Chapter 15.1

Party and place

Across 12 policy issues, Democrats and Republicans in the same kind of deep county differ by 43.1 points in the share taking the Democratic side, 7.6 times the 5.6-point difference between deep-blue and deep-red counties within a party.

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

Still open: the place gap is measured for the county only, and a version at the ZIP code level has not been built. Support by county margin is drawn for a subset of the issues, without an issue selector.

The party gap and the place gap are two differences in the share of respondents who take the Democratic side of a policy question, measured in points. The party gap compares Democrats with Republicans who live in the same kind of county. The place gap compares members of one party who live in deep-blue counties with those who live in deep-red counties, where a deep county has a two-party presidential margin of 20 points or more toward one party. Averaged over 12 issues, the party gap is 43.1 (95% CI 41.8 to 44.5) points and the place gap is 5.6 (95% CI 4.7 to 6.6) points, a ratio of 7.6 (95% CI 6.4 to 9.3), and the party gap is larger than the place gap on every issue. Before adjustment, the interval for the place gap lies above zero on all 12 issues, so the place gap is small and not zero. The section on Republican and Democratic place gaps reports how many of them shrink when respondents are compared within the same state.

The data are interviews from the Cooperative Election Study (CES) in 2013 to 2021, 333,335 of which could be matched to a county, and 320,988 of those belong to a party group used in this chapter. Each is matched to the two-party margin of the respondent's county in the most recent presidential election, counting the election of the interview year, and every estimate uses survey weights. The main comparison of party and place uses the 2013 to 2021 interviews. The trend across presidential cycles in Figure 8 uses the longer source window of 2006 to 2021, so it includes surveys that the main comparison leaves out.Unless stated otherwise, every interval in this chapter is a 95% interval from a bootstrap that resamples whole states (50 clusters, 2,000 replicates). The exceptions and the method are in the methods note at the end of the chapter. The Chapter 1.4, Cohort change in partisan sorting chapter follows the growth of party differences within birth cohorts, and this chapter asks how much of what remains depends on where a respondent lives.

Respondents are Democrats or Republicans by the three-point party question, so leaners belong to neither group. Each issue is a yes-or-no question, coded so that 100 is the side that Democrats take more often: support for an assault weapons ban, for example, and opposition to repealing the Affordable Care Act (ACA). The party gap on an issue is the Democratic minus Republican difference in the share taking that side, averaged over deep-red and deep-blue counties, and the place gap is the deep-blue minus deep-red difference in that share, averaged over Democrats and Republicans.

The party gap exceeds the place gap on every issue

Figure 1 shows both gaps for each issue. The party gap runs from 13.3 (95% CI 12.0 to 14.9) points on universal background checks to 64.1 (95% CI 62.5 to 65.8) points on spending for a border wall. The place gap runs from 2.1 (95% CI 0.9 to 3.1) points on background checks to 12.0 (95% CI 9.2 to 14.6) points on abortion. On no issue do the intervals for the two gaps touch.

Figure 1

On every issue the gap between parties is larger than the gap between deep-red and deep-blue counties

Gap in the share taking the Democratic side, in points, by issue; Democrats and Republicans in deep counties; CES survey years 2013 to 2021

Place gap
Party gap

Show data

Data for On every issue the gap between parties is larger than the gap between deep-red and deep-blue counties
Issue	Democratic side of the question	Survey years	Party gap	Party gap, low	Party gap, high	Place gap	Place gap, low	Place gap, high	Ratio	Smallest group, n
Assault weapons ban	Support an assault weapons ban	2013 to 2021	41.9	40.3	43.8	7.5	6.1	9.0	5.6	16,440
Background checks	Support universal background checks for gun buyers	2013 to 2021, not 2020	13.3	12.0	14.9	2.1	0.9	3.1	6.5	13,375
Concealed carry	Oppose easier concealed carry	2013 to 2021	38.8	37.1	40.7	4.8	3.4	6.4	8.1	16,452
Border security	Oppose more spending on border security	2013 to 2021, not 2018	40.2	38.6	42.0	5.3	3.7	6.7	7.6	13,428
Path to legal status	Support a path to legal status for undocumented immigrants	2013 to 2021, not 2018	38.7	36.9	40.3	4.1	2.6	5.8	9.6	13,422
Border wall	Oppose spending on a border wall	2017 to 2021, not 2019	64.1	62.5	65.8	4.1	2.9	5.3	15.5	8,257
Gay marriage	Support legal gay marriage	2013 to 2016	36.4	33.6	39.3	10.2	6.6	13.7	3.6	7,161
Repeal the ACA	Oppose repealing the Affordable Care Act	2013 to 2021	57.2	55.8	58.8	4.2	2.8	5.4	13.7	16,459
EPA regulates CO2	Support letting the EPA regulate carbon dioxide	2014 to 2021	44.4	42.3	46.5	5.2	3.5	6.9	8.6	15,459
Abortion legal	Support always allowing abortion as a matter of choice	2014 to 2021	46.5	44.7	48.7	12.0	9.2	14.6	3.9	15,818
Medicare for All	Support Medicare for All	2018 to 2021	51.3	49.3	53.6	4.2	1.7	6.4	12.1	8,385
Cut immigration	Oppose cutting legal immigration by half	2018 to 2020	44.2	42.6	45.7	4.1	2.5	5.9	10.8	7,157

The party gap, the solid dot, is larger than the place gap, the open dot, on every issue: the averages are 43.1 and 5.6 points. Whiskers are 95% intervals, and the number at the right of each row is the party gap minus the place gap.Note: A deep county is one where the two-party presidential margin is at or beyond the cutoff stated in the text. The share is measured on the side of each issue that Democrats take more often, which the data table names for each question.Note: The issues were asked in different survey years, listed in the data table, so the averages combine periods. The number in a tooltip is the smallest of the four party by county-type groups behind the row.Note: Ratios are computed from unrounded gaps, so a ratio can differ in its last digit from the quotient of the gaps as printed.Note: Intervals are 95% intervals from a bootstrap that resamples whole states; they are model based and ignore the design of the panel.Source: CES cumulative common content, 2013-2021; county returns for the two-party margin; weighted (weight_cumulative)

Repeal of the ACA shows what the two gaps look like in levels.The cumulative CES file codes the 2014 answers to the question on repealing the ACA in the opposite direction from the other years: the Republican minus Democratic share supporting repeal is negative in 2014 and positive in each of the other nine survey years. The 2014 answers are reversed here on that evidence and on the cumulative file's release notes; the 2014 value labels have not been checked against the original questionnaire. The sections on ratios and trends report what changes if they are dropped instead. In deep-red counties 83.1% of Republicans and 25.6% of Democrats support repeal, and in deep-blue counties 78.6% and 21.7%. The party gap on this issue is 57.2 points and the place gap is 4.2 points, a ratio of 13.7. On an assault weapons ban the party gap is 41.9 points and the place gap 7.5 points, a ratio of 5.6. On spending for a border wall the party gap is 64.1 points and the place gap 4.1 points, a ratio of 15.5.

The issues were asked in different years, listed in the data table under the figure. Border wall spending was asked only from 2017, and legal gay marriage only through 2016, so the averages combine periods and describe a set of questions rather than a single year.

The ratio varies widely by issue

Figure 2 divides the party gap by the place gap on each issue. The ratio is 3.6 (95% CI 2.6 to 5.6) on legal gay marriage and 15.5 (95% CI 12.0 to 22.0) on spending for a border wall. The issues with the two smallest ratios, legal gay marriage and abortion, are the two with the largest place gaps, 10.2 and 12.0 points, against party gaps of 36.4 and 46.5 points. The ratios for repeal of the ACA and Medicare for All are the next largest after the border wall's, and the intervals for the three largest ratios overlap, so their order is not settled. If the 2014 answers to the ACA question are dropped instead of reversed, repeal of the ACA has the largest ratio, 16.7 (95% CI 12.0 to 28.8). The ratio for Medicare for All, 12.1 (95% CI 7.9 to 31.4), has the widest interval, because its place gap of 4.2 points is small and a small denominator makes a ratio move with the sample. The median ratio across issues is 8.4 (95% CI 6.7 to 11.0) and the ratio of the average gaps is 7.6 (95% CI 6.4 to 9.3); each interval lies above equal gaps.

