Consumer Behavior V Scholarly Activity

The Effect of Group Membership Salience on the Avoidance of Negative Outcomes: Implications for Social and Consumer Decisions

DONNEL A. BRILEY ROBERT S. WYER, JR.

Calling consumers' attention to their cultural Identity can make them aware of their membership in a group and, therefore, can induce a group mind-set. This mind- set, in furn, leads fhem to make decisions that minimize the risk of negative out- comes to both themselves and others. The effects of this mind-set generalize over both group and individual choice situations. These possibilities were confirmed in a series of six experiments. Results showed that making people feel part of an ad hoc group increased not only their use of equality as a basts for allocating resources to themselves and others, but also their tendency to compromise in individual consumer choice situations. Moreover, calling Asian and Western participants' attention to their cultural identity also induced feelings of being part of a group and, as a result, had analogous effects on decisions in both group and consumer choice situations.

R esearch and theory on consumer behavior must takeinto account the way in which individuals" cultural backgrounds influence their judgments and decisions. An understanding of this influence, however, is complicated by a number of factors. For one thing, cultural differences in behavior may only be apparent when the culture-related norms and values that bear on a decision to engage in this behavior are accessible in memory at the time the decision is made (Aaker and Lee 2001; Briley, Morris, and Simon- son 2000; see also Briley and Wyer 2001; Hong et al. 2000). This culture-related knowledge is mo.st likely to have an influence when people are not clearly aware of the factors that have led it to come to mind. In fact, calling people's cultural identity to their attention could have effects that override the influence of culture-related norms and vai- ues. Moreover, these effects may be similar across individ- uals regardless of the specific culture that they represent.

The present research examined this possibility. Specif- ically, we hypothesized that calling individuals' attention to their national or cultural identity (like making them

*Donnel A. Briley is assisiani professor of ni^irketing und Ruben S. Wyer. Jr. i.s professor of tnarketing, both al the Hong Konj; University of Science and Technology. School of Business. Clear Water Bay. Kowlotm, Hong Kong (e-mails: [email protected]. mkwyer(?'ust.hk). This research was suppnneti by grants from the Hong Kong government (RGC HKUST fiO22/(H)H and DAG (X)/Ol BM67).

aware of their membership in a more circumscribed group) can stimulate them to adopt a group mind-set. This mind- set, in turn, may induce a prevention focus (Higgins 1997, 1998), that is. a disposition to minimize the negative con- sequences of decisions with little consideration given to possible benefits. This motivational disposition contrasts with a promotion focus (i.e.. a disposition to emphasize the potential benefits of a decision rather than its costs). Prevention and promotion goals are conceptualized by Higgins and his colleagues as "representational structures that guide the system in its pursuit of a reference or end state" (Markman and Brendl 2000. p. 98). These goal ori- entations, once activated, can generalize over situations and decision domains (Liberman et al. 1999). That is, they may influence not only behavior in interpersonal .situations but also individual consumer decisions in which others are not involved.

An understanding of the influence of a prevention focus is of general importance in analyzing the effects of adver- ti.senients and otber marketing communications. These com- munications often activate concepts associated with inter- personal relations and social affiliation. If these commu- nications induce a prevention focus, they could influence purchase decisions for reasons other than tho.se that mar- keters assume. Therefore, the impact of a group mind-set on decision processes, and the extent to which people's

400

2002 by JOURNAL OF CONSl'MER RESEARCH. Inc. • Vol, 2') • December 2(X)2

All righis reserved. lX)93-53t}l/2003/2y(l.1-(KH)8SIO.OO

EFFECT OF GROUP MEMBERSHIP SALIENCE 401

awareness of their cultural identity induces this mind-set, are well worth investigating.

EFFECTS OF GROUP MIND-SET ON JUDGMENTS AND DECISIONS

Effects of Group Membership on Concern with Negative Outcomes

People who become aware of their membership in a group are likely to feel a sense of responsibility to its members (both themselves and others). In principle, these feelings could be manifested in behavior that either en- hances the group and its members or prevents them from encountering adversity. However, people are often more concerned about negative consequences of their behavioral decisions than about positive consequences (loss aversion; Kahneman. Knetsch. and Thaler 1991). and this bias is likely to be magnified when decisions affect other persons as well as themselves. This is because negative outcomes of group-relevant decisions (e.g., failure, disgrace and em- barrassment) can erode group cohesiveness. Furthermore, decisions that have more unfavorable consequences for some members than others could create conflict and dis- harmony. For these reasons, feelings of group membership are likely to induce a prevention focus of the sort noted earlier. More specitically. the.se feelings may increase cau- tiousness and cause a stronger tendency to avoid making decisions that could have negative consequences for one- self and other members.

Aaker and Lee (2001) confirmed the assumption that peo- ple are more concerned about the negative consequences of behavior when they think of themselves as members of a group. Participants imagined themselves as the protagonist in an advertising scenario about the finals of a tennis tour- nament. The scenario referred to the protagonist as either an individual or a member of a team and was worded in a way that emphasized either the desirability of winning or the undesirability of losing. Later, some participants recalled aspects of the scenarios they had read, whereas others es- timated their liking for the tennis racquet that was ostensibly being advertised. When participants had been induced to think of themselves as individuals, they recalled relatively more aspects of the story, and evaluated the product more favorably, when the story emphasized winning. When par- ticipants had been stimulated to think of themselves as mem- bers of a team, however, they retained more information and made more favorable product evaluations when the story focused on the possibility of losing. This suggests that stim- ulating people to think of themselves as part of a group heightens their relative concern about negative conse- quences of behavior.

Aaker and Lee's findings provide the basis for the fol- lowing postulate:

P I : Feelings of membership in a group increase one's concern with the negative consequences of one's behavior.

Effects of a Prevention Eocus on Judgments and Decisions

To reiterate, a prevention focus is conceptualized by Hig- gins (1997, 1998) as a general motivational disposition that mediates judgments and decisions in hoth interpersonal sit- uations (e.g., Higgins et al. 1994) and individual choice tasks (e.g., Crowe and Higgins 1997: Roney. Higgins. and Shah 1995). Its effects, which may be either chronic or situa- tionally induced, can be conceptualized in terms suggested by Anderson (1983). In particular, a concern with negative outcomes could activate a set of behavioral decision strat- egies (e.g., cognitive "•productions") that are employed with minimal cognitive mediation under conditions in which they are applicable. Productions theoretically have the form of "If {X}. then (Y)" rules, where jX} is a set of situational and internally generated features and {Y| is a sequence of behaviors that are spontaneously activated when the eliciting conditions are met. The conditions that elicit a production can include both intemally generated motivational concepts (e.g., "avoid negative outcomes") and aspects of the specific situation in which one finds oneself, and the two sets of features in combination may be both necessary and sufficient to activate the production. Thus, the same motivational con- cept, in combination with different sets of situational fea- tures, could activate quite different behavioral sequences, depending on the situation at hand.

Two sets of studies, each in a different content domain, provide compelling examples of this difference. First, col- lege-age participants in a study by Bargh. Chen, and Bur- rows (1996. experiment 2) performed a sentence-construc- tion task that unobtrusively activated concepts associated with the elderly. These participants walked more slowly to the elevator after leaving the experiment than participants who had not performed this task. In a different situation, furthermore, subliminally priming similar elderly related concepts led participants to perform less well on a memory task. As a second example, Bargh et al. (1996, experiment 3) found that subliminally exposing Caucasian participants to faces of African-Americans while performing a boring task led them to manifest more nonverbal expressions of hostility and irritation when they were asked to repeat the task later in the study. However, an identical procedure led participants in a different situation to perform less well on a mathematics test (Colcombe and Wyer 2001). Thus, ac- tivating a concept of African-Americans (who are stereo- typed as both hostile and aggressive and as unmotivated to do well in academic situations) had quite different effects, depending on the situation at hand.'

'in some situations, a prevention focus might also activate a general semantic concepi (e.g.. "negative consequences are bad and should be avoided") that, once activated, is later useii to interpret other situations lo which it is applicable. The processes that underlie these effects and their general inability over situations may be similar to (hose that more generally govern the effects ot" priming semantic concepts in one situation on judg- ments and behavior in other, quite-unrelated situations (for alternalive the- oreiical formulations of these processes and summanes of results bearing on them, see Bargh 1994; Higgins 1996; Wyer and Sruil 1989).

