In small we trust lay theories about small and large groups

657360 PSPXXX10.1177/0146167216657360Personality and Social Psychology BulletinLa Macchia et al research-article2016 Article In Small We Trust: Lay Theories About Small and Large Groups Personality and Social Psychology Bulletin 2016, Vol 42(10) 1321–1334 © 2016 by the Society for Personality and Social Psychology, Inc Reprints and permissions: sagepub.com/journalsPermissions.nav DOI: 10.1177/0146167216657360 pspb.sagepub.com Stephen T La Macchia1, Winnifred R Louis1, Matthew J Hornsey1, and Geoffrey J Leonardelli2 Abstract Day-to-day interactions often involve individuals interacting with groups, but little is known about the criteria that people use to decide which groups to approach or trust and which to avoid or distrust Seven studies provide evidence for a “small = trustworthy” heuristic, such that people perceive numerically smaller groups as more benevolent in their character and intentions As a result of this, individuals in trust-sensitive contexts are more likely to approach and engage with groups that are relatively small than those that are relatively large We provide evidence for this notion across a range of contexts, including analyses of social categories (Studies and 2), ad hoc collections of individuals (Study 3), interacting panels (Studies 4-6), and generalized, abstract judgments (Study 7) Findings suggest the existence of a general lay theory of group size that may influence how individuals interact with groups Keywords group size, trust, lay theories, group perception, heuristics Received February 17, 2015; revision accepted June 5, 2016 Many day-to-day interactions we have are not with isolated individuals, but with groups of people Just like when we interact with other individuals, when we interact with groups, we often face trust-laden decisions: “Should I approach that group of strangers for help?” “Should I be worried about my upcoming meeting with the review panel?” “Should I sign a contract with this company?” These approach–avoid assessments are partly based on inferences about the character and essential trustworthiness of the group, inferences that are in turn based on limited information It is well established that trust—broadly defined as the generalized expectation of predictable and benevolent motives and/ or behavior from others (see Balliet & Van Lange, 2013)—is a core marker of group functioning Trust has implications for a wide range of outcomes including economic and organizational transactions (Balliet & Van Lange, 2013; Berg, Dickhaut, & McCabe, 1995) Less well established, however, are the superficial qualities of a group that would make them appear more or less trustworthy When approaching an individual, a person might draw on a series of intuitions regarding nonverbal cues that together form a reliable impression of (un)trustworthiness: Are they leaning forward? Do they make eye contact? Do they have a trustworthy face? (DeSteno et al., 2012; Willis & Todorov, 2006) But with groups, one cannot rely on the same set of social instincts This raises an important question: When people make instinctive judgments about the trustworthiness of groups, what heuristics or social cues they rely on? In the present research, we focus on the numerical size of the group (relative to other groups or to the context) as a possible basis for people’s lay theories about which groups can be trusted and which groups cannot We chose size because (as described in the subsequent section) there are a number of theoretical accounts that together build a case for why groups’ size might influence trust perceptions and subsequent decisions to approach or avoid them Furthermore, size is one of the more easily recognizable qualities of a group It is one of the basic components distinguishing perceptions of different group types and categories (Denson, Lickel, Curtis, Stenstrom, & Ames, 2006) and also perceptions of groups of the same type (see Lickel et al., 2000) In this article, we present seven studies that examine the role of group size in determining how external individuals make decisions about the trustworthiness of a group We propose, for the first time, that (a) a group’s size influences the extent to which outsiders trust them, and (b) such perceptions affect how outsiders believe they will be treated by small and large groups To begin, we review the relevant literature on The University of Queensland, Brisbane, Australia University of Toronto, Ontario, Canada Corresponding Author: Stephen T La Macchia, School of Psychology, The University of Queensland, St Lucia, Brisbane, Queensland 4072, Australia Email: s.lamacchia@uq.edu.au Downloaded from psp.sagepub.com by guest on September 17, 2016 1322 Personality and Social Psychology Bulletin 42(10) group size, including research on how numerical size moderates the extent to which groups are seen to be internally cooperative Smaller Groups Are “Tighter” and Are Perceived as Such Group size has been studied in several key literatures in social psychology For example, classic research on social loafing, de-individuation, bystander effects, and social facilitation has examined how attitudes and behaviors are influenced by being in the presence of large versus small groups What has received less attention, however, is the question of whether people have different lay theories about the nature of large versus small groups This question is distinct from established concepts and paradigms relating to group size, in that it directly bears upon interactions between individuals and groups This emphasis contrasts with the intragroup focus that has dominated previous research on group size (e.g., Brewer, 1991, 2003; Dunbar, 1995, 1998) and on trust (e.g., Balliet & Van Lange, 2013) The question at the heart of this article is simple: Do outsiders perceive small and large groups as different in terms of their trustworthiness, and these perceptions help shape willingness to engage with such groups? Past research has focused on perceptions of how differently sized groups operate internally (e.g., their levels of intragroup cohesion and cooperation) Such research suggests that numerically smaller groups are seen to be relatively tight, cohesive, and cooperative For example, smaller perceived group size is correlated with greater perceived collective responsibility (Denson et al., 2006) and perceived endorsement of binding moral foundations (Carnes, Lickel, & Janoff-Bulman, 2015) However, these relationships are part of a broader pattern of perceived differences between different basic group types, with intimacy groups (e.g., families and friends) seen as the smallest and most entitative, and loose associations (e.g., people living in the same town) the largest and least entitative (Carnes et al., 2015; Denson et al., 2006; Lickel et al., 2000) Intimacy groups are hence seen as more collectively responsible (Denson et al., 2006) and more bound by loyalty and care (Carnes et al., 2015) than loose associations Another theoretical approach that implies small groups are seen to be more internally cooperative is provided by optimal distinctiveness theory (ODT; Brewer, 1991, 2003; for a review, see Leonardelli, Pickett, & Brewer, 2010) ODT posits that evolved needs for interdependence and identity make people seek membership in groups that are “optimally” sized—that is, large enough to provide inclusion and security, but small enough to provide a sense of distinctiveness Provided that they are adequately inclusive and secure, groups that are small relative to the surrounding population will generally be preferred, because they are seen as more socially distinctive and trust-bound (Brewer, 2003; Leonardelli & Loyd, 2016; Pickett, Silver, & Brewer, 2002) Guided by ODT, Leonardelli and Loyd (2016) examined perceptions of intragroup trust among minimal minority and majority groups When there was a substantial proportional difference between the groups (i.e., 20% of ostensible participants in the study vs 80%), participants rated the minority group as being higher in intragroup trust than the majority group, even when they themselves were not members of the group As a result, they were more inclined to join the minority group In this respect, the participants’ intuitions that the distinctive minority group is highly cooperative is surprising, especially in light of the fact that most research to date had assumed that membership in numerical minorities would be presumed to be negative and something to be avoided (e.g., Lücken & Simon, 2005) What is striking, however, is that it parallels what research has discovered with the study of absolute group size, wherein groups with a small absolute number of people (such as three) exhibit greater cooperation than those in much larger groups, such as with 10 or 30 members (Brewer & Kramer, 1986; De Cremer & Leonardelli, 2003; Messick & Liebrand, 1995) Finally, there