Figure 2

The party gap is between 3.6 and 15.5 times the place gap, depending on the issue

Party gap divided by place gap, by issue, with 95% intervals; Democrats and Republicans in deep counties; CES survey years 2013 to 2021

Show data

Data for The party gap is between 3.6 and 15.5 times the place gap, depending on the issue
issue	n	ratio	ratio_hi	ratio_lo
Border wall	8,257	15.48	21.99	12.04
Repeal the ACA	16,459	13.75	20.22	10.67
Medicare for All	8,385	12.13	31.41	7.86
Abortion legal	15,818	3.89	5.16	3.12
EPA regulates CO2	15,459	8.60	13.07	6.22
Cut immigration	7,157	10.76	17.65	7.43
Assault weapons ban	16,440	5.62	7.09	4.55
Border security	13,428	7.61	10.85	6.02
Concealed carry	16,452	8.15	11.59	6.08
Path to legal status	13,422	9.55	15.43	6.43
Gay marriage	7,161	3.56	5.62	2.63
Background checks	13,375	6.47	14.12	4.23

The party gap is larger than the place gap on every issue, by a ratio from 3.6 on legal gay marriage to 15.5 on the border wall (point estimates). The median across issues is 8.4 and the ratio of the average gaps is 7.6. Each dot is the party gap on one issue divided by its place gap, and the dashed line marks equal gaps.Note: The number in a tooltip is the smallest of the four party by county-type groups behind the ratio. The interval is skewed to the right where the place gap is small. The intervals for the three largest ratios overlap, and with the 2014 answers to the ACA question dropped instead of reversed, repeal of the ACA has the largest ratio.Note: Ratios are computed from unrounded gaps, so a ratio can differ in its last digit from the quotient of the gaps as printed.Note: Intervals are 95% intervals from a bootstrap that resamples whole states; they are model based and ignore the design of the panel.Source: CES cumulative common content, 2013-2021; county returns for the two-party margin; weighted (weight_cumulative)

Sensitivity to the county cutoff and to leaners

Two choices in the definition of the gaps move their levels. A deep county is one at or beyond a cutoff on the county margin, and a stricter cutoff compares more extreme places. Figure 3 reports the averages at cutoffs of 10 and 40 points as well as the 20 used elsewhere, and with leaners counted as partisans. The place gap is larger when counties are more extreme, the ratio falls as the cutoff rises, and the party gap is larger than the place gap under every definition.

Figure 3

The party gap exceeds the place gap under every definition, and the ratio falls as the county cutoff rises

Average party gap and place gap over the issues, in points (share taking the Democratic side), and their ratio, with 95% intervals; CES survey years 2013 to 2021

Definition	Party gap	Party gap, low	Party gap, high	Place gap	Place gap, low	Place gap, high	Ratio	Ratio, low	Ratio, high
Deep county: margin of 10 points or more	43.8	42.7	45.0	4.5	3.8	5.2	9.8	8.4	11.7
Deep county: margin of 20 points or more (used in this chapter)	43.1	41.8	44.5	5.6	4.7	6.6	7.6	6.4	9.3
Deep county: margin of 40 points or more	42.3	40.1	44.5	7.3	5.8	8.9	5.8	4.6	7.5
Leaners counted as partisans, margin of 20 points or more	44.6	43.4	46.0	5.4	4.5	6.3	8.3	7.1	10.1

Show data

Data for The party gap exceeds the place gap under every definition, and the ratio falls as the county cutoff rises
definition	party_gap	party_gap_hi	party_gap_lo	place_gap	place_gap_hi	place_gap_lo	ratio	ratio_hi	ratio_lo
Deep county: margin of 10 points or more	43.76	45.00	42.72	4.49	5.21	3.78	9.76	11.65	8.35
Deep county: margin of 20 points or more (used in this chapter)	43.09	44.48	41.80	5.63	6.62	4.68	7.65	9.32	6.43
Deep county: margin of 40 points or more	42.27	44.49	40.10	7.35	8.88	5.81	5.75	7.48	4.60
Leaners counted as partisans, margin of 20 points or more	44.64	46.01	43.36	5.36	6.27	4.45	8.33	10.08	7.07

The place gap grows as the cutoff for a deep county moves from 10 to 40 points, and the ratio falls. The last row counts leaners as partisans. Each row is an average over the issues with a 95% interval.Note: Ratios are computed from unrounded gaps, so a ratio can differ in its last digit from the quotient of the gaps as printed.Note: Intervals are 95% intervals from a bootstrap that resamples whole states; they are model based and ignore the design of the panel.Source: CES cumulative common content, 2013-2021; county returns for the two-party margin; weighted (weight_cumulative)

Support varies less across county margins than between the parties

The deep-county comparison uses only the two ends of the distribution of counties. Figure 4 uses every county, grouped into seven bins by two-party margin, for the issues in the panels. In every panel the Democratic and Republican lines are at least 33.0 points apart in every bin. In 9 of the 10 lines the change from the most Republican to the most Democratic bin is toward the Democratic side of the issue, with an interval that excludes zero. The exception is Republican support for a path to legal status, which changes by +3.1 (95% CI −1.1 to 8.5) points.

Figure 4

Within each party, support changes less across county margins than it differs between the parties

Share supporting the policy (%) by party and two-party margin of the respondent's county, seven bins of counties; five issues; CES survey years 2013 to 2021

Democrats
Republicans

County two-party margin (points, Democratic minus Republican)

Show data

Data for Within each party, support changes less across county margins than it differs between the parties
bin	conf_high	conf_low	estimate	issue	margin	n	party
Below -40	83.49	78.63	81.16	Assault weapons ban	−53.35	8,553	Democrats
-40 to -20	84.61	81.24	83.00	Assault weapons ban	−29.09	13,631	Democrats
-20 to -5	86.74	82.80	84.81	Assault weapons ban	−12.70	13,578	Democrats
-5 to 5	87.03	82.65	84.91	Assault weapons ban	0.112	13,026	Democrats
5 to 20	87.40	85.78	86.56	Assault weapons ban	12.20	25,300	Democrats
20 to 40	88.29	85.24	86.68	Assault weapons ban	28.48	21,675	Democrats
Above 40	88.59	86.32	87.29	Assault weapons ban	56.26	27,716	Democrats
Below -40	38.65	34.48	36.47	Assault weapons ban	−53.35	14,432	Republicans
-40 to -20	40.55	36.62	38.73	Assault weapons ban	−29.09	15,619	Republicans
-20 to -5	42.39	37.62	40.22	Assault weapons ban	−12.70	11,652	Republicans
-5 to 5	43.05	37.73	40.29	Assault weapons ban	0.112	9,253	Republicans
5 to 20	45.48	40.82	43.35	Assault weapons ban	12.20	14,228	Republicans
20 to 40	48.90	44.40	46.96	Assault weapons ban	28.48	9,629	Republicans
Above 40	53.43	44.84	49.04	Assault weapons ban	56.26	6,811	Republicans
Below -40	75.61	69.08	72.62	Path to legal status	−53.35	6,731	Democrats
-40 to -20	74.12	70.02	72.14	Path to legal status	−29.09	10,838	Democrats
-20 to -5	74.83	70.90	72.89	Path to legal status	−12.70	11,500	Democrats
-5 to 5	75.21	70.21	72.70	Path to legal status	0.112	10,868	Democrats
5 to 20	75.95	73.97	74.97	Path to legal status	12.20	21,339	Democrats
20 to 40	78.10	74.79	76.47	Path to legal status	28.48	18,012	Democrats
Above 40	79.18	76.62	77.95	Path to legal status	56.26	23,266	Democrats
Below -40	36.16	32.90	34.40	Path to legal status	−53.35	11,088	Republicans
-40 to -20	36.32	33.01	34.63	Path to legal status	−29.09	12,294	Republicans
-20 to -5	35.04	32.07	33.68	Path to legal status	−12.70	9,881	Republicans
-5 to 5	37.36	33.83	35.49	Path to legal status	0.112	7,640	Republicans
5 to 20	37.94	32.68	35.41	Path to legal status	12.20	11,721	Republicans
20 to 40	39.71	35.64	37.72	Path to legal status	28.48	7,837	Republicans
Above 40	42.74	34.03	37.53	Path to legal status	56.26	5,585	Republicans
Below -40	29.34	25.13	27.44	Repeal the ACA (support for repeal)	−53.35	8,556	Democrats
-40 to -20	26.22	22.46	24.38	Repeal the ACA (support for repeal)	−29.09	13,619	Democrats
-20 to -5	25.96	22.44	24.14	Repeal the ACA (support for repeal)	−12.70	13,555	Democrats
-5 to 5	24.41	21.30	22.93	Repeal the ACA (support for repeal)	0.112	13,014	Democrats
5 to 20	23.21	19.73	21.69	Repeal the ACA (support for repeal)	12.20	25,277	Democrats
20 to 40	24.37	19.24	21.95	Repeal the ACA (support for repeal)	28.48	21,665	Democrats
Above 40	23.79	18.73	21.58	Repeal the ACA (support for repeal)	56.26	27,714	Democrats
Below -40	83.72	80.90	82.42	Repeal the ACA (support for repeal)	−53.35	14,470	Republicans
-40 to -20	84.83	82.53	83.71	Repeal the ACA (support for repeal)	−29.09	15,630	Republicans
-20 to -5	86.79	84.59	85.73	Repeal the ACA (support for repeal)	−12.70	11,652	Republicans
-5 to 5	84.54	82.01	83.16	Repeal the ACA (support for repeal)	0.112	9,251	Republicans
5 to 20	83.07	80.21	81.47	Repeal the ACA (support for repeal)	12.20	14,220	Republicans
20 to 40	80.89	77.95	79.42	Repeal the ACA (support for repeal)	28.48	9,633	Republicans
Above 40	79.84	76.22	77.40	Repeal the ACA (support for repeal)	56.26	6,826	Republicans
Below -40	73.89	66.74	70.37	Abortion legal	−53.35	8,255	Democrats
-40 to -20	76.25	71.90	74.15	Abortion legal	−29.09	13,015	Democrats
-20 to -5	79.82	76.77	78.39	Abortion legal	−12.70	12,696	Democrats
-5 to 5	80.64	75.54	78.21	Abortion legal	0.112	12,287	Democrats
5 to 20	82.08	78.67	80.43	Abortion legal	12.20	24,028	Democrats
20 to 40	82.96	77.85	80.38	Abortion legal	28.48	20,804	Democrats
Above 40	84.98	81.98	83.59	Abortion legal	56.26	26,619	Democrats
Below -40	23.12	19.97	21.57	Abortion legal	−53.35	14,058	Republicans
-40 to -20	27.82	23.39	25.76	Abortion legal	−29.09	15,033	Republicans
-20 to -5	31.62	26.02	28.88	Abortion legal	−12.70	10,900	Republicans
-5 to 5	34.39	29.48	32.33	Abortion legal	0.112	8,783	Republicans
5 to 20	35.55	29.43	32.95	Abortion legal	12.20	13,516	Republicans
20 to 40	38.48	32.15	35.59	Abortion legal	28.48	9,260	Republicans
Above 40	45.05	36.31	41.62	Abortion legal	56.26	6,558	Republicans
Below -40	68.37	59.64	63.90	Gay marriage	−53.35	3,353	Democrats
-40 to -20	71.41	64.60	67.99	Gay marriage	−29.09	6,079	Democrats
-20 to -5	75.93	68.93	72.89	Gay marriage	−12.70	6,730	Democrats
-5 to 5	76.62	67.97	72.01	Gay marriage	0.112	6,378	Democrats
5 to 20	77.24	70.62	74.08	Gay marriage	12.20	11,861	Democrats
20 to 40	77.28	67.94	72.45	Gay marriage	28.48	9,357	Democrats
Above 40	79.15	70.81	75.20	Gay marriage	56.26	12,598	Democrats
Below -40	27.81	22.79	25.22	Gay marriage	−53.35	5,044	Republicans
-40 to -20	31.00	26.70	28.98	Gay marriage	−29.09	6,668	Republicans
-20 to -5	35.05	28.89	31.87	Gay marriage	−12.70	5,720	Republicans
-5 to 5	39.46	33.00	36.45	Gay marriage	0.112	4,399	Republicans
5 to 20	40.11	34.02	37.51	Gay marriage	12.20	6,431	Republicans
20 to 40	41.78	35.92	38.96	Gay marriage	28.48	4,079	Republicans
Above 40	45.50	35.94	42.18	Gay marriage	56.26	3,082	Republicans

