Effects of Self-Construal Differences on Cognitive Dissonance Examined by Priming the Independent and Interdependent Self

Prior cross-cultural research on dissonance has relied on cultural stereotypes in assuming that individuals from Western cultures are individualistic and have independent self-construals whereas individuals from Asian cultures are collectivistic and have interdependent self-construals. The current research made use of priming to avoid relying on cultural stereotypes and examined how having an independent or interdependent self-construal accounted for differences in dissonance experienced. A total of 120 participants who were Singapore citizens or permanent residents were randomly assigned to one of four conditions. Participants received either an independent or interdependent prime, and rated and ranked CDs before and after they made a choice between closely valued alternatives either for oneself or a close other. Results indicated that independently primed participants demonstrated significant dissonance when they made choices for themselves and close others whereas interdependently primed participants demonstrated significant dissonance when they made choices for close others but not for themselves. The study’s findings suggest that having either self-construal accounted for differences in dissonance experienced.

after making a choice between their fifth and sixth favorite objects (Brehm, 1956). Participants are intentionally asked to choose between their fifth and sixth choices so as to allow them the opportunity to experience dissonance. Dissonance is experienced as a result of negative aspects of the chosen alternative and positive aspects of the rejected alternative that are dissonant with the decision (Harmon-Jones & Mills, 1999). To reduce dissonance, individuals justify their decisions by increasing the desirability of the chosen alternative, decreasing the desirability of the rejected alternative, or both (Kitayama, Ishii, Imada, Takemura, & Ramaswamy, 2006). This process is termed the post-decisional spreading of alternatives where a spread of alternatives score is calculated as an objective measure of dissonance reduction and is taken to be indicative of dissonance experienced (Hoshino-Browne et al., 2005;Kitayama, Snibbe, Markus, & Suzuki, 2004).
Researchers have concluded that the experience of dissonance is culture specific. For example, Kitayama et al. (2004) examined cross-cultural dissonance effects within a free choice paradigm and found that Japanese demonstrated dissonance only when the presence of a social other was made salient through the priming of "social eyes" (Study 4) and when they were asked to think of college students' preferences (Study 2) whereas European Americans demonstrated dissonance regardless of whether social cues were made salient. Similarly, Hoshino-Browne et al. (2005) found that European Canadians justified choices for themselves whereas Asian Canadians and Japanese justified choices for close friends (Hoshino-Browne et al., 2005).
On the basis of the theory of independent and interdependent self-construals (Markus & Kitayama, 1991), researchers thus surmised that observed differences in dissonance findings were due to participants' culturally sanctioned self-construals (Hoshino-Browne et al., 2005;Kitayama et al., 2004). Self-affirmation theory asserts that a self-affirming, imagemaintaining process is activated when significant aspects of a person's self-concept are threatened (Steele & Liu, 1983;Steele, Spencer, & Lynch, 1993). A suboptimal decision that involved the self was therefore threatening to those with independent self-construals, whereas a suboptimal decision that involved social others was threatening to those with interdependent self-construals, thereby leading them to experience dissonance under different circumstances (Hoshino-Browne et al., 2005).

Limitations in Cross-Cultural Dissonance Research
In cross-cultural dissonance research, cultures are sampled by comparing individuals from different countries or ethnic groups (e.g., Americans vs. Japanese, European Canadians vs. Asian Canadians). Notably, prior research has assumed that individuals from Western countries are individualistic and have independent self-construals whereas individuals from Asian countries are collectivistic and have interdependent self-construals (e.g., Heine & Lehman, 1997;Hoshino-Browne et al., 2005;Imada & Kitayama, 2010;Kitayama et al., 2004). As such, studies purportedly investigating cultural differences between Westerners and Asians cannot conclude with certainty that between-group differences have a cultural source (i.e., due to culturally sanctioned self-construals; Matsumoto, 1999;Matsumoto & Yoo, 2006). Yet, past studies have done so, committing the "cultural attribution fallacy"-where culture is assumed to be responsible for observed differences without empirical justification (Matsumoto & Yoo, 2006, p. 235). In addition, any conclusions drawn regarding self-construal and dissonance processes might have confounded with other variables such as socioeconomic, religious, and demographic variables (Matsumoto, 1999) that differ between individuals from Western and Asian samples (Haberstroh, Oyserman, Schwarz, Kühnen, & Ji, 2002).
To justify the causal statement that differences in culturally sanctioned self-construals between individuals from Western and Asian groups account for differences in dissonance findings, further research is warranted to demonstrate that (a) individuals tested come from individualistic and collectivistic cultures, (b) individuals tested have independent and interdependent self-construals, (c) the cultural dimensions of individualism and collectivism are associated with the specified self-construals, and (d) the specified self-construals account for differences in dissonance findings (Matsumoto, 1999).