is evidence that when groups grow beyond a certain size, the networks of interpersonal connections that underpin trust can become strained Surveys of the size of communal and social groups in humans and nonhuman primates led Dunbar (1995, 1998) to propose that the maximum functional size of these groups is 100 to 200 members Research on online social networks appears to validate this figure (Gonỗalves, Perra, & Vespignani, 2011) The proposed reason for Dunbars number is a neurological limitation regarding the number of social contacts one can effectively monitor at any one time, which evolved to match the number at which primates are able to maintain the greatest amount and variety of stable interpersonal contact Does the Size-Trust Heuristic Extend to Outsiders? Studies on group size that we have discussed so far share one important boundary condition: They all relate to perceptions of the group’s internal dynamic, rather than perceptions of how the group might treat outsiders The empirical story so far suggests not that small groups are seen as trustworthy, but that they are seen to be cooperative among themselves However, if it is true that members or relatively small groups are seen to be relatively benevolent toward each other, would outsiders expect this benevolence to extend to them? None of the theoretical accounts of group perception presented so far predict such a generalization Indeed, there is some evidence that intragroup cooperativeness can translate into less-thanbenevolent behavior toward nonmembers For example, numerical minorities in minimal-group experiments exhibit higher levels of ingroup bias and outgroup discrimination than numerical majorities (Leonardelli & Brewer, 2001) It might be the case that people have an intuitive sense of this, seeing smaller groups as tighter and therefore perhaps as Downloaded from psp.sagepub.com by guest on September 17, 2016 1323 La Macchia et al more cliquish or clannish This might lead to larger outgroups being trusted more (“large = trustworthy” heuristic) The opposite prediction is equally plausible, however Given that intimacy groups (e.g., family, friends) form the smallest basic group type, it is possible that relatively small outgroups evoke perceptions related to intimacy groups— namely, greater internal cohesion and trust (Carnes et al., 2015; Denson et al., 2006; Lickel et al., 2000) Another possibility is suggested by social impact theory (Latané, 1981): People may have a lay sense of social impact, such that they see smaller groups as easier to influence In a transactional encounter, people would therefore expect a smaller outgroup to be more likely to the right thing by them, or to give them what they want received two small-group and two large-group scenarios in alternating order The scenarios were presented in random order for each participant, so that scenario varied independently of group size Outcome variables were trust toward the group and willingness to engage with them Materials and procedure Four scenarios were used Overview of Studies This article reports seven studies investigating the relationship between individuals’ perceptions of outgroup size and of trustworthiness, and the resulting effects on individuals’ expectations and approach intentions toward groups Studies and examine size and trustworthiness of category-type groups, exploring outsiders’ consideration of risky economic interactions with groups (Study 1) and whether group members, when attempting to describe their groups to outsiders as trustworthy, describe them as smaller (Study 2) Studies to examine size and trustworthiness of interacting groups, with participants imagining approaching a nonspecific group of people in a room (Study 3) or being evaluated by a panel described as large or small (Studies 4-6) Study examines possible mechanisms of an apparent group-size preference by directly asking participants why they trust smaller or larger groups more in general Study In Study 1, we devised four scenarios in which the participant makes a trust-sensitive financial decision (e.g., signing a contract with a company) In each scenario, the group is described as either relatively small or relatively large We conducted the study as an exploratory test of whether participants’ levels of trust toward a group would be influenced by its size Method Participants and design A sample of 120 first-year psychology students (69.2% female, Mage = 20.23 years) signed up for what was advertised as a study on individual attitudes and decision making Each participant read four scenarios Each scenario described an ambiguous risk–reward situation in which the participant could potentially gain or lose something by trusting a target group The independent variable was manipulated such that in each scenario the group was described as either relatively small or large All participants A company is seeking to develop and market an invention by the participant Participants were told “the company would own the prototype and design patents, and your control over future development would rely greatly on their good will.” A municipal council owns a vacant lot next to a house the participant is thinking of purchasing, and has promised that rumors it will use the site for an industrial waste facility are false, even though “local rumors persist.” A company is selling a laptop computer online, which is the exact model that the participant is looking for, for half the price offered by its competitors Participants were told, “you had a previous experience buying an expensive item from a previous website, in which you received the wrong item and it took several months for the situation to be fixed.” The participant has lost his or her wallet in a seaside town and, hours later, has to decide whether to cancel their cards or wait until later to see whether the wallet will be turned in Group-size manipulation In each scenario, the target group was described as the same numerical size in the small-group and large-group conditions, but this size was made to look smaller or larger than average (e.g., “The town’s population is around 10,000 people, making it around [half/twice] the size of most other towns in the region”) This was done so that any psychological effects were not confounded with pragmatic considerations associated with objective group size (e.g., a town of 20,000 people being more densely populated than a town of 10,000 people) Manipulation check A single written-response item followed each scenario description, asking participants whether the group was small, large, or average size Trust Within each scenario, trust was measured using five items adapted from Tyler and Huo (2002) Items included “I would trust the [target group]” and “I think that the [target group] would try hard to the right thing by me” (1 = strongly disagree to = strongly agree) A preliminary (size: small, large) × (scenario order) ANOVA, with size as a within-subjects factor and scenario as a between-subjects factor, showed that the effect of size was equivalent across scenario order, interaction: F(1, 118) = 0.54, p = 466 Thus, responses were collapsed Downloaded from psp.sagepub.com by guest on September 17, 2016 1324 Personality and Social Psychology Bulletin 42(10) Table Study 1: Means, Standard Deviations, and Correlations Among Measures Small Variable Small Trust Willingness Large Trust Willingness Large M (SD) Trust Willingness 3.81 (0.96) 3.64 (1.02) — 61*** — 3.56 (0.98) 3.37 (0.97) 21* 10 09 17 Trust Willingness — 50*** — *p < 05 ***p < 001 across scenarios by combining participants’ responses to the two targets of each size This created separate indices of trust toward small (α = 86) and large groups (α = 87) Willingness Willingness to take the risk by positively engaging with the group was measured using three items specific to each scenario Examples for the council scenario include “On the basis of the information provided, how willing would you be to purchase the property?” and “On the basis of the information provided, how willing would you be to look for another property to purchase instead?” (reversescored) All items were measured on a (very unwilling) to (very willing) scale As with trust, a (size) × (scenario order) ANOVA revealed no significant interaction, indicating stability of the size effect across scenarios, F(1, 118) = 0.02, p = 891 Thus, participants’ responses to the two targets of each size were averaged, creating separate indices of willingness to take a risk toward small (α = 72) and large groups (α = 69) Results and Discussion The size manipulation was successful Across all scenarios, participants were more likely to answer all manipulation check items correctly (i.e., by recognizing both small groups as small and both large groups as large; n = 90) than to fail any of the manipulation check items (n = 29), χ2(1) = 31.27, p < 001, ϕ = 1.04 Means, standard deviations, and