Chapter 10.1

Nationalization of the county vote

Counties moved more nearly in step in 2020 to 2024 than in any other pair of presidential elections since 1868, with a vote-weighted standard deviation of county swings of 4.5 points against 5.0 in 2008 to 2012, the next lowest.

Author: Vishal Singh
Affiliation: NYU Stern
Published: 30 September 2026

DraftFigures and numbers may change as the analysis is rebuilt.what is still being checked

County boundaries are not harmonized across elections; a boundary-consistent rerun is pending.

A county's swing is the change in its two-party margin from one presidential election to the next, in points of margin, and a positive swing is a move toward the Democrats. The spread of swings across counties, measured by its vote-weighted standard deviation, was above 10 points in 28 of the 29 pairs of consecutive elections through 1980 to 1984 and below 10 points in 9 of the 10 pairs from 1984 to 1988 on. It was 4.5 points in 2020 to 2024.

The national swing is the same change in the summed margin of the matched counties, the ones that reported in both elections. A county's relative swing is its swing minus the national swing, which is zero for a county that moved exactly with the country. If every county did, the spread of relative swings would be zero too. Political scientists call the loss of local variation in electoral change nationalization, and the spread of swings is this chapter's measure of it. That measure differs from the one that Amlani and Algara [1] apply to the same county returns for 1872 to 2020.Counties are weighted by their two-party votes in the later election unless a figure says otherwise, so a large county counts for more than a small one. Each pair of elections uses the counties that reported in both, matched on FIPS code, which gives 1,687 counties in 1868 to 1872 and 3,111 in 2020 to 2024. Florida, Mississippi, Texas and Virginia have no county rows for 1868, and Colorado, admitted in 1876, first appears in 1880. The first pair therefore covers 91.7% of the two-party votes cast in 1872, mainly because of those gaps, and every later pair covers at least 97.7% of the two-party votes cast in its later election.

The spread of county swings narrowed after 1984

Figure 1 plots the spread for every pair, at the later election. Through 1980 to 1984 only 1940 to 1944, at 6.4 points, was at or below 10 points. From 1984 to 1988 on only 2012 to 2016, at 10.8 points, was at or above it. The 9 later pairs below the cutoff ranged from 4.5 to 7.9 points, with the lowest value in 2020 to 2024 after 5.7 in 2016 to 2020. That value is 0.5 points below 2008 to 2012, the next lowest pair at 5.0.The counts of pairs above and below 10 points depend on the cutoff. At 8 points they are 28 of 29 and 9 of 10, and at 12 points they are 22 of 29 and 10 of 10.

Figure 1

The spread of county swings was above 10 points in 28 of 29 pairs through 1984 and below it in 9 of 10 since

Standard deviation of county swings in the two-party margin, percentage points; one point per pair of consecutive presidential elections, plotted at the later election

Standard deviation of county swings (points)

Show data

Data for The spread of county swings was above 10 points in 28 of 29 pairs through 1984 and below it in 9 of 10 since
View	Pair of elections	Measure	Spread (points)	Counties
weighted	1868–72	Vote-weighted	31.88	1,687
weighted	1872–76	Vote-weighted	29.11	2,151
weighted	1876–80	Vote-weighted	12.83	2,216
weighted	1880–84	Vote-weighted	11.84	2,283
weighted	1884–88	Vote-weighted	10.25	2,345
weighted	1888–92	Vote-weighted	17.31	2,424
weighted	1892–96	Vote-weighted	39.03	2,654
weighted	1896–1900	Vote-weighted	22.85	2,695
weighted	1900–04	Vote-weighted	16.53	2,680
weighted	1904–08	Vote-weighted	14.91	2,707
weighted	1908–12	Vote-weighted	23.94	2,832
weighted	1912–16	Vote-weighted	26.08	2,942
weighted	1916–20	Vote-weighted	17.04	3,018
weighted	1920–24	Vote-weighted	16.74	3,048
weighted	1924–28	Vote-weighted	31.49	3,083
weighted	1928–32	Vote-weighted	20.79	3,095
weighted	1932–36	Vote-weighted	14.55	3,095
weighted	1936–40	Vote-weighted	12.05	3,094
weighted	1940–44	Vote-weighted	6.44	3,093
weighted	1944–48	Vote-weighted	11.85	3,095
weighted	1948–52	Vote-weighted	13.28	3,094
weighted	1952–56	Vote-weighted	10.04	3,097
weighted	1956–60	Vote-weighted	14.58	3,100
weighted	1960–64	Vote-weighted	21.24	3,102
weighted	1964–68	Vote-weighted	21.29	3,107
weighted	1968–72	Vote-weighted	17.72	3,109
weighted	1972–76	Vote-weighted	19.29	3,110
weighted	1976–80	Vote-weighted	10.90	3,112
weighted	1980–84	Vote-weighted	11.19	3,112
weighted	1984–88	Vote-weighted	6.15	3,114
weighted	1988–92	Vote-weighted	7.61	3,114
weighted	1992–96	Vote-weighted	7.86	3,114
weighted	1996–2000	Vote-weighted	7.85	3,113
weighted	2000–04	Vote-weighted	6.13	3,113
weighted	2004–08	Vote-weighted	7.68	3,114
weighted	2008–12	Vote-weighted	4.97	3,114
weighted	2012–16	Vote-weighted	10.81	3,112
weighted	2016–20	Vote-weighted	5.70	3,113
weighted	2020–24	Vote-weighted	4.47	3,111
both	1868–72	Vote-weighted	31.88	1,687
both	1872–76	Vote-weighted	29.11	2,151
both	1876–80	Vote-weighted	12.83	2,216
both	1880–84	Vote-weighted	11.84	2,283
both	1884–88	Vote-weighted	10.25	2,345
both	1888–92	Vote-weighted	17.31	2,424
both	1892–96	Vote-weighted	39.03	2,654
both	1896–1900	Vote-weighted	22.85	2,695
both	1900–04	Vote-weighted	16.53	2,680
both	1904–08	Vote-weighted	14.91	2,707
both	1908–12	Vote-weighted	23.94	2,832
both	1912–16	Vote-weighted	26.08	2,942
both	1916–20	Vote-weighted	17.04	3,018
both	1920–24	Vote-weighted	16.74	3,048
both	1924–28	Vote-weighted	31.49	3,083
both	1928–32	Vote-weighted	20.79	3,095
both	1932–36	Vote-weighted	14.55	3,095
both	1936–40	Vote-weighted	12.05	3,094
both	1940–44	Vote-weighted	6.44	3,093
both	1944–48	Vote-weighted	11.85	3,095
both	1948–52	Vote-weighted	13.28	3,094
both	1952–56	Vote-weighted	10.04	3,097
both	1956–60	Vote-weighted	14.58	3,100
both	1960–64	Vote-weighted	21.24	3,102
both	1964–68	Vote-weighted	21.29	3,107
both	1968–72	Vote-weighted	17.72	3,109
both	1972–76	Vote-weighted	19.29	3,110
both	1976–80	Vote-weighted	10.90	3,112
both	1980–84	Vote-weighted	11.19	3,112
both	1984–88	Vote-weighted	6.15	3,114
both	1988–92	Vote-weighted	7.61	3,114
both	1992–96	Vote-weighted	7.86	3,114
both	1996–2000	Vote-weighted	7.85	3,113
both	2000–04	Vote-weighted	6.13	3,113
both	2004–08	Vote-weighted	7.68	3,114
both	2008–12	Vote-weighted	4.97	3,114
both	2012–16	Vote-weighted	10.81	3,112
both	2016–20	Vote-weighted	5.70	3,113
both	2020–24	Vote-weighted	4.47	3,111
both	1868–72	Each county counted once	40.76	1,687
both	1872–76	Each county counted once	29.19	2,151
both	1876–80	Each county counted once	18.71	2,216
both	1880–84	Each county counted once	16.52	2,283
both	1884–88	Each county counted once	16.63	2,345
both	1888–92	Each county counted once	30.51	2,424
both	1892–96	Each county counted once	49.35	2,654
both	1896–1900	Each county counted once	25.60	2,695
both	1900–04	Each county counted once	23.91	2,680
both	1904–08	Each county counted once	18.98	2,707
both	1908–12	Each county counted once	27.53	2,832
both	1912–16	Each county counted once	26.23	2,942
both	1916–20	Each county counted once	21.79	3,018
both	1920–24	Each county counted once	18.48	3,048
both	1924–28	Each county counted once	36.48	3,083
both	1928–32	Each county counted once	23.56	3,095
both	1932–36	Each county counted once	14.41	3,095
both	1936–40	Each county counted once	16.03	3,094
both	1940–44	Each county counted once	8.90	3,093
both	1944–48	Each county counted once	16.75	3,095
both	1948–52	Each county counted once	18.50	3,094
both	1952–56	Each county counted once	15.20	3,097
both	1956–60	Each county counted once	17.14	3,100
both	1960–64	Each county counted once	36.80	3,102
both	1964–68	Each county counted once	37.96	3,107
both	1968–72	Each county counted once	27.04	3,109
both	1972–76	Each county counted once	25.91	3,110
both	1976–80	Each county counted once	12.75	3,112
both	1980–84	Each county counted once	14.79	3,112
both	1984–88	Each county counted once	8.57	3,114
both	1988–92	Each county counted once	10.12	3,114
both	1992–96	Each county counted once	8.22	3,114
both	1996–2000	Each county counted once	9.40	3,113
both	2000–04	Each county counted once	7.25	3,113
both	2004–08	Each county counted once	10.18	3,114
both	2008–12	Each county counted once	6.23	3,114
both	2012–16	Each county counted once	10.44	3,112
both	2016–20	Each county counted once	5.19	3,113
both	2020–24	Each county counted once	3.36	3,111

