Group Processes &Intergroup Relations

2006 Vol 9(1) 43–61

GPIR

The Roles of Entitativityand Essentiality inJudgments of CollectiveResponsibility

Thomas F. Denson, Brian Lickel, Mathew Curtis and DouglasM. StenstromUniversity of Southern California

Daniel R. AmesColumbia University

Two studies investigated the roles of entitativity and essentiality in judgments of collectiveresponsibility. Analyses focused on four group types (i.e. intimacy groups, task groups, socialcategories, and loose associations). Repeated measures analyses revealed that intimacy groupsand task groups were rated highest in entitativity while intimacy groups and social categorieswere rated highest in essentiality. Correlational analyses revealed that entitativity played a morecentral role in judgments of collective responsibility for all four group types. However, tests ofinteraction effects revealed that essentiality moderated the effect of entitativity on blamejudgments. Implications of the role of collective responsibility in intergroup relations arediscussed.

keywords collective responsibility, entitativity, essentialism, lay theories,intergroup relations

SHOULD parents be held responsible for thebad acts of their children? Were all Germans toblame for the Holocaust? Are all the employeesof a company accountable for the criminal actsof a few coworkers? Are all Palestinians toblame for the act of a single suicide bomber?Each of these questions concerns group-based,or collective, responsibility. The formal legalsystem provides one answer to such questionsand, with few exceptions, holds responsible andpunishes only the direct causal agents of a badact. However, a growing body of research indi-cates that lay people cast a broader net in

assigning blame. In many instances, lay peopleappear to believe that shared group member-ship with a wrongdoer is a basis for blame.

Collective responsibility refers to how per-ceivers assign blame to individuals who werenot direct causal agents of negative events but

Copyright © 2006 SAGE Publications(London, Thousand Oaks, CA and New Delhi)

9:1; 43–61; DOI: 10.1177/1368430206059857

Author’s noteAddress correspondence to: Thomas F.Denson, Department of Psychology, SGM 501,University of Southern California, Los Angeles,CA 90089-1061, USA [email: denson@usc.edu]

do share a social association with a wrongdoer(e.g. Chiu & Hong, 1992; Hamilton, 1978;Lickel, Schmader, & Hamilton, 2003; Sanders etal., 1996; Shultz, Jaggi, & Schleifer, 1987). Suchjudgments of collective responsibility are oftheoretical and practical interest because theyfrequently play a role in intergroup hostilityand aggression. For example, Israelis mayassign responsibility to the Palestinian people asa whole for the acts of a single suicide bomber.Conversely, Palestinians may blame all Israelicitizens for the acts of the government andmilitary.

In this paper, we will introduce a folk theor-etical perspective that will provide answers totwo central questions concerning lay notions ofgroups and responsibility that underlie suchjudgments. First, for which kinds of groups areperceivers more likely to make judgments ofcollective responsibility? Our openingexamples came from quite different groups. Isit reasonable to argue that lay people assessresponsibility for a social category (such as‘Germans’) in the same manner as for a familyor work group? The two studies presented inthis paper are the first to systematically addressthis question. Second, from the lay person’spoint of view, what is the ‘glue’ (Hamilton,Sherman, & Rodgers, 2004) that makes a par-ticular social association worthy of collectiveblame? We will empirically examine two folktheories, group entitativity and group essentiality,that lay perceivers may use when making suchjudgments. Each of these folk theories is a plau-sible basis for a lay person to assign collectiveresponsibility—the current empirical researchis the first to our knowledge to clearly differen-tiate the roles of entitativity and essentiality injudgments of collective responsibility.

Collective responsibility in differenttypes of groups

We opened this paper by asking how lay peoplethink about collective responsibility in verydifferent types of groups. Is it reasonable toargue that a single framework for group-basedresponsibility can be applied to such differenttypes of groups such as nationality, family, and

company? Certainly, it seems that these groupsdiffer in many respects and past research indi-cates that perceivers possess an implicit folktaxonomy distinguishing between four types ofgroups (Lickel et al., 2000; Lickel, Hamilton, &Sherman, 2001). These four group types areintimacy groups (e.g. friends, families), socialcategories (e.g. women, Jews), task groups (e.g.a jury, co-workers), and loose associations (e.g.people in line at a bank, people who likeclassical music).

These different types of groups in the folktaxonomy are to some degree mirrored in howsocial psychologists have divided their field ofstudy with close relationships researchersfocusing on intimacy groups, organizationalscholars focusing on task groups, and stereotyp-ing and social identity researchers focusing onsocial categories (and perhaps, with techniqueslike the minimal group paradigm, loose associ-ations). Some scholars who focus on studyinghow people think about social category mem-berships might argue that the application ofcollective responsibility to such groups is of acompletely different nature than applyingresponsibility to a task or intimacy group. Weargue that there may be important lay theoriesused in collective responsibility judgments,such as entitativity and essentiality, which cutacross different group memberships. However,we agree that such a claim requires evidence.Therefore, in our studies, we will examine howentitativity and essentiality predict collectiveresponsibility judgments across different typesof groups (e.g. task groups versus socialcategories) but also within a particular type ofgroup (e.g. among different social categories).

Entitativity and collectiveresponsibility

Social psychologists have long been interestedin the social influence of interpersonal andintragroup relationships on people’s behavior(Asch, 1955; Cialdini, 2000; Hovland, Janis, &Kelley, 1953; Janis, 1971; Milgram, 1974;Moscovici & Zavalloni, 1969; Sherif, 1966).However, lay people appear to also have astrong (if imperfect) understanding of the

Group Processes & Intergroup Relations 9(1)

44

effects of social influence on behavior. Weargue that folk theories of social influence playa key role in collective responsibility judgments.Furthermore, we argue that group entitativity istightly connected, if not synonymous, with per-ceptions of social influence within a group. Theterm entitativity was coined by Campbell (1958)to describe the degree to which groups are per-ceived as coherent entities. Campbell’s analysisessentially applied visual gestalt principles (e.g.physical similarity, common movement, bound-edness, etc.) to groups and the first subsequentwork on entitativity (Knowles, 1976) kept to thisperspective. Some later work stressed ideas ofhomogeneity or the behavioral consistency ofgroup members (e.g. Castano & Yzerbyt, 1998;McConnell, Sherman, & Hamilton, 1997;McGarty, Haslam, Hutchinson, & Grace, 1995).

However, another line of recent work hassought to differentiate entitativity from percep-tions of group member homogeneity and todefine entitativity in terms of the perceivedinterdependence of group members, ratherthan their similarity. For example, Gaertner andSchopler (1998) defined entitativity as the inter-connection of group members and showed thatsocial interaction influenced the extent towhich people saw an ingroup (and a relevantoutgroup) as high in entitativity. Welbourne(1999) argued for a distinction between behav-ioral consistency versus the coordination ofgroup members’ goals and actions; her researchindicated that perceptions of high levels of jointgoals and coordinated action had information-processing effects that high levels of behavioralconsistency did not. Lickel et al. (2000)examined the extent to which a variety of vari-ables predicted perceptions of entitativity andfound that variables related to group memberinterdependence (e.g. interaction, commongoals) were strongly related to perceptions ofentitativity, whereas other relevant variablessuch as the group’s size, its permanence, andthe impermeability of membership in the groupwere much less strongly related to perceptionsof entitativity. Thus, many (though certainly notall) researchers’ working definitions of entitativ-ity revolve around the idea of perceived inter-dependence among group members. One value

of such a definition is that it is fairly distinctfrom other concepts such as group-memberhomogeneity or, as we will discuss, essentiality.

