wrest control from the GOP. Victory will be harder to reach in the latter body, where 10 of the seats Democrats are defending are in states Donald Trump carried in 2016. But recent shifts among the electorate have given Democrats a greater chance at seizing power in the House.

Four groups of voters could play an outsize role in both the outcome of the election and the potential realignment of the two parties’ coalitions in the years to come. They’re voters who live in counties with distinct demographic and voting profiles: majority-minorities counties, majority-white suburban counties, pro-Trump counties with a manufacturing base, and counties that voted for former President Barack Obama in 2012 and then for Trump in 2016. Some of these areas saw shifts in the last election that helped Trump win the White House. Their votes this year could be particularly instructive for Republicans and Democrats alike who are trying to plan their party’s future.

Using data from three sources—the 2017 American Community Survey (ACS), an annual sampling of the U.S. population; the Bureau of Labor Statistics’ employment data for 2017; and the Associated Press’s election results—we’ve tracked how voting preferences for House candidates in those four groups of counties have changed over time, beginning with the 2012 election. (Figures also include uncontested and noncompetitive races.)

Here, a breakdown of those patterns—which we’ll be analyzing after the 2018 election, too.


Majority-Minorities Counties

The data here reflect counties where minorities are in the majority—where the share of white, non-Hispanic residents is less than 50 percent, based on the ACS. Many of these counties are clustered in specific areas of the country, such as Southern California, Texas, and Alaska.

Electoral patterns in these areas remained relatively steady over the past three elections, with most voters consistently pulling the lever for Democratic House candidates. But voter behavior there could nevertheless be significant, especially in statewide elections.

That’s because, in these counties, turnout is an important X factor. If Democrats can convince more of these voters to cast a ballot this year, it could help the party make up for the turnout gap that’s common between Democrats and Republicans in midterm cycles. The 2014 midterm-election results are illustrative of this trend. While the share of voters who picked Democrats was roughly the same in 2012 and 2016, there was a slight dip in 2014. Democratic organizers are well aware of this challenge, and are pushing to mobilize voters accordingly.


White Suburban Counties

A shift toward Democrats this year could flip these reliably Republican areas, which are scattered across the country. Majority-white suburban counties voted for Republican House candidates 58 percent in 2012 and 62 percent in 2014. Democrats made slight inroads in 2016 from the previous election, but this could be the year they pull these areas firmly into their camp.

College-educated white women disillusioned by Trump could be the key for Democrats in these counties. Compared with their male counterparts, these women disapprove of Trump at higher rates. And that’s significant: As my colleague Ron Brownstein has reported, “The group ordinarily leans Democrat, but only slightly.” Democratic House candidates have not carried more than 52 percent of college-educated white women since 1992, according to exit polls. Hillary Clinton got close to that share in 2016, when she won 51 percent, but she still fell short of Obama’s 52 percent in 2008. According to a Washington Post/Schar School poll released this month, Democrats have a 23-point advantage over Republicans among college-educated white women.

Trump’s low approval rating among these women could push them toward Democrats in greater numbers. Recent research has shown that voters’ attitudes of the president correlate with their preference in House and Senate races.


Pro-Trump Manufacturing Counties

The president has visited some of these areas while campaigning for Republican candidates ahead of the midterm elections. These counties supported him in 2016 and have manufacturing at the heart of their economy. Many of them are in the Rust Belt, a region Trump dominated in 2016.

The country’s manufacturing regions weren’t always a Republican stronghold. Support for GOP House candidates has grown since the 2012 presidential election—from 63 percent that year to 67 percent in 2014, and to 70 percent in 2016.

Whether that trend will hold is one of the more interesting questions of the 2018 election. Since Trump took office, manufacturing industries have been “growing at their fastest rates since the financial crisis,” according to the Brookings Institution. But in the Rust Belt in particular, voters are wary of Trump’s trade policy, including his aluminum and steel tariffs. According to a recent Reuters report, it’s “become a thorny issue for congressional candidates as they seek to win votes” this year.


Obama 2012–Trump 2016 Counties

Compared with the other areas we analyzed, these counties—which voted for Obama in 2012 and Trump in 2016—saw the most dramatic shift from Democrat to Republican House candidates between the past two presidential elections. In 2012, 51 percent of voters backed the Democratic contender, compared with 46 percent in 2016. Just as support for Democrats dipped, support for Republicans increased, from 47 percent in 2012 to 50 percent in 2016.

Trump’s surprise victory in the last election was due in part to the voters in these areas, which overlap in some parts of the country with the pro-Trump manufacturing kind: One-third of the nearly 700 counties that supported Obama in 2008 and 2012 voted for Trump in the last presidential election.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.