THE INFLUENCE OF THE MESSAGE SOURCE’S NATIONAL IDENTITY ON CHINESE INTERNATIONAL STUDENTS’ ATTITUDES By Xiaowen Xu A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Communication -- Master of Arts 2014 ABSTRACT THE INFLUENCE OF THE MESSAGE SOURCE’S NATIONAL IDENTITY ON CHINESE INTERNATIONAL STUDENTS’ ATTITUDES By Xiaowen Xu Although international students’ acculturation and psychological status have been studied by cross-cultural scholars, the question of whether they are more likely to take recommendations from their own national group or the host national group has been minimally studied. Based on Social Identity Theory, this paper predicted that Chinese international students in the U.S. are more easily persuaded by a co-national (Chinese) source than an American source. A 2 (source national identity) by 2 (message valence) factor design was conducted among 144 Chinese international students. Results were not in the predicted direction. Participants were persuaded more by an American source than a Chinese source, though the difference was only marginally significant. There was no significant difference in perceived source credibility or similarity. However, when co-national identification level was controlled, source identity and message valence interacted to influence perceived credibility. Possible reasons for these findings are discussed. Keyword: persuasion, source identity, co-national identification, in- and out-group communication ACKNOWLEDGEMENTS For completing this thesis, my deep gratitude first goes to Dr. Mary Bresnahan, my advisor and thesis committee chair, who expertly guided me through the process of the study. Her enthusiasm for research and education has encouraged my engagement with my own research. Her general care has helped make my study and life in Michigan State University a wonderful experience. Equal gratitude is also given to the other two members of my thesis committee: Dr. Frank Boster and Dr. Gary Hsieh, who was my previous advisor and is now at University of Washington. Both of them gave very valuable suggestions and revision ideas for this thesis. My appreciation extends to my graduate classmates in the Department of Communication. I enjoyed sharing my idea with Bingqing Wang, Sonia Shaikh and Xun Zhu, all of whom have made some insightful comments on this work. My gratefulness further goes to a group of other friends in Michigan State University, who helped me distribute questionnaires among Chinese international students. Thanks to them, the data collection went much more efficient. Above all, I am indebted to my beloved family back in China, my father and my mother, for their constant support and encouragement of my graduate study in the United States. iii TABLE OF CONTENTS LIST OF TABLES .........................................................................................................................v LIST OF FIGURES ..................................................................................................................... vi CHAPTER 1 Introduction ....................................................................................................................................1 CHAPTER 2 Literature Review ..........................................................................................................................3 Social Identity and Persuasion .................................................................................................3 Co-national Identity Matching Effect in Persuasion .............................................................4 Source Credibility and Perceived Similarity ..........................................................................7 Co-national Identification and Communication with Americans .........................................8 CHAPTER 3 Hypotheses and Research Questions ..........................................................................................10 CHAPTER 4 Method ..........................................................................................................................................12 Participants and Research Design .........................................................................................12 Materials and Procedure ........................................................................................................12 Measures ..................................................................................................................................13 Manipulation Check ..............................................................................................................13 Main Variables......................................................................................................................14 Scale Validity and Reliability .................................................................................................15 CHAPTER 5 Results ...........................................................................................................................................18 Manipulation Checks ..............................................................................................................18 Hypotheses Test .......................................................................................................................19 CHAPTER 6 Discussion .....................................................................................................................................29 CHAPTER 7 Limitations ...................................................................................................................................35 APPENDICES ..............................................................................................................................37 Appendix A: Questionnaire ....................................................................................................38 Appendix B: Induction Messages ...........................................................................................44 REFERENCES ............................................................................................................................46 iv LIST OF TABLES Table 1 Descriptive statistics of all dependent variables for each group ........................................................................................................................................................23 Table 2 Correlations among co-national identification, communication with Americans and dependent variables ......................................................................................................................24 v LIST OF FIGURES Figure 1 The effect of source identity and message valence on source credibility with co-national identification as a covariate .......................................................................................................................................................25 Figure 2 The effect of source identity and message valence on trust advice with co-national identification as a covariate .......................................................................................................................................................26 vi CHAPTER 1 Introduction The number of international students from China attending American colleges and universities has been steadily increasing, almost tripling since the school year of 2005-2006. From 2009, China has sent more international students to public and private institutions in the United States than any other nation. In 2011-2012 Chinese international students numbered 194,029, 25.4% of the total number of international students in the U.S. (Institute of International Education, 2012). Because of the rapidly rising number, the cultural and psychological status of Chinese students in the U.S. has become an important area of study. International students are described as traveling voluntarily to a new culture for education goals, and normally plan to stay for a finite period of time with an expectation of finally returning to their country