The Democratic and Republican lines are far apart in every bin, and each line changes less across county margins than the distance between the two. Each panel is one issue; the solid blue line is Democrats and the dashed red line is Republicans. Each point is a bin of counties placed at its average margin, and bands are 95% intervals. The repeal panel plots support for repeal.Note: Counties are grouped by the Democratic minus Republican margin of the most recent presidential election; the data table names the bins. The two outer bins cover a wider range of margins than the others.Note: Intervals are 95% intervals from a bootstrap that resamples whole states; they are model based and ignore the design of the panel.Source: CES cumulative common content, 2013-2021; county returns for the two-party margin; weighted (weight_cumulative)

The two largest changes are among Republicans, +20.1 (95% CI 14.8 to 24.0) points on abortion and +17.0 (95% CI 10.3 to 21.5) points on legal gay marriage, which are also the issues with the largest place gaps. On the assault weapons ban, Republican support rises from 36.5% in the most Republican bin to 49.0% in the most Democratic bin, a change of +12.6 (95% CI 8.6 to 16.9) points, and Democratic support goes from 81.2% to 87.3%. Republican respondents in Democratic counties support a ban more often than Republican respondents in Republican counties, and the change is shallow next to the 41.9 points that separate the parties on this issue.

Republican place gaps are larger on most issues, not reliably

Figure 5 splits the place gap by party. Of the 24 place gaps, one for each party on each issue, all 24 have deep-blue counties on the more Democratic side, and 22 have an interval above zero. The Republican gap is larger than the Democratic gap on 9 of the 12 issues, and the two are equal to one decimal on easier concealed carry. The interval for the difference lies above zero on 5 of them, and on 4 of those its lower end is below 1 point. It lies below zero on no issue. Averaged over issues, the Republican place gap is 6.8 (95% CI 5.0 to 8.5) points and the Democratic place gap 4.5 (95% CI 2.8 to 6.3) points. The difference, 2.3 (95% CI −0.7 to 5.1) points, has an interval that includes zero.

Figure 5

The Republican place gap is larger on most issues, but the interval for the average difference includes zero

Place gap in points (deep-blue minus deep-red counties, share taking the Democratic side) for Democrats and Republicans, by issue; CES survey years 2013 to 2021

