“It is critical that we condemn and confront domestic terrorism regardless of the particular ideology that motivates individuals to violence.… The definition of domestic terrorism in our law makes no distinction based on political views—left, right, or center—and neither should we.”—National Strategy for Countering Domestic Terrorism
In the spring of 2021, shortly after the riot at the US Capitol on January 6, 2021, the National Security Council (NSC) issued a new National Strategy for Countering Domestic Terrorism (NSCDT). Along with affirming that laws regarding terrorism make no distinction on political views, the document states that “in a democracy, there is no justification for resorting to violence to resolve political differences” (National Security Council Reference O’Harrow, Tran and Hawkins2021, 13). Although this is an admirable sentiment that would presumably garner near-unanimous support, there are in fact many contexts in which Americans believe political violence is justified, with recent studies indicating that as many as one in three Americans view the use of force as permissible for some political ends (Cox Reference Cox2021; Diamond et al. Reference Diamond, Drutman, Lindberg, Kalmoe and Mason2020).Footnote 1
When then do Americans view political violence as legitimate, and when do they condemn it as terrorism? Previous research has indicated that public opinion on what constitutes terrorism varies depending on elements of the perpetrator’s identity, including race and ethnicity (Abbas Reference Abbas2017; D’Orazio and Salehyan Reference D’Orazio and Salehyan2018) and on the form and severity of violence (Huff and Kertzer Reference Huff and Kruszewska2018). Yet there has been little research to date on the role of observers’ ideological biases in perceptions of the legitimacy of political violence, especially in contexts of domestic terrorism. In this article, I address this gap by employing a US-based experimental survey (YouGov, N = 3,640) to examine how political ideology influences public perceptions of domestic political violence perpetrated to advance right-wing or left-wing agendas regarding abortion, climate change, and immigration.
This article has several objectives. First and foremost, it aims to examine the extent to which the alignment of observers’ political ideology with a perpetrator’s political cause influences our perception of objectively equivalent acts of violence across highly partisan issues. In addition to flipping the focus from perpetrators’ to observers’ characteristics, the specific focus on political ideology is important, because ideology (in the conservative/liberal sense) has been understudied in terrorism literature, as well as in studies of social identity. In this article, I am interested in the extent to which ideology functions both as a social identity and as a framework for policy preference. Second, the article seeks to go beyond the usual foci on jihadi and, increasingly, far-right extremism to include both left-wing and right-wing actions.Footnote 2 The inclusion of left-wing actions is rare in terrorism literature; analyses of actions perpetrated to advance causes relatable to most participants are also uncommon. I contend that examining political violence through more familiar lenses helps us better understand our biases and blind spots when it comes to assessing what is or is not legitimate. Finally, the article aims to unpack the idea of legitimacy by distinguishing between disapproval, condemnation, and labeling an act as terrorism, going beyond the “is it terrorism” question to gauge perceptions of moral and strategic justifiability.
Why do public perceptions on terrorism and the legitimacy of political violence matter? As Huff and Kertzer (Reference Huff and Kruszewska2018) note, examining public opinion does not “resolve normative debates about what should or should not be considered terrorism” but is crucial, given the “central role that public opinion plays in our understanding of how terrorism works” (55) and of political violence more broadly. Although there are numerous definitions of terrorism and political violence, most agree that such acts are perpetrated to draw public attention to a particular cause. The way the public perceives the legitimacy of such acts is thus central to our understanding of terrorism and violence and has normative, legal, and policy implications. Indeed, the findings from this study underscore how political ideological biases can skew our sense of which acts of violence are justifiable, thereby influencing what we rationalize as morally permissible or defensible. This in turn has implications for how we prosecute such crimes, who we deem in need of “de-radicalization,” and how we develop policies to prevent or counter political violence.
How views on the legitimacy of political violence vary when respondents are ideologically aligned with a cause is the focus of this study: it examines the extent to which conservative/liberal sympathies for a given issue shape views on what constitutes terrorism and what is seen as morally and strategically justifiable. Controlling for variables such as perpetrator identity and type of violence, I demonstrate that political ideology influences how members of the public perceive the legitimacy of political violence in relation to their alignment with a cause, even when the severity of violence is identical. The article is organized as follows. First, I discuss how we perceive the legitimacy of violent actions and the role of political ideology in those assessments. Second, I explain my hypotheses and the methodology of the survey experiment. Third, I present the results, followed by a discussion of the findings. I conclude by discussing the relevance of the findings for academics, policy makers, and practitioners.
Terrorism, Legitimacy, and Political Ideology
Public Perceptions of Terrorism
What factors affect how members of the public view acts of political violence? Previous studies have focused this question on what constitutes terrorism in the public view, considering variables such as actor characteristics, the type of violence, and the type of target. Research has found that actors’ characteristics, particularly race and ethnicity, shape the extent to which respondents view acts as terrorism or not, with attacks perpetrated by Muslims or Arab Americans more likely to be considered terrorism than those perpetrated by Christians or white Americans (D’Orazio and Salehyan Reference D’Orazio and Salehyan2018; Huff and Kertzer Reference Huff and Kruszewska2018). Similarly, Piazza (Reference Schelling2015) finds that the public supports extraordinary detention measures for Muslim terror suspects more than for right-wing (usually white) suspects. This public perception also mirrors media coverage: Muslims are overrepresented as terrorists on US network and cable news (Dixon and Williams Reference Dixon and Williams2015), even though acts of right-wing and left-wing extremism outnumber Islamist attacks (Jones, Doxsee, and Harrison Reference Jost, Glaser, Kruglanski and Sulloway2020; O’Harrow, Tran, and Hawkins Reference Pape2021; Silva et al. Reference Tafjel and Tafjel2020). In addition to ethnicity and religion, a perpetrator’s group membership also influences public opinion, with individuals (often called “lone wolves”) less likely to be viewed as terrorists than members of an organization (Huff and Kertzer Reference Huff and Kruszewska2018).