of origin (Ward & Kennedy, 1994). As they enter a new nation, the unfamiliar cultural environment and different everyday practices increase their uncertainty level, which they have to cope with during the process of cultural adaptation (Ward & Kennedy, 1994; Ward & Rana-Deuba, 1999). Information seeking is one approach to decrease uncertainty due to novelty (Berger & Calabrese, 1975). Thus, international students may seek and take recommendations from others related to their study and life abroad. In this way, it is important to explore what kind of factors may influence international students’ likelihood to be persuaded in their lives. Although international students have been studied in terms of their psychological adjustment (Ward & Kennedy, 1993; Ward & Kennedy, 1994; Ward & Rana-Deuba, 1999), acculturative stress (Wang & Mallinckrodt, 2006), intergroup anxiety (Greenland & Brown, 2005), and in-group difference sensitivity (Sodowsky & Plake, 1992), there are limited studies on this group regarding persuasion. This thesis investigates whether Chinese 1 international students are more persuaded by Chinese than by Americans. A further question asks whether the strength of Chinese international students’ co-national identification, i.e. identification to their national communities of origin (Ward & Kennedy, 1994), and communication with Americans influences their attitudes to follow a recommendation from a national in-group (Chinese) versus a national out-group (American) member. 2 CHAPTER 2 Literature Review Social Identity and Persuasion Studies focusing on the role of social identity in persuasion have shown the potential power of in-group peers on persuasion, not only by imposing compliance, but also by influencing private attitudes. Earlier findings were based on the Social Identity Theory (SIT, Tajfel & Turner, 1986; Tajfel, 1981). SIT explains that identity, part of persons’ self-concept, derives from the social group or groups into which they categorize themselves as a member. People tend to categorize themselves as members of some groups (in-group) and to differentiate themselves from others (out-group). Members from the in-group are perceived as similar to oneself whereas members from the out-group are perceived as different (Tajfel & Turner, 1986; Tajfel & Turner, 1979; Tajfel, 1981). Under conditions in which group identity is salient, people tend to have more positive attitudes toward in-groups than toward out-groups. Studies demonstrated that people are more motivated to agree with their in-group peers than with out-group members (Festinger, 1950; Kelman, 1958; Kelman, 1961; Mackie, 1986). Even an in-group message with weak arguments was more persuasive than an out-group message with either strong or weak arguments (Fleming & Penny, 2000). Fleming and Penny (2000) clarified that it is not source-receiver membership congruence but the subjects’ identification with their own in-groups that affected the persuasive power of the message. Their studies showed that the in-group source worked more effectively than the outgroup source only when subjects reported a high level of in-group identification. Later researchers began to ask an additional question regarding how group membership affects persuasive consequences. Directed by the Elaboration Likelihood Model (ELM), researchers studied whether and when the source’s group membership was used as a peripheral or a 3 systematic cue. In one way, the source’s group identity could be a peripheral cue in that people might view an in-group source more credible in defining reality (Festinger, 1950; Turner, 1987) than an out-group source. In another way, by increasing the perceived group relevance of the issue, it could also be a systematic cue that motivated people to process the message through a central route (Mackie, Worth & Asuncion, 1990; Platow, Mills & Morrison, 2000; van Knippenberg & Wilke, 1992). Although both processes might be invoked, several moderators were examined, for example, issue relevance (Mackie, Worth and Asuncion, 1990), accessibility of heuristic cues (Mackie, Gastardo-Conaco & Skelly, 1992), and group prototypicality of the message position (van Knippenberg, Lossie & Wilke, 1994). Common within all these studies was that the out-group source had no significant persuasive effect. However, Kerr (2001) found that a message from a relevant out-group, for example, a competing group, might gain as much scrutiny as an in-group message, since “the presence of an (relevant) outgroup source activates ingroup identity may lead to outgroup sources producing similar processing outcomes as those found…for ingroup source” (Mackie & Queller, 2000). Another argument is that when people are highly motivated to align with (for an ingroup source) or to differentiate from a source group (for a relevant outgroup source), they are likely to analyze the information carefully (Kerr, 2001). Co-national Identity Matching Effect in Persuasion Research regarding source-audience co-national identity matching in persuasion has mostly been set in the advertisement or commercial contexts. International ads research supported the proposition that customized ads that included local models and other cultural cues were preferred to standardized ads by consumers in different countries (Shoham, 1996; Butt & de Run, 2012). Additional studies drew similar conclusions from samples of racial or ethnic 4 permanent residents, including Chinese or Asian minorities (Deshpande & Stayman, 1994; Elias, Appiah & Gong, 2011; Martin, Kwai-Choi Lee & Feng, 2004; Sierra, Hyman & Torres, 2009; Whittler, 1991; Lee, Fernandez & Martin, 2002). These studies have important implications for the current study on Chinese international students. First, they suggested a potential explanation for the effect of ethnic targeting ads. Audiences perceived the same-ethnic spokesperson more similar to themselves, thus more favorable and more trustworthy than the different-ethnic spokesperson (Morimoto & La Ferle, 2008). Second, some studies found this effect most prominent when an ethnic minority group was targeted (McGuire, McGuire, Child & Fujioka, 1978). This result was said to occur because for the members of a minority group, ethnicity is more distinct and thus the cue of source ethnicity is more prominent for them than for a majority group; and the more salient a variable is to one’s self-concept, the higher its likelihood to elicit the persuasive effect (Deshpande & Stayman, 1994; Appiah & Liu, 2009; Appiah, 2002; Sierra, Hyman & Torres, 2009; Lee, Fernandez & Martin, 2002). Third, similar to what was found with the persuasion effect of ingroup and out-group sources, the stronger people identified with their own ethnic group, the more perceived credibility and preference one would assign to the ethnic in-group message source (Williams & Qualls, 1989; Arpan, 2002; Sierra, Hyman & Torres, 2009; Elias, Appiah & Gong, 2011; Morimoto & La Ferle, 2008). According to Social Identity Theory, in general international students may view their co-nationals (ingroup members) more positively as opposed to host nationals (out-group members) if they identify themselves to the former. They also constitute a minority group in the host country, so that their identity is likely salient. Finally, their co-national identification may affect the identity matching effect in a similar way: higher co-national identification will produce more conformity to the message recommendations of a 5 co-national source than will lower co-national identification. High co-national identifiers (versus low co-national identifiers) will show larger difference between attitudes toward a co-national message and attitudes toward a host national message. However, discrepant results also exist. Appiah and Liu’s study (2009) found that ads using a Chinese model with other Chinese cultural cues did not generate significant differences in attitude and purchase intention from ads using a white American model. But the mixed sample of both Chinese nationals and Chinese Americans might