When discussing “collectivism,” many treat it as a singular cross-partisan construct: people who are more group-oriented rather than individualist, period, end of story. But the politics of “we” aren’t always so neat and uniform. On the contemporary right, collectivism can grow out of threat vigilance and group fusion, whereas on the left, it can arise from justice/trust worldviews and corrective norming. If we box those into a single factor, we risk diluting real differences in why people coordinate (‘collectively’) and how they enforce norms. Furthermore, not all ideological differences have straightforward relationships with ideology. It is possible to see similar scores at both more extreme ideological ends and a dip or rise among moderates relative to those ends. Different motives can produce similar levels of “we” at the extremes. 

Our guiding hypothesis is that human cooperation at the group level evolves through (at least) two significant attractors: binding under threat and coordinating under trust. Both are adaptive, both are moral, and both seem to have been essential for stable multilevel selection in our evolutionary history as a species. Our moral pluralism model treats them as complementary regulatory modes rather than as good vs. bad ideologies.

Evolutionary function	Defensive / Binding	Integrative / Coordinative
Selection pressure	External threat	Internal fairness optimization
Mechanism	Group fusion, punishment, conformity	Procedural trust, corrective norms
Constraint density & Information flow	Reduce entropy by narrowing permissible states (less diversity, more certainty)	Increase adaptive potential by expanding the system’s accessible configuration space while maintaining coherence.
Social phenotype	Parochial altruist	Reciprocal institutionalist
System property	Boundary tightening	Boundary permeability with control loops

Parochial cooperation is bounded: it invests in the welfare of “us” and often defines itself against “them.” Evolutionary models (e.g., Bowles & Gintis 2003) show that parochial altruism can evolve when between-group competition is strong—helping one’s in-group and harming out-groups both raise inclusive fitness at the group level.

• Adaptive context: high intergroup threat, small effective group size, low mobility.
• Mechanism: shared identity, fusion, punishment of defectors, reputation within the group.
• Typical psychology: loyalty, conformity, honor, retribution.
  

Reciprocal cooperation, by contrast, relies on cross-boundary exchange and iterated trust (Trivers 1971; Nowak 2006). It flourishes when repeated interactions, institutions, and information flows make defection costly and trust worthwhile.

• Adaptive context: lower intergroup threat, higher interdependence, dense networks.
• Mechanism: fairness norms, transparency, corrective procedures.
• Typical psychology: trust, forgiveness, proceduralism, cosmopolitan fairness.

Another hypothesis is that large-scale civilizations oscillate between two high level systemic regulation modes: defensive vs. integrative cooperation. When Parochialists are dominant, they tend towards defensive regulation. When Reciprocalists are dominant they tend towards integrative regulation. We view these regulation modes as functional phases in multilevel systems, not merely representative of political or moral stances. 

• Parochial collectivism activates when environmental signals indicate danger or scarcity. It tightens coordination, elevates conformity, and suppresses internal variation to preserve integrity (Think of it as a homeostatic immune response—high gain, low tolerance for error).
  
• Reciprocal collectivism activates when the environment appears stable enough to invest in exploration and cross-boundary integration. It increases diversity of inputs and extends trust outward (Functionally a developmental or morphogenetic mode—building new structure rather than merely defending existing form).
• 

Our recent survey research seeks to measure collectivism in ways that reflect these attractors. This approach is important because the mechanisms for collectivist motivations and subsequent behaviors are distinct: feeling threatened by outgroups is not the same psychological phenomenon as feeling the world should be more fair. Collapsing them hides patterning across ideology and weakens predictions.

With these collective behavior attractors in mind, we built two theory-driven formative composites of parochial and reciprocal collectivism. One that captures a binding/defensive style (right-leaning), and one that captures a coordinative justice style (left-leaning). Rather than forcing everything into a single latent factor, we assemble clearly defined ingredients (sub-composites) that each play a distinct theoretical role. The result is a set of measures that are interpretable,transparent, and predictive. 

Traditional (reflective) and more typical scales presume that items are interchangeable symptoms of one underlying factor and should all correlate strongly (high reliability/alpha). Our targets are different: they are multicomponent orientations. A person can be high on threat vigilance but moderate on fusion, or strong on corrective norm keeping but lower on justice/trust beliefs. These formative components are non-redundant causes of distinct collectivist styles, so we treat them as formative parts (and combine them in straightforward, unit-weighted row means after standardizing). 

