HOSTILE RUMORS AND CONTEXTS OF CONFLICT

Sharing hostile political rumors serves social goals beyond promoting the perceived truth, and is often linked to situations of social conflict (Horowitz Reference Horowitz2001; Petersen, Osmundsen, and Tooby Reference Osmundsen, Bor, Bjerregaard Vahlstrup, Bechmann and Bang Petersen2021). In a comprehensive review of the factors involved in the onset of ethnic riots, Horowitz (Reference Horowitz2001, 74) contends that hostile rumors are more effective than accurate information at mobilizing violence, concluding, “Rumor prevails because it orders and organizes action-in-process.”

The functions of hostile rumor sharing in the context of social conflict are manifold. The act of sharing signals the sharer’s commitment to aggression, potentially creating mobilization cascades among others with similar preferences (Petersen, Osmundsen, and Tooby Reference Osmundsen, Bor, Bjerregaard Vahlstrup, Bechmann and Bang Petersen2021); the content of hostile rumors may persuade neutral audiences to turn against the outgroup (Horowitz Reference Horowitz2001); and flooding informational ecosystems with false, hostile information can sow distraction and confusion among outgroup members as they spend energy on countering the misinformation (Heath Reference Heath2021). Finally, sharing offensive information may signal to others that the sharer has an aggressive and dominant personality (De Araujo et al. Reference De Araujo, Altay, Bor and Mercier2021).

HELPING YOUR PARTY OR DESTROYING THE SYSTEM?

Partisanship offers a common explanation for why people share hostile political rumors in the contemporary United States (Allcott and Gentzkow Reference Allcott and Gentzkow2017; Van Bavel and Pereira Reference Van Bavel and Pereira2018). Deepening political conflict has caused partisans in the United States and beyond to develop hostile feelings across party lines (Iyengar et al. Reference Iyengar, Lelkes, Levendusky, Malhotra and Westwood2019), making it plausible that deeply committed partisans will strategically share hostile political rumors to target members of the opposing party and mobilize co-partisans. Consistent with this, studies find that partisanship predicts political conspiracy beliefs about the out-party (Miller, Saunders, and Farhart Reference Miller, Saunders and Farhart2016) as well as belief in and sharing of political “fake news” (Allcott and Gentzkow Reference Allcott and Gentzkow2017; Osmundsen et al. Reference Osmundsen, Bor, Bjerregaard Vahlstrup, Bechmann and Bang Petersen2021; Van Bavel and Pereira Reference Van Bavel and Pereira2018).

Yet, motivations to share hostile political rumors may also reflect dissatisfaction with the entire political system, including all its traditional actors. In fact, alongside growing partisan polarization, democracies have experienced rising levels of income inequality and stagnation in real wages (Turchin Reference Turchin2016). As social well-being has worsened, some citizens increasingly express feelings of “losing out” and discontent with the political establishment (Kriesi and Schulte-Cloos Reference Kriesi and Schulte-Cloos2020). These feelings of status loss and marginalization, even if imagined rather than real, have shaped recent political events, including the election of Donald Trump as president of the United States (Mutz Reference Mutz2018) and the rise of populism in Europe (Kriesi and Schulte-Cloos Reference Kriesi and Schulte-Cloos2020). Frustrations about status loss have been observed among members of traditionally privileged groups (e.g., white men), but actual experiences of historical injustices by members of marginalized groups can also trigger deep dissatisfaction with the political system (e.g., among Black individuals; Kimmel Reference Kimmel2017; Mutz Reference Mutz2018).

Prior research suggests that experiences of marginalization activate disruptive views and behaviors. For example, indicators of marginalization, such as lack of trust, predict conspiratorial beliefs beyond the effects of partisanship (Miller, Saunders, and Farhart Reference Miller, Saunders and Farhart2016). Psychological research also shows that uncertainty about one’s own social standing predicts identification with radicalized groups (Hogg, Kruglanski, and van den Bos Reference Hogg, Kruglanski and van den Bos2013) and support for political violence (Gøtzsche-Astrup Reference Gøtzsche-Astrup2019).

Individuals who are motivated by partisan animus to share hostile political rumors tend to share only those that strategically help their “own” party by denigrating the opposing party (Osmundsen et al. Reference Osmundsen, Bor, Bjerregaard Vahlstrup, Bechmann and Bang Petersen2021). In contrast, individuals’ discontent with the entire political system should motivate sharing of hostile political rumors that they believe could damage the system itself. As a result, these individuals should share hostile political rumors, irrespective of which party it helps or hurts.

A THEORETICAL MODEL: THE DUAL ROLE OF STATUS ORIENTATIONS AND SOCIAL MARGINALIZATION

We hypothesize that some individuals are so disaffected with current society and their (perceived) status in it that they indiscriminately share hostile political rumors as a way to disrupt the established democratic “cosmos” and start anew.

Prior research has made considerable progress in understanding nonmainstream but legitimate forms of political activism such as voting for populist or radical right-wing parties and how this is fueled by feelings of marginalization in the face of globalization, unemployment, and immigration (Kriesi and Schulte-Cloos Reference Kriesi and Schulte-Cloos2020). Nonetheless, the current spread of hostile political rumors is linked to events that go far beyond legitimate activism, such as the violent protests against state interventions during the corona pandemic (Bartusevičius et al. Reference Bartusevičius, Bor, Jørgensen and Petersen2021) or the insurrection against the U.S. Congress in January 2021.