Democrats
Republicans

Show data

Data for The Republican place gap is larger on most issues, but the interval for the average difference includes zero
Model	Issue	Democrats	Democrats, low	Democrats, high	Republicans	Republicans, low	Republicans, high	Republican minus Democratic	Difference, low	Difference, high	Democrats, n	Republicans, n	Democrats, states	Republicans, states
raw	Border wall	1.9	−0.5	4.3	6.4	3.8	8.4	+4.5	+0.2	+8.4	36,261	24,614	50	50
raw	Repeal the ACA	3.8	1.4	6.3	4.5	2.6	6.0	+0.7	−3.0	+3.9	71,554	46,559	50	50
raw	Medicare for All	2.7	1.2	4.2	5.7	0.8	9.7	+3.0	−1.9	+7.1	36,118	25,023	50	50
raw	Abortion legal	9.5	6.1	12.9	14.4	10.9	17.5	+4.9	+0.5	+8.5	68,693	44,909	50	50
raw	EPA regulates CO2	3.7	1.9	5.5	6.6	3.5	10.0	+2.9	−0.9	+7.1	66,941	43,698	50	50
raw	Cut immigration	3.5	1.3	6.0	4.8	2.8	6.8	+1.3	−1.5	+4.1	30,037	21,417	50	50
raw	Assault weapons ban	4.7	2.9	6.9	10.2	7.9	12.6	+5.4	+2.0	+8.6	71,575	46,491	50	50
raw	Border security	6.2	3.3	9.1	4.3	2.1	6.4	−1.9	−6.0	+2.4	58,832	36,812	50	50
raw	Concealed carry	4.8	2.8	7.2	4.8	2.4	7.4	0.0	−3.5	+3.6	71,622	46,533	50	50
raw	Path to legal status	5.0	2.8	7.3	3.1	0.5	6.0	−1.9	−5.7	+1.9	58,847	36,804	50	50
raw	Gay marriage	7.5	2.4	12.5	13.0	9.3	16.2	+5.5	+0.7	+10.2	31,387	18,873	50	50
raw	Background checks	0.5	−0.2	1.3	3.6	1.4	5.6	+3.0	+0.5	+5.3	58,365	37,872	50	50
demographics	Border wall	1.2	−1.0	3.3	2.3	0.5	4.1	+1.1	−2.3	+4.5	36,193	24,604	50	50
demographics	Repeal the ACA	3.6	1.7	5.5	2.7	1.1	4.2	−0.9	−3.4	+1.5	71,486	46,549	50	50
demographics	Medicare for All	2.0	0.6	3.5	3.3	0.5	6.1	+1.3	−1.7	+4.2	36,050	25,013	50	50
demographics	Abortion legal	10.1	7.1	13.1	12.8	9.5	16.0	+2.7	−0.3	+5.6	68,625	44,899	50	50
demographics	EPA regulates CO2	3.8	2.2	5.4	5.7	3.1	8.2	+1.9	−1.3	+5.0	66,873	43,688	50	50
demographics	Cut immigration	2.6	0.6	4.7	0.4	−1.5	2.3	−2.3	−4.7	+0.2	30,037	21,417	50	50
demographics	Assault weapons ban	5.7	4.0	7.5	10.5	8.2	12.8	+4.8	+1.9	+7.7	71,507	46,481	50	50
demographics	Border security	4.9	2.4	7.4	1.7	−0.1	3.5	−3.1	−6.7	+0.4	58,764	36,802	50	50
demographics	Concealed carry	5.7	3.9	7.5	5.1	2.6	7.6	−0.7	−3.7	+2.4	71,554	46,523	50	50
demographics	Path to legal status	3.2	1.3	5.1	1.1	−1.3	3.5	−2.1	−5.2	+1.0	58,779	36,794	50	50
demographics	Gay marriage	10.2	6.6	13.8	12.7	9.0	16.3	+2.5	−1.7	+6.6	31,387	18,873	50	50
demographics	Background checks	1.0	0.3	1.8	4.1	1.8	6.4	+3.1	+0.5	+5.6	58,297	37,862	50	50
demographics_state	Border wall	0.7	−1.6	2.9	2.1	−0.1	4.4	+1.4	−2.5	+5.3	36,193	24,604	37	35
demographics_state	Repeal the ACA	2.8	0.6	4.9	1.5	−0.2	3.1	−1.3	−4.1	+1.5	71,486	46,549	38	35
demographics_state	Medicare for All	1.7	0.2	3.2	2.5	−0.2	5.2	+0.8	−2.2	+3.9	36,050	25,013	37	35
demographics_state	Abortion legal	6.4	3.4	9.4	8.1	6.1	10.1	+1.7	−1.9	+5.2	68,625	44,899	38	35
demographics_state	EPA regulates CO2	2.6	0.8	4.3	5.5	3.1	7.9	+2.9	−0.6	+6.5	66,873	43,688	38	35
demographics_state	Cut immigration	1.9	0.1	3.8	1.0	−1.3	3.3	−0.9	−3.1	+1.3	30,037	21,417	37	34
demographics_state	Assault weapons ban	4.4	2.3	6.6	8.7	6.8	10.7	+4.3	+1.3	+7.3	71,507	46,481	37	35
demographics_state	Border security	3.1	0.5	5.7	1.6	−0.2	3.5	−1.4	−5.2	+2.4	58,764	36,802	38	35
demographics_state	Concealed carry	4.5	2.5	6.4	4.9	2.6	7.2	+0.5	−3.1	+4.0	71,554	46,523	38	35
demographics_state	Path to legal status	2.7	0.7	4.7	2.0	−0.4	4.4	−0.7	−4.0	+2.6	58,779	36,794	38	35
demographics_state	Gay marriage	5.6	1.2	10.1	8.6	4.7	12.5	+3.0	−1.7	+7.6	31,387	18,873	36	34
demographics_state	Background checks	0.7	−0.3	1.6	3.3	1.3	5.3	+2.6	+0.5	+4.7	58,297	37,862	37	35

The Republican dot lies to the right of the Democratic dot on 9 of 12 issues, and the two are equal on easier concealed carry. The open dot is the Democratic place gap and the solid dot is the Republican place gap, and the number at the right is the Republican gap minus the Democratic gap. Whiskers are 95% intervals.Note: Raw gaps and the raw difference come from a bootstrap over states. Adjusted gaps are coefficients from weighted linear regressions among Democrats or Republicans in deep counties, with standard errors clustered by state and normal intervals. The adjusted difference is the coefficient on deep-blue by Republican in a regression with every term interacted with party, clustered the same way.Note: Rows are in the order of the first figure (party gap, largest first) in every view. The numbers in a tooltip are the respondents behind each dot. The data table gives the difference, its interval and the number of states behind each gap.Source: CES cumulative common content, 2013-2021; county returns for the two-party margin; weighted (weight_cumulative)

The second and third views of the figure adjust for education, race, age and gender in a weighted linear probability model, and the third adds state fixed effects, so that the comparison is between respondents in the same state. Adjusting for demographics alone leaves 15 of the 24 gaps smaller than the raw gap, and adding state fixed effects leaves 22, with 16 of the 24 having an interval above zero. The state fixed effects compare respondents within states, so they rest on the 34 to 38 states, depending on party and issue, that hold both deep-red and deep-blue counties, and the change from the raw gap combines the adjustment with a change in sample. With both adjustments the Republican gap is larger than the Democratic gap on 8 of the 12 issues, and the interval for the difference lies above zero on the assault weapons ban and background checks only. On the assault weapons ban the adjusted Republican gap is 8.7 (95% CI 6.8 to 10.7) points and the Democratic gap 4.4 (95% CI 2.3 to 6.6) points, a difference of +4.3 (95% CI 1.3 to 7.3) points. On abortion and legal gay marriage the adjusted intervals for the two parties overlap.

Independents with no lean have larger place gaps than partisans

Here an independent is a respondent at the middle of the seven-point party scale who did not name a party, and the two groups of leaners are shown apart. Across the 12 issues, the average place gap of independents with no lean is 9.2 (95% CI 7.4 to 10.7) points, which is 3.5 (95% CI 1.9 to 5.0) points more than the average for Democrats and Republicans together. Figure 6 shows all five groups. In point estimates the place gap of independents with no lean is larger than both partisan place gaps on 9 of the 12 issues. Democratic leaners have an average place gap of 3.3 points and Republican leaners of 6.7 points.

Figure 6

The average place gap is larger among independents with no lean than among Democrats and Republicans

Place gap averaged over the issues, in points (deep-blue minus deep-red counties, share taking the Democratic side), by party group, with 95% intervals; CES survey years 2013 to 2021

Show data

Data for The average place gap is larger among independents with no lean than among Democrats and Republicans
n	party_id	place_gap	place_gap_hi	place_gap_lo
9,432	Democrats	4.49	6.29	2.79
2,804	Democratic leaners	3.33	4.59	1.98
5,761	Independents (no lean)	9.18	10.70	7.37
2,783	Republican leaners	6.69	8.78	4.58
7,157	Republicans	6.78	8.47	5.00

Independents with no lean have the largest average place gap of the five groups in point estimates, and the interval for their difference from the average of Democrats and Republicans lies above zero. Each dot is a place gap averaged over the issues, with a 95% interval.Note: The tooltip shows the smallest of the two county-type groups behind each average. The place gaps of Democrats and Republicans in this figure are the same as the averages in the figure on Republican and Democratic place gaps. Intervals are 95% intervals from a bootstrap that resamples whole states; they are model based and ignore the design of the panel.Source: CES cumulative common content, 2013-2021; county returns for the two-party margin; weighted (weight_cumulative)

Figure 7 follows one issue, support for universal background checks, which has the smallest party gap and the smallest place gap of the issues in Figure 1, across the bins of county margin for all the party groups. Democrats support the policy at 95.0% to 96.3% in every bin, near the top of the scale, which leaves little room to rise, and Republicans at 80.3% to 84.4%. Independents with no lean go from 79.0% support in the most Republican bin to 86.1% in the most Democratic bin, a change of +7.1 (95% CI 3.2 to 10.8) points, and Republican leaners go from 71.0% to 79.3%, a change of +8.3 (95% CI 3.8 to 11.9) points. Both intervals lie above zero. For Democrats and Democratic leaners the interval for the change includes zero, and for Republicans the change is +3.4 (95% CI 0.0 to 5.9) points and the interval reaches zero.