Furthermore, entitativity (defined in terms ofinterdependence) has been shown to predictgroup-based responsibility. In our research ofpeople’s judgments of collective responsibilityfor the Columbine High School shootings, wefound that entitativity predicted the extent towhich different groups were blamed for theactions of the shooters (Lickel et al., 2003).Lickel et al. (2003) did not analyze the data bygroup type, but did find that groups ratedhigher in entitativity were assigned greaterlevels of collective responsibility than groupsrated lower in entitativity. For example,intimacy groups such as the killers’ parents andfriendship group (i.e. the Trenchcoat mafia)were rated highest in entitativity and collectiveresponsibility while less entitative loose associ-ations (e.g. people in their neighborhood)were rated lower in both variables. Thisresearch also showed that two inferences aboutgroup social influence by fellow groupmembers may explain why collective responsi-bility is linked to entitativity. These inferences—responsibility by omission (failure to preventthe bad act) and commission (indirectlyencouraging it)—were both predicted byratings of entitativity of the group. However, forsome groups that were peers of the shooters,commission was a stronger predictor ofresponsibility, whereas for groups with anauthority relationship to the shooters (such astheir families) omission was more important.Thus, although both omission and commissionwere related to perceptions of entitativity, theywere not simply synonymous with it. The studiesof the Columbine shootings provided a startingpoint for investigating the role of entitativity incollective responsibility judgments. However,the studies were bound by particular details ofthe Columbine shootings. More important,those studies did not contrast entitativity withother variables that might account for responsi-bility judgments. In particular, perceptions ofessentiality were not considered. Thus, we focuson the contrast between entitativity and essen-tiality in the current research.

Denson et al. collective responsibility

45

Essentiality and collectiveresponsibility

Lay theories of essentiality generally refer toperceptions of a ‘deep’, inalterable, biologicalquality that gives rise to surface features.Drawing on work examining people’s under-standing of other biological objects (e.g. Atran,1994; Malt, 1995; Medin, 1989), social psychol-ogists have begun to consider the extent towhich lay people think of some groups asdefined by an underlying essence. Allport(1954) first noted the importance of essential-ity in social categories and its potential role inprejudice. More recently, Rothbart and Taylor(1992) argued that social categories are per-ceived as natural kinds (objects with an under-lying essence as opposed to being humanartifacts). Lickel et al. (2000) examined twovariables (the group’s duration and its imper-meability) derived from Rothbart and Taylor’s(1992) analysis and found these variables to bedistinct from perceptions of entitativity. Haslamand colleagues (Haslam, Rothschild, & Ernst,2000, 2002; Rothschild & Haslam, 2003) havealso argued that lay perceivers use essentialismas an explanatory mechanism for socialcategories. Furthermore, they developed a two-factor scale for assessing essentialist beliefsabout social categories. The ‘natural kinds’dimension (which most closely resembles con-ceptualizations of essentiality) consists of beliefsin the naturalness, immutability, historicalinvariance, permeability, and prerequisitefeatures of groups. They demonstrated thatessentiality varies across social categories, with‘women’ (for example) rated high on thenatural kinds dimension, while ‘midwesterners’rated low on the natural kinds dimension.

Empirical work suggests that essentialitymight play a role in collective responsibilityjudgments. For example, essentiality is associ-ated with dispositional judgments (Yzerbyt,Corneille, & Estrada, 2001; Yzerbyt, Rocher, &Schadron, 1997) and sexual prejudice (Haslamet al., 2002), suggesting that perceivers doindeed use a folk theory of essence to inferinformation about group members. It isbelieved that perceivers use essence as an

explanatory mechanism by attributing featuresof the group to all members (Yzerbyt, et al.,1997). It is conceivable that if a perceiverbelieves that social categories all possess a deepgenotypic similarity, s/he may infer that allmembers of the social category may behave ina similar manner when placed in a similar situ-ation. The perceiver might also believe that theother members of the group possess the samebad character as the actor, and be blamed forbeing ‘bad people’. Thus, for either of thesereasons, essentiality might lead to group blamefor the act of the wrongdoer. Another indirectpiece of evidence comes from the individualdifference literature, which demonstrates thatperceivers who believe individual humanspossess an unalterable fixed nature (i.e. anessence) are more likely to punish moral trans-gressors than individuals who believe in mal-leable human nature (Chiu, Dweck, Tong, &Fu, 1997).

However, it remains unclear whether laytheories of essentiality simply influence disposi-tional judgments or whether they influencegroup blame and responsibility judgments aswell. While previous research (Lickel et al.,2003) has demonstrated that the entitativity ofa group may influence judgments of collectiveresponsibility, it is less clear that essentialityinfluences these perceptions and judgments.For instance, while lay theories of entitativityrelate to the functioning of the group and thedegree of social influence exerted by groupmembers (see Lickel et al., 2000, 2003;Moreland & McMinn, 2004), lay theories ofessentiality do not provide such information.

Yzerbyt and colleagues (Yzerbyt et al., 2001;Yzerbyt & Rocher, 2002) have discussedinstances in which entitativity and essentialityare capable of bidirectional influence. There-fore, in addition to our main analyses, we alsoexplored whether entitativity and essentialitywould have an interactive effect on judgmentsof collective responsibility. While such effectshave not been tested before, we found it plau-sible that essentiality could magnify the effectsof entitativity. If one believes that groupmembers are both interdependent and possessa shared essence, then judgments of collective

Group Processes & Intergroup Relations 9(1)

46

responsibility might be especially strong. Insuch cases, group members are perceived assimultaneously being ‘in cahoots’ and ‘cut fromthe same cloth’.

Study 1

The first goal of Study 1 was to examinewhether perceivers believe that membership indifferent types of groups (i.e. intimacy groups,task groups, social categories, and loose associ-ations) generally entails different levels ofcollective responsibility. Based on past research(Lickel et al., 2000, 2003) we predicted thatgroups highest in perceived entitativity (i.e.intimacy groups) would be rated highest indegree of collective responsibility followed bytask groups, social categories, and loose associ-ations.

Our second goal was to more closely examinethe extent to which participants’ perceptions ofentitativity of the groups would predict collec-tive responsibility. In our past research on enti-tativity (Lickel et al., 2000), we found thatperceptions of entitativity were strongly relatedto perceptions of the degree of interdepen-dence among members of the group. Thus, asin our analysis of the Columbine shootings(Lickel et al., 2003), our operational definitionof entitativity revolved around the idea of per-ceived social interdependence among groupmembers. However, there are other group vari-ables, notably group size, and essentialist vari-ables (i.e. duration and permeability) thatmight also account for differences amonggroups in collective responsibility. We chose tooperationalize essentiality with these latter twovariables because previous writings andresearch have demonstrated that a key featureof essentialism is a notion of ‘inalterability’(Haslam et al., 2000; Lickel et al., 2000;Rothbart & Taylor, 1992). Thus, Study 1attempted to tease apart the contributions ofentitativity and essentiality to judgments ofcollective responsibility.

MethodParticipants Study participants were 211 Uni-versity of California, Santa Barbara students

who participated in the study for partial com-pletion of a class requirement. Gender was notassessed.

Materials and procedure Participants wereinstructed that they would be completing asurvey of their perceptions of different socialgroups. Participants were presented with thequestionnaire packet and asked to workthrough it at their own pace.