Previous research has also considered how attributes of the action itself, including the type and severity of violence, influence perceptions of legitimacy. As Huff and Kertzer (Reference Huff and Kruszewska2018) indicate, incidents with higher casualties are more likely to be considered terrorism than those without casualties, as might be expected. But the type of action is even more statistically significant than casualty rates, with bombings more likely to be considered terrorism than other tactics such as shootings or hostage taking (63). Huff and Kertzer also find no significant difference in how the public perceives the type of target—namely, whether the target is a military, governmental, civilian, or religious center—even though many definitions of terrorism distinguish between the use of violence against civilian targets and that against military targets (64). However, Avdan and Webb (Reference Avdan and Webb2019) demonstrate that the type of target matters in how the public perceives terrorist threats, based on both physical proximity and personal proximity to the victims, particularly in terms of race; that is, if observers assume victims of a terrorist attack are the same racial or national (though not religious) group as themselves, they are more likely to perceive a terrorist threat.
The question of the type of target is often intertwined with the severity of violence in discussions of domestic terrorism in particular; many scholars argue that actions targeting property are not as violent or “terroristic” as those that target people (Amster Reference Amster2006; Cook Reference Cook2003; Liddick Reference Manz2006; Smith Reference Tafjel, Turner, Austin and Worchel2008, Vanderheiden 2005). This distinction is addressed at length in the literature on “eco-terrorism,” actions by radical environmental and animal rights groups (Carson, LaFree, and Dugan Reference Carson, LaFree and Dugan2012, Hirsch-Hoefler and Mudde Reference Hirsch-Hoefler and Mudde2014) in which perpetrators typically focus more on destroying or damaging infrastructure than physically harming individuals.Footnote 3 Similar debates are ongoing regarding attacks against abortion clinics, with the federal government only investigating those acts as terrorism after 9/11 (Mason Reference Mason2004; Wilson and Lynxwiler Reference Wouters1988).
This study seeks to control for these variables, including actor identity and the severity of violence, in assessing public opinion on what constitutes an act of terrorism. It also builds on these earlier studies in important ways. Primarily, I shift the focus from perpetrator (or victim) identity to observer identity, bringing new attention to the role of political ideological biases in assessing acts of violence. Indeed, the role of ideology (in the conservative/liberal sense) for both perpetrators and observers has been understudied. This study facilitates attention to ideology by situating political violence in the context of left-wing and right-wing issue-oriented actions, rather than the usual “jihadi” or “white supremacist” contexts.
Further, although designating an action as terrorism implies a legal and normative rejection of its legitimacy, the opposite is not necessarily true; that is, not viewing an action as terrorism does not necessarily equate to seeing it as legitimate, especially if it still falls within the bounds of political violence. Indeed, it should not be assumed that terrorism is inherently less legitimate or more immoral than hate crimes or other forms of violence. Thus, in this study, I go beyond the “is it terrorism” question to assess public perceptions on the moral permissibility and strategic value of violent actions and how they interact with terrorist designations.
Legitimacy of Violence: Legality, Morality, and Strategic Rationale
Legitimacy has both legal and normative dimensions. Weber (Reference Westwood, Grimmer, Tyler and Nall1947, 131), for example, equates legitimacy with legality and political order in systems of legal-rational authority but also notes that there needs to be a public belief in that order to give it validity. This underscores why investigating public opinion on political violence is important; most acts of violence are illegal, yet the public does not always agree with or “believe in” that legal basis. At the same time, as Cook (Reference Cook2003, 122) notes, “a demonstrable claim to legitimacy bestows moral authority on its bearer.” I contend that the reverse can also be true: an act seen as morally justifiable can enhance the sense of legitimacy, if not legality, ascribed to a given act of violence.
Of course, perceptions of legitimacy regarding acts of violence depend on the eyes of the beholder, especially in terms of morality, with views on acts of political violence influenced by one’s political sympathies and orientations. Indeed, if one views the target of violence as immoral or illegitimate, then violence becomes not only legitimate but also perhaps morally justified. As McWilliams (Reference Michael and Silke1970, 623) states, “Violence itself cannot be judged as immoral. Violence is ‘immoral’ only to the degree to which the will opposing it is ‘good.’” Regarding terrorism specifically, as Crenshaw (Reference Crenshaw and Crenshaw1983, 2) writes, left-wing terrorists typically “deny the legitimacy of the state and claim that the use of violence against it is morally justified,” whereas right-wing terrorists generally “deny the legitimacy of opposition and hold that violence in the service of order is sanctioned by the value of the status quo” (see also Cook Reference Cook2003, 109). In other words, violence can appear relatively legitimate or moral when serving an end with which we agree.
Further, even an action that lacks moral justifiability may be seen as legitimate by some if viewed as strategically effective for a broader cause; that is, if the ends are perceived as justifying the means. Both Crenshaw (Reference Crenshaw and Reich1998) and Pape (Reference Perlstein2003) have written on the “strategic logic” of terrorism, in terms of agenda setting, inspiring resistance, and provoking government responses to demonstrate injustices. Similarly, Lake (Reference Liddick2002, 26) suggests that extremist violence can be understood from a rationalist perspective as bargaining over ends, with the purpose to “provoke the target into a disproportionate response, radicalize moderates, and build support for… ambitious goals over the long term.” However, other scholars argue that bargaining theory does not hold in practice because terrorism rarely works to garner concessions, especially when civilians are targeted, as opposed to military or police targets (Abrahms Reference Abrahms2006; Reference Abrahms2012, Cronin Reference Cronin2009). Referencing Schelling (Reference Schuurman1966, 75–76), Abrahms (Reference Abrahms2013, 660) notes, “For coercion to work, the challenger must signal not only a credible threat to inflict pain when concessions are withheld, but also a credible promise to remove the pain in the event concessions are forthcoming.” However, terrorist actors generally suffer from a “credibility paradox” (Abrahms Reference Abrahms2013) that decreases incentives for states to negotiate or grant concessions or for moderates to support more extreme actions.