The line is mostly above the dashed horizontal line, at 10 points, through 1980 to 1984 and mostly below it from 1984 to 1988 on. Its lowest point is 2020 to 2024, after a rise to 10.8 points in 2012 to 2016. Each point is the standard deviation of county swings for one pair, weighted by two-party votes in the later election. The other view also counts each county once. Shaded columns mark the two Reconstruction pairs and pairs that include an election in which candidates outside the two major parties won at least 10% of the vote.Note: Full counts of votes: there is no sampling, so the estimates have no confidence interval.Source: County presidential returns, 1868-2024 (Algara and Amlani for 1868-2020, MIT Election Data and Science Lab for 2024); full counts, not survey estimates; county swings weighted by two-party votes in the later election unless stated

Shaded columns mark pairs that are hard to compare with the rest. The two Reconstruction pairs come first. The others include an election in which candidates outside the two major parties won at least 10% of all votes, which happened in 1892, 1912, 1924, 1968, 1992 and 1996, and the widest pair, 1892 to 1896, at 39.0 points, is one of them. The two-party margin leaves those votes out, and 7 of the ten largest spreads fall in shaded pairs.

The spread rose in 4 of the 10 pairs from 1984 to 1988 on, and the largest rise, 5.8 points, came in 2012 to 2016. That pair had the lowest share of votes within 5 points of the national swing, 34.0%, of any pair after 1980 to 1984, and the correlation of county margins with the previous election's fell to 0.96 from 0.99. Both recovered in 2016 to 2020. The spread measures how unevenly counties moved in 2016 and not which kinds of counties moved; Chapter 12.7, Education, ethnicity and the county shift, 2016 to 2024 follows the county shift from that year to 2024.

Nationalization has a long literature. Stokes [2] distinguished national from sub-national electoral forces, and Bartels [3] estimated them from state-level presidential returns for 1868 to 1996 and found a system more nationalized than a century earlier. Claggett, Flanigan and Zingale [4] found no trend toward increasing nationalization of the vote from 1842 to 1970. Amlani and Algara [1] use the county file this chapter draws on, for presidential, Senate and gubernatorial elections from 1872 to 2020, and conclude that contemporary elections are highly nationalized and that presidential and Senate elections are more partisan than in any period since the Civil War. Caramani [5] traced the same process across Western European constituencies, and Hopkins [6] documents it in American attitudes and behavior with surveys and experiments. The spread of county swings does not reproduce the measures of these studies. How far counties sit from the national result in a single election, as against how far they moved, is the subject of Chapter 10.2, Distance from the national vote, 1868 to 2024.

Relative swings narrowed to a ridge near zero

Figure 2 shows the distribution behind the spread for nine pairs, the ones ending in 1876, 1896, 1916, 1936, 1956, 1976, 1996 and 2016, and the latest pair. The earliest ridges are low and reach past the edges of the plot: in 1892 to 1896, 10.0% of votes were cast in counties whose relative swing lay more than 60 points from zero on one side or the other. The last ridge is narrow, and 73.1% of the votes in 2020 to 2024 fall within 5 points of zero, against 22.7% in 1892 to 1896 and 20.3% in 1972 to 1976. Its mass sits above zero. The vote-weighted median county swing was +1.8 points relative to the national swing, 19.8% of votes were cast in counties that swung 5 or more points more toward the Democrats than the nation, and 7.1% in counties that swung that far more toward the Republicans. The reference behind this offset is discussed with Figure 4.

Figure 2

Relative county swings lay far from zero in early pairs and sat in a narrow band just above zero in 2020 to 2024

Relative swing is a county's swing in the Democratic minus Republican margin minus the national swing, percentage points; height is the share of a pair's votes in each 2-point bin, nine pairs of presidential elections