The questionnaire packet contained 15rating tasks, each page of which assessed partici-pants’ judgments of a different group property(e.g. group size, collective responsibility, etc.)to be rated on a 9-point scale with labeledanchors. For each judgment task, participantswere presented with a list of the same 30groups. The sample of 30 groups presented toparticipants was selected from past research onperceptions of group entitativity and the folktaxonomy of group types (Lickel et al., 2000).These group types are intimacy groups (e.g.romantic partners, close friends), task groups(e.g. jury, airline flight crew), social categories(e.g. Blacks, citizens of Poland), and looseassociations (e.g. people who like classicalmusic, people waiting at a bus stop). The 30groups were presented in a different order foreach rating task. Four different versions of thequestionnaire were developed that placed therating tasks in random orders in the packet.

Collective responsibility Participants’ perceptionsof the degree to which membership in eachgroup entailed collective responsibility read asfollows: ‘Rate how responsible a member of thegroup should feel if another member of thegroup committed a serious negative act’.Because it may be construed that asking partici-pants to judge how blameworthy a groupmember should feel if a group membercommits a wrong-doing only assesses one aspectof collective responsibility, we ran two follow upstudies (N = 50 in each study) in which we assedtwo different collective responsibility items incomparison to the one used in the studies in thepresent paper. Specifically, participants in thefirst follow-up sample rated ‘how blameworthy amember of the group should be considered’

Denson et al. collective responsibility

47

while a separate group of participants inanother follow up sample rated ‘how much thewrongdoer’s group should be blamed’ if agroup member committed a wrongdoing. Wethen computed means for each of the 30groups in these two new samples and correlatedthese means with the group means for the‘should feel responsible’ item from Study 1.These two new items correlated highly with our‘feel’ item (r = .94, p < .001, and r = .91, p <.001), for the first and second follow up samplesrespectively. Thus, it is clear that the collectiveresponsibility item used in Studies 1 and 2 ishighly correlated with other items that moreblatantly assess collective blame.

Entitativity Participants’ perceptions of thedegree of interdependence among members ofthe 30 groups were assessed with six items.These consisted of ratings of the degree of inter-action (‘For each group, we would like youropinion about the extent to which the peoplein the group interact with each other’), behav-ioral influence (‘the degree to which thebehavior of individuals in the group can be con-trolled or influenced by other people in thegroup’), norms (‘the degree to which the grouphas formal and informal rules’), interpersonalbonds (‘the degree to which you think there arestrong interpersonal bonds among the peoplein the group’), shared knowledge (‘the degree towhich the members of the group share knowl-edge and information’), and common goals (‘theextent to which the people in the group havecommon goals’) among members of each of the30 groups.

Essentiality Participants rated the duration(‘the extent to which each group is a long-termor short-term group’), and permeability (‘Groupsthat are easy to join and leave are very perme-able, whereas groups that are difficult to joinand leave are not very permeable. For eachgroup below, rate how permeable the group is’)of the 30 groups.

Participants also rated the homogeneity (‘thedegree to which the members of the group arelikely to possess the same personality traits andabilities’), size, and the extent to which each of

the target groups ‘qualified as a group’. The‘qualify as a group’ item was a general measureof entitativity used in prior research (Lickelet al., 2000).

Omission and commission Participants’ justifica-tions of omission were assessed as follows: ‘Ratethe degree to which members of the groupshould be expected to prevent others in thegroup from committing serious negative acts’.Justifications of commission were assessed withthe following statement: ‘Rate the degree towhich it would be suspected that a member ofthe group might have contributed in some wayto a serious negative act committed by anothermember of the group’.

ResultsTo examine our hypotheses, we conductedthree sets of analyses. First, we conductedrepeated measures analyses to examine theextent to which the four types of groups (i.e.intimacy groups, task groups, social categories,and loose associations) were perceived to differin collective responsibility, entitativity, andessentiality. Second, we examined the extent towhich entitativity predicted judgments ofcollective responsibility while teasing out theeffects of other group properties (size andgroup-member homogeneity) and essentialistvariables (duration and permeability). Third,because previous research has demonstratedthat entitativity may influence perceptions ofessentiality and vice versa (see Yzerbyt et al.,2001; Yzerbyt & Rocher, 2002), we examinedpossible interactive effects of these two dimen-sions on collective responsibility.

Repeated measures analyses As previously dis-cussed, we selected groups based on pastresearch (Lickel et al., 2000) distinguishingfour different types of groups (i.e. intimacygroups, task groups, social categories, and looseassociations). We present results based on thecluster assignments for these groups identifiedin past research (Lickel et al., 2000). However,we did verify that the same group clusters couldbe identified with the present data. The resultsof these clustering analyses on the present data

Group Processes & Intergroup Relations 9(1)

48

did replicate (with some minor exceptions) theresults of our past research. The patterns ofmeans and the significance of differences donot differ when using the clusters generatedfrom the present analysis.

In order to examine participants’ beliefsabout the different types of groups, we createdcomposite ratings for each variable for eachgroup. We then ran a repeated measures analysisof variance using ‘type of group’ as the indepen-dent variable to examine the extent to which thefour types of groups differed with respect tocollective responsibility, entitativity, and essen-tiality. We created an essentiality composite ofduration and permeability ratings (� = .62) andan entitativity composite (interaction, behav-ioral influence, interpersonal bonds, commongoals, shared knowledge, and norms; � = .89).

Figure 1 presents the means by group type.As predicted, there was a main effect for grouptype on entitativity (F(3,208) = 1026.80, p <

.001). Post hoc analyses revealed that intimacygroups were rated highest in entitativity,followed by task groups, social categories, andloose associations (all ps < .05). There was alsoa main effect of group type on collectiveresponsibility (F(3,208) = 325.95, p < .001).Consistent with our hypothesis that groupsrated higher in entitativity should be ratedhigher in collective responsibility, the patternof means paralleled the pattern for entitativity(ps < .05) except the difference between socialcategories and loose associations. Also asexpected, groups differed in essentiality (i.e.duration and permeability) (F(3,207) = 36.45,p < .001). Participants rated intimacy groupsand social categories higher in essentiality thantask groups and loose associations (ps < .05).Task groups, in turn, were rated higher inessentiality than loose associations. Intimacygroups and social categories did not differ fromeach other in essentiality.

Denson et al. collective responsibility

49

Figure 1. Ratings of entitativity, essentiality (composite of duration and permeability), and collectiveresponsibility as a function of group type, Study 1.

1

2

3

4

5

6

7

8

9

Intimacy Groups Task Groups Social

Categories

Loose

Associations

Group Type

gnit

aR

Entitativity

Essentiality

Coll. Resp.

Entitativity and essentiality predicting collectiveresponsibility We first examined the zero-order correlations among the key variables ofinterest. Because each participant rated all 30groups, it was necessary to account for depen-dence of observations when examining thesecorrelations. Zero-order correlations betweenall the rating tasks were calculated for each par-ticipant, which created 211 correlation matrices(one for each participant). The median corre-lations were then obtained. Entitativity andessentiality were moderately correlated (r = .50,p < .001). Entitativity and essentiality were bothcorrelated with collective responsibility (rs =.43, p < .001 and .20, p < .05, respectively).

