Explaining the Endowment Effect through Ownership: The ... · Explaining the Endowment Effect through Ownership: The Role of Identity, Gender, and Self-Threat SARA LOUGHRAN DOMMER

Explaining the Endowment Effect throughOwnership: The Role of Identity, Gender,and Self-Threat

SARA LOUGHRAN DOMMERVANITHA SWAMINATHAN

The price people are willing to pay for a good is often less than the price they arewilling to accept to give up the same good, a phenomenon called the endowmenteffect. Loss aversion has typically accounted for the endowment effect, but analternative explanation suggests that ownership creates an association betweenthe item and the self, and this possession-self link increases the value of the good.To test the ownership account, this research examines three moderators that theorysuggests should affect the possession-self link and consequently the endowmenteffect: self-threat, identity associations of a good, and gender. After a social self-threat, the endowment effect is strengthened for in-group goods among both menand women but is eliminated for out-group goods among men (but not women).These results are consistent with a possession-self link explanation and thereforesuggest that ownership offers a better explanation for the endowment effect.

The price people are willing to pay for a particular goodis often significantly less than the price they are willing

to accept to give up the same good (Kahneman, Knetsch,and Thaler 1990, 1991). Many studies have shown that thevaluations of a good from those randomly assigned to re-ceive it (i.e., sellers "endowed" with the good) exceed thevaluations of the good from those randomly assigned notto receive it (i.e., buyers; Kahneman et al. 1990, 1991;Knetsch 1989; Knetsch and Sinden 1984; Thaler 1980; VanBoven, Dunning, and Loewenstein 2000). Thaler (1980)called this increase in valuation of a good when the goodbecomes part of one's endowment the "endowment effect."

Sara Loughran Dommer ([email protected]) is assistantprofessor of marketing at the College of Management, Georgia Instituteof Technology, Atlanta, GA 30308. Vanitha Swaminathan ([email protected]) is associate professor of marketing at the Joseph M. Katz GraduateSchool of Business, University of Pittsburgh, Pittsburgh, PA 15260. Cor-respondence: Sara Loughran Dommer. The authors thank Jeff Inman, Ni-cole Verrochi Coleman, and participants of the Behavioral Decision Re-search Lab at Carnegie Mellon University for comments on an earlierversion of this article. This research is part of the first author's dissertationand was partially funded by a Dean's Small Research Grant from theUniversity of Pittsburgh.

Ann McGill served as editor and Aimee Drolet Rossi served as associateeditor for this article.

Electronically published June 18, 2012

The key difference between buyers and sellers is the pos-session of a good; sellers have a good in their possession,while buyers do not. A question that has yet to be answeredfully, however, is when and why possession of a good con-tributes to disparities in valuations between sellers and buy-ers.

Traditionally, possession has been thought to matter be-cause it causes selling to feel like a loss (Kahneman andTversky 1979; Thaler 1980). This "loss aversion account"suggests that a good is evaluated as a loss when it is givenup and as a gain when it is acquired, and because lossesloom larger than gains, sellers expect to suffer more thanbuyers expect to benefit. As a result, sellers demand morecompensation than buyers are willing to provide, resultingin higher selling prices than buying prices for an identicalgood. Theorists have argued that possession of a good doesnot affect its attractiveness or appeal but rather simply thepain associated with giving it up (Kahneman et al. 1991).Their argument is simply that sellers are averse to losing apossession and thus increase their valuations to avoid doingso (Kahneman et al. 1991).

More recently, however, researchers have begun to chal-lenge this traditional view. The "ownership account" arguesthat ownership indeed affects the appeal of a good. Own-ership involves a sense of possession toward an object andresults in objects being perceived as "mine" (Belk 1988;Pierce, Kostova, and Dirks 2003). Morewedge et al. (2009)find support for the ownership account for the endowment

1034

© 2012 by JOURNAL OF CONSUMER RESEARCH. Inc. • Vol. 39 • February 2013

All rights reserved. 0093-530l/2013/3905-0009$10.00. DOI: 10.1086/666737

OWNERSHIP AND THE ENDOWMENT EFFECT 1035

effect in that buyers and sellers have similar valuations ofa good when buyers already own an identical good. Theyargue that this result occurs "in part because people aremotivated to justify their choices, and in part because own-ing an item creates an association between the item and theself" (948). Because of the intrinsic tendency to enhanceone's self, this "possession-self link" resulting from own-ership subsequently boosts the possession's perceived value(Aggarwal 2004; Aggarwal and Law 2005; Belk 1988; Ga-wronski, Bodenhausen, and Becker 2007; Kleine, Kleine,and Kernan 1993). In other words, people value objects theyown more than objects they do not, and therefore ownershipaffects the appeal of a good. Consistent with this self-en-hancement view, Maddux et al. (2010) demonstrate that theendowment effect is stronger in Western cultures, in whichself-enhancement tendencies are greater, than in Eastern cul-tures, in which self-enhancement goals are less salient.

The current research posits that if ownership accounts forvaluation gaps between sellers and buyers, varying the pos-session-self link will have an impact on the magnitude ofthe endowment effect. Research on self-threat suggests thatpeople increase the value of their possessions as a way toindirectly enhance their self (Beggan 1992; Pettit and Si-vanathan 2011; Sivanathan and Pettit 2010). Identity theoryalso implies that linking a good to an in-group or an out-group changes the possession-self link (Escalas and Bettman2005; Reed 2004). The possession-self link should be stron-ger (weaker) for goods associated with one's in-group (out-group). Previous research, however, suggests that there maybe attentional differences among men and women to suchidentity associations. Consequently, we argue that the effectof identity associations on the possession-self link, and sub-sequently the endowment effect, vary by gender. Therefore,we investigate whether social self-threat, the social identityassociations of a good, and gender moderate the endowmenteffect.

In summary, the contributions of this research are three-fold. First, these findings add to a growing stream of researchon the role of ownership in explaining the endowment effect(;Morewedge et al. 2009). By looking at social self-threat,identity associations, and gender as theoretically meaningfulmoderators, we provide evidence for the ownership accountfor the endowment effect. In doing so, we call into questionthe loss-aversion explanation that has traditionally domi-nated understanding of the discrepancies in buyer and sellervaluations. Second, we show that gender moderates the en-dowment effect for a good with out-group identity associ-ations. We build on past research that suggests that menhave a more agentic orientation, whereas women have amore communal orientation (Bakan 1966; Carison 1971).Due to this difference in agency-communion orientation, wehypothesize that women are less likely than men to attendto intergroup comparisons, and therefore their possession-self link may be less affected by out-group associations.Thus, we find that women sellers demand higher sellingprices than men for an out-group good. We show that whenintergroup comparisons are made salient, however, these

gender differences are eliminated. A third key contributionof this research is shedding light on how possession-selfassociations can help individuals cope with a threat to one'ssocial self and the downstream impact of this on valuationsof goods. It appears that men and women cope with socialself-threat by increasing their valuations of goods that areclosely linked to their identities. In addition, some individ-uals (i.e., males) may cope with social self-threat by de-valuing goods that are in opposition to their identities. Takentogether, the findings from this research examine how iden-tity associations can influence economic value associatedwith goods and how the motivations of individuals can in-fluence the extent of this valuation. The next section outlinesthe theory and develops the hypotheses for subsequent test-ing.

THEORY AND HYPOTHESES

The ownership account for the endowment effect suggeststhat owning a good creates an association between the itemand the self and that this possession-self link increases thevalue of the good (Maddux et al. 2010; Morewedge et al.2009). As noted previously, variables that theoreticallyshould increase (decrease) the possession-self link should,according to the ownership account, also increase (decrease)the price sellers are willing to accept for a good, thus mod-erating the endowment effect. Next, we outline argumentsfor why a social self-threat provides an impetus for drivingup sellers' valuations of goods. After that, we develop ar-guments for why social identity and gender moderate theendowment effect.

Social Self-Threat

Extensive literature documents that people look for waysto enhance their self after a threat to their self-concept (Argo,White, and Dahl 2006; Campbell and Sedikides 1999; Col-lange, Fiske, and Sanitioso 2009; Frey and Stahlberg 1986;Gao, Wheeler, and Shiv 2009; Jordan and Monin 2008; Parkand Maner 2009; Sivanathan and Pettit 2010). Given theintimate connection between the self and possessions (Belk1988), it follows to reason that people will attempt to in-crease the value of their possessions as a way to indirectlyenhance their self (Beggan 1992; Pettit and Sivanathan2011; Sivanathan and Pettit 2010). For example, Beggan(1991) demonstrates that people attribute characteristics totheir possessions that they feel they themselves lack. Thisargument is consistent with prior research showing thatproducts can be used to restore self-views (Gao et al. 2009;Pettit and Sivanathan 2011; Sivanathan and Pettit 2010) andto cope with existential self-threats (Kasser and Sheldon2000; Rindfleisch, Burroughs, and Wong 2009). Thus, aftera self-threat, when self-enhancement tendencies are greater,people can use possessions to affirm their self, and endow-ment effects are likely to be exaggerated (Beggan 1992;Maddux et al. 2010).

We argue that a threat to a person's social self is especiallydamaging and results in self-enhancement through posses-

1036 JOURNAL OF CONSUMER RESEARCH

sions. The sociometer hypothesis proposes that humans aredriven to maintain a stable level of belongingness (Learyet al. 1995). If they detect evidence of social rejection, theyexperience lower levels of belongingness and subsequentlylower levels of self-esteem, which in turn lead to attemptsto restore feelings of self-worth (Gardner et al. 2005; Love-land, Smeesters, and Mandel 2010; Mead et al. 2011; Pick-ett, Gardner, and Knowles 2004).

Taken together, the literature on self-threat and social be-longingness suggests that when a person's social self isthreatened, a way to enhance the self is to increase thevaluation of his or her possessions. We expect this increasein valuation to hold under selhng conditions only, not underbuying conditions. Consistent with the ownership account,objects in a person's possession become connected with theself (Gawronski et al. 2007; Morewedge et al. 2009), andthis possession-self link is less pronounced in buying con-ditions in which the person is required to consider the costsof the purchase. This argument is in line with recent workby Pettit and Sivanathan (2011), who show that buyingprices for luxury goods typically remain unchanged after aself-threat, when people are forced to consider the psycho-logical cost of the expenditure involved.

