Journal of Marketing Research Vol. XLVIII (August 2011), 741–754 *Didem Kurt is a doctoral student (e-mail: firstname.lastname@example.org), and J. Jeffrey Inman is Associate Dean for Research and Faculty and the Alfred Wesley Frey Professor of Marketing (e-mail: email@example.com), Katz Graduate School of Business, University of Pittsburgh. Jennifer J. Argo is Cormie Professor of Marketing, University of Alberta (e-mail: Jennifer. Argo@ualberta.ca). The authors gratefully acknowledge the help of the Point of Purchase Advertising Institute and Yesilyurt Shopping Mall, Samsun, Turkey. Gita Johar served as associate editor for this article. DiDem Kurt, J. Jeffrey inman, and Jennifer J. argo* four studies investigate the interactive influence of the presence of an accompanying friend and a consumer’s agency–communion orientation on the consumer’s spending behaviors. in general, the authors find that shopping with a friend can be expensive for agency-oriented consumers (e.g., males) but not for communion-oriented consumers (e.g., females). that is, consumers who are agency oriented spend significantly more when they shop with a friend (vs. when they shop alone), whereas this effect is attenuated for consumers who are communion oriented. the results also show that this interactive effect is moderated by individual differences in self-monitoring such that friends are especially influential for consumers who are high in self-monitoring, but the effects occur in opposite directions for agency- and communion-oriented consumers (i.e., agentic consumers spend more with a friend, while communal consumers spend less when accompanied by a friend). finally, the authors test the underlying process and document that the interaction of agency–communion orientation, the presence of a friend, and selfmonitoring is reversed when the focal context is changed from “spending for the self” to “donating to a charity.” they conclude with a discussion of implications for research and practice. Keywords: social influence, agency–communion theory, self-monitoring, impression management the influence of friends on Consumer Spending: the role of agency– Communion orientation and Self-monitoring © 2011, American Marketing Association ISSN: 0022-2437 (print), 1547-7193 (electronic) 741 Social influences play a pervasive role in shaping consumers’ affect, cognitions, and behaviors (e.g., Argo, Dahl, and Manchanda 2005; Dahl, Manchanda, and Argo 2001; Ratner and Kahn 2002). To date, behavioral researchers have studied the impact of several social characteristics to determine the likelihood and the extent to which the social context will be influential. For example, while high levels of attractiveness and credibility of a salesperson have been shown to enhance the effectiveness of an influence attempt (e.g., Argo, Dahl, and Morales 2008), high levels of persuasion knowledge and cognitive capacity on the part of consumers have been shown to inoculate them from such an influence (e.g., Campbell and Kirmani 2000). Because occurrences of social influence are not always readily apparent or intentional, it seems likely that consumers may not always be prepared to draw from their repertoire of protective strategies to shield themselves from the influence. An example of such an occurrence is when the social influence arises from an unexpected source, such as other shoppers present in the store. Indeed, Argo, Dahl, and Manchanda (2005) find that the mere physical presence of another shopper in a store aisle is sufficient to elicit emotional and behavioral responses in consumers that benefit the retail establishment. In the current research, we aim to push the envelope even further to determine whether the presence of a friend can also create an unintentional cost to the consumer when in the marketplace. We use the term “friend” to refer to relationships ranging from the stage in which the two parties like each other and seek out each other’s company to the stage of friendly relations (Price and Arnould 1999). Previous research has indicated that the behavioral implications of the interaction between two parties, such as compliance to a request, tend to be similar across this range (e.g., Burger et al. 2001; Dolinski, Nawrat, and Rudak 2001). In general, we predict and find that consumers’spending decisions are influenced by accompanying friends as a result of consumers’ impression management concerns. Importantly, we find that the direction of a friend’s effect on consumer spending is moderated by the consumer’s agency– communion orientation (i.e., the tendency to focus on the self or others; Bakan 1966).1 That is, agentic consumers (i.e., males) spend more when they shop with a friend than when they shop alone, whereas communal consumers (i.e., females) are more likely to control their shopping while in the presence of a friend. We also find that this interactive effect is moderated by individual differences in self-monitoring such that friends are especially influential for consumers who are high in self-monitoring, though the effects occur in opposite directions for agency- and communion-oriented consumers (i.e., agentic [communal] consumers spend more [less] when shopping with a friend). Finally, consistent with our impression management explanation, we find that the interactive effect of a friend’s presence, agency–communion orientation, and self-monitoring is reversed when consumers make a donation to a charity. Communion-primed people with high self-monitoring donate more when accompanied by a friend than when they are alone, but we do not observe this effect for agency-primed people. Our research contributes to the social influence literature by extending our understanding of the impact of friends in consumption. First and foremost, the limited research that has studied a friend’s influence has assessed respondents’ perceptions of an imaginary shopper’s likelihood of making unplanned purchases and spending more money in the context of hypothetical shopping situations (Luo 2005). The use of such an artificial methodology is questionable, especially because typical influence agents are salespeople and marketers (Friestad and Wright 1994); thus, consumers may not be cognizant of the extent to which their friends may influence their spending behaviors. Therefore, we study consumers’ behavior in both actual shopping settings (i.e., mass merchandise stores, a bookstore, and a mall) and an experimental setting. Second, we contribute to social influence research (e.g., Argo, Dahl, and Manchanda 2005; Luo 2005; Ratner and Kahn 2002) by showing that the effect of the social environment (i.e., presence vs. absence of a friend) on consumer spending is qualified by individual differences in agency–communion orientations. We achieve this by (1) using gender as a proxy for agency–communion orientation in the pilot study and Study 1, (2) measuring the orientations directly using an individual difference scale in Study 2, and (3) priming the orientations in Study 3. Next, we present our conceptual development. We then define the models used to test our hypotheses and report the results from a pilot and three studies. We conclude with a discussion of the implications of our results and directions for further research. CONCEPTUAL DEVELOPMENT Social influence has been described as one of the primary factors that affect consumers’ decisions. Indeed, Yang and Allenby (2003, p. 291) suggest that “people live in a world in which they are interconnected, information is shared, recommendations are made and social acceptance is important.” Therefore, it is not surprising that the research studying social influence has found that the social environment can shape and sometimes misconstrue consumers’ opinions, preferences, and choice behaviors as they strive for social acceptance (e.g., Argo, Dahl, and Manchanda 2005; Bearden and Etzel 1982; Dahl, Manchanda, and Argo 2001; Ratner and Kahn 2002). To illustrate, Ariely and Levav (2000) find that consumer choices made in group contexts differ systematically from those made in private consumption contexts, because the choices made in the former setting provide an opportunity for consumers to engage in impression management efforts. Netemeyer, Bearden, and Teel (1992, p. 381) note that “in purchasing and using products, people are social actors whose behavior is open to observation of others.… Individuals use products as a form of impression management to influence the ascriptions others might make about them (i.e., form favorable attributions).” While the majority of research to date has studied the impact of social influence in a rather nondescript fashion (i.e., it has studied public vs. private settings), the impact of the specific source of the influence is not as clear. This is an important void to address, because it seems that impression management concerns may be very different if the shopper is standing in the store aisle with a friend versus a stranger (as studied in Argo, Dahl, and Manchanda 2005). Consistent with this expectation, research has shown that the presence of friends (compared with when the shopper is alone) can be highly influential, serving as not only sources of information related to the product (e.g., Urbany, Dickson, and Wilkie 1989) but also activators of impression management concerns on the part of the consumers (e.g., Childers and Rao 1992). However, extending previous research, we argue that the influence of an accompanying friend on consumers’ shopping decisions and spending is moderated by consumers’ agency–communion orientation because agentic and communal people are socialized differently regarding the relative emphasis placed on self- and other-oriented goals (Bakan 1966; Eagly 1987), leading them to have different impression management concerns in the presence of their friends.2 742 Journal of marKeting reSearCh, auguSt 2011 2Previous research (e.g., Funder and Colvin 1988; Stinson and Ickes 1992) has pointed out that unlike strangers, our friends have a history of prior interaction with us and develop a store of knowledge regarding our personalities. Thus, the presence of friends provides people with both the opportunity and the motivation to conform to the expectations their friends have of them, which would bring about social rewards and help avoid social sanctions. 