Our second question concerned the extentto which entitativity or essentiality uniquely pre-dicted ratings of collective responsibility.Because each participant in the study ratedeach of the 30 groups on each of the variables,the study has a nested correlational design.Because of this nested design, we used hierar-chical linear modeling (HLM; Raudenbush &Bryk, 2001) as our analytic method. HLM is astatistical method that takes into account thedependent, hierarchical nature of data. In thecurrent study, group ratings (level-1 model) arenested within individuals (level-2 model). Forthese analyses, which utilized the HLM softwarepackage (Bryk, Congdon, Cheong, & Rauden-bush, 2000), models were specified with arandom effect for variables (e.g. they were freeto vary), when analysis of variance componentsrevealed significant variability. Taking this indi-vidual variation into account, mean slopes foreach variable of interest were computed acrossindividuals. All variables were participantcentered, entered as participant centered pre-dictors and Z-transformed, such that standard-ized coefficients are reported throughout thispaper. Nine participants were removed fromanalyses due to missing data on at least onevariable rating, leaving a final sample size of 202participants.

In this first set of HLM analyses, group size,homogeneity, essentiality, and entitativity wereentered as predictors of collective responsi-bility. Results of this analysis confirm that, con-trolling for other factors that might account for

collective responsibility ratings, perceptions ofentitativity accounted for a significant portionof the variance in participants’ ratings. Thus,size (� = –.21, p < .001), homogeneity (� = .10,p < .001), and essentiality (� = .06, p < .001)each had a smaller role than entitativity (� =.42, p < .001) when all variables were entered aspredictors of collective responsibility ratings.These results indicate that entitativity plays aunique role in predicting people’s judgmentsof collective responsibility. It should be notedthat this present analysis uses those group prop-erties that we propose lay people are directlyrelying upon to assess a group’s entitativity.However, the results are parallel when using thedistal ‘qualify as a group’ variable rather thanthe composite. When used in lieu of the com-posite measure of entitativity, this variable was asignificant predictor of collective responsibility(� = .29, p < .001) when controlling for size,essentiality, and group-member homogeneity.

Entitativity as a predictor within different types ofgroups The preceding analysis shows that,across different types of groups, entitativity is apredictor of collective responsibility. However, areader might question the extent to which this isdriven by high ratings of collective responsibilityfor intimacy and task groups, with no meaning-ful variation in collective responsibility for socialcategories or loose associations. One way toexamine this is to focus analyses within eachgroup type. Thus, for example, insofar as thereis variation in how much collective responsibilityis applied to different social categories, is thisvariation predicted by ratings of entitativity?

To examine variation in the strength ofrelationship between entitativity and collectiveresponsibility within each of the four grouptypes, we conducted separate HLM models foreach of the four group types. Entitativity signifi-cantly predicted variation in collective responsi-bility among the intimacy groups (� = .17, p <.01), task groups (� = .33, p < .001), socialcategories (� = .17, p < .01), and loose associ-ations (� = .22, p < .001), when controlling forhomogeneity, size, and essentiality (Figure 2).Thus, entitativity is a predictor not only ofbetween group-type differences (e.g. it explains

Group Processes & Intergroup Relations 9(1)

50

the differences between task groups and socialcategories) it is also a predictor of within group-type differences (e.g. differences among socialcategories).

The interactive effects of entitativity and essentialityFinally, to investigate the possible interactiveeffects of entitativity and essentiality (Yzerbytet al., 2001, Yzerbyt & Rocher, 2002), we con-ducted a set of analyses that included anEssence � Entitativity interaction term. Theinteraction term significantly predicted judg-ments of collective responsibility (� = .04, p =.001). Post hoc analyses of simple slopes (Aiken& West, 1991; Bauer & Curran, in press)revealed that the effect of entitativity on collec-tive responsibility was greatest at high levels ofessentiality (one SD above the mean) (� = .62,p < .01). This relationship remained significantbut weakened at mean levels of essentiality (� =.58, p < .001), and low levels of essentiality (one

SD below the mean) (� = .54, p < .001). Itshould be noted that due to the correlationalnature of our data, it is equally possible thatentitativity moderates the effects of essentialityon collective responsibility or vice versa.However, because prior analyses in Study 1revealed entitativity (but not essentiality) to bea significant predictor of collective responsi-bility, we focused on the moderating role ofessentiality on the effects of entitativity oncollective responsibility judgments.

In a secondary set of analyses, we examinedthe extent to which participants’ ratings of theappropriateness of applying justifications ofcommission and omission were related to enti-tativity, essentiality, and collective responsibility.Because lay perceivers use entitativity infor-mation to explain the functioning of groups, itwas hypothesized that entitativity, but not essen-tiality, would predict variability in inferences ofomission and commission. Entitativity was

Denson et al. collective responsibility

51

Figure 2. Relationship between entitativity and collective responsibility. Slopes represent partial coefficients,controlling for homogeneity, size, duration, and permeability for each group type, Study 1. The length ofeach line indicates the range of means of entitativity ratings for the groups in each group type.

Entitativity

Co

lle

ctive

Re

sp

on

sib

ility

9

8

7

6

5

4

3

2

1

1 2 3 4 5 6 7 98

Loose Associations

Social Categories

All Groups

Task Groups

Intimacy Groups

moderately correlated with omission (r = .64,p < .001), and commission (r = .51, p < .001).Essentiality was weakly correlated with omission(r = .24, p < .001), and commission (r = .25, p <.001). These results suggest that lay perceiversmay use entitativity information when makingjudgments of collective responsibility to agreater degree than essentialist beliefs.However, when entered simultaneously andcontrolling for size and homogeneity, entitativ-ity predicted omission (� = .44, p < .001), andcommission (� = .32, p < .001), but essentialityonly weakly predicted justifications of com-mission (� = .07, p < .01), and not omission.

DiscussionThe results of Study 1 supported all of ourhypotheses concerning judgments of collectiveresponsibility. Our first goal was to demonstratethat different types of groups (Lickel et al.,2000) differ in the extent to which perceiversbelieve that membership in the group entailscollective responsibility. In support of this,intimacy groups were highest in collectiveresponsibility, followed by task groups, and thenby social categories and loose associations. Wealso demonstrated the role of entitativity in pre-dicting judgments of collective responsibilityboth between and within each group type.Finally, we demonstrated that the effects of enti-tativity on collective blame increase withincreasing perceptions of essentiality.

Study 2

The primary purpose of Study 2 was to concep-tually replicate and extend the findings fromStudy 1. There were several limitations to Study1. First, we did not use equivalent numbers ofdifferent types of groups (i.e. people rateddifferent numbers of social categories and taskgroups). In Study 2, participants rated five ofeach of the four types of groups. Furthermore,we operationalized both essentiality and entita-tivity somewhat differently in order to examinethe conceptual generalizability of the effects wefound in Study 1. In Study 2, we used an estab-lished measure of entitativity and essentialistbeliefs (Haslam et al., 2000). Haslam et al.’s

(2000) scale was designed to assess perceptionsof essentiality in social categories. Their scalecontains two subscales. One, indexing theextent to which the social category is viewed asa ‘natural kind’, is most similar to how wedefined essentiality in our introduction (and tothe index of duration and impermeability inStudy 1). The second subscale is referred to as‘entitativity’ and consists of questions assessinginformativeness, inherence, uniformity, andexclusivity. We were interested in the extent towhich this entitativity subscale would be consist-ent with the items we used in Study 1. Thus,Study 2 uses a different sample of groups as wellas a different operationalization of essentialityand entitativity in order to provide a robustreplication of Study 1.

MethodParticipants Study participants were 80 Uni-versity of Southern California students (60females, 20 males) who volunteered for extrapsychology course credit. There were nogender differences on the entitativity, essential-ity, and collective responsibility measures bothwithin group type or for the four group typescombined.