Sub-composites: The Underlying Mechanisms of Collective Styles

We start with individual survey items (z-scored and direction aligned) and group them into sub-composites that represent distinct mechanisms. Then further downstream, we combine  sub-composites into the two broader collectivism measures.

Threat Vigilance

The first is Threat Vigilance. Created from items that ask about outgroup acceptance, the relationship with those who oppose one’s self-reported political-ideological values, and the acceptability of physical confrontation for political goals. Taken together, these items capture a participant’s sense of antagonism towards outsiders and their readiness to combine that antagonism with force. In theoretical terms, this sense of Threat Vigilance provides signals for the perception of threat in a social environment and a willingness to confront it, making it the starting point for a defensive form of collective cohesion. Specifically, it can be understood as appraising outsiders/opponents as harmful and endorsing forceful responses when necessary.

Measures:

• Perceived disruption from out-group presence
  • At a neighborhood barbecue/church gathering, someone with very different lifestyle/political values wants to join. “How do you feel about them joining the event? They would likely enhance … They would probably ruin the event.
• Readiness to neutralize opposed groups
  • “Foster mutual understanding with dialogue and cooperation … Eliminate their influence and presence, whatever it takes.” when dealing with political rivals.
• Acceptance of physical force in conflict situations
  • “Physical confrontations with authorities or opposing groups” is justified for political purposes.

Fusion (In-Group Identification)

A second sub-composite is Collective Identification (sometimes described as fusion). This measure is based on items related to group pride/identification and loyalty. These indicators are scored so that higher scores represent stronger feelings of unity with one’s group and greater loyalty to its members. Collective Identification captures a sense of “felt oneness” with one’s in-group, representing a sort of psychological glue that binds individuals together and channels energy into coordinated enforcement (especially in response to perceived threats). Our construct specifically captures how tightly one feels tied to their group; identifying when personal identity and group identity strongly overlap.

Measures:

• Strength of affective stake in group reputation (higher offense = stronger fusion)
  • “How offended would you feel if the reputation of a group you’re a part of (family, circle of friends, church, political party, etc.) were slandered in public?” (Very offended … Not offended at all)
• Prioritizing collective norms over personal ties
  • “When someone close to me in my community acts immorally, my loyalty to them is ___ compared to what is best for the community.” (reversed so higher = community over person)
    

Corrective Tightness

The third sub-composite is Corrective Tightness. The construct is built from items that tap into tendencies towards correction processes and warning frameworks (whether outgroup members in a social setting require confrontation by addressing them directly or with a private formal warning). People who score higher on Corrective Tightness demonstrate a preference for clear rules and procedural steps that correct deviance before it escalates into harsher forms of sanction. This sub-composite reflects an orientation toward pre-emptive clarity, emphasizing procedures that keep the group aligned without immediately resorting to severe punishment. Our specific construct measures the preference to retain and repair members: to address, warn, and reintegrate before escalating.

Measures:

• “Their actions confuse other members and need to be addressed.”
  
• “Give them a formal warning privately.”
  

Justice/Trust Worldview

Finally, we derive a measure of what we call a Justice/Trust Worldview, which includes items that concern beliefs in a safe world (perceiving people as trustworthy), perceptions of an unfair world (people generally don’t get what they deserve), rejection of coercive action towards in-group members for norm violations, and prosocial tendencies. Together, these items form a picture of a worldview where people trust others and expect to work together to hold others and institutions accountable to get more fair outcomes. High scores on this sub-composite suggest that individuals are more likely to legitimize corrective norms and restraint, because they believe that others will also respond in reciprocal ways once they are corrected. Our specific measure captures a prosocial, low-threat stance toward society—trusting strangers, sees the world as not perfectly just, tolerates non-action for minor issues, and personally contributes to collective goods.

Measures:

• generalized interpersonal trust
  • “Most people are very untrustworthy”
• sensitivity to unfairness/systemic luck (low “just-world”)
  • “People generally get what they deserve in life, whether good or bad.” (reverse-coded)
• benign tolerance/leniency when harm is minimal
  • “Not harmful to the group, no action is needed.”
• voluntary, costly community contribution
  • “I engage in activities that help build a sense of community, even when they don’t interest me personally.”
    

