Antisemitism is predicted by anti-hierarchical aggression, totalitarianism, and belief in malevolent global conspiracies

The study investigates which factors predict both ‘old’ (Judeophobic) and ‘new’ (antizionist) forms of antisemitism, moving beyond the traditional left-right political spectrum. Prior work shows anti-Israel and anti-Jewish attitudes are correlated, but it is unclear what underlying factors drive both. Given changing expressions of antisemitism across political contexts, the authors hypothesize that specific ideological traits—rather than self-placement as left or right—may be key predictors. Candidate predictors grounded in political psychology and prior scholarship include Social Dominance Orientation (SDO), Right- and Left-Wing Authoritarianism (RWA/LWA; especially anti-hierarchical aggression), conspiracy beliefs (especially malevolent global conspiracies), support for totalitarian government, ethnic nationalism, and dispositions like political/social cynicism and trust. Demographic predictors (age, gender, education, ethnicity) are also examined given mixed prior findings.

• Antisemitism manifests in both traditional Jew-focused prejudice and in antizionist expressions; multiple studies report significant associations between measures of anti-Jewish and anti-Israel attitudes (e.g., ADL 2023; Allington et al. 2022; Baum & Nakazawa 2007; Kaplan & Small 2006; Staetsky 2017, 2020). The IHRA definition encompasses both.
• Political alignment alone is an imprecise predictor; pockets of both antisemitism and opposition to it exist on both far left and far right (Staetsky 2020). Populist conspiracy narratives on left and right may facilitate antisemitism (Bolton & Pitts 2018; Elman 2022).
• Social Dominance Theory suggests prejudice may stem from preferences for group-based hierarchy (SDO), while RWA captures support for existing hierarchies; prior studies found mixed links to antisemitism (Frindte et al. 2005; Swami 2012; Ekehammar et al. 2004).
• LWA, defined as enthusiasm for overturning existing social orders, may relate to antisemitism; prior UK work links sympathy for violent extremism and desire to overthrow capitalism to antisemitism (Staetsky 2017, 2020).
• Conspiracism historically aligns with antisemitism; mixed prior findings on generic conspiracy belief predicting antisemitism motivate re-testing (Swami 2012; Allington & Joshi 2020; Byford 2011).
• Support for totalitarianism and ethnic nationalism are plausible predictors given historical regimes’ antisemitism and exclusionary nationhood projects; empirical testing as predictors has been sparse.
• Demographics: prior work finds associations with age, education, and higher antisemitism among certain minority groups; mixed findings on gender and education (ADL 2011; Weil 1980, 1985; Greene & Kingsbury 2017; Greene et al. 2021; Jikeli 2015; Staetsky 2017, 2020).

Design: Two cross-sectional surveys in the UK.

• Study 1: Self-selecting sample via Prolific (n=809; Oct 30–31, 2020), quotas for age (18–25 vs 26+) and gender (male/female), UK-resident. Bivariate analyses of demographics and a broad set of ideological scales with GeAs and its subscales (JpAs, AzAs). Scales normalized to 0–1.
• Study 2: Representative YouGov panel sample (n=1853; Dec 16–17, 2020), UK-resident, demographic quotas; YouGov weights applied in linear models (univariate/bivariate unweighted). Selected top non-demographic predictors from Study 1 (Totalitarianism; belief in malevolent global conspiracies; LWA anti-hierarchical aggression [abbreviated to 6 items]) plus demographics. Multivariable linear models with standardized predictors assessed GeAs, JpAs, and AzAs. Measures:
• Antisemitism: Generalised Antisemitism (GeAs) scale with two subscales: Judeophobic Antisemitism (JpAs; ‘old’ antisemitism) and Antizionist Antisemitism (AzAs; ‘new’ antisemitism). Prior validation indicates a common latent factor plus subtraits (Allington et al., 2022b).
• Ideological/psychological scales: Generic Conspiracist Beliefs (GCB: malevolent global, personal wellbeing, government malfeasance, extraterrestrial, information control; Brotherton et al., 2013); Totalitarianism (McClosky & Chong, 1985); Ethnic Nationalism (adapted BES 2017 items incl. ancestry, Christianity, being white); Political Cynicism, Social Cynicism, and Political Trust (Pattyn et al., 2012); Social Dominance Orientation (SDO7: Dominance, Anti-egalitarianism; Ho et al., 2015); Sympathies for Radicalisation (SyFoR; modified; Bhui et al., 2014); Right-Wing Authoritarianism (RWA: submission, aggression, conventionalism; Duckitt et al., 2010); Left-Wing Authoritarianism (LWA: anti-hierarchical aggression, anti-conventionalism, top-down censorship; Costello et al., 2021). Demographics: age (continuous), gender (operationalized female=1; other=0), ethnicity (other-than-white=1; white=0), education (degree=1; non-degree=0). Self-placement on left-right 7-point scale reported descriptively but not used as predictor. Analytic approach:
• Study 1: Welch t-tests for binary demographic predictors; Pearson correlations (r) for age and for scale predictors with GeAs, JpAs, AzAs. Power analyses indicated adequate power for moderate effects; r≥0.10 considered reportable.
• Study 2: Bivariate tests as in Study 1. Nested linear models with standardized predictors tested demographics alone (Model I), then adding Totalitarianism (Model II), GCB malevolent global (Model III), LWA anti-hierarchical aggression (Model IV), and the full model with all three scales plus demographics. Effect sizes via r², adjusted r², Cohen’s f². Weights applied to models. Code and data available via OSF (https://osf.io/9p2f8/).