Materials and procedure Participants wereinstructed that they would be completing asurvey of their perceptions of different socialgroups. As in Study 1, participants were pre-sented with the questionnaire packet with onerating task per page and asked to work throughit at their own pace. There were 14 rating tasks.For each judgment task, participants were pre-sented with a list of the 20 groups to be ratedon 9-point scales with labeled anchors. Thesample of 20 groups presented to participantswas selected from past research (Lickel et al.,2000), such that participants were presentedwith five of each group type (e.g. intimacygroups, social categories, etc.). Examples of thegroups included a local street gang (intimacygroup), cast of a play (task group), citizens ofPoland (social category), and people in a movieaudience (loose association). As in Study 1, thegroups were presented in a different order foreach rating scale.

Group Processes & Intergroup Relations 9(1)

52

Collective responsibility, omission, and commissionThese items were identical to those used inStudy 1.

Entitativity Six items assessed perceptions ofgroup entitativity. These were the four itemsfrom Haslam et al.’s (2000) scale that loaded ontheir entitativity factor and two items from ourprevious research (Lickel et al., 2003). Theitems from Haslam et al.’s (2000) scale assessedthe uniformity of group members (uniformity),inductive potential of group membership(informativeness), the degree to which groupmembers share an underlying reality (inher-ence), and the extent to which membership inthe group precludes membership in othergroups (exclusivity). Two items from Study 1(interaction and common goals) were also usedto assess perceptions of entitativity. These sixitems formed a reliable composite (� = .82).

Essentiality To assess perceptions of essential-ity, the five items from Haslam et al.’s (2000)essentialist beliefs scale that correspond to the‘natural kinds’ dimension were administered.These items assessed group boundaries (dis-creteness), the naturalness of the group (natu-ralness), degree to which a member couldbecome a non-member (immutability), histori-cal invariance (stability), and degree to whichthe group is defined by necessary features(necessity). Because the immutability and his-torical invariance items were fairly equivalent tothe wording of the permeability and durationitems used in Study 1, we did not include theseStudy 1 items in Study 2. The five items formeda fairly reliable scale (� = .67).

ResultsRepeated measures analyses Our first hypo-thesis was that the four group types (intimacygroups, task groups, social categories, and looseassociations) would differ in terms of perceivedessentiality, entitativity, and collective responsi-bility. As in Study 1, we conducted repeatedmeasures analyses of variance with group typeas the single repeated measures factor. The fourtypes of groups differed in collective responsi-bility (F(3,77) = 172.91, p < .001), entitativity

(F(3,77) = 147.12, p < .001), and essentiality(F(3,77) = 134.26, p < .001) (Figure 3). Post hocanalyses revealed that intimacy groups wererated highest in collective responsibilityfollowed by task groups, social categories, andloose associations (all ps < .05). This samepattern was observed for entitativity (all ps <.05). As predicted, social categories were ratedhigher (ps < .05) in essentiality than task groupsand loose associations, and were equivalent tointimacy groups. These patterns replicatedStudy 1.

Entitativity and essentiality predicting collectiveresponsibility We hypothesized that percep-tions of entitativity contain information for thelay perceiver about the actual functioning ofsocial relationships within the group, while per-ceptions of essentiality do not provide thisinformation. Thus, perceived entitativity shouldbe associated with increased judgments ofcollective responsibility, but essentiality shouldbe weakly or not at all associated with such judg-ments. As in Study 1, we used HLM to accountfor the nested nature of our data. One partici-pant was removed from analyses due to incom-plete data.

Median zero-order correlations between allthe rating tasks were calculated for each partic-ipant. Entitativity was moderately correlatedwith essentiality (r = .65, p < .001). More import-ant, the median correlation between entitativityand collective responsibility was larger (r = .58,p < .001), than the correlation between essen-tiality and collective responsibility (r = .38, p <.001, z = 1.63, p = .05), one-tailed (see Cohen &Cohen, 1983). To examine the direct effects ofentitativity and essentiality on judgments ofcollective responsibility, we analyzed a modelwith these two composites simultaneously pre-dicting collective responsibility. Entitativitystrongly predicted collective responsibilityjudgments (� = .58, p < .001), but essentialitydid not (� = .02, ns).

As in Study 1, we also conducted a secondaryset of analyses investigating justifications ofomission and commission. Entitativity wasmoderately correlated with omission (r = .60,p < .001), and commission (r = .55, p < .001).

Denson et al. collective responsibility

53

Essentiality was weakly correlated withomission (r = .33, p < .01), and commission (r = .31, p < .01). When entered simultaneously,entitativity predicted omission (� = .60, p <.001), and commission (� = .63, p < .001), butessentiality did not predict either justification.

Entitativity as a predictor within different types ofgroups We also examined the relationshipbetween entitativity and collective responsibilitywithin each group type. Entitativity predictedvariation in collective responsibility withinintimacy groups (� = .18, p < .001), task groups(� = .18, p < .001), social categories (� = .16,p < .001), and loose associations (� = .37, p <.001) when essentiality was entered simul-taneously in the model. Figure 4 shows theseresults, with the length of each line represent-ing the range of entitativity ratings of thegroups within the group type.

The interactive effects of entitativity and essentialityAs in Study 1, we investigated the possible inter-active effects of entitativity and essentiality. TheEntitativity � Essentiality interaction term sig-nificantly predicted judgments of collectiveresponsibility (� = .07, p < .005). As in Study 1,post hoc analyses of simple slopes (Aiken &West, 1991; Bauer & Curran, in press) revealedthat the effect of entitativity on collectiveresponsibility was greatest at high levels ofessentiality (one SD above the mean) (� = .66,p < .001). This relationship remained signifi-cant but was slightly weakened at mean levels ofessentiality (� = .59, p < .001), and low levels ofessentiality (one SD below the mean) (� = .52,p < .001).

DiscussionUsing a broad, established measure of essential-ist beliefs and entitativity (Haslam et al., 2000)

Group Processes & Intergroup Relations 9(1)

54

Figure 3. Ratings of entitativity, essentiality, and collective responsibility as a function of group type, Study 2.

1

2

3

4

5

6

7

8

9

Intimacy

Groups

Task Groups Social

Categories

Loose

Associations

Group Type

gnit

aR

Entitativity

Essentiality

Coll. Resp.

and a varied analytic approach, Study 2 concep-tually replicated the major findings of Study 1.We replicated the finding that the four types ofgroups differ in perceptions of collectiveresponsibility, essentiality, and entitativity. Fur-thermore, we again found that entitativity was astronger predictor of collective responsibilitythan essentiality. Even when examining vari-ation in collective responsibility within socialcategories that are defined more by essencethan entitativity, entitativity was a significantpredictor of collective responsibility. As inStudy 1, a significant interaction revealed thatthe effect of entitativity on collective responsi-bility was stronger with increasing levels ofessentiality. Overall, these data support our the-orizing that lay perceivers use primarily folktheories of entitativity as an interpretativemechanism for understanding social influenceon moral behavior in groups, but that essential-ity may play a small moderating role in influ-encing these judgments.

General discussion

When bad events occur, humans have an insa-tiable appetite for understanding and explain-ing those events. This search for understandingoccurs whether one is ascribing responsibilityfor the event to an individual or a group. Whenmaking sense of an individual, the perceiver’sfocus is on those invisible but vital mental statesthat help explain the actor’s deeds. Did heintend to do this? Was it an accident? Is heunrepentant, or does he regret his bad acts?These mental states cannot be seen directly, butpeople’s inferences about them are crucial indetermining judgments of individual responsi-bility. For collective responsibility, we argue thatinferences of another invisible psychologicalconcept are important, namely the ties thatbind people together in social relationships andgroups. We argued that entitativity is, at its core,the lay perceiver’s distilled understanding ofthe social relationships and interdependence

Denson et al. collective responsibility

55

Figure 4. Relationship between entitativity and collective responsibility. Slopes represent the partialcoefficient controlling for essentiality for each group type, Study 2. The length of each line indicates therange of means of entitativity ratings for the groups in each group type.