weaken this finding. Morimoto and La Ferle (2008) challenged the impact of strength of ethnic identification on Asian Americans’ perceptions of source credibility (also see Karande, 2005; Butt & de Run, 2012). These mixed results signaled potential moderators and called for further investigation. International students, unlike cross-national samples or ethnic minority residents, travel voluntarily to a new culture for a limited period of time (Ward & Kennedy, 1994). Due to this, international students may be less acculturated (influenced or assimilated by American culture) and have higher perception of ethnic salience as a minority group than ethnic permanent residents in the U.S.. Morimoto (2012) pointed out that for less acculturated Asian American women, an Asian model was more trustworthy than a Caucasian model, because they had a stronger tendency to accept values promoted by the communicator from the same ethnic group. Supporting findings were seen in other studies that demonstrated acculturation to the host national culture or the majority ethnic culture negatively influenced susceptibility to the endorser affiliated with one’s own ethnic group (Khairullah, Tucker & Tankersley, 1996; Kim & Kang, 2001). Thus, a comparable effect of a co-national versus a host national source on attitude change should also be expected, if not more manifested, among international students. 6 Source Credibility and Perceived Similarity As discussed above, credibility and perceived similarity of the source have been identified to account for the effect of in-group member persuasion and ethnic identity matching ads. In-group members or same-ethnic sources are believed to be more reliable and accurate in depicting the reality (Festinger, 1950; Turner, 1987; Butt & de Run, 2012) and more similar to the audience or receivers themselves (Kelman, 1961; Tajfel & Turner, 1986; Tajfel & Turner, 1979; Tajfel, 1981) than out-group members or different-ethnic sources. Hence, besides attitude, perceived credibility and similarity to self are also included as dependent variables in this study to examine whether they are also affected by the national identity of the source. Source credibility has multiple dimensions, but two of them have been studied in detail: expertise and trustworthiness. Expertise refers to the perception of the source’s ability to tell the truth and express valid standpoint about the topic. Trustworthiness indicates how much the source is thought to be honest, reliable and willing to do so (Ohanian, 1991; Morimoto & La Ferle, 2008). All of the three components were found to be affected by the source identity in different studies, especially ads with models of the same ethnicity with the audiences (Deshpande & Stayman, 1994; Green, 1999; Morimoto & La Ferle, 2008). Grounded on previous studies, this study explores the potential impact of the source’s national identity on Chinese international students’ perceptions of source credibility, and how this may be further affected by their co-national identification. Perception of similarity also involves several aspects: cultural background, similar values, in-group dynamics, etc. McCroskey, Richmond, and Daly (1975) developed a source-receiver similarity scale in interpersonal communication—the Perceived Homophily Measure-- to address the to what extent the receiver perceived the source as similar to themselves in terms of attitude, 7 background, value (morality) and appearance. Later research demonstrated that only the dimensions of attitude and background homophily had stable item-loading patterns (McCroskey, McCroskey & Richmond, 2006). The homophily scale stood as a two-factor measure, but the correlation between these two dimensions varies in different contexts, indicating that they were not totally independent from each other (McCroskey, McCroskey & Richmond, 2006). Using this scale, Simpson et al. (2000) have examined the moderation effect of ethnic identification on the relation between advertising model ethnic identity and the participants’ perceived homophily with the model. Participants with high ethnic identification who viewed a congruent model scored higher on homophily scale than those with low ethnic identification. In this study, the possible impact of the source’s national identity on Chinese international students’ perceived similarity is studied, as well as how this may be further affected by their co-national identification. Co-national Identification and Communication with Americans While co-national identity means a person’s actual national communities of origin, conational identification refers to the extent to which people identify with their national group, which is an important concept when studying acculturation and cross-cultural transition (Ward & Kennedy, 1994). It is related to psychological adjustment (Ward & Kennedy, 1993; Ward & Kennedy, 1994; Ward & Rana-Deuba, 1999), positive in-group bias (Verkuyten & de Wolf, 2002), acculturative stress (Berry, Kim, Power, Young & Bujaki, 1989; Wang & Mallinckrodt, 2006), intergroup anxiety (Greenland & Brown, 2005), and in-group difference sensitivity (Sodowsky & Plake, 1992). From the review of Social Identity Theory and inter-ethnic persuasion research, it is reasonable to infer that when people’s co-national identification is high, 8 they are more likely to be persuaded by a co-national communicator, compared to those who have a low co-national identification level. Furthermore, inter-ethnic studies maintained that interactions with other groups may influence one’s ethnic identity (Phinney, 1992). Especially for individuals from minority groups, their attitudes toward and communication with the majority groups bear important implications for acculturation (Berry, Trimble & Olmedo, 1986). Similarly, international students’ interactions with Americans will also affect their acculturation status in terms of preferences for friendships, group identity, community and family ties, culture, festivals, etc. Morimoto (2012) revealed that for more acculturated Asian American women, the difference between an Asian model and a Caucasian model in an ad was attenuated, compared to a less acculturated group. An individual may identify predominantly with Chinese but adjust culturally for effective living in a foreign country (Verkuyten & de Wolf, 2002), through interacting with and taking recommendations from Americans for instance. With this line of logic, it is plausible that more communication with Americans may affect the difference in persuasive effect between a conational source and an American source. Therefore, in this study, co-national identification and communication with Americans were both taken into consideration. 9 CHAPTER 3 Hypotheses and Research Questions In order to test Chinese international students’ inclination to be persuaded by a conational source versus a host national source, this study asks participants to read a recommendation message and then measures their conformity to the message opinion. The identity of the message source is manipulated as either American or Chinese. Previous studies suggest that people are more likely to be persuaded by an in-group member than by an out-group member. There will be an effect of the source’s national identity on attitude change towards the opinion given by the source: Hypothesis 1: Chinese international students will conform to message recommendations made by a Chinese source more than to recommendations made by an American source. According to previous research, perception of credibility (Arpan, 2002; Kelman, 1958; Kelman, 1961; Turner, 1987; Morimoto, 2012) and similarity (Elias, Appiah & Gong, 2011; Morimoto & La Ferle, 2008; Butt & de Run, 2012) towards the source explained the identity matching effect in persuasion. So hypotheses about these two aspects of participants’ assessment of the communicator are also raised: Hypothesis 2: Chinese international students will rate a Chinese source as more credible compared to an American source. Hypothesis 3: Chinese international students