Abstract
“Cancel culture” has become a prominent phrase in US political commentary, with supporters and opponents relying on different value-based arguments to express their views. At the same time, these camps often fall along predictable partisan political lines. What, then, are the real motivations for promoting and opposing behaviors sometimes labeled as “canceling” in contemporary American politics? We explore this question through survey data from Pew in 2020 and two original survey experiments, conducted on the 2021 CES and a 2023 online sample. We examine how Americans define the term “cancel culture,” to what extent cancel culture is linked to both a range of core democratic values, and the role of partisanship in shaping support or opposition to specific behaviors. We observe a significant range of views about cancel culture and document connections to various political values. From our experiments, we find that partisanship is a potent driver of support for and understanding of cancel culture and that value-based framing has a weaker impact as compared to partisanship. Cancel culture provides an important case study of how Americans process conflicting norms and values, including free speech and political accountability, beyond the typical constraints of the formal political system.
Introduction
In 2019, comedian Shane Gillis was fired from Saturday Night Live over racist and homophobic slurs he used on a podcast the year before. Some bemoaned the firing as hypocritical, rash, and unfair, attributing it to the phenomenon of “cancel culture.” Others, however, disagreed and supported Gillis’s firing as appropriate and holding him accountable for his public misbehavior (Asmelash 2019; Garvey 2019). These disagreements extend far beyond comedians, touching on Dr. Seuss, politicians, free speech, and even Socrates (Knowles and Bella 2021; Thomas 2020; Wallace-Wells 2021). Americans are increasingly aware of these discussions and debates: 61 percent of Americans in 2022 reported they have heard a fair amount or more about cancel culture, substantially up from 44 percent in 2020 (Vogels 2022).
What does it mean to be canceled, and what role does cancel culture play in the current partisan American climate? Both supporters and opponents of “canceling” frame their claims by referencing fundamental democratic values. Those who call for canceling frame it as holding people publicly accountable for their words and actions or providing political voice to groups that historically have less access to formal political representation. Opponents of canceling often position canceling as a threat to free speech or unjust punishment outside of due process. Simultaneously, cancel culture has taken on a partisan dimension, as Republican politicians have made opposition to cancel culture a public rallying cry and affective polarization encourages members of both parties to protect co-partisans while censuring or punishing the out-party.
Are views of cancel culture rooted to the democratic values around which the debate is often framed, or are they no more than ways to signal partisan loyalties? To answer this question, we build on political psychology research about how partisanship and democratic value frames shape people’s policy preferences and tolerance for the democratic rights of other citizens (Hsiao and Radnitz 2021; Nelson, Clawson, and Oxley 1997). Additionally, affective polarization is strongly related to support for out-party rights, moral treatment of the out-party, and willingness to compromise. (Barber and Davis 2022; Iyengar and Krupenkin 2018; Iyengar, Sood, and Lelkes 2012). However, recent scholarship suggests that ideological disagreements and fundamental democratic values may be at the root of the partisan divide (Costa 2021; Rogowski and Sutherland 2016; Webster and Abramowitz 2017). Frames about these key policy and political ideas can be very influential on public opinion and behavior, especially when such frames link with important and influential values (Aarøe 2011; Chong and Druckman 2007, 2013). As a partisan touchstone of growing importance, cancel culture provides an ideal case study to evaluate the relative import of both people’s commitment to fundamental democratic values and their partisan attachments.
In this paper, we use multiple sources of data to examine Americans’ definitions of cancel culture, how those views shift in response to values-based frames and the partisan conditions, and how the actions citizens support do or do not align with more abstract notions of canceling. We find that Americans have a range of definitions of cancel culture, with both positive and negative connotations. We find that behaviors linked to the term “canceling” are primarily viewed as consisting of two separate dimensions of support-based and punitive actions. Additionally, we find that both the fundamental definition and the endorsement of specific canceling behavior are dependent on the perceived partisan implications of those actions. While democratic values and corresponding frames are a part of the rhetoric of cancel culture, we find that they have a much weaker impact on people’s support for canceling in a given scenario.
Theoretical Framework
The term “cancel culture” comes from 1980’s slang, where “cancel” referred to breaking up with someone (Schaeffer 2021). However, the political roots of canceling begin with Black Twitter. The earliest appearances of the use of the term canceling are from Black women using it as an effort for justice, allowing disadvantaged minorities an opportunity to give the problem attention and prescribe a solution (Clark 2020; Ng 2022). One early example occurred in 2013 when celebrity chef Paula Deen’s public use of the n-word in response to the Trayvon Martin murder case triggered a hashtag storm (#BBQBecky, #PoolPatrolPaula) on Black Twitter. In circumstances like this, marginalized groups can use social media platforms like Twitter to put ordinary people’s racist and offensive actions into the news, allowing minorities to express frustration and set rules for social acceptance (Clark 2020).