Entitativity

Co

lle

ctive

Re

sp

on

sib

ility

9

8

7

6

5

4

3

2

1

1 2 3 4 5 6 7 98

Loose Associations

Social Categories

All Groups Task Groups

Intimacy Groups

among members of a group. We contrasted thiswith essentiality, which we argued was particu-larly linked to seeing a group as having bothpermanence and inalterability. In our efforts tounderstand these lay theories of groups, we firstestablished that group types differ in collectiveresponsibility, entitativity, and essentiality. Ourmore specific analyses identified that lay per-ceivers use primarily entitativity informationwhen making judgments of collective responsi-bility. Finally, we found a moderating effect ofessentiality such that increasing perceptions ofessentiality slightly increased the effect of enti-tativity on collective blame judgments.

We acknowledge limitations to the currentset of studies. First, our data are correlational innature. Although we did replicate the patternof results, we cannot definitively state thatentitativity perceptions increase perceptionsof collective responsibility. In addition, theexplanatory power of our essentiality com-posites may have been limited do to relativelylow reliability coefficients. Future researchshould focus on laboratory manipulations ofentitativity and essentiality. Second, we investi-gated third party perceptions of groups ratherthan groups to which members have anemotional investment. We would caution aboutfirmly extrapolating the present results to allintergroup settings.

Below, we discuss some implications andfuture directions for research related to laytheories and also theories of intergrouprelations and prejudice.

Implications for development and content oftheories of groupsIn this paper, we argued that perceivers relyupon intuitive theories of groups in order tomake judgments of collective responsibility. Todate, there is some evidence that psychologicalessentialism as a processing heuristic for socialcategories emerges early in life. For example,Hirschfeld and Gelman (1997) found thatpreschoolers believed that people of differentraces would speak different languages and thateven if swapped at birth, the children wouldgrow up to speak the language of their birthparents. There is also evidence indicating that

children use a theory of essentiality whenreasoning about racial categories (Hirschfeld,1995). Similar essentialist theories in childrenhold true for gender as well (Taylor, 1996) andone study found that 12-year-old Indianchildren believed that caste was fixed at birth(Mahalingam, 1999). It remains to be seenwhether folk theories of essentiality for othergroup types emerge early in life as well.

One intriguing finding in our data is thatintimacy groups and social categories are bothperceived as high in essentiality. To our knowl-edge no prior research has conclusively shownthis. At first glance, ascribing equal levels ofessence to both types of groups may appearunusual. However, this simply suggests that bothintimacy groups and social categories are viewedas inalterable, impermeable, and historicallyinvariant. When placed in this context, one mayeasily imagine how lay perceivers view a family aspossessing a deep essence that transcends timeand social circumstance in much the samemanner as gender or racial categories. However,it is less clear how and why people would ascribeessentiality to other intimacy groups such asfriendship groups. There clearly remains muchto be done to understand how essentialist beliefsmay be applied to groups other than socialcategories such as race or gender.

There is an even greater gap in our knowl-edge about the development of intuitivetheories about entitativity. While research hasdemonstrated that the four types of groups pre-sented in the current article are used spon-taneously by adults to encode socialinformation (e.g. Sherman, Castelli, &Hamilton, 2002), there are no data concerningwhen or under which circumstances childrenmay develop their theory of group entitativity.Likewise, while there is a growing body ofresearch about the effects of entitativity onsocial judgments, little is known about howchildren’s beliefs about groups and socialrelationships develop. We would hypothesizethat children reach an understanding of entita-tivity by first developing an understanding ofthe relational principles used in dyadic relation-ships in which the child is involved. As we’vediscussed elsewhere (Lickel et al., in press),

Group Processes & Intergroup Relations 9(1)

56

people’s beliefs about the relational principlesused in a group are tightly connected to theperceived entitativity of the group. Fiske (1991)has argued that there is a developmental orderto children’s understanding of relational prin-ciples (beginning with an understanding ofcommunal sharing and only later developingan understanding of authority ranking, equalitymatching, and much later, market pricing).However, little structured empirical research onchildren’s understanding of the operation ofthese relational styles has been conducted.Developmental research on relational prin-ciples and entitativity would seem to be particu-larly important for future research.

The last issue we would raise in this section isthe connection of our present research withother work that has taken a perceiver-focusedapproach to the study of group perception(Hong, Levy, & Chiu, 2001; Plaks, Levy, Dweck,Stroessner, 2004). One particularly relevant laytheory approach has been conducted by Dweckand colleagues (e.g. Dweck, Chiu, & Hong,1995). These researchers have identified indi-vidual differences in the extent to which peoplebelieve that traits are fixed versus malleable.Entity theorists focus on traits and endorsestereotypes to a greater extent than incremen-tal theorists, while incremental theorists focusmore on situational and mediating psychologi-cal mechanisms (and stereotype less) than theirentity counterparts (see Plaks et al., 2004).Plaks et al. (2004) propose that entity theoristsmay be especially likely to perceive essentialismin groups, while incremental theorists may beespecially likely to perceive entitativity ingroups. If so, this would have interesting ramifi-cations for collective responsibility research.Entity theorists should assign less blame to thegroup (but endorse dispositional judgments toa greater degree) while incremental theoristsshould assign more blame (but not endorse dis-positional judgments). This seemingly counter-intuitive prediction deserves further study.

Implications for intergroup relations andprejudiceWhat are the broader implications of thesefindings for intergroup relations? First, we

would suggest that collective responsibility (andtherefore entitativity) plays a strong role infueling many intergroup conflicts. In inter-group conflicts, retaliation for the actions of anindividual is often spread or displaced to otheroutgroup members beyond the provocateur.These other outgroup members are targetedfor collective blame and therefore retaliationbecause of their shared group membershipwith the provocateur. Although categorizationof the situation in intergroup terms (‘them’and ‘us’) is a crucial first step in the process bywhich collective blame and retaliation occur inintergroup contexts, we would argue that per-ceptions of the entitativity of the outgroup alsoplay a role in moderating the extent to whichretaliation is directed at other outgroupmembers beyond the provocateur. Further-more, the argument that members of theoutgroup are ‘in cahoots’ is a powerful justifi-cation for collective blame and retaliationagainst individuals who have a tenuous directconnection to the event that precipitated theintergroup conflict. The observed interactionbetween essentiality and entitativity on judg-ments of collective responsibility also indicatesthat increasing perceptions of groups as imper-meable and stable over time magnifies theeffect of entitativity on collective responsibilityjudgments. Therefore, groups who are per-ceived as both ‘in cahoots’ and ‘cut from thesame cloth’ may be especially vulnerable togroup-based retaliation. Consider the followingobservation from a Western journalist in theongoing conflict in Iraq:

When an Iraqi man loses a family member to anAmerican missile, he must take another Americanlife to even the score. He may not subscribe to thenotion that some Americans are noncombatants,viewing them instead as the members of a super-tribe that has come to invade his land. (Robertson,Salon, September 23, 2004)

According to this correspondent’s interpret-ation, Iraqis perceive the Americans to be ahighly interdependent group (i.e. ‘supertribe’)in which members are interchangeable for thepurpose of retaliation. This may be furtherexaggerated by perceptions that Americansshare an inalterable, underlying essence.