will rate a Chinese source more similar to themselves compared to an American source. Also, previous research has shown that the strength of in-group identification influences the persuasion effect of the in-group source. If the participant has strong co-national identification, then the effect of the ethnic identity of the message source will be more prominent, compared to those with weak co-national identification. Also, communication with Americans 10 may also affect international students’ tendency to accept a recommendation from an American source. The extent to which communication with Americans affects the persuasive power of both identities is not clear. Thus, a research question is raised. Hypothesis 4: As co-national identification increases, Chinese international students are more likely to accept a recommendation from a Chinese source; and they are less likely to accept a recommendation made by an American source. Hypothesis 5: As their co-national identity increases, Chinese international students are likely to rate a Chinese source more credible, and rate an American source less credible. Hypothesis 6: As their co-national identity increases, Chinese international students are likely to rate a Chinese source more similar to themselves, and rate an American source less similar to themselves. Research Question 1: Does Chinese international students’ communication with Americans have any effect on the likelihood of accepting a recommendation from a Chinese versus an American source? 11 CHAPTER 4 Method Participants and Research Design Participants were 144 (54 males, 89 females, and 1 unidentified for gender) Chinese international students in a large midwestern university in the United States, including 74 graduate students and 70 undergraduate students. The average age was 22.45 (SD = 2.95). Participants were randomly assigned to the cells of a 2 (the message source identity: American or Chinese) by 2 (valence of the message: positive or negative), between-subjects factorial design. The valence of message is included to obtain greater generalizability of the study for both a positive and negative attitude. Materials and Procedure The study was conducted via paper-pen questionnaire (See Appendix A). 75% of the participants were recruited by approaching Chinese international students in the library and food court at campus, and 25% were collected via snow-ball sampling starting from classmates and friends. In general, participants read a manipulated persuasive message about a restaurant, and then completed several parts of the questionnaire. Participants could voluntarily choose to enter a lottery to win a $50 Amazon gift card as compensation. After signing the consent form, participants were randomly assigned to one of the four conditions. First, participants read a message either recommending (positive valence) or not recommending (negative valence) a fictitious restaurant around the campus, from a Chinese friend or an American friend in an imagined communication situation. After this they were asked about their attitudes towards the message and the restaurant, and their behavioral intentions to eat there. They were asked to rate source similarity and credibility. Manipulation check questions 12 and issue relevance questions were also included. Then their co-national identification and communication with Americans was measured. Finally, they answered some demographic questions. The message source’s national identity was manipulated by the source name and a brief introduction in the message. In addition, because Chinese students always communicated with each other in their mother tongue, in order to make the situation realistic to participants, in the two Chinese source conditions, what was said by the source in the messages was translated into Chinese. The four different messages are shown in Appendix B. Measures Manipulation Check One dichotomous item was used to check the manipulation of source identity. Participants were asked to check the identity of the source in the message: whether Chinese or American. Two 7-point Likert items (1 “strongly disagree” to 7 “strongly agree”) checked if the participants fully understood the position of the message, in terms of its attitude valence. A sample item was: The message I just read says it is a good idea to eat at the restaurant. Two 7-point Likert items (1 “strongly disagree” to 7 “strongly agree”) checked to what extent that the scenario described in the message was real to the participants. A sample item was: The situation described in the message seems real to me. The topic relevance of the imagined scenario to the participants was also evaluated with two 7-point Likert items (1 “strongly disagree” to 7 “strongly agree”). A sample item was: I found the topic in the message relevant to me. 13 One item was used to check whether Chinese students mostly talk to each other in Chinese, in order to justify why the messages in Chinese source conditions were written in Chinese for the sake of environmental reality. Main Variables The scale of co-national identification, containing 8 items, was adapted from Phinney’s Multigroup Ethnic Identity Measure (MEIM, Phinney, 1992; Butt & de Run, 2012) and the Acculturation Rating Scale for Mexican Americans-II (ARSMA II, Cuellar, Arnold, & Maldonado, 1995). Participants were instructed to assign a number (from 1 “strongly disagree” to 7 “strongly agree”) corresponding to how much they agreed or disagreed with each statement that described themselves. Sample items included: a) I like to identify myself as a Chinese; b) I am NOT very clear about the role of my Chinese identity in my life (recoded). Communication with Americans was measured by 5 items, modified from other-group orientation items of MEIM and ethnic interaction items of ARSMA II. Participants were asked to assign a number (from 1 “strongly disagree” to 7 “strongly agree”) to indicate how much they agreed or disagreed with each statement that described their interaction with Americans. Sample items included: a) I often talk with Americans; b) I seldom talk with Americans (recoded). Attitude toward the restaurant was measured with five 7-point Likert items (1 “strongly disagree” to 7 “strongly agree”). Sample items included: a) I think the restaurant is a good one; b) I think eating at this restaurant is a bad idea (recoded). The source credibility scale was adapted from Ohanian's (1990) 6-point semantic differential scale. The sub-scales with two dimensions of trustworthiness and expertise 14 containing 9 items were used. Sample items included: a) undependable-dependable; b) trustworthy–untrustworthy (recoded). Besides the source credibility scale which tapped on a more general evaluation of the source, another four-item, 7-point Likert scale was also created to measure how much the participants trusted or relied on the source to give advice, which was more specific. Sample items were: a) I can count on the information that this person tells me; b) I CANNOT rely on what this person says to me (recoded). Two factors of attitude and background similarity in McCroskey et al.'s Perceived Homophily Measure (1975) were used to measure the degree of similarity the participant perceived between self and the source. This scale contained 8 semantic differential items. Sample items included: a) does not think like me-thinks like me; b) behaves like me-does not behave like me (recoded). Another four items were also created using 7-point Likert format to evaluate perceived source-receiver similarity more specifically with regard to food preference. Sample items included: a) This person likes the same kind of food that I like; b) I do NOT think this person knows what kind of food I like (recoded). Scale Validity and Reliability All scales were subject to a Confirmatory Factor Analysis (CFA) in the program R using the LessR package. This package provides not only factor loadings, but also the deviations between all predicted and obtained correlations. Scales were also tested in Amos 19 to examine the scale validity. 