Strengthening Methodology Using Priming
Priming can be understood as the temporary activation of mental concepts (Cross et al., 2011;Gilovich, Keltner, & Nisbett, 2010). The rationale underpinning priming is that multiple modes of self-construal exist in all individuals from any culture (Brewer & Gardner, 1996;Gardner, Gabriel, & Lee, 1999;Hong, Morris, Chiu, & Benet-Martínez, 2000), and the use of priming enables temporary access to either an independent or interdependent self-construal, allowing the examination of either self-construal on behavior (Cross et al., 2011).
Research has demonstrated the effectiveness of priming the independent and interdependent self using the Similarities and Differences With Family and Friends (SDFF) task (Oyserman & Lee, 2008;Trafimow, Triandis, & Goto, 1991). For example, Trafimow et al. (1991) found that irrespective of cultural background, North Americans and Chinese undergraduates who received a private self (independent) prime provided more idiocentric responses referring to personal qualities, beliefs, and attitudes and less group responses referring to experiences of common fate or group membership than those who received a collective self (interdependent) prime (Cross et al., 2011;Kühnen, Hannover, & Schubert, 2001). Similarly, students from New Mexico were more likely to be influenced by personal attitudes and goals when the private self was primed whereas subjective norms had a greater impact on behavioral intentions when the collective self was primed (Ybarra & Trafimow, 1998).

Present Research
In the present study, prior research undertaken by Hoshino-Browne et al. (2005) was extended by adding the SDFF priming task to experimentally manipulate participants' selfconstruals prior to their completing the free choice paradigm. Participants would receive an independent or interdependent prime, then rate and rank CDs before and after completing a dissonance inducing task. In this dissonance inducing task, they would be asked to make a less than ideal decision between their fifth and sixth favorite CDs either for themselves or a close social other. Without relying on assumptions that perpetuate cultural stereotypes, the present study sought to provide empirical evidence demonstrating that having either an independent or interdependent self-construal accounted for differences in dissonance effects. The study was restricted to Singapore citizens or permanent residents, so that culture was held constant to investigate selfconstrual differences.

Hypotheses
Prior research has shown that a suboptimal decision involving the self is threatening for those with independent selfconstruals, whereas a suboptimal decision involving a close social other is threatening for those with interdependent selfconstruals (Hoshino-Browne et al., 2005;Kitayama et al., 2004). It was thus hypothesized that independently and interdependently primed participants would experience significant amounts of dissonance depending on whether they made a suboptimal choice for themselves or close others. Based on past research findings, the following sub-hypotheses were formulated: Hypothesis 1a: Independently primed participants would demonstrate significant dissonance when they made choices for themselves but not for close others. Hypothesis 1b: Interdependently primed participants would demonstrate significant dissonance when they made choices for close others but not for themselves.

Design
The current study had a 2 (priming type: independent vs. interdependent) × 2 (target person: self vs. other) betweensubjects factorial design. The dependent variable was the spread of alternatives, calculated by summing the increase in rating of the chosen CD and decrease in rating of the rejected CD between Times 1 and 2. The spread of alternatives score is a measure of dissonance reduction and is taken to be indicative of dissonance experienced. A greater spread of alternatives indicates greater dissonance experienced.