The social movement literature reflects similar debates. Winter (Reference Zschirnt2013) classifies movements into three sectors: mainstream (operating within legal bounds), militant (using civil disobedience but nonviolent), and extremist (using violence). These categories are not absolute, and the lines between them may be blurry. But in general, as Winter states, the sectors “hold a great deal in common ideologically, [yet] they use different tactics to achieve their aim and assert that ideology, from the most mainstream and legitimate tactics to the most violent and extreme” (7). How, then, do those in the mainstream of the ideological spectrum—presumably the majority—respond to acts of violence by those on the extremist end? In contrast to coercive utility arguments, research indicates that the use of violence, especially against civilians, can backfire on a movement by decreasing support rather than mobilizing moderates. For example, Feinberg, Kovacheff, and Willer (Reference Finkel, Bail, Cikara, Ditto, Iyengar and Klar2020, 1) demonstrate that extreme protest actions decrease support for social movements and for movements’ central causes, largely independent of prior views on the issue, and Huff and Kruszewska (Reference Iyengar and Westwood2016) similarly show that extreme tactics decrease public support for government negotiations and concessions.
Research on the use of nonviolent or less extreme tactics is mixed; Wouters (Reference Wouters2019) finds that public support increases for protests in which participants behave in a “worthy” (Tilly 2004) or nonviolent manner. However, Hsiao and Radnitz (Reference Huff and Kertzer2021, 493) show that “even if a protest tactic is carried out with the intention of demonstrating nonviolence, bias stemming from ideology or partisanship can militate against the expected effects on observers.” In other words, if observers are biased against a cause based on political ideology, the assumed strategic advantage of using nonviolent tactics rather than militant methods can be dampened. In this study, I do not employ a nonviolent counterfactual, although that is an area for additional research. I am instead interested in how political ideology intersects with the extent to which observers view violent actions as relatively strategic and how these perceptions of effectiveness align with views on the morality of such actions. In sum, although there are many dimensions to legitimacy, examining perceptions of both morality and strategic effectiveness can help us better understand whether attitudes toward political violence are rooted more in ethics or expediency.
Political Ideology, Othering, and Violence
Understanding the influence of political ideology on perceptions of violence is especially important in the context of domestic terrorism. Although there has been increasing scholarly attention to far-right extremism (Ahmed and Lynch Reference Ahmed and Lynch2021; Byman Reference Byman2021; Freilich et al. Reference Hirsch-Hoefler and Mudde2018; Manz Reference Mason2018; Michael 2019; Weinberg Reference Wilson and Lynxwiler2013), there has been little research comparing public opinion on left-wing and right-wing extremism within the context of domestic terrorism. A notable exception is Huff and Kertzer (Reference Huff and Kruszewska2018), who show that acts perpetrated by right-wing actors are slightly more likely to be viewed as terrorism (and thus less legitimate) than those perpetrated by left-wing actors. However, as noted earlier, although their study considers the ideology of actors, their analysis does not focus on the ideology of respondents.
How then does political ideology affect how ordinary citizens perceive acts of political violence related to causes to which they are aligned? According to social identity theory (SIT) (Tafjel Reference Tilly1978; Tafjel and Turner Reference Vanderheiden1979), much of how people assess the legitimacy of actions relates to the degree to which respondents see the actor as being in their in-group or out-group. As D’Orazio and Salehyan (Reference D’Orazio and Salehyan2018, 1018; emphasis added) write, “People categorize others into in-groups and out-groups, and have different reactions to objectively similar behavior depending on which category they fall into.” As noted earlier, race, ethnicity, and religion can affect in-group and out-group perceptions. However, in times of salient political polarization, political identity can also contribute to in-group and out-group designations. Indeed, with political ideology now being one of the most salient identities in the United States (Iyengar and Westwood Reference Jones2015), similar in-group and out-group trends are emerging along political lines between self-proclaimed liberals and conservatives and between Democrats and Republicans as those identifiers have emerged as key social identities (Mason Reference Mason and Wronski2016; Mason and Wronski Reference McAlexander2018).
Although political solidarity is nothing new, the conflation of politics with social identity has led to an unambiguous increase in affective polarization, with partisans viewing not only their own party positively but also the opposition more negatively; negative stereotypes, low tolerance, and often out-group animosity outweigh in-group favoritism (Cassese Reference Cassese2021; Iyengar and Westwood Reference Jones2015). Indeed, according to Cassese (Reference Cassese2021, 45), in-group/out-group dynamics have extended beyond mere othering to include dehumanization, with “partisans on both sides of the aisle dehumaniz[ing] their opponents in subtle ways—captured by the attribution of animalistic and mechanistic traits—and also in more blatant ways, in which political opponents are explicitly viewed as subhuman” (see also Cassese Reference Cassese2020). Of relevance to this study, Kalmoe, Gubler, and Wood (Reference Kearns, Betus and Lemieux2018) show that exposure to violence, or even violent metaphors, can promote in-group solidarity against out-groups, further reinforcing partisan polarization.
Most studies applying SIT to political polarization have focused on partisanship (political party) as the core in-group/out-group identity, yet political ideology—though distinct—reflects similar patterns and may even be a more useful indicator in some contexts. Traditionally, political scientists have viewed partisanship as a social identity while viewing ideology as a belief system, such that Republicans and Democrats may dislike one another more today than in the past, even though political beliefs on issues have not diverged to the same extent (Mason Reference Mason2015). With that framing, we would expect more in-group/out-group dynamics to occur along party lines rather than ideological lines. However, ideology is itself emerging as a social identity (Devine Reference Devine2015) and largely with affective patterns; Zschirnt (Reference Zschirnt2011), for example, has shown that conservative self-identification was largely a reaction against liberalism and its associated groups, while liberal self-identification has become more aligned with hostility toward conservative groups in recent years.
In this research, I am interested in the emergence of ideology as a social identity. Rather than viewing partisanship as purely identity based and ideology as purely belief based, I view ideology as a fusion of the two, with identity reinforced by positions on certain policies. As such, I am less concerned with participants “misidentifying” their ideology when self-reporting (Kalmoe Reference Kalmoe, Gubler and Wood2020) and more interested in their perception of where they are on the ideological spectrum and how that influences which acts of violence they see as legitimate. Specifically, if the opposing policy position is perceived as a marker of the out-group (just as earlier studies have demonstrated that demographic traits such as race, ethnicity, and religion demarcate the in-group and out-group), then, by extension, individuals and actions that reflect the out-group’s beliefs may be subject to bias, which in turn feeds views about legitimacy.