Show data

Data for Relative county swings lay far from zero in early pairs and sat in a narrow band just above zero in 2020 to 2024
Pair of elections	Bin center (points)	Share of votes (%)
1872–76	−59	0.01
1872–76	−57	0.00
1872–76	−55	0.56
1872–76	−53	0.05
1872–76	−51	0.06
1872–76	−49	0.02
1872–76	−47	0.15
1872–76	−45	0.04
1872–76	−43	0.13
1872–76	−41	0.06
1872–76	−39	0.15
1872–76	−37	0.08
1872–76	−35	0.05
1872–76	−33	0.15
1872–76	−31	0.10
1872–76	−29	0.23
1872–76	−27	1.29
1872–76	−25	0.38
1872–76	−23	0.91
1872–76	−21	1.60
1872–76	−19	2.39
1872–76	−17	3.03
1872–76	−15	4.91
1872–76	−13	6.25
1872–76	−11	8.41
1872–76	−9	6.32
1872–76	−7	5.53
1872–76	−5	6.06
1872–76	−3	8.93
1872–76	−1	4.71
1872–76	+1	5.20
1872–76	+3	3.39
1872–76	+5	2.84
1872–76	+7	2.46
1872–76	+9	1.55
1872–76	+11	1.53
1872–76	+13	1.20
1872–76	+15	0.64
1872–76	+17	0.40
1872–76	+19	0.57
1872–76	+21	2.02
1872–76	+23	0.50
1872–76	+25	0.34
1872–76	+27	0.37
1872–76	+29	0.29
1872–76	+31	0.34
1872–76	+33	0.39
1872–76	+35	0.29
1872–76	+37	0.17
1872–76	+39	0.33
1872–76	+41	0.42
1872–76	+43	0.16
1872–76	+45	0.35
1872–76	+47	0.21
1872–76	+49	0.33
1872–76	+51	0.35
1872–76	+53	0.37
1872–76	+55	0.13
1872–76	+57	0.29
1872–76	+59	0.34
1892–96	−59	0.03
1892–96	−57	0.10
1892–96	−55	0.13
1892–96	−53	0.17
1892–96	−51	0.06
1892–96	−49	0.04
1892–96	−47	0.31
1892–96	−45	0.03
1892–96	−43	0.33
1892–96	−41	0.15
1892–96	−39	0.47
1892–96	−37	1.13
1892–96	−35	0.37
1892–96	−33	0.67
1892–96	−31	1.39
1892–96	−29	0.61
1892–96	−27	0.94
1892–96	−25	3.79
1892–96	−23	4.79
1892–96	−21	3.71
1892–96	−19	2.95
1892–96	−17	1.22
1892–96	−15	2.40
1892–96	−13	1.99
1892–96	−11	1.78
1892–96	−9	3.55
1892–96	−7	2.70
1892–96	−5	5.26
1892–96	−3	3.25
1892–96	−1	5.92
1892–96	+1	4.88
1892–96	+3	3.98
1892–96	+5	3.71
1892–96	+7	4.95
1892–96	+9	3.67
1892–96	+11	3.57
1892–96	+13	2.96
1892–96	+15	2.38
1892–96	+17	1.44
1892–96	+19	1.47
1892–96	+21	0.69
1892–96	+23	0.80
1892–96	+25	1.08
1892–96	+27	0.57
1892–96	+29	0.40
1892–96	+31	0.25
1892–96	+33	0.16
1892–96	+35	0.55
1892–96	+37	0.33
1892–96	+39	0.19
1892–96	+41	0.28
1892–96	+43	0.17
1892–96	+45	0.24
1892–96	+47	0.13
1892–96	+49	0.08
1892–96	+51	0.16
1892–96	+53	0.22
1892–96	+55	0.13
1892–96	+57	0.06
1892–96	+59	0.23
1912–16	−59	0.22
1912–16	−57	0.12
1912–16	−55	0.28
1912–16	−53	0.15
1912–16	−51	0.06
1912–16	−49	0.19
1912–16	−47	0.58
1912–16	−45	0.39
1912–16	−43	0.32
1912–16	−41	0.28
1912–16	−39	0.05
1912–16	−37	0.69
1912–16	−35	0.29
1912–16	−33	0.76
1912–16	−31	0.64
1912–16	−29	1.54
1912–16	−27	1.03
1912–16	−25	0.75
1912–16	−23	1.16
1912–16	−21	1.22
1912–16	−19	1.89
1912–16	−17	1.78
1912–16	−15	2.36
1912–16	−13	1.61
1912–16	−11	2.12
1912–16	−9	11.26
1912–16	−7	2.83
1912–16	−5	2.36
1912–16	−3	1.67
1912–16	−1	2.50
1912–16	+1	2.69
1912–16	+3	2.47
1912–16	+5	4.58
1912–16	+7	4.85
1912–16	+9	3.33
1912–16	+11	3.17
1912–16	+13	4.13
1912–16	+15	3.24
1912–16	+17	3.95
1912–16	+19	3.85
1912–16	+21	2.13
1912–16	+23	2.77
1912–16	+25	2.18
1912–16	+27	3.12
1912–16	+29	1.55
1912–16	+31	1.01
1912–16	+33	0.63
1912–16	+35	1.20
1912–16	+37	0.38
1912–16	+39	0.27
1912–16	+41	0.16
1912–16	+43	0.32
1912–16	+45	0.21
1912–16	+47	0.18
1912–16	+49	0.13
1912–16	+51	0.10
1912–16	+53	0.07
1912–16	+55	0.20
1912–16	+57	0.03
1912–16	+59	0.07
1932–36	−59	0.00
1932–36	−57	0.01
1932–36	−55	0.00
1932–36	−53	0.03
1932–36	−51	0.00
1932–36	−49	0.01
1932–36	−47	0.02
1932–36	−45	0.04
1932–36	−43	0.04
1932–36	−41	0.06
1932–36	−39	0.10
1932–36	−37	0.31
1932–36	−35	0.24
1932–36	−33	0.46
1932–36	−31	0.44
1932–36	−29	0.71
1932–36	−27	1.11
1932–36	−25	1.51
1932–36	−23	1.15
1932–36	−21	1.86
1932–36	−19	2.11
1932–36	−17	2.38
1932–36	−15	2.98
1932–36	−13	2.77
1932–36	−11	3.84
1932–36	−9	2.73
1932–36	−7	3.72
1932–36	−5	7.98
1932–36	−3	7.22
1932–36	−1	6.08
1932–36	+1	2.85
1932–36	+3	5.23
1932–36	+5	3.83
1932–36	+7	8.02
1932–36	+9	6.06
1932–36	+11	7.33
1932–36	+13	2.14
1932–36	+15	2.38
1932–36	+17	2.71
1932–36	+19	1.56
1932–36	+21	0.69
1932–36	+23	0.94
1932–36	+25	1.00
1932–36	+27	0.49
1932–36	+29	3.36
1932–36	+31	0.42
1932–36	+33	0.00
1932–36	+35	0.00
1932–36	+37	0.16
1932–36	+39	0.22
1932–36	+41	0.28
1932–36	+43	0.03
1932–36	+45	0.00
1932–36	+47	0.20
1932–36	+49	0.00
1932–36	+51	0.08
1932–36	+53	0.00
1932–36	+55	0.00
1932–36	+57	0.00
1932–36	+59	0.00
1952–56	−59	0.01
1952–56	−57	0.01
1952–56	−55	0.02
1952–56	−53	0.01
1952–56	−51	0.00
1952–56	−49	0.00
1952–56	−47	0.04
1952–56	−45	0.02
1952–56	−43	0.01
1952–56	−41	0.03
1952–56	−39	0.02
1952–56	−37	0.01
1952–56	−35	0.01
1952–56	−33	0.03
1952–56	−31	0.01
1952–56	−29	0.03
1952–56	−27	0.06
1952–56	−25	0.60
1952–56	−23	0.18
1952–56	−21	0.21
1952–56	−19	0.37
1952–56	−17	0.51
1952–56	−15	0.94
1952–56	−13	2.41
1952–56	−11	4.47
1952–56	−9	8.54
1952–56	−7	6.62
1952–56	−5	7.28
1952–56	−3	8.11
1952–56	−1	6.47
1952–56	+1	10.87
1952–56	+3	4.37
1952–56	+5	13.83
1952–56	+7	7.39
1952–56	+9	3.87
1952–56	+11	3.39
1952–56	+13	1.94
1952–56	+15	2.02
1952–56	+17	1.12
1952–56	+19	1.05
1952–56	+21	0.89
1952–56	+23	0.58
1952–56	+25	0.56
1952–56	+27	0.21
1952–56	+29	0.27
1952–56	+31	0.10
1952–56	+33	0.05
1952–56	+35	0.07
1952–56	+37	0.07
1952–56	+39	0.02
1952–56	+41	0.02
1952–56	+43	0.02
1952–56	+45	0.00
1952–56	+47	0.00
1952–56	+49	0.00
1952–56	+51	0.01
1952–56	+53	0.00
1952–56	+55	0.00
1952–56	+57	0.01
1952–56	+59	0.00
1972–76	−59	0.00
1972–76	−57	0.00
1972–76	−55	0.00
1972–76	−53	0.00
1972–76	−51	0.01
1972–76	−49	0.00
1972–76	−47	0.00
1972–76	−45	0.00
1972–76	−43	0.00
1972–76	−41	0.23
1972–76	−39	0.01
1972–76	−37	0.00
1972–76	−35	0.13
1972–76	−33	0.15
1972–76	−31	0.53
1972–76	−29	0.27
1972–76	−27	0.56
1972–76	−25	0.50
1972–76	−23	1.27
1972–76	−21	1.42
1972–76	−19	1.76
1972–76	−17	4.16
1972–76	−15	3.96
1972–76	−13	4.54
1972–76	−11	10.97
1972–76	−9	9.86
1972–76	−7	4.73
1972–76	−5	4.16
1972–76	−3	6.08
1972–76	−1	3.83
1972–76	+1	3.49
1972–76	+3	4.15
1972–76	+5	2.25
1972–76	+7	3.87
1972–76	+9	3.85
1972–76	+11	3.14
1972–76	+13	1.31
1972–76	+15	1.95
1972–76	+17	1.64
1972–76	+19	1.06
1972–76	+21	0.99
1972–76	+23	1.73
1972–76	+25	1.20
1972–76	+27	0.92
1972–76	+29	0.85
1972–76	+31	0.97
1972–76	+33	0.52
1972–76	+35	0.84
1972–76	+37	0.62
1972–76	+39	0.51
1972–76	+41	0.61
1972–76	+43	0.39
1972–76	+45	0.39
1972–76	+47	0.34
1972–76	+49	0.37
1972–76	+51	0.26
1972–76	+53	0.23
1972–76	+55	0.20
1972–76	+57	0.38
1972–76	+59	0.12
1992–96	−59	0.00
1992–96	−57	0.00
1992–96	−55	0.00
1992–96	−53	0.00
1992–96	−51	0.00
1992–96	−49	0.00
1992–96	−47	0.00
1992–96	−45	0.00
1992–96	−43	0.00
1992–96	−41	0.00
1992–96	−39	0.00
1992–96	−37	0.00
1992–96	−35	0.00
1992–96	−33	0.02
1992–96	−31	0.02
1992–96	−29	0.01
1992–96	−27	0.02
1992–96	−25	0.06
1992–96	−23	0.05
1992–96	−21	0.36
1992–96	−19	0.43
1992–96	−17	0.40
1992–96	−15	1.22
1992–96	−13	2.14
1992–96	−11	2.31
1992–96	−9	4.32
1992–96	−7	7.22
1992–96	−5	6.41
1992–96	−3	9.24
1992–96	−1	10.91
1992–96	+1	13.27
1992–96	+3	9.34
1992–96	+5	6.44
1992–96	+7	6.25
1992–96	+9	4.87
1992–96	+11	4.88
1992–96	+13	3.78
1992–96	+15	3.93
1992–96	+17	1.13
1992–96	+19	0.52
1992–96	+21	0.19
1992–96	+23	0.00
1992–96	+25	0.01
1992–96	+27	0.21
1992–96	+29	0.01
1992–96	+31	0.00
1992–96	+33	0.00
1992–96	+35	0.00
1992–96	+37	0.00
1992–96	+39	0.00
1992–96	+41	0.00
1992–96	+43	0.00
1992–96	+45	0.00
1992–96	+47	0.00
1992–96	+49	0.00
1992–96	+51	0.00
1992–96	+53	0.00
1992–96	+55	0.00
1992–96	+57	0.00
1992–96	+59	0.00
2012–16	−59	0.00
2012–16	−57	0.00
2012–16	−55	0.00
2012–16	−53	0.00
2012–16	−51	0.00
2012–16	−49	0.00
2012–16	−47	0.00
2012–16	−45	0.00
2012–16	−43	0.00
2012–16	−41	0.01
2012–16	−39	0.01
2012–16	−37	0.03
2012–16	−35	0.08
2012–16	−33	0.11
2012–16	−31	0.28
2012–16	−29	0.49
2012–16	−27	0.49
2012–16	−25	0.72
2012–16	−23	1.21
2012–16	−21	1.13
2012–16	−19	1.41
2012–16	−17	2.19
2012–16	−15	2.64
2012–16	−13	2.38
2012–16	−11	4.13
2012–16	−9	3.47
2012–16	−7	6.19
2012–16	−5	5.31
2012–16	−3	5.33
2012–16	−1	9.77
2012–16	+1	4.89
2012–16	+3	7.38
2012–16	+5	7.62
2012–16	+7	6.08
2012–16	+9	7.27
2012–16	+11	6.75
2012–16	+13	3.44
2012–16	+15	5.03
2012–16	+17	2.72
2012–16	+19	0.78
2012–16	+21	0.11
2012–16	+23	0.01
2012–16	+25	0.21
2012–16	+27	0.00
2012–16	+29	0.09
2012–16	+31	0.00
2012–16	+33	0.00
2012–16	+35	0.25
2012–16	+37	0.00
2012–16	+39	0.00
2012–16	+41	0.00
2012–16	+43	0.00
2012–16	+45	0.00
2012–16	+47	0.00
2012–16	+49	0.00
2012–16	+51	0.00
2012–16	+53	0.00
2012–16	+55	0.00
2012–16	+57	0.00
2012–16	+59	0.00
2020–24	−59	0.00
2020–24	−57	0.00
2020–24	−55	0.00
2020–24	−53	0.00
2020–24	−51	0.00
2020–24	−49	0.00
2020–24	−47	0.00
2020–24	−45	0.00
2020–24	−43	0.00
2020–24	−41	0.00
2020–24	−39	0.00
2020–24	−37	0.00
2020–24	−35	0.00
2020–24	−33	0.00
2020–24	−31	0.00
2020–24	−29	0.00
2020–24	−27	0.00
2020–24	−25	0.00
2020–24	−23	0.01
2020–24	−21	0.13