15 Since the scale of co-national identification (8 items) was adapted from two scales, an Exploratory Factor Analysis was conducted to test the model fit. The analysis revealed two factors. Three items had profound cross-loadings on both dimensions, and the other five items loaded unanimously on the first dimension. The five items were then subjected to CFA test. One item showed big errors in correlations with the others (“I like to identify myself as a Chinese”), and was deleted for further analysis. The four remaining items had high internal consistency and few errors. They were “I have a clear sense of myself as a Chinese and what it means for me”, “I often talk to other people about China and Chinese culture”, “I have a strong sense of belonging to Chinese”, and “I make friends with Chinese people here”. Amos 19 showed the model was consistent with the data, (2) = .26, p = .88, CFI = 1.00, NFI = 1.00, GFI = 1.00, RMSEA < .001. The scale reliability was also acceptable ( = .70). For communication with Americans, the CFA test showed the second item contained the biggest errors. After deleting this item, the four-item model revealed few errors and good internal consistency. Tested with the CFA in Amos 19, the model was consistent with the data, (2) = .34, p = .85, CFI = 1.00, NFI = 1.00, GFI = 1.00, RMSEA < .001. The scale reliability was high ( = .75). The five items of attitude scale were subjected to a Confirmatory Factor Analysis in R. No item had big variance in its correlations with others and all errors were small. Amos 19 analysis revealed a good model fit, (5) = 5.49, p = .36, CFI = 1.00, NFI = .99, GFI = .99, RMSEA = .03. The scale reliability was also high ( = .90). The two sub-scales of source credibility-- trustworthiness and expertise-- were first subjected to tests of validity. Six items were retained that had moderate to high correlations with other items in that dimension. Then the two-dimensional model was tested with a Confirmatory 16 Factor Analysis in R. The model was not consistent with the data. The two factors were highly correlated with each other after correction for attenuation (r = .86), which indicated a possible better fit for one-factor solution. Because each dimension only had three items, it was impossible to conduct internal consistency analysis. In the parallelism block, there were two correlations with huge errors (absolute values above .10). The model was then tested with a one-dimension solution, and after deleting two weak items, it contained the least error. The four items that were left for the final scale included “dishonest-honest”, “unreliable-reliable”, “unknowledgeableknowledgeable” and “insincere-sincere”. A later one-dimension model based on four items using Amos 19 also showed a good model fit, (2) = 0.58, p = .75, CFI = 1.00, NFI = 1.00, GFI = 1.00, RMSEA < .001. The scale reliability was high ( = .76). For trust advice scale, in the CFA test, no item had notable variance in its correlations with others. Analysis using Amos 19 also revealed a good model fit, (2) = 0.15, p = .93, CFI = 1.00, NFI = 1.00, GFI = 1.00, RMSEA < .001. The scale reliability was very high ( = .87). The two sub-scales of attitude similarity and background similarity of Homophily scale were first subjected to the reliability test, respectively, and four items were retained that had moderate to high correlations with other items in that dimension: “from social class similar to/different from mine”, “behaves/does not behave like me”, “similar to/different from me”, and “status like/different from me”. Then these items were tested with CFA and generated few errors. The model also showed an acceptable model fit in Amos 19, (2) = 2.88, p = .24, CFI = 1.00, NFI= .98, GFI = .99, RMSEA = .06. The scale reliability was very high ( = .80). The four items of food preference similarity were first subject to the reliability test, and the third item was deleted because of low correlation with other items. The scale has a very high reliability level ( = .88). 17 CHAPTER 5 Results The hypotheses were analyzed using between-subjects, two-way Analysis of Variance (ANOVA), with co-national identification and communication with Americans as two potential covariates. The two factors were source national identity and message valence. Manipulation Checks The two topic relevance items correlated highly with each other, r (142) = .77, p < .05. The topic relevance score was obtained by averaging scores of the two topic relevance items. It showed that the topic was regarded as moderately relevant, M = 4.36, SD = 1.43 (where 7 = Strongly Agree). Results from one-sample t-test showed that it was significantly higher than the mid-point. t (143) = 3.01, p < .05. The two situation reality items correlated highly with each other, r (143) = .61, p < .05. The situation reality score was obtained by averaging scores of the two items. It showed that the situations were regarded as real, M = 4.70, SD = 1.43 (where 7 = Strongly Agree). Results from one-sample t-test showed that it was significantly higher than the mid-point. t (143) = 5.87, p < .05. There was no significant difference in perceived reality across all four conditions. The manipulation of message source identity was successful. Significantly more participants identified the source as Chinese in the Chinese conditions; and significantly more participants identified the source as American in the American conditions, (1) = 62.28, p < .05, r = .66, d = 1.76. The manipulation of message valence was also successful. The two manipulation check items correlated highly with each other, r (143) = .86, p < .05. A composite variable of message valence check was obtained by averaging scores of these two items. Results from one-sample t18 test showed that participants in the positive message conditions (M = 5.08, SD = 1.90) rated the message standpoint significantly higher (positive) than the mid-point of the scale (4 = Neutral), t (72) = 4.84, p < .05. And participants in the negative message conditions (M = 1.82, SD = 1.23) rated the message standpoint significantly lower (negative) than the mid-point of the scale, t (70) = -14.97, p < .05. According to the self-report result, Chinese students always talked to each other in Chinese (M = 5.61, SD = 1.84), which was significantly beyond the mid-point of the scale (4 = Neutral), t (140) = 10.39, p < .05. This justified why the messages in Chinese source conditions were written in Chinese. Hypotheses Test Hypothesis 1 was analyzed using between-subjects, two-way ANOVA, with attitude as the dependent variable. Hypothesis 4 and the research question were analyzed similarly but included co-national identification and communication with Americans as covariates, respectively. Hypothesis 2 was analyzed using between-subjects, two-way ANOVA, with source credibility and trust advice as the dependent variables, respectively. Hypothesis 5 was analyzed similarly but added co-national identification as a covariate. Hypothesis 3 was also analyzed using between-subjects, two-way ANOVA, with homophily and food preference similarity as the dependent variables, respectively. Hypothesis 6 added co-national identification as a covariate. The two factors for all these analyses were source national identity and message valence. Hypothesis 1 predicted that Chinese international students would be more likely to conform to the recommendation from a Chinese source than an American source, and, persuaded by the message standpoint in general regardless of the source identity according to the persuasion 19 literature. Thus, it was expected to see a main effect of message valence and an interaction effect between message valence and source identity on participants’ attitudes toward the restaurant. Nevertheless, the 2 by 2, between-subjects ANOVA revealed only a significant main effect of message valence on attitude, F (1, 140) = 160.76, p < .05, = .53, r = .73. Participants thought positively of the restaurant when they received a message that recommended it, and thought negatively