JASNH, 2005, Vol. 3, No. 2, 51-72. Copyright 2005 by Reysen Group. 1539-8714
  Evaluative Affect Display toward Male & Female Leaders II: Transmission among Group Members and Leader Reactions  
  Sabine C. Koch
University of Heidelberg, Germany

Two experiments investigated evaluative affect display (EAD, i.e., nonverbal evaluative reactions) toward leaders in small face-to-face groups, its transmission among participants, and leaders’ reactions. Experiment 1 focused on the transmission process of negative affect, Experiment 2 on group leaders’ perceptions, attributions and reactions when faced with negative group affect. Both experiments used confederates to create negative affect. Results suggest robust observations of EAD. Contradicting our hypothesis, EAD was not more negative toward women. In accordance with the hypothesis, participants showed no preference for male over female leaders in competence ratings. Experiment 1 partially supported the existence of an emotional contagion effect. Women in both experiments showed higher awareness of affect display from the group, as was hypothesized. In addition, Experiment 2 suggested that women leaders used a more internal attribution style and that their competence self-ratings dropped significantly after negative group reactions. Conversely, there were no gender differences in leaders’ displayed nervousness. The null hypothesis was confirmed in 50% of the case, indicating inconclusiveness and need of further studies on this topic

pp. 52


Evaluative Affect Display toward Male and Female Leaders. Transmission among Group Members and Leader Reactions

Why is there still such a low percentage of women in higher leadership positions despite the high percentage of women in higher education? As a contribution to this applied research question we investigated gendered reactions toward male and female leaders1 in task-oriented face-to-face groups with a focus on nonverbal communication. Two studies were designed to demonstrate that evaluative affect display toward authorities is a robust phenomenon. The reported experiments extend earlier studies (Brown & Geis, 1984; Butler & Geis, 1990; Koch, 2005) in several ways. Butler and Geis (1990) had found that female group leaders received more negative affect displays than male leaders while they were judged equally or more competent on rating scales. Koch (2005), in a replication of the Butler and Geis findings, developed a more efficient method to assess evaluative affect displays in small groups. Now, -- using confederates for affect manipulation -- the potential role of contagion effects in the origin of nonverbal group consensus needed to be demonstrated. And, reactions of group leaders exposed to negative evaluative affect display needed to be assessed. For the assessment of gender-differences the investigation followed a “doing gender” approach (West & Zimmermann, 1987).

Interaction research has long concentrated on the actor side and almost forgotten the recipient side (“pseudounilaterality”, Duncan, Kanki, Mokros and Fiske; 1984). However, focusing on the active contribution of recipients and their role in the communicative process has experienced a recent renaissance in social psychology (e.g., Bavelas & Chovil, 2000; Bavelas, Coats, & Johnson, 2000; Snyder & Stukas, 1999). At the same time there has been an increasing interest in a closer investigation of the actor side, e.g., the styles of men and women in leadership positions (Eagly & Johannesen-Schmidt, 2001).

Leadership, competence and the decrease of positive self-efficacy beliefs during the first professional years of highly qualified women

An important aspect of leadership is the perception of competence. There is substantial agreement in the research literature that men are by default perceived as more competent than women (Biernat & Fuegen, 2001; Eagly & Karau, 2002; Geis, 1993; Heilman, 2001; Ridgeway, 2001). Berger and colleagues in their expectation states theory (EST) offer an explanation of this phenomenon (Berger, Connor, & Fisek, 1974; Berger, Fisek, Norman, & Zelditch, 1977). According to expectation states theory men are expected to have a higher competence in male tasks, and women are expected to have a higher competence in female tasks. Plus there should be a higher competence expectation for men in neutral tasks in the absence of other diagnostic cues, because under these circumstances gender functions as a diffuse status characteristic with higher performance expectations toward the higher status group. Social role theory (Eagly, 1987), and role congruity theory (RCT; Eagly & Karau, 2002) make the same predictions as EST: since we constantly observe more men in higher positions, we would expect them to be more competent in gender neutral

pp. 53

tasks as well. Eagly and Karau (2002) propose two forms of prejudice toward female leaders. Women are perceived less favorably than men as potential candidates for leadership roles (descriptive aspect) and they are evaluated less favorably for behavior that fulfills the prescriptions of a leader role (prescriptive aspect). As a consequence attitudes are less positive toward female than male leaders or potential leaders.

These negative evaluations can be communicated verbally and nonverbally. The nonverbal channel of communication will be preferred particularly in situations where the target of evaluation is present and has a higher status or position than oneself (cf. Mehrabian, 1971, Krämer, 2001). This paper, therefore, focuses on the nonverbal communication of attitudes, specifically of negative attitudes as they are the critical case in the communication of prejudice. Furthermore, on the basis of earlier research by Hall and others (Hall, 1978, 1984; Hall & Bernieri, 2001), it is to be expected that men and women in positions of authority will be differentially sensitive to the display of nonverbal cues and will notice group reactions to different degrees; they will further attribute different causes to those reactions, and react in different ways. If, however, leaders are not aware of negative reactions at all, will they be susceptible to behavior confirmation processes (Snyder, 1984) or will they be immune?

There is recent evidence that self-efficacy beliefs of women drop dramatically within their first professional years (Abele-Brehm, 2000a). Reactions toward male and female authorities on low levels of organizational hierarchies, that is, within their first years after graduation, may be crucial for further career aspirations of men and women. In a study with graduates from the University of Erlangen, Germany (n=1930), Abele-Brehm showed that although men and women set out in their jobs with equal self-efficacy beliefs, these beliefs obviously underwent a different development in their early career experiences: three years after they had started their jobs, professional women’s self-efficacy had decreased considerably whereas men’s had remained similar to the level at graduation time (Abele-Brehm, 2000a, b; Phillips & Imhoff, 1997).

Evaluative affect display toward men and women in authority positions

Evaluative affect display (EAD) is the expressive component of an attitude, such as a facial expression, or a gesture. It is a nonverbal reaction of approval or disapproval toward a target person and transmits an attitude toward that target. Research findings support the idea that negative attitudes toward authorities will be communicated mostly by nonverbal cues (Mehrabian, 1971). Once perceived by the recipient, they may have behavioral consequences and set off interaction dynamics, such as self-fulfilling prophecies and behavior confirmation (Blanck, 1993; Geis, 1993; Merton, 1948; Snyder, 1984). Affective reactions are bipolar (Cacioppo, Berntson, & Gardner, 1999) and often non-deliberate (Chen & Bargh, 1999; Brief & Weiss, 2002).

Prior findings about the phenomenon of EAD include two studies by Florence Geis (Brown & Geis, 1984, Butler & Geis, 1990) and a recent study by the author (Koch 2005). In 1990, Butler and Geis demonstrated that women in authority positions received more negative affect from both men and women (Butler & Geis, 1990), while competence ratings on a cognitive measure revealed no gender differences. They demonstrated this in an experimental setting, where male and female confederates – playing the role of the group leader – acted according to verbatim scripts. The two confederates talked with two

pp. 54

participants, and that conversation was observed by two raters through a one-way mirror. Observers used the Facial Action Coding System (FACS, Ekman & Friesen, 1978) to code affective facial reactions of participants toward the leaders online. Findings suggest that, while people can control their answers and suppress their negative attitudes on rating-scales, evaluative affect display is not necessarily under conscious control of participants. Nonverbal signals can be encoded and decoded while conscious attention is focused on the content of discussion. This makes a strong case for the use of behavior observations in stereotype and attitude research.