Figure 7

Support for background checks rises with county margin among independents and Republican leaners, not among Democrats

Share supporting universal background checks (%) by party identification and two-party margin of the respondent's county, seven bins of counties; CES survey years 2013 to 2021

Show data

Data for Support for background checks rises with county margin among independents and Republican leaners, not among Democrats
bin	conf_high	conf_low	estimate	margin	n	party_id
Below -40	96.81	94.81	95.91	−53.35	7,003	Democrats
-40 to -20	95.87	94.21	95.02	−29.09	11,448	Democrats
-20 to -5	96.78	95.46	96.20	−12.70	11,168	Democrats
-5 to 5	97.09	95.31	96.29	0.112	10,899	Democrats
5 to 20	96.46	95.23	95.87	12.20	20,631	Democrats
20 to 40	96.70	95.37	96.04	28.48	17,396	Democrats
Above 40	96.61	95.06	95.80	56.26	22,518	Democrats
Below -40	96.66	92.91	94.77	−53.35	2,183	Lean Dem.
-40 to -20	96.23	93.87	95.06	−29.09	3,477	Lean Dem.
-20 to -5	95.66	92.41	94.06	−12.70	3,152	Lean Dem.
-5 to 5	96.77	93.52	95.18	0.112	2,952	Lean Dem.
5 to 20	97.06	95.64	96.37	12.20	5,520	Lean Dem.
20 to 40	97.23	94.11	95.61	28.48	4,321	Lean Dem.
Above 40	97.13	95.46	96.21	56.26	4,813	Lean Dem.
Below -40	82.12	75.65	78.95	−53.35	5,500	Independents
-40 to -20	83.92	79.89	82.06	−29.09	6,604	Independents
-20 to -5	85.36	78.83	82.51	−12.70	5,603	Independents
-5 to 5	87.87	82.44	85.18	0.112	4,769	Independents
5 to 20	86.61	83.58	85.13	12.20	8,452	Independents
20 to 40	87.70	83.30	85.67	28.48	6,078	Independents
Above 40	87.81	83.76	86.09	56.26	5,701	Independents
Below -40	73.68	68.67	70.95	−53.35	3,870	Lean Rep.
-40 to -20	74.74	68.63	71.91	−29.09	4,417	Lean Rep.
-20 to -5	77.88	70.76	74.35	−12.70	3,652	Lean Rep.
-5 to 5	78.91	73.10	76.15	0.112	2,888	Lean Rep.
5 to 20	81.74	76.67	79.38	12.20	4,687	Lean Rep.
20 to 40	83.15	77.37	80.47	28.48	3,132	Lean Rep.
Above 40	82.15	75.74	79.28	56.26	2,270	Lean Rep.
Below -40	81.79	78.63	80.31	−53.35	11,459	Republicans
-40 to -20	82.36	78.79	80.73	−29.09	13,038	Republicans
-20 to -5	83.15	79.25	81.50	−12.70	9,468	Republicans
-5 to 5	84.38	80.03	82.26	0.112	7,663	Republicans
5 to 20	84.28	80.85	82.84	12.20	11,619	Republicans
20 to 40	86.17	82.45	84.42	28.48	7,804	Republicans
Above 40	85.67	80.29	83.66	56.26	5,571	Republicans

Support is higher in the most Democratic bin than in the most Republican bin for independents with no lean and for Republican leaners, and shows no clear change for Democrats and Democratic leaners. Solid lines are Democrats, Republicans and independents, dashed lines are leaners (Lean Dem. and Lean Rep.), and bands are 95% intervals.Note: The vertical axis starts at 60%, not at zero, so slopes look steeper than in the county-margin figure with a full axis.Note: Intervals are 95% intervals from a bootstrap that resamples whole states; they are model based and ignore the design of the panel.Source: CES cumulative common content, 2013-2021; county returns for the two-party margin; weighted (weight_cumulative)

The party gap rose on most issues

Figure 8 follows 4 issues across presidential cycles, including the two with the longest history, border security and a path to legal status. A presidential cycle pools the surveys from one presidential election to the next and uses the county margin of that election, so the 2016 cycle covers surveys from 2016 to 2019. The trend draws on the 2006 to 2021 source window and not only on the 2013 to 2021 interviews of the main comparison, so its earlier cycles include surveys from before 2013. Between the first and last cycle in which an issue was asked, the party gap rose on 3 of the 4 issues, with an interval above zero. On repeal of the ACA the party gap went from 53.8 points in the 2012 cycle to 55.8 points in the 2020 cycle, a change of +2.0 (95% CI 0.0 to 4.1) points whose interval reaches zero. With the 2014 answers dropped instead of reversed, the change is +5.9 (95% CI 3.6 to 8.3) points, with an interval above zero.Not every issue was asked in every survey, so an endpoint can rest on one or two surveys. The 2004 cycle is the 2007 survey alone and its groups are small, universal background checks in the 2020 cycle rest on the 2021 survey alone, and abortion in the 2012 cycle rests on the 2014 and 2015 surveys. The survey years behind every point are in the data tables.

Figure 8

The party gap rose on 3 of these 4 issues across presidential cycles and stayed above both place gaps

Gap in points (share taking the Democratic side) by presidential cycle, Democrats and Republicans in deep counties; four issues; CES surveys 2006 to 2021

Between-party gap
Place gap, Democrats
Place gap, Republicans

Show data

Data for The party gap rose on 3 of these 4 issues across presidential cycles and stayed above both place gaps
Issue	Series	Presidential cycle	Survey years	Estimate	Low	High	Smallest group, n
Path to legal status	Between-party gap	2004	2007	27.8	20.8	33.1	367
Path to legal status	Place gap, Democrats	2004	2007	11.2	2.2	18.4	367
Path to legal status	Place gap, Republicans	2004	2007	1.9	−3.5	9.3	367
Path to legal status	Between-party gap	2008	2010 and 2011	32.1	29.3	34.9	2,932
Path to legal status	Place gap, Democrats	2008	2010 and 2011	9.2	4.2	14.8	2,932
Path to legal status	Place gap, Republicans	2008	2010 and 2011	2.7	−0.6	6.9	2,932
Path to legal status	Between-party gap	2012	2012 to 2015	35.7	32.9	38.1	6,215
Path to legal status	Place gap, Democrats	2012	2012 to 2015	9.0	5.5	12.5	6,215
Path to legal status	Place gap, Republicans	2012	2012 to 2015	3.8	0.2	7.8	6,215
Path to legal status	Between-party gap	2016	2016 and 2017 and 2019	40.0	37.7	42.0	5,394
Path to legal status	Place gap, Democrats	2016	2016 and 2017 and 2019	1.9	0.0	3.7	5,394
Path to legal status	Place gap, Republicans	2016	2016 and 2017 and 2019	3.5	−0.4	7.5	5,394
Path to legal status	Between-party gap	2020	2020 and 2021	43.3	41.4	45.3	4,267
Path to legal status	Place gap, Democrats	2020	2020 and 2021	1.7	−0.2	3.6	4,267
Path to legal status	Place gap, Republicans	2020	2020 and 2021	2.5	−0.4	5.2	4,267
Border security	Between-party gap	2004	2007	30.1	20.5	37.9	367
Border security	Place gap, Democrats	2004	2007	6.2	−2.6	12.7	367
Border security	Place gap, Republicans	2004	2007	−2.8	−11.5	8.4	367
Border security	Between-party gap	2008	2010 and 2011	29.4	27.0	31.4	2,932
Border security	Place gap, Democrats	2008	2010 and 2011	9.5	5.9	13.3	2,932
Border security	Place gap, Republicans	2008	2010 and 2011	2.3	−1.5	5.7	2,932
Border security	Between-party gap	2012	2012 to 2015	33.2	30.6	35.9	6,215
Border security	Place gap, Democrats	2012	2012 to 2015	6.1	2.4	9.6	6,215
Border security	Place gap, Republicans	2012	2012 to 2015	3.7	0.1	6.9	6,215
Border security	Between-party gap	2016	2016 and 2017 and 2019	41.0	38.9	42.9	5,396
Border security	Place gap, Democrats	2016	2016 and 2017 and 2019	6.8	3.4	10.2	5,396
Border security	Place gap, Republicans	2016	2016 and 2017 and 2019	4.0	1.7	6.9	5,396
Border security	Between-party gap	2020	2020 and 2021	46.8	45.1	48.8	4,271
Border security	Place gap, Democrats	2020	2020 and 2021	7.5	3.4	10.8	4,271
Border security	Place gap, Republicans	2020	2020 and 2021	4.6	2.0	6.6	4,271
Repeal the ACA	Between-party gap	2012	2012 to 2015	53.8	51.8	55.8	6,062
Repeal the ACA	Place gap, Democrats	2012	2012 to 2015	8.6	5.5	11.6	6,062
Repeal the ACA	Place gap, Republicans	2012	2012 to 2015	2.3	−0.4	4.7	6,062
Repeal the ACA	Between-party gap	2016	2016 to 2019	59.5	58.0	61.5	8,476
Repeal the ACA	Place gap, Democrats	2016	2016 to 2019	1.2	−1.8	4.0	8,476
Repeal the ACA	Place gap, Republicans	2016	2016 to 2019	5.2	3.4	6.9	8,476
Repeal the ACA	Between-party gap	2020	2020 and 2021	55.8	54.2	57.7	4,262
Repeal the ACA	Place gap, Democrats	2020	2020 and 2021	0.2	−2.8	3.0	4,262
Repeal the ACA	Place gap, Republicans	2020	2020 and 2021	4.5	1.7	7.0	4,262
Abortion legal	Between-party gap	2012	2014 and 2015	39.6	36.9	42.3	3,052
Abortion legal	Place gap, Democrats	2012	2014 and 2015	14.6	9.6	19.2	3,052
Abortion legal	Place gap, Republicans	2012	2014 and 2015	13.8	9.0	17.8	3,052
Abortion legal	Between-party gap	2016	2016 to 2019	46.5	44.3	49.2	8,493
Abortion legal	Place gap, Democrats	2016	2016 to 2019	9.2	5.6	12.7	8,493
Abortion legal	Place gap, Republicans	2016	2016 to 2019	15.6	11.3	19.3	8,493
Abortion legal	Between-party gap	2020	2020 and 2021	53.0	50.4	55.7	4,273
Abortion legal	Place gap, Democrats	2020	2020 and 2021	4.9	2.0	8.2	4,273
Abortion legal	Place gap, Republicans	2020	2020 and 2021	12.0	7.5	15.7	4,273

