The Great Sorting · Region
The standard story of the American South is convergence: the region joined the national party system and stopped being a place apart. The data say something stranger — the South did three different things at once.
For most of the twentieth century the South was the great exception in American politics — a one-party region with its own racial order, its own religion, its own rules. The textbook arc since is one of national absorption: the Solid South cracked, Republicans rose, and the region's politics came to look like everyone else's. V. O. Key's "South apart" became, supposedly, a South like the rest.1
Convergence is half the story. Measure how far Southern attitudes sit from the rest of the country, domain by domain, in standardized units that let religion and redistribution share an axis, and the region splits into three trajectories at once. On some dimensions the gap closed. On some it merely flipped which party owned it. And on others — the cultural ones — the South grew more distinct, not less.
How far the South sits from the rest of the country, by domain
Standardized weighted gap (South minus non-South) over time, in Cohen-d units. Zero = no regional difference. Grouped by what happened.
The realignment is the famous part, and it is genuine. On party identification the South ran more than a quarter of a standard deviation more Democratic than the rest of the country in the mid-century; today it runs Republican. That sign flip — not a narrowing toward zero but a passage through it — is what "realignment" properly means, and it is why the convergence framing misleads. The South did not become average. It crossed to the other side.
The vote crossed over — it did not converge
Democratic share of the major-party presidential vote, South vs. non-South (ANES, weighted).
The South did not become average. On party it crossed to the other side; on culture it pulled further away.
Kuziemko and Washington showed that the partisan exodus of Southern whites was driven by racial conservatism, not economics — defection followed the Democrats' embrace of civil rights, not any divergence in pocketbook interest.2 The attitudinal record here is the sequel to that finding. The economic gap that never explained the realignment has since closed: Southern and non-Southern views on welfare spending and on women's roles have converged toward the national mean. What remains — and has widened — is the cultural and religious distinctiveness that the realignment was built on. Church attendance, ideological self-identification, and traditional views on sexuality all separate the South from the rest of the country more sharply now than in the 1970s.
So the convergence thesis and the persistence thesis are both right, about different things. A region can join the national party system and harden its cultural identity in the same decades; indeed the two may be the same process seen from two angles, as a distinct religious-cultural South found its home in a party that now speaks its language. One country, two timelines — running in opposite directions at once.
Data. ANES Cumulative File (party, vote, racial policy, ideology, church attendance), GSS (gender roles, sexual morality, welfare spending, attendance), and MIT county presidential returns, 1948–2024, each weighted (ANES weight, GSS weight_use, county vote-weighted). South is the Census South (the eleven former Confederate states plus Kentucky and Oklahoma) in every source.
Measure. For each domain and year, the standardized gap is the weighted Southern mean minus the weighted non-Southern mean, divided by the item's weighted standard deviation — Cohen-d units that make domains on different scales comparable. A trajectory is "realigned" if the early and late gaps have opposite signs, "converged" if the late gap is under 70% of the early, and "deepened/persisted" otherwise.
Honest limits. ANES self-reported presidential vote overstates the Democratic share in both regions, so the crossing figure should be read for its gap, not its levels. The South's aggregate vote includes its large Black electorate, which is why the regional Democratic share is higher than a "white South" measure would show; the realignment is sharper still among white Southerners, but the all-respondent gap is the conservative quantity. GSS public geography is region-level only.