In other words, in support of the ownership account, if apossession can enhance the self, we would expect a socialself-threat to increase the possession-self link; this shouldsubsequently drive up the value of a person's possessions(i.e., selling or willingness to accept [WTA] prices) but haveno effect on the value of things outside one's possession(i.e., buying or willingness to pay [WTP] prices). Therefore,we posit the following hypothesis and test it with the firststudy.

HI: Participants in the social self-threat condition willdemand a higher selling price than participants inthe control condition (WTA^ .f > WTA^„„), but nodifference will emerge in buying prices across thetwo conditions (WTA,,|f = WTA,„„).

Social Identity Associations of the Good

Tajfel and Turner (1979, 1985) introduced social identity,or the portion of a person's self-concept derived from hisor her membership in relevant social groups, to explain in-tergroup behavior, such as prejudice, stereotyping, and dis-crimination. According to Tajfel and Turner, people preferthe groups to which they belong (in-groups) over groups towhich they do not (out-groups)—a preference the authorscall "in-group bias." Favoritism toward the in-group elicitsmany positive effects, including an increase in self-esteem(Luhtanen and Crocker 1991; Oakes and Turner 1980). Inthis way, membership in various social groups becomes anintegral part of the self-concept and contributes to people'ssocial identity (Brewer 1991).

Building on social identity theory, consumer behaviorresearch demonstrates that product preferences are oftenbased on the social identity associations inherent in a good

(Escalas and Bettman 2005; Ferraro, Escalas, and Bettman2011; White and Dahl 2007). For example. Escalas andBettman (2005) find that brands consistent with an in-groupenhance self-brand connections for all consumers, whilebrands consistent with an out-group have stronger negativeeffects on independent versus interdependent consumers.Research also suggests that people consume goods both toreaffirm their identities and to communicate their identitiesto others (Escalas and Bettman 2005; Ferraro et al. 2011;Kleine, Kleine, and Allen 1995; Kleine et al. 1993; Reed2004). In this vein. White and Dahl (2007) argue that prod-ucts linked to dissociative reference groups can result innegative self-brand connections, product evaluations, andchoice, particularly when self-presentation concerns are rel-evant.

These findings suggest that products become imbued withmeanings derived from the social identities they symbolize.In other words, social identity theory suggests that in-groupgoods (e.g., a mug with the logo of one's university) andout-group goods (e.g., a mug with the logo of a rival uni-versity) vary in their possession-self hnk. We compare theresults for identity-linked goods with control conditions in-volving a generic good. If, as the ownership account wouldsuggest, the possession-self Hnk is contributing to the en-dowment effect, sellers should value in-group goods (withhigh possession-self links) more, thus exacerbating the en-dowment effect. Similarly, sellers should value out-groupgoods (with low possession-self links) less, thus eliminatingthe endowment effect.

The effect of social identity associations on the posses-sion-self link should be especially prevalent after a socialself-threat. Symbolic self-completion theory (Wicklund andGollwitzer 1981) posits that people use material possessionsand other indicators to communicate aspects of the self,particularly when they feel insecure in such aspects (Braunand Wicklund 1989). A good linked to a person's socialrelationships, therefore, should be especially helpful in re-storing his or her social self. Furthermore, as previouslyargued, a social self-threat increases the connection betweenthe possession and the self as a person strives to restore hisor her self. After a social self-threat, therefore, people shouldadopt generic possessions as part of their identity. When anitem already reflects a person's identity (i.e., an in-groupgood), the possession-self link should be further enhanced,whereas a good that opposes a person's identity (i.e., anout-group good) should weaken the possession-self link.

GenderThere is reason to believe, however, that the downward

pressure on selling prices by out-group identity associationsvaries by gender. Research has long argued that men andwomen differ in their conceptions of the self. Men havewhat is called an individualist, independent, or agentic self-schema (Markus and Kitayama 1991). Men's conception ofthe self comes from striving for power and differentiationfrom others (Bakan 1966; Carlson 1971; Josephs, Markus,and Tafarodi 1992; Wiggins 1991). In contrast, women have


a communal self-schema, which is less focused on differ-entiation and more focused on forming connections withothers (Bakan 1966; Carlson 1971; Markus and Kitayama1991).

Many terms have been used to classify this difference inself-conceptualization between men and women, includingagency communion (Bakan 1966; Carlson 1971). Accordingto Bakan's (1966, 15) definition of agency and communion,"agency manifests itself in the formation of separations;communion in the lack of separations." Agency communion,therefore, is about perceptions of individual distinctiveness([Carlson 1971). Whereas the line between the self and othersis blurred for communal individuals, agentic individuals per-ceive themselves as clearly unique and distinct from others.

Although both men and women may desire relationships,they seem to desire them for very different reasons and actdifferently in relationships. Zarbatany, Conley, and Pepper(2004) show that men are more likely to use friendships toenhance their social prominence. Not surprisingly, in same-sex friendships, men typically exhibit agentic behaviors,while women exhibit more communal behaviors (Suh et al.2004). In consumer behavior literature, gender differencesin agency communion have been shown to influence howmen and women respond to charitable donations, advertis-ing, financial and risky decision making, and social influ-ences while shopping (Byrnes, Miller, and Schäfer 1999;He, Inman, and Mittal 2008; Kurt, Inman, and Argo 2011;Meyers-Levy 1988; Winterich, Mittal, and Ross 2009). Forexample, Winterich et al. (2009) provide evidence that thecommunal orientation of women makes them more likelyto incorporate out-groups into their conceptions of the self.Thus, within a charitable donation context, women with highmoral identity are more favorably disposed to an out-groupthan are men.

As we stated previously, the possession-self link is weakerfor objects associated with one's out-group. Agency-com-munion theory suggests that gender moderates this effect.If men perceive themselves as distinct and different fromothers, out-group distinctions are likely to be significant forthem. Out-group differentiation can serve their needs foruniqueness and separation (Brewer 1991). Thus, the pos-session-self link is likely reduced by out-group associationsfor men. In eonti-ast, women are less likely to classify them-selves as separate from others. The line between themselvesand others is blurred. Therefore, their possession-self linkmay be less affected by out-group associations. In otherwords, the communal nature of women makes them lesslikely to perceive distinctions between self and others, whichshould also make them less likely to classify an out-groupgood as being "not me."

For these reasons, while we expect selling prices to belower for out-group goods than for generic goods amongmen, we do not expect the same pattern for women. For-mally, we predict that under a social self-threat, men willhave low selling prices for an out-group good, thus elimi-nating the difference between selling prices and buyingprices and, consequentiy, the endowment effect. In contrast.

women will show no such out-group derogation and thuscontinue to exhibit the endowment effect for an out-groupgood.

H2: Under a social self-threat, (a) men will exhibit nodifference in selling prices and buying prices of anout-group good (WTA„„, = WTP„J, thus elimi-nating the endowment effect, while (b) women willdemand higher selhng prices than buying prices ofan out-group good (WTA„„, > WTP„„,), thus dem-onstrating the standard endowment effect.

In addition, we suggest that this pattern of results is drivenby the social identity associations of a good. Therefore, weexpect that a contrast between social identity-linked goodsand a generic good will be significant. Specifically, men'spossession-self links will be weaker for out-group goodsthan for generic goods, resulting in lower selling prices ofan out-group good than a generic good for men (but notwomen).

H3: Under a social self-threat, compared with a ge-neric good, (a) men will demand significantly lowerselling prices of an out-group good (WTA„„, <WTAg,„), while (b) women will exhjbit no differ-ence in selling prices (WTA„„, =

We do not expect gender differences, however, to affectthe selling prices of in-group goods. Rather, we believe thatthe possession-self link will be strengthened by the in-groupassociations for both men and women. Thus, in support ofthe ownership account for the endowment effect, we expectthat both men and women will demonstrate exaggeratedendowment effects for in-group goods.

H4: Under a social self-threat, (a) men and (b) womenwill demand higher selling prices than buyingprices of an in-group good (WTAi„ > WTP¡J,thus demonstrating the standard endowment ef-fect.

In addition, because of a strengthened possession-self link,we expect that selling prices of an in-group good will begreater than selling prices of a generic good for both menand women.

H5: Under a social self-threat, compared with a ge-neric good, (a) men and (b) women will demandsignificantly higher selling prices of an in-groupgood (WTAi„ > WT

We test our hypotheses across four studies. In study 1,we test hypothesis 1 by examining the moderating effect ofsocial self-threat on the endowment effect. In study 2, weshow evidence that the social identity associations affect thestrength of the possession-self link and find support for hy-potheses 2-5. In study 3, we find that making intergroupcomparisons salient to women leads them to lower theirselling prices for out-group goods, thus providing additionalevidence of the processes governing gender differences inthe valuation of out-group possessions. Finally, in study 4,


we use a different experimental paradigm with motivationto exchange a good as our dependent measure (instead ofprice valuations) and find similar effects. Study 4 also pro-vides evidence for the role of agency communion in ex-plaining the gender differences within the context of out-group possession valuation.

STUDY 1

In study 1, we tested our hypothesis that a social self-threat would moderate the endowment effect (hypothesis 1).Specifically, we expected that a social self-threat would in-crease selling prices but not buying prices within an en-dowment effect context.

Design and Procedure

The study was a 2 x 2 design with social self-threat (yes,no) and role (seller, buyer) as manipulated between-subjectsfactors. Fifty-five undergraduate students (58% females;42% males) at the University of Pittsburgh were randomlyassigned to the four conditions on the basis of social self-threat and role. The study consisted of two parts: the socialself-threat manipulation or control task and an endowmenteffect experiment. In this study, as well as all the otherstudies, we disassociated the two parts by presenting thestudy as two separate studies, with the social self-threatmanipulation or control task presented as a "life events sur-vey" and the endowment effect experiment presented as a"product evaluation study."

Social Self-Threat Manipulation. We manipulated socialself-threat through a variation of the priming procedure thatBartz and Lydon (2004) use. We asked respondents to imag-ine a previous relationship in which they felt unloved andrejected and further encouraged them to think about howthey felt being in this relationship, to imagine conversationsand interactions with this person, and so forth. Then theywrote about their thoughts and feelings regarding themselvesin relation to this person. Interpersonal rejection is a pow-erful threat to a person's social self and results in a negativeview of self (Shaver and Hazan 1988). The appendix con-tains the social self-threat description and instructions.

