Switzerland is a direct democracy. Therefore, the Swiss Federal Council is structured differently than the governing bodies of other democracies. The Swiss Federal Council consists of seven politicians, ranging from conservative to social-democratic (Dardanelli, 2010; Federal Council, n.d.; Vatter, 2020). It is a heterogenous group which represents the political diversity of the Swiss population. One main idea behind the Federal Council is that it tries to find the best solutions possible, respecting all viewpoints. When a solution is found, all members unanimously represent this position. As a result, the performance of the Federal Council is often evaluated as a group. If the Swiss population is unsatisfied with a solution, it can initiate a referendum. Therefore, how the population perceives the Federal Council as a group is highly relevant (Vatter, 2020).
The only study to compare the evaluation of a list of politicians as a whole with individual evaluations provides evidence that the results of group evaluation are lower than the mean of the group members (Sears, 1983). However, other theories of judgment could be applied to the evaluation of politicians as a group, and this could allow for alternative predictions. A judgment about a group could be based on either equal-weight averaging or on weighted averaging of the group members (Anderson, 1974; Anderson et al., 1973; Skowronski & Carlston, 1989). The direction of the weighted averaging could favor either the evaluation of the group or that of its individual members. In other words, it is possible that the performance or the likeability of the Federal Council is equal to, lower than, or higher than the average of the seven members of the Federal Council.
The aim of this article is to test how the performance and likability of a real group is evaluated. Data from the two studies provide evidence that the performance of the Federal Council is perceived as higher than the mean of its members. The second study provides evidence that this effect is caused by the greater weight given to positive members and not by a reinforcement of a predominant stereotype.
Equal-weight averaging is the simplest averaging model. It is was developed in contrast to the summation model (Willis, 1960). According to the summation model, the valence of attributes of a person are added up, resulting in an overall evaluation. The summation model predicts that a person with two very negative attributes will be evaluated more positively than a person with two very negative and two moderate negative attributes. The averaging model predicts the opposite. The results suggest that evaluators do not sum up attributes but seem to average the attributes (Willis, 1960).
Studies into visual memory provide recent evidence for equal-weight averaging. Observers are quite accurate in estimating the mean of physical attributes of simultaneously presented objects, like the mean size of objects (Ariely, 2001; Chong & Treisman, 2005), the brightness of objects (Bauer, 2009), or the mean hue of a color (Demeyere et al., 2008; Webster et al., 2014). This process is called ensemble perception. In ensemble statistics, every object is given equal weight. However, when attention is involved, observers average with unequal weight (Choi & Chong, 2020; De Fockert & Marchant, 2008; van Osch et al., 2015; Whitney & Yamanashi Leib, 2018).
The only article which compared the evaluation in mean of politicians with an overall impression of a group of politicians based its prediction on the assumption that people are evaluated more positively than all other attitude objects (Sears, 1983). This results in a ‘person-positivity bias’. This general preference for human beings is caused by the similarity between the evaluator and the attitude objects. Because a person is more similar to the evaluator than a group, people are evaluated more positively than groups (Sears 1983).
While Sears (1983) found evidence for the person-positivity bias in evaluations of politicians, in a replication study the person-positivity bias was only observed for negatively evaluated people (Miller & Felicio, 1990). In one study, the person-positivity bias was replicated for unattractive people but not for those considered attractive. In a second study, the person-positivity bias was only replicated when the group members were described with negative attributes but not with positive attributes.
Actually, the person-positivity bias loses an important part of its validity when only negative people become less negative, but not all people become more positive. This contradicts the idea that similarity between individuals increases likeability in general.
The person-positivity bias predicts bias in a judgement of the members of a group and not in a judgment of the group itself (Sears, 1983). Another possibility is that in the evaluation of a group, some members are given higher weights than others. This corresponds to an unequal-weight averaging model (Skowronski & Carlston, 1989). The unequal-weight averaging models do not typically focus on group evaluations, but they could easily be applied to group evaluations. One example is that negative information is given greater weight (Rozin & Royzman, 2016). This negativity bias predicts that negatively rated members are given higher weight than neutrally or positively rated members. This again would lead to predictions that group evaluations are always lower than the mean of member ratings.