In a study by Koch (2005), participants conversed face-to-face in four-person groups. It was a free exchange format about a prior chat-room discussion in which they had just taken part in the same small group without knowing or seeing each other. They were given the debriefing of the chat-room experiment via audio-tape, and were told that this was being done “for standardization reasons”. The tape recording of either the male or female experimenter’s voice was the actual intervention of the study. The key question was: Would participants affect displays differ while listening to the tapes in the male leader vs. female leader condition? Participants’ reactions were coded from videotapes.

Results of the observations showed that women received significantly more negative affect, while at the same time they were rated as more competent. The study was able to successfully replicateof the Butler and Geis (1990) findings. The observational method was refined and at the same time substantially simplified in terms of coding economy. In addition, no male “gender-as-status” bias occurred as would have been predicted by expectation states theory (Berger, Connor, & Fisek, 1974) or role congruity theory (Eagly & Karau, 2002), in fact, the male leader was rated less competent than the female leader. Two control variables, sex of participant and sex composition of group, had no effect. Yet, group membership had a considerable effect on the affect reactions. Participants of some groups unanimously reacted more negatively than participants of other groups -- unrelated to other characteristic of the groups.

Attitude transmission in groups: The potential role of emotional contagion

These findings suggest that there are processes of emotional contagion (Hatfield, Cacioppo, & Rapson, 1994) or spreading attitudes (Walther, 2002) at work. Emotional contagion is defined by Hatfield et al. (1994) as “…the tendency to automatically mimic and synchronize facial expressions, vocalizations, postures, and movements with those of another person and consequently, to converge emotionally” (Hatfield, Cacioppo, & Rapson, 1994, p.5). Hatfield et al. (1994) describe the process of emotional contagion as “relatively automatic, unintentional, uncontrollable and largely inaccessible to conversant awareness”. They describe mechanisms of motor mimicry and synchrony on a behavioral level, and mechanisms of facial, vocal, and postural feedback on a perceptual level. They further define contexts of susceptibility to emotional contagion. The transmission of attitudes in groups emphasizes the crucial role of non-cognitive evaluative processes in attitude formation (Eagly & Chaiken, 1993) and supports Cacioppo’s notion of a bivalenced attitude system (Cacioppo, Berntson & Gardner, 1999). For recent overviews on emotional contagion effects and affect in groups see Kelly and Barsade (2001) and Kelly and Spoor (2005). For experimental evidence of contagion processes in groups see Bartel and Saavedra (2000) and Barsade (2002).

pp. 55

Reactions of Leaders

Men and women might react differently to the display of negative affect. A line of research that Robert Rosenthal and Judith Hall started in the 1970s suggests that men will notice the display of evaluative affect less than women. In a number of meta-analyses, Judith Hall demonstrated that women on average have a higher nonverbal sensitivity than men (Hall, 1978, 1984; see also Snodgrass, 1985). In a meta-analysis of gender-effects over 64 studies, with effect sizes of about r = .30 women showed better encoding and decoding abilities, particularly for facial cues (Hall, 1984) whereas men generally did slightly better in the encoding and decoding of vocal cues (Rosenthal & DePaulo, 1979). Rosenthal and DePaulo assumed a connection of these findings in the fact that facial cues mainly serve the communication of socio-emotional cues, whereas vocal cues mainly serve the communication of dominance-related agentic cues. Evaluative affect is communicated primarily through facial expressions. Not noticing evaluative affect display by part of the group may, thus, make men more immune to negative evaluative reactions than women. Therefore, they may not be as susceptible as women to interactional expectation processes (cf. Blanck, 1993; Heider, 1958; Merton, 1948; Snyder, 1984; but see Dvir, Eden & Banjo, 1995; Eden, 1993), and to causal explanations by internal attributions (e.g., Hall, 1984). In addition, social role expectations may contribute to women questioning their leadership behavior more readily and, thus, enhance processes of self-doubt, nervousness and a general downward spiraling of self-esteem and self-efficacy beliefs in young professional women.

Overview of the Experiments

Expressive affect reactions toward leaders in small groups, and their transmission among participants, were investigated in two experiments (total n=132) based on the findings of Butler and Geis (1990) and Koch (2005). Key research questions were: What communicative mechanism is responsible for the effect of attitude transmission? How much of the effect occurs without conscious verbalizable knowledge? What are leader’s reactions to negative group affect? Are there gender differences in leader’s reactions? Experiment 1 expanded on previous findings that reactions in some discussion groups were much more pronounced than in others. It was investigated whether an emotional contagion effect in face-to-face small groups potentially creates a group consensus reaction. Experiment 2 investigated group leaders’ awareness of, and attributions and reactions to, negative consensual affect display. The studies were based are several basic premises: (a) There is a communicative mediating mechanism of attitude spread in groups, (b) the transmission works rather non-deliberately, most likely via the nonverbal channel, (c) consensual affect cues are able to raise or lower the perceived quality of identical leadership performances in a small group setting (cf. Brown & Geis, 1984), (d) leadership evaluation like any attribution of competence tends to be subject to considerable sex bias (as predicted by expectation states theory and social role theory), and (e) nonverbal consensual affect cues, sex of confederates (reacting) and sex of target (being reacted to) will influence leadership evaluations.

These premises led to the following hypotheses:
Hypothesis 1. There will be more display of negative affect toward female authorities

pp. 56

(F>M; evaluative affect effect), while at the same time
Hypothesis 2. Men will not be rated as more competent (not M>F).
Hypothesis 3. There will be an observable emotional contagion effect between base-line and intervention ratings in the direction of the intervention valence (i.e., given negative affect intervention of confederates, participants affect will change to more negativity in display toward leader accordingly).
Hypothesis 4. Dependent on their gender, leaders will react differently to negative consensual affect display. Expected reactions are that after the affect intervention (a) women will rate their competence lower than before, and lower than men at t2 (W: t2<t1; t2w<t2m), (b) the conscious awareness of the affect intervention will be higher in women than in men (W>M), (c) women will tend to attribute the consensual affect display internally, whereas men will rather attribute it externally (W: int>ext; M: ext>int), and (d) women will display more nervousness than men (W>M at t2).

Experiment 1
The Transmission of Consensual Evaluative Affect

Participants and Design
Fifty-three participants (17 men and 36 women; mean age was 25.7, SD = 6.8), mostly students from the University of Heidelberg, participated in small discussion groups of four persons (2 subjects and 2 confederates as group members per group). Participants were either given partial credit toward a course requirement or received a book for their participation. In a single-factorial pre-post-test design participants were confronted with a male and a female leader. Scripts of leaders were balanced in sequence and role. Two female confederates in each group performed the evaluative affect intervention (negative only). 26 groups were run overall, due to a technical problem only 25 of them were videotaped. Thus overall, there were 4 confederates (male leader, female leader, 2 women confederates) and 2 subjects in each group. Dependent variables were evaluative affect display, emotional contagion and competence ratings.