We ran a pretest to ensure that the manipulation resultedin a negative view of self (i.e., successfully manipulatedself-threat). Participants (n = 46) were either given the so-cial self-threat manipulation or asked to write about activ-ities they engage in on an average day (control condition).We measured participants' general views of self by aver-aging their agreement with two measures on a 7-point scale(1 = "extremely negative," and 7 = "extremely posi-tive"): "What is your opinion of yourself at this point intime?" and "How do you view yourself at this point intime?" (r = .60). We also measured participants' state self-esteem on social, performance, and appearance dimensionsusing Heatherton and Polivy's (1991) scale (ösociai - -86,Op f — .80, cïjpp = .87). Participants in the social self-threat condition reported lower general self-views than those

in the control condition (M „„ = 5.15 vs. M^^¡f = 4.40;F{1,44) = 4.18, p < .05). Participants in the social self-threat condition also reported lower social state self-esteemthan those in the control condition iM^„„ = 3.54 vs. M^,, =3.06; F(l,44) = 4.08, p < .05). No difference occurred ineither performance or appearance state self-esteem amongthe two groups (p > .10).

If women are more communal in nature than men, it ispossible that a social self-threat is more threatening to them.A desire to connect with others is a fundamental humanmotivation, however, and research shows that men also de-sire relationships with others (Baumeister and Sommer1997; Gabriel and Gardner 1999). Given this, men andwomen should respond similarly in terms of both physio-logical and psychological reactions to a social self-threat(Gruenewald et al. 2004). We checked for gender differencesin the manipulation check measures, but neither gender'smain effect or its interaction effect had a significant impact.This finding suggests that the social self-threat manipulationsuccessfully lowered participants' view of self, both gen-erally and with respect to their social self, and that themanipulation was equally effective for men and women.

Endowment Effect Experiment. After the social self-threat manipulation, respondents participated in a typicalendowment effect experiment (Kahneman et al. 1990) inwhich they assumed the roles of buyers and sellers of ageneric ballpoint pen. Participants in the seller (endowed)condition received a pen at the beginning of the study andwere told that it was theirs to keep in appreciation for com-pleting the study. After completing the social self-threat ma-nipulation, these "sellers" indicated whether they would pre-fer to keep the pen or exchange it for a cash amount. Theymade this decision for each of 40 prices, ranging from $0.25to $10.00 (in $0.25 increments). Participants in the buyer(nonendowed) condition saw the pen after the social self-threat manipulation and then chose whether to receive thepen or the cash amount for each of the 40 prices.

Results and Discussion

We analyzed the data using an ANOVA with price as thedependent variable and social self-threat, role, and their in-teractions as independent variables. The overall model wassignificant (F(3,51) = 11.49, p < .0001). The main effectof role was significant (F(l,51) = 24.59, p < .0001), andthe main effect of social self-threat was marginally signif-icant (F(l,51) = 3.04, p < .09). As expected, the interac-tion between social self-threat and role was significant(F(l,51) = 4.01, p = .05). We also included gender as afactor in the model, but its effect was not significant, so weremoved it for parsimony.

In the control condition, the endowment effect was present,with sellers stating higher prices than buyers (M, ,, = 51.59vs. A/b„y = 50.90; F(l,51) = 4.10, p < .05). The endow-ment eff'ect was also present in the social self-threat condition(M,,,, = 52.47 vs. Mb„y = 50.84; F(l,51) = 25.95, p<.0001). As expected, and in support of hypothesis 1, a social


self-threat increased selhng prices (F(l,51) = 6.59, p<.05) but had no effect on buying prices (F(l,51) = .04,NS). Figure 1 presents the least-squares means for eachcondition.

These results support our hypothesis that a social self-threat increases selling prices, thus moderating the endow-ment effect. After a social self-threat, individuals likely havestrong possession-self hnks because possessions can en-hance the self (Beggan 1992; Sivanathan and Pettit 2010)and help individuals cope with the threat. Thus, the own-ership account would predict that a social self-threat wouldmoderate the endowment effect through an increase in sell-ing prices. Our findings, therefore, are consistent with theownership account.

This study provides initial evidence on the possession-self hnk. We find that when people experience a social self-threat, even a generic good can become linked to the self.While it is not obvious what sort of self-associations a ge-neric good may hold, it does appear that valuations of thesegoods increase by virtue of ownership. Consistent with this,a good imbued with social identity associations shouldgreatly strengthen the possession-self link. Study 2 explicitlymanipulates the possession-self link by varying identity as-sociations of a good to gamer more conclusive evidence ofthe role of the possession-self link.

STUDY 2

In study 2, we tested our hypotheses regarding the mod-erating effects of the social identity nature of the good andgender on the endowment effect. We expected that after asocial self-threat, the endowment effect would be greaterfor in-group goods for both men and women, driven by anincrease in the selling prices of in-group goods comparedwith generic goods. In addition, we expected that after asocial self-threat, the endowment effect would be eliminatedfor out-group goods among men, driven by a decrease inthe selling prices of out-group goods compared with genericgoods. We did not expect this latter pattern of results forwomen.


The study was a 3 x 2 x 2 design with object (generic,in-group, out-group) and role (seller, buyer) as the between-subjects manipulated factors and gender as a measured fac-tor. Two hundred fifty-three undergraduate students (58%females; 42% males) at the University of Pittsburgh wererandomly assigned to the six conditions on the basis ofobject and role. All participants completed the social self-threat manipulation used in study 1. The good was a reusableshopping bag that was generic (i.e., no logo), had the par-ticipants' university's (Pitt) logo prominently printed on it(i.e., in-group), or had the rival university's (Penn State)logo printed prominently on it (i.e., out-group).

After completing the social self-threat manipulation, re-spondents participated in an endowment effect experimentsimilar to that of study 1. In this study, however, we ensured

FIGURE 1

MEAN SELLING (WILLINGNESS TO ACCEPT; WTA) ANDBUYING (WILLINGNESS TO PAY; WTP) PRICES FOR A PEN

(STUDY 1)

$4.00

$3.50

$3.00

Û. $2.501-- $2.00g5 $1.50

$1.00 •

$0.50

$0.00

B Sellers D Buyers(WTA) (WTP)

Control Social Self-Threat

incentive compatibility by telling respondents that we wouldrandomly select one of their choices (or prices) to determinewhat they would receive at the end of the experiment(Becker, DeGroot, and Marschak 1964; Kahneman et al.1991; Lemer, Small, and Loewenstein 2004). At the end ofthe study, the experimenter randomly selected one of theprices listed, and the participants received their choice atthat price (either the bag or the money).

Participants in the selling condition also completed sixitems intended to measure possession-self link strength. Weadapted the items from Escalas and Bettman's (2005) self-concept connection scale. Participants rated their agreementwith the following statements on a 7-point scale: "This totebag makes me feel connected to others," "This tote bag isa statement about how I am a part of a group," "This totebag reminds me of who I am," "This tote bag is a part ofme," "This tote bag says a lot about the kind of person Iwould like to be," and "This tote bag is a statement of whoI want to be." We averaged the six items together to forman overall measure of possession-self link strength (a; =.89).


Manipulation Check. We first tested whether the socialidentity associations of a good successfully manipulated thestrength of the possession-self link among sellers. If ourtheorizing is correct, we would find men to have strongerpossession-self links to in-group possessions than to genericgoods and weaker possession-self links to out-group pos-sessions than to generic goods. Comparing the means acrossthe object conditions supports this expectation. Men hadstronger possession-self links to the in-group bag than tothe generic bag {M,, = 2.94 vs. fg,„ = 2.13; F(l, 103) =


6.60, p < .05). Their possession-self links to the out-groupbag, however, were weaker than to the generic bag (M ,„, —1.41 vs. Mg„ = 2.13; F{\, 103) = 4.76, p < .05).

Recall that our prediction was that women would be moreconnected with the in-group bag; however, we also arguedthat their communal nature makes them less affected by out-group associations. In other words, we expected no differ-ence in possession-self link strength between the out-groupbag and the generic (no logo) bag. In support of this the-orizing, women reported greater possession-self link strengthto the in-group bag than to the generic bag {M;„ — 3.56 vs.Mg,„ = 2.35; F(l, 103) = 15.01, p < .001) but reported nodifference in their possession-self link strength to the out-group bag than to the generic bag (M„„, = 2.14 vs. M^^„ =2.35; F(l, 103) = 1.73, NS). Thus, the social identity as-sociations of the good successfully manipulated possession-self link strength in the predicted manner.

Endowment Effect. We analyzed the data using anANOVA with price as the dependent measure and object,role, gender, and their interactions as independent variables.The overall model was significant (F(ll,241) - 8.95,p < .0001). The main effects of object, role, and gender wereall significant (object: F(2,241) = 4.86, p < . 0 1 ; role;F(I,241) = 42.83, /;<.0001; gender: F(l,241) = 19.43,p < .0001). The interaction between object and role wassignificant (F(2,241) = 5.99, p < .01). The three-way in-teraction among object, role, and gender was also significant(F(2,241) = 3.07, p < .05). No other effects were signifi-cant. We ran simple effects tests for men and women sep-arately; we present these results next.

Men. Tests for simple effects revealed that the inter-action between object and role was significant for men(F(2,241) = 6.10, p < .01). In the out-group condition, theendowment effect was not present for men because therewas no difference between buying and selling prices {M^^¡¡= $1.15 vs.Mb j, = 51.69; F(l, 241) = .65, NS). This pat-tern of results supports hypothesis 2a. In contrast, the en-dowment effect was present in the in-group bag condition(M,,,, = 54.12 vs. M,„y = 51.20; F(l,241) = 14.97, p<.001), in support of hypothesis 4a. The endowment effect wasalso present in the generic bag condition (A/s ,, = 52.73 vs.Mb„y = 51.19; F(l, 241) = 5.86, p < .05).

T'he effect of bag condition was significant among malesellers (F(2,241) = 10.77, p < .0001). Men in the out-group bag condition had significantly lower selling pricesthan those in the generic bag condition (F( 1,241) = 5.20,p < .05). Thus, hypothesis 3a was supported. In support ofhypothesis 5a, men in the in-group bag condition demandedsignificantly higher selling prices than those in the genericbag condition (F(l, 241) = 4.36, p < .05). The effect of bagcondition was not significant, however, among male buyers(F(2,241) = .37, NS).

