Cyberostracism: Effects of Being Ignored Over the Internet

The paper investigates whether ostracism—being ignored and excluded—has measurable psychological and behavioral effects even when it occurs in minimal, anonymous, computer-mediated contexts (cyberspace). Motivated by findings that Internet use can relate to loneliness and depression and that being ignored online is common, the authors test whether “cyberostracism” threatens fundamental human needs (belonging, self-esteem, control, meaningful existence) and leads to compensatory behaviors (e.g., conformity). They propose that even brief, remote ostracism will threaten inclusionary needs and cause aversive affect, and that targets will attempt to refortify threatened needs (particularly belonging) by conforming to group norms.

Ostracism is pervasive across species and cultures and is used to regulate group behavior. Prior social psychological research shows that social ostracism produces depressed mood, loneliness, anxiety, and helplessness, with stronger effects when attributed to personal shortcomings or when blatant. Even imagining or role-playing ostracism can harm self-evaluations. Williams (1997) proposed a need-threat model: ostracism uniquely and rapidly threatens belonging, self-esteem, control, and meaningful existence, producing psychological discomfort and motivating coping to repair threatened needs (e.g., enhancing belonging or conforming). The ambiguity inherent in being ignored can intensify aversive reactions. The authors extend this framework to cyber contexts, positing that online interaction’s ambiguity and paucity of cues may modulate impact: it could either exacerbate or buffer perceived ostracism. They hypothesize that cyberostracism will threaten needs (especially belonging and self-esteem) and increase conformity as a coping response.

Two online experiments tested cyberostracism. Experiment 1: - Participants: 1,486 individuals (after excluding 234 early exits) from 62 countries (64% female; modal ages 13–55 distributed across 13–18: 26%, 19–25: 37%, 26–55: 36%). About half were undergraduates; others recruited via listservs/newsgroups/web links. - Design: Between-subjects single factor with four levels of quantity of ostracism: overinclusion (participant received 67% of throws), inclusion (33%), partial ostracism (20%), complete ostracism (0%). Trait self-esteem (Rosenberg, 1965) measured as a predictor. - Procedure: Via a study website, participants consented, provided demographics, and completed Rosenberg Self-Esteem Scale. They were told the study examined mental visualization during a triadic virtual tossing game with two other players (actually computer-controlled). Each chose a color avatar. The game displayed messages/animations for throws and catches (good/bad with probability .9 each for throw and catch). Participants could choose recipients when in possession; otherwise they observed. After one initial throw and catch for all, the algorithm implemented assigned ostracism probabilities. A quit option appeared after the sixth throw; game continued until participants quit, then they completed measures and an open-ended thought listing. - Measures: • Manipulation check: self-reported percentage of throws received. • Persistence: number of turns before quitting. • Need threats (9-point items; higher scores = greater threat after reverse scoring where appropriate): belonging (“How much do you feel you belonged to the group?” reverse scored), self-esteem (socially oriented; “To what extent do you think the other participants value you as a person?” reverse scored), control (“I am in control of my life,” reverse scored), meaningful existence (“Life is meaningless”). • Aversive impact: mood (bad–good, sad–happy, tense–relaxed, rejected–accepted), perceived intensity of ostracism (ignored/excluded vs noticed/included), perceived group cohesiveness (liking of others and perceived others’ liking; reverse scored). Composite aversive impact index created due to intercorrelations (rs ≈ .49–.53). - Analyses: Multiple regressions (dummy-coded ostracism levels), trend analyses (linear, quadratic, cubic using actual ostracism frequencies 0%, 20%, 33%, 67%), and mediation tests (Baron & Kenny) with need threats as mediators of ostracism effects on aversive impact. Experiment 2: - Participants: 501 accessed; 270 did not complete (likely due to technical lags/compatibility). After excluding 18 suspicious, final N = 213 (33% male). 87% PC users, 13% Mac; countries: 14 total (US 59%, Australia 32%). - Design: 3 (group membership of the two cyberball partners: in-group, out-group, mixed) × 2 (interaction type: inclusion vs complete ostracism) between-subjects factorial. Two controls: (1) no-peer-influence perceptual task (N = 24; accuracy baseline), (2) conformity-pressures-only (N = 20; perception task without prior cyberball) to determine whether ostracism increases conformity vs inclusion decreases it. - Procedure: Website with consent, prelim questions (computer type; brief descriptions of typical PC vs Mac users for manipulation check). Cyberball game: animated toss of a Psi-labeled ball among three players for 10 throws total (first 3 standardized so each threw/caught once; participant had ball on throws 1 and 4). From throw 5 onward, inclusion condition: participant continued to receive ~33% of throws; ostracism condition: participant received no further throws. Thought bubbles allowed free-text entries each turn. Following cyberball, participants joined a new six-person perception task (participant always respondent #6 via a mock random wheel). On each of six trials, a simple geometric figure appeared, then six complex figures; participants chose which contained the simple figure. Trials 3, 4, and 6 were critical: the first five ostensible respondents unanimously chose an incorrect option; other trials showed unanimous correct responses. Post-task questionnaire collected manipulation checks and a single-item belonging rating (1–9). - IVs: Group membership cue (others described as PC users, Mac users, or one of each; in-/out-/mixed defined relative to participant’s self-identified platform). Ostracism vs inclusion as above. - DVs: • Conformity measured as (a) mean percentage of critical trials conformed (0–100% across 0, 33, 67, 100%) and (b) percentage of participants who conformed at least once. • Manipulation checks: perceived percentage of throws; recall of partners’ platforms; valence-coded in-/out-group descriptions. • Belonging: single 9-point item regarding sense of belonging with cyberball partners. - Analyses: 2×3 ANOVAs for manipulation checks, conformity, and belonging; Bonferroni post hoc tests; descriptive contrasts with control groups.