One participant noticed that confederates were participating in the discussion and was consequently excluded from the data set. In the course of the debriefing 21 out of 52 participants declared that they had noticed negative affect or a negative attitude communicated by group members in some way (4 men and 17 women). We had tried to keep the intensity of the intervention exactly at the threshold of conscious awareness. Since almost half of the sample had noticed the affect intervention, the manipulation was evaluated as a success.


The participants were informed by a flyer that the experiment was about cooperative problem-solving in small groups. A male and a female experimenter served as group leaders. Participants’ consent to be videotaped was gathered by leader 1. Then they were given a first short questionnaire for completion by leader 2 (for a graphical overview of the procedure see Figure 1). Baseline affect of reaction toward male and female authority was taken from these two short interactions. Groups were then presented with the first topic

pp. 57

of discussion. The topic “workplace values” was chosen, because it proved to be rather gender-neutral in pre-tests. Participants had to rank order six “workplace values” regarding of their subjective importance to them: a. high income, b. little working time - much spare time, c. job security, d. work is fun/pleasure, e. social contacts at work, f. career opportunities. Then they were invited to participate in a five minutes discussion, exchanging arguments for their solution and coming to a group solution of the three most important “workplace values”. Next the topic of “communicative components of workplace communication” was given to them and the procedure started from the beginning. The first topic was given to them by leader 2, who had about three to four minutes of talking time. The second topic was given by leader 1, who also talked to the group for about three to four minutes. When both leaders left the room, one of the two confederates made a negative attitudinal statement about the respective leader and the second confederate vividly agreed to that statement. Confederates were all women, leaders were men and women. We made sure that there was verbal as well as non-verbal comparability of the performance. The main leaders regularly switched script-roles to ensure that the sequence was not responsible for any possible effects. Leader similarity control was verified by having the leaders rated by six research assistants on several important dimensions such as sympathy, dominance, activity, etc. For this purpose, research assistants used bipolar seven-point rating scales. The male and female leaders were rated sufficiently similar at the initial trial (SD on single items 1.0 or less).

We distinguished a baseline and an intervention phase. Only negative affect display interventions were used as they are the critical case in the communication of prejudice. Baseline affect was calculated from the participants’ initial reactions to the leaders, with the confederates not yet in the room. Using the same procedure as in the baseline, participants were then observed after the negative affect intervention by the confederates had set in. Confederates were instructed to display negative affect when the “responsible researcher” started to explain the group task. As we did not know whether a mere nonverbal interaction would be sufficient, we additionally asked the confederates to make a negative statement after the “responsible researcher” had left the room. “Oh, he is really chaotic…” (the second confederate agreeing vividly), or “oh, dear, she’s truly disorganized” (again with second confederate’s agreement). Intervention time as well as baseline time was about four minutes for each leader. Leaders were instructed to appear distracted and stutter a bit during instruction, so that this observation appeared to be partially reality-based. In the end, participants were administered a questionnaire where they rated each participant including the leader on nine items, among them the competence rating. They further completed an attitude rating.

Observational Method and Inter-Rater-Reliability

For baseline coding we observed the initial dyadic interactions of participants with each of the two leaders. Intervention coding was done while confederates displayed consensual negative affect and after confederates had made the negative verbal statement. Coding was done with a simple categorical coding scheme, distinguishing negative/skeptic from positive/open affect. All affect display that was not clearly negative or positive was coded as neutral. A more nuanced coding scheme that was tested before did not prove sufficiently reliable among raters. We used two main raters and one control rater for the codings.

pp. 58

Inter-rater reliability was calculated for the two main female raters on a sample of 143 observations. Simple percentage agreement was 75.17% (ranging from 63.79 to 89.65 for single subjects), Cohen’s kappa = .38 (p < .0003), which is not very high, but acceptable for interpretive categories (Cohen, 1960). A sentence-by-sentence coding method was used tracking target and valence of reaction. One additional experienced rater, familiar with the general research hypothesis, did control-ratings for inter-rater reliability computation purposes. She coded a sample of 17 participants (1/3 of the data). The additional rater agreed 91.79% with one of the two independent blind raters, with an acceptable Cohen’s kappa of .56 (p < .0001) according to Landis & Koch (1977).

Measures and Scales

Competence scale. Competence was assessed by ratings on 4-point scales (from 1, not at all, to 4, very much) embedded within a self-constructed 9-item trait-rating administered at the end of the study. Items were the following: (a) analytic/task-oriented, (b) cooperative / supportive, (c) displays emotions, (d) pleasant, (e) competent, (f) quality of argumentation, (g) talks a lot, (h) influence on opinion of others, and (i) assertive / dominant. The internal consistency of this scale was Cronbach’s Alpha = .81.

Attitude scale. Attitudes toward women and men were measured with eight items from the short version of the Attitude Toward Women Scale (ATW; Spence, Helmreich, & Strapp, 1973) and nine items from the Normative Gender Role Orientation-Scale (NGRO; Athenstaedt, 2000).

Results and Discussion

Affect Display. Standard Pearson Chi-square was computed for affect display effects toward male and female targets in all further analyses. Affect display was calculated for the baseline and the intervention phase. We computed the sums of positive as well as negative affect displays. This procedure yielded four aggregated variables. From the data of main rater R1 we found no significant main effect for sex of leader, with χ2(1, N = 228) = 2.38, p = .12, indicating that neither women nor men received significantly more negative or positive affect, even though in absolute frequencies women received more negative affect. From the data of main rater R2, there was a significant main effect for sex of leader, with χ2(1, N = 228) = 3.87, p < .05, indicating that women received significantly more negative affect.

We generally need to be cautious when interpreting these results, as they are based on a low number of observations of negative affect, NnegAffect = 42 (34) vs. NposAffect = 244 (194; numbers of second rater in parentheses). On average participants did not even display one negative evaluative affect in the course of the contact with the leader, possibly because the cameras were clearly visible in the room.

Contagion Effects. On average, frequencies of negative reactions toward leaders more than tripled from the baseline to the intervention phase. They were 3.25 times as high after the intervention (factor 4.2 for women participants, and 3.8 for men) which was a significant increase. Frequencies of positive affect reactions increased non-significantly. They were 1.3 times as high after the intervention (factor 1.1 for women participants, and 2.2 for men). The main effect for increase of negative vs. positive affect (emotional contagion) was χ2(1, N = 228) = 4.04, p < .05 (p = .044) for the first rater, likewise χ2(1, N = 108) = 7.76, p < .01 (p =

pp. 59

.005) for the second rater. This effect can be interpreted as the result of a contagion process after the onset of the confederates’ reactions. Frequencies of negative affect increased from 8 (9; numbers of second rater in parentheses) to 26 (33), for positive affect from 81 (79) to 133 (165) from baseline to intervention phase. This indicates that negative affect intervention led to significantly more negative affect, with three times as many displays, especially by women, while the increase in positive affect was not significant.