1We use the terms “agency/communion” and “agentic/communal” to refer to the same concepts in the literature, so we use them interchangeably. the influence of friends on Consumer Spending 743 Originally coined by Bakan (1966), the terms “agency” and “communion” capture the notion that people possess two fundamental modalities. In their most simplistic forms, agency refers to a person’s tendency to reflect on his or her individuality and emphasizes the self and its separation from other organisms, whereas communion refers to the merging of a person into a larger organism and social relationships and connections with others (Helgeson 1994). Wiggins (1991) construes agency as a person’s strivings for status and power that facilitate and protect the differentiation of the person from others, whereas communion arises from strivings for cooperation and harmony that protect the unity of the person with a social entity. Accordingly, research has shown that agency involves such qualities as instrumentality, self-confidence, and competence, whereas communion involves such qualities as cooperativeness, concern for others, and kindness (e.g., Eagly 1987). Furthermore, because of differences in their socialization processes, agencyoriented people enjoy putting themselves, their pleasures, and their activities at center stage, whereas communionoriented people refrain from doing so (Bakan 1966). Researchers have explored the usefulness of agency– communion orientations in understanding human behavior in domains such as consumers’ responses to persuasive information (Meyers-Levy 1988), financial risk taking (He, Inman, and Mittal 2008), and donation behavior (Winterich, Mittal, and Ross 2009). To better understand why consumers’ spending behavior in the presence of their friends should be influenced by their agency–communion orientation, we draw from the stereotype literature. This body of work has found that people are motivated to conform to the stereotypic expectations that other people hold about their behavior (e.g., Rosenthal and Rubin 1978). Such a motivation exists because whereas conforming to stereotypic expectations can produce rewards of social approval, violating these expectations risks social sanctions. For example, in the gender domain, females who violate stereotypic expectations by engaging in behaviors typically regarded as masculine (e.g., self-promotion) are rated significantly lower in terms of their social attractiveness (Rudman 1998). Relatedly, research on the “feminine modesty effect” (e.g., Gould and Slone 1982) has shown that in response to normative pressures, females tend to be modest in public contexts. In contrast, society deems it normative and acceptable for males to engage in self-promotion (e.g., Miller et al. 1992). Eagly (1987) and Jost and Kay (2005) suggest that agentic stereotypes (e.g., ambitious, assertive, competent) and communal stereotypes (e.g., warm, considerate, modest) begin to emerge in childhood and are widely held and persistent. Thus, it seems reasonable to expect that these stereotypes would result in different objectives in a social situation and subsequently the use of different self-presentation strategies; in the current context, two specific strategies seem applicable. The first strategy is acquisitive, which focuses on gaining valued outcomes and involves exerting effort to gain admiration, respect, and attention of peers by presenting the self in the most favorable light (Arkin 1981). The second strategy is protective, which is adopted to avoid negative outcomes and is associated with “self-presentations that are cautious, modest, and designed to avoid attention” (Schlenker and Weigold 1992, p. 147; see also Wolfe, Lennox, and Cutler’s  distinction between self-presentations aimed at “getting ahead” of others versus “getting along” with others). On the basis of our conceptual framework, we argue that to conform to the expectations that their friends have of them, agency-oriented consumers will adopt the acquisitive self-presentation style (i.e., “getting ahead”) while shopping with friends and engage in self-promotion through increased spending. In contrast, spending more to impress a friend is not consistent with the modest nature of communion-oriented consumers. Thus, we expect them to adopt the protective self-presentation style (i.e., “getting along”) in the presence of a friend and will control their spending. Although this suggests that communal consumers are not expected to spend more when shopping with a friend, it does not mean that they will decrease their spending. In particular, decreased spending in the presence of a friend represents self-neglect (i.e., focusing on others at the expense of the self) or self-depreciation, and not all communal people have the skills or tendency to perform such behavior (Buss 1990; Fritz and Helgeson 1998). Thus, we do not predict a systematic decline in communal consumers’ spending when they are accompanied by a friend. Rather, we argue that the positive impact of a friend’s presence on agentic consumers’ spending will not be observed in the case of communal consumers. Formally, H1: Agency-oriented, but not communion-oriented, consumers spend more when they shop with a friend than when they shop alone. PILOT STUDY The Point of Purchase Advertising Institute periodically conducts field studies of consumers’ purchasing behavior. It fielded its most recent study in 1995 and provided the data for the present analysis. In-store intercept interviews were conducted at 14 mass-merchandise stores. Consumers were intercepted randomly as they entered the store and were asked several questions. After respondents finished shopping, they returned to the interviewer, who collected their receipts and assessed demographics. The key dependent variable in the study was the amount of money participants spent. Data were collected from 1230 customers, 12 of whom we excluded from the analysis because of missing responses. We also excluded 10 extreme observations identified using studentized residuals, Cook’s D, and hat diagonal.3 Of the 1208 usable respondents, 555 shopped alone, and 72 were accompanied by a friend.4 3We deemed these observations to be outliers because of the extremeness of the magnitude of their dependent variable (i.e., very low or very high actual spending compared with that predicted by the model). For example, six participants spent less than $1.50. Because such observations would have undue impact on the estimated coefficients and their standard errors (as well as on the overall fit of the model), excluding them enables us to avoid reporting potentially misleading results driven by the presence of a few outliers in the data set. Note that the pattern of results is similar when we run the analysis without excluding these outliers. We performed the same outlier diagnostics in other studies as well, and if we excluded any outliers, it is noted. 4We categorized accompaniers into eight groups: friend (72), spouse (138), parent (42), child (298), someone else’s child (48), adult family member (48), someone else (19), and unknown (156); 229 shoppers were accompanied by more than one person. In accordance with previous research (e.g., He, Inman, and Mittal 2008; Winterich, Mittal, and Ross 2009), we operationalized agency–communion orientation as gender in this study (78% of respondents were female). Gender is a reasonable proxy for the orientation, because as Bakan (1966) suggests, and research has demonstrated (for a review, see Guimond et al. 2006), agency orientation is more characteristic of males, whereas communion orientation tends to pertain to females. We analyzed data using ordinary least squares regression, in which the dependent variable was the natural log of the dollar amount the respondent spent.5 We used contrast coding for our two focal independent variables of gender (1 if male, –1 if female) and friend (1 if with friend, –1 if not accompanied by a friend). We controlled for a variety of factors that could potentially affect consumers’ spending, such as the amount of money they planned to spend, the amount of time they spent in the store, and the method they used to pay for their purchases (Inman, Winer, and Ferraro 2009). In addition, we included social variables in the model to control for the impact of other types of relationships (e.g., spouse) and multiple accompaniers (Latané 1981; for details of the model specification, summary statistics, and the complete regression results, refer to the Web Appendix at http://www.marketingpower.com/jmraug11). The overall regression model is significant (F(25, 1182) = 45.24, p < .01), and the model R-square is 48.9%. We find a significant and positive main effect for friend (b2 = .12, p < .05). Importantly, this main effect is qualified by a positive and significant interaction between friend and gender (b18 = .15, p < .01). That is, controlling for planned spending, male (i.e., agentic) consumers spend 56% more when they shop with a friend than when they shop alone, while female (i.e., communal) consumers spend 4% less when they shop with a friend than when they shop alone (though this latter difference is not significant). This result provides initial support for our hypothesis. Figure 1 visually depicts the interaction effect. The results also reveal that the main effects for the other relationship categories are not significant (ps > .17). More important, none of the interactions between gender and other social influence