Something interesting happens if you map the shift in each county from its 2016 presidential vote to the results of its Senate or gubernatorial race: The colors flip.

The reddest parts of the country turn blue, a mark of the margin in 2018 being better for Democrats than it was in 2016. The bluest parts of the country, likewise, largely turn red.

(For this analysis, we compared the 2016 presidential results in each county to the result from the state’s Senate or gubernatorial race. If the state held both contests, the closer of the two race results was used. Data are as of Dec. 1.)

In some cases, the color shifts are a bit of an aberration. In Maine and Vermont, for example, comparing the Democratic and Republican results in the Senate contest excludes votes for the actual winners, independent Sens. Angus King (Maine) and Bernie Sanders (Vt.). If we use their vote totals as the Democratic vote in those states, the map looks like this.

The dark-red-colored counties in Massachusetts, like the ones in Maryland, are not an aberration any more than the dark blue spots in West Virginia are. In each state, results of gubernatorial or Senate elections went in favor of the party that lost the presidential vote in the state. In West Virginia, Sen. Joe Manchin III (D) won reelection despite how red his state has become.

While national politics still colors state races, that West Virginia voters were voting for Manchin and not Hillary Clinton and that Maryland voters were voting for Larry Hogan (R) and not Trump meant each party fared better than it did two years ago.

Look at the Midwest, the region that ended up giving Trump the presidency by narrowly flipping from blue to red two years ago. Chicago is pink, voting more heavily Republican last month than two years ago. Wisconsin, to the northwest, is much more blue, including in the counties adjacent to Milwaukee.

Some of the shift, though, is about the 2016 candidates specifically. Consider the region around Salt Lake City. Utah, more than any red state, rejected Trump in favor of a third-party candidate. In 2018, then, Republicans did a lot better than they had two years prior.

The upshot? The counties that voted most heavily for Trump and Clinton in 2016 were generally the ones that saw the biggest swings back to the opposing party.

The much-noted shift in suburban areas to the Democrats tends to elide that the shift in rural areas was much bigger — a function of how heavily those areas voted for Trump and of how few people live in heavily rural areas, making more-rural counties less likely to swing the results of an election.

All of this is a feature of democracy, not a bug. Residents of Visalia, Calif., choose their own mayor, but then have to agree with others in the area on a representative to serve in the House. They then need to weigh in with the rest of the state on California’s senators and, ultimately, with the rest of the nation on the president. At each step, the candidates are almost necessarily more removed from the locality and its needs. The historic strategy to counter this was to introduce presidential candidates who can appeal broadly to Americans across the board.

That divide, though, may get wider. The 2016 election — with political rhetoric nationalized by cable news and the Internet — allowed Trump to bet that a focus on appealing to Republican base voters would cost him fewer votes than he gained. Clinton, meanwhile, tried to reach across the aisle to Republicans to support her candidacy.

Trump’s bet won — at least in the electoral college. How that bet would hold up against a similarly base-focused Democrat may be the question that defines 2020.