Showing the first 500 of 540 rows.

Early ridges are low and wide; the last ridge is narrow and its mass sits just above zero. Each ridge is one pair, labeled by its years; its height at a point is the share of the pair's votes cast in counties whose relative swing falls in a bin of 2 points, on one height scale for all ridges.Note: Full counts of votes: there is no sampling, so the estimates have no confidence interval.Note: The bins run from minus 60 to plus 60 points and swings beyond that range are not drawn; 1892 to 1896 has the largest share beyond it among the pairs shown.Source: County presidential returns, 1868-2024 (Algara and Amlani for 1868-2020, MIT Election Data and Science Lab for 2024); full counts, not survey estimates; county swings weighted by two-party votes in the later election unless stated

Counties near the national swing, 1868 to 2024

Figure 3 shows the same relative swings county by county, for the pair of elections chosen with the slider. The palest class holds counties within 5 points of the national swing. In 2020 to 2024, 75.6% of matched counties fall in that class, the largest share of the 39 pairs. The smallest share, 8.5%, is in 1924 to 1928, a pair that includes a strong third-party vote, and 1972 to 1976 gives 21.2%. The national swing in the last pair was −6.1 points, a move toward the Republicans, so a blue county in that pair moved less toward the Republicans than the nation did, or toward the Democrats.

Figure 3

The share of counties within 5 points of the national swing rose from 21.2% in 1972 to 1976 to 75.6% in 2020 to 2024

Change in each county's Democratic minus Republican margin, minus the national change, percentage points, for each of 39 pairs of presidential elections

Loading county values

County swing minus national swing (points)

More RepublicanMore Democratic

Values outside −40 to +40 take the end colors.

No data (3,143 counties)

Select a class to show only those places.

Find a county

Hover or tap a county for its value, or search by name. Zoom with the buttons on the map, then drag to move it.

Alaska draws as no data. Its election returns are reported by election district, and the district codes cannot be matched to boroughs and census areas.

In 1972 to 1976 the South is mostly blue: 93.5% of its counties swung 5 or more points more toward the Democrats than the nation did. In 2020 to 2024, 75.6% of counties are in the palest class, within 5 points of the national swing. Blue means a county moved more toward the Democrats, or less toward the Republicans, than the nation did; the national swing in 2020 to 2024 was −6.1 points. Move the slider to change the pair. Each county is colored by its own change in two-party margin minus the change in the summed margin of all matched counties.Note: Full counts of votes: there is no sampling, so the estimates have no confidence interval.Note: Counties are drawn on 2020 boundaries and matched across elections on FIPS code. A county with no return in one of the two elections, or whose code is absent from the 2020 map, draws as no data. Values are rounded to one decimal place, except that a county within 5 points of the national swing is never rounded to the edge of the palest class, so that class holds the counties counted in the text.Source: County presidential returns, 1868-2024 (Algara and Amlani for 1868-2020, MIT Election Data and Science Lab for 2024); full counts, not survey estimates; county swings weighted by two-party votes in the later election unless stated

The share of votes cast in counties within 5 points of the national swing is a coarser summary than the spread, and it is easier to read. Figure 4 plots it. The share was 73.3% in 2008 to 2012, fell to 34.0% in 2012 to 2016 and was 73.1% in 2020 to 2024. No pair through 1980 to 1984 exceeded 62.0%, the value for 1940 to 1944. The two records therefore fall in different pairs: the spread is lowest in 2020 to 2024, and the share of votes near the national swing is highest in 2008 to 2012. The spread gives large deviations more weight than the share does, which is why the two measures can rank neighboring pairs differently. Counted by county, the share was 75.6% in 2020 to 2024 and 58.8% in 2008 to 2012. It is higher than the vote share in 2020 to 2024 and lower in 2008 to 2012.

The reference matters when the spread is small. Measured against the vote-weighted mean of county swings instead of the national swing, 80.0% of votes and 86.3% of counties were within 5 points in 2020 to 2024, against 73.1% and 75.6% for the national swing. This chapter uses the national swing.The national swing is the change in the summed margin, so it reflects where votes were cast as well as how counties voted. The vote-weighted mean of county swings gives each county its later-election weight in both elections and leaves that change out. In 2020 to 2024 the first was −6.1 points and the second −4.8, a gap of −1.3.

Figure 4

The share of votes near the national swing was 73.3% in 2008 to 2012 and 73.1% in 2020 to 2024, the two highest of 39 pairs

Share of votes cast in counties whose swing is within 5 points of the national swing, percent; one point per pair of consecutive presidential elections, plotted at the later election