In order to understand the psychological motivations of these individuals, we must consider the interaction of personality dispositions and situational triggers (Crocker et al. Reference Crocker, Thompson, McGraw and Igerman1987). Recent work on radicalization shows that feelings of marginalization, even if extreme, mostly lead to radicalization for individuals with “dark” personalities (Gøtzsche-Astrup Reference Gøtzsche-Astrup2019).

Individual differences in the importance attributed to social status are particularly important for understanding aggression. Individuals who feel entitled to have a high social status are more prone to engage in antisocial behavior, such as participation in violent events (Bartusevičius, van Leeuwen, and Petersen Reference Bartusevičius, van Leeuwen and Petersen2020) and online bullying (Bor and Petersen Reference Bor and Petersen2022; Buckels, Trapnell, and Paulhus Reference Buckels, Trapnell and Paulhus2014). More specifically, psychological research differentiates between prestige and dominance as differential routes to status (Cheng et al. Reference Cheng, Tracy, Foulsham, Kingstone and Henrich2013) and suggests that dominance orientations are particularly likely to engender feelings of entitlement that motivate aggression. Prestige entails earning status in exchange for competence. Dominance entails achieving status via “the induction of fear, through intimidation and coercion,” which in turn motivates aggressive behavior (Bartusevičius, van Leeuwen, and Petersen Reference Bartusevičius, van Leeuwen and Petersen2020), including in the online domain (Bor and Petersen Reference Bor and Petersen2022). Individuals who are disposed to use dominance to sustain their perceived entitlements to status may therefore be particularly likely to react aggressively in the face of challenges to their status.

While the relationship between dominance and antisocial behavior in politics is well established, we know significantly less about the dynamics between such traits and motivations to disrupt established hierarchies, especially in politics. In fact, dominance-oriented individuals often support existing social hierarchies (Ho et al. Reference Ho, Sidanius, Kteily, Sheehy-Skeffington, Pratto, Henkel, Foels and Stewart2015).

A fuller understanding of the relationship between dominance motivations and discontent requires us to take situational triggers into account—particularly, feelings of marginalization as highlighted in prior research on populism (Kriesi and Schulte-Cloos Reference Kriesi and Schulte-Cloos2020). A key driver of some people’s discontent with the political system is discrepancy between their perceived social standing and the social standing to which they feel entitled (Crocker et al. Reference Crocker, Thompson, McGraw and Igerman1987). Dominance-oriented individuals feel greater entitlement and greater motivation to obtain a superior position within the hierarchy. Marginalization should therefore activate greater feelings of dissatisfaction with the current system among those who are oriented toward dominance-based status.

Marginalization exists in many forms, and some may trigger status-oriented individuals more strongly to share hostile political rumors. First, a consistent finding in psychology is that losing a resource looms larger than failing to gain a resource of a similar size (Tversky and Kahneman Reference Tversky and Kahneman1991). This has also been found to be the case in relation to social status (Osmundsen and Petersen Reference Osmundsen and Petersen2020) and, accordingly, the loss of a privileged position may be particularly likely to generate strong anti-systemic sentiments. This is consistent with research suggesting that feelings of “aggrieved entitlement” are particularly disruptive and found among historically privileged groups such as white men (Kimmel Reference Kimmel2017).

Second, marginalization can be personal, but it can also be felt with reference to one’s social group. Feelings of insecurity about one’s personal social status have consistently been found to propel people toward extremism (Hogg, Kruglanski, and van den Bos Reference Hogg, Kruglanski and van den Bos2013). At the group level, however, prior research suggests that extremism is often related to inflated appraisals of one’s ingroup (e.g., in the form of collective narcissism) rather than feelings of inferiority (Golec de Zavala and Lantos Reference Golec de Zavala and Lantos2020). In fact, such inflated group appraisals often reflect personal marginalization. In their review of the literature on collective narcissism, Golec de Zavala and Lantos (Reference Golec de Zavala and Lantos2020, 276) thus argue that “the in-group’s image is used as a vehicle to satisfy frustrated self-importance and to protect the undermined self-esteem.” Overall, these lines of research converge on the prediction that personal status losses may trigger particularly disruptive responses.

DESTRUCTION AS A DOMINANCE STRATEGY: THE NEED FOR CHAOS

Dominance strategies have mostly been analyzed as ways to protect the status of individuals in high-status groups (Cheng et al. Reference Cheng, Tracy, Foulsham, Kingstone and Henrich2013). In contrast, we are interested in understanding how dominance strategies are adopted by individuals who perceive themselves to be at the periphery of society. We hypothesize that when coupled with feelings of marginalization, dominance-oriented strategies may flip from using targeted aggression toward specific individuals to indiscriminate tactics that sow chaos as a means to disrupt the system and advance up the social hierarchy.

Extant research offers some empirical support for our thesis that marginalization and status loss can motivate some individuals to sow chaos through indiscriminate aggression. For example, the cross-cultural phenomenon of “running amok” emerges in the face of severe status loss (Hempel et al. Reference Hempel, Levine, Meloy and Westermeyer2000). Similarly, in Western societies, shooting rampages are sometimes a response to chronic feelings of marginalization among individuals with antisocial dispositions (Twenge and Campbell Reference Twenge and Campbell2003). In the words of one of the shooters from the Columbine school massacre, “The lonely man strikes with absolute rage!” (Krizan and Johar Reference Krizan and Johar2015, 797). At the less extreme end, psychological studies of exclusion and aggression show that the combination of situational exclusion and dispositional sensitivity to rejection can lead to aggressiveness, even against innocent bystanders (DeWall, Enjaian, and Bell Reference DeWall, Enjaian, Bell, Williams and Nida2016). Similarly, reviews on the frustration–aggression hypothesis find that displaced aggression is a reliable phenomenon (Marcus-Newhall et al. Reference Marcus-Newhall, Pedersen, Carlson and Miller2000) and that it emerges in individuals with antisocial personalities when they face the continuous presence of multiple small provocations (Miller et al. Reference Miller, Pedersen, Earleywine and Pollock2003; Twenge and Campbell Reference Twenge and Campbell2003).