Study 1 (n=809, bivariate):

• Demographics: Other-than-white ethnicity strongly associated with higher antisemitism (GeAs d=0.65 [0.45,0.84]; JpAs d=0.46 [0.27,0.65]; AzAs d=0.60 [0.40,0.80]; all p<0.001). Female gender associated with lower JpAs (d=-0.43 [−0.56,−0.29], p<0.001), marginally higher AzAs (d=0.14 [0.00,0.28], p=0.051), yielding small lower GeAs (d=-0.18 [−0.32,−0.05], p=0.009). Degree education weakly negatively associated with GeAs (d=-0.21 [−0.35,−0.08], p=0.003), driven by JpAs (d=-0.31 [−0.45,−0.17], p<0.001). Age weakly negatively correlated with GeAs (r=-0.11 [−0.18,−0.04], p=0.001) via AzAs (r=-0.18 [−0.24,−0.11], p<0.001); JpAs unrelated to age (r=-0.01).
• Ideological scales with strongest correlations to GeAs (r≥0.30): Totalitarianism r=0.34; GCB malevolent global r=0.34; LWA anti-hierarchical aggression r=0.32; GCB personal wellbeing r=0.30; GCB government malfeasance r=0.30 (all p<0.001). Totalitarianism and GCB global/personal predicted JpAs more than AzAs; LWA anti-hierarchical aggression and GCB government malfeasance predicted AzAs more than JpAs. Ethnic nationalism strongly predicted JpAs (r=0.37) but not AzAs (r≈-0.08), thus weak for GeAs (r=0.18). SDO factors were weak predictors.
• RWA factors positively correlated with JpAs but negatively with AzAs; LWA anti-conventionalism and top-down censorship were positively associated with AzAs and negatively (or not) with JpAs; LWA anti-hierarchical aggression was positively associated with both JpAs and AzAs.

Study 2 (n=1853):

• Bivariate correlations with GeAs: LWA anti-hierarchical aggression r=0.39; GCB malevolent global r=0.36; Totalitarianism r=0.31 (all p<0.001). With JpAs: 0.32, 0.38, 0.40; with AzAs: 0.34, 0.22, 0.09 respectively.
• Demographics (bivariate): Other-than-white ethnicity: GeAs d=0.75 [0.51,0.99], JpAs d=0.65 [0.40,0.89], AzAs d=0.61 [0.36,0.87] (all p<0.001). Female: JpAs d=-0.23 [−0.32,−0.13], AzAs d=0.27 [0.18,0.37]; GeAs d=0.01 (ns). Degree: GeAs d=-0.45 [−0.55,−0.35], JpAs d=-0.57 [−0.67,−0.48], AzAs d=-0.16 [−0.26,−0.05]. Age: AzAs r=-0.15 (p<0.001), GeAs r=-0.08 (p=0.001), JpAs r=0.01 (ns).
• Linear models: • Partial Model I (demographics): GeAs adj r²=0.08 (f²=0.09). Age (β=-0.12), degree (β=-0.53), other-than-white ethnicity (β=0.52) significant; female small negative on GeAs; for subscales, female negative JpAs (β=-0.30) and positive AzAs (β=0.18); age negative AzAs (β=-0.14); degree negative both; ethnicity positive both. • Partial Model II (+Totalitarianism): GeAs adj r²=0.16; Totalitarianism predicts GeAs (β=0.28) and JpAs (β=0.36), weakly AzAs (β=0.09). Gender effect on JpAs disappears after controlling for Totalitarianism. • Partial Model III (+GCB malevolent global): GeAs adj r²=0.18; GCB global predicts GeAs (β=0.32), JpAs (β=0.33), AzAs (β=0.20). Ethnicity and education effects attenuate. • Partial Model IV (+LWA anti-hierarchical aggression): GeAs adj r²=0.22; LWA AHA predicts GeAs (β=0.39), JpAs (β=0.33), AzAs (β=0.32); age and gender effects on GeAs become ns. • Full Model (all three scales + demographics): GeAs adj r²=0.29 (f²=0.40, large). Predictors (β, p<0.001 unless noted): LWA AHA 0.29; GCB global 0.19; Totalitarianism 0.17; ethnicity 0.43; degree -0.25; age and female ns. JpAs adj r²=0.32: Totalitarianism 0.27 strongest; GCB global 0.21; LWA AHA 0.21; female -0.30; degree -0.25; ethnicity 0.41. AzAs adj r²=0.16: LWA AHA 0.29 strongest; GCB global 0.11; Totalitarianism ns (β=-0.01); female 0.18; age -0.08; degree -0.11; ethnicity 0.30.