Awareness Effects. Regarding the level of consciousness of the process, we found that noticing negative affect intervention was not correlated with own display of negative affect. This suggests that emotional contagion processes have worked independently of conscious awareness in this context. Regarding our hypothesis that women would be more likely to consciously notice the negative affect, we found that 17 women and 4 men had noticed it, whereas 18 women and 14 men had not, χ2(1, N = 52) = 2,98, p < .08.

Competence Ratings. An ANOVA was computed for the influence of sex of leader on competence ratings. There were no differences in competence ratings, F(1, 47) = 0.68, p < .42 and F(1, 47) = 0.26, p < .62 for the two raters respectively. Differences in means were M = 3.08 for women rating the female leader, M = 2.91 for men rating the female leader, M = 3.18 for women rating the male leader and M = 3.09 for men rating the male leader. Thus there was a slight tendency for women to be more benevolent in their rating in this sample.
Attitude Effects. Attitudes toward women and men were measured with eight items from the Attitude toward Women Scale (Spence, Helmreich, & Strapp, 1973) and nine items from the NGRO-Scale (Athenstaedt, 2000) -- the latter attitude measure having recently been developed with a German speaking sample. There were no effects of gender attitudes on the display of negative or positive affect, or on competence ratings. Moreover, men and women did not differ significantly in their gender attitudes.

Other effects. We controlled for sex of participant (no effect), acquaintance between experimenters and participants (5 critical cases, no effect), and group composition (same-sex vs. mixed-sex groups; no effect). Regarding group membership: as in the previous study (Koch, 2005), in some groups evaluative affect displays differed markedly from others, suggesting that consensual affect either spread or failed to spread in single groups. However, due to low N, Chi-squares could not be computed in the majority of these cases.
In sum, Experiment 1 supports the notion of affect transmission by emotional contagion processes for negative affect. P-values suggest that the evaluative affect and emotional contagion effect are rather weak. We therefore additionally calculated the effect sizes (Cohen’s d), which were also rather small (see Table 1).

Experiment 2
Reactions of Leaders and Participants to Display of Negative Consensual Affect

Having found hints for emotional contagion processes possibly at work in consensual affect formation and the communication of attitudes in small groups, the next study now focused on the reactions of the leaders to the negative affect display. We were particularly interested in seeing whether processes of self-fulfilling prophecies (Geis, 1993; Merton, 1948) and behavior confirmation (Snyder, 1984; Snyder & Stukas, 1999) occurred on the side of the leaders or whether they were able to ward off negative influences in form of affect displays from the group. Behaviors that mediate self-fulfilling consequences include

pp. 60

the effects of nonverbal signals such as displays of emotional warmth, attention, interest and encouragement by smiles, nods, eye contact, and verbal feedback (e.g., Rosenthal & Jacobsen, 1968).

Initially, we assessed, whether the negative affect display had been noticed at all. Furthermore, we were interested in the following questions: would perceivers hold initial beliefs about leaders (first competence rating) or adopt them from confederates’ nonverbal intervention? Would perceivers join the negative affect reactions? Would targets fit their behavior to perceivers’ actions, i.e., would they become more self-focused and nervous, and would this be related to the degree of awareness of this process? Finally, would perceivers in the second competence rating interpret targets’ behaviors as confirming their beliefs?

Up to this point, we had been conducting a study with only one leader in each gender category. Even though we had attempted to control for similarity in impression formation and person perception in a series of pre-tests (ratings on relevant dimensions such as sympathy, dominance, activity, etc. and adjusting non-congruent features), we were confronted with a resulting stimulus sampling problem (Fiedler, 2000; Wells & Windschitl, 1999). In order to overcome the resulting limitations for construct validity and generalizability Experiment 2 used a total of 40 leaders. How would those leaders react to negative consensual affect, particularly regarding their gender-related causal attributions and their self-ratings on competence scales?

Participants and Design

Eighty participants (59 women and 21 men, mean age 24.8, SD = 6.7), mostly students from the University of Heidelberg, participated in small discussion groups of five persons. They were either given partial credit toward a course requirement or received a book for their participation. There were two real participants and three confederates in each group. Participants were assigned to either the leader role (18 men, 22 women) or a participant role (3 men, 37 women) resulting in two single-factorial pre-post designs (participant and leader design). 40 groups were run and videotaped. Dependent variables of the “leaders design” were awareness (nonverbal sensitivity), nervousness, attribution, and self-ratings of competence, planned dependent variables of the “participant design” were evaluative affect display, emotional contagion, and competence ratings. After data collection the participants design was completely dropped, because of serious methodological limitations, specifically, the fact that “leaders” were not accepted as authority figures by “participants”. Further, “participants” gender distribution was biased with nearly only women in this group. The only design remaining in the analysis was therefore the leaders design.

One participant had noticed that we were working with confederates and was, thus, removed from the data set. 24 out of 74 participants (21 women and 3 men) had noticed that there was some type of negative reaction in the group and had attributed it to different causes (see analysis below). None of the participants had recognized the real purpose of the study.


A procedure similar to Experiment 1 was used in this study (for a graphical overview

pp. 61

see Figure 2). Participants were told that they would participate in a cooperative problem solving task. We used the workplace value discussion task from Experiment 1 because it had proven to be sufficiently gender neutral and to provoke vivid discussions. The first participant to arrive was usually assigned the leader role and was given some time to prepare for the role. The second participant was placed in a different room and assigned the participant role. When both had completed the initial questionnaire and felt comfortable with their roles, they were guided to the group discussion room, where the three confederates (all women) were already waiting.

This experiment used three confederates as we wanted to make sure that the negative evaluative affect was classifiable as a consensual reaction of the group majority. The leaders started their scripts and the group initially complied with the leaders suggestions. Everybody introduced themselves using color labels. Leaders started with the group instructions, however, they had been told to leave the room at a certain point in order to get additional information from the experimenter. When leaders had left the discussion room participants filled in a first rating of leaders’ competence. Up to this point, no negative affect intervention had occurred from the side of the confederates. When leaders came back and started to continue on their scripts, confederates’ negative evaluative affect reactions set in. Leaders as well as participants discussed their “workplace values”, and leaders had been instructed to remind the group to formulate a decision after five minutes of discussion. If, after an additional minute, no group decision was in sight, leaders had been instructed to make a decision themselves. Thus, they were in control of time and decision-making. After the discussion, they guided participants into different rooms for the final questionnaire which evaluated participants, leaders and self.

They were then asked a sequence of questions: Did you notice anything, any reaction from the group? If so, what did you notice? And on the next sheet they were asked: Did the group tend to react positively or negatively to you as the leader? How did you notice? When did you notice? For what reasons do you think the group reacted that way (please name reasons in the sequence of their importance)? After the attribution question, they were asked to indicate their opinion on five-point scales: I really liked the role of the group leader, I would do this again, I would not do this again, I would do it again with this group, I would rather do it with another group, I can easily imagine taking a leadership role after graduation, I can easily imagine working self-employed later, I can easily imagine taking a top leading position in the course of my career. These questions were intended to find out about their degree of awareness of negative affect, their causal attributions in order to explain changes in the behavior of the group, and the behavioral and self-related consequences.