If we accept political ideology as at least partially a social identity, we can consider two different possibilities: first, that conservatives see bombings of an organization advocating for a cause they support as highly illegitimate, but just more legitimate than liberals do (and vice versa); and second, that conservatives see bombings of such groups not just as relatively more legitimate than liberals do but as legitimate in absolute terms as well (and vice versa). My hypotheses focus on the former; indeed, I expect the majority of both liberals and conservatives to reject the legitimacy of extreme acts of political violence, no matter the cause. However, in relative terms, I expect conservatives to see acts with which they are ideologically aligned as more morally and strategically legitimate than liberals do, and vice versa for liberals. Likewise, I expect both liberals and conservatives to more readily designate the “other” group’s actions as terrorism, rather than as actions that align with their ideology.
I expect that affective partisanship will go both ways (liberal bias against conservative actions and conservative bias against liberal actions). Although some research suggests that conservatives are more prone to partisan cues than liberals (Jost et al. Reference Kalmoe2003), other studies show no discernible difference (Iyengar and Westwood Reference Iyengar and Westwood2015). Indeed, as Conway and coauthors (Reference Conway, Houck, Gornick and Repke2018, 1050) write, “Negatively valenced traits once attributed asymmetrically to conservatives may in fact be fairly equally distributed across the political spectrum” (see also Chambers, Schlenker, and Collisson Reference Chambers, Schlenker and Collisson2013; Crawford Reference Crawford2012). As such, my hypotheses are as follows:
H1: Conservatives and liberals are less likely to view acts of political violence as terrorism when they are ideologically aligned with the cause.
H2: Conservatives and liberals are more likely to view acts of political violence as morally justifiable when they are ideologically aligned with the cause.
H3: Conservatives and liberals are more likely to view acts of political violence as strategic when they are ideologically aligned with the cause.
To assess the effect of political ideology on perceptions of legitimacy of political violence, I used a 3 x 2 x 2 survey experiment (N = 3,640), administered in the United States in May 2021 by YouGov. YouGov is a premier survey firm with a prescreened online panel, including diverse racial/ethnic groups, income levels, political ideologies, political parties, age groups, education levels, and other demographics. In this sample, participants reflected a range of income and education levels, with women slightly outnumbering men (54% to 46%). Importantly for this experiment, ideology was relatively evenly distributed, with 33% identifying as liberal or very liberal, 29% identifying as moderate, and 28% identifying as conservative or very conservative. Additional demographic details can be found in the Supplemental Text.
The survey tested public opinion on political violence across three highly partisan domestic issues: abortion, climate change, and immigration. I intentionally chose issues that are not primarily defined by race (though immigration is often framed in racial terms); although more than one-quarter of right-wing incidents in the past five years were related to white supremacy (Jones, Doxsee, and Harrison Reference Jost, Glaser, Kruglanski and Sulloway2020; O’Harrow, Tran, and Hawkins Reference Pape2021), I opted to focus on issues that are driven primarily by political ideology (liberal/conservative) rather than by racial attitudes to avoid conflation. I also intentionally chose issues that varied in terms of the direction of typical “real-world” political violence; abortion-related violence usually involves right-wing attacks on abortion clinics, whereas climate-related violence usually involves left-wing attacks on pipeline construction sites or corporations. Political violence related to immigration is mixed; although there have been right-wing attacks fueled by grievances toward immigrants (McAlexander Reference McWilliams2020), there have also been left-wing attacks on Immigration and Customs Enforcement (ICE) centers.Footnote 4
For the sake of maintaining symmetry and consistency in the experiment, the hypothetical attacks were carried out on organizations advocating for or against access to abortion, environmental regulations, and open immigration policies (rather than on specific sites such as clinics, pipelines, or immigration centers). Each respondent was randomly assigned 1 of 12 possible scenarios that described a violent attack on one of these targets. In each scenario, the attack was described as a bombing; according to Huff and Kertzer (Reference Huff and Kruszewska2018), bombings are the action most likely to be considered terrorism (in comparison to shootings or other forms of violence); thus, bombings are a hard test for ideological sympathy.
Each scenario also included an additional treatment to indicate whether there were “multiple casualties” or whether the attack resulted solely in property damage or destruction. I made this decision for several reasons. First, although a bombing is generally considered an extremist act, the term “bombing” is vague and leaves the scale of destruction open to interpretation. For example, some participants might envision a small pipe bomb or bottle rocket that causes minimal damage, whereas others might imagine a massive explosion that causes an entire structure to implode. The casualty treatment thus adds nuance to understanding potentially different responses to the type and severity of the attack. Second, as noted previously, issue-based domestic terrorism debates often focus on the justifications of targeting infrastructure as opposed to targeting people (see, for example, Eagen Reference Eagan1996; Vanderheiden 2005). Although the casualty treatment does not necessarily indicate intent, it allows us to identify differences between those who see property destruction as legitimate but still view actions that cause personal harm as illegitimate. Finally, whereas previous research has shown that casualty levels matter in public perceptions of terrorism (Huff and Kertzer Reference Huff and Kruszewska2018), I wanted to test how that variable compares to the ideological bias variable in determining participants’ views on a violent action’s legitimacy, moral permissibility, and strategic value.
In each scenario, the perpetrator was defined simply as “a man” to remove ambiguity of gender and to avoid inferences of race, ethnicity, or religion that can influence public perceptions of what constitutes terrorism. Sample scenarios included the following: “A man was convicted of bombing an organization that advocates for increasing access to abortion. The attack caused no injuries.” Or “A man was convicted of bombing an organization that advocates for fossil fuel interests. The attack caused multiple casualties.” Types of actions are summarized in table 1 (see the Supplemental Text for the full script of scenarios).