402 JOURNAL OF CONSUMER RESEARCH

The aforementioned studies do not bear directly on the generalizability of the disposition to avoid negative out- comes. However, Liberman et al. (1999) found evidence that this is the case, ln their study, some participants were asked to write about their hopes for the future, while others wrote about their concerns. On the basis of theory and research by Higgins et al. (1994). the authors assumed that writing about hopes would activate a concern with positive out- comes (a promotion focus). wherea.s writing about concerns would stimulate them to think about avoiding negative out- comes (a prevention focus). Subsequently, participants com- pleted a supposedly unrelated task in which they were asked to imagine that they had received a $5 mug (pen) as a gift and could exchange it for a pen (mug) of similar value. The authors predicted that the prevention focus induced by writ- ing about concerns would increase participants' sensitivity to los.ses and, con.sequently. deter them from giving up their gift. The results confirmed this prediction. Forty-four per- cent of participants who wrote about their goals exchanged their gift, whereas only 19% of those who wrote about con- cerns were willing to do so. The generalizability of a pre- vention focus over decision situations demonstrated by Lib- erman et al. is consistent with other conceptualizations of decision making, which also assume that people often ac- quire general guidelines that they apply to decisions in a variety of different domains (Briley et al. 2000; Shafir, Si- monson, and Tversky 1993). These considerations justify a second postulate;

P2: Once a prevention focus (i.e., a disposition to avoid the risk of negative consequences) is activated in one situation, its effects will generalize to other group and individual choice situations that are un- related to the one in which the focus was induced.

Manifestations of Aversion to Negative Outcomes in Group and Individual Decisions

In the present article, a prevention focus is assumed to be reflected in {a) a preference for equality in interpersonal choice situations, Ih) a tendency to compromise in individual choice situations (cf. Briley et al. 2000). and (c) a tendency in multiple-choice situations to distribute choices over mul- tiple alternatives rather than choosing a single altemative. The basis for these assumptions may require elaboration.

Preference fur Equality. Suppose people are con- fronted with a decision that either (a) benefits one individual at the expense of another or (h) treats all parties equally. The first option could create friction among group members and give rise to resentment. In contrast, the second option is more likely to leave both parties reasonably satisfied and to minimize the social costs that might otherwise be in- curred. Individuals who are motivated by a desire to min- imize the negative consequences of their decision are likely to prefer the second, compromise alternative to the first. Work by Simonson and Tversky (1992; see also Tversky and Simonson 1993) supports this contention. They suggest

that decision makers often seek compromise options be- cause, due to loss aversion (Kahneman et al. 1991). they anticipate the regret they will feel if they choose an option that they later discover is not optimal. Regret anticipation, which reflects a concern about the potential negative con- sequences of the deci.sion. leads individuals to view options in terms of the relative dissatisfaction each would make them feel. The selection of a compromise alternative is attractive because it reduces the maximum dissatisfaction one can in- cur. It follows that increasing one's concern about negative outcomes will increase their tendency to prefer compromise solutions to decision problems. Note that this rationale ap- plies to the other two tasks we examine in this article, dis- cussed below, as they also can be framed as a decision between compromise and noncompromise alternatives.

Compromise in Consumer Choice Situations. Sup- pose consumers are confronted with a choice between (a) a product with a very favorable value along one attribute dimension and a very unfavorable value along a second dimension and (h) a product with moderate values along both dimensions. Assuming that all attribute dimensions are similar in importance, people who are guided by the positive consequences of their decision might generally prefer the first alternative, though those with a desire to minimize neg- ative con.sequences of their choice should prefer the second, compromise option.

Decisions in Multiple Choice Situations. Suppose consumers are offered two equally attractive types of candy and can choose cither (a) two candies of the same type or ib) one candy of each type, in such a situation, they may be uncertain of which candy they prefer. If consumers are motivated by the desire to maximize positive consequences of their choice and have a slight preference for one candy over the other, they should choose two candies of the pre- ferred type. However, consumers with a disposition to avoid negative consequences of their decision should focus on what they stand to lose rather than what they might gain. Because the choice of two candies of the same type would incur a loss of the attractive features of the alternative (not chosen) candy, they are likely to compromise by choosing one candy of each type. (In addition, these participants might be concerned about postdecisional regret from a second source—their own preferences changing over time—and might distribute their choices over candy types for this rea- son as well.)

Other factors might also contribute to decisions in the situations just described. Nevertheless, it seems reasonable to assume that a general tendency to minimize the likelihood of negative consequences of one's choice would be reflected in decisions in all three situations. On the basis of this as- sumption and postulates 1 and 2. we derived the following two hypotheses, which we evaluated in the first two exper- iments to be reported:

H I : Inducing feelings of group membership will in- crease preferences for equality in interpersonal situations.

EFFECT OF GROUP MEMBERSHIP SALIENCE 403

H2: Inducing feelings of group membership will in- crease the tendency to distribute choices over al- ternatives in a multiple-choice situation in which only oneself is involved.

EFFECTS OF CULTURAL SALIENCE ON JUDGMENTS AND DECISIONS

Cultural differences in the norms, values, and standards of behavior that guide people's judgments and decisions have often been conceptualized in terms of individualism and collectivism (Hofstede 1980; Triandis 1989). Individ- ualism, which is common in Western cultures, is character- ized by a tendency to view oneself as an independent being and to pursue personal goals without considering others' interests. Collectivism, which is more typical of Asian so- cieties, is often reflected in a disposition to think of oneself as a member of a group or collective and to evaluate one's own attributes and behavioral outcomes in relation to oth- ers'. (For a similar distinction, see Markus and Kitayama 1991.)

When Asians and Westerners are not conscious of their cultural identity, the norms and values that predominate in their society may influence the decisions they make, and this influence may occur without awareness. (For discus- sions of the nonconscious influences of chronically acces- sible bodies of knowledge on judgments and behavior, see Bargb 1997. 2002; Higgins 1996.) Members of collectivist culture may be generally more prevention focused than members of individualistic societies (Lee, Aaker. and Gard- ner 2000) and consequently may be relatively more disposed to compromise in both interpersonal and individual choice situations. As Hong and Chiu (2001) point out. however, culture-specific norms and values are not the only criteria on which people base judgments and decisions and may not be applied if other relevant criteria arc more salient. To this extent, calling people's attention to their cultural or national origin may induce feelings of group identity that override other effects that culture-related knowledge might otherwise have. In this regard. Druckman (1994) notes that one's cul- tural or national identity is often a central part of one's self- definition, and awareness of this identity can induce feelings of allegiance to one's country and a desire to promote or maintain its interests. Moreover, this may be true regardless of the individualistic or collectivist orientation of the country to which one belongs (Davis 1999).

To this extent, calling people's attention to their cultural or national identity may make both Asians and Westerners conscious of their membership in a group and, therefore, may induce a prevention focus similar to that induced by membership in a more circumscribed group. Under these conditions, then, people are likely to minimize the negative consequences of their decisions in both interpersonal situ- ations and individual choice situations. To formalize these predictions:

H3: Making participants aware of their cultural identity

will increase their preference for equality in in- terpersonal situations.

H4: Making participants aware of their cultural identity will increase their preference for products that will decrease the risk of negative consequences in con- sumer choice situations.

Note that these effects differ from those one might predict on the basis of cultural differences in norms and values per se. As we noted earlier, Asians are generally inclined to think of themselves as part of a group, whereas Westerners are more disposed to think of themselves as individuals (Markus and Kitayama 1991; Triandis 1989). Thus, in the absence of factors that call attention to their cultural identity, Asians may be generally more sensitive to negative con- sequences of their behavior than Americans are. This dif- ference was reported by Aaker and Lee (2001). who found that when participants" cultural identity was not salient to them. Chinese were generally more iikely than Americans to remember aspects of a scenario that focused on losing and were less likely than Americans to remember aspects of a scenario that emphasized winning. However, when peo- ple are consciously aware of their identity as members of the culture they represent, they might have similar concems about negative consequences regardless of their culture.

OVERVIEW

The six experiments to be reported and their implications for the conceptualization we propose can be summarized with reference to figure 1. The solid lines in figure IA show the hypothetical causal relations between feelings of group membership, the adoption of a prevention focus, and both preferences for equality (in interpersonal situations) and compromise-oriented alternatives (in individual judgment situations). Experiments 1 and 2 established the observed effects of inducing feelings of group membership on these preferences, as indicated by the dashed lines in the figure. However. Aaker and Lee's (2001) findings establish the re- lation between feelings of group membership and a pre- vention focus, and experiment 3 establishes that a direct manipulation of emphasis on positive versus negative out- comes influences preferences in the two types of situations we examine in the first two experiments. Moreover, this experiment further shows that the effects of feelings of group membership on these preferences were eliminated when pre- vention focus was experimentally controlled.

Hypotheses 3 and 4 assume that making people aware of their cultural identity induces feelings of group membership. This assumption is confirmed in experiment 4 along with findings by Adaval (2001) and Gardner, Gabriel, and Lee (1999). If this assumption is valid, the effects of making salient participants' cultural identity should be indicated by the solid paths in figure IB. Experiments 5 and 6. which employed both European-American and Hong Kong Chi- nese participants, established the effects of making salient participants' cultural identity on both preferences for equal-

404 JOURNAL OF CONSUMER RESEARCH

FIGURE 1

SUMMARY OF PREDICTED (OR PREVIOUSLY SHOWN) RELATIONSHIPS

A. Group membership effects

AWARENESS OF GROUP

MEMBERSHIP

AAKBt&L£E DESIRE TO AVOID NEG. OUTCOMES

EXP2

PREFERENCE FOR EQUALITY

(proverb endorsement)

COMPROMISING IN INDIVIDUAL CHOICE

TASKS (choosing different vs.

same brands)

B. Cultural identity awareness effects

EXP5

AWARENESS OF CULTURAL IDENTITY

EXP6 DESIRE TO AVOID NEG. OUTCOMES

PREFERENCE FOR EQUALITY

(money allocation)

COMPROMISING IN INDIVIDUAL CHOICE

TASKS (compromise altematives)

ity and compromise choices in a product evaluation ta.sk. as indicated by dashed lines in the figure. However, experiment 6 confirmed thai [a) making participants' cuiturai identity salient induces a concem with negative consequences of their choices (e.g.. a prevention focus), (b) a concern with negative consequence.s leads to a preference for compromise options, and ic) the effect on choice behavior of making cultural identity salient is mediated by thoughts about the negative consequences of decisions. These conclusions and their implications are discussed in the following sections.