Since then, cancel culture has expanded; however, there is no universally accepted definition of the term (Lopes 2022). In general, the literature agrees that cancel culture is a phenomenon where a group withdraws support from someone for doing something offensive (Bouvier 2020; Clark 2020; Ng 2020; Norris 2021; Tandoc et al., 2022). However, significant disagreement underlies this conceptualization. Some contend that canceling must happen on social media, while others include non-social media actions like public shaming and boycotts (Billingham and Parr 2020; Clark 2020; Fahey, Roberts, and Utych 2022; Lewis and Christin 2022; Mueller 2021; Pearson 2021; Schaeffer 2021). Most definitions limit the scope to punishing behaviors, although sometimes the term includes proactive positive support—such as intentionally buying a product that supports one’s views (buycotting).
The behaviors often included in the scope of canceling and cancel culture are not in themselves novel. They are, in many ways, a rebranding of long-running patterns of behaviors among citizens that use consumer behavior to express their exit, voice, or loyalty with regards to their political views (Hirschman 1970). For example, research on political consumerism finds that boycotting and buycotting are undertaken by Americans as a way of expressing their political views (Endres and Panagopoulos 2017), that they are primarily expressive and not an instrumental political behavior (Copeland and Boulianne 2022), and that both have distinctive motivations (Kam and Deichert 2020).
Does cancel culture represent a distinctive set of behaviors and political or psychological processes from other forms of political consumerism or social media use? Regardless of the target or the specific actions, popular culture (Knowles and Bella 2021; Mishan 2020; Romano 2019) and academic literature (Cook et al., 2021; Dias, Druckman, and Levendusky, n.d.; Norris 2021) often portrays “cancel culture” as negative. This fits with many discussions of canceling and cancel culture in the media, especially in conservative circles. In this perspective, cancel culture may be nothing more than a rhetorical branding used by political actors seeking to avoid the effects of, or countermobilize against, political consumerist behaviors.
We identify three separate potential bases for individual support or opposition to cancel culture. After introducing these perspectives, we provide a set of research questions and hypotheses related to the expected effect and relative importance of each. More broadly, because it is clearly linked to partisan and non-partisan cues, cancel culture attitudes provide an excellent case study into how various considerations shape public opinion about tolerance and democratic attitudes in the contemporary United States.
First, different ways of framing cancel culture can link it to a range of core democratic values and processes, which can influence the public to prioritize those democratic elements. We build on research showing that conflicts over core values undergird much of the disagreement in American politics (Marietta and Barker 2019), that values are strong drivers of political beliefs (Feldman 2003; Schwartz et al., 2014), and that value-based framing can powerfully impact evaluations of the behavior and speech of others and narrow the ideas people use to make political judgments (Barker 2005; Brewer 2002; Brewer and Gross 2005; Nelson, Clawson, and Oxley 1997). While not intended to be a comprehensive accounting of all potential value framing, for this research we identify four values linked to frames about cancel culture—two that we expect would increase support for canceling and two that would reduce support for canceling.
Drawing from its roots in Black Twitter, many supporters promote using online platforms or other economic choices to punish elites because minority groups with little access or power over formal political processes have few alternative mechanisms for political influence (Clark 2020; Klein 2022; Tandoc et al., 2022). Additionally, survey research finds that many people view canceling as a way to hold private actors (elite and non-elite) accountable for their words and actions (Atske 2021). On the other side, the primary objection to cancel culture is that it is a limitation of free speech, a form of censorship and ideological control by forcing a person out of the public arena because of political disagreement (Dias, Druckman, and Levendusky, n.d.; Waisbord 2020; Loury 1994). Some opponents express a slightly more narrow and process-based objection, where they oppose canceling specifically because people receive a punishment, such as losing their job, based on “mob rule” and without the benefit of due process protections (Atske 2021; Mueller 2021; Pipyrou 2018). These four values—empowerment, accountability, free speech, and due process—are potentially important frames for thinking about and influencing support for cancel culture in the American public.
Our second basis for understanding ideas about and support for cancel culture connects to parties and partisan cues. As a particular term, “cancel culture” has been politicized in ways that might lead to specific party ownership of the term and strong party cues. Specifically, Republicans may be more likely to weaponize the phrase relative to Democrats (Atske 2021; Fahey, Roberts, and Utych 2022). News organizations have documented how conservative pushback against “political correctness” adapted into the cancel culture debate (Shapiro et al., 2021), until it became a major talking point for Republicans broadly (Ng 2022). In experimental work, scholars find that priming concerns about political correctness or speech restriction increases support of Donald Trump (Conway, Repke, and Houck 2017). This would suggest that Republicans should view cancel culture more negatively and hold more opposition to cancel culture or its associated behaviors, while Democrats should be more supportive.
