Tolerance of speech

In the General Social Survey (GSS) of 2008 to 2018, Republicans were more willing than Democrats to let a speaker the question describes as one "who believes that Blacks are genetically inferior" speak in their community. The shares were 63%63.0% percent who would allow the speaker to speakEstimandweighted share who would allow the racist speaker to speakn2,879Weightsweight_use (wtssnrps from 2004, wtssps before)PopulationRepublicans answering the racist item (leaners included)Period2008-2018 waves pooledSourceGeneral Social Survey of Republicans and 56%56.1% percent who would allow the speaker to speakEstimandweighted share who would allow the racist speaker to speakn4,175Weightsweight_use (wtssnrps from 2004, wtssps before)PopulationDemocrats answering the racist item (leaners included)Period2008-2018 waves pooledSourceGeneral Social Survey of Democrats.Democrats and Republicans include independents who say they lean toward a 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 this essay. The question is whether he should be allowed to speak, not whether the respondent agrees with him. On four of the other five speakers, the parties' answers in those years were close. Whether Democrats or Republicans allowed a speaker more often depended on the speaker.

The party gap depends on the speaker

The essay's labels for the other speakers are short names for the survey's own descriptions (Table 1). The chapter measures the difference as a party gap, Democrats minus Republicans, in pointsPercentage pointThe unit for the difference between two percentages. A change from one share to another is stated in points, not in percent, so that absolute and relative change are not confused.. A negative gap means that Republicans allowed the speaker more often, and the chapter calls it a Republican lead. On the racist speaker the gap in 2008 to 2018 was −6.8−6.8 points (first group minus second group)Estimanddifference in weighted shares who would allow the speaker to speak in the community (Democrats minus Republicans)Interval95% CI −9.7 to −4.0n7,054Weightsweight_use (wtssnrps from 2004, wtssps before)PopulationDemocrats and Republicans (leaners included) answering the itemPeriod2008-2018 waves pooledSourceGeneral Social Survey points, the largest in size of the six, though the interval on its difference from the militarist gap includes zero. The gaps for the Communist, the anti-religionist, the homosexual man and the Muslim clergyman, the four speakers the chapter codes as objectionable to conservatives, had point estimates within 2.12.1 pointsEstimandlargest absolute party gap among the Communist, anti-religionist, homosexual man and Muslim clergyman itemsWeightsweight_use (wtssnrps from 2004, wtssps before)PopulationDemocrats and Republicans (leaners included) answering the itemPeriod2008-2018 waves pooledSourceGeneral Social Survey points of zero. The militarist, like the racist, kept a Republican lead.

In 1976 to 1994 the party gap was negative for every speaker asked, so Republicans allowed each one more often than Democrats did. Figure 1 sets the two periods side by side. The leads that shrank are the leads on the speakers that conservatives are expected to object to. Ideology separates the sides on those four speakers. In 2008 to 2018 self-placed liberals allowed them 7.27.2 pointsEstimandsmallest ideology gap among the Communist, anti-religionist, homosexual man and Muslim clergyman itemsWeightsweight_use (wtssnrps from 2004, wtssps before)Populationliberals and conservatives (self-placed) answering the itemPeriod2008-2018 waves pooledSourceGeneral Social Survey to 13.713.7 pointsEstimandlargest ideology gap among the Communist, anti-religionist, homosexual man and Muslim clergyman itemsWeightsweight_use (wtssnrps from 2004, wtssps before)Populationliberals and conservatives (self-placed) answering the itemPeriod2008-2018 waves pooledSourceGeneral Social Survey points more often than conservatives did, and for each of the four the party gap is smaller than the ideology gap.

