Contribution
Earlier field studies (Totterdell; Bartel & Saavedra 2000) had shown that the moods of team members converge, but only concurrently, so causation was open. This paper manipulates the source of emotion with a trained confederate and shows that mood spreads to group members. It also shows, for the first time by the author’s account, what contagion does to group processes: cooperation, conflict and perceived performance. A second study without a confederate replicates the convergence and the process correlates in naturally formed groups.
Key insight
People are “walking mood inductors”. A group member’s mood changes the others’ mood, and through mood their judgements and conduct, while they take their resulting feelings and judgements to be their own. The effect on the group is not merely shared feeling. It is more cooperation and less conflict, which members attribute to the facts of the task.
Assumptions
- Emotion as a broad label. “Emotion” is used interchangeably with affect, following Hatfield et al. and Ashforth & Humphrey (1995). The object of study is mood, chosen as the most malleable and broad-ranging affect; dispositional affect is treated as not contagious.
- The circumplex model (Russell 1980) structures the design: valence (pleasantness) crossed with energy (activation).
- Mechanism. Contagion runs both through automatic mimicry and afferent feedback (Hatfield, Cacioppo & Rapson 1994) and through conscious social comparison. The study does not separate them.
- Change scores as contagion. Self-reported contagion is Time 2 pleasant mood (recalled for the second half of the session) minus Time 1 mood. Time 2 is retrospective, reported after the exercise.
- Groups. Ad hoc groups of students in a 30-minute mixed-motive task: each member argues for a bonus for “their” employee, while serving the company as a whole. If no agreement is reached, nobody gets a bonus.
Key results
- Manipulation check (Study 1, Table 1). Pleasant confederate rated 6.59 against 3.89 (χ²(1) = 42.67); high energy 7.68 against 3.27 (χ²(1) = 152.52). There was an unexpected valence × energy interaction on perceived energy (χ²(1) = 17.31), controlled for in all tests.
- H1, contagion (supported). Individual level: mood change +.41 with the pleasant confederate, −.26 with the unpleasant one (χ²(1) = 9.97, p < .005). Video-coders, who saw only the members, rated Time 2 mood 2.75 against 2.33 (χ²(1) = 10.30). Group level (Table 3b): aggregated self- and coder ratings, coders’ overall group pleasantness (F = 20.68) and members’ ratings of group pleasantness (F = 10.10) all follow confederate valence.
- H2, unpleasant spreads more (not supported). The absolute changes did not differ (χ² < 1).
- H3, energy amplifies (suggestive only). Unpleasant: high energy −.44, low energy −.07 (χ²(1) = 2.81, p < .10). Pleasant: high energy +.16, low energy +.66 (p < .10), the opposite of the prediction. There was also an unexpected main effect of energy on mood change, with low energy more positive.
- H4, cooperativeness (supported). Individual level (Table 4): mood change predicts self-rated cooperativeness (b = .89, p < .001) and peer-rated cooperativeness (b = .47, p < .05). Group level (Table 5, N = 26): coder-rated contagion correlates with coder-rated cooperativeness (r = .44, p < .05) and with a lower SD of funds distributed (r = −.37, p < .05). Self-reported group contagion gives .34 (p < .10) and −.19 (n.s.).
- H5, conflict (supported). Group contagion correlates with coder-rated conflict: self-report −.48, coder −.42 (both p < .05).
- H6, performance (partly supported). Self-rated performance b = .44 (p < .001); peer-rated b = 3.96 (p < .10).
- Study 2 (113 MBA students, 26 groups). Mood ICC .00 at Time 1, .24 at Time 2. Contagion predicts self-rated cooperation (b = .48, p < .01) and self-rated performance (b = .19, p < .05), but not peer-rated performance. Group contagion correlates with members’ ratings of group cooperation (r = .49, p < .01) and conflict (r = −.49, p < .01), but not with the SD of funds. Peer cooperativeness was dropped because its ICC was .14.
- Exploratory findings. High self-monitors showed more contagion. Friendship within MBA groups correlated with contagion (r = .44, p < .05). In Study 2, contagion was related to winning a larger share of funds (printed “β=23”, p < .05, presumably .23). No demographic variable mattered.