However, negative information does not always result in greater weight. Extreme or diagnostic information can also receive greater weight (Skowronski & Carlston, 1987, 1989). Positive evaluations are more diagnostic than negative information for the evaluation of the ability of a person (Martijn et al., 1992; Skowronski & Carlston, 1987, 1989). A person perceived as smart can say a lot of stupid things before the positive impression of that person changes. So, theories allow for predictions of cases in which the evaluation of the group is more positive than that of the mean members. One example where positive information is given greater weight than negative is in the domain of ability and intelligence (Martijn et al., 1992; Skowronski & Carlston, 1987, 1989). Those studies focused on impression-formation concerning individuals, but not groups. One example of positivity bias in the evaluation of grouped people is the group attractiveness effect (van Osch et al., 2015). Observers of unfamiliar grouped faces evaluate the physical attractiveness of a group more positively than the mean attractiveness of their members. Note that this corresponds to a situation where observers obtain a first impression of unfamiliar faces.
To sum up, equal-weight averaging is mainly observed regarding simple visual objects. Theories of unequal-weight averaging differ in their predictions concerning weighting. Person-positivity bias and negativity bias predict that groups are always evaluated more negatively than the mean of their members. The idea that extreme or diagnostic information receives greater weight allows us to predict that groups are evaluated more positively than the mean of the members.
Studies about the evaluation of actual, known groups are rare. In contrast, there are various studies in which participants rated grouped objects, like a list of persons who had never worked together (Sears, 1983) or grouped faces which had had no previous face-to-face interaction (van Osch et al., 2015).
In the two present studies, we asked participants to evaluate the Federal Council of Switzerland and each member of the council. In the first study, we asked the participants to rate the performance, likability, and attractiveness of the Federal Council, both as a whole and as separate individual members.
The equal-weight averaging is the null hypothesis and states that the evaluation of the Federal Council is equal to the evaluation of the members in mean. The unequal-weight averaging predicts a difference between the Federal Council and the rating of the mean of the members. A negativity bias or person-positivity bias would lead to a more negative evaluation of the Federal Council than the mean of the members. A diagnosticity approach predicts a more positive evaluation of the Federal Council than the mean of the members.
For lack of evidence for estimating the effect size, the sample size could not be precalculated. Instead, we placed the study on a virtual blackboard of a Swiss university and collected data for 21 days.
A total of 167 participants were recruited via online platforms. Due to a programming error, in the first two days 21 participants evaluated the Federal Council several times. Those participants were dismissed. Of the remaining 146 participants, 80 were female and 66 were male. The mean age was 35.53 (SD = 13.68, range = 18 to 82). One hundred and twenty-seven of the participants had the right to vote in Switzerland.
The study employed a within-subjects design. All participants evaluated the seven members of the Federal Council individually and the Federal Council as a group. The order was counterbalanced: Half of the participants evaluated the Federal Council as a group first, while the other half evaluated the members as individuals first. We also counterbalanced three ways the members of the Federal Council were ordered: from social-democratic to conservative (left to right), from conservative to social-democratic, and unrelated to political orientation. The condition assignment was random. Finally, since it can matter whether a member is evaluated within a group or individually (Walker & Vul, 2014), we manipulated how the participants individually evaluated the members of the Federal Council. Half of the participants saw only a single member of the Federal Council at a time, while the other half always saw all seven members, with an arrow indicating which member they should evaluate. Like in the studies into the group attractiveness effect (van Osch et al., 2015), we could not find any difference between these two versions of the evaluation of individual members for performance, t(144) = 0.45, p = .65, or likeability, t(144) = –0.39, p = .69, but there was a marginally significant difference for attractiveness, t(144) = –1.68, p < .010. Therefore, we ignore the between-subject variable in our analysis.
After an introduction, participants were shown a picture of the Federal Council or a member of the Federal Council, and they evaluated their performance, likeability, and physical attractiveness. At the end, participants answered questions concerning their knowledge of the Federal Council and their demographics.