Combining Sub-Composites: The Two Faces of Collectivism

From these sub-composites, we build higher-order composites that represent distinct “faces” of collectivism.

The first is Parochial (“Right-wing”) Collectivism, which combines Threat Vigilance and Collective Identification. This represents a binding defensive path towards the collective “we”. We hypothesize that, for high scorers on this measure, outgroup threat salience leads to tightening the boundaries of belonging and to enforcing norms more strictly. Parochial Collectivism therefore reflects a form of collectivism rooted in defense and cohesion under perceived threat.

The second composite is Reciprocal (“Liberal”) Collectivism, which combines Corrective Tightness with a Justice/Trust Worldview. This composite represents a coordinative justice road to “we”. With liberal collectivism, solidarity does not emerge from fear of outsiders or binding through threat and instead arises from the conviction that the world is fundamentally safe and that people can be expected to behave fairly and prosocially. It is, however, reinforced by skepticism of a just world, the recognition that people do not always get what they deserve, paired with a belief in cooperation. Together, these orientations support the view that rules and procedures are legitimate guides for group life. Individuals who score highly on Liberal Collectivism therefore favor corrective norms and procedural enforcement, leaning on transparency and due process as the means of coordination rather than immediate punitive action.

How these composites sit across ideology

Once constructed, we examined how these composites vary across the four ideological buckets in our sample: Far Left, Moderate Left, Moderate Right, and Far Right (see our first blog report on collectivism and belief-based political-ideology measurement). At first glance, one might expect a simple linear story, with Parochial collectivism steadily climbing as we move rightward, and liberal collectivism steadily climbing as we move leftward. And in many respects, that is true, our measures do capture some predictable monotonic trends.

Figure 1: Subcomposite and Composite Monotonic trends across political ideology (where 1=far-left and 4 = far right)

But collectivist orientations, when built into our formative scales, are not always neatly linear. We compared linear models (collectivism increasing or decreasing steadily across ideology) to quadratic models (collectivism curving up or down at the ends). When a quadratic fit substantially improves model performance (ΔAIC > 2, significant quadratic term), we can conclude that the ideological distribution of that composite followed a horseshoe pattern. For example, Parochial collectivism shows significant curvature (quadratic term p = .002, ΔAIC = 8.1), with scores dipping among moderates and rising again at both ideological extremes. Liberal collectivism also shows a weaker, though still suggestive, quadratic pattern (p = .062), with a modest inverted-U shape. Additionally, some of the sub-composites that feed into these measures also showed strong curvature, Collective Identification (ΔAIC = 22.1), loyalty (ΔAIC = 13.9), and prosociality (ΔAIC = 19.4) among them, so that even at the level of the formative ingredients themselves there is sometimes bending, rather than relating linearly, across ideology. These results emphasize how the different mechanisms, though theoretically driven by ideological differences, can converge on similar levels of collective “we-ness” at the extremes, yielding quadratic patterns when incorporated into our composites.

Figure 2: Parochial and Reciprocal Collectivism across political-ideology.  Linear and quadratic models. 

In this discussion of multiple roads to “we”, empirical support suggests defensive binding and coordinative justice can both generate strong collectivist tendencies, but they do so for different reasons. Moderates, who are lower on threat-fusion but higher on justice-trust orientations, display weaker parochial collectivism but higher reciprocal collectivism. The result is a U-shaped horseshoe pattern for parochial collectivism and an upside-down horseshoe for reciprocal collectivism.

The astute reader will have noticed at this point, that the symmetric parabolic (horseshoe) pattern, contradicts our hypotheses that the parochial collectivism is more right wing, and the reciprocal collectivism is more left wing (though the data is so noisy that any conclusions might be premature, even though p says at least for the parochial collectivism, the data is statistically significant). What this quadratic dependence tells us instead, is that these two modes differ between center and tails on the L/R axis. Either the theory is wrong, or the way that we measured L/R is off, or the way we measured these two constructs (with equal weights of sub-components) is off.