categories are significant (ps > .16), implying that males (i.e., agentic consumers) and females (i.e., communal consumers) do not exhibit differential sensitivity to social influence stemming from sources other than their friends. A key limitation of the pilot study is that participants provided the classification of the “friend,” and thus, it is a subjective perception. To address this limitation, in Study 1, we manipulate the friend’s presence with a trained confederate assuming the role of a friend that is present during a shopping trip. STUDY 1 Method Study 1 uses a retail shopping setting to test a 2 (orientation: agency vs. communion) ¥ 2 (social presence: alone vs. accompanying friend) between-subjects experimental design. The key dependent variable is amount spent. Again, we operationalized orientation as participants’ gender. Eightyseven undergraduate students (43 males and 44 females) from a large North American university completed the study. Procedure. Participants took part in what ostensibly were two unrelated studies. In the first study, participants were run in groups of two or three. In half the groups, unbeknownst to participants, a confederate assumed the role of one of the study participants.6 The goal of the confederate was to become acquainted with the actual participants. We achieved this by having the researcher leave the participants and the confederate alone for an extended period of time (i.e., she went to photocopy more surveys). The confederate followed a script to both initiate and maintain a conversation with the participants during the researcher’s absence. After the researcher returned, the confederate responded to the same survey as the participants. Within the next few days, participants individually completed a second study that took place at the university student center. Upon arrival, they were informed that the purpose of the study was to collect marketing research information for the university bookstore and that to do this they would be asked to go to the store, make a product purchase, and then return to the experimenter to complete a short survey. They were further 744 Journal of marKeting reSearCh, auguSt 2011 5In the model, we control for the amount of money that the shoppers planned to spend. Our results remain unchanged when we use the difference between actual and planned spending as the dependent variable. figure 1 Pilot StuDy: moDerating effeCt of genDer on the relationShiP BetWeen PreSenCe of a frienD anD ConSumerS’ SPenDing 3.3 3.2 3.1 3.0 2.9 2.8 2.7 2.6 2.5 ln(A mount Spent) alone With friend male female Notes: Agentic consumers (i.e., males) spend more when they are accompanied by a friend than when they are alone, whereas there is no difference in the spending of communal consumers (i.e., females) between the two conditions. We calculated the amount spent in each condition at sample means of ln(planned), time spent, age, and ln(income), assuming cash payment and no use of in-store specials. 6To control for potential impact of gender match/mismatch between the confederate and participants, we used two confederates (one woman, one man) in the study, and we randomly assigned participants to each confederate. As we discuss in the “Results” section, neither the gender of the confederate nor the gender match/mismatch between the confederate and participants affects our results. the influence of friends on Consumer Spending 745 told that to determine which product they would purchase, they would select an envelope that contained the name of a product under $5. Unbeknownst to the participants, each of the envelopes identified a package of four AA batteries as the product to purchase. Participants were given $5 and told that they could keep both the product and any remaining change from the purchase. Participants then went to the bookstore to locate and purchase the designated product. The battery display included five brands of AA batteries that varied in price and quality levels. Pretesting established the prices of the five brands to reflect differences in their perceived quality: Duracell/Energizer were rated the best (1; Maverage = 5.92) and were priced at $4.29, Rayovac/Panasonic were rated average (2; Maverage = 4.39) and were priced at $3.99, and Chateau was rated the worst (3; Maverage = 2.85) and was priced at $3.69. Paired-sample t-tests revealed that differences between group means were significant (ps < .01). In the friend condition, when participants entered the store aisle, the confederate they had met previously was standing next to the battery display. In the alone condition, no one else was present in the store aisle. Participants selected and purchased their brand and then returned to the experimenter, where they completed a short questionnaire. In the survey, amidst questions related to the cover story, participants were asked to indicate the brand of batteries they had purchased. We compared participants’ responses to this question with that recorded by an observer situated two aisles away from the battery display with a clear view of participants. In addition, participants indicated their gender, age, and major and completed an open-ended suspicion probe. Examination of the suspicion probe indicated that none of the participants were aware that the two studies were related or guessed the research’s hypotheses. While the confederate cannot be considered a friend per se, previous research (e.g., Burger et al. 2001; Dolinski et al. 2001) has shown that short conversations with strangers lead people to treat them as if they were friends. For example, using a manipulation similar to ours, Burger et al. (2001) find that participants in a conversation (vs. a control) condition complied with a request from the confederate at a higher rate, as if they had been asked by a friend. Similarly, Dolinski, Nawrat, and Rudak (2001, p. 1405) point out that “people involved in a dialogue [but not in a monologue] with a stranger automatically treat him or her as a friend and, consequently comply with his or her request.” Thus, our manipulation enables us not only to control for closeness of friendship and avoid potential problems arising from participant-provided “friend” classifications but also to create an experimental setting in which we can observe participants’ spending decisions as if they were made in the presence of a friend. Results We conducted regression analysis with amount spent as the dependent variable and the accompanying friend, participant’s gender, and their interaction term as the independent variables. We used contrast coding for both gender (1 if male, –1 if female) and friend (1 if with friend, –1 if alone). The overall regression model is significant (F(3, 87) = 8.02, p < .01), and the model R-square is 22.5%. We found significant main effects for friend (b = .08, p < .01) and gender (b = .05, p < .05). Importantly, the analysis reveals a positive and significant friend ¥ gender interaction (b = .06, p < .01). Consistent with H1, males spend significantly more in the presence of a friend than in the alone condition (Mfriend = $4.25 vs. Malone = $3.96, p < .01), whereas the average spending for females did not differ as a function of the social presence (Mfriend = $4.02 vs. Malone = $3.98, p > .54; for the percentage of brands selected in each condition, see the Web Appendix at http://www.marketingpower.com/ jmraug11). Figure 2 visually depicts the interaction effect. Moreover, as Table 1 shows, the confederate’s gender does not affect our results; males increase their spending in both figure 2 StuDy 1: moDerating effeCt of genDer on the relationShiP BetWeen PreSenCe of a frienD anD ConSumerS’ SPenDing $4.40 $4.30 $4.20 $4.10 $4.00 $3.90 $3.80 A mount Spent alone With friend male female Notes: Agentic participants (i.e., males) spend more in the presence of a friend than when they are alone, whereas there is no difference in the spending of communal participants (i.e., females) between the two conditions. table 1 SamPle StatiStiCS anD aVerage aCtual SPenDing aCroSS ConDitionS for StuDy 1 With Friend With Male With Female Difference: Difference: Difference: Alone (1) (2) Friend (2a) Friend (2b) (2) – (1) (2a) – (1) (2b) – (1) Male $3.96 $4.25 $4.26 $4.22 $.29 $.30 $.26 (n = 23) (n = 20) (n = 11) (n = 9) (p < .01) (p < .01) (p < .01) Female $3.98 $4.02 $3.99 $4.07 $.04 $.01 $.09 (n = 24) (n = 20) (n = 13) (n = 7) (p > .54) (p > .88) (p > .37) the male and female friend conditions (vs. alone condition), whereas we observed no significant change in the spending of females across conditions. Furthermore, we reestimated our model by including gender match (1 if the participant and the confederate’s genders match, –1 if otherwise) and gender mismatch (1 if the participant and the confederate’s genders do not match, –1 if otherwise) variables in lieu of the friend variable. Under this specification, both variables being –1 indicates that the participant is alone. Furthermore, we interacted these variables with the gender of the participant. The results reveal positive and significant coefficients for gender match (b = .10, p < .01) and gender mismatch (b = .07, p < .05). The difference between the two coefficients is not significant (F(1, 81) = .82, p > .36). In addition, the coefficients of the interaction terms are positive and significant (bgender ¥ match = .05, p < .10, and bgender ¥ mismatch = .06, p < .05). There is no significant difference between the two coefficients (F(1, 81) = .11, p > .73), indicating that the friend’s effect is not driven by gender match/mismatch. Discussion Study 1 demonstrates that agentic consumers (i.e., males) spend significantly more money when they shop with a friend than when they shop alone, whereas communal consumers (i.e., females) tend to control their spending in the presence of a friend. The finding that males spend more while females are more modest in the presence of