All participants rated each of the seven members of the Federal Council individually and the Federal Council as a group concerning performance, likeability, and attractiveness on a nine-point scale ranging from –4 (very low performance, very unlikeable, very unattractive) to +4 (very high performance, very likeable, very attractive). For the group mean, the seven ratings of the members for each dependent variable were averaged.
Because performance ratings could be based on knowledge, we added 14 questions to assess the participants’ knowledge of the Federal Council. The questions entailed hard facts concerning, for example, which member belonged to which resort (department), or who elects the Federal Council, and soft facts, like in the question: How well does the Federal Council correspond to the diversity of the parliament? There was a high correlation between the answers, so they were averaged to one score of knowledge. The knowledge about the Federal Council does not influence the difference between the group and the member ratings for performance, b = –0.00, t(144) = –0.04, p = 0.97, attractiveness, b = –0.00, t(144) = –0.01, p = 0.99, or likability, b = –0.01, t(144) = –0.39, p = 0.70.
The conceptualization of this research was developed jointly with students in a seminar, where we discussed if the group attractiveness effect could be found in non-visual judgements as well. The idea arose that the performance of the Swiss Federal Council could be evaluated more positively than the mean performance of its members. No part of the study procedures was pre-registered prior to the research being conducted, and no part of the study analysis was pre-registered prior to the research being conducted. However, we replicated the results with a second study. All material and data are available on OSF. https://osf.io/r6x38/?view_only=eb7cd8cbce524347bfd9e82a2a32b307.
Performance of the Federal Council (M = 1.47, SD = 1.49) is higher than the mean performance of the members of the Federal Council (M = 0.64, SD = 1.27), t(145) = 10.4, p < .001, Cohen’s d = .86. Figure 1 illustrates these results.
The likeability of the Federal Council is higher (M = 1.18, SD = 1.67) than the mean likeability of the individual members of the Federal Council (M = 0.70, SD = 1.25), t(145) = 5.49, p < .001, Cohen’s d = .45. Figure 2 illustrates these results.
Contrary to the group attractiveness effect, the Federal Council is physically less attractive (M = –0.47, SD = 1.83) than the mean attractiveness of its members (M = –0.27, SD = 1.28), t(145) = 2.13, p < .01, Cohen’s d = 0.18. Figure 3 illustrates these results. Noteworthily, both attractiveness ratings are below zero, tFederal Council (145) = 3.12, p = .002; tFederal Council (145) = 2.52, p = .01, which indicates that most participants perceive the Federal Council as physically unattractive.
While person-positivity bias argues that the evaluation of group members causes a bias, we expect the unequally weighted averaging models to cause a bias. Evidence for unequally weighted averaging provides correlations of the difference between group ratings and mean members with group ratings and mean member ratings respectively. For every participant we calculated the difference between the evaluation of the Federal Council as a group and the evaluation of the members in mean (Hemmerich 2017). This difference in the performance ratings correlates more strongly with the performance of the Federal Council as a group, rPearson(144) = .54, p < .001, than with the mean performance of the members, rPearson (144) = -.13, p = .13, z = 12.86, p < .001. Likewise, the difference in the likability ratings correlates more strongly with the likability of the group, rPearson (144) = .66, p < .001, than the mean likability of the members, rPearson (144) = .02, p = .77, z = 11.54, p < .001. And finally, the difference in the attractiveness ratings correlates more strongly with the attractiveness of the group, rPearson (144) = .72, p < .001, than with the mean attractiveness of the members, rPearson (144) = .12, p = .15, z = 10.66, p < .001.
The study provides evidence that the performance and the likeability of the Federal Council is higher than the mean of its members. This is not in line with the predictions of the person-positivity bias and is in opposition to the results of Sears (1983). In addition, the results are not in line with the idea that negative members are given greater weight. In fact, these results are in line with the more flexible model, that those attributes are given greater weight which are more diagnostic for the content of the evaluation (Martijn et al., 1992; Skowronski & Carlston, 1987, 1989, 1992). Further evidence for the more flexible weighted averaging model provides the reversal of the group attractiveness effect. We did not predict the reversal of the group attractiveness effect. However, the reversal demonstrates that group attractiveness is not as general as it is proposed by van Osch et al. (2015).