EXPERIMENT 1: EEFECTS OF GROUP MEMBERSHIP ON USE OF EQUALITY-

ORIENTED DECISION PRINCIPLES

Hypothesis I suggests that making people feel they belong to a group induces a di.sposition to avoid negative outcomes to both self and others and that this will be manifested in a preference for equality in interpersonal situations. More-

over, this disposition may generalize to situations other than that in which feelings of group membership were induced (postulale 2; see Higgins 1998; Liberman et al. 1999).

Experiment 1 evaluated this hypothesis. We experimen- tally manipulated participants' feelings of group member- ship by giving them the expectation that they would perform an experimental ta.sk either as individuals or as part of a group. Then, after inducing this expectation, we examined their endorsement of proverbs (hat advocate equality or bal- ance in group situations. Although this measure was not a direct index of decision making, other research provides evidence that people's agreement with proverbs does in fact reflect the values they draw on to make behavioral decisions (Briley et al. 2000; Weber. H.see. and Sokolowska 1998).

Method

Forty-six Chine.se undergraduates at a Hong Kong uni- versity participated in the study for HKSIOO (US$12.82).

EFFECT OF GROUP MEMBERSHIP SALIENCE 405

Subjects were toid that the experiment was intended to test the thinking skills of college students and that to do this, they would be performing a 22-item anagram task. They were randomly assigned to either a group- or individual- focus condition. Those in group-foeus conditions were told they would be performing the task in groups and, on this pretense, were assigned seats at different five-person tables. To encourage cohesiveness. the participants at each table decided on a group name that they would use to identify themselves on their answer sheets. In contrast, participants in individual-focus conditions were told they would be per- forming the task individually and were seated at single- person desks separated by partitions.

After being given these instructions., participants in both conditions were told that before performing the anagram task, they would be asked to complete some other, unrelated questionnaires as part of a different experiment. They were informed that in the first task, they would be asked to read 32 proverbs from different countries and to indicate the extent to which they endorsed them. Five of these proverbs emphasized the importance of balance and equity in social relations: (a) "a single hand cannot clap" (implying that action requires a collective effort), (b) "it takes two to make a quarrel" (both people who quarrel share responsibility for it), ic) "when the shoulder pole is not secure at both ends, the load will slip off (people working together must be responsible for their part), (d) "the pole is easy to carry if the load is balanced" (balancing the work over individuals makes a task easier), and ie) "rivers have two banks and every issue has two sides" (there are two sides to every controversy).

Participants read each proverb and then indicated (a) how much ihey liked the idea of the proverb, ih) how much they would rely on the proverb as a guide in making decisions. and (c) the extent to which they would draw on the proverb as a basis for giving advice. These judgments, each reported along a scale from I (not at all) to 7 (very much), were averaged for each proverb. A homogeneity analysis of re- sponses to the five proverbs assumed to reflect the value attached to equality and balance yielded a coefficient alpha of .82. Responses to the five proverbs were then averaged to yield a single index of each participant".s endorsement of the equal ity-related proverbs.

Results

As we expected, participants endorsed equality-related proverbs more strongly when they had been led to believe they were participating in groups (A/ = 5.60) than when they believed they were taking part as individuals (A/ = 5.07, F(l,44) = 4.4l,/7<.05). Thus, these results support hypothesis 1. that people who are induced to feel part of a group take on a group mind-set and. therefore, are likely to favor equality in interpersonal relations. (An alternative in- terpretation of these results will be evaluated presently.)

EXPERIMENT 2: EFFECTS OF GROUP MEMBERSHIP ON CONSUMER DECISION

MAKING

The prevention focus that is stimulated by consciousness of one's group membership should generalize to consumer decision situations in which only one's own outcomes are affected. Thus, according to hypothesis 2, the equality-re- lated decision strategy that was prompted by feelings of group membership is applied in individual decision situa- tions as well. The next experiment explored this possibility. As in experiment 1. some participants believed they were participating in the experiment as part of a group, whereas others believed they were participating as individuals. Later, on leaving the experiment, they were given an opportunity to choose two candies to take home with them in return for participating. Two types of candy were available. Therefore, participants could choose two candies of the same type or. alternatively, one of each type. For reasons outlined earlier, we assumed that participants who were motivated by a desire to minimize the negative consequences of their choice would be more inclined to choose one candy of each type. Ac- cording to hypothesis 2. this tendency should be greater among participants wbo had participated in the experiment as part of a group than among those who had participated as individuals.

Method

Forty-three Hong Kong undergraduate students were paid HKS100 (USS12.82) to participate. As in experiment 1. they were assigned randomly to either an individual-focus or group-focus conditit>n and told that they would be com- pleting an anagram task to test their thinking skills. The same manipulations used in experiment 1 were applied in the present study. In addition, to encourage cohesiveness in the group-focus condition, participants were told that the group's performance on the anagram task would be eval- uated as a whole, ln both conditions participants were in- structed to work individually and not talk to one another while performing the task.

Participants were then given five minutes to complete a 22-item anagram task. When time was up. they were told that the experiment was over. However, they were told that in addition to the money they were being paid for partici- pating, they could have two pieces of candy. Two types of candy that are popular among students and for which they have similar preferences (Nestle's Crunch and Cadbury Timeout)- were placed in separate bowls on a table near the exit. Each participant was paid individually and. after re- ceiving payment, left the room, taking two pieces of candy on the way out. Participants' candy selections were observed through a one-way mirror, and the nature of each partici-

-ln a preiest 48 subjects were offered a choice between these two brands. Forty-six percent chose Nestle's Crunch, and 54'Ji- chose Cadbury Timeout (z = . 5 9 1 , p>.201

406 JOURNAL OF CONSUMER RESEARCH

pant's choice (two candies of the same kind, or one of each kind) was recorded.

Results and Discussion

As predicted, participants were more likely to choose one candy of each brand when they had participated in a group (.73) than when they had participated individually (.45; X^(l) = 3.74./?< .05). This difference is particularly note- worthy in light of the fact that participants believed that the experiment was over at the time they made their choices and had no reason to believe that these choices were being monitored. These results therefore support our contention that people with a group mind-set are more likely to seek balance and compromise in their choices than are those with an individual mind-set.

EXPERIMENT 3: EEFECTS OE REWARD AND PUNISHMENT INCENTIVES ON

DECISIONS The results of the first two experiments are consistent

with hypotheses I and 2. To establish the validity of the assumptions underlying these hypotheses, however, we must show that ia) a more direct manipulation of participants concern with positive versus negative outcomes will affect participants' preferences for equality and choice behavior in the same way that feelings of group membership affect them and ib) the effects of feelings of group membership are eliminated when concern with these alternative outcomes is experimentally controlled. Experiment 3 attained these objectives.

The procedure we employed in this study was similar to that used in experiment 2. That is, participants were toid they would perform an anagram task either as individuals or as members of a group. However, some participants were led to believe they would receive a reward for doing well, whereas others were told they would be penalized for doing poorly. In anticipation of performing the task, participants completed the proverb-evaluation questionnaire adminis- tered in experiment I and then, on completing the anagram task and leaving the experiment, were given an opportunity to choose candies under conditions used in experiment 2.

We expected that participants would be more likely to endorse proverbs that advocated equality and more likely to diversify their candy choices when penalties for poor performance rather than rewards for good performance were emphasized. Moreover, we expected that a direct manipu- lation of the importance of positive versus negative out- comes would decrease or eliminate the less direct effects of group membership salience on participants' judgments and decisions.

Method

Ninety-three Hong Kong undergraduates, who partici- pated for course credit, were told they would perform a 22- item anagram task under either individual- or group-focus

TABLE 1

MEAN EVALUATIONS OF EQUALITY-ORIENTED PROVERBS AND PROPORTIONS OF PARTICIPANTS WHO CHOSE

DIFFERENT BRANDS OF CANDY AS A FUNCTION OF GROUP VERSUS INDIVIDUAL FOCUS AND OUTCOME FOCUS-

EXPERIMENTS 1-3

Endorsement of equality- oriented proverbs:

Individual focus Group focus Overall

Proportion of partioipants who chose different brands of candy:

Individual foous Group focus Overall

Experiments 1 and 2

(no incentives)

5.07 5.60 5.34

.45

.73

.61

Experiment 3

Reward

5.08 5.07 5.07

.50

.35

.43

Punishment

5.49 5.60 5.55

.59

.81

.71

conditions identical to those employed in experiment 2. In addition, we induced a concern with either positive or neg- ative outcomes using a procedure similar to that employed by Roney et al. (1995). In reward conditions, participants were informed that the individual/group completing the most anagrams would receive HK$IO0/$5O0 (US$I2.82/$64.IO). In punishment conditions, they were told that individuals/ groups wouid receive candy for participating, except those that completed significantly fewer anagrams than average.