Apparatus and Procedure
Participants were randomly assigned to one of four conditions. Before the experiment began, participants were given an information sheet and consent form to complete. The information sheet advertised the research as a study about undergraduates' English music preferences so as to prevent participants from learning the actual purpose of the study and responding in a socially desirable manner.
Priming. Participants were then primed with the SDFF task (Trafimow et al., 1991). Independently primed participants were given the verbal instructions: "For the next two minutes, you will not need to write anything down, think of what makes you different from your family and friends." Interdependently primed participants were given the verbal instructions: "For the next two minutes, you will not need to write anything down, think of what you have in common with your family and friends." This priming technique has been found to be effective in priming independent or interdependent self-construals (Oyserman & Lee, 2008).
Time 1 ranking task. Participants chose 10 out of 15 English music CDs. Participants in the self condition were asked to rank CDs according to personal preferences. Participants in the other condition were asked to pick a close social other (e.g., close friend, relative, or romantic partner) they had in mind and to think of this close other's preferences when ranking CDs. A variable that might have influenced the results in the other condition was the degree of closeness that participants felt toward the social other (i.e., close friend, relative, romantic partner) they thought of. In an effort to control for this variable, participants who chose a gift CD for close others were asked to indicate the degree of felt closeness to that person on a 5-point scale (1 = not close and 5 = really close). On task completion, CDs and participants' responses were removed.
Time 1 rating task. Participants were provided with rating sheets. Depending on the condition they were assigned to, participants rated the 10 CDs according to personal preferences (self condition) or the preferences of the close other they were thinking of (other condition). Participants rated each CD on a 7-point scale (1 = really don't like it and 7 = really like it). On task completion, rating sheets were removed.
Decision making: Choosing one of two CDs. In a separate area, the researcher viewed participants' responses on the ranking sheets to identify their fifth and sixth favorites. Following that, participants were told a cover story that they would receive a gift CD in return for their participation. This was to ensure that a serious choice would be made between two CDs, allowing an opportunity for dissonance. Participants were then presented with two CDs and asked to choose one of them. These CDs corresponded to participants' fifth and sixth favorites, so that a choice would be made between closely valued alternatives as well as to allow movement in ratings when participants rated CDs again later. Participants in the self condition were asked to choose a CD for themselves while participants in the other condition were asked to choose a CD for the close other (i.e., close friend, relative, or romantic partner) they had been thinking of. Participants were given several minutes to ensure sufficient time for them to justify their choices. In addition, they were given the option of having the gift CD mailed to them or their close other, and were asked for their mailing addresses, thereby increasing the credibility of the cover story.
Time 2 ranking task. Participants were asked to rank and rate the same CDs once more. They were provided with a list of CD albums to aid in their ranking and rating of CDs. Ranking sheets were provided and participants ranked the same 10 CDs according to personal preferences or a close other's preferences.
Time 2 rating task. Participants were then provided with rating sheets that were identical to those used at Time 1, except that CD titles were ordered differently so as to prevent them from remembering their previous ratings. Participants then rated the 10 CDs according to personal preferences or a close other's preferences.
"Who am I?" test. Last, participants completed the "Who am I?" test as a manipulation check that assessed the self-construal they identified with after priming. The test was patterned after Kuhn and McPartland's (1954) Twenty Statements Test that is a self-report measure of self-construal. Participants were asked to provide seven open-ended responses to the question "Who am I?" Their responses were coded as either independent or interdependent by two coders. Responses that described a personal attribute, such as an ability or belief (e.g., I am intelligent) were coded independent (Agrawal & Maheswaran, 2005;Trafimow et al., 1991). Responses that described a social role (e.g., I am a son), demographic category (e.g., I am Chinese), or group membership (e.g., I am from James Cook University) were coded interdependent (Agrawal & Maheswaran, 2005;Trafimow et al., 1991). Responses that referred to relationships and qualities of interdependence such as being sensitive and responsive to others (e.g., I am loved by friends and family, I am someone who helps my friends) were also coded interdependent (Trafimow et al., 1991). Participants were then asked to indicate their age, gender, and ethnicity for the purpose of a demographic analysis of the participant sample.
Participants were thanked and debriefed about the purpose of the experiment. The researcher apologized for the deception and informed participants that they would not be receiving a CD due to limited funding. The experiment took approximately 40 min.

Results
Data from 120 participants were collected and analyzed using Statistical Package for the Social Sciences (SPSS) version 18. An alpha level of .05 was used for all statistical tests. Data cleaning was undertaken where a preliminary analysis of the data revealed a single outlier with a standardized score in excess of 3.29 that was subsequently assigned a raw score one unit greater than the next most extreme score in the distribution. Skewness and kurtosis values were within an acceptable range of deviation from normality (i.e., ± 2.00).