Denson et al. collective responsibility

57

Interestingly, though, some of the samefactors that may lead to applying collectiveresponsibility to an outgroup (such as Iraqisblaming Americans) may also lead, under somecircumstances, to people applying collectiveblame to their ingroups. An emerging literatureon group-based emotions has found thatpeople may feel vicarious shame, guilt oringroup-directed anger when they believe thatmembers of the ingroup have behaved in ablameworthy fashion (e.g. Doosje, Branscombe,Spears, & Manstead, 1998; Lickel, Schmader,Curtis, Scarnier, & Ames, 2005). In turn, theseemotional reactions elicit behaviors to copewith the blameworthy action of the ingroupmember. In particular, guilt motivates apologyand reparations, shame motivates distancingfrom the wrongdoer, and ingroup directedanger motivates confrontation and punishmentof the ingroup wrongdoer (e.g. Doojse et al.,1998; Iyer, Schmader, & Lickel, 2005; Lickelet al., 2005). At least in part, these reactions ofingroup blame rely upon perceptions of entita-tivity. In particular, vicarious guilt has beenlinked to the interdependence/entitativity ofthe group to which the wrongdoer and the per-ceiver belong (Lickel et al., 2005). It is alsopossible that vicarious shame is heightened ingroups that are high in essentiality. We clearlybelieve that there is a distinction in people’sthinking between the more relational aspects oftheir group membership versus the identityaspects of the group membership (Brewer &Gardner, 1996). Thus, group-based shame ismore related to the identity aspects of groupmemberships, whereas group-based guilt is par-ticularly linked to the relational aspects ofgroup membership (Lickel et al., 2005).Clearly, there are many connections yet to bemade with regard to different facets of collec-tive responsibility (whether directed towardingroups or outgroups) and folk theories ofgroups.

Finally, we also believe that entitativity mayplay a role in some forms of prejudice, particu-larly in which there is suspicion of coordinatedaction by the group. Thus, for example, anti-Semitism historically involves an ugly preoccu-pation of Jewish conspiracy to undermine or

control government, economy, or culturewithin predominantly Christian societies(Ruotsila, 2000). Such conspiracy theories restupon something different than just ascribingnegative traits to Jews. It is not only that anti-Semites view Jews as a homogeneous group of‘bad people’ who might be dislikable oruntrustworthy. Instead, beyond that, there is atheory of coordinated action amongst Jews, aninterdependence of Jewish goals and actions,which threaten the (non-Jewish) perceiver.Such a theory cannot rest very easily upon per-ceptions of homogeneity of traits or even beliefin a group essence. Instead, conspiracy theoriessuch as this rest upon perceptions of entitativ-ity, the belief that there is interdependence ofgoals, actions, and relationships amongmembers of the group.

In conclusion, groups differ in countlessways, and yet a few key dimensions of socialassociation may explain much of the importantdifferences in how lay people perceive groups.In the present paper, we argued that two keyfolk theories—entitativity and essentiality—differ across groups and also differ in theireffects on collective responsibility judgments.We hope our work can be used as a steppingstone to further examine how such perceptionsof group entitativity, essentiality, and collectiveblame influence social cognition and inter-group relations.

AcknowledgmentsThe current research was supported in part by aJohn Randolph Haynes and Dora HaynesFoundation fellowship to Thomas F. Denson andNational Science Foundation Grant BCS-0112473 tothe second author. Study 1 was part of Brian Lickel’sdissertation conducted at the University ofCalifornia, Santa Barbara. We gratefully acknowledgeDavid L. Hamilton, Jim Blascovich, Nancy Collins,Leda Cosmides, and Diane Mackie who served asBrian Lickel’s dissertation committee.

ReferencesAiken, L. S., & West, S. G. (1991). Multiple regression:

Testing and interpreting interactions. Thousand Oaks,CA: Sage.

Group Processes & Intergroup Relations 9(1)

58

Allport, G. W. (1954). The nature of prejudice. Oxford,UK: Addison-Wesley.

Asch, S. E. (1955). Opinions and social pressure.Scientific American, 193, 31–35.

Atran, S. (1994). Core domains versus scientifictheories: Evidence from systematics and Itza-Mayafolkbiology. In L. A. Hirschfeld & S. A. Gelman(Eds.), Mapping the mind: Domain specificity incognition and culture (pp. 316–340). New York:Cambridge University Press.

Bauer, D. J., & Curran, P. J. (in press). Probinginteractions in fixed and multilevel regression:Inferential and graphical techniques. MultivariateBehavioral Research.

Brewer, M. B., & Gardner, W. (1996). Who is this‘we’? Levels of collective identity and self-representations. Journal of Personality and SocialPsychology, 71, 83–93.

Bryk, A. S., Congdon, R., Cheong, Y. F., &Raudenbush, S. W. (2000). HLM 5: Hierarchicallinear and nonlinear modeling. Lincolnwood, IL:Scientific Software International.

Campbell, D. T. (1958). Common fate, similarity,and other indices of the status of aggregates ofpersons as social entities. Behavioral Science, 3,14–25.

Castano, E., & Yzerbyt, V. Y. (1998). The highs andlows of group homogeneity. Behavioural Processes,42, 219–238.

Chiu, C., & Hong, Y. (1992). The effects ofintentionality and validation on individual andcollective responsibility attribution among HongKong Chinese. Journal of Psychology, 126, 291–300.

Chiu, C., Dweck, C. S., Tong, J. Y., & Fu, J. H. (1997).Implicit theories and conceptions of morality.Journal of Personality and Social Psychology, 73,923–940.

Cialdini, R. B. (2000). Influence: Science and practice(4th ed.) Boston: Allyn & Bacon.

Cohen, J., & Cohen, P. (1983). Applied multipleregression/correlational analysis for the behavioralsciences (2nd ed.) New York: Erlbaum.

Doojse, B., Branscombe, N. R., Spears, R., &Manstead, A. S. R. (1998). Guilty by association:When one’s group has a negative history. Journal ofPersonality and Social Psychology, 75, 872–886.

Dweck, C. S., Chiu, C., & Hong, Y. (1995). Implicittheories and their roles in judgments andreactions: A world from two perspectives.Psychological Inquiry, 6, 267–285.

Fiske, A. P. (1991). Structures of social life. New York:Free Press.

Gaertner, L., & Schopler, J. (1998). Perceivedingroup entitativity and intergroup bias: An

interconnection of self and others. EuropeanJournal of Social Psychology, 28, 963–980.

Hamilton, V. L. (1978). Who is responsible? Towarda social psychology of responsibility attribution.Social Psychology, 41, 316–328.

Hamilton, D. L., Sherman, S. J., & Rodgers, J. S.(2004). Perceiving the groupness of groups:Entitativity, homogeneity, essentialism, andstereotypes. In V. Yzerbyt, C. M. Judd, &O. Corneille (Eds.), The psychology of groupperception: Perceived variability, entitativity, andessentialism. Philadelphia, PA: Psychology Press.

Haslam, N., Rothschild, L., & Ernst, D. (2000).Essentialist beliefs about social categories. BritishJournal of Social Psychology, 39, 113–127.

Haslam, N., Rothschild, L., & Ernst, D. (2002). Areessentialist beliefs associated with prejudice? BritishJournal of Social Psychology, 41, 87–100.

Hirschfeld, L. A. (1995). Do children have a theoryof race? Cognition, 54, 209–252.

Hirschfeld, L. A., & Gelman, S. A. (1997). Whatyoung children think about the relation betweenlanguage variation and social difference. CognitiveDevelopment, 12, 231–238.