The dark between-party gap line rises from the first to the last cycle on 3 of the 4 issues and stays above both place gap lines in every cycle. Each panel is one issue; the solid blue line is the Democratic place gap and the dashed red line is the Republican place gap. Bands are 95% intervals.Note: A presidential cycle pools the surveys from one presidential election to the next and uses the county margin of that election. Not every issue was asked in every survey, so the first and last cycle of an issue can rest on one or two surveys. Survey years behind the first and last cycle of each panel: Path to legal status, 2004 cycle 2007 and 2020 cycle 2020 and 2021; Border security, 2004 cycle 2007 and 2020 cycle 2020 and 2021; Repeal the ACA, 2012 cycle 2012 to 2015 and 2020 cycle 2020 and 2021; Abortion legal, 2012 cycle 2014 and 2015 and 2020 cycle 2020 and 2021. The data table gives the survey years behind every point.Note: The 2004 cycle rests on one survey year, so its bands are wide. A cycle with a party by county-type group below the suppression floor would be dropped; none was. The number in a tooltip is the smallest of the four party by county-type groups behind the estimate.Note: Intervals are 95% intervals from a bootstrap that resamples whole states; they are model based and ignore the design of the panel.Source: CES cumulative common content, 2006-2021; county returns for the two-party margin; weighted (weight_cumulative)

Figure 9 gives the change for all 12 issues, grouped by the span of cycles over which each was asked. The party gap rose on 8 of them with an interval above zero. It fell on universal background checks, where the change is −3.3 (95% CI −6.0 to −0.7) points, and the interval includes zero on 3: legal gay marriage, repeal of the ACA and cutting legal immigration. The issues asked over the longest span, border security and a path to legal status, have the widest intervals: between the 2004 and 2020 cycles the party gap rose by +16.7 points on border security and +15.5 points on a path to legal status. Between the 2012 and 2020 cycles it rose by +13.4 points on abortion and +10.3 points on the assault weapons ban.

Figure 9

The party gap rose on 8 of 12 issues over spans that differ by issue, and fell on background checks

Change in the between-party gap in points (share taking the Democratic side), first to last presidential cycle in which each issue was asked, grouped by span; Democrats and Republicans in deep counties; CES surveys 2006 to 2021

Cycles 2004 to 2020

Cycles 2012 to 2020

Cycles 2012 to 2016

Cycles 2016 to 2020

Change in the party gap (points)

Show data

Data for The party gap rose on 8 of 12 issues over spans that differ by issue, and fell on background checks
Issue	Cycles	Survey years, first cycle	Survey years, last cycle	Party gap change	Low	High	Party gap change per cycle	Place gap change	Place gap change, low	Place gap change, high	Smaller endpoint, n
Border security	Cycles 2004 to 2020	2007	2020 and 2021	+16.7	+9.4	+26.0	+4.2	+4.3	−2.6	+10.6	2,767
Path to legal status	Cycles 2004 to 2020	2007	2020 and 2021	+15.5	+10.2	+22.9	+3.9	−4.5	−9.8	+0.9	2,767
Abortion legal	Cycles 2012 to 2020	2014 and 2015	2020 and 2021	+13.4	+9.8	+17.0	+6.7	−5.8	−8.7	−2.4	20,940
Assault weapons ban	Cycles 2012 to 2020	2013 to 2015	2020 and 2021	+10.3	+7.6	+13.4	+5.2	−1.5	−4.5	+1.4	25,669
Concealed carry	Cycles 2012 to 2020	2013 to 2015	2020 and 2021	+6.8	+3.8	+10.2	+3.4	−0.9	−3.7	+1.8	25,716
EPA regulates CO2	Cycles 2012 to 2020	2014 and 2015	2020 and 2021	+3.5	+0.3	+7.2	+1.8	−1.0	−4.4	+2.3	20,800
Repeal the ACA	Cycles 2012 to 2020	2012 to 2015	2020 and 2021	+2.0	0.0	+4.1	+1.0	−3.1	−5.3	−0.9	31,281
Background checks	Cycles 2012 to 2020	2013 to 2015	2021	−3.3	−6.0	−0.7	−1.6	−1.3	−3.6	+1.1	9,240
Gay marriage	Cycles 2012 to 2016	2012 to 2015	2016	−2.1	−4.3	+0.5	−2.1	−4.2	−6.4	−1.8	24,606
Border wall	Cycles 2016 to 2020	2017 and 2018	2020 and 2021	+6.1	+4.1	+8.3	+6.1	−0.3	−2.2	+1.6	29,562
Medicare for All	Cycles 2016 to 2020	2018 and 2019	2020 and 2021	+5.7	+3.2	+8.5	+5.7	−1.6	−4.1	+0.8	29,844
Cut immigration	Cycles 2016 to 2020	2018 and 2019	2020	−0.5	−3.0	+2.1	−0.5	+0.2	−2.0	+2.4	22,064

The interval for the change lies above zero on 8 issues and includes zero on 3, and the two issues asked over the longest span have the widest intervals. Each dot is the change in the party gap, in points, between the first and the last presidential cycle in which the issue was asked; the panels group issues by that span. Whiskers are 95% intervals.Note: Issues were asked over different spans, so the cycles differ by row. The survey years behind each cycle, the change per cycle and the change in the place gap with its interval are in the data table. The number in a tooltip is the size of the smaller of the two endpoint cycles.Note: Universal background checks were not asked in 2020, so the last cycle for that issue rests on the 2021 survey alone.Note: Intervals are 95% intervals from a bootstrap that resamples whole states; they are model based and ignore the design of the panel.Source: CES cumulative common content, 2006-2021; county returns for the two-party margin; weighted (weight_cumulative)

The average place gap changed by more than sampling error on 3 of the 12 issues, and each change is a fall: −5.8 (95% CI −8.7 to −2.4) points on abortion, −4.2 (95% CI −6.4 to −1.8) on legal gay marriage and −3.1 (95% CI −5.3 to −0.9) on repeal of the ACA. With the 2014 answers to the ACA question dropped instead of reversed, the change on repeal of the ACA is −0.7 (95% CI −3.0 to 1.6) points, and its interval includes zero. The fall on abortion is mostly a Democratic one: the Democratic place gap goes from 14.6 points in the 2012 cycle to 4.9 points in the 2020 cycle, and the Republican gap from 13.8 to 12.0 points. The Democratic place gap on a path to legal status changed by −7.1 (95% CI −10.3 to −4.2) points between the 2012 and 2016 cycles, from 9.0 to 1.9 points, and was 1.7 points in the 2020 cycle, a change of −0.2 (95% CI −2.6 to 2.2) points from 2016. The steps before 2012 have intervals that include zero.

A second survey gives ratios of the same order

The CES records the respondent's county. The American National Election Studies (ANES) record only the state, so a check with ANES uses a coarser place. Figure 10 repeats the calculation for strong and not very strong partisans in deep states, meaning states with a two-party presidential margin of 20 points or more. The four groups of party by state type hold 3,752 to 4,410 respondents in all, depending on the measure, and 3,933 on the liberal-conservative scale.ANES states are known for only part of the 2024 sample, and the gun question was not asked in 2024. The ANES partisans are those who call themselves strong or not very strong Democrats or Republicans, which matches the three-point CES measure. Leaners are excluded. The liberal-conservative scale, rescaled so that 100 is the most liberal answer, has a party gap of 41.3 points and a place gap of 4.7 points, a ratio of 8.8 (95% CI 4.7 to 24.2). The ratios on the ANES issue questions have smaller point estimates: 3.2 for making it harder to buy a gun, 2.9 for not decreasing immigration, 2.8 for a law against job discrimination and 2.5 (95% CI 1.5 to 4.2) for abortion. Among these questions only the interval for abortion lies below the interval for the ideology scale, as the ratio tab of the figure shows.