a friend is consistent with our impression management framework. Study 2 has two primary objectives: First, we directly measure individual differences in consumers’ agency–communion orientation instead of using gender as a proxy, and second, because research has found that consumers differ in their responsiveness to social and interpersonal cues of situationally appropriate behavior (Gangestad and Snyder 2000), we explore the moderating role of self-monitoring. STUDY 2 Self-Monitoring as a Moderator Effective impression management efforts require that people accurately scan the social situation for cues to determine how to respond and adjust their behavior accordingly. The theory of self-monitoring (Lennox and Wolfe 1984; Snyder 1974, 1987) posits that people differ in terms of their ability and willingness to engage in expressive control and strategically manage their public appearances. More specifically, according to Gangestad and Snyder (2000), high (vs. low) self-monitors are better at monitoring their behavior and regulating their self-presentation to convey desired public appearances. However, previous research has documented that selfmonitoring has an asymmetric impact on the public behavior of agentic and communal people (e.g., Bozin and Yoder 2008; Flynn and Ames 2006). For example, Flynn and Ames (2006) find that higher self-monitoring provides additional benefits to communal people (i.e., females), but not to agentic people (i.e., males), in the context of selfenhancement. In their first study, an analysis of peer evaluations of the participants who completed a semester-long group project documents that high- and low-self-monitoring males are rated as equally valuable and influential contributors to the group by their peers. In contrast, female group members who exhibit high (vs. low) self-monitoring are considered more valuable and influential contributors. Moreover, in their second study, results of a dyadic negotiation exercise reveal that males high in self-monitoring do not perform better than those who are low in self-monitoring, whereas the negotiation outcome increases with high selfmonitoring in the case of females. The authors attribute these findings to the notion that males tend to naturally exhibit the valued traits of competence and self-confidence; thus, monitoring the situation and realizing the demand for self-confidence does not boost their performance. In contrast, females increase their portrayal of competence and self-confidence when they are high self-monitors who realize that the situation demands this; thus, they perform better. As Flynn and Ames (2006, p. 279) point out, “We do not predict (nor find evidence) that men and women exhibit different levels of self-monitoring. Instead, we propose that the impact of self-monitoring may be different for men and women because they experience different gender stereotypes.” In light of these findings and the results of our first study, we anticipate that regardless of the level of their selfmonitoring, agency-oriented consumers will spend more when accompanied by a friend than when they are alone. This is because we do not expect higher self-monitoring to provide additional benefits to them in the process of selfenhancement. Specifically, in the context of shopping for the self, increased spending is often associated with selfpromotion (Griskevicius et al. 2007), a typical behavior exhibited by agentic people. Thus, agentic consumers will not obtain additional benefits from monitoring the situation and realizing that engaging in self-promotion through increased spending would be a stereotype-consistent selfpresentation style. However, this should not be the case for communion-oriented consumers, for whom the impact of high self-monitoring on public behavior should be stronger. In particular, communals with high (vs. low) self-monitoring have the ability and tendency to adopt the “protective” selfpresentation strategy that the situation calls for and engage in stereotypically consistent behavior to convey a favorable impression. Thus, high self-monitors will exert even more control over their spending in the presence of a friend and exhibit a heightened level of modesty (or self-depreciation), leading them to reduce their spending compared with when they are alone. In contrast, we do not expect communals with low selfmonitoring to decrease their spending in the presence of a friend. These consumers have difficulty in creating favorable impressions in the eyes of others (Bozin and Yoder 2008; Flynn and Ames 2006) because of a lack of skills in reading cues regarding socially appropriate behavior, which results in them failing to alter their behavior accordingly. Our predictions can be summarized as follows: H2: Self-monitoring moderates the impact of a friend’s presence on spending for communion-oriented consumers but not for agency-oriented consumers. Specifically, (a) agentic consumers with both high and low self-monitoring spend more when they shop with a friend than when they shop alone, and (b) communal consumers with high, but not low, selfmonitoring spend less when they shop with a friend than when they shop alone. 746 Journal of marKeting reSearCh, auguSt 2011 the influence of friends on Consumer Spending 747 Method One hundred thirty-six shoppers were intercepted randomly as they entered a large shopping mall located in Turkey. Respondents were compensated with two movie tickets (worth approximately $10) in exchange for their participation in the study. Only customers shopping alone or accompanied by a single friend were invited to participate in this study.7 Although 136 customers participated in our study, we excluded 7 respondents from the sample because of missing responses. We also excluded 3 extreme observations, leaving 126 respondents, of which 53% were female and 45% were accompanied by a friend. Table 2 summarizes the sample statistics. Respondents were asked to complete two surveys. Following Erdem, Swait, and Valenzuela (2006), to ensure that the items included in the surveys were correctly translated and conveyed the same meaning in Turkish, we used the standard technique of back-translation (from English to Turkish and then back to English).8 The entry survey included questions such as “How often do you visit this shopping mall?” and “How much do you plan to spend in this shopping mall today?” whereas the exit survey assessed the amount they spent, agency–communion orientations, self-monitoring, payment method, and demographics. We also measured buying impulsiveness (Rook and Fisher 1995) as an additional control variable. Measures Agency/communion. We used 16 five-point (1 = “low,” and 5 = “high”) bipolar adjective scales from the extended version of the Personal Attributes Questionnaire (Spence, Helmreich, and Holahan 1979) to measure agency and communion. The reliability and validity of these widely used scales have been well documented (e.g., Helgeson 1994). Examples of items that assess agency are “not at all independent/very independent” and “feels very inferior/feels very superior.” Examples of items that assess communion are “very cold in relations with others/very warm in relation with others” and “not at all aware of others’ feelings/very aware of others’ feelings.” We averaged the responses to create their respective orientations (aagency = .67, and acommunion = .76). Because a person can embody both agency and communion dimensions and a high score on agency or communion does not necessarily suggest a low score on the other dimension, we needed a measure to capture the difference between the two dimensions. Thus, after calculating agency and communion scores for each respondent, we created a new measure to assess relative agency orientation, ACDIF, by subtracting each respondent’s communion score from his or her agency score.9 The ACDIF measure enables us to assess not only the direction but also the relative magnitude of each respondent’s agency–communion orientation. Self-monitoring. We measured self-monitoring using Lennox and Wolfe’s (1984) revised self-monitoring scale, which consists of 13 items rated on seven-point scales (1 = “strongly disagree,” and 7 = “strongly agree”). The scale includes items such as “In social situations, I have the ability to alter my behavior if I feel that something else is called for” and “When I feel that the image I am portraying isn’t working, I can readily change it to something that does.” We combined these items and averaged them together to create a self-monitoring index (a = .75).10 Results The regression model included a contrast-coded variable for being accompanied by a friend (1 if with friend, –1 if alone), and we included relative agency–communion and self-monitoring in the model as continuous variables. The model also includes two-way interactions and a three-way interaction of these variables. To reduce multicollinearity, we mean-centered the continuous variables (Aiken and West 1991). Similar to the analysis in the pilot study, we included several control variables such as income and buying impulsiveness in the model (for details of the model specification and the measures, refer to the Web Appendix at http://www.marketingpower.com/jmraug11). The ordinary least squares regression results indicate that the overall model is significant (F(16, 109) = 12.89, p < .01), and the model R-square is 66.9% (see Table 3). In addition, all variance inflation factors are less than 1.7, suggesting that our results do not suffer from multicollinearity. The main effect for friend is both positive and significant (d2 = .14, p < .05). Furthermore, consistent with H1, the interaction between friend and ACDIF is positive and significant (d14 = .28, p < .05),11 indicating that the level of the table 2 SamPle StatiStiCS for StuDy 2 Frequency % Gender Male (alone) 33 26.19 Female (alone) 35 27.77 Male (with friend) 28 22.22 Female (with friend) 30 25.00 Payment Method Cash 55 43.65 Credit 57 45.24 Cash and credit 14 11.11 7We did not invite shoppers accompanied by more than one friend to participate in the study, because previous research (e.g., Argo, Dahl, and Manchanda 2005; Latané 1981) shows that the strength of a social influence increases with the number of sources, which might have confounded our results. Limiting number of accompanying friends to one provides a more conservative test of our theory. 