Chapter 3.2

The two coalitions

Voters with a four-year degree rose from 19.2% to 43.4% of Democratic voters in the ANES between the presidential elections of 1976 to 1988 and those of 2012 to 2024.

Author: Vishal Singh
Affiliation: NYU Stern
Published: 3 October 2026

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

The ANES and the CES disagree on the degree-holding share of Republican voters after 2008; the chapter reports both and has not resolved the difference.

In the 2024 Cooperative Election Study (CES), 44% of white voters who chose between the two major-party candidates for president voted for the Democrat, so white voters as a group leaned Republican. In the same survey, white voters made up 64% of the Democratic candidate's voters. Both numbers hold because white voters were 72% of all two-party voters that year, and a minority of so large a group still outnumbered the Democratic voters of every other group combined. The two shares divide by different things. The first divides by white voters and says how a group voted; the second divides by Democratic voters and says who a party's voters are. Here a voter is a respondent who reported a vote for one of the two major-party candidates.

Two shares that answer different questions

The same two shares can be followed over time. In the American National Election Studies (ANES), voters with a four-year degree were 19.2% of Democratic voters in the presidential elections of 1976 to 1988 and 43.4% in those of 2012 to 2024. Call that the group's coalition share, with the party's voters as the denominator. Over the same two periods the Democratic share among degree holders rose from 39.2% to 62.0%. Call that the group's vote share, with the group's voters as the denominator. The coalition share says who a party's voters are, and the vote share says how a group voted. A change in one does not identify a change in the other.

Arithmetic links the two. A group's coalition share is its share of all voters multiplied by its relative lean, the ratio of its vote share to the Democratic share of all voters.Axelrod's account of the party coalitions from 1952 to 1968 used the same arithmetic, breaking a group's contribution to a party's vote into its size, its turnout and its loyalty [1]. A group can become a larger part of a coalition because it grew, because it leaned further toward the party, or both. Its coalition share can also fall while its vote share rises. Figure 1 shows both shares in 2024 for every group in the CES.

Figure 1

In 2024 white voters leaned Republican and still made up most of the Democratic coalition

Two shares for each group of two-party presidential voters in 2024: the Democratic share of the group's voters and the group's share of all Democratic voters, percent; 95% intervals from the Kish effective sample size