Figure 10

In both surveys the party gap is larger than the place gap on every measure

Party gap and place gap at the state level, and their ratio, by measure; ANES waves 2012 to 2024 and CES survey years 2013 to 2021

The party gap is larger than the place gap on every measure in both surveys

Gap in points on a 0 to 100 scale, strong and not very strong partisans in deep states; ANES waves 2012 to 2024 and CES survey years 2013 to 2021

Place gap
Party gap

The open dot is the place gap (deep-blue minus deep-red states within a party) and the solid dot is the party gap (Democrats minus Republicans within a state type). Issue rows are the share taking the Democratic side; the ideology row is the liberal-conservative scale rescaled so that liberal is 100. A deep state has a two-party presidential margin of 20 points or more. The last two rows are the CES averages at the state and county levels.

Show data

Data for In both surveys the party gap is larger than the place gap on every measure, Gaps
Measure	Place level	Party gap	Party gap, low	Party gap, high	Place gap	Place gap, low	Place gap, high	Respondents
ANES ideology	state	41.3	38.8	43.6	4.7	1.7	8.8	3,933
ANES abortion	state	42.9	37.7	47.6	17.4	11.2	28.2	4,410
ANES gun purchase	state	44.1	41.0	50.0	13.8	6.7	19.3	3,752
ANES immigration	state	27.7	23.0	30.5	9.4	2.9	18.0	4,005
ANES job discrimination	state	16.5	11.5	19.6	6.0	1.6	13.2	3,889
CES abortion	state	46.4	43.7	48.7	14.3	11.2	18.9	113,602
CES all issues, state	state	43.3	41.2	45.2	5.5	3.3	8.2	67,860
CES all issues, county	county	43.1	41.8	44.5	5.6	4.7	6.6	118,638

The party gap is larger than the place gap on every measure, and the interval for the difference lies above zero on each. The first tab plots the two gaps and the second their ratio with its 95% interval. Abortion is the one question asked in both surveys.Note: ANES partisans are strong and not very strong identifiers, which matches the three-point CES measure. Leaners are excluded. State is known for only part of the 2024 ANES sample, and the gun question was not asked in 2024.Note: Each issue in ANES is a different question from the CES issues except abortion, so only the abortion rows compare the same wording. Intervals come from a bootstrap over the states in the ANES sample and are wide because the deep-state groups are small.Source: ANES cumulative data file, waves 2012-2024, weighted (VCF0009z); CES cumulative common content, 2013-2021, weighted (weight_cumulative); state returns for the two-party margin

Abortion is the one question asked in both surveys. At the state level, ANES gives a party gap of 42.9 points and a place gap of 17.4 points, and the CES gives 46.4 and 14.3, a ratio of 3.2 (95% CI 2.3 to 4.3) against 2.5 in ANES. Over all issues at the state level, the CES party gap is 43.3 points, the place gap 5.5 points and the ratio 7.9 (95% CI 5.1 to 13.7), close to the county-level values in Figure 1. The ANES intervals are wide because each group of party by state type holds a limited number of states and respondents, and the two surveys differ in questions, years and level of place. They agree on the order of magnitude of the ratio, and the intervals are too wide to say more.

The party gap does not vary clearly with state partisanship

Figure 11 asks whether the party gap depends on how Democratic or Republican a state is. Each dot is a state, with its party gap averaged over the issues plotted against the state's mean two-party margin. The state party gap averages 45.7 points, with a standard deviation of 5.4 points across 49 states, and that spread includes the sampling error of each state's estimate, whose standard error is 1.3 points in the median state. The District of Columbia is not plotted because too few issues have enough respondents of each party there. The fitted slope is −0.5 (95% CI −1.3 to 0.4) points of party gap per 10 points of Democratic margin. The point estimate is a fall in the gap as states become more Democratic, and the interval also includes no change.

Figure 11

The party gap averages 45.7 points across states and shows no clear relation to a state's partisanship

State party gap averaged over the issues, in points, against the state's mean two-party presidential margin; Democrats and Republicans; CES survey years 2013 to 2021

State two-party margin (points, Democratic minus Republican)

Show data

Data for The party gap averages 45.7 points across states and shows no clear relation to a state's partisanship
State	State two-party margin	Party gap	Standard error	Party gap, low	Party gap, high	Respondents per issue
AL	−26.3	41.5	1.3	38.9	44.1	2,259
AZ	−4.6	49.9	1.0	48.0	51.9	3,836
AR	−27.1	41.0	1.7	37.7	44.2	1,287
CA	+28.9	44.1	0.6	43.0	45.2	15,069
CO	+7.3	53.3	1.2	51.0	55.7	2,446
CT	+16.6	41.3	1.8	37.8	44.8	1,569
DE	+15.7	43.0	2.4	38.3	47.8	650
FL	−1.0	41.8	0.6	40.7	42.9	12,328
GA	−4.9	42.6	0.9	40.9	44.3	5,087
HI	+36.0	31.5	3.2	25.2	37.7	479
ID	−34.3	51.2	2.2	46.8	55.6	761
IL	+17.6	43.5	0.8	41.8	45.1	6,436
IN	−16.1	44.4	1.0	42.4	46.4	3,457
IA	−4.0	47.9	1.3	45.3	50.5	1,539
KS	−20.7	49.8	1.6	46.7	52.9	1,215
KY	−27.6	39.5	1.4	36.9	42.2	2,683
LA	−19.1	38.0	1.6	34.9	41.0	1,758
ME	+8.5	49.2	2.1	45.1	53.3	789
MD	+28.9	46.7	1.0	44.8	48.7	3,066
MA	+28.3	45.8	1.4	43.1	48.6	2,552
MI	+3.8	44.5	1.0	42.6	46.5	4,873
MN	+5.0	52.2	1.1	50.0	54.4	2,676
MS	−15.8	37.1	1.8	33.7	40.6	1,162
MO	−15.4	48.1	1.0	46.1	50.2	3,397
MT	−18.6	56.2	2.2	51.9	60.5	502
NE	−23.5	52.7	1.6	49.6	55.9	950
NV	+4.0	46.4	1.6	43.3	49.5	1,851
NH	+3.7	50.5	2.2	46.2	54.8	692
NJ	+16.0	42.0	0.9	40.1	43.8	4,431
NM	+10.1	49.3	1.9	45.5	53.1	1,018
NY	+27.8	38.9	0.7	37.6	40.3	10,028
NC	−2.7	44.9	0.8	43.2	46.5	4,781
ND	−31.9	48.4	3.2	42.1	54.6	313
OH	−4.4	42.3	0.7	40.9	43.8	6,810
OK	−36.0	47.4	1.7	44.1	50.7	1,528
OR	+13.2	57.3	1.1	55.2	59.4	2,468
PA	+1.8	43.3	0.6	42.1	44.5	9,050
RI	+21.5	39.6	3.4	33.0	46.2	404
SC	−12.8	43.4	1.1	41.2	45.7	2,259
SD	−26.2	52.7	2.5	47.8	57.6	413
TN	−24.3	45.5	1.2	43.2	47.9	3,129
TX	−10.5	43.4	0.6	42.2	44.6	10,983
UT	−32.2	45.1	1.7	41.7	48.5	1,215
VT	+34.0	41.2	3.9	33.6	48.8	298
VA	+6.1	45.5	1.0	43.6	47.3	4,569
WA	+17.3	53.2	1.1	51.0	55.5	3,329
WV	−37.4	37.8	1.8	34.2	41.3	1,078
WI	+2.2	48.4	0.9	46.5	50.2	3,156
WY	−47.1	54.4	3.0	48.5	60.4	265

State party gaps show no clear rise or fall from the most Republican to the most Democratic states: the fitted slope is −0.5 points of party gap per 10 points of Democratic margin, and its interval includes zero. The labelled dots are the three lowest and the three highest state gaps; every state is in the data table. The line is a least squares fit with a 95% band from a bootstrap over states.Note: The dots carry no whiskers because the scatter chart draws none. The data table gives each state's standard error and a 95% interval (normal, from a linearised standard error that treats respondents within a state as independent). The largest standard error is 3.9 points, and the smallest state has 265 Democrats and Republicans per issue on average.Note: A state needs at least 50 Democrats and 50 Republicans on an issue for that issue to count, and at least ten issues to appear. The tooltip shows the average number of respondents per issue. The District of Columbia has fewer than ten such issues and is not plotted, so the scatter has 49 states while the bootstrap for the other estimates has 50 clusters.Source: CES cumulative common content, 2013-2021; state presidential returns (MEDSL) for the state two-party margin; weighted (weight_cumulative)

Hopkins argued that American political behavior has nationalized, with voters more engaged with national politics than with the politics of places nearer home [1]. If that account holds, the party gap should depend little on a state's partisanship, and the flat slope is consistent with it, though the interval also allows a modest fall. This chapter does not test the account, and a flat relation across states says nothing about influence inside a county or a neighborhood, which Chapter 15.3, Variation in opinion within places (planned) takes up.