An alternative explanation for the weighted averaging model is an accentuation of the evaluation of the group. Similar to the processes of outgroup derogation and ingroup favoritism (Hewstone et al., 2003), it is possible that evaluators accentuate their overall impression, so that evaluators who have the impression that most members perform well accentuate their impression in their group ratings.
The second study thus had two goals: One was to replicate the results of study one for the evaluation of the performance of the Federal Council, while the other was to rule out the alternative explanation for the group accentuation effect. For this reason, all participants were asked to build their ideal Federal Council. Like with the real Federal Council, it was only possible to select members of the two chambers of the Swiss Parliament. The ideal Federal Council is naturally considered to be more positive than the actual Federal Council. In addition, the ideal Federal Council is less heterogenous. Accentuation refers to the whole group. For example, ingroups are upvalued whereas outgroups are devalued. For most evaluators, it is not possible to accentuate a general impression of the actual Federal Council, because they evaluate some members positively and some negatively. However, the ideal Federal Council is more homogenous than the actual Federal Council, and members are evaluated mostly positively. This offers the possibility for an accentuation of the group impression. Positive accentuation of a homogenous group should be more positive than for a heterogenous group. To sum up, an accentuation predicts a greater difference between group and mean members for homogeneous than for heterogenous groups. In contrast, an unequally weighted averaging model predicts the opposite. The members of the ideal Federal Councils are less heterogenous than the actual Federal Council. Therefore have unequal weights lesser influence on group ratings. In an extreme homogeneous case in which all members of the ideal Federal Council were estimated with the same performance, any unequal weights would result in the same group evaluation.
The sample size of Study 2 size was determined to be comparable to that of Study 1. A total of 150 participants were recruited via online platforms. Eighty-seven were female and 63 were male. The mean age was 27.79 (SD = 10.04, range = 17 to 66). One hundred and thirty-two of the participants had the right to vote in Switzerland.
The design differs from Study 1 in four regards. First, participants saw only the names of the politicians and not the faces. Second, we omitted the condition where participants evaluated group members individually; instead, the names of the seven members were presented at the top of the page, while participants evaluated each member individually. Third, all participants selected and evaluated their ideal Federal Council. Fourth, we measured only performance, not likeability or attractiveness.
After an introduction, participants first either evaluated the actual Federal Council or selected and evaluated their ideal Federal Council. To select the ideal Federal Council, participants were given a list of all the members of both chambers of the Swiss Parliament. The list was sorted by political party and alphabetically. In the end, participants answered questions concerning their knowledge of the Federal Council and their demographics.
All participants rated the performance of the seven members of the actual Federal Council, the Federal Council as a group, the seven individual members of their ideal Federal Councils, and their ideal Federal Councils as a group on the same nine-point scale as in Study 1. As in Study 1, the knowledge about the Federal Council does not influence the difference between the group and the member performance ratings for the actual Federal Council, b = 0.01, t(147) = 0.55, p = 0.58. However, there was a marginally significant effect of knowledge on the difference between group and member performance rating for the ideal Federal Council, b = –0.03, t(147) = –1.76, p = 0.08.
A repeated measures ANOVA revealed two main effects and one interaction. The evaluation of the performance of the ideal Federal Councils (M = 2.48, CIlower = 2.32, CIupper = 2.63) is higher than the performance of the actual Federal Council (M = 0.97, CIlower = 0.81, CIupper = 1.12), F(1, 149) = 237.07, p < .001, η2 = .47. Like in Study 1, the performance of the group is higher (M = 1.93, CIlower = 1.80, CIupper = 2.07) than the mean performance of the members (M = 1.51, CIlower = 1.51, CIupper = 1.64), F(1, 149) = 50.75, p < .001, η2 = 0.04. In line with the predictions of the unequal weighted averaging and in contrast to the group accentuation effect, the difference between the group and the mean members is stronger for the actual Federal Council (Mgroup = 1.28, SD = 1.37; Mmembers = 0.65, SD = 1.06) than for the ideal Federal Councils (Mgroup = 2.59, SD = 0.94; Mmembers = 2.37, SD = 0.85), F(1, 149) = 16.86, p < .001, η2 = .009. Figure 4 illustrates these results.