- Unawareness. In both studies members attributed their effectiveness to money won and to their own abilities, not to their mood.
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | A group member’s displayed mood causally changes other members’ mood | strong for pleasant vs unpleasant valence in this setting: randomised confederate, two independent measures | Study 1, H1; Table 3 |
| C2 | Pleasant mood contagion increases cooperation and reduces conflict in the group | moderate: Study 1 correlates contagion (not the manipulation) with processes, 26–29 groups; replicated correlationally in Study 2 | Tables 4–5, 7–8 |
| C3 | Contagion is a group-level phenomenon, not only dyadic transmission from a strong source | moderate: mood convergence without a confederate (ICC .00 → .24); not a test of mechanism | Study 2 |
| C4 | People relate their pleasantness to others’ conduct but show no awareness that their mood shapes their judged effectiveness, which they attribute to task factors | weak to moderate: self-report attribution items, not a designed test | General Discussion |
| C5 | Unpleasant emotion spreads more than pleasant; energy amplifies contagion | not supported (C5a) / mixed and marginal (C5b) | H2, H3 |
Concepts
- emotional contagion: after Schoenewolf (1990), “a process in which a person or group influences the emotions, or behavior of another person or group through the conscious or unconscious induction of emotion states and behavioral attitudes”.
- primitive contagion: automatic mimicry of facial, vocal and postural expression, followed by afferent feedback that produces the feeling (Hatfield, Cacioppo & Rapson).
- emotional energy: the intensity with which an emotion is expressed: pitch, loudness, tempo, gesture.
- group contagion (Study 2): convergence of members’ moods, measured as the rise in intraclass correlation from before to after the interaction.
Connections
- von Scheve & Ismer 2013 (LIT-696). Their “face-to-face” perspective and their definition of collective emotion as convergence frame this study. Study 2’s ICC is that definition operationalised. They do not cite this paper.
- Ashforth & Humphrey 1995 (LIT-703), Deferred. Cited (p. 6) for using “emotion” as a broad label for subjective feeling.
- Frijda 1988 (LIT-704), Deferred. Cited (p. 12) for mood contagion as “a direct source of information … about how the group is doing”.
- Sanchez-Burks & Huy 2009 (LIT-699), Deferred. The perceiver’s complement, going by their abstract. The General Discussion’s call for leaders who are “emotionally attuned to and influenced by their followers” is the competence they name emotional aperture.
- Lazarus. Cited (1991: 47) for the definition of mood as “a transient reaction to specific encounters”. The record holds that book unread (LIT-550).
Bearing on the record
- It is the evidence a THEORY on group emotion would need for its mechanism: that convergence of affect in a group is produced in part by contagion and has consequences for the group’s conduct. A candidate is given in the report, not filed.
- No ML instruction. Nothing here belongs in the anthology.
Limitations
- Pre-acceptance manuscript. Numbers may differ from the published article.
- Small group samples. 29 and 26 groups. Table 3b gives 7, 7, 8 and 7 groups per cell, while Table 5 reports N = 26. The manuscript does not explain the difference.
- The process effects are correlational. H4–H6 regress outcomes on measured contagion, not on the manipulated condition, so a third factor (for example, a member’s own reaction to the task) could drive both mood and cooperation. Study 2 has no manipulation at all.
- Retrospective Time 2 mood, reported after the exercise and recalled for its second half.
- One male drama-student confederate played all four conditions, in a competitive student task; the author herself suspects context reversed the energy effect for pleasant mood.
- Mechanism untested. Mimicry-feedback and social comparison are both invoked; nothing separates them.
Open questions
- What happens when members convey different or contradictory emotions? The author calls this “fascinating” and untested.
- Do ongoing teams habituate to one another’s moods or become more sensitive? The friendship correlation in Study 2 hints at sensitivity.
- Does contagion between people who identify as a group differ from contagion between strangers, as von Scheve & Ismer’s I-mode/We-mode distinction would predict?
Corrections
- none to a seeded skim (there was no seed)
- Typo in the manuscript. It cites “Barsade & Gibson, 1988” once (p. 3) for the 1998 chapter.