correlations among the dependent measures are shown in Table by group size Consistent with a “small = trustworthy” heuristic, the groups were trusted more when they were described as relatively small than when they were described as relatively large, t(119) = −2.18, p = 031, 95% confidence interval (95% CI) = [−0.46, −0.02], d = 0.40 Participants also showed a stronger willingness to engage with the groups when they were described as relatively small than when they were described as relatively large, t(119) = −2.28, p = 025, 95% CI = [−0.50, −0.03], d = 0.42 Unsurprisingly, the ratings of trust were closely linked with willingness to engage Specifically, difference scores between trust ratings for the small and the large groups were closely correlated with difference scores between willingness ratings for the small and the large groups, r = 57, p < 001 In sum, groups were seen to be more trustworthy when they were described as relatively small, and these ratings of trustworthiness were associated with increased willingness to take a risk by engaging positively with that group The effect was modest in size but consistent across a range of scenarios It should be noted that these effects emerged even though we kept the objective size of the groups constant It was the mere perception that the group was more or less large that was enough to trigger the “small = trustworthy” heuristic Study Study results suggested a “small = trustworthy” heuristic when it comes to engaging with other groups If this lay theory about groups is truly entrenched, then it may be the case that these attributions should also apply reflexively, such that people believe that the relative size of their own groups says something about their own trustworthiness in the eyes of others Study examines this question: Do people use the size of the groups to which they belong as a social signal for trustworthiness? If so, one would expect that participants who are asked to self-present as highly trustworthy will report their groups (e.g., their home towns; their schools) as significantly smaller than participants who are asked to self-present as untrustworthy Method Participants and design A community sample of 202 U.S residents took part via Amazon Mechanical Turk (MTurk) in return for a payment of US$0.50 Data were deleted for participants who failed an attention check,1 leaving a final sample of 190 participants (59% males; Mage = 33.84 years) Participants were randomly assigned to one of two faking conditions: trustworthy or untrustworthy Materials and procedure Participants read the following instruction: Downloaded from psp.sagepub.com by guest on September 17, 2016 1325 La Macchia et al we’d like you to answer in a way that makes you look like a completely [trustworthy/untrustworthy] person, even if it means answering untruthfully We want you to fake your responses wherever and however you need to in order to come across as [trustworthy/untrustworthy] as possible This may seem like a strange exercise, but please just the best you can Participants then rated the size of five groups: “the church/ synagogue/mosque/temple you attend,” “your high school,” “the main organization you work for/study at,” “the town/ city you were born in,” and “the town/city you live in currently.” The response scale was (very small) to (very large), with the midpoint anchor average size Responses on the five items were averaged to form an index of membership group size, α = 84 Results and Discussion Consistent with hypotheses, participants faking being trustworthy reported their groups as significantly smaller (M = 4.26) than participants faking being untrustworthy (M = 4.82), t(154.09) = −2.28, p = 024, 95% CI = [−1.05, −0.07], d = 0.37 Thus, not only people harbor a general belief that members of smaller groups are more trustworthy than members of larger groups, but they also believe that other people share this lay theory This suggests that people might, under certain circumstances, self-present by adjusting the relative size of their category groups A person trying to look trustworthy might mention that they grew up in a small town or city, for example, and withhold the fact that they were educated in a large school or work in a large company Study Studies and demonstrate a clear “small = trustworthy” heuristic in the context of social categories (e.g., towns, companies) However, if this heuristic is a deep-seated general lay theory of groups, it should be demonstrable across a range of contexts and with respect to a range of groups (not just social categories) In Study 3, we explored the robustness of the group-size trust effect by seeing whether it could be found between smaller and larger clusters of interactive individuals To this, we asked participants to think of a small or large group of people in a room and rate how they felt toward that group Method Participants and design Eighty-nine first-year psychology students (62% female, Mage = 18.54 years) were randomly allocated to imagine either a large or a small group of people in a room Dependent variables comprised participants’ ratings of their feelings toward the group on various dimensions related to trust and sociability, as well as their willingness to meet and interact with the group Materials and procedure Following the demographic items, a paragraph informed participants that the following sections of the questionnaire were aimed at examining “how people form impressions of groups of people when only given minimal information about them.” Participants were then given the following instruction: “Imagine there’s a [small/large] group of people in a room For the items below, please circle a mark on each line to indicate how you feel toward the group.” This was followed by 12 semantic-differential items, responded to on a line with eight evenly spaced tick-marks (coded as −4 to +4 with no zero/midpoint) Half the items were positively scored (e.g., “unfriendly–friendly”) and the other half reversescored (e.g., “trusting–distrustful”) Factor analysis using oblimin rotation revealed two factors that together accounted for 55% of the variance.2 After one item was deleted that cross-loaded >.40 (“unwelcome–welcome”), scores on the two factors were extracted for further analysis Items related to trust loaded on a single factor (Trusting) defined by seven items (e.g., “trusting,” “safe,” “relaxed”) Participants’ sociable feelings toward the group formed a second factor (Sociable), defined by four items (“approaching,” “outgoing,” “friendly,” and “interested”) The two factors were significantly positively correlated, r = 45, p < 001 Willingness to approach Willingness to approach the imaginary group in the room was measured using the items “To what extent you think you would want to enter the room?” and “To what extent you think you would want to meet or interact with the group?” (1 = not at all, = very much; r = 65, p < 001) Manipulation check Participants were asked, “How many people you think are in the group?” As expected, participants instructed to think of a large group imagined significantly more people in the room (M = 69.32) than those instructed to think of a small group (M = 7.21), t(33.03) = 3.06, p = 004, 95% CI = [−103.46, −20.76], d = 1.06 Results and Discussion To test the unique effects of group size on trusting and social feelings toward the group, one-way between-subjects ANCOVAs were conducted on each of the extracted factors (Trusting and Sociable) with the other factor as a covariate Consistent with Studies and 2, participants felt significantly more trusting (controlling for sociable feelings) toward a small group (Madjusted = 0.22) than toward a large group (Madjusted = −0.20), F(1, 85) = 4.92, p = 029, η2 = 04 However, there was no significant difference in how sociable participants felt (controlling for trust) toward a small group (Madjusted = −0.16) compared with a large group (Madjusted = 0.11), F(1, 85) = 2.05, p = 156, η2 = 02 A between-groups t test showed that participants were also no more willing to meet or interact with a small group (M = 4.69) than a large group (M = 4.86), t(87) = 0.56, p = 580, 95% CI = [−0.46, 0.82], d = 0.12 Downloaded from psp.sagepub.com by guest on September 17, 2016 1326 Personality and Social Psychology Bulletin 42(10) Figure Indirect effect model of the effect of group size (−1 = small, +1 = large) on willingness to approach the group (Study 3) Note Coefficients shown are unstandardized Follow-up analyses were conducted to determine whether there was an indirect effect of group size on approach willingness via the Trusting factor, controlling for sociability This analysis was conducted with Hayes’s (2013) PROCESS model of parallel mediation (Model 4), using bootstrapping with 10,000 resamples and 95% CIs As shown in Figure 1, Trusting (95% CI = [0.26, 0.72]) and Sociable (95% CI = [0.66, 1.11]) feelings toward the target group were both uniquely