Democratic share of the group's voters
Group's share of Democratic voters

Education

Race and ethnicity

Religion

Age

Gender

Percent of two-party voters

Show data

Data for In 2024 white voters leaned Republican and still made up most of the Democratic coalition
Dimension	Group	Share	Estimate (%)	95% CI low	95% CI high	n voters in the denominator	Group's share of all voters (%)	Group's share of Republican voters (%)
Education	Postgraduate degree	Democratic share of the group's voters	61.9	60.3	63.4	6,565	16.0	12.0
Education	Postgraduate degree	Group's share of Democratic voters	20.3	19.4	21.1	23,517	16.0	12.0
Education	College graduate	Democratic share of the group's voters	52.6	51.2	53.9	10,429	25.8	24.0
Education	College graduate	Group's share of Democratic voters	27.7	26.7	28.6	23,517	25.8	24.0
Education	Some college	Democratic share of the group's voters	47.2	45.8	48.5	14,759	31.1	32.2
Education	Some college	Group's share of Democratic voters	30.0	29.0	30.9	23,517	31.1	32.2
Education	High school or less	Democratic share of the group's voters	40.0	38.3	41.6	10,255	27.1	31.8
Education	High school or less	Group's share of Democratic voters	22.1	21.2	22.9	23,517	27.1	31.8
Race and ethnicity	White	Democratic share of the group's voters	43.5	42.7	44.4	31,857	72.3	80.0
Race and ethnicity	White	Group's share of Democratic voters	64.3	63.3	65.3	23,517	72.3	80.0
Race and ethnicity	Black	Democratic share of the group's voters	81.9	79.8	84.0	3,753	9.3	3.3
Race and ethnicity	Black	Group's share of Democratic voters	15.6	14.8	16.3	23,517	9.3	3.3
Race and ethnicity	Hispanic	Democratic share of the group's voters	54.7	51.9	57.6	3,564	10.4	9.2
Race and ethnicity	Hispanic	Group's share of Democratic voters	11.6	11.0	12.3	23,517	10.4	9.2
Race and ethnicity	Asian	Democratic share of the group's voters	63.0	58.6	67.4	1,065	3.7	2.6
Race and ethnicity	Asian	Group's share of Democratic voters	4.7	4.3	5.1	23,517	3.7	2.6
Race and ethnicity	Other	Democratic share of the group's voters	43.0	39.1	47.0	1,769	4.3	4.9
Race and ethnicity	Other	Group's share of Democratic voters	3.8	3.4	4.2	23,517	4.3	4.9
Religion	Protestant	Democratic share of the group's voters	32.9	31.8	34.1	14,201	36.9	48.5
Religion	Protestant	Group's share of Democratic voters	24.8	23.9	25.7	23,506	36.9	48.5
Religion	Catholic	Democratic share of the group's voters	43.3	41.6	45.0	7,973	18.5	20.6
Religion	Catholic	Group's share of Democratic voters	16.4	15.6	17.1	23,506	18.5	20.6
Religion	Jewish	Democratic share of the group's voters	68.0	64.0	72.1	1,219	2.7	1.7
Religion	Jewish	Group's share of Democratic voters	3.7	3.3	4.1	23,506	2.7	1.7
Religion	Other religion	Democratic share of the group's voters	45.7	43.1	48.2	4,287	10.9	11.5
Religion	Other religion	Group's share of Democratic voters	10.1	9.5	10.7	23,506	10.9	11.5
Religion	No religion	Democratic share of the group's voters	70.9	69.7	72.2	14,311	31.0	17.7
Religion	No religion	Group's share of Democratic voters	45.0	43.9	46.0	23,506	31.0	17.7
Age	18 to 29	Democratic share of the group's voters	60.6	57.6	63.6	3,082	11.7	9.1
Age	18 to 29	Group's share of Democratic voters	14.5	13.8	15.3	23,517	11.7	9.1
Age	30 to 44	Democratic share of the group's voters	55.1	53.5	56.8	7,073	23.1	20.3
Age	30 to 44	Group's share of Democratic voters	26.1	25.1	27.0	23,517	23.1	20.3
Age	45 to 64	Democratic share of the group's voters	43.9	42.8	45.1	16,848	35.3	38.8
Age	45 to 64	Group's share of Democratic voters	31.7	30.7	32.7	23,517	35.3	38.8
Age	65 and over	Democratic share of the group's voters	45.5	44.4	46.7	15,005	29.8	31.8
Age	65 and over	Group's share of Democratic voters	27.7	26.8	28.6	23,517	29.8	31.8
Gender	Women	Democratic share of the group's voters	50.8	49.8	51.8	21,759	50.7	48.7
Gender	Women	Group's share of Democratic voters	52.9	51.8	53.9	23,300	50.7	48.7
Gender	Men	Democratic share of the group's voters	46.6	45.5	47.8	19,998	49.3	51.3
Gender	Men	Group's share of Democratic voters	47.1	46.1	48.2	23,300	49.3	51.3

In the race and ethnicity panel, the vote-share dot for white voters sits to the left of their coalition-share dot, and the two dots for Black voters sit far apart in the other order. 82% of Black voters voted Democratic, the highest vote share of any racial or ethnic group, and Black voters were 16% of Democratic voters. Every row has two dots with their 95% intervals, the Democratic share of the group's voters and the group's share of Democratic voters, among two-party presidential voters in the 2024 CES. Values, with each group's share of all voters and of Republican voters, in Table 1.

Note: The CES is an opt-in online panel, so its intervals are model based and do not cover how respondents were recruited.Source: Cooperative Election Study cumulative file, 2024; weighted (weight_cumulative); two-party presidential voters

Three surveys and two pooled periods

The ANES interviews different people in each election. Its cumulative file has a presidential vote and demographics for every election from 1952 to 2024, with 34,423 respondents who reported a vote for the Democratic or the Republican candidate. The CES is an online opt-in panel that begins in 2008 and records finer racial and religious categories. The Current Population Survey (CPS) voting supplement asks about turnout and records no vote choice, so the chapter uses it only for the make-up of eligible adults and of voters. Every estimate is weighted. For comparisons across time the chapter pools four elections at each end, 1976 to 1988 and 2012 to 2024, so that small groups have enough respondents, and it gives each election the same weight. Vote and turnout are self-reports. The qualification that matters is survey lean. Set against the certified count, the ANES leans Democratic in its later elections, so changes in vote shares between the two periods are larger in the ANES than in the count. The surveys also disagree about who the voters are. From 2008 to 2024 the share of Republican voters holding a four-year degree changed by −8.5 points in the ANES and by +7.8 points in the CES. The ANES fall is absent from its web samples, which like the CES are collected online. The report gives every estimate with its interval and sample size.

