The Atlanta Ring Has Not Reversed Once Since 2000

Sort every county casting at least 25,000 votes by how far its partisan lean has moved since 2000 and the top four entries are all in one metro area. (Drop the turnout floor and the top three are still Atlanta's, but Teton County, Wyoming slots in fourth at 53.11.) Rockdale County, Georgia has moved 80.06 points. Henry, 68.63. Douglas, 59.54. Gwinnett, 51.79. The fifth-largest mover on the list, Travis County, Texas, at 48.08 points, is the first entry that is not part of metro Atlanta.

Large movement over twenty-four years is not, by itself, remarkable. Counties change. What separates the Atlanta ring is that the movement never once turned around. Seven metro Atlanta counties in our file — Rockdale, Henry, Douglas, Gwinnett, Newton, Fayette and Cobb — each recorded six consecutive cycles of same-direction movement. Six is the ceiling. Seven presidential elections from 2000 through 2024 give you exactly six cycle-to-cycle comparisons, and these counties won all six.

That is the finding. It is worth being careful about what it means, because "moved Democratic in every cycle" is not the same sentence as "Democrats did better in every cycle." In at least two of these counties, the two are not the same thing at all.

Six of six, and why that is hard

Lean is a county's two-party margin minus the nation's two-party margin. It strips out the national environment so you can see the county move on its own. Cobb in 2016 is the cleanest illustration in the file: its margin was D+2.2, the country's margin was D+2.2, and its lean was D+0.1. Cobb that year voted essentially like America.

The national environment over this span was not a smooth ramp. It ran D+0.6, R+2.5, D+7.4, D+4.0, D+2.2, D+4.5, R+1.5 — six swings, and the direction of those swings reversed four times out of a possible five. Against that background, a county with an unbroken streak is one whose lean moved the same way through four cycles in which the nation swung Republican and two in which it swung Democratic, without a single cycle of give-back.

Streaks are not a Georgia monopoly. Of the twelve largest movers in our national ranking, seven carry six-cycle streaks; Henrico County, Virginia (44.23 points) and Charles County, Maryland (43.54) both do. But five of those seven are metro Atlanta counties, and the four biggest movers anywhere in the ranking are all in the same ring. The other large movers — Travis (48.08), Hamilton County, Indiana (47.56), Boulder County, Colorado (43.63), Albemarle County, Virginia (42.65) and Williamson County, Texas (41.63) — carry current runs of three, meaning they gave a cycle back somewhere earlier in the span.

The ring, county by county

Rockdale's lean was R+30.7 in 2000 and D+49.3 in 2024. Nothing else in our national file has traveled that far.

These are not rounding-error jurisdictions. The seven counties together recorded about 581,000 votes in 2000 and roughly 1.2 million in 2024. Gwinnett alone went from 191,111 to 420,681; Henry from 38,867 to 129,034; Newton from 18,375 to 59,051. The lean was moving and the electorate was growing at the same time, which is a different situation from a shrinking county changing its mind.

The thinnest link in the chain

Cobb's streak survives on two-tenths of a point. In 2012 its raw margin got worse for Democrats, going from R+9.4 to R+12.4 — about three points of Republican gain. But the country moved further, from D+7.4 to D+4.0. So Cobb's lean improved anyway, from R+16.9 to R+16.7: a relative swing of 0.23, the smallest link in any of these seven chains.

Whether 0.23 deserves to count is a fair argument, and we would rather have it in the open than settle it silently. Round that swing to zero and Cobb's streak is three cycles, not six. We report it as movement because that is what the arithmetic says, and because rounding small numbers to "flat" quietly destroys the exact thing a streak is built to measure. Cobb is also the least dramatic county on this list by total shift, 40.89 points, and it is the one whose unbroken run is most sensitive to how you round.

2024: the two largest stalled, four others accelerated

Gwinnett is the county most likely to be misread from raw results. Its margin fell from D+18.2 in 2020 to D+16.5 in 2024 — 1.7 points the other way. Its lean rose, from D+14.0 to D+18.2, a relative swing of 4.28, because the national margin moved about six points toward Republicans over the same cycle, from D+4.5 to R+1.5. Both statements are true. Gwinnett got less Democratic and more Democratic-leaning in the same election.

Still, 4.28 was Gwinnett's smallest relative swing since 2004, and Cobb's 6.53 was down from 9.95 in 2020. The two largest counties in the group — 420,681 and 401,369 votes respectively — were the two that slowed most. Four of the five smaller counties did the opposite. Rockdale (12.65), Henry (15.23), Douglas (12.28) and Newton (10.23) each posted their largest relative swing since 2008. Fayette, at 9.74, eased slightly from 10.98. That is a split inside a single metro area, in a single cycle, and it is the kind of thing a headline margin will not show you.

What this does not tell you

Fayette is the obvious county to watch, and the obvious place to be careful. It has never produced a Democratic margin in this span, but its 2024 lean was R+1.7 and its raw margin R+3.2. If the pattern of the last six cycles held, it would be the next to cross. A streak is a description of what has already happened, not a mechanism that guarantees more of it, and every streak in this file has the same property: it is intact right up until the cycle it isn't.

Two more limits worth stating. This is county-level presidential lean; it says nothing here about who carried Georgia, about down-ballot races, or about why any of this happened — our file records votes, not reasons. And the shifts are relative. If the national environment keeps moving, a county can continue climbing the lean table while its raw margins go sideways, exactly as Gwinnett's did in 2024.

What the data supports is narrower and, we think, more interesting than a forecast: over seven elections and four reversals in the national environment, seven counties in one metro area moved the same direction every single time, and four of them moved further than any other county we track. If the streak breaks in 2028, that will be a real event with a date on it — which is the point of counting this way.