A final way of thinking about cancel culture in the U.S. approaches partisanship differently. Rather than identifying a partisan asymmetry, partisanship might identify in-groups and out-groups, such that self-interested group members are more likely to punish out-party members and reward those who share their partisan identity (Cohen 2003; Zell, Stockus, and Bernstein 2021). People’s partisan attachments strongly influence their views on specific policies and social topics, far beyond the realm of explicitly political subjects (Huber and Malhotra 2016; Shafranek 2019). Much like Barber and Pope’s (2019) finding that party cues overwhelm ideological policy commitments, pundits on both sides revel in pointing out the hypocrisy of parties that take public positions for or against cancel culture and then directly contradict that position in practice (Porter n.d.; Romano 2021). Consistent with research on partisanship and political tolerance (Argyle, Terman, and Nelimarkka 2022; Barber and Davis 2022; Hsiao and Radnitz 2021), this may be one manifestation of how affective polarization can erode democratic norms (Finkel et al., 2020; Kingzette et al., 2021). In this perspective, for both Republicans and Democrats, stated values are justifications for taking whatever position is most advantageous to one’s own party.
Research Questions and Hypothesis
Given the varied definitions of cancel culture in the research literature and media, we begin by examining how Americans define cancel culture. We then use two survey experiments to evaluate how the definition of and support for cancel culture depend on the value framing, the respondent’s partisanship, and the partisan context of a hypothetical scenario. We include the definition of cancel culture as an experimental outcome, on the theory that one way people can reduce cognitive dissonance when abstract partisan or democratic values frames conflict with their self-interest in a particular situation might be to change the very definition of cancel culture. In this way, it is possible for people to remain opposed to cancel culture in general, but to redefine the term such that particular behaviors are no longer seen as “canceling” and can therefore be supported.
We begin with a set of ideas about what Americans’ definitions of cancel culture might contain. Existing research gives some guidance about what these definitions might mention, but not in enough detail to warrant a precise hypothesis. As a result, we focus on research questions here rather than specific predictions. Our evaluation of these research questions largely falls within the descriptive, open-ended survey data in the analyses that follow.
In our first research question, we explore the degree to which definitions of cancel culture make reference to the democratic values that are often used to frame cancel culture in the contemporary United States. We emphasize four ideas that are prominent in discussions of cancel culture, as noted earlier: increased voice for minorities, accountability, freedom of speech, and due process. RQ1. To what extent do Americans’ definitions of cancel culture reference the democratic values frames of giving voice to minorities, accountability, freedom of speech, and due process?
Our second research question considers whether and how ideas about cancel culture center on partisanship and political disagreement. RQ2. To what extent and in what manner do Americans’ definitions of cancel culture reference specific political parties (e.g., Republicans and Democrats)?
Based on prior literature, we expect that both partisanship and values will be connected to public ideas about cancel culture, but we do not have a priori expectations of which of these will be more salient in practice. RQ3 proposes exploring the relative frequency of the values and partisan dimensions uncovered in RQ1 and RQ2. RQ3. What is the relative weight of partisanship and political values on public opinion about cancel culture?
Additionally, we acknowledge that Americans’ ideas about cancel culture likely contain a range of content that goes beyond just references to partisanship and these specific political values. As existing work does not articulate the set of other things that could or should be a part of Americans’ views of cancel culture, we propose exploring the additional content expressed in definitions of cancel culture in a more exploratory way, as suggested by RQ4. RQ4. What other types of content are contained in the public’s understanding of cancel culture?
In addition to these four research questions, we also approach this research with three hypotheses about factors that shift Americans’ ideas about what cancel culture involves and influences their support for different canceling behaviors. The first is that we expect different frames referencing various political values will shift support for behaviors connected to canceling. Values frames can push respondents towards a specific position on a political issue and focus people on that frame to the exclusion of other competing values (Brewer and Gross 2005). In the case of cancel culture, frames referencing the values of empowering minorities should increase support for canceling behavior, and frames focusing on freedom of speech should decrease support for canceling. This expectation is rooted in the idea that frames about empowerment will make considerations about the protection of vulnerable people more salient in respondents’ minds and align with media coverage in support for canceling behavior, encouraging more support for activities that fall under the umbrella of canceling. Freedom of speech frames, on the other hand, will do the opposite—bringing to mind ideas about protecting even offensive and problematic political speech (e.g., Nelson, Clawson, and Oxley 1997) and fall in line with larger narratives discouraging actions associated with canceling. This prediction is summarized in H1.