Women. The interaction between object and role wasonly marginally significant for women (F(2,241) = 2.38,p<.10). Unlike men, and in support of hypothesis 2b,women demonstrated the endowment effect for the out-

group bag (M,,|, = 53.85 vs. M,„y = 51.79; F(l,241) =11.36, p< .001). Like men, and in support of hypothesis4b, women also demonstrated the endowment effect for thein-group bags (M,,,, = 55.10 vs. M^,^ = 52.10; F(l,241) =29.18, p < .0001). The endowment effect was also presentfor the generic bags (M,,,, = 53.69 vs. M^ j, = 52.44;F ( l , 2 4 1 ) = 4.11, p<.05).

The effect of bag condition was significant among femalesellers (F(2,241) = 3.09, p < .05). While no differences oc-curred in selling prices across the out-group and generic bagconditions (F( 1,241) = .06, NS), women in the in-groupcondition demanded significantly higher selling prices thanthose in the generic bag condition (F(l,241) = 4.95, p <.05). This pattern of results supports hypotheses 3b and 5b.There was no effect of bag condition, however, among fe-male buyers (F(2,241) = .83, NS). Figure 2 presents theleast-squares means for each condition.

The results from study 2 demonstrate that social identityplays a moderating role in the endowment effect by affectingselling prices, thus providing further support for the own-ership account. We find that sellers experiencing a socialself-threat have higher valuations of in-group goods than ofgeneric goods, thus exacerbating the endowment effect. Re-garding out-group goods, after a social self-threat, in theselling condition men had lower valuations of such posses-sions than of generic goods, while female sellers exhibitedno such change in valuations. Therefore, the endowmenteffect for an out-group good was not present for men butremained for women.

We also find evidence in study 2 that social identity as-sociations affect possession-self link strength, as we ex-pected. The possession-self link strength for the in-groupgood was the strongest for both men and women, but onlymen demonstrated a weakened possession-self link for out-group goods. Women had an equal connection with out-group and generic possessions. The ownership account forthe endowment effect would suggest that the endowmenteffect for an out-group good is not eliminated for womenbecause their possession-self link is not altered. In supportof this proposition, we find that possession-self link strengthand selling prices were correlated (r = .24, p < .02). Toensure robustness, we also replicated these findings using adifferent type of product (i.e., pens). The results were iden-tical to those obtained with tote bags. (Detailed results arenot provided here to save space but are available on request.)

The preceding results suggest that out-group identity as-sociations do not have as strong an effect on women as onmen. Women's valuations of out-group goods are not dif-ferent from their valuations of generic goods. We have ar-gued that this is because women's communal nature blursthe line between the self and others, making them less pro-ficient at recognizing and comprehending intergroup com-parisons than men. While research has shown that womenmay be more likely to incorporate out-groups into their con-ceptions of the self (Winterich et al. 2009), to our knowledgeno one has examined attentional differences among men andwomen with regard to intergroup comparisons. If this is the


FIGURE 2 •

MEAN SELLING (WILLINGNESS TO ACCEPT; WTA) AND BUYING (WILLINGNESS TO PAY: WTP) PRICES FOR OUT-GROUPIN-GROUP, AND GENERIC BAGS UNDER A SOCIAL SELF-THREAT (STUDY 2)

B

$6.00

$5.00

$4.00

; $3.00 •

$2.00 •

$1.00

$0.00

$5.00

$5.00

$4.00

I

- $3.00

$2.00

$1.00 •

$0.00

;

'Q Sellers D Buyers. • (WTA) . (WTP)

t • . .

Out-group Bag In-group Bag Generic Bag

3 Sellers • Buyers(WTA) (WTP)

/

I

Cut-group Bag In-group Bag Generic Bag

NOTE.—A, Men: S, women.

primary reason for the gender differences observed, makingintergroup comparisons salient across both men and womenshould eliminate the differences based on gender. We setout to test this hypothesis in study 3, which incorporatedan explicit intergroup comparison condition in which par-ticipants received both in-group and out-group goods si-multaneously.

STUDY 3

The purpose of study 3 was to determine whetherwomen's pattern of valuations in an endowment effect forout-group goods would mimic that of men when intergroupcomparisons were salient. Therefore, we modeled .study 3after the work of Nayakankuppam and Mishra (2005). Study3 was similar to study 2, with one key difference. We addeda condition in which participants saw both the in-group andthe out-group bags next to each other before valuing one orthe other. Presumably, seeing the bags next lo each othershould force women to attend to. features they may not au-tomatically attend to, thus increasing the likelihood of in-

tergroup comparison. To establish robustness of the effects,we- also used a different manipulation of social self-threatbased on an.unscrambled sentences task.


One hundred thirteen undergraduate students at the Uni-versity of Pittsburgh took part in the experiment (51% fe-males; 49% males). All participants began by participatingin the social self-threat manipulation. They were asked tounscramble a series of sentences, many of which were re-lated to social self fears (e.g., "I often feel alone," "Peopleare often unsupportive"). The remaining sentences werefiller sentences (e.g., "Jogging is a good way to exercise").

After the social self-threat manipulation, participantsagain received tote bags. Some of the participants (the com-parison group) saw both the in-group (Pitt) and the out-group (Penn State) tote bags. Other participants (the controlgroup) saw only one of the bags. Participants who saw bothtote bags evaluated one of the bags only. Rather than assignparticipants to buying and selling conditions, in this study.


we had them act as both buyers and sellers, imagining them-selves in both roles. Research has shown that simply imag-ining owning an object is sufficient to induce feelings ofownership (Brinbaum et al. 1992; Strahilevitz and Loew-enstein 1998). The order of buying and selling prices wascounterbalanced. Thus, the design of the experiment was a2 (role: buyer vs. seller) x 2 (group: control vs. comparison)X 2 (object: in-group vs. out-group) x 2 (gender: male vs.

female) design, with group and object as between-subjectsfactors, role as a within-subject factor, and gender as a mea-sured variable.


Manipulation Checks. We pretested the scrambled sen-tence task (n = 58) to ensure that it successfully loweredparticipants' view of self and social self-esteem. Half theparticipants completed the scrambled sentence task, and theother half completed a control task in which they wroteabout their trip to campus that day. We again measuredparticipants' general view of self by averaging their agree-ment with two measures on a 7-point scale (1 = "extremelynegative," and 7 = "extremely positive"): "What is youropinion of yourself at this point in time?" and "How do youview yourself at this point in time?" (r = .95). We also usedHeatherton and Polivy's (1991) scale for participants' stateself-esteem on social, performance, and appearance dimen-sions (a^,,,,,, = .91, ccperf = -83, o;„pp = .90).

We found that the scrambled sentence task significantlylowered overall self-view {M^^„ = 5.66 vs. M^^¡( = 4.61;F(l, 56) = 8.87, p < .01). This task also lowered socialstate self-esteem (W,„„ = 3.50 vs. M,,,f = 3.02; F(l, 56) =4.13, p < .05) but had no effect on performance or appear-ance state self-esteem (p > .10). We also checked for genderdifferences in the manipulation check, but neither its maineffect nor its interaction effect had a significant impact.

To check whether showing the two bags together successfullymanipulated intergroup comparison salience among women,we asked the participants to rate their connection with andattachment to their university on 7-point scales (1 = "not atall," and 7 = "very"; r = .73). Women reported being moreattached to their university in tbe comparison condition thanin the control condition (M^^^ = 5.53 vs. M^^^^ — 6.13;F(l, 110) - 3.87, p = .05). Showing the two bags togetherhad no effect, however, on men's attachment to their uni-versity (M,„„ = 5.96 vs. W_p = 5.53; F(l, 110) = 1.94,NS). This pattern of results supports our theorizing aboutgender. If men already incorporate intergroup comparisonsinto their judgments, seeing the bags together should haveno effect on them. If, however, women are less likely toincorporate intergroup comparisons into their judgments,seeing the two bags together should make the comparisonsalient and thus increase their attachment to their university.Therefore, the comparison condition successfully manipu-lated intergroup comparison salience for women.

Endowment Effect. We ran a 2 (role: buyer vs. seller) x2 (group: control vs. comparison) x 2 (object: in-group vs.

out-group) X 2 (gender: male vs. female) mixed ANOVA.To simplify the results, we present them for men and womenseparately.

For men, the three-way interaction among role, group,and object was not significant (F(l,52) = .34, NS). Thiswas not unexpected; we assumed that the comparison wouldhave no effect on men because they likely automaticallyincorporate intergroup comparisons into their judgments. Inthe conti-ol condition, men demonstrated the endowment ef-fect for the in-group bag but not for the out-group bag (in-group: W,,,, = 54.28 vs. Mt,y = 53.57; F(l, 105) = 4.56,p < .05; out-group: M, ,, = $2.08 vs. M ^ = $1.33;F(l,]05) = 1.71, NS). This same pattern of results alsoemerged in the comparison condition (in-group: M^.^^ =53.87 vs. Mb„y = 53.09; F(l,105) = 5.22, /? < .05;' out-group: M,,,, = 52.45 vs. M,„y = 52.00; F(l, 105) = 1.53,NS).

For women, the three-way interaction among role, group,and object was significant (F(l, 53) = 4.05, p < .05). Con-sistent with previous findings, women in the control con-dition 'reported higher selling prices than buying prices forboth the in-group bag and the out-group bag (in-group:A/,,|, = 53.13 vs. M^ j, = 52.40; F(l, 105) = 4.08, p < .05;out-group: M,,,, = 51.54 vs. M,„y = 50.63; F(l, 105) =7.91, p < .01). In the comparison condition, women contin-ued to demonstrate the endowment effect for the in-groupbag (M,,,, = 54.45 vs. M,„y = 52.61; F(l, 105) = 23.93,p < .0001), but the endowment effect was not present forthe out-group bag (M,,,, = $1.06 vs. M^ j, = $0.89; F(l,105) = .13, NS). Tbe least-squares means for each conditionappear in figure 3.