Our argumentation of ruling out the accentuation effect is based on the higher homogeneity of the ideal Federal Council and the actual Federal Council. We therefore calculated for each person the standard deviation of the evaluation of the seven ideal councilors and the seven actual councilors. Actually, the standard deviation of the ideal councilors (M = 0.76, SD = 0.51) was lower than the standard deviation of the actual councilors (M = 1.83, SD = 0.8), t(149) = 14.1, p < .001, Cohen’s d = 1.15.
The aim of this paper was to compare the evaluation of the Federal Council as a group with the mean of the evaluation of the members of the Federal Council. The results are in line with the predictions of a model in which some members are given greater weight. In contrast, the results are not in line with the idea that all members are given equal weight. Likewise, the data contradict the idea of a person-positivity bias, a negativity bias and the group accentuation effect.
The results of both studies contradict previous studies which provided evidence that a group of politicians is evaluated more negatively than the mean of its members (Sears, 1983). However, there are a few differences between the present studies and those of Sears. One significant difference is that participants in previous studies got a list of 26 politicians from several political groups in the USA like senators, governors, retired Ford-Administration officials, state attorneys general, and members of the United States Congress. These were grouped on the list, but they did not belong to the same group. Therefore, typical attributes of groups, such as shared identity, face-to-face interaction, role differentiation, or common goals, were missing. It is impossible to rate the group performance of a mix of people who have never worked together. The goal of the studies of Sears was not to measure the evaluation of the performance of a group, but to measure an evaluation of stereotypes of politicians (Miller & Felicio, 1990). Previous studies were thus interested in how stereotypes of a group differ from the evaluation of individuals. The present studies had the goal of comparing evaluations of the performance and the likability of a real group and not the stereotypes of an abstract group.
While person-positivity bias states that the difference between group ratings and mean ratings of its members is caused by a more positive evaluation of the members, unequal averaging models locate the origin of the difference in the evaluation of the group. One such example is the negativity bias (Rozin & Royzman, 2016; Skowronski & Carlston, 1989), which predicts a higher weight to negatively rated members.
The data of the present studies do not support the negativity bias. However, the data are in line with unequal averaging. The higher correlation of the difference of the group and mean members ratings with group ratings than with the mean ratings of the members provides evidence that the rating of the group and not the rating of the members caused the effect.
What the data do not show is which person was given greater weight and for what reasons. However, the results are in line with the model that extreme or diagnostic members are given higher weight (Skowronski & Carlston, 1989). Ability is one domain in which positive information looms larger than negative information (Martijn et al., 1992; Skowronski & Carlston, 1987, 1989). The diagnosticity approach predicts that a given weight depends on the evaluation dimension so that for evaluations of performance other people get greater attention than for morality.
Noteworthily, the results of the first study contradict the theoretical propositions of the group attractiveness effect, according to which observers always place more weight on the more attractive group members when they are evaluating group attractiveness (van Osch et al., 2015). It thus seems that in the first study the observers placed more weight on less attractive members of the group.
The group accentuation effect provides an alternative explanation to the unequal weight averaging: the enhancement or devaluation of group evaluations. One example of accentuation of group evaluations is the enhancement of an ingroup and the degradation of an outgroup (Efferson et al., 2008). The group accentuation effect predicts that a predominant impression will be accentuated in the group evaluation. This would mean that the more positive (negative) the predominant impression of the group, the more positive (negative) the difference between the evaluation of the group and the mean of the members. Our second study provides evidence that this is not the case in the evaluation of the Federal Council. Participants created their ideal Federal Council, which is more positive than the actual Federal Council. However, the difference between the group and the evaluation of the members in mean was smaller for the ideal Federal Council than for the actual one. However, this is in line with the prediction of unequal weight averaging. An ideal Federal Council is less heterogeneous than the actual Federal Council. Therefore, the influences of weighted averaging are smaller. The more similar the members in their performance, the lesser the influence of weight on the average.