When potential canceling is framed as giving minority groups political voice, respondents will be more supportive of behaviors associated with cancel culture than when canceling is framed around protecting free speech. Second, we expect differences between Democrats and Republicans in responding to situations involving canceling. These patterns could be due to a variety of different factors and mechanisms. As noted earlier, the history and current rhetoric around cancel culture in American politics often positions cancel culture as something enacted and supported by Democrats and primarily targeted at and opposed by Republicans. In addition, some scholars find that the Republican Party is more connected by political identity than Democrats (e.g., Grossmann and Hopkins 2016) and others document more mobilization among Republicans in response to social and culturally divisive issues (Deshpande et al., 2023). As a result of these factors, we expect Republicans to be less supportive of cancel culture generally and more sensitive to canceling when they are targeted at their own group. These predictions are described in H2 and H3.
Republicans will be less supportive of canceling behaviors than Democrats.
Republicans will be more responsive to in-party and out-party cues in situations involving canceling than Democrats. In contrast to H2 and H3, it may be the case that partisans on both sides of the political spectrum are willing to engage in canceling, depending on the partisan identity of the entity being canceled. This is consistent with larger research in political science about the role of group identities and intergroup conflict with respect to partisanship, as noted in the previous section (Cohen 2003; Goren, Federico, and Kittilson 2009; Munro, Weih, and Tsai 2010; Zell, Stockus, and Bernstein 2021). Applied to our specific research, we expect a more expansive definition of canceling and less support for canceling behaviors when a co-partisan is the target than when an out-partisan is the target (H4).
Respondents will have a more expansive definition of canceling behaviors and be less supportive of canceling when a co-partisan is the target in a canceling situation relative to when an out-partisan is the target.
Data
To explore ideas about cancel culture and test our hypotheses, we use multiple sources of survey data. First, we use open-ended survey responses from the Pew Research Center’s American Trends Panel in September 2020. Second, we use two original survey experiments. We explain how each speaks to different aspects of our predictions below.
The Pew Research Center recruits participants into the American Trends Panel based on nation-wide random address sampling to complete surveys online. In the September 2020 survey wave, Pew asked respondents to first indicate if they had heard the term “cancel culture.” The 4,220 respondents who had heard of cancel culture were then asked: “In your own words, what do you think cancel culture means?” We analyze the text of these open-ended responses to evaluate how Americans’ think about cancel culture in their own words, and test our definitions-related research questions. 1
To consider the influence of the different considerations related to cancel culture, we conducted two survey experiments. Both experiments have ∼1,000 respondents; the first is conducted as part of a team module on the 2021 CES administered by YouGov and matched to national demographic distributions, and the second is conducted on a sample of Prolific respondents in 2023 which are balanced using quotas to national proportions of age, gender, and partisanship.
In both surveys, respondents are initially asked a number of questions about their own demographic backgrounds and political views. The experiment then presented respondents with a vignette that described a person named Casey 2 whose circumstances led them to consider engaging in behaviors that could be considered canceling. The CES experiment asks about a restaurant owner who “donated large sums of money to {Democratic/Republican/no party label} state legislators who are sponsoring a controversial bill that Casey finds offensive and problematic” with value treatments on free speech or empowerment (full text in the Online Appendix). In the Prolific experiment, the situation is about a “true crime podcast that has become incredibly popular nationally,” but Casey “just found out that the podcaster has also been actively promoting a controversial bill on social media, including making rude comments about the bill’s opponents.” We include both to evaluate canceling as it occurs in different context and with different levels of prominence on platforms like social media. In both cases, the vignette includes a fully crossed 3 (partisanship) × 2 (values frame) experimental design.
These vignettes describe a circumstance where there may be some grounds for canceling but where the behavior is not so severe as to make the choice to cancel obvious and nearly universal. We selected this kind of situation as a place where value framing and party treatments might have an impact; in other scenarios, where the behavior is either too egregious or not at all controversial, these kinds of treatments are less likely to be consequential. 3 In the partisanship manipulation, there are three conditions: (1) Casey is a Democrat and the state legislators are Republicans, (2) Casey is a Republican and the state legislators are Democrats, or (3) a control in which Casey is a citizen and the state legislators do not have any listed party. In the values frames treatments, respondents either see a justification that opposes canceling on the basis of free speech or a justification that supports canceling on the basis of non-traditional representation for marginalized groups. 4 Based on the respondent’s self-reported party affiliation, 5 we code the treatment for analysis based on whether Casey is a co-partisan with the survey respondent (who is canceling an out-partisan) or whether Casey is an out-partisan (who is canceling a co-partisan), or has no party label (no-party control). This approach allows us to test H3 and H4. The value frame treatments allow us to test our predictions in H1. We also present analyses separately by Democratic and Republican subgroups of respondents, to evaluate the effect of party positioning expressed in H2.