Show data

Data for The share of votes near the national swing was 73.3% in 2008 to 2012 and 73.1% in 2020 to 2024, the two highest of 39 pairs
Counted by	Pair of elections	Reference swing	Share within the band (%)	Counties
Votes	1868–72	National swing	22.5	1,687
Votes	1868–72	Mean county swing	22.7	1,687
Counties	1868–72	National swing	16.5	1,687
Counties	1868–72	Mean county swing	16.7	1,687
Votes	1872–76	National swing	26.1	2,151
Votes	1872–76	Mean county swing	16.8	2,151
Counties	1872–76	National swing	22.9	2,151
Counties	1872–76	Mean county swing	17.4	2,151
Votes	1876–80	National swing	61.1	2,216
Votes	1876–80	Mean county swing	60.4	2,216
Counties	1876–80	National swing	47.9	2,216
Counties	1876–80	Mean county swing	47.7	2,216
Votes	1880–84	National swing	51.5	2,283
Votes	1880–84	Mean county swing	51.5	2,283
Counties	1880–84	National swing	39.5	2,283
Counties	1880–84	Mean county swing	39.5	2,283
Votes	1884–88	National swing	60.8	2,345
Votes	1884–88	Mean county swing	58.7	2,345
Counties	1884–88	National swing	50.1	2,345
Counties	1884–88	Mean county swing	48.3	2,345
Votes	1888–92	National swing	60.6	2,424
Votes	1888–92	Mean county swing	61.6	2,424
Counties	1888–92	National swing	43.5	2,424
Counties	1888–92	Mean county swing	44.1	2,424
Votes	1892–96	National swing	22.7	2,654
Votes	1892–96	Mean county swing	23.6	2,654
Counties	1892–96	National swing	18.4	2,654
Counties	1892–96	Mean county swing	21.3	2,654
Votes	1896–1900	National swing	37.8	2,695
Votes	1896–1900	Mean county swing	32.6	2,695
Counties	1896–1900	National swing	30.1	2,695
Counties	1896–1900	Mean county swing	25.3	2,695
Votes	1900–04	National swing	29.1	2,680
Votes	1900–04	Mean county swing	28.3	2,680
Counties	1900–04	National swing	21.2	2,680
Counties	1900–04	Mean county swing	22.0	2,680
Votes	1904–08	National swing	25.4	2,707
Votes	1904–08	Mean county swing	25.2	2,707
Counties	1904–08	National swing	23.1	2,707
Counties	1904–08	Mean county swing	22.8	2,707
Votes	1908–12	National swing	16.8	2,832
Votes	1908–12	Mean county swing	17.7	2,832
Counties	1908–12	National swing	14.8	2,832
Counties	1908–12	Mean county swing	15.8	2,832
Votes	1912–16	National swing	11.7	2,942
Votes	1912–16	Mean county swing	11.5	2,942
Counties	1912–16	National swing	12.7	2,942
Counties	1912–16	Mean county swing	11.9	2,942
Votes	1916–20	National swing	21.4	3,018
Votes	1916–20	Mean county swing	23.1	3,018
Counties	1916–20	National swing	15.8	3,018
Counties	1916–20	Mean county swing	15.9	3,018
Votes	1920–24	National swing	30.6	3,048
Votes	1920–24	Mean county swing	30.7	3,048
Counties	1920–24	National swing	24.8	3,048
Counties	1920–24	Mean county swing	25.1	3,048
Votes	1924–28	National swing	10.7	3,083
Votes	1924–28	Mean county swing	9.9	3,083
Counties	1924–28	National swing	8.5	3,083
Counties	1924–28	Mean county swing	7.9	3,083
Votes	1928–32	National swing	16.9	3,095
Votes	1928–32	Mean county swing	16.7	3,095
Counties	1928–32	National swing	17.0	3,095
Counties	1928–32	Mean county swing	16.5	3,095
Votes	1932–36	National swing	28.1	3,095
Votes	1932–36	Mean county swing	27.0	3,095
Counties	1932–36	National swing	22.6	3,095
Counties	1932–36	Mean county swing	20.8	3,095
Votes	1936–40	National swing	41.0	3,094
Votes	1936–40	Mean county swing	39.9	3,094
Counties	1936–40	National swing	25.0	3,094
Counties	1936–40	Mean county swing	24.4	3,094
Votes	1940–44	National swing	62.0	3,093
Votes	1940–44	Mean county swing	62.4	3,093
Counties	1940–44	National swing	54.3	3,093
Counties	1940–44	Mean county swing	54.8	3,093
Votes	1944–48	National swing	32.9	3,095
Votes	1944–48	Mean county swing	32.1	3,095
Counties	1944–48	National swing	27.3	3,095
Counties	1944–48	Mean county swing	27.5	3,095
Votes	1948–52	National swing	36.7	3,094
Votes	1948–52	Mean county swing	35.3	3,094
Counties	1948–52	National swing	24.3	3,094
Counties	1948–52	Mean county swing	24.2	3,094
Votes	1952–56	National swing	39.1	3,097
Votes	1952–56	Mean county swing	40.1	3,097
Counties	1952–56	National swing	31.9	3,097
Counties	1952–56	Mean county swing	32.4	3,097
Votes	1956–60	National swing	27.1	3,100
Votes	1956–60	Mean county swing	26.9	3,100
Counties	1956–60	National swing	19.7	3,100
Counties	1956–60	Mean county swing	19.3	3,100
Votes	1960–64	National swing	29.4	3,102
Votes	1960–64	Mean county swing	29.2	3,102
Counties	1960–64	National swing	22.2	3,102
Counties	1960–64	Mean county swing	22.0	3,102
Votes	1964–68	National swing	37.1	3,107
Votes	1964–68	Mean county swing	37.5	3,107
Counties	1964–68	National swing	20.6	3,107
Counties	1964–68	Mean county swing	20.8	3,107
Votes	1968–72	National swing	28.9	3,109
Votes	1968–72	Mean county swing	33.6	3,109
Counties	1968–72	National swing	18.6	3,109
Counties	1968–72	Mean county swing	19.6	3,109
Votes	1972–76	National swing	20.3	3,110
Votes	1972–76	Mean county swing	20.4	3,110
Counties	1972–76	National swing	21.2	3,110
Counties	1972–76	Mean county swing	20.4	3,110
Votes	1976–80	National swing	37.1	3,112
Votes	1976–80	Mean county swing	37.8	3,112
Counties	1976–80	National swing	29.7	3,112
Counties	1976–80	Mean county swing	29.3	3,112
Votes	1980–84	National swing	33.7	3,112
Votes	1980–84	Mean county swing	33.6	3,112
Counties	1980–84	National swing	27.8	3,112
Counties	1980–84	Mean county swing	27.9	3,112
Votes	1984–88	National swing	61.7	3,114
Votes	1984–88	Mean county swing	62.4	3,114
Counties	1984–88	National swing	42.2	3,114
Counties	1984–88	Mean county swing	42.4	3,114
Votes	1988–92	National swing	48.6	3,114
Votes	1988–92	Mean county swing	49.6	3,114
Counties	1988–92	National swing	40.0	3,114
Counties	1988–92	Mean county swing	39.4	3,114
Votes	1992–96	National swing	48.8	3,114
Votes	1992–96	Mean county swing	48.8	3,114
Counties	1992–96	National swing	47.5	3,114
Counties	1992–96	Mean county swing	44.9	3,114
Votes	1996–2000	National swing	44.8	3,113
Votes	1996–2000	Mean county swing	46.3	3,113
Counties	1996–2000	National swing	32.2	3,113
Counties	1996–2000	Mean county swing	31.4	3,113
Votes	2000–04	National swing	61.6	3,113
Votes	2000–04	Mean county swing	61.8	3,113
Counties	2000–04	National swing	54.2	3,113
Counties	2000–04	Mean county swing	53.7	3,113
Votes	2004–08	National swing	56.5	3,114
Votes	2004–08	Mean county swing	56.1	3,114
Counties	2004–08	National swing	43.9	3,114
Counties	2004–08	Mean county swing	43.5	3,114
Votes	2008–12	National swing	73.3	3,114
Votes	2008–12	Mean county swing	73.0	3,114
Counties	2008–12	National swing	58.8	3,114
Counties	2008–12	Mean county swing	58.3	3,114
Votes	2012–16	National swing	34.0	3,112
Votes	2012–16	Mean county swing	34.2	3,112
Counties	2012–16	National swing	24.6	3,112
Counties	2012–16	Mean county swing	23.6	3,112
Votes	2016–20	National swing	69.1	3,113
Votes	2016–20	Mean county swing	69.1	3,113
Counties	2016–20	National swing	73.9	3,113
Counties	2016–20	Mean county swing	73.9	3,113
Votes	2020–24	National swing	73.1	3,111
Votes	2020–24	Mean county swing	80.0	3,111
Counties	2020–24	National swing	75.6	3,111
Counties	2020–24	Mean county swing	86.3	3,111

The line reaches its highest values in 2008 to 2012 and 2020 to 2024, and no pair through 1980 to 1984 exceeds 62.0%. Each county's swing is measured against the national swing, the change in the summed margin; the text compares that reference with the vote-weighted mean of county swings. Switch views to count counties instead of votes.Note: Full counts of votes: there is no sampling, so the estimates have no confidence interval.Note: A county is near the national swing if its own swing differs from it by less than 5 points.Source: County presidential returns, 1868-2024 (Algara and Amlani for 1868-2020, MIT Election Data and Science Lab for 2024); full counts, not survey estimates; county swings weighted by two-party votes in the later election unless stated

County swings in the South were wider until 1992

Figure 5 splits the matched counties into the Census Bureau's South region and the rest and gives the spread inside each. Within the South the spread was larger than in the rest of the country in each of 13 consecutive pairs from 1940 to 1944 through 1988 to 1992. The run ends on small gaps: the South's spread exceeded the rest's by 1.6 points in 1984 to 1988 and 0.1 points in 1988 to 1992, and the rest of the country's was the larger by 1.7 points in 1992 to 1996. In 1960 to 1964 the spread was 33.5 points inside the South and 11.4 outside, and in 1972 to 1976 it was 20.7 and 9.7. The shaded pair of 1964 to 1968, which includes the 1968 third-party vote, was higher still inside the South, at 39.2 points. The shaded columns in Figure 5 mark the same pairs as in Figure 1. Before 1940 the pattern is mixed, and the rest of the country had the larger spread in 10 of the 18 pairs through 1936 to 1940.

Switching to South against the nation compares the South's own margin swing with the national swing. A negative value means the South moved further toward the Republicans than the nation did. The South's margin swing differed from the nation's by +21.9 points in 1972 to 1976. Since 1984 to 1988 the difference has stayed within 2.4 points, and in 2020 to 2024 the spread is 4.4 points in the South and 4.5 elsewhere.

