Party and ideology

Among Democrats who placed themselves on the liberal-conservative scale of the American National Election Studies (ANES), the share who called themselves conservative fell from 25%25% percent of the party's identifiers who placed themselvesEstimandweighted share in the conservative camp, Wilson intervalInterval95% CI 22% to 28%n776Weightsweight (VCF0009z)PopulationANES Democrat identifiers (leaners included) who placed themselves on the liberal-conservative scalePeriod1972SourceANES Cumulative Data File in 1972 to 8%8% percent of the party's identifiers who placed themselvesEstimandweighted share in the conservative camp, Wilson intervalInterval95% CI 7% to 10%n2,310Weightsweight (VCF0009z)PopulationANES Democrat identifiers (leaners included) who placed themselves on the liberal-conservative scalePeriod2024SourceANES Cumulative Data File in 2024.Democrats and Republicans here include independents who say they lean toward the party, the leanersPartisan leanerA respondent who first says they are independent but, when asked, says they lean toward one party. Most figures group leaners with the party they lean toward, and each chapter says which grouping it uses., throughout the chapter. Republicans who called themselves liberal thinned out over the same years. As these opposite-label camps shrank, a person's party came to say more about which ideological label they would choose. In 1972 a Democrat could be found under either label; by 2024 a conservative Democrat or a liberal Republican was the exception in each party.

The two labels drew together on all three surveys

The camp shares follow one party at a time. A correlationCorrelationA number from minus 1 to plus 1 for how closely two measures move together in a straight-line pattern. Zero means no straight-line pattern; a correlation does not say that one measure causes the other. puts the whole link between party and label into one number, from minus one to plus one, for how closely two answers move together along a straight line. Near zero, no straight line links the two answers; near one, the party answer nearly fixes the placement. Among ANES respondents who answered both questions, the correlation between party identification and liberal-conservative self-placement was +0.32+0.32 correlation coefficient, minus 1 to plus 1Estimandweighted Pearson correlation between party identification and liberal-conservative self-placementInterval95% CI 0.27 to 0.36n1,545Weightsweight (VCF0009z)PopulationANES respondents with a party identification (leaners kept) and a liberal-conservative self-placementPeriod1972SourceANES Cumulative Data File in 1972 and +0.72+0.72 correlation coefficient, minus 1 to plus 1Estimandweighted Pearson correlation between party identification and liberal-conservative self-placementInterval95% CI 0.70 to 0.74n4,716Weightsweight (VCF0009z)PopulationANES respondents with a party identification (leaners kept) and a liberal-conservative self-placementPeriod2024SourceANES Cumulative Data File in 2024.

Figure 1 plots the correlation for every survey year. Every ANES reading from 1992 on is above every reading from 1972 through 1990. Both the ANES and the General Social Survey (GSS) had also risen before then, and their first readings of the 1990s do not differ detectably from the earlier peaks, so neither survey dates the step to a single year. The GSS runs below the ANES in every year in which both surveys ran, and it also ends higher than it starts. The Cooperative Election Study (CES), an online survey that begins in 2006, starts above both, falls between 2008 and 2016, and ends higher than it began. The three surveys agree on the direction and disagree on the level, so the chapter compares changes within each survey and never pools them.

How the three surveys measure it

The ANES has asked both questions since 1972, the GSS since 1974 and the CES since 2006. Each interviews its own sample of Americans, each estimate is weightedSurvey weightsNumbers that scale each respondent so the sample matches the population on known traits such as age, race, education and region. A weighted average counts a respondent with a large weight for more than one with a small weight. so that the sample matches the population on known traits, and the three share no respondents, so every number here is computed inside one survey. Party identification is a seven-point scale from strong Democrat to strong Republican on which leaners keep their own points. Self-placement runs from extremely liberal to extremely conservative on a seven-point scale in the ANES and the GSS and on a five-point scale in the CES. People who decline to place themselves are left out, so every estimate describes people who did. The ANES shows no further rise between 2020 and 2024, where the difference is −0.01 (95% CI −0.04 to 0.01)−0.01 difference in correlation coefficientsEstimand2024 minus 2020 ANES correlation; standard errors of independent estimates combined in quadratureInterval95% CI −0.04 to 0.01n11,757Weightsweight (VCF0009z)PopulationANES respondents with a party identification (leaners kept) and a liberal-conservative self-placementPeriod2020 and 2024SourceANES Cumulative Data File, an interval that includes zero. The report gives every estimate with its interval and sample size.