Observational Method and Observer Agreement

The observational method was identical to the one in Experiment 1 for affect ratings. Leaders were rated regarding their degree of nervousness and insecurity at four distinct points in time in the course of experiment 2 (t1: initial nervousness; t2: baseline nervousness; t3: intervention nervousness, beginning of discussion; t4: intervention nervousness, end of discussion). We computed increase in nervousness from t2 (pre-intervention) to t3 (post-intervention). For nervousness ratings inter-rater reliability of 2 x 2 independent blind raters was 82% and 79% respectively, with Cohen’s kappa = .60, p < .0001 (82%) and Cohen’s kappa = .56, p < .0001 (79%). Participants’ EAD toward leaders was rated by two main independent blind raters (one man, one woman) and two control raters. Agreement for the

pp. 62

first two raters was calculated for 1/4 of the data (219 observations). Simple agreement was 76.5% (range 62.9% to 87.5% for single subjects), with kappa = .35, p < .0001, which is not great but acceptable for interpretive ratings. The two control raters additionally rated 1/5 of the data and agreed to 89.1% (range 76% to 96% per participant), with an acceptable kappa of .55 (p < .0001) according to Landis & Koch (1977).

Measures and Scales

Competence scale. Competence was assessed by ratings on 4-point scales (from 1, not at all, to 4, very much) embedded within a self-constructed 9-item trait-rating administered before and after the affect intervention. Items were the following: (a) analytic/task-oriented, (b) cooperative / supportive, (c) displays emotions, (d) pleasant, (e) competent, (f) quality of argumentation, (g) talks a lot, (h) influence on opinion of others, and (i) assertive / dominant. The internal consistency of this scale was Cronbach’s Alpha = .83.

Attitude scale. Attitudes toward women and men were measured with eight items from the short version of the Attitude Toward Women Scale (ATW; Spence, Helmreich, & Strapp, 1973) and nine items from the Normative Gender Role Orientation-Scale (NGRO; Athenstaedt, 2000).

Results and Discussion

As noted earlier the participant condition was excluded from the analysis as the experiment failed to create the conditions that leaders were accepted as authority figures. Affect display, contagion process, and competence ratings were thus excluded from the analysis. Awareness effects and leaders’ reactions were analyszed.

Awareness Effects. Noticing negative affect (24 out of 74 cases) and displaying own negative affect again were unrelated, suggesting independence and non-deliberateness of processes. Regarding the gender hypothesis (increased nonverbal sensitivity of women), we found that overall 21 women and 3 men recognized negative affect in the group, whereas 34 women and 16 men did not. Standard Pearson chi-square values were computed. In the leader role (and thus under high cognitive load), 10 women and 3 men recognized negative affect (whereas 11 women and 16 men did not), with χ2(1, N = 74) = 4.61, p = .039 (p < .04). In the participant role (under lower cognitive load), 11 women and no man noticed the intervention whereas 22 women and 3 men did not (non-computable chi-square). Overall, at least three times as many women noticed and mentioned affect display by the group. Taking into account their baseline-ratios in the sample, 38% of women and only 16% of men noticed the negative affect intervention. Looking at the leaders’ nonverbal sensitivity alone, however, perceptual confirmation of negative expectations after negative evaluative affect intervention happened in about one third of the cases and significantly more so in female leaders than in male leaders with χ2(1, N = 40) = 6.06, p = .041. Thus, the nonverbal sensitivity hypothesis was partially confirmed.

Self-ratings of Competence. Self-ratings on competence showed no gender difference before the intervention. However, after the negative consensual affect intervention the women of our sample rated their competence as lower than the men, with F(1, 74) = 3.41, p = .069; a repeated measurement ANOVA showed that women’s change in self-rating of competence was significant, with p < .05.

pp. 63

Attributions of leaders. We were interested in how leaders reasoned, especially when they had noticed negative affect. In this part of the questionnaire, 7 women and 3 men out of 40 leaders stated directly to have noticed negative affect. However, from their attributions we clearly infer that a total of 12 women and 3 men had noticed. So, there were five women who were too hesitant to directly state that they had noticed negative reactions from the group as revealed in the debriefing (Did you notice a reaction from the group? If yes, was it rather positive or rather negative (mark field)?). Among the persons who mentioned awareness of consensual affect in the questionnaires, the 3 men mentioned six external reasons and one internal reason for the negative reaction, whereas the 12 women mentioned 14 external and 11 internal reasons. Subtracting external realistic reasons (“part of the participants was instructed by the researcher to react this way”), men had four external and one internal reason, and women had ten external and eleven internal reasons. The remaining external reasons mentioned were situational (e.g., experimental situation, camera presence, roles one had to play, too little time and highly demanding task) as well as personal (e.g., yellow/ blue were responsible for negative atmosphere, blue was passive and resistant, yellow was dominant/little cooperative, the others were not in discussion mood). Internal reasons mentioned were “my leadership style” (by the only men) and “I am younger than the others (3x), probably my instructions were unclear, I was not competent enough, not well enough prepared, I was no “respect-person”, my arguments were too weak” or “I was not sovereign, not assertive, however, this may have left a more sympathetic impression”, ”my unclear communication, insufficient sovereignty in my leadership style” (all by women).

Nervousness of leaders. No significant differences were found on nervousness observations between male and female leaders or between baseline and intervention phase (coded by four independent raters). I.e., nervousness of the leaders did not rise observably, when exposed to the negative consensus reactions.

Attitude Effects. Attitudes toward women and men were measured with nine items from the Attitude Toward Women Scale (Spence, Helmreich, & Strapp, 1973) and nine items from the NGRO-Scale (Athenstaedt, 2000). Progressiveness or traditionalism values had no influence on either affect display, or contagion, or competence ratings, nor was there a sex of participant effect. Differences in means showed that women and men had the same amount of traditionalism (very low: m = 14.63 for women and m = 14.38 for men), but women had the more progressive gender attitudes (m = 33.24 for women and m = 28.90 for men), with F(1, 74) = 3.03, p = .085.

Other effects. Besides sex of participant (with effect on self-ratings and attributions, see above), we controlled for acquaintance between leaders and participants (3 critical cases, no effect), and group membership (no effect). Motivational and emotional factors as assessed in the questionnaire did not have a significant influence on any of the dependent variables.

The low awareness of negative affect coming from the group may have been partially due to the high cognitive load of leaders. They had to coordinate the task, watch the time, the decision process and monitor the equal contribution opportunities for all participants, as well as find their own solution and argue for it -- all of this in 10 minutes. We assume that it was the role-play character of the group discussion that let the leaders remain “cool” with no increase in nervousness. Self-ratings of competence showed that before the negative affect intervention there was no significant difference between male and female leaders. However, after the intervention the women of our sample rated their competence considerably lower than the men (with F(1, 74) = 3.41, p = .069). The fact that the confederates’ ratings of female competence was at the same time significantly higher (p


pp. 64

< .05) shows how perception of self and others can differ in group situations.