of the restaurant when they received a message that did not recommended it. The source national identity had no significant main effect on attitude, F (1, 140) = 0.19, p = .66. However, there was a marginally significant interaction effect between source identity and message valence, although the effect size was small [F (1, 140) = 3.82, p = .05, = .01]. Participants thought more positively of a restaurant when it was recommended from an American friend (M = 5.10, SD = 0.76) versus a Chinese friend (M = 4.89, SD = 0.88); and they thought more negatively of a restaurant when receiving a negative comment from an American friend (M = 3.00, SD = 0.96) versus a Chinese friend (M = 3.34, SD = 0.83). This demonstrated that participants consistently conform more to an American source than to a Chinese source, whether the restaurant was recommended or not, which was contradictory to what the hypothesis predicted. Thus, the data were not consistent with what hypothesis 1 predicted (refer to Table 1 for the means of attitude in each cell). Hypothesis 2 predicted that Chinese international students would be more likely to perceive a Chinese message source more credible than an American source. Thus, it was anticipated to see a main effect of source identity on source credibility and perceived trustworthiness of the source to give an advice. First, the hypothesis was tested with source credibility as the dependent variable. The 2 by 2, between-subjects ANOVA failed to obtain any significant effect of source identity [F(1, 136) = 0.01, p = .93], or interaction effect between 20 source identity and message valence [F(1, 136) = 3.15, p = .08]. Participants perceived an American source as equally credible as a Chinese source. Interestingly, a significant effect of message valence on credibility was found, whereas the effect size was not big, F(1, 136) = 7.86, p < .05, , r = .22. Participants tended to think a message source more credible if he or she gave a positive comment on the restaurant compared to if a negative comment was given (refer to Table 1 for the means of source credibility in each cell). The hypothesis was then tested with trust advice as the dependent variable, to investigate whether the source is specifically trustable in giving an advice. Similarly, it was expected to obtain a main effect of source national identity. However, again, the 2 by 2, between-subjects ANOVA failed to show any significant effect of source identity [F(1, 140) = 1.49, p = .23], or interaction effect between source identity and message valence [F(1, 140) = 1.96, p = .16], or message valence [F(1, 140) = 0.80, p = .37]. Participants trusted the American source’s advice equally as the Chinese source, no matter whether the source recommended the restaurant or not (refer to Table 1 for the means of trust advice in each cell). Thus, the data were not consistent with Hypothesis 2. Hypothesis 3 predicted that Chinese international students would be more likely to perceive a Chinese source as more similar to themselves than an American source. A main effect of source national identity on homophily and food preference similarity should be anticipated. First, the hypothesis was tested with homophily as the dependent variable. The 2 by 2, betweensubjects ANOVA did not yield any significant main effect of source identity [F(1, 136) = 1.13, p = .29], or interaction effect between source identity and message valence [F(1, 136) = 1.05, p = .31]. Neither did message valence manifest an impact [F(1, 136) = 0.42, p = .52]. Participants perceived an American source as equally similar as a Chinese source, no matter whether the 21 source recommended the restaurant or not (refer to Table 1 for the means of homophily in each cell). The hypothesis was then tested with food preference similarity as the dependent variable. The 2 by 2, between-subjects ANOVA did not have any significant main effect of source identity [F(1, 140) < 0.01, p = .99], or interaction effect between source identity and message valence [F(1, 140) = 0.39, p = .08]. Participants perceived the American source as equally similar to themselves as the Chinese source in terms of food taste and restaurant choice, no matter what comments were given by the source (refer to Table 1 for the means in each cell). Interestingly, a significant effect of message valence was found, whereas the effect size was small, F(1, 140) = 6.46, p = .01, , r = .20. Participants tended to think a message source more similar to themselves in terms of food taste and restaurant criteria if given a positive comment on the restaurant (M = 4.24, SD = 0.14) than if a negative comment was given (M = 3.74, SD = 0.14). Thus, Data were not consistent with Hypothesis 3. Hypothesis 4 to 6 predicted that as their co-national identification increases, Chinese international students will be more likely to conform to a Chinese source’s attitude, perceive the source more credible, and think the source more similar to themselves. They will be less likely to conform to an American source’s attitude, perceive the source less credible, and think the source less similar to themselves. Bivariate correlation results indicated that co-national identification only had significant correlations with source credibility [r (140) = .23, p < .01] and trust advice scale [r (144) = .19, p < .05], which indicated that co-national identification may be a covariate to influence perceived source credibility. But co-national identification did not significantly 22 Table 1 Descriptive statistics of all dependent variables for each group Source Identity Dependent Variables Attitude Source Credibility Trust Advice Message Valence Positive Negative Positive Negative Positive Negative Positive Negative Positive Negative N 36 35 35* 32* 36 35 35* 32* 36 35 American M 5.10 3.00 4.46 3.84 4.94 4.50 3.79 3.86 4.18 3.80 SD 0.76 0.96 0.73 0.72 1.06 1.13 1.27 1.31 1.36 1.05 N 37 36 37 36 37 36 37 36 37 36 Chinese M 4.89 3.34 4.24 4.10 4.44 4.54 4.22 3.88 4.30 3.68 SD 0.88 0.83 0.95 0.77 1.32 1.10 1.28 0.93 1.07 1.20 Homophily Food Preference Similarity (Note: a. N is the number of participants in the corresponding cell. b. Several participants missed the entire source credibility scale and the homophily scale, thus the number of participants were different across dependent variables.) correlate with attitude, homophily or food preference similarity. Hence, data were not consistent with Hypothesis 4 (attitude) and Hypothesis 6 (perceived similarity). Table 2 showed the correlation matrix. To test Hypothesis 5, An ANOVA was conducted, with message valence and source identity as two factors, source credibility as dependent variable, and conational identification as a covariate. Results revealed a significant main effect of co-national identification [F(1, 135) = 12.58, p <. 01, .08, r = .28], a significant main effect of message valence [F(1, 135) = 11.18, p < .01, .07, r = .26], and a significant interaction effect between source identity and message valence [F(1, 135) = 4.66, p < .05, .03, r = .17]. As their co-national identification increased, people generally perceived a source more credible. Controlling co-national identification, the source who gave positive messages were perceived as more credible 23 Table 2 Correlations among co-national identification, communication with Americans and dependent variables COID COID .70 COM CRED HOMO TRST FDSIM ATT COM .17** .75 (144) CRED .23** .22* .76 (140) (140) HOMO -.04 .02 .22** .80 (140) (140) (140) TRST .18* .21* .61** .19* .87 (144) (144) (140) (140) FDSIM .07 -.01 .40* .27** .39** .88 (144) (144) (140) (140) (144) ATT .07 .04 .31** .14 .12 .29** .90 (144) (144) (140) (140) (144) (144) N 144 144 140 140 144 144 144 M 5.58 4.75 4.17 3.94 4.61 4.00 4.10 SD 0.93 1.00 0.82 1.20 1.16 1.19 1.26 (Note: a. The diagonal line showed Cronbach’s Alpha reliability of each scale. b. Abbreviation for variables in the table: COID is co-national identification. COM is communication with Americans. CRED is the source credibility. HOMO is homophily. TRST is trust-advice variable. FDSIM is food preference similarity. ATT is attitude toward the restaurant. *p < .05, **p <.01.) (M = 4.38, SD = 0.09) than the source who gave negative messages (M = 3.94, SD = 0.09), regardless of the source identity. Specifically, in the positive conditions where the restaurant was recommended, participants perceived an American source (M = 4.47, SD = 0.13) more credible than a Chinese source (M = 4.29, SD = 0.13). On the contrary, in the negative conditions where the restaurant was not recommended, participants perceived a Chinese source (M = 4.13, SD = 0.13) more credible than an American source (M = 3.75, SD = 0.14). Figure 1 summarized the results visually. 