Hong, Y., Levy, S. R., & Chiu, C. (2001). Thecontribution of the lay theories approach to thestudy of groups. Personality and Social PsychologyReview, 5, 98–106.

Hovland, C. I., Janis, I. L., & Kelley, H. H. (1953).Communication and persuasion: Psychological studies ofopinion change. New Haven, CT: Yale University Press.

Iyer, A., Schmader, T., & Lickel, B. (2005). PredictingAmerican and British opposition to the occupation ofIraq: The role of group-based anger, shame, and guilt.Unpublished manuscript.

Janis, I. L. (1971, November). Groupthink. PsychologyToday, pp. 43–46.

Knowles, E. S. (1976). Group size and the extensionof social space boundaries. Journal of Personalityand Social Psychology, 33, 647–654.

Lickel, B., Hamilton, D. L., & Sherman, S. J. (2001).Elements of a lay theory of groups: Types ofgroups, relational styles, and the perception ofgroup entitativity. Personality and Social PsychologyReview, 5, 129–140.

Lickel, B., Hamilton, D. L., Wieczorkowska, G., LewisA., Sherman, S. J., & Uhles, N. (2000). Varieties ofgroups and the perception of group entitativity.Journal of Personality and Social Psychology, 78,223–246.

Lickel, B., Rutchick, A., Hamilton, D. L., & Sherman,S. J. (in press). Intuitive theories of group typesand relational principles. Journal of ExperimentalSocial Psychology.

Denson et al. collective responsibility

59

Lickel, B., Schmader, T., Curtis, M, Scarnier, M., &Ames, D. R. (2005) Vicarious shame and guilt.Group Processes & Intergroup Relations, 8, 145–157.

Lickel, B., Schmader, T., and Hamilton, D. L. (2003).A case of collective responsibility: Who else was toblame for the Columbine High School shootings?Personality and Social Psychology Bulletin, 29,194–204.

Mahalingam, R. (1999). Essentialism, power, andrepresentation of caste: A developmental study.Dissertation Abstracts International, 60, 856.

Malt, B. C. (1995). Category coherence in cross-cultural perspective. Cognitive Psychology, 29,85–148.

McConnell, A. R., Sherman, S. J., & Hamilton, D. L.(1997). Target entitativity: Implications forinformation processing about individual andgroup targets. Journal of Personality and SocialPsychology, 72, 750–762.

McGarty, C., Haslam, S. A., Hutchinson, K. J., &Grace, D. M. (1995). Determinants of perceived consistency: The relationship betweengroup entitativity and the meaningfulness ofcategories. British Journal of Social Psychology, 34,237–256.

Medin, D. L. (1989). Concepts and conceptualstructure. American Psychologist, 44, 1469–1481.

Milgram, S. (1974). Obedience to authority. New York:Harper and Row.

Moreland, R. L., & McMinn, J. G. (2004). Entitativityand social integration: Managing beliefs about thereality of groups. In V. Yzerbyt, C. M. Judd, & O.Corneille (Eds.), The psychology of group perception:Perceived variability, entitativity, and essentialism.Philadelphia, PA: Psychology Press.

Moscovici, S., & Zavalloni, M. (1969). The group as apolarizer of attitudes. Journal of Personality andSocial Psychology, 12, 125–135.

Plaks, J. E., Levy, S. R., Dweck, C. S., & Stroessner,S. J. (2004). In the eye of the beholder: Laytheories and the perception of group entitativity,variability, and essence. In V. Yzerbyt, C. M. Judd,& O. Corneille (Eds.), The psychology of groupperception: Perceived variability, entitativity, andessentialism. Philadelphia, PA: Psychology Press.

Raudenbush, S. W., & Bryk, A. S. (2001). Hierarchicallinear models: Applications and data analysis methods.Thousand Oaks, CA: Sage.

Robertson, P. (2004). Hell: Salon’s warcorrespondent on the Iraq inferno. SalonMagazine, September 23.

Rothbart, M., & Taylor, M. (1992). Category labels

and social reality: Do we view social categories asnatural kinds? In K. Fiedler & G. R. Semin (Eds.),Language, interaction, and social cognition(pp. 11–36), Newbury Park, CA: Sage.

Rothschild, L., & Haslam, N. (2003). Thirsty for H-sub-20? Multiple essences and psychologicalessentialism. New Ideas in Psychology, 21, 31–41.

Ruotsila, M. (2000). Lord Sydenham of Combe’sworld Jewish conspiracy. Patterns of Prejudice, 34,47–64.

Sanders, J., Hamilton, V. L., Denisovsky, G., Kato, N.,Kawai, M., Kozyreva, P., et al. (1996). Distributingresponsibility for wrongdoing inside corporatehierarchies: Public judgments in three societies.Law and Social Inquiry, 21, 815–855.

Sherif, M. (1966). The psychology of social norms.Oxford, UK: Harper Torchbooks.

Sherman, S. J., Castelli, L, & Hamilton, D. L. (2002).The spontaneous use of a group typology as anorganizing principle in memory. Journal ofPersonality and Social Psychology, 82, 328–342.

Shultz, T. R., Jaggi, C., & Schleifer, M. (1987).Assigning vicarious responsibility. European Journalof Social Psychology, 17, 377–380.

Taylor, M. (1996). The development of children’sbeliefs about social and biological aspects ofgender differences. Child Development, 67,1555–1571.

Welbourne, J. L. (1999). The impact of perceivedentitativity on inconsistency resolution for groupsand individuals. Journal of Experimental SocialPsychology, 35, 481–508.

Yzerbyt, V., Corneille, O., & Estrada, C. (2001). Theinterplay of subjective essentialism and entitativityin the formation of stereotypes. Personality andSocial Psychology Review, 5, 141–155.

Yzerbyt, V., & Rocher, S. (2002). Subjectiveessentialism and the emergence of stereotypes. InC. McGarty & V. Y. Yzerbyt (Eds.), Stereotypes asexplanations: The formation of meaningful beliefs aboutsocial groups (pp. 38–66), New York: CambridgeUniversity Press.

Yzerbyt, V., Rocher, S., & Schadron, G. (1997).Stereotypes as explanations: A subjectiveessentialistic view of group perception. InR. Spears, & P. J. Oakes (Eds.), The social psychologyof stereotyping and group life (pp. 20–50). Malden,MA: Blackwell.

Paper received 12 October 2004; revised version accepted27 June 2005.

Group Processes & Intergroup Relations 9(1)

60

Biographical notesTHOMAS F. DENSON is pursuing a doctorate in social

psychology at the University of SouthernCalifornia. His research interests include laytheories of groups, intergroup relations, andaggressive behavior.

BRIAN LICKEL is an assistant professor in thedepartment of psychology at the University ofSouthern California. His work focuses on emotionand social cognition related to groups, inparticular to folk theories of groups, socialemotions such as shame and guilt, and the role ofgroup-based responsibility in intergroup behavior.

MATHEW CURTIS is pursuing a doctorate in socialpsychology at the University of Southern

California. His current research examinesvicarious shame and guilt in close relationships,linguistic intergroup bias, and legal decision-making.

DOUGLAS M. STENSTROM is pursuing a doctorate insocial psychology at the University of SouthernCalifornia. His research interests involveintergroup relations, aggression, and law andpsychology.

DANIEL AMES is an assistant professor in themanagement division at Columbia BusinessSchool. His research focuses on social judgment,including impressions of individuals and groups,and social behavior, including cooperation andconflict.

Denson et al. collective responsibility

61