Another explanation for the results lies in the possibility that evaluators could base their evaluation of the group on attributes other than the evaluation of the members. Group performance could be based on the ability to find appropriate solutions which suit the variety of interests of the Swiss population. An evaluation of the performance of the members could be based more on their performance as a head of a department. For the performance ratings, we could not rule out this alternative explanation. However, this alternative explanation does not apply to likeability or attractiveness, where the data also provided evidence for unequal weight averaging.
In addition, it is possible that irrelevant factors influence performance evaluations. The evaluation of the Federal Council could be influenced by attitudes toward the Swiss political system and the evaluation of the councilors is influenced by attitudes toward the party more than towards the Federal Council. Here the risk of ‘comparing apples with oranges’ arises. However, this problem is inherent with the idea that a group could be more than its parts. The advantage and disadvantages of groups rely on processes which are less relevant to individuals, like reaching compromises or coordination losses. To sum up, that evaluation of a group and evaluations of their members could rely on different criteria is an alternative explanation which cannot be fully ruled out.
To sum up, the present studies provide evidence that the evaluation of the Federal Council of Switzerland is more positive than that its average members. This effect is not caused by an accentuation of the group evaluation, but by unequal-weight averaging. The results are in line with the diagnosticity approach, where in the domain of ability positive information is given greater weight than negative information.
We thank two reviewers for their helpful comments. Materials and data are available at https://osf.io/r6x38/?view_only=eb7cd8cbce524347bfd9e82a2a32b307.
The study was carried out in accordance with the recommendation of the Federal Act on Research involving Human Beings of the Swiss Confederation. The studies were approved by the ethics committee of the Faculty of Business, Economics, and Social Science of the University of Bern (Project Number:112021).
The authors have no competing interests to declare.
All authors gave conceptional input to this paper, analyzed the data and proofread the manuscript. Mattia Carnelli programmed both studies and provided the link for recruiting participants. He prepared data files for data analysis. He provided conceptional input, especially for the design of Study 2. Claude Messner initiated this project, analyzed data, and wrote the manuscript. Patrick Höhener double-checked the data analysis, proofread the manuscript, and wrote part of the revision.
Anderson, N. H. (1974). Cognitive Algebra: Integration Theory Applied to Social Attribution. Advances in Experimental Social Psychology, 7(C), 1–101. DOI: https://doi.org/10.1016/S0065-2601(08)60035-0
Anderson, N. H., Lindner, R., & Lopes, L. L. (1973). Integration theory applied to judgments of group attractiveness. Journal of Personality and Social Psychology, 26(3), 400–408. DOI: https://doi.org/10.1037/h0034441
Ariely, D. (2001). Seeing Sets: Representation by Statistical Properties. Psychological Science, 12(2), 157–162. DOI: https://doi.org/10.1111/1467-9280.00327
Bauer, B. (2009). Does Stevens’s power law for brightness extend to perceptual brightness averaging? Psychological Record, 59(2), 171–186. DOI: https://doi.org/10.1007/BF03395657
Choi, Y. M., & Chong, S. C. (2020). Effects of Selective Attention on Mean-Size Computation: Weighted Averaging and Perceptual Enlargement. Psychological Science, 31(10), 1261–1271. DOI: https://doi.org/10.1177/0956797620943834
Chong, S. C., & Treisman, A. (2005). Statistical processing: Computing the average size in perceptual groups. Vision Research, 45(7), 891–900. DOI: https://doi.org/10.1016/j.visres.2004.10.004