The opposite camps shrank

Figure 2 returns to the labels one party at a time, as the share of each party's identifiers who call themselves liberal, moderate or conservative. Among Republicans, 53%53% percent of the party's identifiers who placed themselvesEstimandweighted share in the conservative camp, Wilson intervalInterval95% CI 49% to 57%n595Weightsweight (VCF0009z)PopulationANES Republican identifiers (leaners included) who placed themselves on the liberal-conservative scalePeriod1972SourceANES Cumulative Data File chose a conservative label in 1972 and 77%77% percent of the party's identifiers who placed themselvesEstimandweighted share in the conservative camp, Wilson intervalInterval95% CI 75% to 80%n2,172Weightsweight (VCF0009z)PopulationANES Republican identifiers (leaners included) who placed themselves on the liberal-conservative scalePeriod2024SourceANES Cumulative Data File did in 2024, while the liberal share of Republicans fell. Among Democrats the liberal share rose from 36%36% percent of the party's identifiers who placed themselvesEstimandweighted share in the liberal camp, Wilson intervalInterval95% CI 33% to 40%n776Weightsweight (VCF0009z)PopulationANES Democrat identifiers (leaners included) who placed themselves on the liberal-conservative scalePeriod1972SourceANES Cumulative Data File to 62%62% percent of the party's identifiers who placed themselvesEstimandweighted share in the liberal camp, Wilson intervalInterval95% CI 59% to 65%n2,310Weightsweight (VCF0009z)PopulationANES Democrat identifiers (leaners included) who placed themselves on the liberal-conservative scalePeriod2024SourceANES Cumulative Data File, and the conservative share fell, as the opening numbers showed. The GSS shows the same direction on both sides. In the CES, which starts in 2006, the liberal share of Democrats rose and the conservative share of Republicans did not change detectably.

These shares describe people who placed themselves, and that group grew. In 1972, 57%57% percent of respondents with a party identificationEstimandweighted share who placed themselves on the liberal-conservative scale, Wilson intervalInterval95% CI 55% to 59%n2,695Weightsweight (VCF0009z)PopulationANES respondents with a party identification (other party and not sure excluded)Period1972SourceANES Cumulative Data File of ANES respondents with a party answer placed themselves on the scale; in 2024, 84%84% percent of respondents with a party identificationEstimandweighted share who placed themselves on the liberal-conservative scale, Wilson intervalInterval95% CI 82% to 85%n5,483Weightsweight (VCF0009z)PopulationANES respondents with a party identification (other party and not sure excluded)Period2024SourceANES Cumulative Data File did. The 1972 correlation therefore describes a smaller share of respondents than the 2024 one. A change in who answers cannot be the only source of the rise. In the GSS, 96%96% percent of respondents with a party identificationEstimandweighted share who placed themselves on the liberal-conservative scale, Wilson intervalInterval95% CI 95% to 97%n1,402Weightsweight_usePopulationGSS respondents with a party identification (other party and not sure excluded)Period1974SourceGeneral Social Survey of respondents with a party answer placed themselves in 1974 and 96%96% percent of respondents with a party identificationEstimandweighted share who placed themselves on the liberal-conservative scale, Wilson intervalInterval95% CI 94% to 96%n3,185Weightsweight_usePopulationGSS respondents with a party identification (other party and not sure excluded)Period2024SourceGeneral Social Survey in 2024, and its correlation rose all the same.

Web respondents show a higher GSS correlation

The GSS rise of the 2020s mixes two changes. Both the ANES and the GSS added web interviews, and a change in survey modeSurvey modeHow respondents were interviewed: in person, by phone, on paper or online. Mode can change answers, so a change in mode can look like a change in opinion. can move a correlation without any change in the public. Figure 3 splits the GSS by mode for the years that record it. Web respondents show a higher correlation than in-person respondents in 2022 and 2024, and the in-person reading for 2024 is level with the one for 2018, while the reading over all interviews climbs. The GSS rise of the 2020s therefore combines any change among Americans with a change in how the GSS interviewed them, and these data cannot separate the two. In the ANES, the web and face-to-face readings do not differ detectably in 2012, 2016 or 2024, and mode does not remove the level difference between the two surveys.