White voters' two shares moved in opposite directions

Between the two pooled periods the Democratic vote share among white non-Hispanic voters rose, from 39.4% to 44.4%. Their coalition share fell, from 72.7% of Democratic voters to 58.0%. Both changes are in the ANES, and both follow from the arithmetic. White non-Hispanic voters became a smaller share of all voters, which lowers their coalition share. Their vote share rose by less than the Democratic share of all voters did, so their relative lean fell as well. Their share of Republican voters fell too, so both coalitions became less white, and the Democratic coalition changed by the larger amount. A reader who hears that white voters moved toward the Democratic candidate and that the Democratic coalition became less white has heard two true statements about one group.

Figure 2 shows the make-up of each coalition in the two periods along six dimensions. In the race view the white segment shrinks in both pairs of bars, and the segment that grows, in both coalitions, is Hispanic voters and voters of other races. The education view carries the chapter's lead finding, the four-year degree segment growing in both coalitions and by more in the Democratic one. The income view shows the highest income third growing as a share of Democratic voters, and in the religion view Protestants fall in both coalitions.

Figure 2

White non-Hispanic voters fell from 72.7% to 58.0% of Democratic voters and from 94.8% to 85.9% of Republican voters

Make-up of each party's two-party presidential voters, percent, in 1976 to 1988 and 2012 to 2024 (the buttons switch the dimension); pooled elections; 95% intervals from the Kish effective sample size