Analysis of Manipulation Check for Self-Construal Priming
A two-way mixed ANOVA was conducted to verify the selfconstrual priming manipulation. Proportions of each type of response were calculated by taking the number of independent or interdependent responses each participant provided and dividing by the total number of responses made by the participant (Trafimow et al., 1991). To increase reliability of response coding, a third party coded participants' responses and a kappa statistic was generated using the SPSS Crosstabs procedure. Kappa was .92, demonstrating substantial agreement between raters. Figure 1 shows the proportion of independent and interdependent responses participants in each condition provided. There was a significant interaction between priming and response type, F(1, 118) = 46.24, p < .001, η 2 = .23.
Two-tailed independent-sample t tests were conducted to examine the difference in responses at each level of priming.

Effects of Priming and Decision Making on the Spread of Alternatives
A two-way between-subjects ANOVA was undertaken to test the prediction that the level of dissonance experienced was dependent on the kind of priming participants received and whether they were asked to choose for themselves or for close others. Figure 2 shows the mean spread of alternatives of independently and interdependently primed participants. Results indicated that the interaction between priming and target person was significant, F(1, 116) = 8.48, p = .004, η 2 = .07. Although not the main research focus, there was a significant main effect for target person, F(1, 116) = 5.05, p = .03, η 2 = .04, but no significant main effect for priming, F(1, 116) = 0.56, p = .46.
Two-tailed independent-sample t tests were conducted to explore the significant interaction effect obtained. Results indicated that among participants in the self condition, those who were independently primed (M = 0.53, SD = 1.20) had a significantly higher mean spread of alternatives than those who were interdependently primed (M = −0.20, SD = 1.24), t(58) = 2.33, p = .02, 95% CI = [0.10, 1.36], η 2 = .09. Among independently (M = 0.40, SD = 0.86) and interdependently primed (M = 0.83, SD = 1.06) participants in the other condition, no significant difference in mean spread of alternatives was obtained, t(58) = −1.75, p = .09, 95% CI = [−0.93, 0.06]. Table 1 shows the results of four one-sample t tests against zero that were conducted to test the prediction that independently primed participants would justify decisions for themselves and not for close others, whereas interdependently primed participants would justify decisions for close others and not for themselves. In line with prior research (Hoshino-Browne et al., 2005), an alpha level of .05 was used without correcting for familywise error.
Contrary to Hypothesis 1a, independently primed participants justified choices for themselves but also justified choices for close others. Supporting Hypothesis 1b, interdependently primed participants justified choices for close others but not for themselves.

Degree of Closeness in Other Condition
Participants who chose a gift CD for close others were asked to indicate the degree of felt closeness to that person on a 5-point scale. An independent-samples t test indicated no significant difference in ratings between independently primed (M = 4.33, SD = 0.66) and interdependently primed participants (M = 4.47, SD = 0.57), t(58) = −0.84, p = .41, 95% CI = [0.45, 0.19].

Discussion
The present study examined how having either an independent or interdependent self-construal affected the experience of dissonance. Consistent with prior research (Trafimow et al., 1991;Ybarra & Trafimow, 1998), a significant interaction between priming and response type showed that the priming manipulation was successful-independently primed participants provided more independent responses than interdependently primed participants, and interdependently primed participants provided more interdependent responses than independently primed participants.
A significant interaction between priming type and target person supported the hypothesis that the amount of dissonance experienced was dependent on the type of priming received and whether a choice was made for oneself or a close other. Although individuals from Western and Asian cultures displayed dissonance depending on whether a choice Note. Error bars represent standard errors of the means. An asterisk indicates that the mean spread of alternatives is significantly different from zero.
was made for oneself or a close friend (Hoshino-Browne et al., 2005), the present results corroborated and extended this finding by providing empirical evidence suggesting that differences in self-construal accounted for differences in dissonance experienced.