From Measurement to Hypothesis Tests

With these formative composites that separate a binding defensive road to “we” (Parochial Collectivism: Threat Vigilance + Collective Identification) from a coordinative justice road (Liberal Collectivism: Corrective Tightness + Justice/Trust), we sought to address a second question: How do these mechanisms actually behave when they predict downstream enforcement and action? To answer that, we tested a series of hypotheses. 

Similar to our collectivism formative scales, we generated other formative scales from survey responses with which to test our collectivism measures.

• Punitive Tightness 
• Proxy for RWA
• Binding Socialization

Predisposition, not destiny

We asked whether an orientation toward authoritarian conservatism (measured with an abbreviated RWA-Proxy scale including authoritarian submission, government norm enforcement and preference for strictness/low-permissiveness) would predispose individuals toward more punitive enforcement, and whether this effect would diminish once more immediate threat cues were taken into account. The results provide some clear empirical support for this view. On its own, RWA-Proxy significantly predicted greater endorsement of punitive sanctions (b ≈ .22, p < .001), accounting for roughly 4% of the variance in punitive attitudes (R² ≈ .043). In other words, people higher in dispositional conservatism are more sanction-prone, consistent with previous of work on right-wing authoritarianism and norm enforcement. However, when situational predictors of Threat and Fusion were added to the model, the story sharpens. The effect of RWA-Proxy attenuates to b ≈ .16, while Threat emerges as the dominant predictor of punitive enforcement (b ≈ .26, p < .001). The explanatory power of the model more than doubled (R² jumping to ≈ .11).

Figure 3: Punitive Tightness Models. The left panel shows RWA-Proxy predicting Punitive Tightness. The right panel shows Threat Vigilance (with Fusion added to the model), demonstrating improved model fit. Each includes standardized coefficients and fitted regression lines with 95% confidence intervals.

These findings are consistent with our theoretical framing of the binding–defensive road to collectivism. RWA-Proxy sets a baseline predisposition: it tilts some individuals toward seeing punishment as legitimate or desirable in the abstract. But this predisposition is not destiny. What actually activates punitive enforcement is the perception of threat in the immediate social environment. 

Downstream Action

We next asked whether the two collectivism related composites would show downstream consequences for coercive action acceptance. Our claim was straightforward: if binding–defensive mechanisms energize enforcement, they should also lift support for coercive action. However, if coordinative–justice mechanisms emphasize restraint and corrective processes, they should predict lower acceptance of coercion.

The results are consistent with this expectation. Our CT_core composite (capturing the joint influence of Threat Vigilance and Collective Identification) is a strong positive predictor of coercive action acceptance (b ≈ .18, p < .001). Punitive Tightness also independently predicts greater endorsement of coercive responses (b ≈ .21, p < .001). Together, these mechanisms account for a sizeable share of the variance (R² ≈ .19). On the other hand, the CL composite (formed from Corrective Tightness and Justice/Trust worldviews) was negatively related to coercive action acceptance (b ≈ –.14, p < .01), indicating that individuals who prioritize fairness, trust, and corrective norms were less likely to endorse coercive measures.

Figure 4: Predictors of Coercive Action Acceptance. CT_core shows the strongest positive association, Punitive contributes modestly, and CL shows a small negative effect.

Aligning with our multiple roads to ‘we’, defensive binding mechanisms seem to heighten appetite for coercion, consistent with a worldview where threat and group fusion justify hard enforcement. In contrast, justice-first coordination guides enforcement away from coercive action and toward corrective, rule-based responses. Essentially, the two composites not only arise from different motivations, they also diverge sharply in their downstream consequences.

Escalation mechanisms: Threat x Fusion

We also examined whether threat cues interact with group fusion to escalate punitive enforcement. If defensive binding really operates as a two-part mechanism, then threat should predict punitive enforcement most strongly when people feel highly fused with their group. In other words, a strong Threat and Fusion interaction would indicate that threat only converts into punishment when individuals also experience deep identification and loyalty to the in-group.

Our results provide partial support for this claim. Threat Vigilance itself emerges as a robust predictor of punitive enforcement (b ≈ .26, p < .001), confirming the central role of threat in predicted sanctioning. However, the Threat × Fusion interaction did not reach significance, and the coefficient for Fusion on its own was modest and non-significant. Johnson–Neyman probing showed that the effect of Threat on punitive responses remained significant across the entire range of Fusion scores. So, the threat to punitive link holds regardless of how fused someone feels with their group.