White non-Hispanic
Black non-Hispanic
Hispanic or another race

Show data

Data for Four-year degree holders rose from 19.2% to 43.4% of Democratic voters and from 25.2% to 31.4% of Republican voters
Dimension	Party and period	Group	Share of the party's voters (%)	95% CI low	95% CI high	n party voters	n in group
Education	Democratic voters, 1976 to 1988	No four-year degree	80.8	79.1	82.5	2,173	1,754
Education	Democratic voters, 1976 to 1988	Four-year degree or more	19.2	17.5	20.9	2,173	419
Education	Democratic voters, 2012 to 2024	No four-year degree	56.6	55.2	58.0	9,449	4,639
Education	Democratic voters, 2012 to 2024	Four-year degree or more	43.4	42.0	44.8	9,449	4,810
Education	Republican voters, 1976 to 1988	No four-year degree	74.8	73.1	76.5	2,563	1,919
Education	Republican voters, 1976 to 1988	Four-year degree or more	25.2	23.5	26.9	2,563	644
Education	Republican voters, 2012 to 2024	No four-year degree	68.6	67.2	70.1	7,156	4,340
Education	Republican voters, 2012 to 2024	Four-year degree or more	31.4	29.9	32.8	7,156	2,816
Income	Democratic voters, 1976 to 1988	Lowest third	33.5	31.4	35.6	2,011	664
Income	Democratic voters, 1976 to 1988	Middle third	35.7	33.5	37.8	2,011	716
Income	Democratic voters, 1976 to 1988	Highest third	30.9	28.8	32.9	2,011	631
Income	Democratic voters, 2012 to 2024	Lowest third	24.9	23.6	26.2	9,236	2,855
Income	Democratic voters, 2012 to 2024	Middle third	32.5	31.2	33.9	9,236	3,111
Income	Democratic voters, 2012 to 2024	Highest third	42.6	41.1	44.0	9,236	3,270
Income	Republican voters, 1976 to 1988	Lowest third	20.3	18.6	21.9	2,376	474
Income	Republican voters, 1976 to 1988	Middle third	34.6	32.7	36.6	2,376	813
Income	Republican voters, 1976 to 1988	Highest third	45.1	43.0	47.1	2,376	1,089
Income	Republican voters, 2012 to 2024	Lowest third	21.1	19.8	22.4	6,908	1,911
Income	Republican voters, 2012 to 2024	Middle third	35.6	34.1	37.2	6,908	2,506
Income	Republican voters, 2012 to 2024	Highest third	43.3	41.7	44.8	6,908	2,491
Race and ethnicity	Democratic voters, 1976 to 1988	White non-Hispanic	72.7	70.8	74.6	2,179	1,594
Race and ethnicity	Democratic voters, 1976 to 1988	Black non-Hispanic	20.0	18.3	21.7	2,179	425
Race and ethnicity	Democratic voters, 1976 to 1988	Hispanic or another race	7.3	6.2	8.4	2,179	160
Race and ethnicity	Democratic voters, 2012 to 2024	White non-Hispanic	58.0	56.6	59.4	9,493	5,863
Race and ethnicity	Democratic voters, 2012 to 2024	Black non-Hispanic	19.9	18.7	21.0	9,493	1,778
Race and ethnicity	Democratic voters, 2012 to 2024	Hispanic or another race	22.1	20.9	23.3	9,493	1,852
Race and ethnicity	Republican voters, 1976 to 1988	White non-Hispanic	94.8	93.9	95.7	2,570	2,436
Race and ethnicity	Republican voters, 1976 to 1988	Black non-Hispanic	1.3	0.9	1.8	2,570	33
Race and ethnicity	Republican voters, 1976 to 1988	Hispanic or another race	3.9	3.1	4.6	2,570	101
Race and ethnicity	Republican voters, 2012 to 2024	White non-Hispanic	85.9	84.8	87.0	7,190	6,234
Race and ethnicity	Republican voters, 2012 to 2024	Black non-Hispanic	1.6	1.2	2.0	7,190	82
Race and ethnicity	Republican voters, 2012 to 2024	Hispanic or another race	12.5	11.5	13.5	7,190	874
Religion	Democratic voters, 1976 to 1988	Protestant	59.5	57.4	61.6	2,186	1,293
Religion	Democratic voters, 1976 to 1988	Catholic	28.0	26.0	29.9	2,186	619
Religion	Democratic voters, 1976 to 1988	Jewish, another religion or none	12.6	11.1	14.0	2,186	274
Religion	Democratic voters, 2012 to 2020	Protestant	36.0	34.5	37.6	7,349	2,676
Religion	Democratic voters, 2012 to 2020	Catholic	21.8	20.4	23.1	7,349	1,571
Religion	Democratic voters, 2012 to 2020	Jewish, another religion or none	42.2	40.6	43.8	7,349	3,102
Religion	Republican voters, 1976 to 1988	Protestant	68.8	67.0	70.7	2,569	1,773
Religion	Republican voters, 1976 to 1988	Catholic	23.1	21.5	24.8	2,569	595
Religion	Republican voters, 1976 to 1988	Jewish, another religion or none	8.0	7.0	9.1	2,569	201
Religion	Republican voters, 2012 to 2020	Protestant	52.3	50.5	54.0	5,497	2,881
Religion	Republican voters, 2012 to 2020	Catholic	22.5	21.0	24.0	5,497	1,258
Religion	Republican voters, 2012 to 2020	Jewish, another religion or none	25.2	23.7	26.7	5,497	1,358
Age	Democratic voters, 1976 to 1988	18 to 29	19.2	17.6	20.9	2,182	409
Age	Democratic voters, 1976 to 1988	30 to 44	30.6	28.6	32.6	2,182	675
Age	Democratic voters, 1976 to 1988	45 to 64	31.2	29.2	33.1	2,182	687
Age	Democratic voters, 1976 to 1988	65 and over	19.0	17.3	20.7	2,182	411
Age	Democratic voters, 2012 to 2024	18 to 29	18.5	17.4	19.6	9,304	1,125
Age	Democratic voters, 2012 to 2024	30 to 44	24.6	23.4	25.9	9,304	2,255
Age	Democratic voters, 2012 to 2024	45 to 64	34.6	33.2	35.9	9,304	3,301
Age	Democratic voters, 2012 to 2024	65 and over	22.3	21.1	23.5	9,304	2,623
Age	Republican voters, 1976 to 1988	18 to 29	18.9	17.4	20.5	2,564	472
Age	Republican voters, 1976 to 1988	30 to 44	33.0	31.2	34.9	2,564	851
Age	Republican voters, 1976 to 1988	45 to 64	29.6	27.8	31.4	2,564	765
Age	Republican voters, 1976 to 1988	65 and over	18.5	16.9	20.0	2,564	476
Age	Republican voters, 2012 to 2024	18 to 29	13.1	12.0	14.2	6,976	548
Age	Republican voters, 2012 to 2024	30 to 44	21.5	20.2	22.8	6,976	1,431
Age	Republican voters, 2012 to 2024	45 to 64	40.0	38.5	41.6	6,976	2,685
Age	Republican voters, 2012 to 2024	65 and over	25.4	24.0	26.8	6,976	2,312
Gender	Democratic voters, 1976 to 1988	Women	58.6	56.5	60.7	2,191	1,286
Gender	Democratic voters, 1976 to 1988	Men	41.4	39.3	43.5	2,191	905
Gender	Democratic voters, 2012 to 2024	Women	55.5	54.0	56.9	9,495	5,391
Gender	Democratic voters, 2012 to 2024	Men	44.5	43.1	46.0	9,495	4,104
Gender	Republican voters, 1976 to 1988	Women	53.2	51.2	55.2	2,579	1,373
Gender	Republican voters, 1976 to 1988	Men	46.8	44.8	48.8	2,579	1,206
Gender	Republican voters, 2012 to 2024	Women	49.6	48.0	51.1	7,205	3,564
Gender	Republican voters, 2012 to 2024	Men	50.4	48.9	52.0	7,205	3,641

