Participants were 131 undergraduate students from Jaume I University (Spain) recruited via campus posters and social media advertisements (mean [M] = 21.22 years, standard deviation [SD] = 2.97; range = 18–33; 81.7% female).

Participants were tested individually in laboratory rooms equipped with desktop computers. They were randomly assigned to an inclusion (exclusion = 1; 48.1%) or exclusion (inclusion = 0; 51.9%) condition using the Cyberball paradigm (Cyberball 5.5.0.2; Empirisoft Corporation, New York, NY, USA; programmed by Williams and Jarvis, 2006). Both participants and experimenters were blind to the condition. Participants were informed that they would play an online ball-tossing game with two same-age students participating from other rooms; in reality, the other players were computer-controlled.

A standardized instruction script informed participants that the task measured visualization ability and social coordination. After the Cyberball game, participants completed manipulation checks and post-task measures, followed by a full debriefing.

A post hoc sensitivity analysis using G*Power 3.1.9.7 (Heinrich Heine University Düsseldorf, Düsseldorf, Germany) (Faul et al., 2007) indicated that the final sample (N = 131) provided 83.1% power to detect a small-to-moderate interaction effect (f² = 0.062) at $\alpha$ = 0.05 for a two-predictor moderation model. Because this analysis was conducted post hoc, its results should be interpreted cautiously.

All questionnaires were completed on the same computer before and after the experimental manipulation. Higher scores reflect greater levels of the construct assessed, unless stated otherwise.

The Cyberball game consisted of 60 throws over approximately 3 minutes (Hartgerink et al., 2015).

⚫ Inclusion condition (0): approximately 20 passes to the participant, with an equal distribution.

⚫ Exclusion condition (1): approximately 10 passes to the participant early in the game, followed by continued omission.

To enhance ecological validity and the perceived stakes of inclusion, participants were informed that they would receive €5 for participation, plus an additional €2 bonus if they received $\ge$15 passes—a threshold achievable only in the inclusion condition. All participants received the full €7 after debriefing.

This hybrid manipulation, combining ostracism with distributive unfairness, was designed to heighten participants’ appraisals of personal injustice—a strength for testing the current moderation hypothesis but a limitation for isolating “pure” exclusion effects (see Discussion for implications). An experimental flowchart of the study is shown in Fig. 1.

Fig. 1.

Experimental flowchart of the cyberball ostracism paradigm study.

Analyses were conducted in SPSS 29.0 (IBM Corp., Armonk, NY, USA) using the PROCESS macro version 4.2 (Hayes, 2022). We tested a moderation model (PROCESS Model 1) predicting post-task anger with the following predictors:

⚫ Experimental condition (1 = exclusion; 0 = inclusion).

⚫ M-centered trait perceived injustice.

⚫ The interaction between condition and trait perceived injustice.

⚫ Covariates: baseline anger and baseline anxiety.

Significant interactions were probed at -1 SD (low), the M, and +1 SD (high) levels of trait perceived injustice. All effects were reported with unstandardized coefficients, standardized betas, 95% confidence intervals (CIs), and exact p-values. Assumptions were checked prior to analysis.

Ms, SDs, and zero-order correlations are presented in Table 1. All variables demonstrated acceptable skewness (–0.72 to 0.84) and kurtosis (–0.65 to 1.02). As shown in Table 1, post-Cyberball anger levels were positively associated with trait perceived injustice, baseline anger, and baseline anxiety, supporting expected convergent patterns.

A moderation analysis tested whether trait perceived injustice (T-IEQ) moderated the effect of ostracism (1 = exclusion, 0 = inclusion) on post-task anger, controlling for pre-task anger and anxiety (PROCESS Model 1; Hayes, 2022). The model accounted for 26.3% of the variance in post-task anger, R² = 0.26, F(5,125) = 8.92, p 0.001.

The interaction between ostracism and trait perceived injustice was significant, $\beta$ = 0.23, t(125) = 2.41, p = 0.017, 95% CI [0.04, 0.43]. The main effects of ostracism, $\beta$ = –3.60, t(125) = –1.32, p = 0.189, and trait perceived injustice, $\beta$ = –0.05, t(125) = –0.63, p = 0.528, were not significant when covariates were included. Pre-task anger and pre-task anxiety contributed significantly to post-task anger, $\beta$ = 0.49, t(125) = 3.27, p = 0.001, and $\beta$ = 0.14, t(125) = 1.93, p = 0.056, respectively.

Although the main effect of ostracism appears negative, this reflects M-centering of predictors in the moderation model. Simple slope decomposition clarified the conditional effects:

⚫ Low trait perceived injustice (16th percentile, T-IEQ = 20): Exclusion (vs. inclusion) did not significantly affect anger, $\beta$ = 1.08, t = 1.07, p = 0.288.

⚫ Average trait perceived injustice (50th percentile, T-IEQ = 25): Exclusion significantly increased anger, $\beta$ = 2.26, t = 2.89, p = 0.005.

⚫ High trait perceived injustice (84th percentile, T-IEQ = 33): Exclusion showed the strongest effect on anger, $\beta$ = 4.13, t = 4.30, p 0.001.