Respondents were then given three questions in randomized order to gauge perceptions of legitimacy regarding (1) whether the action was morally justifiable, (2) whether the action was strategic for achieving a political end, and (3) whether the action was terrorism. As noted earlier, the designation of an act as terrorism has inherent normative implications. The first two questions aim to unpack that normative framing further—and extend beyond the terrorism designation—by distinguishing between the morality of the act (is it ethically right or wrong) and the strategic value of the act (does it work).Footnote 5 Responses to the questions were based on a 7-point Likert scale ranging from “Strongly disagree” to “Strongly agree.” (In the data analysis, I rescaled the question responses to a scale of −3 to 3, where −3 indicated strong disagreement, 0 indicated a neutral answer, and 3 indicated strong agreement.)
I then ran linear and logistic regressions to examine the relationship between respondents’ political ideology and perceptions of ideologically similar actions as morally justifiable, strategically justifiable, or constituting terrorism. I chose to focus on ideology (rather than partisanship) for several reasons. First, the experiment focuses on issue-based attacks, not identity-based attacks. Second, as noted previously, I am interested in the emergence of ideology as a social identity; that is, how political beliefs and policy positions map onto in-group/out-group dynamics. Third, focusing on ideology may be more useful in the (post) Trump era, when many traditionally conservative Republicans left the party or voted Democrat, and many voters who had supported Obama voted for Trump in one or both elections. As such, partisan identity may be more fluid at this time of writing than usual,Footnote 6 whereas ideological identity has arguably become more entrenched. Further, although ideology was relatively evenly distributed across the sample—between (very) liberal, moderate, and (very) conservative—political party was skewed and less representative of the national population, with 39% identifying as Democrat, 22% as Republican, and 43% as Independent. As expected, a chi-squared cross-tabulation indicated a statistically significant relationship between ideology and political party; however, computing Cramer’s V identified it only as a moderately strong relationship (coefficient 0.41). The correlation between party and ideology is shown in figure 1.
In the survey, participants were asked to identify their political ideology as “Very Liberal,” “Liberal,” “Moderate,” “Conservative,” “Very Conservative,” or “Don’t Know.”Footnote 7 To measure the impact of respondents’ ideology, I first used an interval-level measure to recover the average effect of moving a respondent one space up or down the ideology scale. For this analysis, I removed the “Don’t Know” respondents and scaled the answers on a -2 to 2 scale, where -2 indicated a very liberal respondent and +2 indicated a very conservative respondent. Second, I operationalized ideology using a categorical measure to recover the effect of switching a respondent from one ideology to another (i.e., switching from moderate to liberal); I coded respondents as liberal if they recorded their ideology as “liberal” or “very liberal,” moderate if they described themselves as “moderate,” or conservative if they described themselves as “conservative” or “very conservative.” I first ran the regressions with the three issue areas combined and then for each issue separately.
The results indicate that political ideology is important in influencing perceptions on what constitutes terrorism (H1) and the morality of political violence (H2), but not on perceptions of strategic effectiveness (H3). They also demonstrate that actions that result in casualties are perceived as less legitimate than those that do not result in casualties, but the effect is outweighed by political ideology. The results were most pronounced on the issue of abortion. As hypothesized, the findings do not indicate that most conservatives/liberals view ideologically aligned actions as justified; indeed, a strong majority (77%) of respondents—conservative, moderate, and liberal—viewed the actions as terrorism, and a strong majority (72%) viewed the actions as morally unjustifiable (see the Supplemental Text for additional data on average response levels for liberals and conservatives and full results). However, where there were differences, political ideology was statistically significant in explaining them.
Ideology on Perceptions of Actions as Terrorism
Across all the models, moving a respondent rightward along the ideology scale is associated with a decrease in identifying conservative-aligned actions as terrorism, supporting Hypothesis 1 that conservative respondents are more likely to view conservative actions as less terroristic (table 2; figure 2). Similarly, liberal respondents identify conservative actions as more terroristic than liberal actions. This is a substantively large effect: one step along the ideology scale corresponds to the same difference in perceptions associated with an action that causes casualties. This finding is even more notable when considering that the standard deviation of the ideology measurement is a sizable 1.49.
Note: * p < 0.1; ** p <0.05; *** p <0.01.
The effect with liberal actions is slightly different: liberals are much less likely to designate liberal actions as terrorism relative to conservative actions, but in absolute terms, liberals still identify liberal actions as slightly more terroristic than conservatives do. This likely occurred because liberal respondents were more likely to view all the actions as terrorism, with the relationship between ideology and terrorism statistically significant (e.g., moving a respondent from the moderate to liberal category results in a 0.18 increase in the terrorism rating, regardless of the action). Still, the ideology effect is larger for liberals than for conservatives: the difference in liberals’ terrorism scores for liberal versus conservative actions is twice the difference between conservatives’ terrorism scores. This finding is counter to some previous studies that suggest conservatives are more prone to partisan cues and display larger double standards than liberals do (Jost et al. Reference Kalmoe2003). There is also evidence of left-leaning bias among moderate respondents; an action being conservative is associated with an increase in perceptions of it as terrorism, relative to liberal actions.
When analyzed by issue area (table 3), the strongest and most significant effect was on the issue of abortion, likely reflecting the highly contentious nature of the abortion debate in the United States. In the experiment, moving a respondent from the moderate to very liberal category results in a 0.76 higher terrorism score when the target is an organization that favors abortion rights. In contrast, moving a respondent from the moderate to very conservative category results in 0.66 times lower terrorism rating for the same target. Other issues were more nuanced. Moving a respondent from moderate to liberal results in a 0.40 increase in terrorism rating when the target is an organization advocating for environmental protections, whereas moving a respondent from moderate to conservative decreases their terrorism rating of a pro-environment group by 0.49 and increases their terrorism rating of a fossil fuel organization by 0.47. Effects on immigration were not statistically significant.
Notes: All continuous predictors arc mean-centered and scaled by 1 standard deviation. *** p <0.001; ** p <0.01: * p <0.05.