8Two Turkish doctoral students studying at a North American university and a translation agency operating in Turkey translated the scales. Disagreements were resolved through discussion. 9Dindia (2006, p. 11) rationalizes this measure: “Women and men differ in degree if both possess the same trait or display the same behavior but one possesses or displays more of it. Thus, if both women and men are agentic and communal, but women are more communal and men are more agentic, then with respect to agency and communion, they differ in degree, not kind.” Winterich, Mittal, and Ross (2009, p. 213) also point out that “individuals who are not distinctly categorized as either masculine or feminine may experience identity conflict.” In their study examining donation behavior, they find that the pattern of results for androgynous and undifferentiated participants is inconsistent with the pattern exhibited by either those with masculine gender identity or those with feminine gender identity. 10Consistent with Flynn and Ames (2006), we do not find a significant correlation between agency–communion orientation (i.e., ACDIF) and selfmonitoring (r = –.15, p > .10). 11Consistent with our results in previous studies, when we use gender as a proxy for agency–communion, there is again a positive and significant interaction between friend and gender (b = .14, p < .05), as well as a significant main effect for friend (b = .12, p < .05). difference between agency and communion orientation of a person affects the degree to which he or she is influenced by the presence of a friend during a shopping trip. More important, as H2 predicts, there is a significant three-way interaction of friend, ACDIF, and self-monitoring (d17 = .31, p < .05).12 In addition, the coefficients of all the main control variables have the expected signs, but only planned amount (d1 = .71, p < .01) and paying with a credit card (d5 = .14, p < .05) are statistically significant. We also conducted several reestimations of the model (e.g., correcting for potential sample selection bias) to provide more insight into our findings. The results are substantively unchanged. We present these analyses in the Web Appendix (http://www. marketingpower.com/jmraug11). To facilitate the interpretation of the three-way interaction, we followed the post hoc probing procedure Aiken and West (1991) recommend. We first calculated high (low) values for ACDIF and self-monitoring by adding (subtracting) the standard deviation to (from) the mean. We then conducted simple slope analysis, which examines the interaction between ACDIF and presence of a friend during a shopping trip on amount spent at low and high levels of self-monitoring. This analysis enables us to assess whether the pattern of results is consistent with the specific predictions of the second hypothesis. Panels A and B of Figure 3 visually depict the moderating effect of self-monitoring for high and low self-monitors, respectively. We also report the average actual spending across conditions in Table 4. As H2a predicts, we find that the slopes for high selfmonitor/high ACDIF (b = .378, t = 2.90, p < .01) and low self-monitor/high ACDIF (b = .243, t = 2.15, p < .05) are both significantly different from zero, indicating that regardless of their level of self-monitoring, consumers with high 748 Journal of marKeting reSearCh, auguSt 2011 table 3 regreSSion reSultS for StuDy 2 Equation: ln(amount spent) Parameter Estimate t-Value Intercept 3.99** 34.46 ln(planned amount) .71** 9.90 Friend .14* 2.16 Time spent in the store .003 1.61 ln(income) .02 .24 Credit .14* 2.00 Mixed payment (cash and credit) .15 1.21 In-store special .12 1.82 Buying impulsiveness .01 .23 Visit –.04 –.99 Self-monitoring (SM) –.13 –1.40 ACDIF .08 .76 Age .001 .17 Gender –.03 –.50 Friend ¥ ACDIF .28* 2.60 Friend ¥ SM –.10 –1.19 SM ¥ ACDIF .10 .70 Friend ¥ ACDIF ¥ SM .31* 2.12 R2 67.0% *p < .05. **p < .01. 12We also estimated a modified version of our model in which the dependent variable was the difference between actual and planned spending and we dropped ln(planned) from the right-hand side of the model. The coefficients on the friend ¥ ACDIF and friend ¥ ACDIF ¥ self-monitoring are still positive and significant at the 5% level, whereas the coefficient on friend is positive but not significant. 4.2 4.0 3.8 3.6 3.4 ln(A mount Spent) alone With friend w/ Friend H high aCDif low aCDif w/ Friend w/ Friend 4.2 4.0 3.8 3.6 3.4 ln(A mount Spent) alone With friend w/ Friend H high aCDif low aCDif w/ Friend w/ Friend figure 3 StuDy 2: three-Way interaCtion BetWeen PreSenCe of a frienD, aCDif SCore, anD Self-monitoring A: High Monitora B: Low Monitorb aHigh (low) ACDIF consumers spend more (less) when they are accompanied by a friend than when they are alone. bBoth high and low ACDIF consumers spend more when they are accompanied by a friend than when they are alone, though the increase in the spending of the latter group is not significant. table 4 aVerage aCtual SPenDing aCroSS ConDitionS for StuDy 2 Alone With Friend Low SM High SM Low SM High SM High ACDIF 58.29 68.31 90.72 177.61 (n = 19) (n = 16) (n = 17) (n = 12) Low ACDIF 75.48 87.46 82.07 64.61 (n = 16) (n = 17) (n = 11) (n = 18) Notes: We used median splitting to assign the participants to high/low ACDIF and self-monitoring conditions. We report mean spending in Turkish Lira (TL); at the time of the study, $1 = 1.2TL. the influence of friends on Consumer Spending 749 ACDIF scores spend significantly more when they are with a friend than when they are alone. Although the slope for high self-monitors is greater than that of low self-monitors, the difference between the slopes is not significant (t = .82, p > .41). Furthermore, the slope for high self-monitor/low ACDIF is negative and significant (b = –.254, t = 1.84, p < .05, onetailed), implying that high self-monitors with low ACDIF spend less when they shop with a friend than when they shop alone. Although the slope for low self-monitor/low ACDIF is positive, it is not statistically significant (b = .203, t = 1.33, p > .18). Finally, the difference between the two slopes is statistically significant (t = 2.06, p < .05). These results support H2b. Discussion The results support our thesis that self-monitoring qualifies the impact of the presence of a friend on spending for communion-oriented consumers but not for agency-oriented consumers. We find that agentics consistently spend more when they are accompanied by a friend than when they are alone, regardless of self-monitoring. However, higher selfmonitoring moderates the impact of a friend’s presence on the spending of the communion-oriented consumers. Communals with high self-monitoring spend less in the presence of a friend, whereas there is no difference in the amount communals with low self-monitoring spend when they shop alone and with a friend. Thus far in our analysis, we either use gender as a proxy for the agentic or communal nature of participants (pilot study and Study 1) or measure agency–communion orientation (Study 2). Although previous research has employed both methods (e.g., He, Inman, and Mittal 2008; Helgeson 1994; Winterich, Mittal, and Ross 2009), priming agency– communion orientation will enable us to test our hypotheses in a more controlled setting. This is one of the objectives of Study 3. Furthermore, a basic premise of our research is that communion-oriented consumers’ impression management concerns lead them to control their spending in the presence of their friends. While the spending context utilized in the first three studies is self-focused (i.e., counter to a communionoriented perspective), there are certain instances in which increased spending is consistent with communal stereotypes. One such instance might involve donations to a charity, because the communal stereotypic beliefs mainly describe a concern with the welfare of other people and communionoriented people embody such traits as being caring, helpful, and sympathetic (i.e., characteristics inherent in a charity; Eagly 1987). Conversely, agency-oriented people place emphasis on independence from others and embody such traits as being self-reliant (i.e., characteristics not inherent in donating to a charity). Thus, if differences in impression management concerns of agency- and communion-oriented consumers are the underlying reason for their differential sensitivity to a friend’s influence, our findings should reverse when we examine communion- and agency-primed people’s donation behavior in the presence of their friend.13 Specifically, agency-primed consumers with both high and low self-monitoring should adapt a “protective” self-presentation strategy (rather than “acquisitive” style as they do in the spending context); thus, they should neither increase nor decrease their donation in the presence of their friend. It is important to note that although decreased donation is consistent with the self-reliant nature of agentic consumers, it will increase the risk of being perceived as “greedy” and “cheap,” which is inconsistent with agentics’ aspiration for status among their peers. As a result, we predict no effect (rather than a decline) of the presence of a friend on the amount donated by agentic consumers, regardless of their selfmonitoring. We expect communion-primed people with high self-monitoring to donate more to a charity in the presence of a friend (vs. when they are alone), whereas communionprimed people with