associated with greater willingness to engage with the target However, Trusting was marginally associated with size, p = 079, 95% CI = [−0.39, 0.02], whereas Sociable was not, p = 659, 95% CI = [−0.16, 0.26] Furthermore, a significant indirect effect of group size on willingness emerged via Trusting feelings, 95% CI = [−0.2220, −0.0004], but not via Sociable feelings, 95% CI = [−0.14, 0.22] When the Sociable factor was excluded from the model, the indirect effect via the focal Trusting variable was still significant, 95% CI = [−0.34, −0.03] In sum, participants felt more trusting when thinking of approaching a small group of people This trust preference toward the smaller group, however, was not part of a global preference; there was no social halo effect In particular, participants felt equally sociable toward small and large groups, and overall were equally willing to approach small and large groups Unlike in Study 1, the effect of group size via trust on willingness appears to have been balanced by other factors suppressing the influence of trust and/or favoring the large group (e.g., inclusiveness) The absence of established (vs ad hoc) trust measures and items relating to other basic social attributes (e.g., agency/competence) also makes interpretation of these processes difficult Study To create a more vivid experimental context, Study compared perceptions of smaller and larger groups, but this time in an interaction scenario with an emphasis on expected outcomes Specifically, participants imagined they were called in front of a disciplinary panel to account for a transgression they had committed Two scenarios were used to assess the generalizability of effects: The transgression was either academic plagiarism or unprofessional workplace behavior Participants were asked questions about the likely attributes of the panel, how much they trusted it, and the severity of the outcome they expected They answered these questions twice (in counterbalanced order): once for a scenario involving a small panel (three people), and once for the same scenario involving a large panel (10 people) On the basis of the “small = trustworthy” effect found in Studies to 3, we expected that the smaller panel would be trusted more than the larger panel, and would be expected to punish less severely Although Studies to supported the hypothesis of smaller groups being perceived more favorably than larger groups, we not hypothesize this favorability as a general “halo effect.” Rather, we present a more focused hypothesis that groups described as relatively small should be seen as more benevolent to outsiders Thus, to sharpen our analysis of people’s lay theories around group size, we tested the prediction that smaller groups would be seen as relatively warm in character (a benevolence-related trait) but not necessarily more competent (a trait unrelated to benevolence), and we analyzed these perceptions as mediators of the trust difference Method Participants A community sample of 234 participants took part via MTurk in return for a payment of US$0.50 Data were deleted for participants who did not progress through all screens of the questionnaire or who failed any of the attention checks,3 leaving a final sample of 191 participants (50% female, Mage = 33.95 years) Design The independent variable was panel size (small or large), which was manipulated within subjects (and in counterbalanced order) Scenario (academic or workplace) was manipulated between subjects Materials and procedure Each scenario was two paragraphs long and written in second person In both scenarios, the description made it clear that the participant had committed a transgression, but the appropriate level of disciplinary action (which would be decided by the panel) was ambiguous The academic scenario described that the participant was studying full time at university, and had recently submitted an assignment for which they had plagiarized the last section due to being in a hurry to finish The participant was therefore to appear before an academic integrity panel, having been reported by their professor for academic misconduct The workplace scenario described that the participant worked in an office in a job that they were reasonably paid for and satisfied with The participant had been working in the job for years, but during that time, “[had]n’t really been maintaining the boundaries between work and personal life the way that company policy [said they] should.” Examples Downloaded from psp.sagepub.com by guest on September 17, 2016 1327 La Macchia et al Figure Illustrations included as part of the within-subjects manipulation of panel size condition in Studies and Note a = small panel condition, b = large panel condition included taking home stationery from the office, using a company laptop for personal and recreational purposes, and meeting with their significant other for coffee during work hours After repeated warnings, the participant had been informed by their boss that they were to face a disciplinary panel of their superiors “as part of a new crackdown on infringements of company policy.” Size manipulation Following the scenario description was the panel size manipulation: “Imagine that the disciplinary panel mentioned above consists of [three/10] people, none of whom you know personally.” The three-person panel and 10-person panel defined the small and large conditions, respectively To enhance the salience of the manipulation, a picture appeared underneath the sentence showing a grayscale, silhouetted panel of three or 10 businesspeople, with all people on the panel facing the viewer (see Figure 2) This picture appeared again at the top of each new screen (i.e., above the outcome measures) Measures On a scale from (not at all) to (very much), participants first rated the group on nine traits Four traits related to warmth: “warm,” “helpful,” “kind,” and “sensitive” (small α = 97, large α = 92); and four related to competence: “competent,” “ambitious,” “confident,” and “assertive” (small α = 79, large α = 72) These items were taken from Jeffries, Hornsey, Sutton, Douglas, and Bain (2012) As a check of the group-size manipulation, participants also rated the group according to the attribute “large.” Trust toward the panel was measured using the same five items used in Study (small α = 90, large α = 88), with the object of each statement changed to “the panel” (e.g., “I would trust the panel”) Three items measured the severity of the expected disciplinary outcome The first two items were “Think about the panel’s final decision How likely is it that the outcome of this decision will be harsh?” and “How likely is it that the panel will go easy on you?” (reverse-scored) They were responded to on a scale of (not likely at all) to (very likely) The third item asked participants to choose what they thought would be the panel’s most likely decision from a scale of seven outcomes ranging from (least severe) to (most severe) For example, the least severe outcome in the academic plagiarism scenario was “Give you an official warning,” and the most severe was “Deny credit from the course + expel you from the university.” Participants’ responses on the three items were averaged (small α = 76, large α = 72), with higher scores indicating more severe expected outcomes Results Manipulation check Confirming the manipulation, the 10-person panel (M = 7.03) was rated as significantly larger than the three-person panel (M = 3.12), F(1, 189) = 525.08, p < 001, η2 = 74 Overview of analyses A series of (size) × (scenario) mixed ANOVAs was conducted The Size × Scenario interaction was not significant in any of these analyses, Fs ≤ 0.36, ps ≥ 551, η2s < 01, indicating that all effects of size condition were consistent between the two scenarios Two main effects of scenario emerged, such that the academic panel was seen as more competent and trustworthy than the workplace panel, but because they are not relevant to the hypotheses they will not be discussed further In the following section, we focus on the effects of panel size condition, which were significant on all measures Effects of group size Main effects of panel size were found on warmth, F(1, 189) = 12.11, p < 001, η2 = 06; competence, F(1, 188) = 11.75, p < 001, η2 = 06; and trust, F(1, 189) = 8.37, p < 001, η2 = 04 As expected, the small panel was seen to be warmer (M = 4.90) and more trustworthy (M = 3.38) than the larger panel (Ms = 4.51 and 3.20, respectively) In contrast, the larger panel (M = 6.93) was seen to be more competent than the smaller panel (M = 6.70) Downloaded from psp.sagepub.com by guest on September 17, 2016 1328 Personality and Social Psychology Bulletin 42(10) Figure Multiple mediation model for Study (showing unstandardized coefficients) Note Variables were analyzed as difference scores of large panel condition minus small panel condition (L − S), with the sum scores of all mediators entered as