After the vignette, respondents were asked whether a good citizen in Casey’s situation should take a series of eight actions asked individually with a randomized order (wording can be found in the Online Appendix—respondents indicated yes or no for each). These items can be seen in the results section; some actions are punitive, whereas others support the restaurant owner/podcaster. Following this question about what Casey should do, respondents indicated if they felt each of these same eight actions should be considered “canceling.” We solicited these evaluations to better understand the scope of behaviors people include in their definition of canceling, and how those ideas vary based on context.
Results
We first consider the definitions Americans have of cancel culture. We begin with unsupervised, automated text analysis of the Pew survey data. We use structural topic modeling (STM) to evaluate the themes and topics that emerge from the text responses. STM is a well-established, unsupervised method of analyzing texts that allows researchers to explore the content of those texts with few assumptions (Roberts et al., 2014). As such, it allows us to consider other ideas in these definitions of cancel culture that go beyond the specific themes and values we have already identified. More details can be found in the Online Appendix.
STM Results.
Addressing RQ1, the topics that emerge using STM include three of the four democratic values that we identified. Specifically, topic 5 addresses due process and mob rule concerns, topic 7 is about accountability, and topic 8 (and topic 9 to a lesser extent) articulates threats to freedom of speech. The value that is missing from the STM analysis is giving political voice to minority groups. The central idea of using cancel culture to withdraw support from individuals with objectionable views is contained in topic 3, but there is not clear mention of underrepresented or minority groups in any topic.
Interestingly, we find minimal support for the idea that respondents explicitly discuss parties in their definitions of canceling. We do not observe any specific discussion of political parties or partisans in the defining words for any topic (RQ2); the closest element is the mention of liberals and conservatives in topic 1, but those terms seem to reflect disagreements between factions more than one political side having ownership over or a defined position on the term. When we specifically code for mentions of the two main political parties (in a dictionary-based way outside of STM), we see only very rare mentions of the parties in these definitions. 6 To be clear, this finding indicates that people do not spontaneously bring up the names of the political parties in their definition of cancel culture; people may still evaluate specific instances of canceling through a lens of partisanship (a point we turn directly to in the survey experiment that follows).
There are several other themes that emerge from the STM analysis, which speaks to RQ4. Mentions of the financial and economic costs of cancel culture occur in both topics 2 and 4. This indicates that at least some people recognize it as something with consequences outside of the political system. Many of the topics identify themes of punishment, losing support, or silencing disagreement, which is consistent with how cancel culture is defined in academic work.
Survey Experiment
Respondents may not explicitly name parties in their definitions of cancel culture but use partisanship to evaluate which specific cases of canceling are justified and which are not. We directly examine the role of both value framing and partisanship in the survey experiments. Following the previously described vignettes about Casey, respondents were asked first if a good citizen would take each of eight actions given the hypothetical scenario, followed by identifying which of those behaviors count as canceling. As noted earlier, all of these items are binary, with a yes (coded as 1) or no (coded as 0) response for each action.
Without respect to treatment condition, more than 60 percent of the CES sample and 59 percent of the Prolific sample felt that calling for a boycott and posting negative comments on social media counted as cancel culture (see the Online Appendix for full distributions). Interestingly, these two items have the least support in the question about what a good citizen should do in the situation at hand—less than 25 percent of the CES sample and the Prolific sample felt that good citizens should call for a boycott or post a negative comment on social media. As expected from the negative rhetoric around canceling, there seems to be a gap in many people’s minds between what canceling entails and what people ought to do. The opposite was true of behaviors people felt did not count as canceling. Only 25 percent of the CES sample and 8 percent of the Prolific sample felt that contacting the restaurant owner counted as canceling, but more than 70 percent of both the CES sample and the Prolific sample felt that this was what a good citizen would do in Casey’s situation. This juxtaposition suggests that most Americans view canceling behaviors as antithetical to democratic citizenship.
Factor Analysis.
Note: Bolded items indicate questions that load well on the respective factor.