While generalized destructive tendencies, such as displaced aggression, are often interpreted as maladaptive (Miller et al. Reference Miller, Pedersen, Earleywine and Pollock2003), there may be functional benefits to displays of destructive intent for marginalized individuals. First, displays of destructive tendencies may serve as hard-to-fake signals of the motivation to impose costs and, hence, operate as a general deterrence device (Fessler et al. Reference Fessler, Tiokhin, Holbrook, Gervais and Snyder2014). Second, if individuals react with severe aggression to rejection, others may be pressured to invest in burnishing the person’s sense of self as a way to reduce the person’s ire. For example, among vulnerable narcissists, minor rejections may trigger “narcissistic rage” involving disproportionate aggression with the goal of avoiding further rejection (Krizan and Johar Reference Krizan and Johar2015). Finally, as a dominance strategy, marginalized individuals may see destruction as a form of “niche construction” in which they cultivate a social ecology where they are more likely to be successful. Antisocial individuals are better at navigating social conflict than others (Sell, Tooby, and Cosmides Reference Sell, Tooby and Cosmides2009). If they can stir up conflict, they create a context in which they have an advantage.

Our aim is to measure the destructive mindset that emerges from the combination of dominance dispositions and social marginalization. We call this mindset the Need for Chaos and define it as a desire for a new beginning through the destruction of order and established structures. We chose the term “chaos” to reflect both the modern meaning of disorder and the original Greek meaning where it refers to a state from which order is produced. Need for Chaos is thus a mindset to gain status by disrupting the established order. We use the term “need” in the same way as it is used in studies of other psychological individual differences—that is, as something that “…directs behavior towards a goal and cause tension when this goal is not attained” (Cacioppo and Petty Reference Cacioppo and Petty1982, 117). It is not a biological need in the sense of thirst or hunger. Rather, the underlying evolutionary problem that the desire for chaos expresses relates to status maintenance and acquisition. Furthermore, the Need for Chaos is not a stable personality trait, but is expected to emerge in the interplay of environments related to personal status loss and traits related to dominance. It is thus a “characteristic adaptation” (McAdams and Pals Reference McAdams and Pals2006, 208–9) that matches behavioral tendencies in dispositions to a specific context.

We contend that individuals who are high in Need for Chaos share hostile rumors as a way to destabilize the established political system. As discussed above, these individuals may believe that sharing hostile rumors mobilizes like-minded others and creates confusion among those with the status they seek. In addition, the normative transgressions involved in the sharing of offensive and outrageous information may in itself be seen as attempts to assert the dominance that people high in Need for Chaos so strongly crave.

OVERVIEW OF RESEARCH QUESTIONS AND STUDIES

Empirically, we address the two key research questions outlined above: First, to what extent do extreme anti-systemic sentiments encapsulated in a Need for Chaos predict the sharing of hostile political rumors over and beyond partisan motivations? Second, does the Need for Chaos emerge through the combination of dominance-oriented dispositions and marginalization? We address these research questions through six empirical tests. The empirical basis is eight survey studies (combined N = 10,921) conducted in the United States in the period of February 2018 to February 2022. Table 1 provides an overview of the surveys, and Section S1 of the Supplementary Material (SM) details the sample characteristics and procedures.

Table 1. Overview of Studies

Note: The number of respondents in the Sample size column indicates those who consented to participate. In some studies, the actual number of respondents used for analysis will be slightly lower due to failed (preregistered) attention checks and missing values on key variables (see Section S1 of the SM for details).

The first three empirical tests focus on the Need for Chaos as an alternative motivation for hostile rumor sharing over and beyond partisan animosity. Test 1 develops and validates a one-dimensional measure of chaotic sentiments to tear down the political system—the Need for Chaos. Test 2 marshals data from six studies to demonstrate the relevance of the Need for Chaos for understanding the spread of hostile rumors about political elites affiliated with the Democratic and Republican Party. We show that while party identifiers favor rumors targeting political opponents, individuals with high levels of Need for Chaos (“chaos-seekers”) are motivated to indiscriminately share hostile rumors targeting elites from both parties. These analyses also reveal that chaos-seekers share hostile rumors about both Democrats and Republicans, even if they report to identify with one of these parties.

Test 3 examines why a Need for Chaos motivates indiscriminate sharing of rumors, even against political elites with whom they share a partisan identity. Test 3 shows that chaos-seekers view hostile rumor sharing as a tool for mobilizing against all political elites they consider corrupt, whether from the in-party or the out-party. This is motivated by deeply negative views of voters and elites from both parties, including from their party, reflecting in part a feeling of “abandonment” by their own party.

The role of feelings of abandonment provides a bridge to our second research question on the causes of a Need for Chaos. In Test 4, we ask: Who needs chaos? In this test, we demonstrate that the Need for Chaos is pronounced among individuals with a combination of Status-Driven Risk-Taking, a key measure of dominance orientations (Ashton et al. Reference Ashton, Lee, Pozzebon, Visser and Worth2010), and feelings of marginalization.