Ideology and Perceptions of Actions as Morally Justifiable
When the issues are analyzed collectively (table 4; figure 3), there is a significant positive relationship between a respondent’s ideology on a 5-point scale and perceptions of moral justifiability: in support of Hypothesis 2, conservatives rate conservative-aligned actions as more justifiable than liberals (and moderates), and liberals rate liberal-aligned actions as more justifiable than conservatives. In terms of (a)symmetry, liberals and conservatives indicate similar levels of moral justification for actions with which they are aligned, but liberals see less justification for conservative actions than conservatives do for liberal actions. For moderate respondents, an action being conservative is associated with a decrease in perceptions of it being morally justifiable. Surprisingly, moderates indicate slightly higher moral justification for liberal actions than even those who identify as liberal or very liberal.
Note: * p <0.1; ** p <0.05; *** p < 0.01.
When analyzed separately, there are notable differences among the issue areas (table 5), with the effects most pronounced and statistically significant on the issue of abortion. Moving a respondent from the moderate to very liberal category results in a 0.79 lower moral legitimacy rating for attacks on pro-choice organizations, whereas moving a respondent from the moderate to very conservative category results in a 0.99 higher moral legitimacy score when the target is pro-choice. Again, this finding likely reflects the highly contentious nature of the abortion debate in the United States and the salience of religious and moral codes in influencing views on this issue. Other results were mixed, however, with ideological lines more blurred on environmental and immigration issues; these findings suggest the need for further research on the complementary mechanism of issue salience.
Notes: All continuous predictors are mean-centered and scaled by I standard deviation.*** p < 0.001;** p < 0.0 1;* p < 0.05.
Perceptions of Actions as Strategic
In contrast to Hypothesis 3, political ideology alignment had no significant effect on public opinion on the strategic justifiability of an action; that is, whether it would be strategic for achieving a political end. Insignificant interactions between the respondent ideology term and the conservative/liberal action indicate that a respondent’s ideology does not influence the effect that an action being liberal or conservative has on their view of it as being strategic. The same was true in the categorical measure and the single-issue measures; insignificant interactions between the ideology variables indicate that a respondent’s ideology does not influence how they react to an action being conservative or liberal when assessing potential effectiveness.
The lack of a significant relationship between ideology alignment and perceptions of legitimacy may be due to several factors. First, questions related to perceptions of terrorism and morality are fairly straightforward and specific to the respondent’s own opinions. However, a question regarding strategic justifiability may compel some respondents to consider what other observers might view as strategic effectiveness. Relatedly, some respondents may view any potential broader outcome regarding strategic effectiveness as plausible, even if they personally were opposed to the motive of the tactic. Finally, the lack of relationship may reflect the general ambiguity on the effectiveness of violence discussed previously; some will see extreme actions as furthering a cause or forcing concessions, whereas others will see extreme actions as backfiring on a movement.
Yet, when controlling for the interaction between respondents’ ideologies and an action being conservative or liberal, there is a positive relationship between ideology on a 5-point scale and perceptions of the actions as strategic. The coefficient associated with a respondent being liberal is significant and negative across all models, indicating that, relative to moderate respondents, liberals saw all actions as less strategically effective. However, the coefficient associated with a respondent being conservative is not significant across all models, indicating that there were no significant differences between how conservative respondents and moderate respondents perceived the actions as strategic. This may reflect less tolerance from liberals for violence in general (which would require further research to fully assess) or perhaps a traditional left-leaning preference (in leftist/Marxist groups for example) for more cohesive organizing over individual violent attacks.
Severity of Violence
The experiment also indicates that severity of violence matters, but not as much as ideological alignment. Across all the models, the coefficient associated with casualties was positive, indicating that respondents of all groups found actions that resulted in casualties to be more terroristic. However, the effect was still less than that of ideological alignment. For example, liberals were more likely to view a conservative-leaning action without casualties as terrorism than they were to view a liberal-leaning action with casualties as terrorism. This suggests that ideology has a relatively large effect on perceptions of terrorism: the difference between how moderates and conservative/liberal respondents respond to an action is almost twice as large as the average change in response associated with an action causing casualties. This finding is similar to that of Huff and Kertzer (Reference Huff and Kruszewska2018, 63) who found that “although higher-casualty incidents are more likely considered to be terrorism, the size of the effect is small relative to [other factors].”
Surprisingly, there was no significant relationship between the presence of casualties and perceptions of moral justifiability, indicating that the severity of violence has no significant effect on overall perceptions of the morality of acts of violence. This finding should be tempered somewhat, however. As noted earlier, because the majority of respondents found all the actions, with or without causalities, to lack moral justifiability, the absence of a significant relationship may reflect this reality more than a lack of moral concern. Further, where respondents do see moral justification, it is important to note that they have not been presented with a nonviolent tactic counterfactual. Indeed, across all models, there was a significant negative relationship between an action resulting in casualties and perceptions of the action as strategic, indicating that, holding all other factors constant, an action resulting in casualties was associated with lower perceptions of strategic effectiveness relative to an identical action that did not produce casualties. This is in accordance with other studies (Abrahms Reference Abrahms2013; Cronin Reference Cronin2009) mentioned previously that argue against the strategic rationale of extreme actions, especially those that target civilians.
Covariates and Demographic Trends
Although my focus was on political ideology, I also tested party affiliation but found that ideology was in fact a better explanatory variable. When the issues were analyzed collectively, coding respondents as Democrat or Republican did not reveal significant relationships between partisan identity and the perceived legitimacy of actions with an associated ideology overall. When broken down by issue, however, partisan identity achieved statistical significance on the issue of abortion: Republicans were 0.68 times more likely than Democrats to consider an attack targeting a pro-choice organization as morally legitimate and 1.03 times less likely than Democrats to consider it terrorism. Republicans were also 0.46 times less likely than Democrats to consider an attack against a pro-environment facility as terrorism. However, these differences may be due to the underlying trend of Democrats (like liberals) being more prone to identifying a violent action as terrorism than Republicans or Independents. Further, as noted, we should be cautious of overstating these findings because the single-issue analyses were based on relatively small Ns from which it is difficult to draw strong conclusions.