As shown in Table 2, the factor analysis suggests that these actions involve two distinct underlying factors in people’s minds—the amount of supportive behavior and a more actively punitive element. The questions about continuing to eat at the restaurant/listen to the podcast, eating at the restaurant/listening to the podcast more frequently, posting positive reviews on social media, and encouraging more people to eat at the restaurant/listen to the podcast are all closely related to each other in the first factor. In the second factor, people link stopping eating at the restaurant/listening to the podcast, posting negative reviews on social media, and calling for a widespread boycott of the restaurant/podcaster as a separate punishing dimension. This suggests that public outcry or shaming and boycotting behaviors are both connected to cancel culture in respondents’ minds. These patterns are linked as well to the Pew data, where respondents discuss both boycotting as well as being critical of public figures.
When thinking about canceling, people both reduce supportive behaviors and engage in active punishment, and they generally do not think good citizens should engage in canceling behaviors. 7 The distinctness of these factors—as opposed to all loading on the same factor with negative loadings, for example—indicates that support and punishing are not exact inverses of each other and that we ought to group the two sets of behaviors separately. 8 We integrate these factor analysis results into the experimental section that follows and analyze “punishing behaviors” (Factor 2) separately from “supportive behaviors” (Factor 1).
To more fully consider how much attitudes about cancel culture are driven or influenced by partisanship or frames relating to larger political values, we turn to the treatments in the survey experiments. As noted earlier, recall that this was a vignette-based experiment asking subjects to consider a situation where an individual (Casey) had objections to the political behavior of a local restaurant owner or podcaster. The experiments randomly interjected (1) value frame treatments emphasizing empowerment or freedom of speech and (2) partisan cues. Given the results of the factor analyses, we combined the larger batteries of items into four indices that become our four dependent variables—(1) a support index for the good citizen perceptions; (2) a punishing index for good citizen perceptions; (3) a support index for definitions of cancel culture; and (4) a punishing index for definitions of cancel culture. The four indices are minimally correlated (see Online Appendix). For the punishing indices, a positive effect is an indicator of more endorsement of canceling; for the support behaviors, a negative effect is an indicator of less support (and presumably more implicit endorsement of canceling).
Value Frame Treatment Effects.
Note: Significance in Comparison to Freedom of Speech Value, estimated with OLS
*p < 0.05, **p < 0.01, p < 0.001.
We find some support for H1 in the 2021 data, but none in the 2023 data. As shown in the first two models for the CES data, the value frame treatment has the expected effects on increased normative endorsement of punishing behaviors and reduced endorsement of supportive behaviors. Respondents who received this treatment were more likely to say that a good citizen should punish the restaurant owner/podcaster and that a good citizen should reduce their support for the restaurant owner/podcaster. Both of these effects are relatively small, and, coupled with the lack of an effect in 2023, suggest only weak and inconsistent support for H1.
At the same time, these treatments did not influence respondents’ larger notions of what cancel culture is in either experiment. We see no parallel effects on the cancel culture definition questions, either on the support or punishing dimensions. Furthermore, we do not see evidence of an interaction between the value frame and the party treatments in either the 2021 or 2023 experiments (see Tables A5 and A6 in the Online Appendix).
We next consider the connection between partisan attachments and perceptions of cancel culture (addressing H2, H3, and H4). To consider H2, that Republicans will be less supportive of canceling than Democrats, we consider how much partisans approve of supporting and punishing behaviors in these situations. These findings are summarized in Figure 1, broken out by the party-based treatments. Support for punishing and supporting behaviors by party and treatment. Notes: Bars represent unadjusted mean values for each party by experimental treatment combination. 95% confidence intervals.
This figure illustrates consistent partisan differences. With respect to respondents’ perceptions of what counts as canceling, we observe that Republicans are more likely to count punitive actions as canceling than Democrats in both years, although the size and significance of these differences varies by treatment condition. When reporting what a good citizen ought to do in these circumstances, Democrats are more likely to approve of punishing actions than Republicans in both years—this is the case for every year-treatment combination except for when Casey is an out-party member (and the respondents’ party is therefore being canceled) in 2023. Republicans in both years, on the other hand, are generally more likely to indicate that a good citizen should support the person under consideration than Democrats (again, we observe this except when Casey is an out-partisan in 2023). We take this as some support for H2 and evidence that Republicans are generally less supportive of canceling behaviors than Democrats.
To evaluate H3 and H4, we compare our dependent variables when party cues suggest that an out-party restaurant owner will be canceled, when an in-party restaurant owner will be canceled, and when partisan cues are absent. We again use OLS regression to estimate these treatment effects, and we present the treatment effects separately by party.