In Test 5, we provide additional evidence on the type marginalization most likely to trigger a Need for Chaos. Specifically, we examine the association between Need for Chaos and perceptions of status gains and status losses at the level of groups and individuals, respectively. We find that a Need for Chaos is most strongly associated with perceptions of personal status losses.

In the final test, Test 6, we use these insights to shed light on the politics of contemporary American society and how a Need for Chaos is distributed across indicators of historically marginalized groups based on gender and race. We find that Need for Chaos is highest among racial groups facing historical injustice—in particular, Black males—reflecting their higher concerns about their societal standing. At the same time, consistent with notions of aggrieved entitlement among members of historically privileged groups, we show that white men react more aggressively than any other group to perceived status challenges. While white men do not feel highly status-challenged on average, they are more likely to seek chaos when they do. Test 6 thus shows that a desire to “burn it all down” may emerge from multiple pathways shaped by current societal cleavages, yet for different reasons.

Figure 1 presents an overview of the proposed theoretical model and how the empirical assessments of the individual parts of the model are distributed across tests.

Figure 1. Overview of the Relationship between the Theoretical Model and Six Empirical Tests

All studies were conducted in accordance with the Code of Conduct for Research Integrity of the primary investigator’s university and associated policies of the National Committee for Health Research Ethics. Studies were conducted online and involved voluntary and informed consent with the possibility of ending participation at any time. In Section S1 of the SM, we outline participant compensation. For studies conducted via a survey company, consent for participation and data processing was obtained by the company on the author’s behalf. For studies conducted via a crowd-sourcing platform, we obtained consent for participation and data processing ourselves. Given the potential sensitive nature of the topic of hostile political rumors, no studies in the main text used deception. Data and codes to reproduce all results are available on Dataverse (Petersen, Osmundsen, and Arceneaux Reference Petersen, Osmundsen and Arceneaux2023).

TEST 1. DEVELOPING THE NEED FOR CHAOS SCALE

The aim of Test 1 is to develop the Need for Chaos scale. To establish the scale, we use survey data from a socially diverse sample of 1,004 American adults (“Study 1” of Table 1). The survey fielded an initial pool of 11 items intended to reflect the central features of the Need for Chaos scale: a desire for upending social institutions, tearing down established political hierarchies, and rebuilding society from scratch. Participants were asked on seven-point scales whether they disagreed or agreed with each item. We then conducted an exploratory factor analysis to identify the latent construct(s) underlying this battery of measured items, anticipating that the items mapped onto a one-dimensional construct of chaotic motivations. The details of the analysis are presented in Section S2 of the SM.

Results

Table 2 presents the final items used to measure the Need for Chaos (NFC_Chaos for short). Based on item clarity, item loadings, and cross-loadings from an exploratory factor analysis, we retained seven of the items that most comprehensively described the one-dimensional structure of the data (i.e., items in bold font in Table 2). This one-dimensional solution produced an eigenvalue of 4.53 and explained 65% of the shared variance (see Table S3 in the SM). As Table 2 shows, the items making up NFC_Chaos tap disruptive political sentiments—for example, “When I think about our political and social institutions, I cannot help thinking ‘just let them all burn’”—as well as general tendencies toward destruction—for example, “Sometimes I just feel like destroying beautiful things.” The final scale displayed high reliability (α = 0.90).

Table 2. Final Items of the Need for Chaos Scale

Note: The items in bold are used to construct NFC_Chaos. Reverse-coded items in italics are added in construction of NFC_Chaos-R.

The Need for Chaos captures motivations to tear down society’s political order, but how commonplace are such sentiments? Figure 2 shows the distribution of NFC_Chaos, scaled to range from 0 to 1, in the socially diverse Study 1. The measure skews toward the low end of the distribution, with a mean value of 0.17 and a standard deviation of 0.22. In other words, most participants accept the current political system. Nonetheless, a nontrivial minority of participants place themselves toward the scale’s higher end, expressing “some” to “strong” agreement with many of the statements. The skewed nature of the distribution of the Need for Chaos presents analytical challenges (e.g., outliers), which we deal with in subsequent tests.

Figure 2. Distributions of Need for Chaos and Alternative Measure

Note: The figure displays histograms of Need for Chaos (left) and alternative NFC_Chaos-R measure with two reverse-coded items (right).

All items of NFC_Chaos are worded such that agreement reflects strong disruptive motivations. This question format may induce acquiescence bias in which some participants tend to agree with statements irrespective of content. Study 2 therefore developed a revised measure of the Need for Chaos (NFC_Chaos-R) that includes two reverse-coded items on the top of the original items (presented in italics in Table 2 and used in Studies 2, 7, and 8). Section S2 of the SM describes the construction of NFC_Chaos-R, whereas the right-hand panel of Figure 2 plots its distribution. Reassuringly, the distributions of the two versions are very similar.

Supporting the distinct contribution of the Need for Chaos measure, analyses presented in Section S2 of the SM (6–9) reveal that the Need for Chaos can be reliably distinguished from a range of measures of antisocial dispositions (e.g., psychopathy and social dominance orientation). Finally, the analyses presented in Section S3 of the SM, Figure S3 in the SM, and associated tests provide evidence for the measurement stability of the scale.

TEST 2. THE NEED FOR CHAOS AND HOSTILE POLITICAL RUMORS

Test 1 established the Need for Chaos scale. Test 2 examines NFC_Chaos as a predictor of people’s beliefs and motivations to share hostile political rumors. Our claim is that chaos-seeking individuals will endorse vilifying rumors about political elites because they have grown disillusioned with the political system. In American politics, the Democratic and Republican Parties are the established parties within mainstream politics. We therefore expect a strong Need for Chaos to predict endorsement of rumors about politicians from both the Democratic and Republican parties.