Figure 5: Punitive vs Threat stratified by Fusion. Fusion raises baseline, but not Threat sensitivity.

This pattern fits our broader framework. The escalation pathway is real, in that threat consistently drives endorsement of punitive sanctions, but Fusion is not a necessary multiplier in this dataset. 

Restorative buffer: Threat × Corrective

Next, we turned to the question of whether Corrective Tightness might serve as a restorative buffer against threat-driven escalation. The logic behind this is that, if it is the case that corrective norms make it so that group enforcement translates into clear rules and procedural steps, then the presence of those norms should reduce the degree to which perceived threat translates into punitive sanctioning. In short – we expected a negative Threat × Corrective interaction, which would suggest that individuals high on corrective orientations would be less likely to respond to threat with harsher punitive measures.

The results offer only partial support for this. Both predictors entered the model as significant independent effects. Threat Vigilance remained a strong, positive predictor of punitive enforcement (b ≈ .26, p < .001), while Corrective Tightness was also positively related to punitive responses (b ≈ .15, p < .01). But the interaction term was non-significant, meaning that Corrective Tightness did not buffer or moderate the effect of threat. Instead, both factors simply add to one another in predicting higher levels of punitive endorsement.

Corrective and punitive orientations do not seem to be mutually exclusive. People who value clear corrective procedures can still turn to punitive enforcement when they perceive high threat. In this sense, the data suggest that the ingredients of the restorative pathway matter and exert influence, but they operate independently. The restorative buffer we anticipated (in which corrective orientations would dampen threat’s effect) did not hold.

Figure 6: Punitive vs Threat stratified by Corrective: near-parallel across levels of corrective. Corrective adds but does not buffer Threat Vigilance.

Binding Social vs. RWA‑Proxy (discriminant prediction)

RWA-Proxy reflects a broad dispositional tilt toward authority and norm enforcement, built from authoritarian submission, government norm endorsement, and rejection of permissiveness. In contrast, BIND-SOC, is measured through religious commitment and intergenerational values, capturing a more relational, community-anchored orientation.

We tested whether RWA-Proxy and BIND-SOC would differentially predict punitive enforcement and government norm endorsement. We hypothesized that both should incline individuals toward stronger sanctioning, but in different ways. The logic being that, while RWA-Proxy expresses a general authoritarian predisposition, BIND-SOC represents socially embedded commitments that might amplify sanctioning when threat and group fusion are salient.The results provided partial support for this. 

RWA-Proxy significantly predicts both punitive enforcement and government norm endorsement, independent of threat. For punitive, RWA-Proxy showed a modest but reliable effect (b ≈ .11, p < .05), while threat remained the dominant situational driver (b ≈ .25, p < .001). For government norm endorsement, RWA-Proxy was especially strong (b ≈ 2.74, p < .001), explaining substantial variance in support for government enforcement even after accounting for threat and fusion. In both models, RWA-Proxy predicted endorsement even at low threat, confirming it as a predispositional tilt toward authority and norm defense.In contrast, BIND-SOC contributed little to these outcomes. It showed a small positive direct effect on punitive enforcement (b ≈ .09, p < .05), but did not significantly predict government norm endorsement. Moreover, BIND-SOC did not strengthen the threat × fusion escalation pathway, contrary to our hypothesis.

Figure 7a: Punitive and Govnorm vs RWA-Proxy by Threat: main effects for both and a moderate interaction for GovNorm. Threat and RWA each increase Punitive and GovNorm, but the effect of RWA is strongest under higher perceived threat, particularly for GovNorm.

Figure 7b: GovNorm and Punitive vs Threat across Fusion and BIND-SOC: interaction plotted across −1, 0, +1 SD of BIND-SOC with lines representing Fusion z = −1, 0, +1. Threat increases Punitive and GovNorm, but effects strengthen with higher Fusion and BIND-SOC, especially for GovNorm. Fusion effects are more pronounced at low levels of BIND-SOC. 

These findings suggest that RWA-Proxy is a more powerful and consistent driver of sanctioning and government norm enforcement. Binding-Social orientations, while theoretically important, played a more limited role here, adding only modestly to punitive enforcement and failing to interact with threat or fusion as expected. 