Experiment 1: - Manipulation check: Participants accurately perceived inclusion/ostracism gradients. Reported mean percent throws decreased across conditions: Overinclusion M ≈ 50% (SD 19), Inclusion M ≈ 34% (13), Partial M ≈ 26% (14), Complete M ≈ 12% (9). Strong main effect of ostracism level (F values up to 449.1, p < .01). Trend analysis: significant linear effect (β ≈ .76, t ≈ 14.45, p < .01) with a small quadratic component; linear model explained most variance (R² ≈ .472). - Persistence: Ostracism affected number of turns before quitting (ΔR² ≈ .012, p < .01). Complete ostracism reduced persistence (M 11.0 turns) vs inclusion (M 13.7, p < .05) and partial ostracism (M 15.0, p < .01); no difference vs overinclusion (M 12.7). Trend was predominantly quadratic: persistence increased from overinclusion → inclusion → partial, then dropped at complete ostracism. - Aversive impact: Robust main effect of ostracism (ΔR² ≈ .293, F ≈ 219.3, p < .01). Planned contrasts showed stepwise increases in aversive impact from overinclusion to inclusion to partial to complete (all ps < .01). Trait self-esteem had a main effect (ΔR² ≈ .046): lower self-esteem predicted greater aversive impact; no interaction with ostracism level. Trend: significant linear and quadratic effects (ΔR² ≈ .29), indicating increasing aversive impact with more ostracism. - Need threats and mediation: Ostracism increased threats to belonging (R² ≈ .332, p < .01) and socially oriented self-esteem (R² ≈ .097, p < .01) but did not affect control or meaningful existence. Both belonging and self-esteem partially mediated ostracism’s effect on aversive impact; each uniquely predicted aversive impact when entered together (betas ≈ 3.3 and 1.24, ps < .001). Perceived threats to control and meaningful existence independently correlated with aversive impact but were not influenced by ostracism manipulation. - Overinclusion vs inclusion: Overinclusion did not produce aversive responses, supporting the distinctiveness of being ignored versus receiving attention. Experiment 2: - Manipulation checks: Inclusion led to higher perceived throws than ostracism (M 39% vs 21%, F(1,163) = 63.2, p < .05). Group membership main effect and interaction emerged because participants ostracized by in-group overestimated how often they received the ball; recall of partners’ platforms was mostly accurate (82%). In-/out-group descriptions showed expected ingroup favoritism. - Conformity: Ostracized participants conformed more to unanimous incorrect judgments (M 27%, SD 30%) than included participants (M 18%, SD 28%), F(1,163) = 4.01, p < .05. No main effect of group membership and no interaction. The no-peer-influence control showed no errors; the conformity-pressures-only control averaged 16.65% conformity, comparable to inclusion (18%), indicating that ostracism increased conformity rather than inclusion reducing it. - Belonging: Ostracized participants reported lower belonging (M 3.11) than included (M 4.65), F(1,163) = 14.4, p < .01, with an interaction: ostracism by out-group and mixed-group reduced belonging relative to inclusion, but ostracism by in-group did not reduce reported belonging versus in-group inclusion (paradox). Belonging did not statistically mediate conformity (belonging measured post-conformity; regression ns).