This expectation differs from accounts focused on partisan motivations. These accounts predict that in polarized political contexts, partisans—due to warm feelings toward their party and/or animus toward the out-party—will accept rumors targeting members of opposition parties, but reject rumors reflecting badly on their own party. From a partisan perspective, the party affiliation of the rumor target should significantly influence the strength and direction of the association between partisanship and rumor uptake.

Results

Is the Need for Chaos associated with motivations to believe and share hostile political rumors? We examine this question in Figure 3. The figure shows estimated coefficients from models that regress motivations to believe (left panel) and share (right panel) hostile rumors about Democratic (black) and Republican (gray) elites on the Need for Chaos, partisanship, and a set of covariates (education, gender, income, age, and race).Footnote ² Figure 3 displays the coefficients separately for each study. The top panels show the association between Need for Chaos and rumor endorsement, whereas the lower panels show the association between partisanship and rumor endorsement. Tables S8–S13 in the SM present the regression tables.

Figure 3. Estimated Regression Coefficients from Models That Examine the Relationship between the Need for Chaos (Top) and Partisanship (Bottom) and the Motivation to Believe (Left) and Share (Right) Hostile Political Rumors about Democratic (Black) and Republican (Gray) Elites

Note: The horizontal bands give 95% confidence intervals. Dependent variables scaled to range from 0 to 1, with higher values indicating greater belief and intention to share the rumors. The Need for Chaos has been z-scored. Partisanship is a dichotomous variable (0 = Republican Identifier; 1 = Democratic Identifier). All models adjust for education, gender, income, age, and race.

As expected, across the various samples, chaos-seeking participants report a significantly greater acceptance of the rumors. Taken as a whole, when the Need for Chaos increases by one standard deviation, intentions to share hostile rumors increase by approximately 10 percentage points of the outcome scale (ranging from 4 in Study 3 to 19 in Study 6; all p’s < 0.05). The increase in rumor beliefs is slightly smaller—about seven points across samples (ranging from 2 percentage points in Study 4 to 15 percentage points in Study 6)—but it remains statistically significant and in the direction of our expectations. Figure 3 also supports the argument that disruptive sentiments fuel animosity of politicians regardless of their party affiliation. Accordingly, participants with a high Need for Chaos are more likely to endorse rumors targeting both Democrats and Republicans.Footnote ³

What about partisanship? The lower panels of Figure 3 reveal that Democratic respondents generally endorse anti-Republican rumors, but are less persuaded by rumors reflecting badly on their in-party. Among Republican identifiers, the opposite pattern of results occurs. These results confirm a partisan story of rumor endorsement. However, they come with a caveat: in contrast to the Need for Chaos, partisanship matters more for rumor beliefs than intentions to share the rumors on social media. Believing is a private act, whereas sharing is a public—and therefore more activistic—act. Furthermore, the absolute size of the coefficients when averaging across rumors appears to be roughly similar for the Need for Chaos and partisanship. Together, these observations suggest that the Need for Chaos and partisanship may be equally strong, yet different drivers of nonnormative forms of political activism.

A final question concerns the interplay between partisanship and the Need for Chaos. Theoretically, a strong party bond could dampen the inclination to share maligning rumors about one’s in-party, even among chaos-seeking citizens. On the other hand, chaos-seekers may feel so politically disillusioned that they will share harmful rumors even about the party with which they nominally identify. To examine this question, Figure 4 splits the top panels of Figure 3 by respondents’ party identification. The figure supports the latter expectation: irrespective of party allegiances, a greater Need for Chaos is significantly associated with endorsement of both in-party and out-party rumors.

Figure 4. Estimated Regression Coefficients from Models That Examine the Relationship between the Need for Chaos and the Motivation to Believe (Left) and Share (Right) Hostile Political Rumors about Democratic (Black) and Republican (Gray) Elites among Democratic (Top) and Republican Identifiers (Bottom)

Note: The horizontal bands give 95% confidence intervals. Dependent variables scaled to range from 0 to 1, with higher values indicating greater belief and intention to share the rumors. The Need for Chaos has been z-scored. All models adjust for education, gender, income, age, and race.

TEST 3. WHY IS NEED FOR CHAOS ASSOCIATED WITH MOTIVATIONS TO SHARE RUMORS?

Test 2 demonstrated a consistent association between the Need for Chaos and motivations to share hostile political rumors. This association existed beyond partisan animosity and even extended to rumors targeting party elites with whom the sharer ostensibly identified. Test 3 investigates the goals that individuals high in Need for Chaos seek to accomplish by indiscriminately sharing hostile information targeting partisan elites.

Our theoretical framework contends that the sharing of hostile rumors is a social and instrumental act aimed at escalating social conflict by, for example, mobilizing like-minded others. Yet, there are multiple reasons for why people share rumors—even hostile ones—including, first, epistemic reasons oriented toward enlightening audiences about information believed to be true (Pennycook et al. Reference Pennycook, Epstein, Mosleh, Arechar, Eckles and Rand2021) and, second, for social but noninstrumental entertainment purposes as encapsulated by Buckels, Trapnell, and Paulhus’s (Reference Buckels, Trapnell and Paulhus2014) study on how “trolls just want to have fun.” Finally, sometimes misinformation is shared with the explicit goal of countering it—for example, by exposing or ridiculing those believing in it (Johansen et al. Reference Johansen, Marjanovic, Kjaer, Baglini and Adler-Nissen2022). This latter possibility opens for an alternative explanation for chaos-seekers’ indiscriminate sharing, which could be in line with the traditional partisan account: perhaps those high in Need for Chaos share rumors about their in-party not to hurt the in-party, but rather to expose the out-party’s lies. In contrast, our account argues that such behavior reflects disdain for all actors within the political system, even those with whom the chaos-seeker ostensibly identifies. In Test 3, we use data from Studies 4 and 7 to test all these accounts (see Section S4 of the SM for details).