After receiving these instructions, however, participants were told that before performing the anagram task, we wanted them to complete an unrelated questionnaire as a part of a different experiment. On this pretense, they were administered the same 32-item proverb endorsement ques- tionnaire used in experiment 1. Then, they completed the anagram task. Participants under reward conditions were then told that they would be notified later if they had won, and those in punishment conditions were told they had done well enough to receive candy. Then, all participants were dismissed, making candy selections as they left according to procedures described in experiment 2.

Results and Discussion

Participants' evaluations of equality-oriented proverbs and their likelihood of choosing variety in their candy se- lections are shown in table 1 as a function of outcome focus (reward vs. punishment) and group versus individual task conditions. Each set of data is discussed in tum.

EFFECT OF GROUP MEMBERSHIP SALIENCE 407

Proverb Endorsement.'^ We expected that participants would evaluate proverbs that endorsed equality more pos- itively under punishment conditions than under reward con- ditions. Moreover, we speculated that this direct manipu- lation of outcome focus would override the effects of inducing a group mind-set on these evaluations. This was in fact the case. Participants judged equality-oriented prov- erbs more useful as decision guides under punishment con- ditions (M = 5.55) than under reward conditions (M — 5.07;F(l,89) = 4.84./? < .05). However, the effect of group (vs. individual) focus was not evident (F< 1); participants' evaluations of equality-oriented proverbs were virtually identical regardless of whether they participated as a group or as individuals (5.33 vs. 5.28). The interaction of group/ individual focus and outcome focus also did not approach significance ( F < I).

Candy Preferences. Participants were more likely to choose different types of candy in punishment conditions (.71) than in reward conditions (.43). In contrast, the pro- portions of individuals who chose variety in group- and individual-focus conditions were virtually identical (.57 vs. .53, respectively). These conclusions are confirmed by a logistic regression analysis, which indicated that the effect of outcome focus was quite reliable (x" = 6.50, p< .01), whereas the main effect of group versus individual focus did not even approach significance ( x ' < 0- Interestingly, the interaction term was significant (x" = 7.46, /;<.O1), reflecting the relatively high tendency of subjects in the group-focus/punishment conditions to choose candies of dif- ferent brands (.81; see table I).

The.se results indicate that providing explicit incentives that lead persons to focus on positive versus negative out- comes influences candy choice variety in the way we would expect and, moreover, overrides the less direct effect of a group versus individual mind-set that is apparent in the ab- sence of these incentives. A comparison of the data obtained in experiment 3 with those of experiments 1 and 2 (see table 1) indicates that the mean preference for equality oriented proverbs under reward conditions of experiment 3 is vir- tually identical to that observed in individual-focus condi- tions of experiment I (5.07 in both cases), whereas the mean preference for these proverbs under punishment conditions is very similar to that obtained in group-focus conditions of the first experiment (5.55 vs. 5.60, respectively). Corre- spondingly, the proportion of participants who chose two candies in punishment conditions of the present study (.71) is very similar to the proportion who did so in group-focus conditions of experiment 2 (.73), and the proportion who chose two candies under reward conditions (.43) is similar to the proportion who did so under individual-focus con- ditions of the earlier experiment (.45). Thus in the earlier

'Participants attempted more anagrams in reward than punishment con- ditions (7.34 vs. 6.47, r ( l , 8 9 ) = 2.09, / J > . I O ) . Though this difference did nol reach significance, the direction is consistent wilh Roney et al.'s (1995) finding that task persistence is greater for individuals who are prompted to use approach vs. avoidance strategies.

experiment where explicit incentives were absent, partici- pants under group conditions behaved as if they had a pun- ishment focus, whereas those under individual conditions behaved as if they had been given a reward focus. However, explicit incentives to focus on reward or punishment de- creased or eliminated any effects that a group or individual focus might otherwise have had.

An alternative interpretation of the results of experiment 1 should be noted in this regard. That is, group membership might induce a more general equality norm that influences the endorsement of proverbs independently of a concern about negative outcomes per se. That is. the effects of group versus individual focus in experiment 1 and the effects of punishment versus reward in the present study could con- ceivably be due to different factors. However, note that group versus individual focus was manipulated in the present study as well as in the first experiment. If group focus in- duces an equality norm that influences proverb endorsement independently of its impact on concem with negative out- comes, it should have had an effect in the present study over and above the effects of reward versus punishment. There was obviously no evidence of this; judgments were virtually identical in group- versus individual-focus condi- tions regardless of whether reward or punishment was em- phasized. Thus, any effect of group focus on preference for equality was apparently induced by a concern about negative outcomes, and this effect was eliminated when this concern was induced more directly.

It is of course conceivable that participants based their preferences on reward and punishment considerations in ex- periment 3, where these outcomes were called to their at- tention, but on a more general equality norm in experiment 1. when group membership was relatively more prominent. This interpretation, however, assumes that group member- ship does not in itself increase attention to negative con- sequences, and Aaker and Lee's (2001) findings argue against this assumption. Finally, the assumption that the effect of group membership is mediated by its impact on the salience of a more general equality norm could not easily account for its effects on candy selections. In combination, therefore, experiments 1-3 provide good support for both hypotheses 1 and 2 and the assumptions underlying them.

EXPERIMENT 4: EEFECTS OE CULTURAL AWARENESS ON CONSCIOUSNESS OF

GROUP MEMBERSHIP

According to hypotheses 3 and 4, making persons aware of their cultural identity should have effects similar to the effects of group membership observed in experiments I and 2. These hypotheses, however, are based on the assumption that calling people's attention to their cultural identity would induce feelings of belonging to a group and, therefore, a concern with the consequences of one's behavior in group situations. This assumption is consistent with findings by Druckman (1994) and Davis (1999), as noted earlier. It is nevertheless desirable to confirm the assumption using the

408 JOURNAL OF CONSUMER RESEARCH

same procedure to be used for testing hypotheses 3 and 4 (in experiments 5 and 6) for inducing awareness of cultural identity.

Our hypotheses assume that thoughts about group mem- bership are activated spontaneously by awareness of cultural identity, in the absence of situational demands. We therefore evaluated this assumption in a way that did not require asking participants explicitly for their thoughts about their cultural identity or. for that matter, about group membership more generally. The procedure we used was suggested by studies conducted by Gardner et al. (1999) and Adaval (2001). ln these studies, participants were exposed to either first person singular pronouns i\, me, mine) or first person plural pronouns (we, us, our) in the course of performing an ostensibly irrelevant reading task. Gardner et al. (1999) found that participants in the second condition were more likely to describe themselves on the basis of their group memberships and roles (e.g., "I am a sister in Kappa Kappa Gamma" [p. 322]) and endorse group oriented values (e.g., belongingness, family safety, national security).

In Adaval's study, participants performed a sentence con- struction task that required the use of either first person singular or first person plural pronouns. After doing so, they were exposed to one of two ads for Pepsi-Cola. The ads. wbich showed cartoon pictures of different individuals (a teenager, a baseball player, a policeman, etc.), were identical except for the banner head. In some cases, the banner read "Stand out from the crowd!" and in other cases, it read "Don't stand out from the crowd!" Participants were asked to read the ad and to report their liking for it along a scale from - 3 to +3. Participants who had been primed to use first person singular pronouns reported liking the first ("Stand out . . .") ad more than the second ("Don't stand o u t . . .") ad (1.56 vs. 0.70). whereas participants who had been primed with first person plural pronouns liked the sec- ond ad more than the fist (1.27 vs. 0.60). This study, in combination with Gardner et al.'s (1999), suggests that per- sons who use the first person plural are more inclined to think of themselves as part of a group and to value group membership more than those who use the first person sin- gular. The present study was based on this assumption.

After first exposing participants to either symbols of their own culture or symbols of a different culture, we asked them to perform a sentence construction task in which they could choose either a first person singular or first person plural pronoun to form the sentences they were given. We expected that participants would be more inclined to use the first person plural pronoun in the first case than the second. If this occurs, and if the use of the first person plural pronoun reflects a tendency to think of oneself as members of a group (Gardner et al. 1999), it would support the assumption that exposure to symbols of one's own culture induces feelings of group membership.

Method

Design and Participants. Forty-five United States and 35 Hong Kong Chinese college students participated for

course credit. Before performing the sentence-construction task, participants were exposed to pictures depicting either American or Chinese cultural icons. Thus, the design was a 2 (cultural icon exposure: American or Chinese) by 2 (participant sample: Chinese or European-American) fac- torial.