control variables Comp = competence A main effect of panel size also emerged on severity, F(1, 189) = 16.12, p < 001, η2 = 08, with the smaller panel (M = 4.28) expected to punish less severely than the larger panel (M = 4.69) Mediation of the size effect on severity A mediation model was analyzed (using PROCESS Model 4) with the effect of panel size on expected punishment severity mediated by warmth and competence (in parallel) and then by trust (see Figure 3) Panel size could not be entered into the model because it was manipulated within subjects Instead, difference scores were calculated (large minus small) for warmth, competence, trust, and severity In line with Judd, Kenny, and McClelland’s (2001) method for regression-based within-subjects mediation analysis, sum scores (large + small) were also calculated for warmth, competence, and trust, and entered into the models as control variables Controlling for the sum score of a within-subjects variable when analyzing its unique relationship with another (using difference scores) serves to counteract the problematic covariance between sets of difference scores (Judd et al., 2001) The model revealed that warmth significantly predicted trust, 95% CI = [0.33, 0.45], whereas competence did not, 95% CI = [−0.11, 0.10] Furthermore, the greater perceived warmth of the small panel was associated with less severe expected punishment, 95% CI = [−0.70, −0.51] The indirect effect of small-group warmth preference via trust was significant, 95% CI = [−0.45, −0.23], although the residual association of smallgroup warmth preference with lower perceived severity was still significant, 95% CI = [−0.38, −0.15] By contrast, the overall path from competence to severity (controlling for warmth) was not significant, 95% CI = [−0.08, 0.26] Neither was its indirect effect via trust, 95% CI = [−0.10, 0.10], indicating that competence was not a significant mediator of the panel size difference in expected punishment severity.4 Discussion As expected, participants’ expectations when facing a disciplinary panel varied according to whether the panel comprised three or 10 people, such that the smaller panel was expected to punish less severely Consistent with the findings of Studies and 2, this group-size effect was significantly mediated by a trust preference in favor of the small panel Furthermore, the difference in perceived warmth completely accounted for the size effect on trust In other words, the panel of three was seen as warmer than the panel of 10, which also meant that it was trusted more This increased trust was in turn associated with expecting the small panel to be more lenient and fair The larger panel was seen as more competent, but this did not affect trustworthiness perceptions or outcome expectations Together with Study 3, this study elaborates our understanding of the group-size trust effect by venturing beyond the large-scale, category-type groups in Studies and to demonstrate the effect among small-scale, interactive groups The same trust preference toward the smaller group appears to emerge both when there is no explicit purpose or importance to the interaction (Study 3) and when the scenario presented involves a clear purpose and serious ramifications (Study 4) Study We wanted to investigate whether these key effects found in Study (i.e., a trust preference for the smaller group, leading to the expectation of more favorable outcomes) would still be found when not explicitly contrasting the small and large groups Study was therefore conducted as a between-subjects replication of Study To simplify the design and analyses, and because effects were not moderated by scenario in Study 4, only the academic plagiarism scenario was used Method Participants and design A community sample of 400 U.S residents took part via MTurk in return for a payment of US$0.75 Data were deleted for participants who did not progress through all screens of the questionnaire or who failed any of the attention checks (see Note 3), leaving a final sample of 370 participants (54% male, Mage = 38.24 years) The study used a one-way between-subjects design, with panel size (small vs large) as the independent variable Materials and procedure The academic plagiarism scenario description and panel size manipulation were as per Study An extra sentence was added to the panel size manipulation for Study to ensure that the small and large panel would not be seen as consisting of different kinds of members: “The panel has been selected to include a mix of gender, age and seniority.” Measures The same manipulation check item and three-item severity measure (α = 81) were used as in Study However, trust toward the panel was measured using a longer scale of Downloaded from psp.sagepub.com by guest on September 17, 2016 1329 La Macchia et al was significantly mediated by a trust preference in favor of the small panel These effects remained significant when participants only contemplated one size of panel (i.e., when they were tested between subjects as opposed to within subjects) Those contemplating facing a panel of three people were more trusting and expected more leniency than those considering a panel of 10 people Study Figure Indirect effect model of the effect of panel size (−1 = small, +1 = large) on expected punishment severity via trust of the panel (Study 5) Note Coefficients shown are unstandardized nine items The first four items were scenario specific (e.g., “How much you think the panel would listen to you/give you a voice?” = not at all, = completely) The remaining items consisted of the same set of five trust items used in Study The nine items were averaged together onto a to scale (α = 93), such that higher scores indicated greater trust of the panel Results Manipulation check In line with the manipulation, the 10-person panel (M = 5.99) was rated as significantly larger than the three-person panel (M = 2.98), t(362.17) = −15.85, p < 001, d = 1.67 Effects of group size Consistent with earlier results, there was a significant effect of panel size on trust, t(368) = 2.12, p = 035, d = 0.22 As expected, the small panel (M = 4.45) was trusted more than the larger panel (M = 4.20) There was also a significant panel size effect on expected punishment severity, t(368) = −2.20, p = 029, d = 0.23, such that less severe punishments were expected at the hands of the small panel (M = 4.65) compared with the large panel (M = 4.91) Mediation of the size effect on severity A simple mediation model was analyzed (using PROCESS Model 4) testing mediation of the effect of panel size on expected punishment severity via trust (see Figure 4) The model confirmed that trust significantly negatively predicted severity, 95% CI = [−0.65, −0.48], and showed a significant indirect effect of size on severity via trust (i.e., mediation by trust) as predicted, 95% CI = [0.01, 0.14] The residual association between size and severity controlling for trust was not significant, 95% CI = [−0.04, 0.15] Discussion As in Study 4, participants’ expectations when facing a disciplinary panel varied according to whether the panel comprised three or 10 people, such that the smaller panel was expected to punish less severely Again, this group-size effect In Study 6, we again used a trust context involving evaluation by a panel, but one that involved a positive contingency (reward) rather than a negative one (punishment) This transition allows us to partly address potential effects of panel size on impressions regarding the situation, which may influence participants’ panel preference In a reward context, if anything, a larger panel might indicate a more substantial and desirable reward, as opposed to a more severe transgression and punishment It also allows us to see whether trust and approach preferences toward smaller and larger interacting groups follow different patterns in positive- and negative-contingency situations Another methodological shift in Study is that in the current scenario participants are not physically present in front of the panel In Studies to 5, where the scenario dictated that the participants imagine attending the hearing, it could be that large panels would have been simply more anxietyprovoking, and that this may have helped drive the results By removing the need to physically attend the panel discussion, Study removes this possibility Participants were asked to imagine having their job application reviewed by a panel, and to directly compare what the panel would be like depending on whether it comprised or 10 people In line with the preceding studies, we predicted that participants would expect the smaller panel (compared with the larger panel) to be seen as more trustworthy and (as a result) more preferable We expected this effect to emerge on both trust toward the panel and participants’ anticipated likelihood that their application would be successful Method Participants and design A community sample