Results

Do individuals high in Need for Chaos share rumors to escalate social conflict or for other purposes? To examine this, Figure 5 presents estimated coefficients from OLS models that regress responses to the questions about the four different potential motives for rumor sharing (0–1-coded) on NFC_Chaos (z-scored) and a set of sociodemographic covariates (see Table S15 in the SM for regression tables).Footnote ⁴ Consistent with findings on “trolling,” we find a significant association between NFC_Chaos and intentions to share hostile rumors to amuse friends (β _Amuse = 0.04, p < 0.001). In contrast, we find no evidence that people with disruptive intentions share rumors because they believe that the rumors are true—in fact, the coefficient is negative (β _Truth = −0.03, p < 0.001)—or because they believe that they could have major consequences (β _Consequence = 0.01, p = 0.36). Finally, corroborating our core theoretical expectation, we find that chaos-seekers share rumors to mobilize against disliked groups, Democrats and Republicans alike (β _Mobilize = 0.04, p < 0.01).

Figure 5. Estimated Regression Coefficients for Association between Need for Chaos and Four Motivations for Sharing Hostile Political Rumors: (1) because the Story Is True, (2) to Mobilize against Disliked Groups, (3) because the Story Has Major Consequences If True, and (4) to Amuse Friends

Note: The horizontal bands give 95% confidence intervals. Dependent variables scaled to range from 0 to 1. The Need for Chaos has been standardized. All models adjust for gender, age, education, and race.

Does this motivation to mobilize against a disliked group only apply to sharing hostile rumors that target the out-party, or does it also apply to rumors that target the in-party? Using data from Study 7, Figure 6 shows predictions from OLS models regressing motives for sharing rumors on NFC_Chaos-R, partisanship, the interaction between the two variables, and a set of covariates (see Table S17 in the SM for regression tables). The top panels of the figure plot the two motives for sharing hostile rumors about Democratic elites—that is, to mobilize against either corrupt Democrats (left) or lying Republicans (right)—among Democratic (black lines) and Republican (gray lines) identifiers across levels of NFC_Chaos-R (x-axis). The bottom panels present results for sharing anti-Republican rumors.

Figure 6. Predicted Values from Regression Models That Regress Motivations for Sharing Rumors about Democrats (Upper Panels) and Republicans (Lower Panels) on the Need for Chaos (X-Axis) among Democratic (Black Lines) and Republican (Gray Lines) Identifiers

Note: Dependent variables have been scaled to range from 0 to 1. The Need for Chaos has been z-scored. Histograms show the distribution of the Need for Chaos and the number of Democratic and Republican identifiers at each bin. All models adjust for gender, age, ethnicity, education, and the interactions between the covariates and the Need for Chaos. The predictions are made by setting covariates equal to their mean (continuous covariate) or median (categorical covariate) value.

Figure 6 shows that when the Need for Chaos is low, partisans of different stripes have widely different motives for sharing rumors. Consider first the anti-Democratic rumors (top panels). Here, Republican identifiers are much more motivated than Democratic identifiers to mobilize against corrupt Democratic politicians (left panel), whereas Democratic identifiers largely concentrate on exposing Republicans of propagating hurtful lies about their party (right panel). At the lowest level of Need for Chaos, these partisan gaps in motives amount to approximately 35 percentage points of the outcome scales. Moving to the bottom panels, we see a similar picture with efforts to mobilize against the out-party underpinning sharing of anti-Republican rumors. Here, Republican identifiers share rumors to expose the lying Democrats, whereas Democratic identifiers share rumors to organize opposition against corrupt Republicans. These are clear signs of politically motivated reasoning.

But what happens when the Need for Chaos increases? In each of the four models, the interaction between partisanship and NFC_Chaos-R was statistically significant (all p’s < 0.005). The result of this interaction is readily apparent in Figure 6: as the Need for Chaos grows stronger, partisanship gradually loses power and the partisan gaps shrink. In fact, when the Need for Chaos reaches its peak level, partisans of both stripes become equally—and strongly—willing to share rumors to hurt both parties. Consequently, for chaos-seekers, political sympathies toward political parties appear to matter little for sharing decisions; instead, what matters is that rumors can be used as an instrument to mobilize against the entire political establishment.

The above findings suggest that chaos-seekers’ motivations to share hostile political rumors about both parties—even the party with which they ostensibly identify—reflect a genuine disdain for both parties. In Section S4 of the SM (37–40), we provide additional analyses demonstrating that partisans with strong chaotic motivations feel explicit hostility toward political actors—both elites and rank-and-file voters—from both parties. Furthermore, the analyses show that this reflects a feeling of being abandoned: even if they feel close to a particular party, they feel that both party elites and other ordinary party members have turned their back on them and, as a consequence, they want to punish them by sharing denigrating rumors.

TEST 4. WHO NEEDS CHAOS?