Other covariates had little or no effect, with the exception of gender. Men were 0.17 times more likely than women to consider any violent action as morally legitimate (and on the issue of abortion, men were 0.33 times more likely than women to consider an attack against a pro-abortion organization as morally justifiable). Men were also 0.29 times more likely than women to view any violent action as strategically legitimate. However, men were also 0.13 times more likely than women to designate violent actions as terrorism.
Discussion: Political Ideology and the Legitimacy of Violence
Political ideology has a statistically significant effect on how the public perceives acts of political violence. Respondents were more likely to view actions aligned with their political identity as more legitimate than those unaligned in two ways: they were less likely to view those actions as terrorism, and they were more likely to see violent actions as more morally justifiable. Likewise, respondents were more likely to assess objectively equivalent actions with which they were ideologically opposed as less legitimate, viewing them as morally unjustifiable and as acts of terrorism. In terms of designating an action as terrorism, the effect of political ideology was nearly double that of the severity of violence (represented by loss of life or serious injury).
There are different ways to measure legitimacy. As discussed previously, the actions in the experiment were all objectively illegal but could arguably be seen as legitimate if respondents considered them to be morally justified, strategically effective, or both. The findings indicate that, for actions aligned with political identity, both conservatives and liberals are more likely to view those actions as more morally justified but not necessarily as strategically effective. In other words, when ideologically aligned, the actions are more likely to be seen as morally legitimate but not necessarily useful in advancing a cause. This has some pragmatic implications: even relative moral “support” for an action does not equate to approval of its use if it is not seen as effective. This finding may help temper other broad conclusions of surveys that indicate “support” for the use of force or violence for political ends; moral support (or at least lack of moral condemnation), although notable and perhaps worrisome, does not necessarily lead to endorsement or actualization. This is particularly important when thinking about what steps we take to prevent violent extremism.
As noted previously, this study sought to go beyond the “is it terrorism or not” dichotomy. But that question is still another useful indicator of the extent to which we view certain actions as legitimate. For example, one could view a violent action that is aligned with their political identity as neither morally justifiable nor strategically effective, but still not consider it terrorism, whereas a violent action attacking something one believes in could easily be interpreted as a terrorist act. In the experiment, the interrelationship between the three dependent variables was correlated, but perhaps not as strongly as might be expected; moral and strategic justification had a correlation coefficient of 0.43, and moral justification and terrorism had a correlation coefficient of -0.32. Empirically, this lower-than-expected correlation might be attributed to the scale of responses. As shown in the Supplemental Text, although the majority of respondents saw all the acts as morally unjustifiable and as terrorism, they indicated moral unjustness more strongly than they did terrorism (e.g., strongly disagreeing that the act was morally justified, while being more spread on the positive side of the scale indicating that the act was terrorism). Conceptually, this may be attributable to respondents’ uncertainty about the legal definition of terrorism,Footnote 8 particularly in domestic cases.
However, the relatively low correlation may also be explained by the fact that the terrorism label is typically used to describe not merely actions with which we disagree but also those that we actively condemn as distinct from other types of violence or crime. This creates a feedback loop of sorts in that public consensus around legitimacy (or lack thereof) affects the designation as terrorist but also, by designating acts as terrorism—in public discourse, in the media, or by elites—the group and its ideology are delegitimized. As Rubinstein (Reference Schock1988) states, “to call an act of political violence terrorist is not merely to describe it but to judge it… it implies illegitimacy” (quoted in Eagan Reference Eagan1996, 1). Indeed, classifying actions as terrorism has normative implications: “to categorize something as terrorism is to de-legitimate its goals” (Huff and Kertzer Reference Huff and Kruszewska2018, 56). It is thus not surprising that we are less likely to strongly condemn actions as terrorism when we sympathize with the cause, even if we disagree with the tactics. Likewise, in accordance with affective polarization, we are more likely to actively condemn another group’s actions as terrorism. As Rubinstein (Reference Schock1988) aptly states, “Nobody wants to be called a terrorist; terrorism is what the other side is up to” (quoted in Eagan Reference Eagan1996, 2).
Unpacking terrorism and legitimacy in these ways is more than just a semantic exercise; indeed, the terrorism label carries real consequences, more so than the abstract notion of legitimacy. As Huff and Kertzer (Reference Huff and Kruszewska2018, 56) note, “Classifying actions as terrorism has direct policy implications for how the perpetrators are prosecuted”: terrorism is a federal charge, whereas most violent crime is prosecuted at the state level. Identifying certain actions as terrorism not only affects how specific acts are prosecuted but also changes the extent to which state, local, and federal officials can take steps to monitor the activity of groups or individuals to prevent acts of “terrorism.”
For example, during the War on Terror, the US Patriot Act allowed for expanded tools and strategies to “intercept and obstruct” terrorism. Most of the focus in those years was on external terrorist threats, although the National Security Agency (NSA)’s surveillance program targeted both foreign and domestic nationals. As the focus turns to domestic terrorism (US Patriot Act 2001), there will no doubt be further concerns about the broad use of the terrorism label, especially if applied more broadly to one ideological group. The NSC indicates an awareness of the need to balance civil liberties and free speech with security and violence prevention; the 2021 NSCDT states, “New criminal laws, in particular, should be sought only after careful consideration of whether and how they are needed to assist the government in tackling complex, multifaceted challenges like the one posed by domestic terrorism and only while ensuring the protection of civil rights and civil liberties” (25; emphasis added). However, even absent new criminal laws, the application of the terrorism label to domestic political violence will no doubt increase normative and political pressure to crack down on potential “terrorists,” especially if public opinion supports preventive actions. The findings from the experiment underscore the need to be clear-eyed about when and how to use the terrorism label, given that political ideology can skew our perspective of what counts as terrorism and who should be considered a terrorist.