How the GSS measures tolerance of speech

The battery descends from Stouffer's national survey of 1954 [1] and keeps its wording across survey years. The descriptions of the speakers are the questionnaire's, dating from the 1970s except for the Muslim clergyman, added in 2008, and phrases such as "a man who admits that he is a homosexual" and "a man who admits he is a Communist" are the survey's language. The GSS has asked national samples of American adults the speaker questions since 1972, through interviewers, mostly in person, until 2021. In all, about 45,00044,878 respondentsEstimandcount of respondentsWeightsnone (count)Populationrespondents who answered at least one speak itemPeriod1972-2024SourceGeneral Social Survey respondents answered at least one of the six speak items between 1972 and 2024. Each share 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 independents and moderates drop out of the group comparisons. Ideology uses the seven-point self-placement, with liberals at the three left points and conservatives at the three right points. The qualification that matters most is 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.. In 2021 the GSS moved mostly online, and 86%86% percent of respondentsEstimandshare interviewed on the web (unweighted count)n1,310Weightsnone (count)Populationrespondents who answered the racist speak itemPeriod2021SourceGeneral Social Survey of the respondents who answered the racist item that year were interviewed on the web; 2022 and 2024 mix web and interviewer-administered respondents. Mode can change answers to sensitive questions [2], so estimates from 2021 on carry a mode label. The report gives every estimate with its interval and sample size.

Web respondents allowed the racist speaker less often

Figure 2 follows the party gap on the racist speaker one survey year at a time. The estimate is below zero in every survey year since 1976 but 2008, and it is farthest below zero in 2021, at −22.9−22.9 points (first group minus second group)Estimanddifference in weighted shares who would allow the speaker to speak in the community (Democrats minus Republicans)Interval95% CI −30.3 to −15.4n990Weightsweight_use (wtssnrps from 2004, wtssps before)PopulationDemocrats and Republicans (leaners included) answering the itemPeriod2021 waveSourceGeneral Social Survey points. In 2022 and 2024, when the GSS used both modes, 41%40.7% percent who would allow the speaker to speakEstimandweighted share who would allow the speechInterval95% CI 36.5% to 44.9%n1,153Weightsweight_use (wtssnrps from 2004, wtssps before)Populationadults interviewed on the web (Racist)Period2022 and 2024SourceGeneral Social Survey of web respondents would allow the racist speaker and 55%54.7% percent who would allow the speaker to speakEstimandweighted share who would allow the speechInterval95% CI 50.0% to 59.4%n1,118Weightsweight_use (wtssnrps from 2004, wtssps before)Populationadults interviewed by an interviewer (Racist)Period2022 and 2024SourceGeneral Social Survey of respondents interviewed by an interviewer would. The difference is similar among Democrats and among Republicans. For the Communist, the anti-religionist and the Muslim clergyman, the mode differences in the share allowing the speech have intervals that include zero.

The mode difference the data can detect is in the share allowing the racist speaker, not in the size of the party gap. The gap is negative in both modes, and the difference between the two modes has an interval that includes zero. Pooled across modes, the 2022 to 2024 gap differs from the 2008 to 2018 gap by −7.3 (95% CI −15.5 to 0.9)−7.3 points (2022-2024 party gap minus 2008-2018 party gap)Estimanddifference between the pooled 2022-2024 and the 2008-2018 racist party gaps, all interview modes (independent samples)Interval95% CI −15.5 to 0.9Weightsweight_use (wtssnrps from 2004, wtssps before)PopulationDemocrats and Republicans (leaners included) answering the itemPeriod2022 and 2024 against 2008-2018SourceGeneral Social Survey points, an interval that includes zero. Every estimate for 2021 to 2024 is more negative than the 2008 to 2018 gap, but only the 2021 wave on its own separates from it, and the data do not separate mode from period. Nor can they say whether web respondents answer differently or are different people.

The exception year, 2008, opens the 2008 to 2018 window, and in that wave alone every party gap except the militarist's was positive. With 2008 included, the change in the racist gap between the two periods has an interval that includes zero. Without it, the 2010 to 2018 gap differs from the 1976 to 1994 gap by −3.9 (95% CI −7.5 to −0.3)−3.9 points (2010-2018 gap minus 1976-1994 gap)Estimandchange between the two periods in the party gap, 2008 wave left out (independent samples)Interval95% CI −7.5 to −0.3Weightsweight_use (wtssnrps from 2004, wtssps before)PopulationDemocrats and Republicans (leaners included) answering the itemPeriod1976-1994 to 2010-2018SourceGeneral Social Survey points, an interval that excludes zero. The persistence of the racist and militarist leads does not depend on that wave; whether the racist lead grew does.