Attributions of leaders shed some more light on internal processes involved in this decrease of women’s self-esteem. Content analysis of attributions showed that (a) leaders did not particularly like confederates (in many more cases than participants not liking confederates), (b) women were more hesitant than men to state that they had noticed negative affect coming from the group, even though they noticed it more often (nonverbal sensitivity hypothesis) and (c) regarding attribution patterns: when women noticed a negative reaction they more often attributed it internally, whereas men more often attributed it externally.

General Discussion

The aim of the present research was to find out whether there was an emotional contagion effect of EAD in small grous and whether negative consensual affect might have differential effects on male and female leaders. Previous studies had shown differential evaluative affect display toward male and female leaders. While the results for the evaluative affect and contagion effect were not as conclusive as expected (H3), competence ratings of leaders on rating scales were not affected: we found no pro-male bias in competence ratings of leaders. That is, Hypothesis 2 was confirmed. Hypothesis 1 was partially confirmed: in Experiment1 the second rater observed significantly more negative affect toward women, while the first rater did not. As hypothesized in H3, emotional contagion was presumably happening: negative reactions more than tripled after the negative affect intervention.

As evidenced in Experiment 2, leaders’ reactions (H4) were in many cases gender-specific: while EAD was not more negative toward women as we had hypothesized (H1), women in both experiments showed a higher awareness of negative affect displays from the group (H4b). This fact combined with a more internal attribution style of women leaders (H4c) and lower ratings of their own competence after negative affect intervention under real job conditions (H4a) supports explanations of an inner emigration or an outer rejection of leadership roles early in professional women’s careers. Differences in awareness and processing of negative reactions did, however, not cause differential displays of nervousness in men and women (H4d was rejected).

The experiments reported here extend – empirically and methodologically – the few prior studies that address bias in evaluative affect display toward male and female leaders (Brown & Geis, 1984; Butler & Geis, 1990; Koch, 2005). Affect display was reliably observable by trained judges. Experiment 1 supported the assumption that evaluative affect can spread in groups by processes of emotional contagion. Experiment 2 demonstrated that self-ratings of competence dropped considerably in female leaders from baseline to intervention measure, whereas there was no difference between t1 and t2 ratings of men. This decrease of competence self-ratings in women may be a possible explanation of the general decrease in self-efficacy beliefs during women’s first professional years and, thus, an explanation for higher drop-out rates and a lower aspiration level of women in professional contexts. Interestingly, women’s decrease in self-ratings of competence was not paralleled by an increase in women’s nervousness. Ratings of a change in leaders’ nervousness from t1 to t2 revealed absolutely no gender difference. Women generally noticed the attitude change more and attributed it more internal, yet, they did not display more insecurity.

pp. 65

Methodological limitations and suggestions for further investigations

In contrast to the lack of evidence of the assumed gender bias in EAD in our experimental face-to-face situations, there was some evidence for the transmission of negative evaluative affect from confederates’ intervention in Experiment 1. In sum, however, the experiment does not make a particularly strong point for the assumed contagion effect. We cannot completely exclude that each of the members acquired the same affective reaction on their own. Gender composition of groups puts another limitation to the experiments: groups were mostly either female with a male leader, or female with a female leader. With a different group composition, results may have been different. This possibility needs testing in future studies. Finally, while the leaders functioned fine as authorities in Experiment 1, this manipulation did not work in Experiment 2. Presumably because then leaders were peers of the same age as participants and confederates. A future option would be to replace (lay-) confederates with professional actors, hoping that this would lead to more control of the affect intervention. However, such experiments cost much time and staff.

From Hall’s meta-analyses on nonverbal gender differences we know that women are better en- and decoders of nonverbal signals (Hall, 1978, 1984; Hall & Bernieri, 2001). This finding has a number of implications for our research: Regarding our hypotheses, we expected women to be more prone to emotional contagion and female leaders to notice more and be more susceptible to negative consensual affect display by the group. Moreover, we expected female raters to have a higher detection rate of evaluative affect. We worked with female raters in most cases. However, in order to avoid gender bias we introduced a male rater to the affect codings. We found no gender difference in detection rate and agreement.

Results suggest that the transmission of evaluative affect occurs non-deliberately because awareness of intervention across experiments was unrelated to any particular degree of positive or negative evaluative affect reactions. Research in automatic processing (Bargh, Chen, & Burrows, 1996, Chen & Bargh 1997; Bargh & Wegner, 1998) has proposed that automatically activated (and chronically accessible) expectations can induce perceivers to act in line with these expectations without verbalizable knowledge (e.g., through “ideomotor processes”, Bargh, Chen, & Burrows, 1996). One difference between the two experiments consisted in a higher cognitive load for leaders of the group in Experiment 2. Even though we do not know whether this was the reason for less awareness, it might have at least contributed to it. As hypothesized, women noticed negative affect more often than men. Overall, awareness was 36.6%, awareness in women was 40% (n=38) in men 18.4% (n=7). This higher nonverbal sensitivity in women is one of the most consistent findings in gender differences in communication (Burgoon, 1994; Hall, 1984).


In these two experiments, the main interest was to assess nonverbal reactions toward leaders and their transmission in small groups. Specifically, the two studies investigated how negative reactions spread in groups, whether leaders noticed negative consensual affect, and how they reacted to it. Both, Experiment 1 and 2, were conducted as role-play social influence studies. The situation of the face-to-face encounters was socially realistic and relevant, yet trials were not totally identical and, thus, experimental control was low.

pp. 66

The observational method yielded an effective non-obtrusive measure to assess nonverbal reactions as indicators of discriminative processes in stereotype research.

It was our aim to investigate emotional contagion processes and leaders’ reactions in face-to-face small group communication. From all we have learned, young leaders in the working world may react differently on the basis of their gender when they find themselves in situations where they are confronted with negative evaluative affect. A decrease in self-esteem, measured by lower self-competence ratings, may be more likely in women. However, this depends on a number of moderators like the gender and status salience of the situation, the gender-schemata of the recipient, and the reactions of the leader. Dependent on these moderators, leaders will make different choices in situations of negative group feedback dependent on its frequency, pervasiveness, quality and intensity, their nonverbal sensitivity, their attributional style, and on their gender. Moderators like communicativeness of the situation, degree of dependency on authority, etc., need to be taken into account to come closer to an understanding of the “genderedness” of certain behaviors in the context of evaluative affect reactions and their transmission in groups. Taken together, the studies reported in this article provide partial support for the hypothesis that attitudes toward persons can be formed through simple processes of evaluative affect displays mediated by emotional contagion processes. These processes have perceptual, attributional and behavioral consequences that need further investigation.


Abele-Brehm, A. (2000a). A dual-impact model of gender and career-related processes. In T. Eckes, & H. M. Trautner (Eds.). The developmental social psychology of gender (pp. 361-388). Mahwah, NJ: Erlbaum.

Abele-Brehm, A. (2000b). Gender-gaps in early career development of university graduates. Why are women less successful than men? European Bulletin of Social Psychology, 12, 22-37.

