The ripple effect: emotional contagion and its influence on group behavior

Administrative Science Quarterly,  Dec, 2002  by Sigal G. Barsade

• E-mail
• Print
• Link

Emotional Contagion Measures

Emotional contagion was measured by both participants' self-reports and observers' ratings of mood via video-tape ratings of the participants interacting in the group exercise. The use of these dual measures of the emotional contagion construct is necessary for several reasons. First, having access to both types of measures of mood is important, as each has been shown to influence the contagion process (Hsee, Hatfield, and Chemtob, 1992) and yet not always give matching information. For example, Bartel and Saavedra (2000) found that observers' ratings of group mood matched self-report ratings for high-energy affect arousal better than low-energy affect (two of the experimental conditions). Methodologically, the video-coder data allow the benefit of better access to the mood being expressed by participants in real time, while self-report of mood (traditionally used in mood research, Larsen and Diener, 1992) allows a different type of access to participants' internal feeling states. Last, while facial expression is certainly a powerful gauge of emotions (Ekman and Friesen, 1975), significant differences have been found in the emotions employees express versus the emotions they display in organizational settings (e.g., Rafaeli and Sutton, 1991; Pugh, 2001).

Video-coder measures. Four video-coders were extensively trained in coding emotion through facial expression, body language, and verbal tone but were intentionally kept unaware of the experimental conditions or the purpose of the study. Much support has been found for video-coders' abilities to reliably judge facial expression and non-verbal behavior (e.g., Ekman and Friesen, 1975; Gump and Kulik, 1997), overall group mood (e.g., Bartel and Saavedra, 2000), and group dynamics (e.g., Jehn and Shah, 1996).

This set of coders viewed only the participants, not the confederate, so as to lessen the chance of coding bias due to the confederate's behavior. The coders were trained using the same work-group emotion scale created by Bartel and Saavedra (2000), which provides coders with an extensive list of behaviors indicative of work-group mood and has been shown to be valid and reliable. The coders measured emotional contagion by watching participants' facial expressions, body language, and verbal tone throughout the course of the experiment and rating the level of a participant's pleasant mood every two minutes (at the sound of a beep) on a scale of 1 (very slightly or not at all) to 5 (very much). The two-minute segments were aggregated across coders for the second part of the experiment to create a Time 2 mood scale based on video-coders' ratings. This scale had a mean of 2.56 (s.d. = .50), with a within-rater Cronbach alpha of .82 (each two-minute segment used as an item in the Time 2 participant contagion video-coder scale). The ICC interrater reliability among the video-coders for participants' Time 2 contagion was .77. Given that this was a laboratory experiment with randomly assigned participants, who started out at the same mood level across groups (no significant difference in participants' self-reported pre-experiment Time 1 mood across experimental conditions; F = .87, n.s.), it is possible to infer that the experimental conditions caused the differences in participants' mood at Time 2. On a group level, video-coders also rated their perceptions of each group's overall level of pleasant mood on a 1-7 scale (mean = 3.75, s.d. = 1.22; ICC = .72).