The studies demonstrate that ostracism experienced in a minimal, anonymous online context reliably produces immediate psychological costs and behavioral consequences. In Experiment 1, cyberostracism increased aversive affect and threatened belonging and socially based self-esteem in a graded, linear fashion as exclusion intensified, consistent with Williams’s need-threat model. Control and meaningful existence did not change under these brief cyber conditions, possibly due to context irrelevance, measurement specificity, or the interface allowing minimal perceived control. Behaviorally, partial ostracism maintained engagement, whereas complete ostracism reduced persistence, paralleling partial reinforcement effects. Experiment 2 extends findings beyond self-report: brief cyberostracism increased conformity to a new group’s unanimous errors, consistent with the prediction that targets attempt to refortify belonging after exclusion. Conformity rose irrespective of whether ostracizers were in-group or out-group members, suggesting that ostracism by anyone is sufficiently aversive to trigger affiliative, norm-following behavior toward others. Self-report anomalies in the in-group ostracism condition (overestimated inclusion; unchanged belonging) coupled with elevated conformity suggest participants may mask or reinterpret in-group exclusion or that categorization alone bolstered perceived inclusion. Overall, results underscore ostracism’s potency as a social influence process, even in highly minimal cyber settings, and support the need-threat framework linking short-term need deprivation to immediate coping responses.

This work introduces and validates an online paradigm for studying cyberostracism and shows that even minimal exclusion by anonymous others over the Internet threatens belonging and sociometer self-esteem, worsens mood, and increases conformity to new groups. The findings provide empirical support for the need-threat model of ostracism, demonstrating both affective consequences and compensatory behavioral strategies (conformity) following short-term exclusion. The paradigm also illustrates the feasibility of rigorous social psychological research online with diverse samples. Future research should (a) refine measures of control and meaningful existence in cyber contexts and examine time courses of need threats, (b) test stronger group identification manipulations and examine conformity toward ostracizers versus new groups, (c) explore long-term cyberostracism effects (e.g., depression, alienation) and boundary conditions (ambiguity, anonymity), and (d) address methodological challenges of online experimentation (dropout, technical variability) while leveraging its reach.

- Control and meaningful existence did not vary with ostracism in Experiment 1; measures may have been too trait-like or the cyber context/brief duration made these needs less salient. The interface may have afforded minimal perceived control (clicking to proceed), attenuating control threat. - High dropout in Experiment 2 (technical lags/compatibility), raising self-selection concerns; effects may be conservative if those most distressed left. - Limited experimenter control over participants’ environments and identities (possible misreporting of demographics; repeated participation risk via false emails), common to online research. - Belonging measured after conformity in Experiment 2, precluding classical mediation testing; the in-group ostracism paradox suggests potential self-presentation or categorization effects not fully captured by measures. - The cyber paradigms, while ecologically relevant to some online interactions, still represent brief, artificial tasks; generalizability to sustained or consequential online exclusion may differ.