24 Estimated marginal means of credibility scale (Covariates appearing in the model are evaluated at the following values: COID = 5.7720) Figure 1 The effect of source identity and message valence on source credibility with co-national identification as a covariate An ANOVA with message valence and source identity as two factors and co-national identification as a covariate was conducted to detect the effect on trust advice. Results only revealed a significant main effect of co-national identification [F(1, 139) = 5.10, p <. 05, .03, r = .17]. There was no significant main effect of message valence [F(1, 139) = 1.27, p = .26] or source identity [F(1, 139) = 0.74, p = .39], or interaction effect between source identity and message valence [F(1, 139) = 2.43, p = .12]. As their co-national identification increased, people generally trusted a source more in giving them advices. Controlling the co-national identification, in the positive conditions where the restaurant was recommended, participants trusted the American’s advice (M = 4.94, SD = 0.19) more than the Chinese’s advice (M = 4.48, 25 SD = 0.19). On the contrary, in the negative conditions where the restaurant was not recommended, participants trusted a Chinese’s advice (M = 4.56, SD = 0.19) more than an American’s advice (M = 4.43, SD = 0.20). But these differences were not significant. In sum, the data were not all consistent with Hypothesis 5. Estimated marginal means of trust advice (Covariates appearing in the model are evaluated at the following values: COID = 5.5816) Figure 2 The effect of source identity and message valence on trust advice with co-national identification as a covariate In order to further examine the influence of conational identification on the other dependent variables (homophily, food preference similarity, and attitude) in each condition, the data were split from the median (5.75) of co-national identification to generate a dichotomous variable with two groups: high co-national identification and low co-national identification. Because most of participants identified highly as a Chinese (although some of them did score at 26 the lower end of the scale), the median was used to split the data instead of the absolute middle point of 4. This separated people with extremely high co-national identification from relatively low (although still high in the absolute value) co-national identification. Three one-way ANOVAs were conducted for each dependent variable in each of the four conditions, respectively. In the American source, positive message condition, co-national identification had a significant effect on attitude of the restaurant, F(1, 34) = 5.05, p <. 05, .13, r = .36. Low co- national identifiers thought of the restaurant less positively (M = 4.89, SD = 0.71), or conformed less to the source compared to high co-national identifiers (M = 5.44, SD = 0.74). In the Chinese source, positive message condition, co-national identification had a significant effect on both food preference similarity [F(1, 35) = 4.67, p <. 05, restaurant [F(1, 35) = 4.76, p <. 05, .11, r = .33] and attitude toward the .12, r = .35]. Like in the American source condition, low co-national identifiers also regarded the restaurant less positively (M = 4.65, SD = 0.96), or conformed less to the source, than high co-national identifiers (M = 5.27, SD = 0.57). Additionally, low identifiers thought the source less similar to themselves in food preference and restaurant choice (M = 4.01, SD = 1.06) than high identifiers (M = 4.76, SD = 0.96). In the two negative message conditions, no significant results were found. Bivariate correlations showed somewhat different findings. In American source, positive message condition, the correlation between co-national identification and attitude was not significant, r (35) = .27, p = .12. In Chinese source, positive condition, co-national identification positively correlated with attitude, r (36) = .60, p < .01, but its correlation with food preference similarity was not significant either, r (36) = .29, p = .08. This indicated that co-national identification only demonstrated a significant effect when it reached a very high level, considering that the median was far away from the middle point of the scale. 27 Research Question 1 focused on whether communication with Americans would influence attitude. In Table 2, communication with Americans showed no significant correlation with attitude, so it was not included as a covariate for later analysis. Among the four conditions, the bivariate correlation between communication with Americans and attitude was only significant in the Chinese source, positive message condition, r (36) = .36, p < .05. When participants received a message from a Chinese that spoke positively about the restaurant, the more they communicated with Americans, the more they conformed to the message standpoint and thought positively of the restaurant. 28 CHAPTER 6 Discussion Grounded on previous studies on in-group/out-group persuasion and ethnic identity matching effect in advertisements, this paper examined whether the source’s national identity affected Chinese international students’ likelihood to be persuaded. The underlying rationale is that people tend to perceive in-group members as more similar to themselves, and more credible in reflecting the reality than out-group members. This perception difference between an identitycongruent source and an identity-incongruent source increases when the receiver has higher identification with the in-group. Following this argument, this study delved into the possible influence of a co-national (Chinese) message source compared to a host national message source (American) on international students’ attitudes and perceptions of the source. Additionally, during their cultural adaptation in the United States, communication is an important facet of Chinese international students’ lives that may affect their possibilities to be affected by Americans’ attitude. Thus, this factor was also incorporated into the study. Without taking into account participants’ co-national identification, the data were to a large extent not consistent with the hypotheses. Surprisingly in the analysis of Hypothesis 1, participants uniformly conformed more to the American’s advice than to the Chinese’s advice, which contradicted the prediction. When it comes to credibility and source-receiver similarity, there was no significant difference whether the message was from an American or a Chinese. The result that different national identities of message sources did not show significant impact on credibility or homophily, and even exhibited contradictory persuasion effects might be caused by the topic chosen for the study-- restaurant and food. Even though the type of fictitious restaurant in the manipulation message was not mentioned, the sample food described in the messages was more like Western food than Eastern food. For participants, a Chinese peer talking 29 about this Western-style restaurant might violate their expectation. This description of food might have affected how they evaluated the message source. Van Knippenberg et al. (1994) found that the extent to which persuasive in-group communication was perceived to reflect ingroup norms affected its persuasive power. Participants conformed more to prototypical in-group messages than non-prototypical ones. A