What the gaps mean, and what they do not

The gaps describe who would let a speaker speak, and the answer moves with the speaker. The militarist item asks about speech, not about support for army rule, which Chapter 5.3, Support for democracy takes up, and willingness to let a speaker speak in principle is not the same as behavior toward a speaker. The raw party gaps also mix party with the traits that differ between the parties. Among people of the same age, education, race and sex in the same survey year, Democrats allowed the four speakers objectionable to conservatives more often than Republicans did in 2008 to 2018. Each of those adjusted gaps has an interval above zero, and the adjusted racist gap stayed negative, at −3.9−3.9 points (first group minus second group)Estimanddifference in the share who would allow the speaker to speak, Democrats minus Republicans, adjusted for age, education, race, sex and survey year (weighted linear probability model, HC3 errors)Interval95% CI −6.9 to −0.8n7,037Weightsweight_use (wtssnrps from 2004, wtssps before)PopulationDemocrats and Republicans (leaners included) answering the itemPeriod2008-2018 waves pooledSourceGeneral Social Survey points. The racist party gap is negative on all three actions the battery asks about, speaking in the community, teaching in a college and keeping a book in the public library.

A single average of which side allows more depends on which speakers are counted. Figure 3 averages each respondent's answers over the five speakers asked in every period, split into the two families. The party gap for the family that conservatives are expected to object to went from a Republican lead in 1976 to 1994 to an interval that includes zero in 2008 to 2018. The gap for the racist and the militarist stayed negative, at −5.2−5.2 points (first group minus second group)Estimanddifference between groups in the mean share of the family's speakers a respondent would allow (Democrats minus Republicans)Interval95% CI −7.6 to −2.9n6,991Weightsweight_use (wtssnrps from 2004, wtssps before)PopulationDemocrats and Republicans (leaners included) answering the itemPeriod2008-2018 waves pooledSourceGeneral Social Survey points in 2008 to 2018, and was farthest below zero in the mostly web 2021 wave. Whether a pooled party gap over all five speakers has an interval that excludes zero depends on whether each speaker or each family counts once. One account links conservatism to needs for order and certainty [3]; another holds that liberals and conservatives are each intolerant of the groups that threaten their values [4] [5]. The battery cannot test the dispositions behind either account.

What these answers cannot tell you

The gaps are differences between groups in one survey question, so they do not show why the groups differ, whether party or ideology shapes tolerance, or whether tolerance shapes party or ideology. After 2020 the mode of interview is tied to time, and the 2022 and 2024 samples are small once cut by party or ideology, so the size of the racist gap after 2020 is uncertain. The battery asks about a fixed list of speakers, and grouping them by which side is expected to object is a coding choice. The racist and Muslim clergyman items describe the speech itself, so the levels of the items are not comparable with one another.

Go deeper

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

Reading paths that include this chapter:

• Trust and democracy

Data sources

• General Social SurveyGeneral Social Survey cumulative file 1972-2024, allow-to-speak items (spk), college-teaching items (col) and library-book items (lib); interview mode from the raw fileYears: 1972-2024 (35 study years) · Weights: weight_use (wtssnrps from 2004, wtssps before) · License: See the provider's terms of use.

All data sources

References

1. Stouffer, Samuel A. (1955). Communism, Conformity, and Civil Liberties: A Cross-Section of the Nation Speaks Its Mind. Doubleday.
2. Tourangeau, Roger; Yan, Ting (2007). Sensitive Questions in Surveys. Psychological Bulletin 133(5): 859-883. doi:10.1037/0033-2909.133.5.859
3. Jost, John T.; Glaser, Jack; Kruglanski, Arie W.; Sulloway, Frank J. (2003). Political Conservatism as Motivated Social Cognition. Psychological Bulletin 129(3): 339-375. doi:10.1037/0033-2909.129.3.339
4. Brandt, Mark J.; Reyna, Christine; Chambers, John R.; Crawford, Jarret T.; Wetherell, Geoffrey (2014). The Ideological-Conflict Hypothesis: Intolerance Among Both Liberals and Conservatives. Current Directions in Psychological Science 23(1): 27-34. doi:10.1177/0963721413510932
5. Crawford, Jarret T.; Pilanski, Jane M. (2014). Political Intolerance, Right and Left. Political Psychology 35(6): 841-851. doi:10.1111/j.1467-9221.2012.00926.x

How to cite

@misc{singh2026toleranceofspeech,
  author = {Singh, Vishal},
  title = {Tolerance of speech},
  year = {2026},
  url = {https://vishalsingh.org/politics/tolerance-of-speech},
  note = {In The Public Mind. Published 30 September 2026.}
}