low self-monitoring should not change their donation when they are accompanied by a friend (vs. alone). STUDY 3 Method We employed a 2 (orientation: agency vs. communion) ¥ 2 (social presence: alone vs. friend) ¥ 2 (self-monitoring: high vs. low) between-subjects experimental design. We manipulated orientation and social presence, whereas we measured self-monitoring. One hundred ninety-two undergraduate students from a large North American University completed the study in exchange for course credit. Procedure. Undergraduate students signed up for the study with a friend who was also a registered undergraduate student at the same university. Upon arrival, the pairs of friends were informed that three randomly selected participants would receive $50 after the study. They were then randomly assigned to agency/communion and friend/alone conditions. Participants in the alone condition were told that they would complete the study in separate rooms, whereas participants who were assigned to the friend condition remained in the same room until the end of the study. The experimenters (two males and two females) were also randomly assigned to the different conditions. The first part of the survey included a priming task and manipulation check exercise. The experimenter then presented participants with a list of eight (fictitious) charities and verbally asked each participant whether he or she would like to donate to a charity if he or she won the $50, and if so how much. In the friend condition, although the experimenter asked the donation question to both participants simultaneously, we included only the first participant’s answer in the analysis because the other participant’s response might be influenced by the first participant’s response. Finally, participants in both the friend and alone conditions were given the second part of the survey, which contained the self-monitoring scale,14 demographic questions, and an open-ended suspicion probe. None of the participants guessed the focal hypotheses of the research. Measures Agency–communion prime and self-monitoring. To manipulate agency–communion orientation, we used a 13We thank the review team for this suggestion. 14Neither the agency–communion prime nor the presence of a friend had a significant impact on the self-monitoring scores (rAC – SM = .14, p > .10; rF – SM = –.06, p > .50). scrambled-sentence task. Participants were presented with 20 scrambled sentences, of which 15 were related to agency or communion orientation, depending on the prime. The remaining 5 sentences were not related to either prime and were categorized as neutral (for the full list of sentences, see the Appendix). We took words and phrases used for each prime from Eagly (1987), Meyers-Levy (1988), and Winterich, Mittal, and Ross (2009). For example, the agency prime included sentences such as “personal beliefs are important” and “I try to be assertive,” while the communion prime included sentences such as “social norms are important” and “I try to be selfless.” We again measured selfmonitoring (a = .74; Lennox and Wolfe 1984). Manipulation check. To verify that the agency–communion prime was successful, we used Kuhn and McPartland’s (1954) task in which participants completed ten “I am ...” statements. Two independent research assistants coded each response as either agentic or communal (95% agreement, with any disagreements resolved through discussion). Agentic statements referred to a personal description, attitude, or belief focusing on self (e.g., “I am independent,” “I am tall”). Communal statements referred to either relationships or sensitivity to others (e.g., “I am helpful,” “I am a daughter”) or a demographic group or category to which the participant belongs to (e.g., “I am a marketing major,” “I am a Christian”). Coders classified statements that did not relate to either category as other (e.g., “I am hungry”), and we excluded these statements from the analysis. Participants in the agency-priming condition wrote more agentic statements than those in the communion-priming condition (Magency = 5.30 vs. Mcommunion = 4.62; t = 2.38, p < .05). Participants in the communion-priming condition wrote more communal statements than those in the agency-priming condition (Magency = 3.42 vs. Mcommunion = 4.55; t = 4.00, p < .01). These results indicate that agency–communion orientation was successfully primed.15 Results Excluding accompanying friends and two outliers from the analysis resulted in a final sample size of 124 respondents (45% female; 52% with friend). The average donation was $23.14 (SD = $17.44). Table 5 reports the average donation across conditions. We use regression to test our hypothesis, with selfmonitoring mean-centered to minimize multicollinearity (all variance inflation factors < 1.5). We used contrast coding for agency–communion prime (1 if agency, –1 if communion), friend (1 if with friend, –1 if alone), and gender (1 if male, –1 if female). We regressed donation amount on the agency–communion prime, friend, self-monitoring, twoway interactions, three-way interaction of these variables, and gender. The results indicate that the overall model is significant (F(8, 115) = 2.17, p < .05), and the model Rsquare is 13.14% (see Table 6). As we predicted, there was a significant three-way interaction of agency–communion prime, friend, and self-monitoring (b = –6.67, p < .05). The friend ¥ self-monitoring interaction was also significant (b = 7.00, p < .01). No other effects were significant. To facilitate the interpretation of the three-way interaction, we followed the post hoc probing procedure that Aiken and West (1991) recommend (see Figure 4). Consistent with our hypothesis, for both high and low self-monitors in the agency-priming condition, there is no significant relationship between presence of a friend and donation amount (bA-HSM = –1.32, t = –.42, p > .60; bA-LSM = –1.73, t = –.59, p > .55). Conversely, for those with the communion prime and high self-monitoring, the presence of a friend has a positive and significant impact on donation amount (bC-HSM = 4.93, t = 1.65, p = .05, one-tailed test); the presence of a friend (vs. alone) leads to higher donations by communion-primed participants with high self-monitoring. This result is consistent with the argument that because communal people tend to be caring and nurturing, displaying a portrait that is consistent with these characteristics in front of a friend may bring about social rewards. Conversely, communion-primed participants with low selfmonitoring donated less in the presence of a friend than those in the alone condition (bC-LSM = –11.85, t = –3.29, p < .01). This result is unexpected, and we speculate on why it arose in the “Discussion” section. However, overall, our results provide support for the predicted reversal of the friend effect in the donation (vs. spending) context. Discussion Study 3 demonstrates that the presence of a friend and self-monitoring interact to influence donation behavior of 750 Journal of marKeting reSearCh, auguSt 2011 15Given that agency–communion orientation may be difficult to manipulate because of internalization of these characteristics through socialization at an early age (Eagly 1987; Winterich, Mittal, and Ross 2009), an experimental design in which agency–communion orientation is primed provides a conservative test of our hypothesis. table 5 aVerage aCtual Donation aCroSS ConDitionS for StuDy 3 Alone With Friend Low SM High SM Low SM High SM Agency priming $25.28 $29.17 $22.67 $23.42 (n = 18) (n = 12) (n = 15) (n = 19) Communion priming $28.13 $17.50 $12.69 $24.71 (n = 16) (n = 14) (n = 13) (n = 17) Notes: We used median splitting to assign the participants to high-/lowself-monitoring conditions. table 6 regreSSion reSultS for StuDy 3 Equation: Amount Donated Parameter Estimate t-Value Intercept 23.06** 14.86 Friend –2.49 –1.60 Agency–communion (AC) 2.38 1.54 Self-monitoring (SM) 3.54 1.33 Gender –1.23 –.80 Friend ¥ AC .97 .63 Friend ¥ SM –.10 –1.19 SM ¥ AC 7.00** 2.64 Friend ¥ AC ¥ SM –6.67* –2.51 R2 13.1% *p < .05. **p < .01. the influence of friends on Consumer Spending 751 communion-oriented people but not agency-oriented people. Stated differently, the direction of the interaction between presence of a friend, agency–communion orientation, and self-monitoring documented in Study 2 reversed when the consumer decision under examination changed from spending for the self to donation to a charity. Jointly, Studies 2 and 3 provide a test of the underlying role of impression management concerns. We document that people, with the exception of communals with low self-monitoring, shape their spending decisions in the presence of their peers to avoid counteracting the stereotypes associated with their orientation. A puzzling finding we obtained in this study is that communals with low self-monitoring decrease their donation to a charity in the presence of a friend (vs. alone). A possible explanation for this finding is that low self-monitors simultaneously exert less effort and try to avoid being perceived as making an effort to create a good impression, which may sometimes lead them to exhibit context inappropriate behavior (i.e., donating less to a charity in the presence of a friend because they were primed to think that others value nurturance). This is consistent with the items in Snyder’s (1987) self-monitoring scale that describe low self-monitors (e.g., “At parties and social gatherings, I do not attempt to do or say things that others will like”; “I feel a bit awkward in company and do not show up quite so well as I should”). It is noteworthy that previous research has also documented unexpected findings regarding the public behavior of low self-monitors (e.g., Ratner and Kahn 2002; White and Gerstein 1987). Further research is needed to reconcile theoretical arguments and empirical findings on the behavior of low self-monitors in