Figure 5

The spread of county swings was 33.5 points in the South and 11.4 elsewhere in 1960 to 1964, and 4.4 and 4.5 in 2020 to 2024

Standard deviation of county swings within each region, percentage points of the two-party margin, weighted by votes; one point per pair of elections, plotted at the later election

Standard deviation of county swings within the region (points)

Show data

Data for The spread of county swings was 33.5 points in the South and 11.4 elsewhere in 1960 to 1964, and 4.4 and 4.5 in 2020 to 2024
View	Pair of elections	Region	Value (points)	Counties
spread	1868–72	South	39.98	670
spread	1868–72	Rest of the country	28.60	1,017
offset	1868–72	South	−6.13	670
offset	1868–72	Rest of the country	1.02	1,017
spread	1872–76	South	23.21	1,059
spread	1872–76	Rest of the country	31.15	1,092
offset	1872–76	South	5.38	1,059
offset	1872–76	Rest of the country	−1.96	1,092
spread	1876–80	South	22.54	1,084
spread	1876–80	Rest of the country	6.53	1,132
offset	1876–80	South	3.57	1,084
offset	1876–80	Rest of the country	−0.47	1,132
spread	1880–84	South	16.59	1,106
spread	1880–84	Rest of the country	9.22	1,177
offset	1880–84	South	−3.32	1,106
offset	1880–84	Rest of the country	1.05	1,177
spread	1884–88	South	17.09	1,142
spread	1884–88	Rest of the country	6.22	1,203
offset	1884–88	South	1.77	1,142
offset	1884–88	Rest of the country	−0.39	1,203
spread	1888–92	South	23.35	1,164
spread	1888–92	Rest of the country	14.20	1,260
offset	1888–92	South	7.85	1,164
offset	1888–92	Rest of the country	−1.93	1,260
spread	1892–96	South	24.96	1,197
spread	1892–96	Rest of the country	42.19	1,457
offset	1892–96	South	−4.18	1,197
offset	1892–96	Rest of the country	1.29	1,457
spread	1896–1900	South	17.22	1,188
spread	1896–1900	Rest of the country	24.15	1,507
offset	1896–1900	South	−0.17	1,188
offset	1896–1900	Rest of the country	0.79	1,507
spread	1900–04	South	16.19	1,164
spread	1900–04	Rest of the country	13.74	1,516
offset	1900–04	South	18.70	1,164
offset	1900–04	Rest of the country	−3.18	1,516
spread	1904–08	South	12.32	1,185
spread	1904–08	Rest of the country	13.67	1,522
offset	1904–08	South	−12.42	1,185
offset	1904–08	Rest of the country	2.79	1,522
spread	1908–12	South	22.53	1,320
spread	1908–12	Rest of the country	24.00	1,512
offset	1908–12	South	−5.59	1,320
offset	1908–12	Rest of the country	0.38	1,512
spread	1912–16	South	23.35	1,356
spread	1912–16	Rest of the country	26.50	1,586
offset	1912–16	South	4.51	1,356
offset	1912–16	Rest of the country	0.40	1,586
spread	1916–20	South	15.08	1,375
spread	1916–20	Rest of the country	15.27	1,643
offset	1916–20	South	10.84	1,375
offset	1916–20	Rest of the country	−3.53	1,643
spread	1920–24	South	14.72	1,382
spread	1920–24	Rest of the country	16.16	1,666
offset	1920–24	South	12.63	1,382
offset	1920–24	Rest of the country	−2.66	1,666
spread	1924–28	South	25.15	1,405
spread	1924–28	Rest of the country	24.93	1,678
offset	1924–28	South	−38.58	1,405
offset	1924–28	Rest of the country	9.72	1,678
spread	1928–32	South	24.02	1,411
spread	1928–32	Rest of the country	18.64	1,684
offset	1928–32	South	15.17	1,411
offset	1928–32	Rest of the country	−3.45	1,684
spread	1932–36	South	13.20	1,412
spread	1932–36	Rest of the country	14.49	1,683
offset	1932–36	South	−6.15	1,412
offset	1932–36	Rest of the country	1.58	1,683
spread	1936–40	South	10.70	1,411
spread	1936–40	Rest of the country	11.44	1,683
offset	1936–40	South	10.11	1,411
offset	1936–40	Rest of the country	−1.79	1,683
spread	1940–44	South	9.64	1,411
spread	1940–44	Rest of the country	5.25	1,682
offset	1940–44	South	−4.90	1,411
offset	1940–44	Rest of the country	0.95	1,682
spread	1944–48	South	17.49	1,412
spread	1944–48	Rest of the country	10.30	1,683
offset	1944–48	South	−5.27	1,412
offset	1944–48	Rest of the country	1.25	1,683
spread	1948–52	South	18.79	1,411
spread	1948–52	Rest of the country	10.40	1,683
offset	1948–52	South	−6.54	1,411
offset	1948–52	Rest of the country	0.39	1,683
spread	1952–56	South	13.77	1,414
spread	1952–56	Rest of the country	8.76	1,683
offset	1952–56	South	−2.82	1,414
offset	1952–56	Rest of the country	0.79	1,683
spread	1956–60	South	18.23	1,417
spread	1956–60	Rest of the country	12.63	1,683
offset	1956–60	South	−8.70	1,417
offset	1956–60	Rest of the country	2.11	1,683
spread	1960–64	South	33.49	1,415
spread	1960–64	Rest of the country	11.44	1,687
offset	1960–64	South	−16.52	1,415
offset	1960–64	Rest of the country	4.91	1,687
spread	1964–68	South	39.16	1,419
spread	1964–68	Rest of the country	9.43	1,688
offset	1964–68	South	13.63	1,419
offset	1964–68	Rest of the country	−4.20	1,688
spread	1968–72	South	25.37	1,422
spread	1968–72	Rest of the country	9.99	1,687
offset	1968–72	South	−14.73	1,422
offset	1968–72	Rest of the country	5.15	1,687
spread	1972–76	South	20.70	1,422
spread	1972–76	Rest of the country	9.66	1,688
offset	1972–76	South	21.93	1,422
offset	1972–76	Rest of the country	−8.04	1,688
spread	1976–80	South	11.08	1,425
spread	1976–80	Rest of the country	10.57	1,687
offset	1976–80	South	−2.92	1,425
offset	1976–80	Rest of the country	0.96	1,687
spread	1980–84	South	12.04	1,425
spread	1980–84	Rest of the country	8.01	1,687
offset	1980–84	South	−9.28	1,425
offset	1980–84	Rest of the country	4.21	1,687
spread	1984–88	South	6.97	1,425
spread	1984–88	Rest of the country	5.39	1,689
offset	1984–88	South	−2.43	1,425
offset	1984–88	Rest of the country	1.12	1,689
spread	1988–92	South	7.65	1,425
spread	1988–92	Rest of the country	7.54	1,689
offset	1988–92	South	0.91	1,425
offset	1988–92	Rest of the country	−0.09	1,689
spread	1992–96	South	6.66	1,424
spread	1992–96	Rest of the country	8.33	1,690
offset	1992–96	South	−1.24	1,424
offset	1992–96	Rest of the country	0.67	1,690
spread	1996–2000	South	9.13	1,423
spread	1996–2000	Rest of the country	6.92	1,690
offset	1996–2000	South	−1.98	1,423
offset	1996–2000	Rest of the country	1.08	1,690
spread	2000–04	South	6.58	1,423
spread	2000–04	Rest of the country	5.83	1,690
offset	2000–04	South	−0.73	1,423
offset	2000–04	Rest of the country	0.49	1,690
spread	2004–08	South	9.30	1,423
spread	2004–08	Rest of the country	6.34	1,691
offset	2004–08	South	−1.83	1,423
offset	2004–08	Rest of the country	1.32	1,691
spread	2008–12	South	4.50	1,423
spread	2008–12	Rest of the country	5.18	1,691
offset	2008–12	South	1.17	1,423
offset	2008–12	Rest of the country	−0.46	1,691
spread	2012–16	South	9.20	1,422
spread	2012–16	Rest of the country	11.61	1,690
offset	2012–16	South	1.04	1,422
offset	2012–16	Rest of the country	−0.34	1,690
spread	2016–20	South	6.68	1,422
spread	2016–20	Rest of the country	5.03	1,691
offset	2016–20	South	0.14	1,422
offset	2016–20	Rest of the country	−0.20	1,691
spread	2020–24	South	4.40	1,422
spread	2020–24	Rest of the country	4.50	1,689
offset	2020–24	South	0.36	1,422
offset	2020–24	Rest of the country	−0.01	1,689

The South's line sits above the rest of the country's from 1944 to 1992 and the two lines are level in 2020 to 2024. The South is the Census Bureau's South region. Each line is the vote-weighted standard deviation of county swings among the counties in the region. The other view gives the South's own margin swing compared with the national swing.Note: Full counts of votes: there is no sampling, so the estimates have no confidence interval.Note: The South is the Census Bureau's South region: the states from Delaware, Maryland and West Virginia through Texas and Oklahoma, and the District of Columbia, which has returns from 1964. The national swing used in the second view is the one computed on all matched counties.Source: County presidential returns, 1868-2024 (Algara and Amlani for 1868-2020, MIT Election Data and Science Lab for 2024); full counts, not survey estimates; county swings weighted by two-party votes in the later election unless stated

County margins and winners track the last election more closely

A second view of nationalization asks how well a county's margin in one election predicts its margin in the next. Figure 6 plots the vote-weighted correlation between county margins in each pair, and between margins four elections apart. The correlation with the previous election was 0.33 in 1892 to 1896, 0.51 in 1960 to 1964 and 0.51 in 1972 to 1976. It was above 0.9 in 3 of the 29 pairs through 1980 to 1984 and in all 10 pairs from 1984 to 1988 on. It is 0.99 in each of 2008 to 2012, 2016 to 2020 and 2020 to 2024, and 0.96 in 2012 to 2016. Across four elections the correlation was 0.85 between 1972 and 1988 and is 0.90 between 2008 and 2024.