Findings indicate that antisemitism is less about left-right self-identification and more about specific ideological traits. A conspiracist worldview (belief in malevolent global conspiracies), a desire to overturn the social order (anti-hierarchical aggression), and support for authoritarian forms of government (totalitarianism) robustly predict antisemitism. Anti-hierarchical aggression predicts both ‘old’ and ‘new’ antisemitic attitudes, including in multivariable models, suggesting a broad link to the latent trait of generalized antisemitism. Totalitarianism is specific to Judeophobic attitudes and does not predict antizionist antisemitism once controls are applied. Conspiracy belief predicts both subtypes but is comparatively stronger for Judeophobic antisemitism than for antizionist antisemitism in multivariable models. RWA facets align positively with Judeophobic but negatively with antizionist antisemitism; LWA facets (anti-conventionalism, top-down censorship) show the reverse, highlighting the complexity behind ‘left’ and ‘right’ labels. Demographically, other-than-white ethnicity is consistently associated with higher GeAs, JpAs, and AzAs; degree-level education associates with lower antisemitism (especially JpAs); age relates negatively only to AzAs. Gender shows opposite associations for JpAs (lower among women) and AzAs (higher among women), canceling out in GeAs. Overall, results suggest that would-be totalitarians, conspiracists, and those desiring revolutionary social change—found across the political spectrum—are most likely to hold antisemitic attitudes.

Across two UK samples, antisemitism is predicted most strongly by anti-hierarchical aggression, with additional contributions from belief in malevolent global conspiracies and support for totalitarian government. Totalitarianism is linked to ‘old’ antisemitism only, whereas anti-hierarchical aggression and conspiracist beliefs predict both ‘old’ and ‘new’ forms. Demographic effects—particularly higher antisemitism among other-than-white ethnic groups and lower antisemitism among degree-holders—replicate prior findings; the pattern of lower JpAs but higher AzAs among women appears novel. The study advances measurement by focusing on specific ideological dispositions rather than broad left-right identity, providing actionable insight for research and policy. Future work should broaden predictors (e.g., additional conspiracy domains, ethnic nationalism, RWA/LWA subfactors) in multivariable models, test causal pathways (e.g., path models), and use alternative antisemitism measures (e.g., experimental scenarios) to address social desirability.

• Only three non-demographic predictors (totalitarianism, malevolent global conspiracy belief, anti-hierarchical aggression) were re-tested in multivariable models in Study 2 due to resource constraints; strongest AzAs predictors from Study 1 (e.g., LWA anti-conventionalism, government malfeasance conspiracies) were omitted, likely reducing explained variance for AzAs.
• Linear models may be suboptimal given two correlated subscales; path or structural models may better capture relations to the common latent factor and subtraits.
• Self-report agreement with antisemitic statements may be affected by social desirability and construct transparency, potentially varying by education level.
• Ethnicity analyses constrained by smaller non-white subsample (Study 1 power limitations) and wide CIs in Study 2.
• Education measured unidimensionally; field of study not distinguished, possibly masking countervailing effects.
• Left-right self-placement had substantial nonresponse; not used as predictor.
• Cross-sectional design prevents causal inference.