Tests 1–3 addressed our first research question: does Need for Chaos shape motivations to indiscriminately share hostile political rumors, above and beyond partisanship? The answer was affirmative and provided evidence that chaos-seekers actively seek to mobilize others against all actors within a loathed political system. In Tests 4–6, we turn toward our second research question: does Need for Chaos—as a characteristic adaptation—emerge from the combination of a dominance-oriented personality and social experiences of personal status loss? Test 4 provides an initial test with a focus on broad experiences of marginalization, whereas Tests 5 and 6 differentiate between distinct types of marginalization.

Results

Is the Need for Chaos associated with dominance orientations and social marginalization? Yes. Figure 7a displays estimated coefficients from models that regress the NFC_Chaos on status-driven risk-taking, perceived social ladder placement and loneliness (see Table S20 in the SM for regression tables). These models show that NFC_Chaos is significantly higher among participants who readily take risks to obtain status and among participants who feel lonely. In contrast, participants believing they belong at the top of the social ladder are significantly less willing to express chaotic motivations. Collectively, these findings support our characterization: chaos-seekers strive for status and feel rejected by friends and society as a whole.Footnote ⁵

Figure 7. (a) Estimated Regression Coefficients with 95% Confidence Intervals from Models That Regress Need for Chaos on Status-Driven Risk-Taking, Loneliness and Perceived Social Status without Interaction Terms. (b,c) Marginal Effect of Status-Driven Risk-Taking on Need for Chaos, Conditional on Perceived Social Status and Loneliness

Note: The hollow circles give marginal effects from a binning estimator that discretizes the data into five equally sized bins (Source: Hainmueller, Mummulo, and Xu Reference Hainmueller, Mummolo and Xu2019). The models adjust for gender, age, educational level, and race. In all panels, the Need for Chaos has been scaled to range from 0 to 1, whereas the independent variables have been z-scored.

But how does an orientation toward status interact with social marginalization? Figure 7b,c shows the interaction between status orientations, on the one hand, and perceived social status and loneliness, on the other hand. These panels—where the y-axes display the marginal effect of status-driven risk-taking—allow us to determine whether the association between the Need for Chaos and status aspirations grow stronger as people feel more socially marginalized. In the panels, the Need for Chaos is 0–1-coded, whereas our measures of status aspirations and social marginalization are z-scored.

We find support for this prediction. The association between status aspirations and NFC_Chaos becomes significantly stronger as perceived loneliness increases (β _{Status-Driven Risk-Taking × Loneliness} = 0.06, p < 0.001), whereas the association becomes weaker as people’s perceived placement on society’s ladder grows stronger (β _{Status-Driven Risk-Taking × Social Ladder Placement} = −0.05, p < 0.001). These results suggest that status aspirations only lead to extreme political discontent among people who feel ostracized; among participants who perceive themselves to be at the top of society’s hierarchy, or do not feel lonely at all, the association between Status-Driven Risk-Taking and chaotic motivations disappears. As such, the findings highlight the importance of considering both dispositional traits as well as situational triggers of destructive political motivations.

TEST 5. THE NEED FOR CHAOS AND DISTINCT TYPES OF MARGINALIZATION

Test 4 demonstrates that a Need for Chaos is activated when status-oriented individuals feel socially marginalized. As theoretically discussed, however, these situational status challenges can take many forms. First, marginalization could reflect a loss of status or the inability to gain status. Second, unmet expectations can relate to personal status expectations or to expectations about the status of one’s group. Whereas Test 4 did not distinguish between these different forms of status challenges, the aim of Test 5 is to directly examine how status losses and gains at the personal and group levels, respectively, predict a Need for Chaos. As discussed above, there are theoretical reasons to expect that personal status losses are particularly likely to fuel the form of extreme dissatisfaction that triggers a Need for Chaos.

Results

We regressed the NFC_Chaos-R on the four status measures as well as a set of sociodemographic covariates (age, gender, education, and income). Figure 8 displays estimated coefficients from these models (see Table S22 in the SM for regression tables). It shows that the NFC_Chaos-R is most strongly associated with worries about losing one’s own position in the social hierarchy (β _{Personal Status: Loss} = 0.05, p < 0.001) and—to a lesser, but still significant extent—the perception that one is personally being kept back from climbing the social status ladder (β _{Personal Status: Gain} = 0.02, p < 0.001). In this model, where all four status measures are entered simultaneously, considerations about status experienced as part of a group seem far less influential: the association between the NFC_Chaos-R and perceptions about one’s group’s inability to gain status is almost nil (β _{Group Status: Gain} = −0.00, p = 0.86), whereas considerations about group-based status loss may be associated with less willingness to disrupt the system (β _{Group Status: Loss} = −0.02, p < 0.001).Footnote ⁶ This latter finding is consistent with work suggesting that inflated perceptions of the ingroup’s worth, rather than the opposite, is associated with extreme behavior (Golec de Zavala and Lantos Reference Golec de Zavala and Lantos2020).

Figure 8. Estimated Regression Coefficients with 95% Confidence Intervals from the Model That Regress Need for Chaos on Four Measures of Status

Note: The models adjust for gender, age, educational level, and race. Need for Chaos has been scaled to range from 0 to 1, whereas the status measures have been z-scored.

Taken together, the results indicate that personal status concerns—especially the threat of losing status—motivate citizens to disrupt the status hierarchy. When people are confronted with threats to their own position in the hierarchy, they may endorse chaotic policies as a defense mechanism to regain a sense of dominance.