Findings That Address Hypothesis 1a
Consistent with prior research (Hoshino-Browne et al., 2005;Kitayama et al., 2004), independently primed participants significantly justified choices for themselves, partially supporting Hypothesis 1a. This finding supports the notion that for those with independent self-construals, self-identity is associated with inner attributes perceived to be stable and unchanging across situations (Cross et al., 2011). Individuals' behaviors are deemed to be a reflection of their attitudes and emotions (Markus & Kitayama, 1991). Thus, behavior inconsistent with attitudes threatens those with independent self-construals (Heine & Lehman, 1997) and produces dissonance leading them to justify decisions for themselves. In contrast to an earlier study that showed that European Canadians (assumed to have independent self-construals) justified choices for themselves but not for close friends (Hoshino-Browne et al., 2005), independently primed participants in the current study significantly justified choices for close others. The present finding suggests that individuals with independent self-construals may have experienced interpersonal worry and therefore adjusted their behavior according to the feelings and actions of close others (Kitayama et al., 2004;Markus & Kitayama, 1991). Cross, Bacon, and Morris (2000) distinguished between the nature of interdependence for individuals from individualistic and collectivistic cultures; close relationships are more likely to play a part in self-definition for individuals from individualistic societies whereas roles and group memberships are more likely to play a part in self-definition for individuals from collectivistic societies. Accordingly, individuals with independent self-construals could experience a different form of interpersonal worry from individuals with interdependent self-construals. Further research is required to examine this distinction between interpersonal worry for individuals with independent self-construals and individuals with interdependent self-construals.
Results showed that when participants chose CDs for others, interdependently primed participants justified their choices more than independently primed participants, although this difference was not significant. Thus, the extent that individuals with independent self-construals are influenced by close social others may be interpreted to be less than that of individuals with interdependent self-construals. Moreover, when participants chose CDs for themselves, results indicated that independently primed participants demonstrated significantly greater dissonance than interdependently primed participants, suggesting that individuals with independent self-construals found suboptimal decisions for themselves more threatening than individuals with interdependent self-construals did.
The difference in results between Hoshino-Browne et al. 's (2005) study and the present study may be due to improved methodology with the implementation of priming. Notably, this extends current understandings of those with independent self-construals-whereas prior research relying on cultural stereotypes suggests that individuals with independent self-construals experience dissonance only when making choices for themselves (Hoshino-Browne et al., 2005), the present findings suggest that having an independent selfconstrual does not preclude one from experiencing interpersonal worry and demonstrating dissonance when choices implicate close others.

Findings That Address Hypothesis 1b
Consistent with research by Hoshino-Browne et al. (2005), results supported Hypothesis 1b, as interdependently primed participants significantly justified their choices for close others but not for themselves. In a similar vein, prior research (Imada & Kitayama, 2010;Kitayama et al., 2004) indicated that Japanese (assumed to have interdependent self-construals) demonstrated dissonance when the presence of a social other was made salient through the priming of "social eyes" as perceived social influence was relevant to the interdependent self but not in the absence of such priming. Note. CI = confidence interval; LL = lower limit; UL = upper limit.
The present findings indicate that those with interdependent self-construals are threatened by the possibility of having made inconsiderate decisions for close others but not when decisions are made for themselves, supporting the theory that social roles, group memberships, and close relationships encompass important aspects of self-identity for individuals with interdependent self-construals (Cross et al., 2011). As such, these individuals are largely influenced by the perceived feelings and actions of others (Markus & Kitayama, 1991). Inconsiderate decisions that implicate close others therefore threaten individuals with interdependent self-construals (Hoshino-Browne et al., 2005) and produce dissonance leading them to justify decisions for close others.

Alternative Explanations for Participants' Justification in the Other Condition
There are three alternative explanations as to why participants justified their choices for close others.
Vicarious dissonance. One explanation is that participants experienced dissonance vicariously. Norton, Monin, Cooper, and Hogg (2003) demonstrated that undergraduates displayed dissonance on witnessing close in-group members deliver speeches that were inconsistent with their beliefs. Vicarious dissonance was experienced as a result of the discomfort participants imagined experiencing if they were in the same position as that close in-group member (Norton et al., 2003). However, participants in the present research did not witness close others engage in choice-justifying behavior, nor were they asked to imagine themselves in a close other's place. Therefore, participants likely experienced psychological discomfort evoked by making choices for close others and not vicarious dissonance.
Using personal preferences. Another explanation is that participants substituted that close other's music preferences with their own preferences. The subsequent choices they made and the post-decisional justification were then for themselves and not for that close other. However, measures were implemented to prevent this by asking participants to indicate to the researcher who they were thinking of (i.e., a close friend, relative, or romantic partner). Second, all questionnaires provided to participants were clear in their instructions as to whose preferences participants were to use. Third, as part of the cover story, participants were given the option of mailing the gift CD to that close other if they so desired. Thus, efforts were made to ensure that participants in the other condition did not use their own music preferences.
Degree of felt closeness. The degree of closeness that participants felt toward the person they chose a gift CD for could have influenced levels of dissonance experienced (Hoshino-Browne et al., 2005); hence, participants were asked to indicate the degree of closeness they felt toward their friend, relative, or romantic partner. Results indicated no significant differences in subjective closeness between independently and interdependently primed participants, thereby eliminating variations in felt closeness as a factor accounting for differences in spread of alternatives between the groups.