Figure 2 details the results of regressing the party treatment conditions on both the supporting and punishing behavior indices in 2021; Figure 3 does the same for 2023. Parallel tables can be found in the Online Appendix. The marginal effects presented in the graph are the difference between each partisan treatment condition and the control condition in which Casey is a citizen and the restaurant owner/podcaster is assigned no partisanship. If Casey is a co-partisan with the respondent, then the restaurant owner/podcaster at risk of being canceled is an out-partisan, and we would expect relatively higher support for canceling. By contrast, when Casey is an out-partisan in the vignette, then the restaurant owner/podcaster at risk of being canceled is an in-partisan and we would expect lower overall support for canceling. The gray dots and bars on the figure indicate the results for the whole sample, while the red and blue dots and lines show the patterns only for Republican and Democratic respondents (respectively). The baseline for comparison in this figure is the no-party condition, either for the entire sample or for only the corresponding partisan group. Treatment Effect Estimates by Party, 2021 CES data. Notes: Point estimates are OLS coefficients from a regression predicting the percent of “yes” responses. Baseline (dashed line) is the no-party control condition. For estimates for Republicans and Democrats, the baseline is the no-party condition for only that group of partisans. Models are run separately for each party and the full sample. 95% confidence intervals. Treatment Effect Estimates by Party, 2023 Prolific data. Notes: Point estimates are OLS coefficients from a regression predicting the percent of “yes” responses. Baseline (dashed line) is the no party control condition. For the separate estimates for Republicans and Democrats, the baseline is the no-party condition for only that group of partisans. Models are run separately for each party and the full sample. 95% confidence intervals.
As shown in Figure 2, we find strong support for H4 in the 2021 CES data which expected less support of canceling when the in-group is threatened. The partisan identity treatments clearly impact the ways in which people view the appropriateness and scope of different types canceling: when Casey is not the same party as the respondent and is contemplating canceling a member of the respondent’s in-party, there is significantly less support for canceling behaviors. The effect of the treatment is most pronounced on perceptions of supportive actions a good citizen should take; we observe smaller reduction in support for punishing behaviors on the good citizen questions. We see the same patterns in the overall estimates in the 2023 Prolific data, as shown in Figure 3.
When it comes to the definition of cancel culture, the treatment has a much smaller effect in 2021 and no effect in 2023. In the 2021 CES data, the effect is only significant for the punishing behaviors, but is consistent with a more expansive definition of canceling in a situation where respondents have reason to oppose it. When one’s own party is being canceled, people seem to be more willing to label various behaviors as canceling, suggesting partisan and identity-based reasons for these views (as opposed to more general or permanent views of values). The same effect does not appear in the 2023 results, suggesting that H4 may be more applicable to perceptions of what people ought to do as opposed to ideas about what cancel culture is in the abstract.
Next, we interacted the partisan treatment with respondents’ partisanship to determine how group attachment may affect Democrats and Republicans differently (see H4). We utilized the seven-point party identification variable and combined strong, weak, and leaners into the party they most identified with, leaving only “true” independents out of the sample. For the 2021 CES data, although the treatment effect in the figure appears larger for Democrats, we do not observe statistically significant interactions between the treatments and party. 9 This is a stark lack of support for H3. As noted in the prior section, we also interacted the party cue treatments with the value frames treatments and failed to find any evidence of an interaction between the treatments (see Tables A5 and A6 in the Online Appendix).
In the 2023 Prolific data, we do observe some significant interactions with party, primarily observing treatment effects overall and for Democrats. We do not observe significant treatment effects among Republicans. This is even stronger support against H3 than in the 2021 data; we observe the exact opposite of this prediction, only seeing effects for Democrats. 10
Finally, comparing the results in Figure 3 to those in Table 3, we observe substantively larger effects from the party treatments across both experiments. This leads us to conclude that partisan cues have a more dramatic effect on the behaviors people categorize as cancel culture and think other people should actually do in these areas (giving some further insight into RQ3).
In addition, we also included follow-up questions in the 2023 Prolific sample to evaluate respondents’ perceptions of the treatments and the podcaster. This included a set of judgments about the podcaster, asking respondents if they thought the podcaster was elitist, anti-religion, unpatriotic, woke, socialist, bigoted, intolerant, racist, sexist, or homophobic. Examining these items suggests that the party-based treatments did shift respondents’ reactions to the podcaster; when respondents shared a political party with the supporters of the bill and presumably the podcaster, evaluations of the podcaster improved. 11 This suggests that the partisan treatments shifted respondents evaluations of the podcaster for both Republicans and Democrats even if treatment effects were weaker for Republicans in the 2021 experiment. This again gives support to H4 and a lack of support for H3. Figures A9 and A10 of the Online Appendix display these results.