The findings from this study shed light on why out-groupidentity associations have less impact on women's sellingprices in an endowment effect experiment. Because of theiragentic nature, men tend to adopt intergroup comparisonsinto their judgments. In contrast, the communal nature ofwomen blurs the line between the self and others, makingwomen less likely to notice in-group and out-group dis-tinctions. Thus, agency-communion theory combined withsocial identity theory would suggest that men's possession-self links are more affected by out-group identity associa-tions than are women's. When intergroup comparisons aremade salient, however, we find that out-group associationsaffect women's possession-self links and, subsequentiy, theirselling prices.

We have argued that the gender differences within thecontext of an out-group good are due to the agentic natureof men and the communal nature of women. If our resultsare primarily due to these agentic-communal differences, weexpect that in the presence of agency-communion measures,the role of gender effects will be minimized. To obtain fur-ther support for these arguments, in study 4, we measureagency and communion across both men and women. Spe-cifically, we predict that under a social self-threat, partici-pants high in agency will have lower preference for an out-group good than do participants high in communion. Thefindings from the previous studies seem to be a consequence


FIGURE 3

MEAN SELLING (WILLINGNESS TO ACCEPT; WTA) AND BUYING (WILLINGNESS TO PAY; WTP) PRICES FOR IN-GROUPAND OUT-GROUP BAGS (STUDY 3)

B

$6.00

$5.00

$4.00

- $3.00

$2.00

$1.00

$0.00

$6.00

$5.00

$4.00 •L

- $3.00C

$2.00

$1.00 .

$0.00

Q Sellers a Buyers(WTA) (WTP)

Ç

1

Control Control , i ComparisonOut-ijroup Bag In-group Bag Out-group Bag In-group Bag

B Sellers a Buyers(WTA) (WTP)

V

Control Comparison ComparisonOul-group Bag In-group Bag Out-group Bog tn-group Bag

NOTE.—/i\, Men; e, women. Error bars are not provided because this experiment includes a within-subject factor; confidence interval calculationsare based on an alternative calculation that accounts for the within-subject factor (Loffus and Masson 1994).

of the social self-threat, which makes social identity an av-enue to restore one's view of self. We test this assumptionin study 4, by examining whether the effects of social iden-tity are mitigated when people are given an opportunity torestore their self-view before the endowment effect.

Furthermore, an unanswered question from the precedingfindings is whether gender effects driven by agency com-munion persist in situations in which a lower risk is asso-ciated with the transaction. Brown (2005) argues that a cer-tain amount of risk is present in an endowment effectexperiment because of the ambiguity surrounding the truevalue of a good. Specifically, within an endowment effectcontext, sellers are at risk of selling at a price below value,and buyers are at risk of paying a price above value. Thus,sellers (buyers) protect themselves from getting a bad dealby increasing (;decreasing) their offers. To examine the ro-bustness of our findings in situations in which the transactioncost is lowered, we broaden our experimental paradigm instudy 4 by using a riskless choice paradigm similar to thoseused previously in the literature on loss aversion (Inesi

2010). Specifically, we asked participants to imagine ex-changing their identity-linked good for another similar ge-neric good. This reduces the risk of selling at a price belowvalue. If the findings regarding gender persist in this choiceparadigm, it will help establish robustness of the precedingfindings.

STUDY 4

Study 4 had three primary purposes. First, as describedpreviously, we examined whether the results hold when par-ticipants are given an opportunity to affirm their self afterthe social self-threat manipulation (but before the exchangemotivation measures). Second, we included measures ofagency and communion to shed further light on why genderdifferences were obtained in the preceding studies. Third,for reasons stated previously, we changed the experimentalparadigm and examined people's stated propensity to ex-change one good for another. Consequently, we use a dif-


ferent dependent variable to measure valuation in study 4—that is, motivation to exchange.


The study was a 2 x 2 x 2 design with object (in-group,out-group) and self-affirmation (yes, no) as the between-subjects manipulated factors and gender as a measured var-iable. One hundred eighteen college students (48% females;51% males) from an online panel were initially asked toprovide the name of their own university and the name ofa rival university (that competed with their own universityin both academics and athletics). Then they completed 165-point bipolar adjective scales from the Extended Versionof Personal Attributes Questionnaire (Spence, Helmreich,and Holahan 1979), which taps people's agency and com-munion. Examples of agency items are "not at all indepen-dent/very independent" and "feels very inferior/feels verysuperior." Examples of communion items are "very cold inrelation with others/very warm in relation with others" and"not at all aware of others' feelings/very aware of others'feelings." We averaged the eight agency items and the eightcommunion items to form overall scores («agency = -83;«comm = -82). In hne with previous research using suchmeasures (Kurt et al. 2011), we created a new measure toassess relative agency and relative communion orientationby subtracting each respondent's communion score from hisor her agency score. Creating a difference variable of thetwo scores enables us to capture the difference between thetwo dimensions. Higher scores on this difference variableindicate a relative agentic orientation, and lower scores in-dicate a relative communal orientation.

Participants were then exposed to one of the followingitems; the social self-threat manipulation used in studies 1and 2 or the social self-threat manipulation followed by aself-affirmation task. The self-affirmation task involvedranking 11 personal characteristics and values and then elab-orating on their choice for most important (Cohen, Aronson,and Steele 2000). Next, all participants were asked to imag-ine the following situation: "Suppose it is pouring rain out-side and you have to be outside running errands. Unfortu-nately, you have forgotten your umbrella, so you are. forcedto visit the only store nearby that sells umbrellas. Becauseumbrellas are in high demand that day, when you arrivethere is only one umbrella left. It has the colors and logoof [name of own university/name of rival university] on it,and you are forced to purchase it. You take it, walk up tothe register, and pay for it."

Thus, we essentially made all participants "sellers" of anumbrella and then presented them with an opportunity toexchange the umbrella: "After you pay for the umbrella,you start to walk outside when you see the manager bringingout more umbrellas from the stock room. The umbrellas areplain black. The manager tells you that you are able toexchange the umbrella currently in your possession for theplain black umbrella." Then we asked the participants torespond to the following two questions on 7-point scales:"How good would it feel to exchange the umbrella?" ("not

at all good"/"very good") and "How motivated would yoube to exchange your umbrella?" ("not at all motivated"/"very motivated"). We averaged the two items (r = .75) tocreate the dependent measure of exchange motivation.

Results

We conducted an ANOVA with exchange motivation asthe dependent measure and object, gender, and self-affir-mation as independent variables in a full factorial model.The overall model was significant (F(7,110) = 13.53, p <.0001). The main effect of object was significant (F(l, UO)= 74.82, p < .0001), as was the main effect of self-affir-mation (F(l, 110) = 4.55, /?< .05). The two-way interac-tion between object and self-affirmation was also significant(F(l, 110) = 13.36, p < .001). Finally, the three-way inter-action among object, gender, and self-affirmation was sig-nificant (F(l, 110) = 4.30, p < .05). No other effects weresignificant. The least-squares means for each condition ap-pear in figure 4. To simplify the presentation, we discussthe results of the analysis of simple effects conducted formen first and then women.

Men. The interaction between object and self-affirmationwas significant for men (F(l,110) - 17.23, p < .0001).Planned contrasts indicated that in the absence of self-af-firmation, men had greater exchange motivations for the out-group umbrella than for the in-group umbrella (M^ , =5.68 vs. Mi„ = 2.97; F(l, UO) = 53.44, p < .0001). How-ever, a social self-threat with the addition of self-affirmationmitigated this effect (M„„, = 4.79 vs. M,, = 4.25; F(l, UO)= 2.22, p > .13). The addition of the self-affirmation taskdecreased exchange motivation for the out-group umbrella(F(l, UO) = 5.85, p < .05) and increased exchange moti-vation for the in-group umbrella (F(l, UO) = U.95, p<.001).

Women. In contrast, the interaction between object andself-affirmation was not significant for women (F(I, UO) =1.18, NS). Planned contrasts revealed that women hadgreater exchange motivations for the out-group good in boththe presence and the absence of the self-affirmation task (noself-affirmation; M„„, = 4.94 vs. M.,„ = 3.00; F(l, UO) =23.12, /?< .0001; self-affirmation: M„„ = 5.25 vs. M;„ =3.91; F(1,UO) = 13.14, p < .001). However, while self-affirmation had no effect on the exchange motivation forthe out-group good (F(l, UO) = .71, NS), women were lessmotivated to exchange the in-group good in the absence ofthe self-affirmation task (F(l, UO) = 4.93, p < .05).

Recall that in the previous study, out-group identity as-sociations had stronger effects on men's evaluations thanwomen's. In this study, consistent with the previous findings,in the absence of a self-affirmation task, men had greaterexchange motivations for the out-group good than didwomen (F(l, UO) = 4.07, p < .05). This difference was notsignificant, however, in the presence of the self-affirmationtask (F(l, UO) = 1.56, NS). In addition, no gender differ-ences occurred in the exchange motivations for the in-group


FIGURE 4

MEAN EXCHANGE MOTIVATION FOR OUT-GROUP AND IN-GROUP UMBIRELLAS UNDER A SOCIAL-SELF THREAT (STUDY 4)

0Oul-group

Din-group

V

No Self- Self-AffirmationAffirmation

NOTE.—A, Men; S, women.

B

"— S

0 Out-group

O In-group

No Self- Self-AffirmationAffirmation

good in either condition {p > .35), again replicating our pre-vious findings.

In summary, both men and women were more motivatedto exchange the out-group good than the in-group good, butthis result was mitigated somewhat in the presence of a self-affirmation task. Men were more motivated than women toexchange the out-group good, but only in the absence ofthe self-affirmation task. Next, we examined whether theseresults replicated using measures of agency and communion,rather than gender, as a factor.

Agency and Gommunion Measures. As expected, menscored significantly higher than women on the agency-com-munion difference variable (F(l, 116) = 34.92, p < .0001),suggesting that men are more agentic and women are morecommunal. To discover whether our results replicated usingmeasures of agency and communion, we next ran anANCOVA with object, self-affirmation, and agency com-munion in a full factorial model predicting exchange mo-tivation. We also included gender as a covariate. The overallmodel was significant (F(8,109) = 13.83, p < .0001). Themain effects of object and self-affirmation were signifi-cant (object: F(l, 109) = 60.50,/?<.0001; self-affirmation:F(l, 109) = 4.26, p < .05). The two-way interaction be-tween object and self-affirmation was significant (F(l, 109)= 21.22, p < .0001), as was the three-way interactionamong object, self-affirmation, and agency communion(F(I, 109) = 11.52, p < .01). No other effects were signif-icant, including gender as a covariate (F(l, 109) = .01,NS).To further explore the three-way interaction, we examinedthe effect of object and agency communion in the presenceand absence of self-affirmation separately.