Discussion

This study set out to test a moral-pluralism account of collectivism: the idea that there are multiple engines of “we-ness” rather than one generic collectivist trait. The data support that view on three fronts—scale construction, ideological distribution, and behavioral consequence.

Scale Construction
Treating collectivism as formative—assembled from non-interchangeable parts—proved useful. A binding-defensive engine (Threat Vigilance + Fusion) and a coordinative-justice engine (Corrective Tightness + Justice/Trust) behaved differently throughout the analyses, vindicating the choice to avoid a single reflective factor.

Ideological Distribution
Across ideology, relationships bend rather than run straight. There are significant horseshoe-shaped components underlying political ideology. Parochial Collectivism shows a U-shape (lowest among moderates, higher at both extremes). Liberal Collectivism shows a milder inverted-U, peaking near the center-left. Several ingredients (fusion, loyalty, prosociality) also curve. These horseshoe patterns explain how both ends of the spectrum can look “collective” while being powered by distinct moral logic.

Behavioral Consequence
The engines diverge downstream. Binding-defensive ingredients (Threat + Fusion) and Punitive Tightness are associated with greater acceptance of coercive action, whereas the coordinative-justice composite relates to lower coercion. At a mechanism level, Threat is the proximate throttle of punitive endorsement. Fusion raises the baseline willingness to punish but does not multiply threat (nearly parallel slopes). Corrective Tightness does not buffer threat either; it adds modestly to punishment when alarm is high, suggesting it blends a restorative desire for procedure with a zero-tolerance facet under pressure.

Disentangling Disposition & External Perception of Threat
A second dissociation clarifies dispositional vs situational control. RWA strongly predicts government-enforced norms even at low threat, marking it as a broad orientation toward authority and rule enforcement. By contrast, willingness to punish in specific situations is primarily situationally gated by perceived threat, with only modest contributions from RWA once threat is in the model. In short: who you are (RWA) tracks support for formal authority; what’s happening now (threat) drives punitive appetite.

Moral Pluralism Isn’t a Spectrum – It’s a Stack
The evidence from our study suggests that moral pluralism isn’t two tribal preferences locked in opposition across groups (as is commonly understood); it means two moral ingredients within every group that stack. Under alarm, even people who value clear procedures can favor harsher measures. The clearest candidate for a true brake is Justice/Trust—the expectation of norms of fairness, reciprocity, and restraint—which already shows a negative association with coercion at the composite level and should be tested directly as a moderator of threat. This makes sense – when people experience their wider social environment as more reciprocal and trustworthy, it is harder for collective threat circuits to become activated. 

Limitations
Cross-sectional, self-report data limit causal claims and are vulnerable to social-desirability and common-method bias. Corrective Tightness likely mixes at least two facets (procedural clarity vs severity); separating them in future efforts should sharpen interpretation. Robustness to demographic covariates and measurement invariance across ideology buckets should be verified in future studies, and behavioral outcomes (policy choices, sanction decisions in vignettes, real-effort/donation tasks) would strengthen external validity.

Conclusion

Collectivism is not one thing. People arrive at strong “we-ness” through at least two moral roads: one that binds under alarm and tightens boundaries, and another that coordinates under justice and legitimizes cooperation. Treating “collectivism” as a single reflective factor would mask these functionally distinct engines.These roads distribute differently across ideology, predict opposite enforcement tendencies, and are governed by different levers—threat for situational punishment, RWA for stable endorsement of government enforcement. The practical lesson is simple: if the goal is to reduce coercive spirals, lower perceived threat and raise credible expectations of fairness. More rules may improve coordination, but under alarm they are not a brake. We will continue to explore these questions and welcome questions and comments.

Future tests / prereg?

• Mediation: RWA → Threat → Punitive/Coercion (with Fusion as covariate).
• Moderation: Threat × Justice/Trust (expect negative).
• Corrective Split: separate Corrective-Clarity vs Corrective-Severity subscales and make sure Punitive is separate also.
• Shape robustness? restricted cubic splines; ideology-bucket contrasts.
• Controls & invariance: age, education, religiosity; multi-group tests by ideology.
• Experimental: threat inductions and justice/trust primes.