What the rise means, and what it does not

The rise is not the work of one group. In every birth cohortCohortPeople born in the same period, such as the same decade. Comparing cohorts shows whether a difference comes from when people were born, as distinct from their age or the year they were surveyed., education group and region the ANES can define, the correlation ends above where it starts. College graduates, who started highest, rose less than the other education groups, and the differences between groups are smaller in 2020 to 2024 than in 1980 to 1988. The link also shows up as distance between the parties' averages. The average Republican identifier placed themselves 0.90.9 points on the seven-point liberal-conservative scaleEstimandweighted mean self-placement of Republican identifiers minus that of Democratic identifiersInterval95% CI 0.7 to 1.0n1,371Weightsweight (VCF0009z)PopulationANES identifiers (leaners included; pure independents excluded) who placed themselvesPeriod1972SourceANES Cumulative Data File points to the right of the average Democratic identifier on the seven-point scale in 1972, and 2.42.4 points on the seven-point liberal-conservative scaleEstimandweighted mean self-placement of Republican identifiers minus that of Democratic identifiersInterval95% CI 2.3 to 2.5n4,482Weightsweight (VCF0009z)PopulationANES identifiers (leaners included; pure independents excluded) who placed themselvesPeriod2024SourceANES Cumulative Data File points to the right in 2024.

Both measures are labels. Agreement between the party label and the ideology label is not agreement on issues, and the public is more likely to call itself conservative than liberal while favoring more government action on many problems [1]. A rise in the correlation between party and label is therefore not a rise in the correlation between party and any set of policy positions. Planned, Constraint among issue positions (planned) asks whether issue positions themselves bundled more tightly. The correlation also says nothing about extremity.

What these surveys cannot tell you

The data are repeated cross-sections, so they show that party and self-placement became more tightly linked and do not show which of the two changed; a panel study that followed the same people found movement in both directions, as some changed party to match their positions and others changed positions to match their party [2]. The three surveys give different levels, and these data cannot say which is closer to the level in the population. The intervals treat each weighted sample as a simple random sample and do not follow the sampling design, and where that could be checked, in the GSS cells behind the mode comparison, the design-based standard errors were larger. The data hold no measure of what party leaders did, so they cannot say whether the electorate moved toward parties that had moved apart or the parties moved apart because the electorate had; Planned, Polarization in Congress (planned) takes up the parties in Congress.

Go deeper

• The report: every estimate with its interval and sample size.
• Tables: every figure of the report as a table.
• World values: the explorer for this part.

Reading paths that include this chapter:

• The core argument in eight chapters
• A tour of the book, one chapter per part

Data sources

• ANES Cumulative Data FileANES Time Series Cumulative Data File, 1972-2024: seven-point party identification and seven-point liberal-conservative scale; interview mode from the mode weights (2008, 2012, 2016, 2024)Years: 1948-2024 (33 study years; even years only after 2004; no 2006, 2010 or 2014 study) · Weights: weight (VCF0009z); weight_ftf and weight_web for interview mode · License: Free download after accepting the ANES terms of use; cite the release.
• General Social SurveyGeneral Social Survey, 1974-2024: partyid and polviews; interview mode joined from the raw file (from 2004)Years: 1972-2024 (35 study years) · Weights: weight_use · License: See the provider's terms of use.
• Cooperative Election StudyCooperative Election Study cumulative file, 2006-2024: pid7 and ideo5Years: 2006-2024 (every year; self-reported turnout is available in the 10 even years) · Weights: weight (single wave); weight_cumulative (pooled waves) · License: See the provider's terms of use.

All data sources

References

1. Ellis, Christopher; Stimson, James A. (2012). Ideology in America. Cambridge University Press.
2. Carsey, Thomas M.; Layman, Geoffrey C. (2006). Changing Sides or Changing Minds? Party Identification and Policy Preferences in the American Electorate. American Journal of Political Science 50(2): 464-477. doi:10.1111/j.1540-5907.2006.00196.x

How to cite

@misc{singh2026partyandideology,
  author = {Singh, Vishal},
  title = {Party and ideology},
  year = {2026},
  url = {https://vishalsingh.org/politics/party-and-ideology},
  note = {In The Public Mind. Published 30 September 2026.}
}