In the absence of self-affirmation, the interaction betweenobject and agency communion was significant (F(l, 109) -

6.38, p < .05). Further simple effects tests revealed that theeffect of agency communion was significant in the out-groupumbrella condition (F(l, 109) = 9.58, p < .01) but not inthe in-group umbrella condition (F(l, 109) - 1.01, NS).Plotting the means at 1 standard deviation above and belowthe mean of agency communion revealed that participantswith a relative agentic orientation had greater exchange mo-tivation for the out-group umbrella than did those with arelative communal orientation. In addition, a spotlight anal-ysis at 1 standard deviation above the mean of agency com-munion revealed a significant difference such that partici-pants with a relative agentic orientation reported greaterexchange motivation for the out-group umbrella than for thein-group umbreUa (¿, = -3.13, r(109) = -7.15, p < .0001).A similar spotlight analysis at 1 standard deviation belowthe mean of agency communion revealed that participantswith a relative communal orientation also reported greaterexchange motivations for the out-group umbrella than forthe in-group umbrella {b = -1.80, i(109) = -5.60, p<.0001). Figure 5 plots the mean exchange motivation for thein-group umbrella and out-group umbrella conditions at plusor minus 1 standard deviation from the mean of agencycommunion.

In the presence of self-affirmation, the interaction be-tween object and agency communion was also significant(F(l, 109) = 5.21, p < .05). Further simple effects tests re-vealed, however, that the effect of agency communion wasnot significant in the out-group umbrella condition (F(l, 109)= 2.27, p > .13) and only marginally significant in the in-group condition (F(l, 109) = 3.09, p < .09). Thus, the ef-fect of agency communion was mitigated in the presence ofself-affirmation.


FIGURE 5

IMPACT OF AGENCY COMMUNION ON EXCHANGEMOTIVATION FOR OUT-GROUP AND IN-GROUP UMBRELLAS

(STUDY 4)

i-Oul-group

5.65 I

I 4.77

• 3.23

2.78 <

Relative Agentic RelativeOrientation Communal

Orientation

Discussion

In study 4, we showed that the results occur under a socialself-threat because people view social identity as a way toenhance the self, and we explain gender differences on thebasis of agency and communion. By using another depen-dent variable (exchange motivation), we were also able toshow that the results hold outside a traditional endowmenteffect experiment.

We sought evidence that social identity matters under asocial self-threat because the possession-self link allows so-cial identity-linked possessions to restore individuals' self-views through self-enhancement. When participants had anopportunity to affirm their self before stating their exchangemotivations, no significant gender differences emerged inmotivation to exchange an out-group good. This findingprovides evidence that social identity-linked possessions actas avenues for self-enhancement.

Finally, using measures of agency and communion, wewere able to replicate the gender results. Specifically, weshowed that regardless of agency or communion orientation,participants had greater exchange motivations for an out-group good than for an in-group good, but participants withan agentic orientation had greater exchange motivations foran out-group than did those with a communal orientation.These results further support the argument that the gendereffects we observed are due to men's agentic orientationand women's communal orientation.

GENERAL DISCUSSION

The endowment effect, or the result that the price peopleare willing to pay for a particular good is often significantlyless than the price they are willing to accept to give up thesame good, has been the subject of considerable research.

Although loss aversion has been the traditional explanationfor this effect, researchers have begun to argue that own-ership actually enhances the appeal of the good because apossession becomes tied to the self (Gawronski et al. 2007;Morewedge et al. 2009). If ownership accounts for the en-dowment effect, varying the possession-self link using the-oretically meaningful moderators should have an impact onthe magnitude of the endowment effect. The current worktests three such moderators: social self-threat, social identity,and gender. These moderators affect selling prices in themanner predicted by a possession-self link explanation, thusproviding support for the ownership account for the endow-ment effect.

The endowment effect has been described as the mostrobust representation of loss aversion (Rozin and Royzman2001). There is, however, a growing list of research thatcalls into question the ability of loss aversion to adequatelyexplain the endowment effect. The loss aversion accountmaintains that ownership does not increase the attractivenessof a good (Kahneman et al. 1991). Regarding the socialidentity associations of a good, the loss aversion accountwould predict that sellers are equally attracted to a good asare buyers, regardless of the good's social identity associ-ations. In other words, social identity should have a maineffect on prices across both buyers and sellers, and nochange should occur in the magnitude of the endowmenteffect. We find, however, that social identity associationsaffect selling prices, which suggests that such associationshave a stronger effect on owners' evaluations. The owner-ship account would attribute this result to the social identityassociation changing the strength of the possession-self link.These results extend previous findings that demonstrate thatendowment effect is higher for goods that are associatedwith self (Tom 2004) and for goods that sellers have ownedfor a long time (Strahilevitz and Loewenstein 1998). Theseresults also are consistent with Morewedge et al. (2009),who demonstrate that the endowment effect does not occurwhen buyers have previously owned an identical good,thereby providing support for the ownership explanation.

The fact that there is burgeoning support for the own-ership account for the endowment effect means that we mayhave to revise our thinking with regard to other endowmenteffect findings. For instance, Aggarwal and Zhang (2006)demonstrate that sellers have higher selling prices whennorms of communal, rather than exchange, relationships aresalient. Using an ownership explanation, this result suggeststhat people are more closely tied to their possessions whennorms of a communal relationship are salient. This alsoimplies that other moderators of the endowment effect con-text including specific emotions, and attributes of a goodthat have focused on loss aversion explanations, could bereexamined using an ownership lens (Ariely, Huber, andWertenbroch 2005; Brenner et al. 2007; Carmon and Ariely2000; Lemer et al. 2004).

Although our findings question the loss aversion expla-nation for the endowment effect, this does not necessarilyimply that loss aversion does not exist. Loss aversion or the


idea that individuals typically expect losses to be more im-pactful than gains is a well-established phenomenon. Thecurrent research attempts to demonstrate a particular in-stance in which ownership provides a better explanation thanloss aversion. As Kahneman (1992, 301) noted, "loss aver-sion does not affect all transactions. The critical distinctionis between goods held for use and goods held for exchange."Building on this, Mandel (2002) demonstrates the impactof motivational factors (e.g., transaction demand) on theendowment effect. These, along with the findings from ourresearch, imply that motivational factors can often overridethe impact of loss aversion in infiuencing valuations forgoods. Future research should focus on identifying how mo-tivational factors and loss aversion can jointly explain var-ious consumer behavior phenomena.

There are various substantive implications of these find-ings regarding the role of ownership. If ownership increasesthe valuation consumers place on objects, then marketerscould benefit from any action that creates feelings of own-ership before actual purchase. A number of marketing ac-tions, such as free trials, sampling, and coupons, are likelysuccessful because of their ability to create feelings of own-ership, which may be successful in driving up valuations ofa good. For instance, a consumer given a free trial of acouch may be more willing to pay for the couch after 30days than someone purchasing the couch in the store. Cloth-ing stores may increase feelings of ownership by havingcustomers try on products. Sporting goods stores may wantto allow customers to try out equipment in the store toincrease feelings of ownership.

The fact that sampling, free trials, and coupons are suc-cessful in inducing trial is not surprising. However, the find-ings from the current research suggest that there may be avariety of moderators of this effect, which may provideuseful directions for future research. For instance, accordingto Maddux et al. (2010), some cultures or regions may bemore focused on self-enhancement through material goods.If this is the case, these regions or cultures or customersegments where goods are used for self-enhancement mayfind a reluctance to upgrade to new items or return faultyproducts after a recall because of the strength of the pos-session-self link. These individuals who are more prone toself-enhance using goods also may be more willing to en-tertain promotional offers such as gift certificates, sampling,and buybacks (Sen and Johnson 1997). Alternatively, cir-cumstances under which consumers evaluate these offers(e.g., self-threat) may affect their valuations. Further re-search should investigate these possibilities.

Aside from providing support for the ownership accountfor the endowment effect, our research provides insights intogender differences as they relate to intergroup comparisonsand how these relate to valuations of identity-linked goods.Men's agentic nature, which makes them more likely toperceive the self as separate from others, makes them morelikely to devalue out-group goods. In contrast, women havea communal nature, which makes them less likely to attendto out-group differences, unless the intergroup comparisons

are made salient. While prior research has suggested thatwomen are more likely to include out-groups into their self-concepts (Winterich et al. 2009), to our knowledge we arethe first to demonstrate that there are attention differencesamong men and women that result in varying propensitiesto engage in intergroup comparisons. These gender differ-ences have important imphcations for valuations of identity-linked goods among buyers and sellers.

Our findings regarding gender differences also suggestthat in certain situations marketers may benefit from prompt-ing female consumers to make intergroup comparisons, par-ticularly if they are marketing the in-group brand for thesewomen. Without making the comparisons saiient, womenmay be less likely to spontaneously engage in comparisonswith competitor out-group brands. If a company wants todeter purchase of such competitor brands among women,they should use marketing efforts (e.g., advertising) to makesalient the identity differences across the brands. A goodexample of such an advertisement is Apple's Mac versusPC ads that depict the user identities of the two brands.

Some limitations of the current research are worth high-lighting and suggest useful avenues for future research. Onelimitation is that all of the studies in this research focusedon a single type of identity setting (i.e., university identity).Men and women did not differ in terms of their averageattachment toward either their own university or the rivaluniversity. However, it is possible that men are more likelythan women to draw extreme distinctions between in-groupsand out-groups in terms of sports team affiliations. Futureresearch could replicate these findings with other types ofidentity-associations (e.g., sports team affiliations), in aneffort to either strengthen or reverse the gender effects ob-served in this research. Further, it would be interesting toconsider whether men and women had different interpre-tations of what ownership signifies in an endowment effectcontext. Examining how men and women respond to losses,particularly when they pertain to different types of losses(e.g., monetary vs. social), will provide additional insightsconcerning the impact of gender and identity on loss aver-sion.