Athenstaedt, U. (2000). Normative Geschlechtsrollenorientierung (NGRO) – Entwicklung und Validierung eines Fragebogens [Normative gender role attitudes – development and validation of a new questionnaire.]. Zeitschrift für Differenzielle und Diagnostische Psychologie, 1, 61-74.

Bargh, J. A., Chen, M., & Burrows, L. (1996). Automaticity of social behavior: direct effects of trait construct and stereotype activation on action. Journal of Personality and Social Psychology, 71, 230-244.

Bargh, J. A., & Wegner, D. M. (1998). Control and automaticity in social life. In D. Gilbert, S. Fiske, & G. Lindzey (Eds.), The Handbook of Social Psychology (pp. 446-496). Boston: McGraw-Hill.

Barsade, S. G. (2002). The ripple effect: Emotional contagion and its influence on groupbehavior. Administrative Science Quarterly, 47(4), 644-675.

Bartel, C., & Saavedra, R. (2000). The collective construction of work group moods.Administrative Science Quarterly, 45, 802-836.

Bavelas, J. B., Coates, L., & Johnson, T. (2000). Listeners as co-narrators. Journal of Personality and Social Psychology, 79, 941-952.

Bavelas, J. B., & Chovil, N. (2000). Visible acts of meaning: An integrated message model of language in face-to-face dialogue. Journal of Language and Social Psychology, 19, 163-194.

Berger, J., Connor, T. L., & Fisek, M. H. (1974). Expectation states theory. A theoretical research program. Cambridge: Winthorp.

Berger J., Fisek, M. H., Norman, R. Z., & Zeldich, M. Jr. (1977). Status characteristics and social interaction: An expectation-states approach. New York: Elsevier.

Biernat, M., & Fuegen, K. (2001). Shifting standards and the evaluation of competence. Complexity in gender-based judgement and decision making. Journal of Social Issues, 57, 707-724.

Blanck, P. D. (1993). Interpersonal expectations. Theory, research, and applications. Cambridge: Cambridge University Press.

Brief, A. P., & Weiss, H. M. (2002). Organizational behaviour. Affect in the workplace. Annual Review of Psychology, 53, 279-307.

pp. 67

Brown, V., & Geis, F. L. (1984). Turning lead into gold. Evaluations of men and women leaders and the alchemy of social consensus. Journal of Personality and Social Psychology, 46, 811-824.

Burgoon, J. K. (1994). Nonverbal signals. In M.L. Knapp, & G. R. Miller (Eds.), Handbook of Interpersonal Communication (2 ed., pp. 229-285). London: Sage.

Butler, D., & Geis, F. L. (1990). Nonverbal affect responses to male and female leaders. Implications for leadership evaluations. Journal of Personality and Social Psychology, 58, 48-59.

Cacioppo, J. T., Berntson, G. G., & Gardener, W. L. (1999). The affect system has parallel and integrative processing components. Form follows function. Journal of Personality and Social Psychology, 76, 839-855.

Chen, M., & Bargh, J. A. (1997). Nonconscious behavioral confirmation processes: The self-fulfilling consequences of automatic stereotype activation. Journal of Experimental Social Psychology, 33, 541-560.

Chen, M., & Bargh, J. A. (1999). Consequences of automatic evaluation: Immediate behavioral predispositions to approach or avoid the stimulus. Personality and Social Psychology Bulletin, 25, 215-224.

Cohen, J. (1960). A coefficient of agreement for nominal scales. Educational and Psychological Measurement, 20, 37-46.

Cohen, J. (1969). Statistical power analysis for the behavioral sciences. New York: Academic Press.

Deaux, K., & LaFrance, M. (1998). Gender. In D. Gilbert, S. Fiske & G. Lindzey (Eds.), The Handbook of Social Psychology (pp. 788-827). Boston: McGraw-Hill.

Deaux, K., & Major, B. (1987). Putting Gender Into Context: An Interactive Model of Gender-Related Behavior. Psychological Review, 94, 369-389

Dovidio, J. F., Brown, C. E., Heltmann, K., Ellyson, S. L., & Keating, C. F. (1988). Power displays between women and men in discussions of gender-linked tasks. A multichannel study. Journal of Personality and Social Psychology, 55, 580-587.

Duncan, S. Jr., Kanki, B. G., Mokros, H., & Fiske, D. W. (1984). Pseudounilaterality, simple-rate variables and other ills to which interaction research is heir. Journal of Personality and Social Psychology, 46, 1335-1348.

Dvir, T., Eden, D., & Banjo, M. L. (1995). Self-fulfilling prophecies and gender: Can women be Pygmalion and Galatea? Journal of Applied Psychology, 80, 253-270.

Eagly, A. H. (1987). Sex differences in social behavior: A social-role interpretation. Hillsdale, NJ: Erlbaum.

Eagly, A. H., & Chaiken, S. (1993). The psychology of attitudes. San Diego: Harcourt, Brace, Jovanovich.

Eagly, A. H., & Johannesen-Schmidt, M. C. (2001). The leadership styles of women and men. Journal of Social Issues, 57, 781-797.

Eagly, A. H., & Karau, S. (2002). Role congruity theory of prejudice toward female leaders. Psychological Review, 109, 573-598.

Eden, D. (1993). Interpersonal expectations in organizations. In P. D. Blanck (Ed.), Interpersonal expectations: Theory, research, and applications (pp. 154-178). Cambridge: Cambridge University Press.

Ekman, P., & Friesen, W. V. (1978). The facial action coding system. Palo Alto, CA: Consulting Psychology Press.

Fiedler, K. (2000). Beware of samples! A cognitive-ecological sampling approach to judgment biases. Psychological Review, 107, 659-676.

Geis, F. L. (1993). Self-fulfilling prophecies: A social psychological view of gender. In A. E. Beall, & R. J. Sternberg (Eds.), The psychology of gender (pp. 9-54). New York: Guilford Press.

Hall, J. A. (1978). Gender effects in decoding nonverbal cues. Psychological Bulletin, 85, 845-857.

Hall, J. A. (1984). Nonverbal sex differences. Communication accuracy and expressive style. London: Hopkins.

Hall, J. A., & Bernieri, F., J. (2001). Interpersonal sensitivity. Theory and measurement. Mahwah, NJ: Erlbaum.

Hatfield, E., Cacioppo, J. T., & Rapson, R. L. (1994). Emotional contagion. Paris: Cambridge University Press.

Heider, F. (1958). The psychology of interpersonal relations. New York: Wiley

Heilman, M. E. (2001). Description and prescription: How gender stereotypes prevent women’s ascent up the organizational ladder. Journal of Social Issues, 57, 657-674.

Kelly, J. R., & Barsade, S. G. (2001). Mood and emotions in small groups and work teams. Organizational Behavior & Human Decision Processes, 86, 99-130.

Kelley, J.R., & Spoor, J.R. (2005). Affective Influence in Groups.8th Annual Sydney Symposium of Social Psychology, March 14-17, 2005, Sydney, Australia. Available at: www.sydneysymposium.unsw.edu.au/ 2005/Papers/Kelly%20&%20Spoor.pdf

Koch, S. C. (2005). Evaluative affect display toward male and female group leaders. A replication and extension. Small Group Research, 36 (6).