The white non-Hispanic segment is shorter in the later bar of each pair, and by more in the Democratic pair; the segment for Hispanic voters and voters of other races grows in both. Each bar is one party's two-party voters in one pooled period, split into groups, so a segment is a coalition share and the segments in a bar sum to the whole. The buttons switch to education, income, religion, age and gender; the religion view ends in 2020 because the cumulative file's 2024 religion codes are unusable. Values with their intervals in Table 5.

Note: Each period pools four elections, 1976, 1980, 1984 and 1988 against 2012, 2016, 2020 and 2024, and every election has the same total weight. The religion view ends in 2020 because the 2024 religion codes of the cumulative file are unusable.Note: Intervals are model based (Kish effective sample size, normal approximation) and ignore clustering in the sample design.Note: Stacked bars carry no whiskers; the interval and the number of voters in the party's denominator are in the tooltip and the table.Source: ANES Cumulative Data File, 1976 to 1988 and 2012 to 2024; weighted (weight_postelection where the file has one, else weight; each election has the same total weight); two-party presidential voters

A rising vote share is easy to read as a group changing its mind, and repeated cross-sections cannot show that. The Democratic share of all ANES voters rose between the two periods, and Figure 3 splits that rise using cells defined by sex, three racial and ethnic groups and four age bands. The part that comes from a different mix of voters across those cells is +2.9 points, and the part that comes from a different Democratic share inside the cells is +5.3 points. The split is an accounting identity under those cells, so the two parts sum to the change and say nothing about why either moved. The within-cell part holds voters who changed their vote, new voters who replaced those who died, and different people voting in the same cell from one election to the next. It also holds changes in how the surveys sampled each cell and the surveys' own lean. Among degree holders the make-up of graduates by sex, race and age accounts for little of the rise in the vote share, and that does not make the rise persuasion either. Chapter 2.1, Cohort change in partisan sorting separates switching from replacement for ideology by following birth cohorts, which these vote questions are too thin to support.

Figure 3

The mix of voters accounts for +2.9 points of the change in the Democratic share and change within groups of the same sex, race and age for +5.3 points

Change in the Democratic share of two-party presidential voters from 1976 to 1988 to 2012 to 2024, split into the mix of voters across sex, race and age cells and change within cells, percent and percentage points; ANES; the buttons switch the voters

Show data

Data for The mix of voters accounts for +2.9 points of the change in the Democratic share and change within groups of the same sex, race and age for +5.3 points
Voters	Step	Value	95% CI low	95% CI high	n
All two-party voters	Democratic share, 1976 to 1988	45.9	44.5	47.4	4,770
All two-party voters	Change in the mix of voters	2.9	2.1	3.8	21,554
All two-party voters	Change within groups of the same sex, race and age	5.3	3.6	6.8	21,554
All two-party voters	Democratic share, 2012 to 2024	54.1	53.1	55.2	16,784
Voters with a four-year degree	Democratic share, 1976 to 1988	39.2	36.2	42.2	1,063
Voters with a four-year degree	Change in the mix of voters	1.1	−0.1	2.9	8,689
Voters with a four-year degree	Change within groups of the same sex, race and age	21.7	18.0	24.8	8,689
Voters with a four-year degree	Democratic share, 2012 to 2024	62.0	60.6	63.5	7,626
Voters without a four-year degree	Democratic share, 1976 to 1988	47.8	46.1	49.4	3,673
Voters without a four-year degree	Change in the mix of voters	4.7	3.5	5.9	12,652
Voters without a four-year degree	Change within groups of the same sex, race and age	−3.2	−5.4	−1.5	12,652
Voters without a four-year degree	Democratic share, 2012 to 2024	49.3	47.9	50.8	8,979

The first bar is the Democratic share of all two-party voters in 1976 to 1988 and the last bar the share in 2012 to 2024. Of the two steps between them, the change in the mix of voters is the smaller and the change within cells of the same sex, race and age the larger. The steps carry bootstrap intervals and the bars intervals from the Kish effective sample size. The buttons restrict the voters by degree. Among degree holders the mix of voters accounts for little of the change, and among voters without a degree a rise from the mix and a fall within cells leave the share barely moved. Values in Table 8.