Implications
Whereas research has indicated that affirming self-relevant values eliminates dissonance (Steele & Liu, 1983), the present findings imply that the processes of self-affirmation and image-maintenance vary for independent and interdependent selves due to distinct identity related concerns. For example, those with independent self-construals might find positive feedback regarding their uniqueness and competency to be most self-affirming whereas those with interdependent selfconstruals might perceive positive feedback concerning their abilities to cooperate and fit in to be self-affirming.
In line with prior research (Agrawal & Maheswaran, 2005;Trafimow, Silverman, Fan, & Law, 1997), the current study demonstrated the effectiveness of priming an independent and interdependent self among participants from a single cultural group, further supporting the theory that multiple modes of self-construal exist in all individuals regardless of cultural background (Brewer & Gardner, 1996). This suggests that other modes of self-construal can temporarily override individuals' original self-construal shaped by stable factors such as personality or culture (Suh, Diener, & Updegraff, 2008). Implications of the present findings extend to research on acculturation, where acculturation can be understood as changes that take place in individuals who come into contact with cultures that are different from their own (Berry, 1997). Acculturation strategies (viz., assimilation, separation, integration, marginalization; Berry, 1997) may then be influenced by exposure to contextual cues (e.g., being at home and at school) and symbols (e.g., language) that are psychologically associated with either culture (Hong et al., 2000).

Limitations and Future Directions
Some methodological limitations include the use of music CDs that are becoming irrelevant due to improved technology such as portable media players. CDs are therefore objects of less value, possibly arousing lower levels of dissonance in participants. In addition, the present research did not include a control group, where control participants are not exposed to any priming manipulation. In future research, all participants could complete a self-report measure of self-construal, for example, the Self-Construal Scale (Singelis, 1994). Participants in the experimental conditions would receive a prime while participants in the control conditions would not. The responses of participants who received an independent or interdependent prime can then be compared with participants who naturally make use of an independent or interdependent self-construal.
Self-construal in the current research was conceptualized as a binary construct because independence and interdependence have received the most attention (Cross et al., 2011;Suh et al., 2008). However, research has indicated other kinds of self-construal such as the personal, collective and relational selves (Brewer & Gardner, 1996). Of note, parallels have been drawn between the personal and independent self while the interdependent self is considered to have collective and relational components (Cross et al., 2011;Kühnen et al., 2001). Relational-interdependent self-construal describes how individuals define themselves in terms of their close relationships whereas collective-interdependent selfconstrual describes the extent to which individuals define themselves in terms of group memberships (Cross et al., 2011). Research has indicated that European American women describe themselves in more relational terms whereas European American men describe themselves in more collective terms . Accordingly, future research on dissonance can examine cross-cultural gender differences where men and women from different cultures are subjected to independent, relational, and collective priming.

Conclusion
Prior cross-cultural research on dissonance has relied on assumptions regarding the cultural backgrounds and selfconstruals of participants tested in concluding that observed differences between Westerners and Asians are due to selfconstrual differences. As such, the validity of their conclusion is questioned. With the use of priming, the present study avoided relying on cultural stereotypes and provided direct evidence demonstrating that having an independent or interdependent self-construal accounted for differences in dissonance experienced. Although further research is required with participants from Western and Asian countries (Matsumoto, 1999), the present study is a first step toward verifying the claim that observed differences in prior crosscultural dissonance research (e.g., Hoshino-Browne et al., 2005;Kitayama et al., 2004) are a result of self-construal differences sanctioned by an individualistic Western culture or a collectivistic Asian culture.

Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding
The author(s) received no financial support for the research and/or authorship of this article.