Exposure to Cultural Icotis. To increase participants' awareness of their national identity, we used a procedure similar to that employed successfully by Hong et al. (2000). Specifically, participants were told that we were interested in assessing the general knowledge of college students and that as a part of this investigation we wanted to determine individuals' abilities to identify important persons, objects, or events and the time period with which they are primarily associated. On this pretense, participants were given six pic- tures or drawings. In the American icon exposure condition, the pictures portrayed an American flag, a "roaring '20s" dance scene, a Dixieland band, Marilyn Monroe, Superman, and Abraham Lincoln. In the Chinese icon exposure con- dition, they portrayed a Chinese dragon, (he Great Wail, a girl playing a traditional Chinese musical instrument, two persons writing ideographs, an actor from a Chinese opera, and a character in a famous Chinese novel (Journey to the West). Participants in each condition were asked to identify tbe picture's referent and to indicate the approximate period of time in which the referent first existed (or, if fictitious, the time it was first created).

A,ssessment of Group Orientation, To assess their dis- position to think of themselves as individuals or as members of a group, we asked participants to complete a 32-item sentence-construction questionnaire. Each item consisted of five randomly ordered words, and participants were told to underline four of the words that could be used to construct a meaningful sentence. Two sentences could be formed from each set. In 16 of the items, one possible sentence required use of a first-person plural pronoun (e.g., we, us, ours, etc.), and tbe other required use of a first-person singular pronoun (e.g.. I, me, mine, etc.), for exatnpie,

to go work we I.

We uesd the proportion of these items in which participants selected the first person plural pronoun to construct a sen- tence as an index of their group versus individual focus.

Results and Discussion

The proportion of items in which participants used a plural first-person pronoun to construct a sentence was computed for each participant and analyzed as a function of partici- pants" country of origin (Hong Kong vs. United States) and cuiturai exposure condition (same vs. different culture). The influence of exposure condition was significant (F(l,70) ^ 9.12,/?< .01) and is attributable to the fact that participants selected a greater percentage of first-person plu- ral pronouns when they had been exposed to symbols of

EFFECT OF GROUP MEMBERSHIP SALIENCE 409

their own culture (M = .32) than when they had been ex- posed to symbols of a different cuhure (M = .23). More- over, this was true for both Chinese participants {.35 vs. .25, F(l,70) = 6 . 0 7 . / K . 0 2 5 ) and Americans (.30 vs. .22, F(],70) = 5.28. /?<.O5). Neither cultural sample {F{ 1. 70) = 1.68. p = .20) nor its interaction with icon ex- posure condition {F< I) reached significance.

This experiment indicated that exposing participants to symbols of their own culture increased their tendency to use first person plural pronouns and that this was true for both Chinese and Americans. If this is so, and if the aforemen- tioned findings by Gardner et al. (1999) and Adaval (2001) are valid, it follows that exposure to symbols of one's own culture should increase feelings of membership in a group. To this extent, it should induce a prevention focus and, therefore, have effects analogous to those observed when participants are conscious of belonging to a more circum- scribed group. Experiments 5 and 6 confirmed this prop- osition.

EXPERIMENT 5: EFFECTS OF CULTURAL AWARENESS ON PREFERENCES FOR

EQUALITY

Experiment I supported the hypothesis that inducing par- ticipants to think of themselves as members of a group increases the value tbey attach to equality, as refleeted in the endorsement of proverbs that emphasize the desirability of this state of affairs. It further confirmed the implications of postulate 2, that feelings of belonging to a group give rise to a mind-set that generalizes over groups and situations. If this is so, and if consciousness of one's national identity increases the tendency to think of oneself as part of a group, it should induce a preference for equality similar to that induced by feelings of group membership in experiment 1.

To examine this possibility, we constructed a resource- allocation situation in which participants' preferences for equality could be more directly evaluated. Both United States and Hong Kong Chinese college students participated. After making tbeir national identity salient to them in the manner employed in experiment 3. we asked representatives of both cultures to imagine that they were responsible for allocating monetary rewards to themselves and a coworker and to indicate their preferences for alternative allocation schemes. These alternatives varied in terms of the equality of the allocations to self and other. Under these conditions, hypothesis 3 implies that making participants' cultural iden- tity salient should increase their tendency to base their al- locations on equality, and this should be true for both Chi- nese and Americans.

Method

Design and Participants. Thirty-seven United States and 55 Hong Kong Chinese students participated for course credit. Before performing tbe resource-allocation task, some participants were exposed to the pictures of either American

TABLE 2

EIGHT POSSIBLE ALLOCATION SCHEMES CONSIDERED PARTICIPANTS: EXPERIMENT 5 ($1,000s)

Self >

High joint

High equality 8, 7 Low equality 9. 6

other

Low joint

7,6 8, 5

Other >

High joint

7, 6,

8 9

self

BY

Low joint

6, 5,

7 8

NOTE.—Monetary award figures for Hong Kong subjects were multiplied by 10 (e.g., the amounts for the first cell above were HK$80,000 and HK$70,000) so that the value ot outcomes was approximately equivalent for Hong Kong and U.S. subjects (1 US$ = 7.8 HK$)-

or Chinese cultural icons used in experiment 4. Thus, the design was a 2 (cuiturai icon exposure: American vs. Chi- nese) by 2 (participant sample: Chinese ur European-Amer- ican) factorial.

Procedure. Participants' consciousness of their cultural identity was induced using the procedure developed by Hong et al. (2000) and employed in experiment 4. That is, some representatives of each cultural group were exposed to American icons, whereas others were exposed to Chinese icons. Then, all participants were given the resource-allo- cation task. To introduce this task, participants were asked to imagine that a firm for which they are working will pro- vide a monetary bonus to themselves and a coworker for their work on a special project. The instructions indicated that the firm was considering several alternalive possibilities for distributing the money and wished to know the alter- native they would prefer. With this preamble, participants were given eight possible allocation schemes and asked to rank them in order of preference.

The allocations composing each altemative scheme, in units of $1,000, are shown in table 2. In each case, the money allocated to the participant is indicated first, and the allocation to the coworker is indicated second. As the table indicates, the eight allocation possibilities compose a three- factor design involving (a) the equality of the allocations to the participant and the coworker, (b) the joint allocation to the two individuals in combination, and (c) whether the participant's allocation was greater or less than the co- worker's. Thus, the difference in allocations was either $1,000 (high equality) or $3,000 (low equality); tbe total amount awarded to the work group was either $ 15,000 (high joint gain) or $13,000 (low joint gain); and the participant could get more or less money than the coworker (positive vs. negative relative gain). Participants were asked to rank order the eight allocation alternatives using 1 to denote the most preferred allocation, 2 to denote the next most pre- ferred, and so on.

Results and Discussion

Subjects' rankings of the eight .schemes were reverse seored so that higher numbers would reflect stronger pref- erences. Mean (reverse) rankings of the four schemes that

410 JOURNAL OF CONSUMER RESEARCH

TABLE 3

MEAN RANKINGS AND UTILITIES OF EACH MOTIVE BY SAMPLE AND EXPOSURE CONDITIONS: EXPERIMENT 5

Sample and exposure condition Equality Joint gains Self relative to other

Rankings: Hong Kong:

Same culture exposure Different culture exposure

United States: Same culture exposure Different culture exposure

Utilities: Hong Kong;

Same culture exposure Different culture exposure

United States: Same culture exposure Different culture exposure

5.25 5.11

5.46 5.06

.71

.57

.72

.45

5.24 4.84

4.77 4.79

.97

.82

.81

.78

6.09 5.84

5.95 5.83

.84

.48

.59

.37

NOTE.—Rankings are reverse scored so Ihal larger numbers indicate higher (more favorable) rankings.

offer high equality (the schemes composing the top row of table 2). high joint gain (table 2, second and fourth columns), and positive relative gain (table 2. first and second columns) each are shown in table 3 as a function of participants' nationality and priming condition. (Rankings of the re- maining four schemes, which are the mirror images of the first four, are not shown.) These means indicate preference for each of the associated motives. Participants reported a stronger preference for equality when they had seen pictures from their own culture (M = 5.36) than when they had seen pictures of a different culture (M = 5.08: F(l. 90) = 3.83. p < .05). Although this difference appears to be somewhat less for Hong Kong participants (5.25 vs. 5.11) than for Americans (5.46 vs. 5.06). the contingency of this difference on cultural background was not at all reliable ( F < 1).

Although the mean ranking of alternatives that reflected high (vs. low) joint gain was somewhat greater when par- ticipants" own culture had been called to their attention (M = 5.01) than when it had not (M = 4.81), this differ- ence was not significant (F(1.90) = 2.60. /7>.IO). Simi- larly, the preference for allocations in which one's own out- comes exceeded the other's also did not depend significantly on whether participants' own culture was made salient to them {M = 6.02) or not {M = 5.83; f ( l , 9 0 ) = 1.02. p> .10).