TEST 6. HISTORICAL PRIVILEGE AND THE NEED FOR CHAOS

Test 5 demonstrated that chaos-seeking is linked to personal status concerns and, especially, status losses. In Test 6, we ask whether the framework of Need of Chaos—and the central role played by status concerns—helps shed light on contemporary political cleavages. In particular, we focus on the intersection of two of the most potent sociodemographic fault lines in current U.S. politics—gender and race—and whether and how gender and race shape anti-systemic sentiments. From the outset, it is clear that status challenges are a central background of political activism at the intersection of gender and race. For example, overt and subtle forms of discrimination have limited the opportunities for upward social advancement among women and Black people and the resulting frustrations have been fueling political activism, for example, in the context of the Black Lives Matter movement. Furthermore, arguments about status loss are also central to populist rhetoric about being marginalized, which has been shown to resonate with members of traditionally high-status groups such as white people and men (Mutz Reference Mutz2018). These different feelings of disenfranchisement could ultimately lead both historically disadvantaged and historically privileged groups to endorse overthrowing a system that denies them the status and respect to which they feel entitled (whether justified or not). Such multiple pathways to destructive sentiments may help explain the combustibility of current U.S. politics.

On this background, Test 6 is an exploratory analysis of how the intersection of gender and race shapes levels of the Need for Chaos as well as levels of the different status concerns that underlies the Need for Chaos. Furthermore, we explore how the intersection of gender and race shapes the predictors of Need for Chaos, specifically by examining whether different status concerns are stronger predictors for some groups than others.

Results

The left-hand panel of Figure 9 displays the average levels of Need for Chaos across groups defined by gender and race. The figure reveals a small but clear set of differences in the felt Need for Chaos (see also Table S23 in the SM). White women have, on average, the lowest score of Need for Chaos (M = 0.22, 95% CI = [0.20–0.24] on a 0–1 scale), followed by white men (M = 0.25, 95% CI = [0.23–0.27]), Black women (M = 0.28, 95% CI = [0.26–0.30]) and Black men (M = 0.30, 95% CI = [0.28–0.33]). This initial observation suggests that a Need for Chaos is particularly related to race-based historical disadvantages and less so to gender-based, potentially because of lower general aggressiveness among women (Bartusevičius, van Leeuwen, and Petersen Reference Bartusevičius, van Leeuwen and Petersen2020). In the following set of analyses, we explore whether status concerns can help us understand this pattern.

Figure 9. Average Level of the Need for Chaos (Left) and Status Concerns (Right) among Sociodemographic Subgroups

Note: Status Concerns and the Need for Chaos have been scaled to range from 0 to 1. The horizontal bands give 95% confidence intervals.

The right-hand panel of Figure 9 displays the average levels of each of the four status concerns from Test 5 across the four demographic groups (see also Table S24 in the SM). On each and every measure of status concern, both Black women and Black men score substantially and significantly higher than white women and men. This pattern is particularly evident for those measures that assess group-based status challenges, but it is also the case for the key driver of a Need for Chaos, feelings of personal status loss. These findings suggest that both experiences of racism not only provoke feelings among Black individuals of being held back relative to other groups in American society, but they also fuel anti-systemic sentiments by generating a fear of losing the gains that Black people have achieved compared to previous generations.

Figure 10 examines status concerns as predictors of the Need for Chaos across the intersection of gender and race (see Table S25 in the SM for interaction models). While the levels of status concern illuminate the higher Need for Chaos among Black individuals, Figure 10 provides a deeper understanding of chaos-seeking among white men. As is clear, the association between status concerns and the Need for Chaos is stronger among white men compared with any other group. The differences between white men and all other groups are significant for all status concerns, except personal status loss. While group-based marginalization tempers a Need for Chaos among Black individuals, group-based feelings of being unable to advance in society fuels a Need for Chaos among white men. Consistent with notions of aggrieved entitlement among historically dominant groups (Kimmel Reference Kimmel2017), many white men are preoccupied with their societal standing and react with aggression against any threat. This also suggests that the average level of Need for Chaos among white men in Figure 9 indicates that many white men are quite low in Need for Chaos, but that some—specifically, those feeling threatened in terms of their status—harbor strong chaotic sentiments.

Figure 10. Estimated Regression Coefficients for the Association between Status Concerns and the Need for Chaos among Demographic Subgroups

Note: The Need for Chaos has been scaled to range from 0 to 1, whereas the four measures of status concerns have been z-scored. The horizontal bands give 95% confidence intervals.

For all other groups, the association between status concerns and the Need for Chaos is roughly comparable. In tandem with the observation that both Black men and Black women feel equally threatened in terms of their status, this raises the question of why Black women display a lower level of Need for Chaos. In the SM, we examine another measure: reported fear of opposing the system (i.e., engaging in anti-system actions). Consistent with their historical marginalized position, we find that Black individuals, independently of gender, experience significantly higher fear of opposing the system than white individuals (see Figure S6 in the SM). Importantly, however, for Black women, but not Black men, this fear of opposing the system correlates negatively with a Need for Chaos (see Table S26 in the SM).

In sum, Test 6 suggests that there are multiple routes to having a high Need for Chaos. While both Black men and women feel marginalized, Black women’s Need for Chaos is inhibited by fear, whereas Black men’s marginalization more readily translates into a high Need for Chaos. In contrast, white men—experiencing both less marginalization and less fear of opposing the system—express a Need for Chaos because of extreme reactions to any perceived status threat including, for example, the expansion of racial and gender equality. While chaos-seeking unites these different groups, it is important to note that they are political adversaries rather than political allies and their goal is to address the specific status challenges they personally face. This may help explain the instability of current U.S. politics: dissatisfaction with the current system is widespread, but comes from diametrically opposed challenges.