of 203 U.S residents (57% male, Mage = 31.00 years) took part via MTurk in return for a payment of US$0.50 The study used a oneway within-subjects design, with panel size (small vs large) as the independent variable Materials and procedure A paragraph described the scenario to participants in second person, telling them to imagine they had applied for a job Participants were explicitly assured that they were suitably qualified for the job, and very interested in it According to the description, the participant had received a confirmation email saying the application would be reviewed by a panel of people who would decide whether Downloaded from psp.sagepub.com by guest on September 17, 2016 1330 Personality and Social Psychology Bulletin 42(10) Figure Illustrations included as part of the within-subjects manipulation of panel size condition in Study Note The illustrations accompanied initial instructions to participants explaining that for every question item, they were to consider and compare the two possible sizes of the panel reviewing their job application They also appeared as the scale endpoints for every measure Table Study 6: Means, Standard Deviations, and Correlations Among Measures Variable Large Warmth Competence Trust Success Choice M (SD) 3.26 (1.41) −1.01 (1.33) −0.36 (1.93) −0.85 (1.32) −0.95 (1.52) −1.72 (2.14) — −.20** −.04 −.28*** −.21** −.19** — 44*** 56*** 45*** 42*** — 50*** 41*** 36*** — 64*** 47*** — 62*** Note All variables were scaled from −4 (strongly applies to or favors small panel) to +4 (strongly applies to or favors large panel) *p < 05 **p < 01 ***p < 001 to shortlist them for the job The email also said that the participant would be notified of the outcome within weeks Size manipulation Participants were presented with the following instruction: “For the sets of questions that follow, we would like you to consider and compare two possibilities regarding how many people are on the panel reviewing your job application: either the panel has three people, or it has 10 people.” Below this were labeled, silhouetted illustrations of the two panels (see Figure 5), smaller versions of which would appear as the endpoint or item labels for all measures Measures All panel-related items had 9-point response scales, with responses coded from −4 (the small panel possessed the attribute much more than the large group) to +4 (the large panel possessed the attribute much more than the small group), and indicating no difference As a manipulation check, participants rated the two panels’ relative size, with higher scores representing stronger perception of the large panel as larger than the small panel Perceived warmth and competence comprised the same items used in Study 4, plus an additional, negatively scored item for each (“cold” for warmth and “inept” for competence) Although the five warmth items were able to be averaged into a single scale (α = 77), no subset of the five competence items was reliable at α ≥ 60, so a single item (“competent”) was used to represent the construct (we return to the instability of the competence construct in this paradigm in the discussion) Trust was measured using eight items (α = 82) Participants compared the small and large panel as to which would be more likely to, for example, “try hard to be fair,” “be sympathetic to my situation,” and “make a decision shaped solely by professionalism.” Participants’ expectation of a successful outcome (success) was measured using four items assessing the extent to which the smaller panel (−4) or larger panel (+4) would be more likely to consider hiring the participant (e.g., “ shortlist your application”; α = 82) The final measure (choice) consisted of a single item: “All in all, to what extent would you prefer your application to be reviewed by one panel or the other?” (−4 = strongly prefer panel of people to +4 = strongly prefer panel of 10 people) Results Means, standard deviations, and correlations are shown in Table Panel size effects were assessed using one-sample t tests, testing means against the value of (representing the scale midpoint) The panel size manipulation was again successful, with the 10-person panel rated as significantly larger than the three-person panel, t(202) = 32.86, p < 001, 95% CI = [3.06, 3.45], d = 4.62 Effects of group size Consistent with previous results, the small panel was seen as more trustworthy than the large panel, t(202) = −9.14, p < 001, 95% CI = [−1.03, −0.67], d = 1.29 Downloaded from psp.sagepub.com by guest on September 17, 2016 1331 La Macchia et al Figure Study mediation model of effects on choice of panel (unstandardized coefficients) Note Each construct represents a direct comparison between large and small panels; in each case, the construct represents a small-group preference Participants also expected that the small panel would be warmer, t(202) = −10.81, p < 001, 95% CI = [−1.19, −0.83], d = 1.72, and would be more likely to consider hiring them, t(202) = −8.89, p < 001, 95% CI = [−1.16, −0.74], d = 1.25 Unsurprisingly, participants preferred the small panel overall, t(202) = −11.47, p < 001, 95% CI = [−2.02, −1.43], d = 1.61 Inconsistent with previous studies, the smaller panel was also seen to be more competent, t(202) = −2.66, p = 008, 95% CI = [−0.63, −0.09], d = 0.37 Mediation of size effects on choice A mediation model congruent to that of Study was tested (see Figure 6), examining mediation of the effect of panel size on choice, whereby overall participants preferred to have their application reviewed by the small panel As previously reported, participants trusted this small panel more Differential warmth perceptions, 95% CI = [0.29, 0.53], and competence perceptions, 95% CI = [0.13, 0.30], were both associated with trust, representing significant indirect effects of smaller size generating greater trust via warmth and competence, respectively The overall path from differential warmth to panel choice (total effect) was significant, with greater perceived warmth of the smaller panel associated with a stronger preference for that panel, 95% CI = [0.29, 0.74] The indirect effect of warmth on choice via trust was significant, 95% CI = [0.07, 0.35], although the residual effect of warmth on choice remained significant controlling for trust, 95% CI = [0.08, 0.55] Competence was also associated with choice: its overall path, 95% CI = [0.09, 0.39], and indirect effect via trust, 95% CI = [0.04, 0.19], were both significant However, the direct effect of competence on choice (controlling for trust) was not significant, 95% CI = [−0.02, 0.29] Discussion The job application scenario in this study is conceptually similar to the economic risk scenarios in Study 1, in that it involves a potential payoff or reward However, the scenario was different from those in all preceding studies in that it did not involve any potential interaction or negotiation, nor any potential risk or loss Results show that trust was nonetheless still important in evaluations and expectations toward the group, and still favored the smaller group relative to the larger group Of all the studies in this article, Study makes group size most salient, with the small and large panel directly compared on every question item This might help explain why the group-size trust effect seems much larger in this study than in the previous studies (all ds > 1.00) This suggests that although the effect may often be subtle and context-sensitive, it may become stronger as size becomes more salient Study The results of Studies to show a clear preference for trusting smaller groups over larger groups, which may lead to greater expectations when interacting with smaller groups (Studies to 6) and a greater willingness to interact with them (compared with larger groups) if given the choice (Study 6) Studies and show that the higher trust toward smaller groups is mediated by higher warmth perceptions, but this leaves open the question of underlying process—that is, precisely why it is that a smaller group is generally seen as warmer and more trustworthy than a larger group of the same type Study was conducted to help shed light on this question Participants were directly asked why they saw smaller groups (or larger groups) as more trustworthy, and their open-ended responses were recorded They also answered scaled items aimed at comparing the perceptions and mental models evoked by small and large groups Method Participants and design A community sample of 50 U.S residents took part via MTurk in return for a payment of US$0.30 A correlational survey design was used that included qualitative responses Measures and procedure Participants first read a couple of sentences that explained that there are many types of groups people belong to, and each type of group can vary in size (i.e., number of people) Participants then answered a dichotomous choice item that asked, “For any particular type of group, which statement you