These results indicate that individuals with higher trait perceived injustice exhibit significantly greater anger responses when excluded, supporting the hypothesis that sensitivity to unfairness amplifies emotional reactivity (see Fig. 2).

Fig. 2.

Moderating effect of trait perceived injustice on the relationship between ostracism and anger.

A secondary moderation analysis tested whether trait perceived injustice (T-IEQ) moderated the effect of ostracism (1 = exclusion, 0 = inclusion) on post-task anxiety, controlling for pre-task anxiety and pre-task anger. The model explained a significant proportion of variance in post-task anxiety, R² = 0.61, F(5,125) = 39.62, p 0.001.

Neither the main effect of ostracism, $\beta$ = –0.96, t(125) = –0.46, p = 0.644, nor the interaction between ostracism and trait perceived injustice, $\beta$ = 0.02, t(125) = 0.21, p = 0.833, reached statistical significance. Trait perceived injustice also did not independently predict post-task anxiety, $\beta$ = 0.004, t(125) = 0.07, p = 0.946. In contrast, higher pre-task anxiety strongly predicted higher post-task anxiety, $\beta$ = 0.77, t(125) = 13.75, p 0.001, whereas pre-task anger showed a small but significant negative association, $\beta$ = –0.23, t(125) = –1.99, p = 0.048.

Overall, trait perceived injustice did not moderate anxiety responses to ostracism, indicating that the observed effects of T-IEQ in this study are specific to anger rather than generalized to negative affect.

The Cyberball task successfully induced social exclusion. Participants in the ostracism condition reported receiving significantly fewer ball tosses (15.34% vs. 35.54%) and feeling more excluded and rejected than those in the inclusion condition, all p 0.001. These results support the validity of the experimental manipulation.

First, these findings extend Williams’ (2007) temporal need–threat model by demonstrating that dispositional fairness concerns shape the reflective stage of the ostracism response. Although reflexive distress is widespread, reflective appraisal processes allow personal predispositions—such as heightened expectations of unfair treatment—to amplify negative affect (Poppelaars et al., 2019). This helps explain why not all young people respond uniformly to exclusion (Lynn Mulvey et al., 2017).

Second, our results provide further evidence that fairness-related traits may serve as more proximal predictors of anger than broader affective dispositions, such as neuroticism (McDonald and Donnellan, 2012). Although conceptually related to justice sensitivity, trait perceived injustice captures a generalized tendency to appraise personally relevant situations as unfair (Yakobov et al., 2019), reflecting a stable cognitive-emotional bias rather than a situational state. This aligns with developmental findings showing that empathy deficits and endorsement of aggressive responses to conflict are associated with involvement in school violence (Martos Martínez et al., 2021).

Third, grounded in a person–environment fit perspective (van Vianen, 2018), these results illustrate how susceptibility factors and peer adversity interact. Ostracism is a common social stressor in adolescence and emerging adulthood. When experienced by individuals who strongly interpret experiences as unjust, it may trigger heightened anger, a proximal precursor to conflict escalation.

Although aggression was not measured directly, anger is a well-established proximal risk factor for reactive aggression (Smith et al., 2016). Thus, trait-perceived injustice may help identify individuals who are emotionally vulnerable when exposed to exclusionary behavior, bullying, or relational conflict. Brief, evidence-based strategies could reduce this risk, including:

(1) Cognitive interventions targeting negative appraisals of unfairness and enhancing reappraisal skills (Bissell et al., 2018).

(2) School-based programs promoting empathy, perspective-taking, and nonviolent conflict resolution (Park-Higgerson et al., 2008).

Such approaches may be particularly valuable during developmental periods when concerns about peer belonging are especially prominent.

Interpretation of the present results should consider several limitations. In the hybrid fairness manipulation, the monetary contingency introduced distributive unfairness in addition to social exclusion. Consequently, anger responses may reflect the combined experience of being ostracized and receiving fewer rewards. Future work should isolate interpersonal exclusion from economic unfairness to clarify their independent and interactive effects. Regarding sample characteristics, participants were predominantly female undergraduates, which limits generalizability to populations at higher risk for behavioral aggression (Ashton et al., 2014; Malhi et al., 2020). Although emerging adults share developmental vulnerabilities with adolescents, especially regarding sensitivity to peer rejection and its association with conflict behaviors (Watson and Nesdale, 2012), replication in gender-balanced, school-based samples is needed. Another limitation concerns the focus on a single emotional outcome: anger. Although anger is theoretically central to reactive aggression pathways (Smith et al., 2016), ostracism also affects anxiety, sadness, prosocial repair behaviors, and physiological responses. A broader assessment would clarify whether perceived injustice reflects a general vulnerability to emotion dysregulation or a mechanism specific to anger. Additionally, longitudinal and behavioral studies are required to determine whether individuals high in perceived injustice show greater escalation from exclusion-induced anger to observable aggression, particularly in real-world peer contexts. Future research could also extend this approach to at-risk clinical populations (e.g., youth with depression, eating disorders, or borderline traits), in which rejection sensitivity and fairness concerns may contribute to affective dysregulation.