Note: The first and last bars carry Kish intervals and the two steps carry bootstrap intervals. The cells combine sex, race and ethnicity (white non-Hispanic, Black non-Hispanic, Hispanic or another race) and four age bands; a respondent with a missing value forms a cell of its own. A cell with no respondents in one period takes the other period's Democratic share, so it adds nothing to the within-cell term.Note: Intervals come from a bootstrap that resamples respondents within each election (400 replicates, seed 12345). The split is symmetric between the mix and the within-cell change.Source: ANES Cumulative Data File, 1976 to 1988 and 2012 to 2024; weighted (weight_postelection where the file has one, else weight; each election has the same total weight); two-party presidential voters

What these surveys cannot tell you

The surveys interview different people in each election, so the chapter compares groups and never individuals. It cannot say whether any voter changed party, how much of a change comes from older voters leaving the electorate and younger ones entering it, or how much comes from different people turning out. Vote and turnout are self-reports. The ANES leans Democratic in its later elections, the CES is an opt-in panel and the CPS records no vote, so no survey here measures a party's voters exactly. A check against voter records, the usual test of self-reported vote and turnout, is not available for either vote file [2]. The ANES and the CES also disagree about whether the degree-holding share of Republican voters rose or fell after 2008, and the chapter reports both. The two periods, 1976 to 1988 and 2012 to 2024, were chosen to give each group enough respondents, and a different choice of years would give different numbers.

Go deeper

The report: every estimate with its interval and sample size.
Tables: every figure of the report as a table.

Data sources

ANES Cumulative Data FileCumulative Data File, presidential years 1952 to 2024Years: 1948-2024 (33 study years; even years only after 2004; no 2006, 2010 or 2014 study) · Weights: weight_postelection where available, else VCF0009z · License: Free download after accepting the ANES terms of use; cite the release.
ANES 2024 Time Series2024 Time Series, summary religion code, one comparison with the CESYears: 2024 (pre-election and post-election interviews) · Weights: V240107b · License: Free download after accepting the ANES terms of use; cite the release.
Cooperative Election StudyCumulative Common Content, presidential years 2008 to 2024Years: 2006-2024 (every year; self-reported turnout is available in the 10 even years) · Weights: weight_cumulative · License: See the provider's terms of use.
CPS Voting and Registration SupplementNovember Voting and Registration Supplement, presidential years 1996 to 2024 (and 1976 to 1988 for one comparison), adults aged 18 or olderYears: 1976-2024 (25 November samples, biennial, even years) · Weights: VOSUPPWT · License: IPUMS CPS conditions of use: a free extract after registration, no fee may be charged for the data, IPUMS must be cited, and publications are to be added to the IPUMS bibliography. The underlying Census Bureau data are public domain.
State presidential returnsState presidential returns 1976 to 2024, certified national two-party shareYears: 1976-2024 (13 elections) · Weights: none (full counts) · License: See the provider's terms of use.

All data sources

References

Axelrod, Robert (1972). Where the Votes Come From: An Analysis of Electoral Coalitions, 1952-1968. American Political Science Review 66(1): 11-20. doi:10.2307/1959275
Ansolabehere, Stephen; Hersh, Eitan (2012). Validation: What Big Data Reveal About Survey Misreporting and the Real Electorate. Political Analysis 20(4): 437-459. doi:10.1093/pan/mps023

How to cite

@misc{singh2026thetwocoalitions,
  author = {Singh, Vishal},
  title = {The two coalitions},
  year = {2026},
  url = {https://vishalsingh.org/politics/the-two-coalitions},
  note = {In The Public Mind. Published 3 October 2026.}
}

Two shares that answer different questions

Three surveys and two pooled periods

White voters' two shares moved in opposite directions

A change in a vote share does not identify persuasion

What these surveys cannot tell you

Go deeper

Data sources

References

How to cite