Another limitation of the current research is that we onlyfocus on one type of self-threat (i.e., social self-threat). Aninteresting question is whether these results will hold inendowment effect contexts when a person experiences adifferent type of self-threat (e.g., job loss). Varying the typeof self-threat could affect the type of social identity that ismade salient, which in turn could lead to a different rangeof predictions regarding the types of goods that could bevalued. For example, an object associated with college maybe less effective at helping a person cope with a threat tohis or her intelligence. It would also be interesting to ex-amine how possession-self associations and subsequent val-uation of a good associated with an identity are affected bya threat that pertains to that identity.

Future research could provide further evidence of theownership account, by examining other theoretically mean-ingful moderators and boundary conditions. For example.


Gawronski et al. (2007) find evidence that the act of choos-ing an item creates an association between the self and theitem, resulting in implicit self-evaluations being transferredto the chosen object. In traditional endowment effect ex-periments, participants are given items to evaluate. It wouldbe worthwhile to examine the moderating effect of choiceon the endowment effect. The ownership account wouldpredict that choice leads to a stronger possession-self linkand, consequently, to higher selling prices. Another inter-esting extension of our findings is to examine whether pay-ment method (cash vs. credit card) in an endowment effectcontext will result in changing the effects observed here.According to Sivanathan and Pettit (2010), high-status goodswill be especially attractive to buyers who purchase usinga credit card since there is no psychological pain like thatwhich accompanies cash expenditures. Absent the psycho-logical pain of a cash transaction, even buyers may be ableto mimic the pattern of sellers if they are able to imaginethemselves as owners of a good. This would yield rich in-sights regarding the ownership explanation and help identifyadditional boundary conditions for the endowment effect.This, along with other future research suggestions describedpreviously, would provide useful insights to strengthen ourunderstanding of the effect of social identity on consumers'valuations of goods.

APPENDIX

SOCIAL SELE-THREAT MANIPULATION(STUDIES 1-3)

Please think about a relationship that you have had thatfits the description given below and picture in your mindthe person with whom you have had that relationship. Pleasemake sure that the person and the relationship you havechosen to focus on are meaningful and important to yourlife. After reading the relationship description, turn to thenext page.

Please think about a relationship you have had in whichyou have felt uncomfortable being alone and worried thatthe other person didn 't value you as much as you valuedthem.

Now, take a moment and try to get a visual image in yourmind of this person. What does this person look like? Whatis it like being with this person? You may want to remembera time you were actually with this person. What would heor she say to you? What would you say in return? How doyou feel when you are with this person? How would youfeel if they were here with you now? After the visualization,write a sentence or two about your thoughts and feelingsregarding yourself in relation to this person.

REEERENCESAggarwal, Pankaj (2004), "The Effects of Brand Relationship

Norms on Consumer Attitudes and Behavior," Journal of Con-sumer Research, 31 (1), 87-101.

Aggarwal, Pankaj, and Sharmistha Law (2005), "Role of Rela-

tionship Norms in Processing Brand Information," Journal ofConsumer Research, 32 (3), 453-64.

Aggarwal, Pankaj, and Meng Zhang (2006), "The Moderating Ef-fect of Relationship Norm Salience on Consumers' Loss Aver-sion," Journal of Consumer Research, 33 (3), 413-19.

Argo, Jennifer J., Katherine White, and Darren W. Dahl (2006),"Social Comparison Theory and Deception in the Interper-sonal Exchange of Consumption Information," Journal ofConsumer Research, 33 (1), 99-108.

Ariely, Dan, Joel Huber, and Klaus Wertenbroch (2005), "WhenDo Losses Loom Larger than Cains?" Journal of MarketingResearch, 42 (2), 134-38.

Bakan, David (1966), The Duality of Human Existence, Chicago:Rand.

Bartz, Jennifer A., and John E. Lydon (2004), "Close Relationshipsand the Working-Self Concept: Implicit and Explicit Effectsof Priming Attachment on Agency and Communion," Per-sonality and Social Psychology Bulletin, 30 (11), 1389-1401.

Baumeister, Roy F., and Kristin L. Sommer (1997), "What DoMen Want? Cender Differences and Two Spheres of Belong-ingness: Comment on Cross and Madson (1997)," Psycho-logical Bulletin, 122 (1), 38-44.

Becker, Cordon M., Morris H. DeCroot, and Jacob Marschak(1964), "Measuring Utility by a Single-Response SequentialMethod," Behavioral Science, 9 (3), 226-32.

Beggan, James K. (1991), "Using What You Own to Cet WhatYou Need: The Role of Possessions in Satisfying ControlMotivation," Journal of Social Behavior and Personality, 6(6), 129-46.

(1992), "On the Social Nature of Nonsocial Perception:The Mere Ownership Effect," Journal of Personality and So-cial Psychology, 62 (2), 229-37.

Belk, Russell W. (1988), "Possessions and the Extended Self,"Journal of Consumer Research, 15 (2), 139-68.

Braun, Ottmar L., and Robert A. Wicklund (1989), "PsychologicalAntecedents of Conspicuous Consumption," Journal of Eco-nomic Psychology, 10 (2), 161-87.

Brenner, Lyla, Yuval Rottenstreich, Sanjay Sood, and Baler Bilgin(2007), "On the Psychology of Loss Aversion: Possession,Valence, and Reversals of the Endowment Effect," Journalof Consumer Research, 34 (3), 368-76.

Brewer, Marilynn B. (1991), "The Social Self: On Being the Sameand Different at the Same Time," Personality and Social Psy-chology Bulletin, 17 (5), 475-82.

Brinbaum, Michael A., Cregory Coffey, Barbara A. Meilers, andRobin Weiss (1992), "Utility Measurement: Configural-Weight Theory and the Judge's Point of View," Journal ofExperimental Psychology, 18 (2), 331-46.

Brown, Thomas C. (2005), "Loss Aversion without the EndowmentEffect, and Other Explanations for the WTA-WTP Disparity,"Journal of Economic Behavior and Organization, 57 (3),367-79.

Byrnes, James P., David C. Miller, and William D. Schäfer (1999),"Cender Differences in Risk Taking: A Meta-analysis," Psy-chological Bulletin, 125 (3), 367-83.

Campbell, W. Keith, and Constantine Sedikides (1999), "Self-Threat Magnifies the Self-Serving Bias: A Meta-analytic In-tegration," Review of General Psychology, 3 (1), 23-43.

Carlson, Rae (1971), "Sex Differences in Ego Functioning: Ex-ploratory Studies of Agency and Communion," Journal ofConsulting and Clinical Psychology, 37 (2), 261-11.

Carmon, Ziv, and Dan Ariely (2000), "Focusing on the Forgone:


How Value Can Appear So Different to Buyers and Sellers,"Journal of Consumer Research, 27 (3), 360-70.

Cohen, Ceoffrey, Joshua Aronson, and Claude M. Steele (2000),"When Beliefs Yield to Evidence: Reducing Biased Evalua-tion by Affirming the Self," Personality and Social Psychol-ogy Bulletin, 26 (9), 1151-64.

Collange, Julie, Susan T. Fiske, and Rasyid Sanitioso (2009),"Maintaining a Positive Self-image by Stereotyping Others:Self-Threat and the Stereotype Content Model," Social Cog-nition, 27 (1), 138-49.

Escalas, Jennifer Edson, and James R. Bettman (2005), "Self-Con-strual. Reference Groups, and Brand Meaning," Journal ofConsumer Research, 32 (3), 378-89.

Ferraro, Rosellina, Jennifer Edson Escalas, and James R. Bettman(2011), "Our Possessions, Our Selves: Domains of Self-Worthand the Possession-Self Link," Journal of Consumer Psy-chology, 21 (2), 169-77.

Frey, Dieter, and Dagmar Stahlberg (1986), "Selection of Infor-mation after Receiving More or Less Reliable Self-Threat-ening Information," Personality and Social Psychology Bul-letin, 12 (4), 434-41.

Gabriel, Shira, and Wendi L. Cardner (1999), "Are There 'His'and 'Hers' Types of Interdependence? The Implications ofGender Differences in Collective versus Relational Interde-pendence for Affect, Behavior, and Cognition,". Journal ofPersonality and Social Psychology, 77 (3), 642-55.

Gao, Leilei, S. Christian Wheeler, and Baba Shiv (2009), "The'Shaken Self: Product Choices as a Means of Restoring Self-View Confidence," Journal of Consumer Research, 36 (1)29-38.

Gardner, Wendi L., Cynthia L. Pickett, Valerie Jefferis, and MeganKnowles (2005), "On the Outside Looking In: Loneliness andSocial Monitoring," Personality and Social Psychology Bul-letin, 31 (11), 1549-60.

Gawronski, Bertram, Galen V. Bodenhausen, and Andrew P.Becker (2007), "I Like It, because 1 Like Myself: AssociativeSelf-Anchoring and Post-decisional Change of Implicit Eval-uations," Journal of Experimental Social Psychology 43 (2)221-32.

Gruenewald, Tara L., Margaret E. Kemeny, Najib Aziz, and JohnL. Fahey (2004), "Acute Threat to the Social Self: Shame,Social Self-Esteem, and Cortisol Activity," PsychosomaticMedicine, 66 (6), 915-24.

He, Xin, Jeffrey J. Inman, and Vikas Mittal (2008), "Gender Jeop-ardy in Financial Risk Taking," Journal of Marketing Re-search, 45 (4), 414-24.

Heatherton, Todd F., and Janet Polivy (1991), "Development andValidation of a Scale for Measuring State Self-Esteem," Jour-nal of Personality and Social Psychology, 60 (6), 895-910.

Inesi, M. Ena (2010), "Power and Loss Aversion," OrganizationalBehavior and Human Decision Processes, 112 (1), 58-69.

Jordan, Alexander H., and Benoît Monin (2008), "From Sucker toSaint: Moralization in Response to Self-Threat," Psycholog-ical Science; 19 (8), 809-15.

Josephs, Robert A., Hazel Rose Markus, and Romin W. Tafarodi(1992), "Gender and Self-Esteem," Journal of Personality andSocial Psychology, 63 (3), 391-402.

Kahneman, Daniel (1992), "Reference Points, Anchors, Norms,and Mixed Feelings," Organizational Behavior and HumanDecision Processes, 51 (2), 296-312.