Dardanelli, P. (2010). Federal democracy in Switzerland. Federal Democracies, 142–159. DOI: https://doi.org/10.4324/9780203857571-14/FEDERAL-DEMOCRACY-SWITZERLAND-PAOLO-DARDANELLI
De Fockert, J. W., & Marchant, A. P. (2008). Attention modulates set representation by statistical properties. Perception and Psychophysics, 70(5), 789–794. DOI: https://doi.org/10.3758/PP.70.5.789
Demeyere, N., Rzeskiewicz, A., Humphreys, K. A., & Humphreys, G. W. (2008). Automatic statistical processing of visual properties in simultanagnosia. Neuropsychologia, 46(11), 2861–2864. DOI: https://doi.org/10.1016/j.neuropsychologia.2008.05.014
Efferson, C., Lalive, R., & Fehr, E. (2008). The coevolution of cultural groups and ingroup favoritism. Science, 321(5897), 1844–1849. DOI: https://doi.org/10.1126/science.1155805
Federal Council. (n.d.). Retrieved February 21, 2022, from https://www.admin.ch/gov/en/start/federal-council.html
Hemmerich, W. (2017). StatistikGuru: Korrelationen statistisch vergleichen. Retrieved from https://statistikguru.de/rechner/korrelationen-vergleichen.html
Hewstone, M., Rubin, M., & Willis, H. (2003). Intergroup Bias. 575–604. DOI: https://doi.org/10.1146/annurev.psych.53.100901.135109
Martijn, C., Spears, R., Van Der Pligt, J., & Jakobs, E. (1992). Negativity and positivity effects in person perception and inference: Ability versus morality. European Journal of Social Psychology, 22(5), 453–463. DOI: https://doi.org/10.1002/ejsp.2420220504
Miller, C. T., & Felicio, D. M. (1990). Person-positivity bias: Are individuals liked better than groups? Journal of Experimental Social Psychology, 26(5), 408–420. DOI: https://doi.org/10.1016/0022-1031(90)90066-U
Rozin, P., & Royzman, E. B. (2016). Negativity Bias, Negativity Dominance, and Contagion. 5(4), 296–320. DOI: https://doi.org/10.1207/S15327957PSPR0504_2
Sears, D. O. (1983). The person-positivity bias. Journal of Personality and Social Psychology, 44(2), 233–250. DOI: https://doi.org/10.1037/0022-35188.8.131.52
Skowronski, J. J., & Carlston, D. E. (1987). Social Judgment and Social Memory: The Role of Cue Diagnosticity in Negativity, Positivity, and Extremity Biases. Journal of Personality and Social Psychology, 52(4), 689–699. DOI: https://doi.org/10.1037/0022-35184.108.40.2069
Skowronski, J. J., & Carlston, D. E. (1989). Negativity and Extremity Biases in Impression Formation: A Review of Explanations. Psychological Bulletin, 105(1), 131–142. DOI: https://doi.org/10.1037/0033-2909.105.1.131
Skowronski, J. J., & Carlston, D. E. (1992). Caught in the act: When impressions based on highly diagnostic behaviours are resistant to contradiction. European Journal of Social Psychology, 22(5), 435–452. DOI: https://doi.org/10.1002/ejsp.2420220503
van Osch, Y., Blanken, I., Meijs, M. H. J., & van Wolferen, J. (2015). A Group’s Physical Attractiveness Is Greater Than the Average Attractiveness of Its Members. Personality and Social Psychology Bulletin, 41(4), 559–574. DOI: https://doi.org/10.1177/0146167215572799
Vatter, A. (2020). Der Bundesrat : Die Schweizer Regierung (Vol. 12). NZZ Libro ein Imprint der Schwabe Verlagsgruppe AG. DOI: https://doi.org/10.5771/9783748906810-201
Walker, D., & Vul, E. (2014). Hierarchical Encoding Makes Individuals in a Group Seem More Attractive. Psychological Science, 25(1), 230–235. DOI: https://doi.org/10.1177/0956797613497969
Webster, J., Kay, P., & Webster, M. A. (2014). Perceiving the average hue of color arrays. Journal of the Optical Society of America A, 31(4), A283. DOI: https://doi.org/10.1364/JOSAA.31.00A283
Whitney, D., & Yamanashi Leib, A. (2018). Ensemble Perception. Annual Review of Psychology, 69(1), 105–129. DOI: https://doi.org/10.1146/annurev-psych-010416-044232
Willis, R. (1960). Stimulus pooling and social perception. Journal of Abnormal and Social Psychology, 60(3), 365–373. DOI: https://doi.org/10.1037/h0048912