different types of public contexts. GENERAL DISCUSSION Across three field studies and a lab experiment, we demonstrate the expensive impact of a “friendly” social influence on consumers’ actual spending decisions. In general, our findings suggest that the effect has the greatest implications for agentic consumers (i.e., males) because a decision to shop with a friend (vs. alone) tends to have negative ramifications for their pocketbook—that is, they spend more with an accompanying friend. This caveat does not seem to hold for communal consumers (i.e., females). Indeed, communal consumers with high self-monitoring spend significantly less money when they shop with a friend than when they shop alone. These findings seem to be spending-context dependent, in that we also document that when the spending is for a good cause (i.e., donating to a charity), communals with high self-monitoring loosen their purse strings in the presence of a friend (vs. alone), while donation behavior of agentics is not influenced by an accompanying friend. Our investigation of the impact of an accompanying friend on consumer spending makes important contributions with implications for both consumers and managers. First, previous research on social influences has found that friends influence consumers’ purchase decisions in a positive way by providing information related to the product (Urbany, Dickson, and Wilkie 1989). We extend this research by demonstrating that friends can also have deleterious implications for a shopper’s wallet, in that agentic shoppers spend more when they are accompanied by a friend than when they shop alone. Furthermore, the variability in our empirical design enables us to control for any confounding social factors, including mere presence effect (Argo, Dahl, and Manchanda 2005). For example, we conducted our second study in a large shopping mall, where both solo shoppers and those accompanied by a friend were subject to the mere presence effect of other shoppers in the stores, but we document a significant friend effect beyond any mere presence effect. Our research also explores the underlying mechanism that drives our effects. We theorize that the presence of a friend affects consumer spending because it motivates consumers to engage in impression management. To empirically explore this possibility, we identify and test the moderating roles of a consumer’s agency–communion orientation and 30 25 20 15 10 5 A mount Donated alone With friend w/ Friend A agency Communion w/ Friend w/ Friend 30 25 20 15 10 5 A mount Donated alone With friend w/ Friend A agency Communion w/ Friend w/ Friend figure 4 StuDy 3: three-Way interaCtion BetWeen PreSenCe of a frienD, agenCy–Communion orientation, anD Self-monitoring A: High Monitora B: Low Monitorb aCommunion-primed participants with high self-monitoring donate more to a charity in the presence of a friend than when they are alone, whereas there is no difference in the amount donated by agency-primed participants between the two conditions. bCommunion-primed participants with low self-monitoring donate less to a charity in the presence of a friend than when they are alone, whereas there is no difference in the amount donated by agency-primed participants between the two conditions. individual differences in self-monitoring and test the impact of the spending context. First, we argue that support for an impression management mechanism would be provided if consumers engage in stereotypically consistent behaviors in the presence of their friends. According to the stereotype literature (e.g., Rosenthal and Rubin 1978), people are motivated to engage in behaviors that are consistent with existing stereotypes when they are in public settings. A stereotype of agency-oriented people is that they are selforiented; thus, in the present context, a consistent behavior could be self-promotion manifested through increased spending. Conversely, because a stereotype of communionoriented people is that they are group focused, a behavior consistent with this stereotype in the current research would be one that would prevent the person from standing out (i.e., they would be modest and would limit their spending). We find support for such effects. Second, the definition of self-monitoring revolves around the idea that those who have high levels of this individual difference are likely to adapt and change their behaviors (i.e., manage their impressions) in the presence of others. However, previous research (e.g., Flynn and Ames 2006) has also suggested that higher self-monitoring provides additional benefits to communal but not agentic people in the process of impression management. Thus, Study 2’s demonstration that our effects arise asymmetrically for communal and agentic consumers lends additional credence to the proposed underlying mechanism. The impact of an accompanying friend on the spending of agentic and communal consumers who are high in self-monitoring is positive and negative, respectively, whereas it is positive for both groups who are low in self-monitoring (though the impact is not significant for communal consumers with low self-monitoring). Finally, we provide evidence for the underlying role of impression management concerns by finding that the pattern of results is dependent on the spending context. In particular, agentics/communals with high self-monitoring spend more in the presence of a friend (vs. alone) in contexts in which impression management concerns are paramount (agentic = self-focused situations; communal = other-focused situations). In addition to extending the current literature on social influence in the marketplace, our research contributes to the nascent literature examining how agency–communion orientation affects consumers’ monetary decisions. Whereas previous studies have demonstrated the role of agency and communion on consumers’financial risk taking (He, Inman, and Mittal 2008) and donation behavior (Winterich, Mittal, and Ross 2009) in a private decision-making context, our study focuses on impression management–related spending implications of the orientations. Our findings have important implications for managers. Given that agentic consumers’spending seems to be highly susceptible to the presence of a friend, managers should focus on strategies that will help them attract a higher number of male consumers who shop with their friends. By creating shopping environments that prime an agentic orientation and encourage shopping with friends, retailers may be able to boost sales. For example, offering promotions such as “bring a friend and both get an extra X% off” targeted to male consumers can be effective in that increased spending by both consumers can cover expenses of the promotion and generate additional revenue. Indeed, such a promotion might also legitimize spending for females, because both shoppers will be receiving a discount. Further research is needed to explore whether there are conditions (other than donating to a charity) under which communion-oriented people will purchase more when in the presence of a friend. It seems likely that there are instances in which communion-oriented consumers may not want to convey an impression of being modest, such as when they are purchasing a gift for someone else. In this case, modesty concerns may go by the wayside in favor of creating other types of impressions (e.g., a generous friend). Research should also examine the conditions under which the type of store inhibits communion-oriented consumers from imposing mental constraints on their behavior (i.e., behaving modestly) or spurs them to spend more when in the presence of a friend. For example, it is possible that in more experiential environments (e.g., salon, spa), the physical relaxation from receiving the treatment might simultaneously relax tendencies to engage in the stereotypically consistent behavior of modesty, and as a result, communionoriented people may be more likely to spend more when a friend is present (e.g., try additional services). Few purchase decision are made in a social void. Thus, it is important to examine how others influence consumers’ purchase decisions and spending. In this research, we took a crucial step toward achieving this goal by presenting a comprehensive analysis of the impact of shopping with friends, a major source of social influence in the shopping context, on consumers’spending. However, more research is needed to better understand the nature of the relationship between a friend’s presence and a consumer’s spending. APPENDIX AGENCY–COMMUNION PRIMING FOR STUDY 3 We are interested in how people form meaningful English sentences. Please form meaningful, non-question sentences from the following scrambled words. To complete the exercise successfully, you need to use all of the words given for each sentence. Agency Important personal are beliefs. Being ambitious to success is key. Control individuals seek to others. Usually I on myself focus. Achieve aspiring individuals goals their. Virtuous is a quality displaying self-sufficiency. Convictions I my stand by own. Being important a leader is. Respect get people accomplished. Separate individuals are others from. Try assertive to be I. Competition enjoyable makes life more. Concern for I have well-being my own. Power people for strive. Bring happiness alone spending time may. Communion Important social are norms. Assisting others to happiness is key. Conform with individuals seek to others. 