What these data cannot show

The place gap is a difference between respondents who live in different counties. It is not an estimate of what a county does to a person. People choose where to live, and the respondents who live in deep-blue counties may differ from those in deep-red counties in ways that survey questions on education, race, age and gender do not capture. Mummolo and Nall ask why partisans do not sort more by place and study the constraints on political segregation [2]. Earlier work on political context, from a 1984 survey in South Bend, Indiana, found that people build discussion networks that match their own views, that the political composition of their social context limits this, and that the interaction favors the local majority [3]. A cross-section cannot separate influence from selection, and evidence on people who move is the design that can. Chapter 15.6, Movers and opinion change (planned) takes up that design.

The party gap is not an estimate of what a party does to an opinion either. Party identification and issue positions shape each other, and party has taken on other identities, so a Democrat and a Republican also differ in race, religion and much else [4]. The two gaps are also measured on 12 questions that were available in the survey, on guns, immigration, abortion, marriage, health care and climate. Their ratio would change with a different set of issues, and the ceiling of a share on a yes-or-no question limits how large a gap can be. The policy questions end in 2021, so the chapter says nothing about later years.

The county is one level of place. A finer level, such as a neighborhood, could show larger place gaps, and a coarser one, such as the state, shows the ANES and CES gaps above. The chapter does not measure the variation in opinion among people who live in one place, which Chapter 15.3, Variation in opinion within places (planned) measures, or the share of partisans who live where their party loses, which Chapter 15.4, Partisans in the other party's territory (planned) measures. Place-specific contexts such as recent immigration may matter more for some issues than the county margin does, as Chapter 15.2, Immigration and local context (planned) examines, and the change in a county's vote over time has its own accounting in Chapter 15.5, Composition and behavior in county change (planned).

The intervals describe sampling variation in an opt-in online panel. They treat the states as an independent sample of clusters and the survey weights as correct, and they do not cover error from who joins the panel or from question wording. With few clusters, standard cluster-robust tests can over-reject [5], and the resampling method used here is simpler than the one that paper proposes. The ratio is uncertain where the place gap is small.

Data sources

Cooperative Election StudyCooperative Election Study cumulative common content, 2006-2021 policy items, party identification and county FIPSYears: 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.
County election returnsCounty presidential returns 2004-2020 (repaired file), two-party marginYears: President 1868-2024 (40 elections); House and Senate 1990-2024 · Weights: none (full counts) · 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.
State presidential returnsState presidential returns 1976-2024, two-party marginYears: 1976-2024 (13 elections) · Weights: none (full counts) · License: See the provider's terms of use.
ANES Cumulative Data FileANES Time Series Cumulative Data File, waves 2012-2024, state identifier, issue itemsYears: 1948-2024 (33 study years; even years only after 2004; no 2006, 2010 or 2014 study) · Weights: weight (VCF0009z) · License: Free download after accepting the ANES terms of use; cite the release.

All data sources

Methods

Sample: Cooperative Election Study interviews in 2013 to 2021 (2006 to 2021 for the trend by presidential cycle) that could be matched to a county with presidential returns, Alaska excluded, with a positive weight: 333,335 interviews in 2013 to 2021, of which 320,988 are in a party group used here and 118,638 are Democrats or Republicans in deep counties. Coding: in the cumulative file the 2014 answers to the question on repealing the Affordable Care Act have the partisan pattern of the opposite code from the other years (the Republican minus Democratic share supporting repeal is negative in 2014 and positive in every other year), so the 2014 answers are reversed. The guide to the cumulative policy file records that the 2014 question was worded differently from the other years (a vote to repeal, in Congress, after a description of the law, where the adjacent years asked whether the respondent supported or opposed repeal) and that an earlier release of the file had switched the 2014 answers and the current release does not. The guide gives one pooled set of value labels and no year-specific labels for 2014, so the reversal rests on the partisan pattern and on that record, and the original 2014 labels have not been checked against the questionnaire. The build stops if any issue has a Democratic minus Republican gap at or below zero in any survey year, which tests the expected partisan direction and is not an independent check of the coding. Party: the three-point identification, so leaners are outside the Democrats and Republicans; independents are respondents at the middle of the seven-point scale, and leaners are shown apart. Place: the two-party margin of the respondent's county in the most recent presidential election, counting the election of the interview year, from the repaired county returns file; a deep county has a margin of 20 points or more toward one party. Outcomes: 12 yes-or-no policy questions, each oriented so that 100 is the side that Democrats take more often. Weights: weight_cumulative, which gives every survey year the same total weight. Estimator: weighted means in party by county-type cells; the party gap is the Democratic minus Republican difference averaged over deep-red and deep-blue counties, the place gap is the deep-blue minus deep-red difference averaged over Democrats and Republicans, and the ratio is the party gap divided by the place gap. Adjusted place gaps are coefficients on a deep-blue indicator in weighted linear regressions with education, race, age and gender, with and without state fixed effects, and standard errors clustered by state; the adjusted Republican minus Democratic difference is the coefficient on deep-blue by Republican in a regression with every term interacted with party. Intervals: unless stated otherwise, 95% percentile intervals from a bootstrap that resamples whole states (50 clusters, the states and the District of Columbia, 2,000 replicates, seed 12345), with gaps, ratios and medians computed inside each replicate. Respondents in one state share a political environment, so treating them as independent can give intervals that are too narrow; with few clusters even cluster-robust tests can over-reject (Cameron, Gelbach and Miller 2008), and the percentile bootstrap used here is simpler than the wild cluster bootstrap-t of that paper. The exceptions are the adjusted gaps (normal intervals from the clustered regressions), the ANES intervals (a bootstrap over the states in the ANES sample), the band on the state scatter (a bootstrap over the plotted states) and the state intervals in its data table (linearised standard errors). Resampling counties instead of states gives a narrower interval around the ratio of the average gaps, 7.6. These intervals are model based and ignore the design of the online panel. Suppression: a cell needs at least 100 respondents and an effective sample of 100; the smallest cell behind an issue holds 7,157 respondents and none fell below the floor. The state-level check uses the ANES cumulative file for 2012, 2016, 2020 and 2024 (strong and not very strong partisans, respondents with a known state, weight VCF0009z) and the same estimator with deep states, a two-party margin of 20 points or more from the state returns.

References

Hopkins, Daniel J. (2018). The Increasingly United States: How and Why American Political Behavior Nationalized. University of Chicago Press. press.uchicago.edu/ucp/books/book/chicago/I/bo27596045.html
Mummolo, Jonathan; Nall, Clayton (2017). Why Partisans Do Not Sort: The Constraints on Political Segregation. The Journal of Politics 79(1): 45-59. doi:10.1086/687569
Huckfeldt, Robert; Sprague, John (1987). Networks in Context: The Social Flow of Political Information. American Political Science Review 81(4): 1197-1216. www.cambridge.org/core/journals/american-political-science-review/article/abs/networks-in-context-the-social-flow-of-political-information/393FFEC212ECE14C0E0F79715E086D60
Mason, Lilliana (2018). Uncivil Agreement: How Politics Became Our Identity. University of Chicago Press. press.uchicago.edu/ucp/books/book/chicago/U/bo27527354.html
Cameron, A. Colin; Gelbach, Jonah B.; Miller, Douglas L. (2008). Bootstrap-Based Improvements for Inference with Clustered Errors. The Review of Economics and Statistics 90(3): 414-427. direct.mit.edu/rest/article/90/3/414/57731/Bootstrap-Based-Improvements-for-Inference-with

How to cite

@misc{singh2026partyandplace,
  author = {Singh, Vishal},
  title = {Party and place},
  year = {2026},
  url = {https://vishalsingh.org/politics/party-and-place},
  note = {In The Public Mind. Published 30 September 2026.}
}

Same question at another grain

These chapters take up the same axis with a different unit of analysis: individuals, places or countries.

Party and place
- Chapter 1.4, Cohort change in partisan sorting Individuals
- Chapter 15.5, Composition and behavior in county change (planned) Places

The party gap exceeds the place gap on every issue

The ratio varies widely by issue

Sensitivity to the county cutoff and to leaners

Support varies less across county margins than between the parties

Republican place gaps are larger on most issues, not reliably

Independents with no lean have larger place gaps than partisans

The party gap rose on most issues

A second survey gives ratios of the same order

The party gap does not vary clearly with state partisanship

What these data cannot show

Data sources

Methods

References

How to cite

Same question at another grain

Reading paths