Because participants' rank orderings of the choice alter- natives were not independent, an analysis of variance of these rankings is nol strictly appropriate. However, a con- joint model yielded similar conclusions. This model predicts preference for allocation schemes using dummy variables for the three motives as predictors (0 = low equality = low joint gain = low self gain, I = high equality = high joint gain = high self gain). The coefficients (utilities) of the model, shown in table 3, estimate the relative influence of the three judgmental criteria on preferences. The Pearson correlation between participants' actual rankings and those predicted by the model is significant (r = .703, p< .05). indicating that the model's fit is acceptable.

The utility (or model coefficient) associated with allo- cation options that offer high (rather than low) equality was greater for subjects who had been exposed to icons from their own culture (M = 0.72) rather than another (M = 0.52; F( I. 90) = 4.34, p < .05). and this difference did not significantly depend on whether participants were from Hong Kong (0.71 vs. 0.57) or the United States (0.72 vs. 0.45; F < I). In contrast, comparable analyses of the utilities as.sociated with joint gain and self gain yielded no significant results whatsoever.

ln summary, experiment 5 confirmed implications of the assumption that exposing participants to symbols of their own culture induced a tendency to think of themselves as part of a group rather than as an individual and, therefore, induced a prevention focus. That is. it increased their relative preference for equality when allocating resources to them- selves and a coworker. This was true regardless of whether the participants were Chinese or American. Hypothesis 3 was therefore supported.

EXPERIMENT 6: EFFECTS OF CULTURAL SALIENCE ON INDIVIDUAL DECISION

MAKING

Experiment 2 demonstrated that subjects' participation in a group influenced their individual choice behavior in a later, ostensibly unrelated situation. According to hypothesis 4. making participants conscious of their national identity should have analogous effects. We explored this possibility in experiment 6. To do so. we employed a choice task in which the implications of participants' choices for the rel- ative importance of positive and negative outcomes could be more directly evaluated.

In this task, people are asked to consider two types of products. One type has a very favorable value along one attribute dimension and a very unfavorable value along a

EFFECT OF GROUP MEMBERSHIP SALIENCE 41

second dimension. The second has moderately desirable val- ues along both dimensions. A person who is primarily mo- livated by the attractiveness of positive features is likely to prefer the first type of product (which has a very desirable feature) to the second. In contrast, an individual who wishes to avoid the cost associated with undesirable features is likely to prefer the second, compromise alternative to the first (Simonson and Tversky 1992).

Cultural differences in choice behavior under these con- ditions were identified by Briley et al. (2000). Under con- ditions in which participants were asked to explain their choices, Americans tended to choose a product of the first type, whereas Chinese often chose the compromise alter- native. These results could suggest that Chinese and North Americans differ in the relative emphasis they place on positive and negative attributes. Results obtained by Aaker and Lee (2001), noted earlier, suggest a similar conclusion.

However, the cultural difference observed by Briley et al. (2000) and Aaker and Lee (2001) occurred under con- ditions in which participants' cultural identity was not ex- plicitly called to their attention. Our theorizing suggests that making participants conscious of their country of or- igin is likely to increase their feelings of group identity regardless of their cultural background and. therefore, to produce a general tendency to avoid the risk of negative outcomes. To this extent, it may increase both Asians' and North Americans' preferences to compromise in the situ- ation examined by Briley et al. (2000). Further, if the in- fluence on compromising is mediated by a tendency to avoid negative choice outcomes, as we propose, we might find some evidence of this by examining the frequency with which subjects mention problems they would like to avoid (rather than features they find compelling) when ex- plaining their selections.

Method

Sixty U.S. university students and 127 Hong Kong Chi- nese university students participated to fulfill a course re- quirement. Participants in each cultural sample were ran- domly assigned to one of three cultural icon exposure conditions (North American, Chinese, or none) using pro- cedures employed in experiments 4 and 5.

Participants were told that they would be completing some unrelated studies. Then, those in American- and Chinese- exposure conditions completed the icon-exposure task used in experiments 4 and 5, whereas participants in no-exposure conditions did not. Finally, ail participants were adminis- tered a choice task similar to that employed by Briley et al. (2000). Specitically. they were toid that we were interested in the reasons that underlie people's preferences for choice altematives and that we wished to examine the choices that people make after they have narrowed potential selections down to a few alternatives that differ along two primary dimensions. On this pretense, participants were given four shopping scenarios, each involving a choice between three products. Each set of products was in a different domain (specifically, personal computers, 35 mm cameras, and

stereo receivers and speakers). Each scenario contained a short description of the category and the features of the available alternatives. The features varied in such a way that participants were faced with a decision between two extreme options (i.e., options with a high value along one attribute dimension and a low vaiue aiong a second dimension) and a compromise alternative (i.e.. an option with moderate val- ues along both dimensions). One scenario, for example, de- scribed three 35 mm cameras as follows:

Typical range Option A Option B Option C

Reliability rating of expert panel

40-70 45 55 65

Maximum autofocus range {meters)

12-28 25 20 15

In each case, participants first wrote a sentence or phrase giving a reason for selecting one option over the others, and then indicated their choice.

Results

Manipulation Check, We assumed that participants are more inclined to think of themselves as group members when their cultural identity has been called to their attention than when it has not. If this is so. they might be less inclined to give self-referent explanations (e.g.. "A is the best com- puter for me because it has lots of RAM") in the former condition. To evaluate this possibility, the proportion of ex- planations that included a self-re fere nee was computed for each participant separately and analyzed as a function of cultural sample and icon exposure conditions (same culture vs. different culture). Participants were clearly less likely to make self-references in their explanations after being ex- posed to icons of their own culture (M = .08) than after being exposed to icons of a different culture (M = .19: f ( l , 112) = 3.85, /) = .05), and this pattern did not sig- nificantly depend on whether participants were from Hong Kong or the United States (F < 1). Thus, these results con- firm the conclusions drawn from experiment 4.

Choice Data. Sixteen participants who failed to pro- vide reasons for their selections were excluded from the analysis. The proportion of scenarios in which the compro- mise option was chosen was computed for each of the re- maining participants. These proportions were analyzed as a function of icon exposures conditions and cultural sample. This analysis revealed a main effect of cultural icon exposure (F(2, 165) = 3.09, p < .05) but no other main effects or interactions (F's < i).

In no-exposure conditions, Hong Kong participants made a greater proportion of compromise choices than U.S. par- ticipants did, though this difference did not reach signifi- cance (.58 vs. .51, F{1,55) = 1.55. p > . 1 0 ) . More impor- tant, planned comparisons indicated participants chose a greater proportion of compromise options when they had been exposed to icons from their own culture (M ~ .63) than when they had been exposed to icons from another

412 JOURNAL OF CONSUMER RESEARCH

culture (M = .50: f ( l . n 2 ) = 5.69. ; J < . 0 5 ) . Moreover, this difference did not significantly depend on whether par- ticipants were from Hong Kong or the United States (F < 1.1).

Mediation Analysis. We assumed that making partici- pants aware of their cultural identity should induce a prevention focus, as reflected by a concem with negative outcomes of their choices. This concem for potential disappointment, then, should mediate the influence on choices of cultural identity awareness. Such concems might be reflected in the explanations partici- pants gave for their choices by references to negative features of choice altematives (e.g., "Option A doesn't have enough hard disk space") rather than positive or neutral ones (e.g., "Option C has the most RAM"). To explore this possibility, a coder who was blind to experimental conditions detemiined whether each reason that participants generated mentioned any concem about perceived negative aspects of any of the alter- natives or their attributes.

Three analyses were performed to examine the relation- ship among priming condition, the content of explanations, and choices, and to test for mediation. First, pooled over participants, we found that negative features were more likely to be mentioned in explanations for compromise choices than explanations for extreme ones (.66 vs. .41, F(l,112) = 4.17. p<.05). This finding supports the as- sumption that compromise choices result from a desire to avoid negative consequences.

In the second analysis, the proportion of explanations that mentioned negative features was computed for each partic- ipant and analyzed as a function of priming condition and cultural sample. Hong Kong participants were (insignifi- cantly) more likely to mention negative features in their explanations than U.S. participants were (.41 vs. .34. F(I. 112) - 1.27, /7>.IO). consistent with finding that a prevention focus is more characteristic of Asian cultures than Westem cultures (Lee et al. 2000). More important, participants were more likely to mention negative features when they had been exposed to symbols of their own culture (.44) than when they had been exposed to symbols of a different culture (.31, F(l. 112) = 7.00./? < .01). and this pattem held for both Hong Kong participants (.47 vs. .34, F(l, 112) = 2.94. p<.Ol) and Americans (.41 vs. .28, F(l, 112) = 7.35, p < . 0 1 ) .

If the effect of exposing participants to symbols of their own culture on compromise choices is mediated by its im- pact on participants' concern with negative outcomes, this effect should be eliminated when differences in this concem are controlled. The third analysis confirmed this prediction. In this analysis, priming condition, cultural sample, the in- teraction of these two variables, and the proportion of each subject's explanations that mentioned negative features were used to predict the proportion of compromise choices made by each. As predicted, the content of participants' reasons was predictive of compromising {F(l, 112) = 4.22, p< .05), but priming condition and the other predictors were not (p > . 10). Furthermore, that fact that the fit of this model (adjusted R- = .066) was no better than one in which the

cultural priming condition variable was not included (ad- justed R' = .075) further confirms that participants' con- cems about negative consequences is the proximal cause of compromising.