The correlation depends on how spread out county margins are as well as on the spread of swings, so it accompanies the spread measure and does not replace it.

Figure 6

County margins correlated with the previous election's margins above 0.9 in 3 of 29 pairs through 1984 and in all 10 pairs since

Vote-weighted correlation between a county's two-party margin and its margin in an earlier election, plotted at the later election

Show data

Data for County margins correlated with the previous election's margins above 0.9 in 3 of 29 pairs through 1984 and in all 10 pairs since
Elections compared	Compared with	Correlation	Counties
1868–72	Previous election	0.542	1,687
1872–76	Previous election	0.646	2,151
1876–80	Previous election	0.887	2,216
1880–84	Previous election	0.903	2,283
1884–88	Previous election	0.911	2,345
1888–92	Previous election	0.821	2,424
1892–96	Previous election	0.334	2,654
1896–1900	Previous election	0.741	2,695
1900–04	Previous election	0.863	2,680
1904–08	Previous election	0.897	2,707
1908–12	Previous election	0.664	2,832
1912–16	Previous election	0.590	2,942
1916–20	Previous election	0.821	3,018
1920–24	Previous election	0.895	3,048
1924–28	Previous election	0.527	3,083
1928–32	Previous election	0.685	3,095
1932–36	Previous election	0.838	3,095
1936–40	Previous election	0.887	3,094
1940–44	Previous election	0.968	3,093
1944–48	Previous election	0.874	3,095
1948–52	Previous election	0.842	3,094
1952–56	Previous election	0.897	3,097
1956–60	Previous election	0.761	3,100
1960–64	Previous election	0.514	3,102
1964–68	Previous election	0.551	3,107
1968–72	Previous election	0.674	3,109
1972–76	Previous election	0.510	3,110
1976–80	Previous election	0.854	3,112
1980–84	Previous election	0.862	3,112
1984–88	Previous election	0.959	3,114
1988–92	Previous election	0.940	3,114
1992–96	Previous election	0.943	3,114
1996–2000	Previous election	0.958	3,113
2000–04	Previous election	0.975	3,113
2004–08	Previous election	0.964	3,114
2008–12	Previous election	0.989	3,114
2012–16	Previous election	0.958	3,112
2016–20	Previous election	0.988	3,113
2020–24	Previous election	0.992	3,111
1868–84	Four elections earlier	0.673	1,688
1872–88	Four elections earlier	0.499	2,155
1876–92	Four elections earlier	0.709	2,227
1880–96	Four elections earlier	0.593	2,287
1884–1900	Four elections earlier	0.756	2,336
1888–1904	Four elections earlier	0.784	2,398
1892–1908	Four elections earlier	0.555	2,642
1896–1912	Four elections earlier	0.533	2,693
1900–16	Four elections earlier	0.807	2,710
1904–20	Four elections earlier	0.830	2,720
1908–24	Four elections earlier	0.860	2,850
1912–28	Four elections earlier	0.282	2,943
1916–32	Four elections earlier	0.750	3,020
1920–36	Four elections earlier	0.524	3,052
1924–40	Four elections earlier	0.573	3,080
1928–44	Four elections earlier	0.533	3,094
1932–48	Four elections earlier	0.733	3,095
1936–52	Four elections earlier	0.689	3,092
1940–56	Four elections earlier	0.724	3,091
1944–60	Four elections earlier	0.588	3,092
1948–64	Four elections earlier	0.053	3,090
1952–68	Four elections earlier	0.653	3,094
1956–72	Four elections earlier	0.233	3,095
1960–76	Four elections earlier	0.763	3,100
1964–80	Four elections earlier	0.451	3,104
1968–84	Four elections earlier	0.724	3,108
1972–88	Four elections earlier	0.847	3,109
1976–92	Four elections earlier	0.644	3,112
1980–96	Four elections earlier	0.721	3,111
1984–2000	Four elections earlier	0.828	3,112
1988–2004	Four elections earlier	0.810	3,111
1992–2008	Four elections earlier	0.796	3,112
1996–2012	Four elections earlier	0.842	3,112
2000–16	Four elections earlier	0.842	3,111
2004–20	Four elections earlier	0.864	3,112
2008–24	Four elections earlier	0.900	3,112

The solid line stays above 0.9 from 1984 to 1988 on and is lowest in 1892 to 1896, at 0.33. The solid line correlates each county's margin with its margin in the previous election and the dashed line with its margin four elections earlier, both weighted by two-party votes in the later election.Note: Full counts of votes: there is no sampling, so the estimates have no confidence interval.Note: A county needs returns in both elections to enter a correlation.Source: County presidential returns, 1868-2024 (Algara and Amlani for 1868-2020, MIT Election Data and Science Lab for 2024); full counts, not survey estimates; county swings weighted by two-party votes in the later election unless stated

County winners persisted in the same way. Applying the national swing to every county gives a test: a county's predicted margin is its margin in the earlier election plus the national swing, and its winner is called from the sign. Keeping each county's earlier winner instead gets 50.9% of counties wrong in 1972 to 1976, 1,584 counties, and 2.8% in 2020 to 2024, 87 counties. The national swing cuts the number of counties called wrong by 26% in 1972 to 1976 and 28% in 2020 to 2024, to 37.8% and 2.0% of counties. The fall in the error rate therefore follows the fall in the number of counties that changed winner and does not reflect a better fit of the uniform swing. The national swing comes from the later election's own results, so the test is a fit and not a forecast, and the count depends on how many counties sit close to an even split as well as on how far they swing.

The 2020 to 2024 low holds under other measures

The lead of 2020 to 2024 over 2008 to 2012 is small, so this section asks whether it depends on the measure or on the counties included. Leaving out any one state or the District of Columbia in turn keeps 2020 to 2024 the lowest pair, with a gap to 2008 to 2012 of 0.3 to 0.7 points.

The Votes and counties view of Figure 1 adds a line that counts each county once. That gives a larger spread in 35 of the 39 pairs, so the small counties, which count for less in the vote-weighted figure, usually have the more varied swings. The vote-weighted spread is the larger one in 1932 to 1936, 2012 to 2016, 2016 to 2020 and 2020 to 2024. In 2020 to 2024 it is 4.5 points against 3.4 for the county-equal version, so votes are concentrated in counties whose swings sit further from the mean than the typical county's do. The county-equal lead is wider: 3.4 points in 2020 to 2024 against 5.2 in 2016 to 2020, the next lowest, a gap of 1.8 points.

The main measure could also depend on how small counties, the changing county set or extreme counties are handled. There are five measures, each computed with vote weights and with each county counted once, which gives 10 versions, and 2020 to 2024 is the lowest pair in every one (Figure 7). The measures are the standard deviation, the standard deviation after dropping counties with fewer than 1,000 two-party votes in either election, the standard deviation on the fixed panel of 1,652 counties with votes in all 40 elections, and two that give little weight to extreme counties, the median absolute deviation and the interquartile range.The median absolute deviation is the middle value of the distances of county swings from their median, and the interquartile range is the distance between the swings at the 25th and 75th percentiles. Each is scaled to equal the standard deviation for a normal distribution.