feel is more true in general?” Participants chose between the options “those that are smaller are more trustworthy” and “those that are larger are more trustworthy.” Participants then provided open-ended explanations responding to the question, “Please explain your answer above What is it about smaller/larger groups that makes them more trustworthy?” Participants then answered six Likert-type scaled (1-7) items Four of these items measured the extent to which small groups evoke intimacy groups (e.g., “When I think Downloaded from psp.sagepub.com by guest on September 17, 2016 1332 Personality and Social Psychology Bulletin 42(10) about a small group, it reminds me of my family and friends”; α = 62; items averaged) One item measured the expectation of smaller group benevolence (“A small group would treat me better than a large group would”) The remaining item measured perceived smaller group influenceability (“It would be easier for me to influence a small group than a large group”) Results and Discussion Consistent with the preceding studies, 86% of participants (n = 43) indicated that smaller groups were generally more trustworthy Of these participants, 44% (n = 19) justified this choice by referring to smaller groups being more cohesive or close-knit, and 35% (n = 15) cited people within smaller groups being more accountable Responses to the scaled items were tested against a mean of (scale midpoint) in one-sample t tests They showed, among small-group-trusting participants, high levels of agreement regarding smaller groups being likely to treat them better, M = 5.69, t(42) = 9.64, p < 001, d = 2.97 Small groups were also strongly seen as evoking intimacy groups, M = 5.72, t(42) = 11.98, p < 001, d = 3.70, and being easier to influence, M = 5.84, t(43) = 11.29, p < 001, d = 3.44 These results are consistent with the reasons for a smaller group trust preference outlined earlier First, smaller groups were trusted more because they apparently evoke the positive perceptions associated with intimacy groups; and second, smaller groups are seen as easier to influence, leading to better expected outcomes from interacting with them as an individual outsider General Discussion All else being equal, individuals trust smaller groups more than larger groups This effect was found when people rated large, category-type groups (Study 1), ad hoc clusters of individuals (Study 3), and interactive panels (Studies 4-6) It was found when the scenario involved participants being physically present in front of the group (Studies 3-5), and when it did not (Study 6) It emerged on direct measures of trust (Studies 1-7) as well as on indirect measures (Studies 4-6), using both between-subjects and within-subjects experimental designs.5 Furthermore, it emerged regardless of whether the groups differed in objective numerical size (Studies 3-6) or did not (Study 1) Not only people hold a “small = trustworthy” heuristic when it comes to groups, they appear to be aware that other people hold this heuristic, and adjust their self-presentations accordingly (Study 2) In many cases, the “small = trustworthy” heuristic affected participants’ approach–avoid intentions In Studies and 6, for example, participants preferred to engage with the smaller target groups, effects that were mediated through perceptions of the trustworthiness of the group In the other study that measured approach intentions (Study 3), this direct effect did not emerge However, in this study there was still an indirect effect of the size manipulation on intentions via trust Overall, then, it appears that the “small = trustworthy” heuristic has the potential to guide people’s decisions about which groups to approach and which to avoid, although in some contexts this effect may also be tempered by other pragmatic considerations It should be noted that the preference for small groups appears specific to beliefs about the benevolence (i.e., warmth and trustworthiness) of the group’s character and intentions On criteria unrelated to benevolence (e.g., competence), there was either a weak tendency to favor the small group (Study 6) or a clear preference for the large group (Studies and 5).6 Therefore, small groups are seen as more trustworthy and benevolent in their intentions toward nonmembers, but they not benefit from a global halo effect This is consistent with our hypothesized reasons why small groups might be trusted more than large groups, as supported by the results of Study First, small groups may remind people of intimacy groups and thus elicit perceptions of communal traits, such as warmth (see Carnes et al., 2015; Denson et al., 2006; Lickel et al., 2000) Second, small groups may also be seen as easier to influence (in line with social impact theory; Latané, 1981), which in a trust interaction might make them appear less likely or able to betray one’s trust The small-group trust preference evidenced in our results complements previous research proposing or demonstrating that smaller groups are perceived as more cohesive and cooperative than larger groups (e.g., Brewer, 2003; Carnes et al., 2015; Denson et al., 2006; Leonardelli & Loyd, 2016) However, we designed our contexts specifically to emulate some of the many individual-to-group interactions encountered in modern social living, such as dealing with a commercial or governmental organization (Study 1) and facing a review panel (Studies 4-6) The robustness and reliability of the pattern of results is consistent with a fundamental sense of having less to fear and more to gain from interacting with a smaller group When given the choice between a smaller group and a larger group of the same type (e.g., company), it appears that people will be more inclined to choose the smaller group When having to interact with a group, people will trust it more if they know it is relatively small compared with others Groups could potentially use this to their advantage by making themselves look small compared with others, for example, in advertising Yet, the implications of the results must be understood in the context of their limitations Limitations and Future Directions The scenarios featured in the present studies were explicitly designed to be trust sensitive, so caution should be taken in extrapolating the results In each case, the target group’s intentions and the likely outcomes were ambiguous and likely to depend on the benevolence of the group toward the Downloaded from psp.sagepub.com by guest on September 17, 2016 1333 La Macchia et al perceiver Some scenarios may have included factors that would potentially favor choosing the small group (e.g., potentially better treatment from a smaller company in Study due to them having fewer other clients to service) Nonetheless, the generalization of results across multiple contexts speaks against the effect being reducible to such factors In the trust-sensitive contexts examined, people appear to expect better treatment from smaller groups In more predictable interactions, this preference may not occur; future research could examine this possibility Furthermore, competence generally did not significantly predict group trust in our studies, presumably due to the fact that competence was not a major antecedent of possible outcomes There was no reason for participants to expect that the group’s level of competence would affect the outcome in question (e.g., a small disciplinary panel and a large disciplinary panel would be equally able to inflict a severe punishment) In other contexts, however, it is conceivable that perceived competence would play a larger role If the group’s capability to deliver positive outcomes to the individual is the main unknown quantity, rather than their will to so (e.g., in certain purchase or investment contexts), such conditions might override a preference for interacting with smaller groups based on increased benevolence expectations Future research could seek to replicate the effect and to identify contrasting contexts where a large-group preference might emerge Other possibilities for future research include comparing the effects of group size to those of other attributes (e.g., cohesiveness, intragroup similarity, status), in terms of interaction and relative strength Moreover, research could investigate whether the apparent lay theory of larger groups being less trustworthy contributes to the reversal of the bystander effect under some conditions (see van Bommel, van Prooijen, Elffers, & Van Lange, 2012) Specifically, individuals may sometimes be more likely to intervene in a larger group because they trust the group less to intervene Conclusion Seven studies provide empirical evidence for a general group-size trust effect, whereby individuals outside a group trust the group more if it is relatively small than if it is relatively large Here the “small = trustworthy” effect appears to be consistent across contexts: small-scale and