Kahneman, Daniel, Jack L. Knetsch, and Richard H. Thaler (1990),"Experimental Tests of the Endowment Effect and the CoaseTheorem," Journal of Political Economy, 98 (6), 1325-48.

(1991), "The Endowment Effect, Loss Aversion, andStatus Quo Bias," Journal of Economic Perspectives, 5 (1)193-206."

Kahneman, Daniel, and Amos Tversky (1979), "Prospect Theory:An Analysis of Decision under Risk," Econometrica 47 (2)263-91.

Kasser, Tim, and Kennon M. Sheldon (2000), "Of Wealth andDeath: Materialism, Mortality Salience, and Consumption Be-havior," Psychological Science, 11 (4), 348-51.

Kleine, Robert E., Ill, Susan Schultz Kleine, and Jerome B. Keman(1993), "Mundane Consumption and the Self: A Social-Iden-tity Perspective," Journal of Consumer Psychology, 2 (3)209-35.

Kleine, Susan Schultz, Robert E. Kleine III, and Chris T. Allen(1995), "How Is a Possession 'Me' or 'Not Me'? Character-izing Types and an Antecedent of Material Possession At-tachment," Journal of Consumer Research, 22 (3), 327-43.

Knetsch, Jack L. (1989), "The Endowment Effect and Evidenceof Nonreversible Indifference Curves," American EconomicReview, 79 (5), 1277-84.

Knetsch, Jack L., and J. A. Sinden (1984), "Willingness to Payand Compensation Demanded: Experimental Evidence of anUnexpected Disparity in Measures of Value," Quarterly Jour-nal of Economics, 99 (3), 507-21.

Kurt, Didem, J. Jeffrey Inman, and Jennifer J. Argo (2011), "TheInfluence of Friends on Consumer Spending: The Role ofAgency-Communion Orientation and Self-Monitoring," your-nal of Marketing Research, 48 (3), 741-54.

Leary, Mark R., Ellen S. Tambor, Sonja K. Terdal, and DeborahL. Downs (1995), "Self-Esteem as an Interpersonal Monitor:The Sociometer Hypothesis," Journal of Personality and So-cial Psychology, 68 (3), 518-30.

Lemer, Jennifer S., Deborah A. Small, and George Loewenstein(2004), "Heart Strings and Purse Strings: Carry over Effectsof Emotions on Economic Decisions," Psychological Science15 (5), 337-41.

Loftus, Geoffrey R., and Michael E. Masson (1994), "Using Con-fidence Intervals in Within-Subject Designs," PsychonomicReview and Bulletin, 1 (4), 476-90.

Loveland, Katherine E., Dirk Smeesters, and Naomi Mandel(2010), "Still Preoccupied with 1995: The Need to Belongand Preference for Nostalgic Products," Journal of ConsumerResearch, 37 (3), 393-408.

Luhtanen, Riia, and Jennifer Crocker (1991), "Self-Esteem andIntergroup Comparisons: Toward a Theory of Collective Self-Esteem," in Social Comparison: Contemporary Theory andResearch, ed. Jerry Suis and Thomas Ashby Wills, Hillsdale,NJ: Erlbaum, 211-34.

Maddux, William W, Haiyang Yang, Carl Falk, Hajo Adam, WendiAdair, Yumi Endo, Ziv Carmon, and Steven J. Heine (2010),"For Whom Is Parting with Possessions More Painful? Cul-tural Differences in the Endowment Effect," PsychologicalScience, 21 (12), 1910-17.

Mandel, David R. (2002), "Beyond Mere Ownership: TransactionDemand as a Moderator of the Endowment Effect," Orga-nizational Behavior and Human Decision Processes 88 (2)737-47.

Markus, Hazel Rose, and Shinobu Kitayama (1991), "Culture andthe Self: Implications for Cognition, Emotion, and Motiva-tion," Psychological Review, 98 (2), 224-53.

Mead, Nicole L., Roy F. Baumeister, Tyler F. Stillman, CatherineD. Rawn, and Kathleen D. Vohs (2011), "Social ExclusionCauses People to Spend and Consume Strategically in the


Service of Affiliation," Journal of Consumer Research, 37(5), 902-19.

Meyers-Levy, Joan (1988), "The Influence of Sex Roles on Judg-ment," Journal of Consumer Research, 14 (4), 522-30.

Morewedge, Carey K., Lisa L. Shu, Daniel T. Gilbert, and TimothyD. Wilson (2009), "Bad Riddance or Good Rubbish? Own-ership and Not Loss Aversion Causes the Endowment Effect,"Journal of Experimental Social Psychology, 45 (4), 947-51.

Nayakankuppam, Dhananjay, and Himanshu Mishra (2005), "TheEndowment Effect: Rose-Tinted and Dark-Tinted Glasses,"Journal of Consumer Research, 32 (3), 390-95.

Oakes, Penelope J., and John C. Turner (1980), "Social Catego-rization and Intergroup Behaviour: Does Minimal IntergroupDiscrimination Make Social Identity More Positive?" Euro-pean Journal of Social Psychology, 10 (3), 295-301.

Park, Lora E., and Jon K. Maner (2009), "Does Self-Threat Pro-mote Social Connection? The Role of Self-Esteem and Con-tingencies of Self-Worth," Journal of Personality and SocialPsychology, 96(1), 203-17.

Pettit, Nathan C, and Niro Sivanathan (2011), "The Plastic Trap:Self-Threat Drives Credit Usage and Status Consumption,"Social Psychological and Personality Science, 2 (2), 146-53.

Pickett, Cynthia L., Wendi L. Gardner, and Megan Knowles (2004),"Getting a Cue: The Need to Belong and Enhanced Sensitivityto Social Cues," Personality and Social Psychology Bulletin,30(9), 1095-1107.

Pierce, Jon L., Tatiana Kostova, and Kurt T. Dirks (2003), "TheState of Psychological Ownership: Integrating and Extendinga Century of Research," Review of General Psychology, 1(1), 84-107.

Reed, Americus, II (2004), "Activating the Self-Importance ofConsumer Selves: Exploring Identity Salience Effects onJudgments," Journal of Consumer Research, 31 (2), 286-95.

Rindfleisch, Arie, James E. Burroughs, and Nancy Wong (2009),"The Safety of Objects: Materialism, Existential Insecurity,and Brand Connection," Journal of Consumer Research, 36(1), 1-16.

Rozin, Paul, and Edward B. Royzman (2001), "Negativity Bias,Negativity Dominance, and Contagion," Personality and So-cial Psychology Review, 5 (4), 296-320.

Sen, Sankar, and Eric J. Johnson (1997), "Mere-Possession Effectswithout Possession in Consumer Choice," Journal of Con-sumer Research, 24 (1), 105-17.

Shaver, Philip R., and Cindy Hazan (1988), "A Biased Overviewof the Study of Love," Journal of Social and Personal Re-lationships, 5 (4), 473-501.

Sivanathan, Niro, and Nathan C. Pettit (2010), "Protecting the Selfthrough Consumption: Status Goods as Affirmational Com-modities," Journal of Experimental Social Psychology, 46 (3),564-70.

Spence, Janet T., Robert L. Helmreich, and Carole K. Holahan(1979), "Negative and Positive Components of PsychologicalMasculinity and Femininity and Their Relationships to Self-Reports of Neurotic and Acting out Behaviors," Journal ofPersonality and Social Psychology, 31 (10), 1673-82.

Strahilevitz, Michal A., and George Loewenstein (1998), "TheEffect of Ownership History on the Valuation of Objects,"Journal of Consumer Research, 25 (3), 276-89.

Suh, Eun Jung, D. S. Moskowitz, Marc A. Foumier, and DavidC. Zuroff (2004), "Gender and Relationships: Influences onAgentic and Communal Behaviors," Personal Relationships^-11 (0,41-60.

Tajfel, Henri, and John Turner (1979), "An Integrative Theory ofIntergroup Conflict," in The Social Psychology of IntergroupRelations, ed. William G. Austin and Stephen Worchel, Mon-terey, CA: Brooks/Cole, 33-47.

(1985), "The Social Identity Theory of Intergroup Be-havior," in The Psychology of Intergroup Relations, 2nd ed.,ed. Stephen Worchel and William G. Austin, Chicago: Nelson-Hall, 7-24.

Thaler, Richard H. (1980), "Towards a Positive Theory of Con-sumer Choice," Journal of Economic Behavior and Organi-zation, 1 (1), 39-60.

Tom, Gail (2004), "The Endowment-Institutional Affinity Effect,"Journal of Psychology: Interdisciplinary and Applied, 138(2), 160-70.

Van Boven, Leaf, David Dunning, and George Loewenstein (2000),"Egocentric Empathy Gaps between Owners and Buyers: Mis-perceptions of the Endowment Effect," Journal of Personalityand Social Psychology, 79 (1), 66-76.

White, Katherine, and Darren W. Dahl (2007), "Are All Out-Groups Created Equal? Consumer Identity and DissociativeInfluence," Journal of Consumer Research, 34 (4), 525-36.

Wicklund, Robert A., and Peter M. Gollwitzer (1981), "SymbolicSelf-Completion, Attempted Influence, and Self-Depreca-tion," Basic and Applied Social Psychology, 2 (2), 89-114.

Wiggins, Jerry S. (1991), "Agency and Communion as ConceptualCoordinates for the Understanding and Measurement of In-terpersonal Behavior," in Thinking Clearly about Psychology,ed. William M. Grove and Dante Cicchetti, Minneapolis: Uni-versity of Minnesota Press, 89-113.

Winterich, Karen Page, Vikas Mittal, and William T. Ross (2009),"Donation Behavior toward In-Groups and Out-Groups: TheRole of Gender and Moral Identity," Journal of ConsumerResearch, 36 (2), 199-214.

Zarbatany, Lynne, Ryanne Conley, and Susan Pepper (2004), "Per-sonality and Gender Differences in Friendship Needs and Ex-periences in Preadolescence and Young Adulthood," Inter-national Journal of Behavioral Development, 28 (4), 299-310.

Copyright of Journal of Consumer Research is the property of Journal of Consumer Research, Inc. and its

content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's

express written permission. However, users may print, download, or email articles for individual use.

Documents

Explaining the Endowment Effect through Ownership: The ... · Explaining the Endowment Effect through Ownership: The Role of Identity, Gender, and Self-Threat SARA LOUGHRAN DOMMER