Figure 7

2020 to 2024 is the lowest pair under all 10 versions of the spread

Spread of county swings for 2020 to 2024 and for the next lowest pair under five measures, each weighted by votes and with each county counted once, percentage points

Measure	Counted	2020 to 2024 (points)	Next lowest pair	Its value (points)	Margin (points)
Standard deviation	Weighted by votes	4.47	2008–12	4.97	0.51
Standard deviation without small counties	Weighted by votes	4.47	2008–12	4.97	0.50
Standard deviation on the fixed panel of counties	Weighted by votes	4.19	2008–12	5.00	0.81
Median absolute deviation, scaled	Weighted by votes	3.98	2008–12	4.19	0.21
Interquartile range, scaled	Weighted by votes	3.96	2008–12	4.27	0.31
Standard deviation	Each county counted once	3.36	2016–20	5.19	1.83
Standard deviation without small counties	Each county counted once	3.31	2016–20	5.12	1.82
Standard deviation on the fixed panel of counties	Each county counted once	2.84	2016–20	4.35	1.51
Median absolute deviation, scaled	Each county counted once	2.74	2016–20	4.20	1.47
Interquartile range, scaled	Each county counted once	2.76	2016–20	4.22	1.46

Show data

Data for 2020 to 2024 is the lowest pair under all 10 versions of the spread
counting	latest	margin	measure	next_pair	next_value
Weighted by votes	4.47	0.505	Standard deviation	2008–12	4.97
Weighted by votes	4.47	0.503	Standard deviation without small counties	2008–12	4.97
Weighted by votes	4.19	0.811	Standard deviation on the fixed panel of counties	2008–12	5.00
Weighted by votes	3.98	0.213	Median absolute deviation, scaled	2008–12	4.19
Weighted by votes	3.96	0.312	Interquartile range, scaled	2008–12	4.27
Each county counted once	3.36	1.83	Standard deviation	2016–20	5.19
Each county counted once	3.31	1.82	Standard deviation without small counties	2016–20	5.12
Each county counted once	2.84	1.51	Standard deviation on the fixed panel of counties	2016–20	4.35
Each county counted once	2.74	1.47	Median absolute deviation, scaled	2016–20	4.20
Each county counted once	2.76	1.46	Interquartile range, scaled	2016–20	4.22

In every row the value for 2020 to 2024 is below the next lowest pair's. The smallest margin is 0.21 points.Note: Full counts of votes: there is no sampling, so the estimates have no confidence interval.Note: Dropping small counties removes counties with fewer than 1,000 two-party votes in either election of a pair. The fixed panel keeps the 1,652 counties with a positive two-party vote in all 40 elections. The scaled median absolute deviation and the scaled interquartile range are rescaled to equal the standard deviation for a normal distribution.Source: County presidential returns, 1868-2024 (Algara and Amlani for 1868-2020, MIT Election Data and Science Lab for 2024); full counts, not survey estimates; county swings weighted by two-party votes in the later election unless stated

What these data cannot show

County returns describe places, not voters. A county that swings with the nation can contain groups that moved in opposite directions, and returns cannot separate voters who changed their choice from people who started or stopped voting or moved in or out, a limit known as ecological inference. The national swing has a related limit: it is a change in a summed margin, so it counts where votes were cast as well as how counties voted, and in 2020 to 2024 it differed from the mean county swing by −1.3 points.

The returns are full counts, so the figures carry no sampling intervals, and differences between measures come from how they are defined and which counties enter. The county set is one of those choices. Counties are matched on FIPS code and boundaries are not harmonized, so a county that gained or lost territory is treated as one unit, and the county-equal figures depend on how many small counties exist. The 53 county rows that have Democratic and Republican votes but no total vote count are left out, among them 27 Idaho counties in 1912, and keeping them changes no pair's spread by more than 0.12 points. Returns before 1965 count the votes of the people allowed to vote, and in much of the South barriers to voting kept many Black citizens from the ballot until the Voting Rights Act, so early margins describe smaller electorates than the populations of those counties. The two-party margin also leaves out third-party votes, which is why the shaded pairs in Figure 1 are the least comparable with the rest.

The chapter measures how uniformly counties moved and does not say why. How much of the county vote census characteristics explain is the subject of Chapter 12.1, Five axes of the county vote, and ticket splitting is taken up in Chapter 10.3, Down the ballot. Chapter 10.4, The Electoral College (planned) turns state margins into electoral outcomes, and Chapter 15.1, Party and place compares how far partisans' issue positions differ between deep-red and deep-blue counties with the gap between the parties. The series also makes no forecast: the spread was 5.0 points in 2008 to 2012 and 10.8 in 2012 to 2016.

Data sources

County election returnsCounty presidential returns, 1868-2024: Algara and Amlani (1868-2020) and MIT Election Data and Science Lab (2024), repaired fileYears: President 1868-2024 (40 elections); House and Senate 1990-2024 · Weights: none (full counts); county swings weighted by two-party votes in the later election, county-equal in one view · License: The Heideman file is CC BY 4.0 (doi:10.5281/zenodo.22976553), but the state records behind it keep their own terms, and the sites of Florida, Indiana, Colorado and Wisconsin restrict reuse.

All data sources

Methods

Sample: county presidential returns for the 40 elections from 1868 to 2024 (Algara and Amlani for 1868 to 2020, MIT Election Data and Science Lab for 2024), from the repaired file whose 2024 totals are corrected for 173 counties in South Carolina, Wisconsin and West Virginia. Rows dropped before any calculation: 41 Alaska rows (Alaska reports election districts and has no returns before 2024), 50 rows with no FIPS code (defunct counties), 88 rows with no vote count, 53 more rows with Democratic and Republican votes but no total vote count (27 of them Idaho counties in 1912, so Idaho has no county in the pairs 1908 to 1912 and 1912 to 1916; keeping these rows changes no pair's vote-weighted standard deviation by more than 0.12 points and leaves 2020 to 2024 the lowest pair) and 74 rows with no two-party votes. Coverage shares are computed on the rows kept. Pairs: each pair of consecutive elections uses the counties present in both, matched on FIPS code, so a pair holds 1,687 to 3,114 counties; county boundaries are not harmonized. Estimand: the two-party margin is 100 times the Democratic minus the Republican votes over their sum; a county's swing is its margin in the later election minus its margin in the earlier one; the national swing is the same change in the summed margin of the matched counties; the relative swing is the county swing minus the national swing. Spread: the standard deviation of county swings about their own mean, population divisor, weighted by two-party votes in the later election (county-equal in one view). Share near the national swing: the absolute difference from the reference is below 5 points, counted by votes and by counties, against the national swing and against the vote-weighted mean county swing. Robustness: counties with fewer than 1,000 two-party votes in either election dropped; a fixed panel of 1,652 counties with a positive two-party vote in all 40 elections; the median absolute deviation (multiplied by 1.4826) and the interquartile range (divided by 1.349), each weighted by two-party votes (a weighted quantile is the first value whose cumulative weight reaches the quantile) and county-equal (the sample median, and linear interpolation between order statistics for the quartiles); the vote-weighted standard deviation recomputed for every pair with one state or the District of Columbia left out at a time. Uniform-swing test: a county is called wrong when the sign of its earlier margin plus the national swing differs from the sign of its later margin. Persistence: Pearson correlation of the county margin with the margin one and four elections earlier, weighted by two-party votes in the later election. Regions: the Census South (AL, AR, DE, DC, FL, GA, KY, LA, MD, MS, NC, OK, SC, TN, TX, VA, WV) against all other states. Weights: none from a survey, because the returns are full counts. Interval method: none; a full count has no sampling error, so no confidence interval is reported, and differences between measures come from definitions and the county set, not from chance. Suppression: none for the county series; a regional figure needs at least 30 matched counties in the region, and every pair meets that.

References

Amlani, Sharif; Algara, Carlos (2021). Partisanship & Nationalization in American Elections: Evidence from Presidential, Senatorial, & Gubernatorial Elections in the U.S. Counties, 1872-2020. Electoral Studies 73: 102387. doi:10.1016/j.electstud.2021.102387
Stokes, Donald E. (1967). Parties and the Nationalization of Electoral Forces. In William Nisbet Chambers and Walter Dean Burnham (eds.), The American Party Systems: Stages of Political Development, pp. 182-202. New York: Oxford University Press.
Bartels, Larry M. (1998). Electoral Continuity and Change, 1868-1996. Electoral Studies 17(3): 301-326. doi:10.1016/S0261-3794(98)00035-3
Claggett, William; Flanigan, William; Zingale, Nancy (1984). Nationalization of the American Electorate. American Political Science Review 78(1): 77-91. doi:10.2307/1961250
Caramani, Daniele (2004). The Nationalization of Politics: The Formation of National Electorates and Party Systems in Western Europe. Cambridge University Press.
Hopkins, Daniel J. (2018). The Increasingly United States: How and Why American Political Behavior Nationalized. University of Chicago Press.

How to cite

@misc{singh2026nationalizationofthecountyvote,
  author = {Singh, Vishal},
  title = {Nationalization of the county vote},
  year = {2026},
  url = {https://vishalsingh.org/politics/nationalization-of-the-county-vote},
  note = {In The Public Mind. Published 30 September 2026.}
}

The spread of county swings narrowed after 1984

Relative swings narrowed to a ridge near zero

Counties near the national swing, 1868 to 2024

Share of votes near the national swing

County swings in the South were wider until 1992

County margins and winners track the last election more closely

The 2020 to 2024 low holds under other measures

What these data cannot show

Data sources

Methods

References

How to cite

Reading paths