large-scale, category-type and interactive groups, rewards and punishments Furthermore, our results indicate that the effect may be strongest where trust and group size are most salient When approaching or interacting with a group, many rational and irrational psychological processes will affect our perceptions, expectations, and actions When the situation involves or depends on trust, it appears that group size plays an important role in these processes, with smaller size making us trust a group more and expect more from them Regarding groups, the present findings can be summarized by combining two simple axioms: Size does matter, and less is more Acknowledgments The authors thank Paul Bain, Bill von Hippel, and Tom Denson for their comments on earlier drafts of the manuscript Declaration of Conflicting Interests The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article Funding The author(s) received no financial support for the research, authorship, and/or publication of this article Supplemental Material The online supplemental material is available at http://pspb.sagepub.com/supplemental Notes The attention-check item instructed participants as follows: “To show that you’re responding properly, please select [the minimum-score response] for this item.” Choice of a two-factor solution was based on examination of the scree plot Attention-check items consisted of two comprehension-check items, as well as catch items assessing whether participants were generally inclined to respond correctly to instructions The comprehension items assessed whether participants correctly understood (a) what they were being punished for and (b) how big the group was The catch items consisted of the same attention-check item used in Study 3, as well as Oppenheimer, Meyvis, and Davidenko’s (2009) “Decisions and Activities” task Analyses conducted without excluding poor responders showed reduced mediation effects, but the main effects on trustworthiness were consistent Excluding competence from the model did not substantially affect the strength of the warmth and trust paths to severity Sample sizes in all studies allowed for cell ns in excess of the 30 per cell recommended for measuring mean differences (e.g., Cohen, 1988) Study had the smallest cell size, with 40 to 50 participants per cell in a two-sample between-subjects design All other studies had around 90 or more participants in each theoretically relevant cell In Study 3, this preference was shown on the sociability factor This appears at odds with the established concept of sociability as a communal trait overlapping with warmth and trust However, such overlap may have been blurred in Study due to the abstractness of the “group in a room” paradigm producing noisy variability in the groups and approach strategies imagined References Balliet, D., & Van Lange, P A M (2013) Trust, conflict, and cooperation: A meta-analysis Psychological Bulletin, 139, 1090-1112 doi:10.1037/a0030939 Berg, J., Dickhaut, J., & McCabe, K (1995) Trust, reciprocity, and social history Games and Economic Behavior, 10, 122-142 doi:10.1006/game.1995.1027 Brewer, M B (1991) The social self: On being the same and different at the same time Personality and Social Psychology Bulletin, 17, 475-482 doi:10.1177/0146167291175001 Downloaded from psp.sagepub.com by guest on September 17, 2016 1334 Personality and Social Psychology Bulletin 42(10) Brewer, M B (2003) Optimal distinctiveness, social identity, and the self In M R Leary & J P Tangney (Eds.), Handbook of self and identity (pp 480-491) New York, NY: Guilford Press Brewer, M B., & Kramer, R M (1986) Choice behavior in social dilemmas: Effects of social identity, group size, and decision framing Journal of Personality and Social Psychology, 50, 543-549 doi:10.1037/0022-3514.50.3.543 Carnes, N C., Lickel, B., & Janoff-Bulman, R (2015) Shared perceptions: Morality is embedded in social contexts Personality and Social Psychology Bulletin, 41, 351-362 doi:10.1177/0146167214566187 Cohen, J (1988) Statistical power analysis for the behavioral sciences (2nd ed.) Hillsdale, NJ: Lawrence Erlbaum De Cremer, D., & Leonardelli, G J (2003) Cooperation in social dilemmas and the need to belong: The moderating effect of group size Group Dynamics: Theory, Research, and Practice, 7, 168-174 doi:10.1037/1089-2699.7.2.168 Denson, T F., Lickel, B., Curtis, M., Stenstrom, D M., & Ames, D R (2006) The roles of entitativity and essentiality in judgments of collective responsibility Group Processes & Intergroup Relations, 9, 43-61 doi:10.1177/1368430206059857 DeSteno, D., Breazeal, C., Frank, R H., Pizarro, D., Baumann, J., Dickens, L., Lee, J J (2012) Detecting the trustworthiness of novel partners in economic exchange Psychological Science, 23, 1549-1556 doi:10.1177/0956797612448793 Dunbar, R I M (1995) Neocortex size and group size in primates: A test of the hypothesis Journal of Human Evolution, 28, 287296 doi:10.1006/jhev.1995.1021 Dunbar, R I M (1998) The social brain hypothesis Evolutionary Anthropology, 6, 178-190 Gonỗalves, B., Perra, N., & Vespignani, A (2011) Modeling users’ activity on Twitter networks: Validation of Dunbar’s number PLoS ONE, 6(8), e22656 doi:10.1371/journal.pone.0022656 Hayes, A F (2013) An introduction to mediation, moderation, and conditional process analysis: A regression-based approach New York, NY: Guilford Press Jeffries, C H., Hornsey, M J., Sutton, R M., Douglas, K M., & Bain, P G (2012) The David and Goliath principle: Cultural, ideological, and attitudinal underpinnings of the normative protection of low-status groups from criticism Personality and Social Psychology Bulletin, 38, 1053-1065 doi:10.1177/0146167212444454 Judd, C M., Kenny, D A., & McClelland, G H (2001) Estimating and testing mediation and moderation in within-subject designs Psychological Methods, 6, 115-134 doi:10.1037/1082989x.6.2.115 Latané, B (1981) The psychology of social impact American Psychologist, 36, 343-356 doi:10.1037/0003-066X.36.4.343 Leonardelli, G J., & Brewer, M B (2001) Minority and majority discrimination: When and why Journal of Experimental Social Psychology, 37, 468-485 doi:10.1006/jesp.2001.1475 Leonardelli, G J., & Loyd, D L (2016) Optimal distinctiveness signals membership trust Personality and Social Psychology Bulletin, 42, 843-854 doi:10.1177/0146167216643934 Leonardelli, G J., Pickett, C L., & Brewer, M B (2010) Optimal distinctiveness theory: A framework for social identity, social cognition, and intergroup relations In M P Zanna & J M Olson (Eds.), Advances in experimental social psychology (Vol 43, pp 63-113) San Diego, CA: Academic Press Lickel, B., Hamilton, D L., Wieczorkowska, G., Lewis, A., Sherman, S J., & Uhles, A N (2000) Varieties of groups and the perception of group entitativity Journal of Personality and Social Psychology, 78, 223-246 doi:10.1037/0022-3514.78.2.223 Lücken, M., & Simon, B (2005) Cognitive and affective experiences of minority and majority members: The role of group size, status, and power Journal of Experimental Social Psychology, 41, 396-413 doi:10.1016/j.jesp.2004.08.006 Messick, D M., & Liebrand, W B G (1995) Individual heuristics and the dynamics of cooperation in large groups Psychological Review, 102, 131-145 doi:10.1037/0033-295X.102.1.131 Oppenheimer, D M., Meyvis, T., & Davidenko, N (2009) Instructional manipulation checks: Detecting satisficing to increase statistical power Journal of Experimental Social Psychology, 45, 867-872 doi:10.1016/j.jesp.2009.03.009 Pickett, C L., Silver, M D., & Brewer, M B (2002) The impact of assimilation and differentiation needs on perceived group importance and judgments of ingroup size Personality and Social Psychology Bulletin, 28, 546-558 doi:10.1177/0146167202287011 Tyler, T R., & Huo, Y J (2002) Trust in the law: Encouraging public cooperation with the police and courts New York, NY: Russell Sage van Bommel, M., van Prooijen, J.-W., Elffers, H., & Van Lange, P A M (2012) Be aware to care: Public self-awareness leads to a reversal of the bystander effect Journal of Experimental Social Psychology, 48, 926-930 doi:10.1016/j.jesp.2012.02.011 Willis, J., & Todorov, A (2006) First impressions: Making up your mind after a 100-ms exposure to a face Psychological Science, 17, 592-598 doi:10.1111/j.1467-9280.2006.01750.x Downloaded from psp.sagepub.com by guest on September 17, 2016 ... between willingness ratings for the small and the large groups, r = 57, p < 001 In sum, groups were seen to be more trustworthy when they were described as relatively small, and these ratings... attempting to describe their groups to outsiders as trustworthy, describe them as smaller (Study 2) Studies to examine size and trustworthiness of interacting groups, with participants imagining... group-size trust effect by seeing whether it could be found between smaller and larger clusters of interactive individuals To this, we asked participants to think of a small or large group of people in