We also asked respondents to guess, in an open-ended question, the topic of the bill the podcaster supported. These data suggests that respondents thought the bill was about a wide range of topics (see Table A8 of the Online Appendix for the detailed breakdown). The modal response, 31 percent, mentioned abortion, with another 10 percent indicating something related to LGBTQ+ issues. This suggests that no one topic dominated people’s thoughts about the scenario.
Discussion/Conclusion
Our results suggest that cancel culture is complicated, subject to disagreement over values and influence by the same partisan forces that shape a great deal of American politics. From the Pew open-ended data, we see some discussion of political disagreement and mentions of core political values like the freedom of speech, accountability, and due process. The STM analyses indicate that, in addition to democratic values, respondents bring up different concerns about the economic consequences of cancel culture, the role of social media, and connect it to debates about rewriting history. However, there is a notable lack of explicit discussion of partisanship in these answers. The results of the factor analysis from our CES and Prolific data also indicate that reactions to different types of actions often linked to cancel culture contain a more supportive and more actively punitive dimension.
Our experimental studies help identify and distinguish the relative importance of value frames and partisan attachments in shifting perceptions of what cancel culture is and approval of specific, canceling actions. We observe strong evidence that the partisan identity of the target being canceled shifts evaluations of what behavior is normatively desirable. Relative to partisanship, values are the dominant way in which people talk about cancel culture. We also observe baseline differences between Democrats and Republicans in their willingness to support punishing, canceling behaviors. At the same time, the survey experiments clearly demonstrate that people are far more responsive to party cues than they are to values-based framing. In contrast to expectations of asymmetry in partisan discourse, we do not see a different set of reactions to the party-focused cues from Republicans; either effects are parallel across the parties or Democrats are more responsive than Republicans.
While there are many strengths to the methods used here, there are also some limitations that should motivate further study on this topic. Although reflective of the general population, the majority of the sample in the survey experiments is White, and the sample is not large enough to examine the more specific impact of race. We may have been, for example, better able to detect expression of and response to a minority empowerment value if minority citizens had made up a larger share of the sample. Additionally, the survey measures ask the respondent what a good citizen ought to do, not what the respondent would do. While this is an intentional choice, we cannot draw direct conclusions on what actions people actually perform or would be willing to perform. Given the role we observe of partisan identity and conflict, other studies should also consider the role of the media in influencing perspectives of cancel culture. The specific scenarios used here also represent cases where canceling is possible but not a foregone conclusion; this was intentional as to explore the influence of the values and partisan treatments. At the same time, future work could consider how the influence of the treatments we use here change when the person to be canceled is engaged in more or less severe types of actions.
Bringing together the various conclusions of this research suggests a series of important things about cancel culture and democratic politics more generally. First, the breadth of opinions about cancel culture suggests that political actors have space to influence public opinion on this topic and encourage different interpretations of their actions. This raises the possibility of political elites using the term cancel culture as another tool to mobilize support and demonize opposition without much real substance. Second, our experimental analyses suggest that partisan messages are more likely to be influential than frames targeting specific political values. We therefore provide another data point supporting the role of partisanship in contemporary political conflicts in the United States, and its potential primacy over other democratic values. Third, the inconsistency between what one ought to do and general definitions of cancel culture show how what people talk about in politics can often differ from the actions they are or are not willing to take. This is not unlike research on the so-called “principle-implementation gap” which has long established that people’s abstract principles can be far removed from their immediate behavior (Dixon, Durrheim, and Thomae 2017; Jackman 1978). Cancel culture seems to be no exception to this trend.
Supplemental Material
Supplemental Material - The Political Psychology of Cancel Culture: Value Framing or Group Identity?
Supplemental Material for The Political Psychology of Cancel Culture: Value Framing or Group Identity? by Elle Diether, Suzy Yi, Lisa P. Argyle, and Ethan C. Busby in Political Research Quarterly
Supplemental Material
Supplemental Material - The Political Psychology of Cancel Culture: Value Framing or Group Identity?
Supplemental Material for The Political Psychology of Cancel Culture: Value Framing or Group Identity? by Elle Diether, Suzy Yi, Lisa P. Argyle, and Ethan C. Busby in Political Research Quarterly
Footnotes
Acknowledgments
We would like to thank the editorial team at Political Research Quarterly and three anonymous reviewers for their comments and improvements on this paper. Madison Andersen also provided invaluable research assistance on the project, especially during its early stages. Faculty and students at the Center for the Study of Elections and Democracy at BYU also provided helpful suggestions and advice on the project. We would also like to thank the participants and discussants for the panel where this paper was presented at ISPP in 2022, and MPSA in 2023.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Brigham Young University.
Supplemental Material
Supplemental material for this article is available online.
Notes
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