752 Journal of marKeting reSearCh, auguSt 2011 the influence of friends on Consumer Spending 753 Usually I on others focus. Make caring a difference people. Virtuous is a quality displaying nurturance. While decisions making thoughts others’ consider I. Being important a follower is. Respect get people modest. Connected individuals are others to. Try selfless to be I. Cooperation enjoyable makes life more. Concern for I have of others the welfare. Togetherness people for strive. Bring happiness with others spending time may. Neutral Listening to music our minds clear can. Exercise a good way to jogging is. Events I know college related. Daily life a part of technology is. Guided by life is knowledge. REFERENCES Aiken, Leona S. and Stephen G. West (1991), Multiple Regression: Testing and Interpreting Interactions. Newbury Park, CA: Sage Publications. Argo, Jennifer J., Darren W. Dahl, and Rajesh V. Manchanda (2005), “The Influence of a Mere Social Presence in a Retail Context,” Journal of Consumer Research, 32 (2), 207–212. ———, ———, and Andrea C. Morales (2008), “Positive Consumer Contamination: Responses to Attractive Others in a Retail Context,” Journal of Marketing Research, 45 (December), 690–701. Ariely, Dan and Jonathan Levav (2000), “Sequential Choice in Group Settings: Taking the Road Less Traveled and Less Enjoyed,” Journal of Consumer Research, 27 (4), 279–90. Arkin, Robert M. (1981), “Self-Presentation Styles,” in Impression Management Theory and Social Psychological Research, James T. Tedeschi, ed. New York: Academic Press, 311–33. Bakan, David (1966), The Duality of Human Existence. Chicago: Rand McNally & Company. Bearden, William O. and Michael J. Etzel (1982), “Reference Group Influence on Product and Brand Purchase Decisions,” Journal of Consumer Research, 9 (2), 183–94. Bozin, Marie A. and Janice D. Yoder (2008), “Social Status, Not Gender Alone, Is Implicated in Different Reactions by Women and Men to Social Ostracism,” Sex Roles, 58 (9/10), 713–20. Burger, Jerry M., Shelley Soroka, Katrina Gonzago, Emily Murphy, and Emily Somerwell (2001), “The Effect of Fleeting Attraction on Compliance Request,” Personality and Social Psychology Bulletin, 27 (12), 1578–86. Buss, David M. (1990), “Unmitigated Agency and Unmitigated Communion: An Analysis of the Negative Components of Masculinity and Femininity,” Sex Roles, 22 (9/10), 555–68. Campbell, Margaret and A. Kirmani (2000), “Consumers’ Use of Persuasion Knowledge: The Effects of Accessibility and Cognitive Capacity on Perceptions of an Influence Agent,” Journal of Consumer Research, 27 (1), 69–83. Childers, Terry L. and Akshay R. Rao (1992), “The Influence of Familial and Peer-Based Reference Groups on Consumer Decisions,” Journal of Consumer Research, 19 (2), 198–211. Dahl, Darren W., Rajesh V. Manchanda, and Jennifer J. Argo (2001), “Embarrassment in Consumer Purchase: The Roles of Social Presence and Purchase Familiarity,” Journal of Consumer Research, 28 (3), 473–81. Dindia, Kathryn (2006), “Men Are from North Dakota, Women Are from South Dakota, ” in Sex Differences and Similarities in Communication, Kathryn Dindia and Daniel J. Canary, eds. Mahwah, NJ: Lawrence Erlbaum Associates, 3–20. Dolinski, Dariusz, Magdelena Nawrat, and Izabela Rudak (2001), “Dialogue Involvement as a Social Influence Technique,” Personality and Social Psychology Bulletin, 27 (11), 1395–406. Eagly, Alice H. (1987), Sex Differences in Social Behavior: A Social-Role Interpretation. Hillsdale, NJ: Lawrence Erlbaum Associates. Erdem, Tülin, Joffre Swait, and Ana Valenzuela (2006), “Brands as Signals: A Cross-Country Validation Study,” Journal of Marketing, 70 (January), 34–49. Flynn, Francis J. and Daniel R. Ames (2006), “What’s Good for the Goose May Not Be as Good for the Gander: The Benefits of Self-Monitoring for Men and Women in Task Groups and Dyadic Conflicts,” Journal of Applied Psychology, 91 (2), 272–81. Friestad, Marian and Peter Wright (1994), “The Persuasion Knowledge Model: How People Cope with Persuasion Attempts,” Journal of Consumer Research, 21 (1), 1–31. Fritz, Heidi L. and Vicki S. Helgeson (1998), “Distinctions of Unmitigated Communion from Communion: Self-Neglect and Overinvolvement with Others,” Journal of Personality and Social Psychology, 75 (1), 121–40. Funder, David C. and C. Randall Colvin (1988), “Friends and Strangers: Acquaintanceship, Agreement, and the Accuracy of Personality Judgment,” Journal of Personality and Social Psychology, 55 (1), 149–58. Gangestad, Steven W. and Mark Snyder (2000), “Self-Monitoring: Appraisal and Reappraisal,” Psychological Bulletin, 126 (4), 530–55. Gould, Robert J. and Caroline Slone (1982), “The ‘Feminine Modesty’ Effect: A Self-Presentational Interpretation of Sex Differences in Causal Attribution,” Personality and Social Psychology Bulletin, 8 (3), 477–85. Griskevicius, Vladas, Joshua M. Tybur, Jill M. Sundie, Robert B. Cialdini, Geoffrey F. Miller, and Douglas T. Kenrick (2007), “Blatant Benevolence and Conspicuous Consumption: When Romantic Motives Elicit Strategic Costly Signals,” Journal of Personality and Social Psychology, 93 (1), 85–102. Guimond, Serge, Armand Chatard, Delphine Martinot, Richard J. Crisp, and Sandrine Redersdorff (2006), “Social Comparison, Self-Stereotyping, and Gender Differences in Self-Construals,” Journal of Personality and Social Psychology, 90 (2), 221–42. He, Xin, J. Jeffrey Inman, and Vikas Mittal (2008), “Gender Jeopardy in Financial Risk Taking,” Journal of Marketing Research, 45 (August), 414–24. Helgeson, Vicki S. (1994), “Relation of Agency and Communion to Well Being: Evidence and Potential Explanations,” Psychological Bulletin, 116 (3), 412–28. Inman, J. Jeffrey, Russell S. Winer, and Rosellina Ferraro (2009), “The Interplay Between Category Characteristics and Shopping Trip Factors on In-Store Decision Making,” Journal of Marketing, 73 (September), 19–29. Jost, John T. and Aaron C. Kay (2005), “Exposure to Benevolent Sexism and Complementary Gender Stereotypes: Consequences for Specific and Diffuse Forms of System Justification,” Journal of Personality and Social Psychology, 88 (3), 498–509. Kuhn, Manford H. and Thomas S. McPartland (1954), “An Empirical Investigation of Self-Attitudes,” American Sociological Review, 19 (1), 58–76. Latané, Bibb (1981), “The Psychology of Social Impact,” American Psychologist, 36 (4), 343–56. Lennox, Richard D. and Raymond N. Wolfe (1984), “Revision of the Self-Monitoring Scale,” Journal of Personality and Social Psychology, 46 (6), 1349–64. Luo, Xueming (2005), “How Does Shopping with Others Influence Impulsive Purchasing?” Journal of Consumer Psychology, 15 (4), 288–94. Meyers-Levy, Joan (1988), “The Influence of Sex Roles on Judgment,” Journal of Consumer Research, 14 (4), 522–30. Miller, Lynn C., Linda L. Cooke, Jennifer Tsang, and Faith Morgan (1992), “Should I Brag? Nature and Impact of Positive and Boastful Disclosures for Women and Men,” Human Communication Research, 18 (3), 364–99. Netemeyer, Richard G., William O. Bearden, and Jesse E. Teel (1992), “Consumer Susceptibility to Interpersonal Influence and Attributional Sensitivity,” Psychology and Marketing, 9 (5), 379–94. Point of Purchase Advertising Institute (1995), The1995 POPAI Consumer Buying Habits Study. Englewood, NJ: Point of Purchase Advertising Institute. Price, Linda L. and Eric J. Arnould (1999), “Commercial Friendships: Service Provider–Client Relationships in Context,” Journal of Marketing, 63 (October), 38–56. Ratner, Rebecca K. and Barbara E. Kahn (2002), “The Impact of Private Versus Public Consumption on Variety-Seeking Behavior,” Journal of Consumer Research, 29 (2), 246–57. Rook, Dennis W. and Robert J. Fisher (1995), “Normative Influences of Impulsive Buying Behavior,” Journal of Consumer Research, 22 (3), 305–313. Rosenthal, Robert and Donald B. Rubin (1978), “Interpersonal Expectancy Effects: The First 345 Studies,” Behavioral and Brain Sciences, 1 (3), 377–415. Rudman, Laurie A. (1998), “Self-Promotion as a Risk Factor for Women: The Cost and Benefits of Counterstereotypical Impression Management,” Journal of Personality and Social Psychology, 74 (3), 629–45. Schlenker, Barry R. and Michael F. Weigold (1992), “Interpersonal Processes Involving Impression Regulation and Management,” Annual Review of Psychology, 43, 133–68. Snyder, Mark (1974), “Self-Monitoring of Expressive Behavior,” Journal of Personality and Social Psychology, 30 (4), 526–37. ——— (1987), Public Appearances/Public Realities: The Psychology of Self-Monitoring. New York: Freeman. Spence, Janet T., Robert L. Helmreich, and Carole K. Holahan (1979), “Negative and Positive Components of Psychological Masculinity and Femininity and Their Relationships to SelfReports of Neurotic and Acting Out Behaviors,” Journal of Personality and Social Psychology, 37 (10), 1673–82. Stinson, Linda and William Ickes (1992), “Empathic Accuracy in the Interactions of Male Friends and Male Strangers,” Journal of Personality and Social Psychology, 62 (5), 787–97. Urbany, Joel E., Peter R. Dickson, and William L. Wilkie (1989), “Buyer Uncertainty and Information Search,” Journal of Consumer Research, 16 (2), 208–215. White, Michael J. and Lawrence H. Gerstein (1987), “Helping: The Influence of Anticipated Social Sanctions and Self-Monitoring,” Journal of Personality, 55 (1), 41–54. Wiggins, Jerry S. (1991), “Agency and Communion as Conceptual Coordinates for the Understanding and Measurement of Interpersonal Behavior,” in Thinking Clearly About Psychology, William M. Grove and Dante Cicchetti, eds. Minneapolis: University of Minnesota Press, 89–113. Winterich, Karen Page, Vikas Mittal, and William T. Ross Jr. (2009), “Donation Behavior Toward In-Groups and OutGroups: The Role of Gender and Moral Identity,” Journal of Consumer Research, 36 (2), 199–214. Wolfe, Raymond N., Richard D. Lennox, and Brian L. Cutler (1986), “Getting Along and Getting Ahead: Empirical Support for a Theory of Protective and Acquisitive Self-Presentation,” Journal of Personality and Social Psychology, 50 (2), 356–61. Yang, Sha and Greg M. Allenby (2003), “Modeling Interdependent Consumer Preferences,” Journal of Marketing Research, 40 (August), 282–94. 754 Journal of marKeting reSearCh, auguSt 2011 Copyright of Journal of Marketing Research (JMR) is the property of American Marketing Association 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.