GENERAL DISCUSSION

Asians and North Americans often use different criteria to evaluate products and. therefore, have different prefer- ences when making judgments and choices (e.g.. Aaker and Lee 2001; Briley et al. 2000). As noted earlier, however. these differences are likely to be more apparent when par- ticipants are not consciously aware of their cultural identity. The present research shows that when participants' cultural identity is called to their attention, they become aware of their membership in a group, and this awareness has similar effects on their decisions regardless of the culture to which they belong. Specifically, making Chinese aware of their identity as Chinese stimulates them to avoid decisions that might have negative consequences, as reflected in a tendency to compromise. However, making U.S. participants aware of their identity as Americans has precisely the same effect. To our knowledge, these effects have not previously been identified.

The series of experiments we have conducted, in com- bination with research evidence reported elsewhere, provide a coherent picture of the processes that underlie these effects (see fig. 1). As postulate I implies (see also Aaker and Lee 2001), feelings of belonging to a real or imagined group induce a prevention focus (Higgins 1997. 1998). or a ten- dency to make decisions that minimize the risk of negative outcomes. This disposition is not specific to the group that leads it to be activated and generalizes not only to other interpersonal situations but also to consumer decision sit- uations in which only oneself is involved (postulate 2: see Higgins 1998). Consistent with this proposition, experiments I and 2 showed that real or imagined participation in an ad hoc laboratory group increased the tendency to endorse proverbs that advocated equality in interpersonal situations and to prefer to distribute one's choices over a variety of products in a multiple-choice situation. The assumption that these effects are mediated by a concem with negative out- comes was confirmed by experiment 3. which showed that direct manipulations of participants' concem about receiving negative outcomes had effects very similar to the effects of inducing feelings of group membership. Moreover, this more direct manipulation eliminated any effects that feelings of group membership otherwi.se had.

Furthemiore, we expected that calling people's attention to their cultural identity might induce feelings of group membership similar to those that result from actual partic- ipation in a group. This possibility was supported by evi- dence that exposing participants to symbols of their own culture increased their use of the first person plural pronoun in an unobtrusive sentence-construction task (experiment 4) and that the use of these pronouns is associated with a desire to be part of a group rather than independent (Adaval 2001: Gardner et al. 1999). If this assumption is valid, it implies

EFFECT OF GROUP MEMBERSHIP SALIENCE 413

that calling people's attention to their cultural identity should have effects analogous to the effects of real or imagined group membership that we observed in experiments 1 and 2. This is in fact the case. Exposing participants to symbols of their own culture increased their tendency to use equality as a basis for allocating resources in an interpersonal situ- ation (experiment 5) and to choose products that avoided the risk of negative consequences in a product choice task (experiment 6). Finally, mediational analyses in experiment 6 confirmed the assumption that exposing participants to symbols of their own culture induced a concem with neg- ative consequences of their choices and that this concern, in turn, led them to chose compromise alternatives. In fact, the effect of cultural salience on choices was eliminated when differences in this concern were controlled.

It may be important to note that although people are ob- viously conscious of their feelings of group membership, they may not be clearly aware of the effects of these feelings on their decisions. As we noted earlier, the prevention focus that is induced by these feelings may give rise to a general goal-relevant cognitive procedure (Anderson 1983; Dijk- sterhuis et al. 2000; Markman and Brendl 2000) that, once activated, guides behavior in a number of situations without awareness of the factors that give rise to its activation (Bargh and Gollwitzer 1994). Thus, as our results suggest, increas- ing participants' awareness of their cultural identity, or their membership in a group more generally, can have far-reach- ing effects, being manifested in decisions to which the cul- ture or group is quite irrelevant.

Alternative Interpretations

The hypotheses we evaluated in the present research and the conclusions we have drawn are logical deductions from premises that were empirically validated in either this re- search or elsewhere. Moreover, the assumptions that underlie these hypotheses were generally confirmed empirically. Al- though alternative interpretations of specific findings exist, they generally do not clearly apply to our findings as a whole. For example, the preferences for equality that we ob.served in experiment 1 might be attributed in part to the evocation of a fairness or cooperativeness norm that is ac- tivated by awareness of group membership and applied in- dependently of a concern with negative outcomes. Although this interpretation is plausible if considered in isolation, it could not clearly account for the effects of group and cultural awareness on the consumer decisions that we identified in experiments 2 and 6. Moreover, the findings of experiment 3 render this alternative interpretation less compelling. These results showed that a direct manipulation of concem with positive versus negative outcomes eliminated the ef- fects of awareness of group membership on the experiment 1 dependent variable, proverb endorsements, and on candy choices (experiment 2; see fable I). Thus, any influence that group membership might have had on the salience of a cooperativeness or fairness norm appears to have had no effect on preferences for equality over and above its impact on participants' concern about avoiding negative outcomes.

The evidence that calling attention to participants' cultural identity in experiment 5 induces awareness of group mem- bership also seems unlikely to be mediated by a concern with equality per se. Indeed, a general preference for equality in resource-allocation situations could be the result of past ex- periences in which unequal allocations have had negative social consequences. To this extent, equality norms may not provide an altemative explanation for the effects of group tnembers but rather may be a consequence of the negative consequences that unequal allocations can have. Thus, al- though these and other ad hex: interpretations might be plau- sible for explaining individual findings, the conceptualization we have propcsed provides a parsimonious and systematic account of our results as a whole.

Our conclusions conceming the effects of cultural salience are based largely on the differences in choice behavior when participants were exposed to symbols of their own culture and behavior when they were exposed to symbols of a dif- ferent culture. In some cases, therefore, it is somewhat un- clear whether calling attention to one's own culture increases the tendency to avoid negative outcomes, whether the ex- posure to symbols of a different culture decreases this ten- dency, or both. In this regard, exposing Chinese to American cultural icons in experiment 6 decreased their tendency to compromise relative to conditions in which no cultural sym- bols at all were presented. This decrease may have been due to the association of American icons with individualistic values, which induce a tendency to focus on the positive attributes of choice altematives rather than negative ones (Briley et al. 2000). However, this effect may be independent of the effect of making salient one's own cultural identity (as implied by hypothesis 4).

General Implications and Future Research Directions

Cultural Influences on Consumer Behavior. The findings reported in this article can be viewed in the context of other research on cultural differences in product evalu- ations. When individuals are not conscious of their identity as members of a particular culture or nationality, their pur- chase decisions may be influenced by culture-related norms or behavioral dispositions that they spontaneously apply without awareness of the factors that influence their use. In particular, Asians may be generally more prevention focused than Westerners are (Lee et al. 2000) and consequently may be more inclined to compromise in choice situations (at least when they are asked to explain their decisions to others; see Briley et al. 2000). Making Asians aware of their national or cultural identity may further increase their feelings of belonging to a group and consequently may increase this focus. However, making Americans aware of their national identity may induce feelings of group membership as well and may produce a prevention focus similar to that of Asians. Moreover, the effects of this situationally induced focus may override the effects of cultural differences that might otherwise occur.

414 JOURNAL OF CONSUMER RESEARCH

In this regard. Hong and Chiu (2001; see also Hong et al. 2000) point out that culture-related norms and values are only a subset of the criteria that people can potentially bring to bear on their judgments and decisions, and these criteria may not always be thought of at the time a judgment or decision is made. Thus, the effects of these norms may often be offset by transitory situationa! factors that make alter- native criteria more accessible in memory. Exposure to cul- tural symbols could often increase the accessibility of cul- ture-specific knowledge as well as inducing a prevention focus. However, the influence of this knowledge on judg- ments and behavior may only be apparent under conditions in which participants' judgments have few real or imagined consequences for either themselves or others and, therefore, a concern with negative outcomes of behavior is not par- ticularly relevant (cf. Briley and Wyer 2001; Hong et al. 2000; Oishi, Wyer, and Colcombe 2000).

The overriding influence of transitory situational factors we ob.served in the present research could reflect a more general tendency for people to base their judgments and decisions on the first relevant criterion that comes to mind, without thinking about altemative, equally relevant criteria (Taylor and Fiske 1978; Wyer and Srull 1989). In the con- ditions investigated in the present research, concepts acti- vated by thoughts about one's cultural identity may have been more salient than culture-specific norms and values, and so these latter criteria were not consulted. When people consider the choice situations with which they are con- fronted to be important, they might sample additional judg- ment criteria, and so the effect of culture-based norms and values might be more evident. Further exploration will be required to understand the interplay of cognitive conse- quences of awareness of cultural identify on knowledge ac- cessibility and the motivational consequences of this aware- ness and to identify more precisely the mechanisms that are at work.