Krämer, N. C. (2001). Bewegende Bewegung. Sozio-emotionale Wirkungen nonverbalen Verhaltens und deren experimentelle

pp. 68

    Untersuchung mittels Computeranimation. [Moving movement – socio-emotional impact of nonverbal behavior and its experimental assessment by computer animation] Lengerich: Pabst.

Landis, J. R., & Koch, G. G. (1977). The measurement of observer agreement for categorical data. Biometrics, 33, 159-174.

Lipps, T. (1903). Leitfaden der Psychologie (Kap. 14: Die Einfühlung, pp. 187-201) [Guide to Psychology. Chapter 14: Empathy)]. Leipzig: Wilhelm Engelmann.

Mehrabian, A. (1971). Silent Messages. Belmont: Wadsworth.

Merton, R. K. (1948). The self-fulfilling prophecy. Antioch Review, 8, 193-210.

Phillips, S. D., & Imhoff, A. R. (1997). Women and career development: A decade of research. Annual Review of Psychology, 48, 31-59.

Ridgeway, C. L. (2001). Gender, status, and leadership. Journal of Social Issues, 57, 637-656.

Rosenthal, R., & DePaulo, B. (1979). Sex differences in eavesdropping on nonverbal cues. Journal of Personality and Social Psychology, 37, 273-285.

Rosenthal, R., & Jacobson, L. (1968). Pygmalion in the classroom. New York: Holt, Rinehart, & Winston.

Snodgrass, S. E. (1985). Women’s intuition. The effect of subordinate role on interpersonal sensitivity. Journal of Personality and Social Psychology, 49, 146-155.

Snyder, M. (1984). When Beliefs Create Reality. In L. Berkowicz (Ed.), Advances in Experimental Social Psychology (Vol. 18, pp. 248-305). Orlando, FL: Academic Press.

Snyder, M., & Stukas A. A., Jr. (1999). Interpersonal Processes: The Interplay of Cognitive, Motivational, and Behavioral Activities in Social Interaction. Annual Review of Psychology, 50, 273-303.

Spence, J. T., Helmreich, R. L., & Stapp, J. (1973). A short version of the Attitudes toward Women Scale (AWS). Bulletin of the Psychonomic Society, 2, 219-220.

Walther, E. (2002). Guilty by mere association: Evaluative conditioning and the spreading attitude effect. Journal of Personality and Social Psychology, 82, 919-934.

Wells, G. L., & Windschitl, P. D. (1999). Stimulus sampling and social psychological experimentation. Personality and Social Psychology Bulletin, 25, 1115-1125.

West, C., & Zimmerman, D. H. (1987). Doing gender. Gender & Society, 1, 125-151.


pp. 69

Table 1
Means, standard deviations and effect sizes (Cohen’s d) for negative evaluative affect display toward male and female leaders in Experiment 1

  N Mean SD d (Cohen)a
    Pre Post Pre Post Pre Post

Experiment 1 53            
Neg. reactions tw women (1)   0.28 0.69 0.59 1.23 0.46 0.34
Neg. reactions tw men (1)   0.05 0.35 0.22 0.49    


Note: Effect sizes were small (d = .20 to .50); contagion effects are reflected in the increase in means from pre- to post-test; duration of observational period was: 4 min. (for pre and post observation); aCohen’s d between negative reactions of women and men in baseline phase (pre) and after intervention (post).


pp. 70

Figure 1
Experiment 1: Design and procedure

Figure 1: Experiment 1

Note: N = 53, one-factorial within-group design (male vs. female leader); 27 4-person groups (2 participants; 2 confederates); participants were exposed to a negative consensus reaction; dependent variables: evaluative affect, amount of contagion, awareness, and competence ratings; the female leader was expected to receive more negative EAD; no difference was expected for competence ratings; women were expected to be more aware of negative affect interventions. Amount of negative affect was expected to increase from t1 to t2 (contagion effect). Control variables: sex of participant, baseline affect, team membership, awareness of intervention, and gender attitude. Results of hypothesis testing: (a) EAD: Significant more negative reactions toward women leaders were observed by rater 2, but not by rater 1, (b) Contagion: significant increase of negative affect at t2, (c) Awareness: was higher in women, and (d) Competence: no gender differences.

pp. 71

Figure 2
Experiment 2: Design and procedure

Figure 2: Experiment 2

Note: N = 40 leaders and 40 participants; one-factorial (male vs. female leader); 20 5-person groups with 2 participants and 3 confederates; participant condition was dropped; dependent variables for leader condition: awareness of intervention, attributions, self-ratings of competence, and nervousness ratings; awareness, internal attributions, drop in self-ratings of competence at t2, and increase in nervousness was expected to be more pronounced in female leaders. Control variables: sex of participants, baseline nervousness, team membership, awareness, and gender attitude. Results of hypothesis testing: (a) Awareness: Negative reaction by confederates were noticed by the leaders in 37% of the cases (40% by women, 18% by men). Leaders reactions: (b) More internal attributions in women, (c) Significant drop of self-ratings of competence at t2 in women only (p<.05), and (d) No gender differences in display of nervousness.

pp. 72

Contact Information:

Sabine C. Koch,
Department of Psychology
University of Heidelberg
Hauptstr. 47-51
69117 Heidelberg
Phone: +49-6221-547297
Fax: +49-6221-547703
Email: sabine.koch@urz.uni-heidelberg.de

Author Note

Sabine C. Koch, Department of Psychology, University of Heidelberg, Germany.

This research was partially supported by a grant from the Deutsche Forschungsgemeinschaft (DFG; Kr 505/11-2) to Prof. Dr. Lenelis Kruse and Prof. Dr. Caja Thimm.

I am grateful to all participants and to the raters, who spend many hours in front of the video-tapes. Thanks to Joseph E. McGrath for invaluable comments on earlier drafts of this article, to Klaus Fiedler, Robyn Cruz, Eva Walther, Tilmann Betsch, Jeannette Schmid and Joerg Zumbach for constructive comments on the designs, and to Rebecca Luft, Christina Bähne, Phoebe Ballmann, Katrin Linser, Antje Kubat, Imke Figura, Yue Zhang, Johannes Szylvássi, Almut Stromberger and Sofia Grigoriadis who served as confederates in the experiments.

This article is dedicated to Florence Geis, who died too early.

Correspondence concerning this paper should be addressed to Sabine C. Koch, Department of Psychology, University of Heidelberg, Hauptstrasse 47-51, 69117 Heidelberg, Germany. E-mail: sabine.koch@urz.uni-heidelberg.de



  1. The term /leader/ is used in a broad sense throughout the article: we are talking about authorities with some degree of power over the groups. In Study 1 the leader is the experimenter and time keeper (confederate), in Study 2 the leader is the facilitator of the discussion and time keeper (participant).
  Received: 11/3/04
Revised: 9/3/05
Accepted: 9/10/05


Copyright © Reysen Group 2005