Why Political Ideology Matters
With political polarization seemingly at an unprecedented high in America (Finkel et al. Reference Freilich, Chermak, Gruenewald, Parkin and Klein2020), it is important to locate how and where those divides manifest. We should expect that there will always be differences of opinion on policy issues, including those that were included in the experiment. But the study indicates how those policy positions, and the sense of political identity associated with them, can also shape and skew our beliefs on broader concepts of violence, morality, and legitimacy. Indeed, the results indicate that, even when presented with objectively equivalent acts of violence or, in some cases, more severe acts of violence, we are likely to be biased to some degree by our political ideology. Further research is needed to determine the extent to which this effect is due to belief-based policy convictions, the emergence of ideology as a social identity (with familiar in-group/out-group dynamics), or both, which are no doubt influenced by elite cues and rhetoric informing more ideologically entrenched worldviews. I contend that the two are more intertwined than previous distinctions suggest and argue for more attention to how ideology, especially in the context of issue positions, is emerging as a form of identity—sometimes even superseding partisanship—and thus skewing perspectives on seemingly objective ethical questions and policy considerations.
Indeed, it is important to note that, although political ideology showed clear statistical significance with perceptions of political violence, political party was less explanatory. That is, in most of the scenarios, with the exception of abortion, coding respondents as Democrat or Republican revealed few significant relationships between partisan identity and the perceived legitimacy of actions with an associated ideology. For example, in most cases, Republicans were not more statistically likely to view conservative-leaning actions as legitimate, nor were Democrats more statistically likely to view liberal-leaning actions as legitimate. One explanation for this may be the sample itself, which, as noted earlier, intentionally included a representative ideological spread of liberals, moderates, and conservatives but was less nationally representative by political party, with Republicans underrepresented relative to Democrats and Independents.
Another explanation may be that party members who generally view themselves as non-extremist on such issues may actively seek to distance themselves from acts of violence, even if they are generally sympathetic to the political cause. Indeed, the fact that such individuals are potentially invested in the cause may make them even more likely to try to distinguish themselves from external extremists; as Winter (Reference Zschirnt2013, 1) writes, “Defenders of a cause will attempt to differentiate between the two in order to maintain their legitimacy.” Another explanation (and the rationale for focusing on ideology over party in the analysis) is the fact that the experiment focused on issue-based rather than identity-based scenarios. As such, the usual tribalism and sectarianism that are increasingly associated with partisanship may have been dampened somewhat by more policy-motivated responses. This is especially true when combined with the complementary mechanism of issue salience: the findings indicate that, as might be expected, the effect of ideology is strongest when applied to the most contentious issues. It is clear, however, that political ideology, in both its policy and social identity frames, influences how we think about political violence. More research is needed in this area to investigate the effect of ideology compared with partisanship on questions of violence and terrorism.
This study has indicated how political ideology influences the extent to which we see acts of domestic political violence as morally justified or legitimate. The findings have implications for scholars, policy makers, and practitioners.
In terms of theory, the study contributes to our understanding of public perceptions of “terrorism” in several ways. First, it sheds new light on opinions on domestic terrorism, including left-wing actions that are often overlooked, along with right-wing actions. This approach reveals that, contrary to some assumptions, “double standards” among liberals on questions of extremism may exceed those of conservatives, even as liberals demonstrate less tolerance for violent acts overall. Second, it extends beyond the “is it terrorism or not” dichotomy and its inferred correlation with (il)legitimacy by assessing how legitimacy is defined—whether through moral justifiability, strategic effectiveness, or both—with moral justification proving to be a stronger variable. Third, the study adds to a growing field of knowledge about the impacts of political identity and polarization, integrating political ideology (as distinct from partisanship) in both independent and dependent variables to ascertain its impact in the context of domestic extremism and arguing for more research on political ideology as both a social identity and belief framework. Finally, the study highlights the nuances in public opinion on political violence and will, I hope, inspire further experimental research on this topic.
In terms of policy, if ideological alignment influences what we deem to be terrorism or not and what we see as legitimate or illegitimate, it is important to be aware of such biases when defining terrorism and developing terrorism policies. The new NSCDT mentioned at the start of the article aspires to make no distinction between political ideologies when identifying domestic terrorism, but specific designations may prove to be contentious. For example, should the US Criminal Code include property damage and destruction in its definition of terrorism? The findings from the study suggest that liberals might support such a move to define attacks against pro-choice organizations (or, more likely, abortion providers), whereas conservatives might support such a move to define attacks against pipeline construction firms or infrastructure. Similarly, these designations affect the way we prosecute such crimes, whether as federal terrorism offenses or at the state level as hate crimes, violent crimes, or property destruction. As strategies to counter domestic terrorism develop, policy makers should be aware of the effects of political ideology on how we perceive different actions.
The findings from this study have implications for practitioners as well. For example, there have been a proliferation of preventing violent extremism and countering violent extremism initiatives since 9/11 that are aimed to reduce radicalization, concentrating mostly on perceived jihadi or Islamist threats. Many of these programs have recently started to focus more on interventions with right-wing extremists, especially after the 2017 Unite the Right Rally in Charlottesville, Virginia, and the 2021 riot at the US Capitol (see, for example, the Resolve Network at the US Institute of Peace and Life after Hate). Yet even those initiatives have focused mostly on individuals and groups associated with white supremacy. There have been fewer efforts to prevent or address violent extremism when the issue or ideology itself is widely supported, from either the left or the right. This study should remind practitioners to be self-aware when determining who we deem in need of interventions.
Finally, the study has broader implications for how we understand our conceptions of violence, morality, and legitimacy. We assume that we think of violence as normatively “bad” and illegitimate, but it is evident that we easily rationalize some violent actions as more morally justifiable—or at least not condemnable as terrorism— when we agree with the cause, even if we do not consider such actions to be strategic or legal. At the same time, we are quick to condemn identical actions that go against our interests and are ready to label such actions and movements as illegitimate. These broader questions of perceived legitimacy extend beyond the issues identified in this article to include other recent episodes of political violence in which political ideology has influenced what we see as permissible or justified, further reinforcing our political identities and contributing to further polarization.
To view supplementary material for this article, please visit http://doi.org/10.1017/S1537592722000688.
I would like to thank the anonymous reviewers for their insightful comments and feedback. I am also grateful to Benjamin Green and Benedetta Giocoli for assistance with the data analysis, and Thomas Gift for initial ideas and discussion about the experiment.