General Effects of Group Salience. Our results have implications, not only for the effects of culturai factors on choice behavior, but also for the impact of group salience more generally. The evidence that feelings of membership in a group can influence decisions that are irrelevant to this group is particularly provocative. Numerous situational fac- tors that influence Ihe salience of one's membership in a collective could have effects similar to those we identified in experiments 2 and 6. For example, marketing commu- nications that emphasize social affiliation or group inter- action, or that call attention to family ties, could also induce heightened concern about the negative consequences of a product selection, such as potential product failure. Refer- ences to a group in these communications may be beneficial when marketers wish consumers to focus on features that prevent problems (e.g.. product safety, reliability, and du- rability), but may have less benefit when positive features (e.g., superior performance) are the intended focus. These influences could occur even when the products being con- sidered are of minimal relevance to the particular type of group that was mentioned in the communication.

[Received April 2001. Revised March 2002. David Glen Mick sensed as editor and Wavne D. Hoyer served as

associate editor for this article.]

REFERENCES Aaker. Jennifer L. and Angela Y. Lee (2001). " T Seek Pleasures.

'We' Avoid Pains: The Role of Self Regulatory Goals in Information Processing and Persuasion." Journal of Con- sumer Re.search, 28 (June). 3.^^9.

Adaval. Rashmi (2001), unpublished data. Hong Kong University of Science and Technology, School of Business, Hong Kong.

Anderson, John R. (1983). The Architecture of Cognition. Cam- bridge, MA: Harvard University Press.

Bargh, John A. (1994), "The Four Horsemen of Automaticity: Awareness, Intention. Efficiency and Control in Social Cog- nition," in Handbook of Social Coynition. 2d ed.. Vol. 1. ed. Robert S. Wyer. Jr., and Thomas Srull. Hillsdale. NJ: Erlbaum, 1-40.

(1997). "The Automaticity of Everyday Life." m Advance.'; in Social Cognition, Vol. 10. ed. Robert S. Wyer, Jr., Mahwah. NJ: Erlbaum. 231-246.

(2002). "Losing Consciousness: Automatic Influences on Consumer Judgment. Behavior, and Motivation." Journal of Consumer Research, 29 (September). 280-285.

Bargh, John A.. Mark Chen, and Lara Burrows (1996). "Auio- maticity of Social Behavior: Direct Effects of Trait Construct and Stereotype Activalion on a Action." Journal of Person- ality and Social Fsycholosy, 71(2), 230-244.

Bargh. John A. and Peter M. Gollwitzer (1994). "Environmental Control of Goal-Directed Action: Automatic and Strategic Contingencies between Situations and Behavior." in Integra- tive V/>n'.v of Motivation. Cognition and Emotion: Nebraska Symposium on Motivation. Vol. 41. ed. William D. Spaulding. Lincoln: Universiiy of Nebraska Press. 71-124.

Briley, Donnel A.. Michael Morris, and Itamar Simonson (2000). "Reasons as Carriers of Cullure: Dynamic versus Disposi- tiona! Models of Cultural Influence on Decision Making." Journal of Consumer Research. 27 (September). 157-178.

Briley, Donnel A. and Robert S. Wyer, Jr. (2001). "Transitory Determinants of Values and Decisions: The Utility (or Non- utilily) in individualism and Collectivism in Understanding Cultural Differences." Social Cognition, 19 (June). 198-229.

Colcombe. Stanley J. and Roben S. Wyer. Jr. (2001). "The Effects of image-Based Priming on Lexical Acce.ss. Conceptual Ac- tivation, and Behavior." unpublished manuscript. University of Illinois at Urbana-Champaign.

Crowe. Ellen and E. Tory Higgins (1997). "Regulat{iry Focus and Strategic inclinations: Promotion and Prevention in Decision- Making." Organizational Behavior and Human Decision Pro- cesses. 69 (February), i 17-132.

Davis. Thomas C. (1999). "Revisiting Group Attachment: Ethnic and National identity." Political Psychologv, 20 (March). 2 5 ^ 7 .

Dijksterhuis, Ap. Henk Aarts. John A. Bargh. and Ad van Knip- penberg (20(X)), "On the Relation between Associative Strength and Automatic Behavior," Journal of Experimental Social Psychology. 36 (September). 531-544.

Druckman, Daniel (1994). "Nationalism. Patriotism, and Group Loyalty: A Social Psychological Perspective." International Studies Quarterly. 38 (April). 43-68.

Gardner, Wendi L. Shira Gabriel, and Angela Y. Lee (1999), " T Value Freedom, but 'We' Value Reiationships: Self-Construal

EFFECT OF GROUP MEMBERSHIP SALIENCE 415

Priming Mirrors Cultural Differences in Judgment." Psycho- logical Science. 10 (July), 321-326.

Higgins. E. Tory (1996). "Knowledge Activation: Accessibility. Applicability, and Salience," in Social Psychology: Handbook of Basic Principles, ed. E. Tory Higgins and Arie W. Krug- lanski. New York: Guilford, 133-168.

(1997), "Beyond Pleasure and Pain." American Psychol- ogist, 55 (November). 1217-1233.

(1998). "Promotion and Prevention: Regulatory Focus as a Motivational Principle." in Advances in Experimental Social Psychology, Vol. 30. ed. Mark P. Zanna, San Diego, CA: Academic Press, \-46.

Higgins, E. Tory. Christopher J. R. Roney, Ellen Crowe, and Charles Hymes (1994), "ideal versus Ought Predilections for Approach and Avoidance: Distinct Self-Regulatory Systems," J<mrnat of Personality and Social Psychology, 66 (February), 276-286.' * .

Hofstede. Geert (1980). Culture's Consequences, Newbury Park. CA: Sage.

Hong. Ying-yi and Chi-yue Ciiiu (2001). "Toward a Paradigm Shift: From Cross-Cultural Differences in Social-Cognitive Mediation of Cultural Differences," Social Cognition, 19 (June). 181-196.

Hong, Ying-yi, Michaei Morris, Chi-yue Chiu, and Veronica Benet- Martinez (2000). "Multicultural Minds: A Dynamic Construc- tivist Approach to Culture and Cognition," American P.s-y- chologist, 55 (July). 709-720.

Kahneman, Daniel. Jack L. Knetsch. and Richard Thaler (199i). "The Endowment Effect, Loss Aversion, and Status-Quo Bias," Journal of Economic Perspectives. 5 (Winter). 193-206.

Lee. Angela Y. Jennifer L. Aaker. and Wendi L. Gardner (2000), "The Pleasures and Pains of Distinct Self-Construals: The Role of interdependence in Regulatory Focus." Journal of Personality and Social P.sychotogy, 78 (June), I 122-1 134.

Liberman. Nira, Lorraine C. idson. Christopher J. Camacho, and E. Tory Higgins (1999). "Promotion and Prevention Choices between Stability and Change." Journal of Personality and Social Psychulogy. 11 (June). 1135-1145.

Markman. Arthur B. and Miguel C. Brendl (20(X)). "The influence of Goals on Value and Choice." in Psychology of Learning and Motivation, Vol. 39. ed. Douglas L. Medin. San Diego, CA: Academic Press. 97-128.

Markus, Hazel-Rose and Shinobu Kitayama (1991), "Culture and the Self: Implications for Cognition. Emotion and Motiva- tion." Psychological Review, 98 (April). 224-253.

Oishi. Shigehiro. Robcrl S. Wyer. and Stanley Colcombe (2000). "Cultural Variation in ihe Use of Current Life Satisfaction to Predict the Future." Journal of Personality and Social Psy- chology, 78 (March), 434-445.

Roney, Ciiristopher J. R., E. Tory Higgins, and James Shah (i995). "Goais and Framing: How Outcome Focus Influences Moti- vation and Emotion." Personality and Social Psychology Bul- letin. 2i (November). li-'il-ll60.

Shafir. Eldar. itamar Simonson. and Amos Tversky (1993), "Rea- soning-Based Ciioice," Cognition. 49 (October-November), 11-36.

Simonson, Itamar and Amos Tversky (1992). "Choice in Context: Tradeoff Contrast and Extremeness Aversion." Jounial of Marketing Research, 29 (August). 281-295.

Taylor. Shelley E. and Susan T Fiske (1978). "Salience. Attention, and Attribution: Top of the Head Phenomena." in Advances in Experimental Social P.sychology. Vol. 11, ed. Leonard Ber- kowitz. New York: Academic Press. 249-288.

Triandis, Harry C. (1989). "The Self and Social Behavior in Dif- fering Cultural Contexts." Psychological Review, 93 (July). 506-520.

Tversky. Amos and Itamar Simonson (1993). "Context-Dependent Preferences," Management Science, 39 (October), 1179-1189.

Weber. Elke. Christopher Hsee. and Joanna Sokolowska (1998), "What Folklore Tells Us about Risk and Risk Taking: Cross- Cultural Comparisons of American. German and Chinese Proverbs." Organizational Behavior and Human Decision Proces.ses. 15 (August). 170-186.

Wyer, Roberi S. and Thomas K. Srull (1989). Memory and Cog- nition in Its Social Context. Hillsdale, NJ: Erlbaum.