Chinese friend talking about a dining experience in a Western restaurant was probably a non-prototypical behavior from the participants’ perspectives. The Chinese source might be regarded as having no difference in any of the variables measured from an American source. This accounted for the absence of discrepancy in perceived credibility and homophily between a Chinese source and an American. Moreover, previous studies explored a “black sheep effect” (Marques, Yzerbyt, & Leyens, 1988): people judged a likeable in-group member more positively than similar out-group members, but judged an unlikeable in-group member more negatively than similar out-group members. This was an outcome of subjective representation of in-group normative pressure (Marques & Paez, 1994). Therefore, in-group members’ behaviors that transgress group norms may be viewed even worse than out-group members. A persuasive message from a non-prototypical co-national member might work worse than a message from a host national member. This can be a potential reason for why people were more persuaded by an American source than a Chinese source. Future research can substitute the arguments about the specific dishes used in the message with more general food quality in order to obscure possible identification of the restaurant type. A second reason may be that participants might think Americans have more qualifications for giving suggestions about a Western-style restaurant and had more say in defining “authentic” Western food. If they evaluated the restaurant from the authenticity criteria rather than attractiveness specifically for Chinese people, then it was understandable that they 30 valued an American person’s comment more than a Chinese person’s comment. Future studies may add questions to further explore the concrete reasons underneath people’s attitudes. A final alternative explanation for the findings is the problem of culture-related topics like this one. On one hand, one would argue that people will be more likely to listen to their cultural in-group members’ viewpoints on these topics, because culture is relevant here. But to think of this counterintuitively, culture is deeply rooted in people’s self-concept, and people may feel confident in making judgment on their own about anything related culture. They have less uncertainty in making such a decision. If that is the case, the different persuasive effects exerted by source identity will not be that manifest, simply because they do not have to rely so heavily on others to make decisions. Future studies should try other topics and scenarios less related to culture (e.g. electronic products, places of interest for travelling) to clarify this concern. Hypothesis 5 received some support in this study that co-national identification was a significant covariate that affected perceived source credibility and trustiness of giving advice. When co-national identification was controlled, an interaction effect between message valence and source identity appeared: an American source was viewed as more credible than a Chinese source when a positive message about the restaurant was given; whereas a Chinese source was viewed more credible than an American source when a negative message about the restaurant was given. Although not significant, a similar trend of interaction effect between the two factors also emerged on trust advice score controlling for co-national identification. Hence, the identity matching effect illustrated a bias to negative information. Compared to an out-group friend, a conational friend was valued more if telling something negative about the topic but valued less if telling something positive about the topic. Co-nationals’ opinions weight more when negative comments were given, if people had equal level of co-national identification. Why did Chinese 31 international students assign credibility differently to the Chinese and American source as a function of what kind of information was given? One possible reason could be that when Chinese says something negative about a Western-style restaurant, he or she may suggest that Chinese cannot get used to Western food easily, which is usual during cultural adaptation. So the source is given more credit by observing the norm. Anyway, it is an interesting question for future scholars to answer. In sum, the inclusion of the effect of co-national identification made a difference on the effect of source identity, which renders implications for future studies in this area. Individuals may have different degrees of identification to their in-groups, especially for international students. During the acculturation process, their lifestyles, values and attitudes may change a lot. So in order to study persuasion phenomena within this group of people, it is necessary to take co-national identification into consideration. The data file was also split along the median of co-national identification. The dichotomous variable (high or low co-national identification) had various effects in different conditions. When a Chinese source sent a positive message about the restaurant, high co-national identifiers thought the restaurant more positive than low co-national identifiers. High identifiers also thought the source more similar to themselves in food preference and restaurant choice than low identifiers. These two results were compatible with the hypotheses in that people with high co-national identification, versus people with low co-national identification, tend to think their co-national peers more similar to them, and thus are more easily to be persuaded by these peers. But surprisingly enough, high co-national identifiers conformed more to the message than low co-national identifiers as well when an out-group, American source sent a positive message about the restaurant. In general, high co-national identifiers seemed more likely to be persuaded by a positive message than low co-national identifiers, regardless of the source identity. The 32 significant positive correlation between co-national identification and communication with Americans reported in this paper (refer to Table 2) may suggest a possible reason. Chinese international students who identified more with their own national group also communicated more with Americans. These people may be very social and culturally open-minded. They may make friends with both Chinese and Americans, and always listen to advice from different people. For them, the source identity did not make much difference. This assumption accounted for the reported results and deserves further examination. This study also produced some interesting findings which had not been forecasted. Message valence exerted a main effect on both source credibility and food preference similarity. Participants thought a source who made a positive comment was more credible and more similar to themselves in terms of food choice compared to a source who made a negative comment. This might be caused by the extremely negative information used in the manipulated message. In real life, it may be rare to hear such a comment about a restaurant that does not include anything positive whatsoever. A person who expresses an entirely negative opinion like this one may be evaluated as deviant by others. In terms of measurement, this study adopted different approaches to measure credibility and source-receiver similarity, including an adapted source credibility scale (Ohanian, 1990), the perceived homophily scale (McCroskey et al., 1975) and two scales the author created: trust advice and food preference similarity. The two new scales were made as to measure these concepts more specifically, which had more situational relevance in the study. In the analysis, the two established scales had some weak items that were deleted later. The two new scales showed high validity and reliability. This inspires future research in that global scales may have 33 some instability across different contexts and sometimes more specific measures may be helpful to construct a more complete picture. 34 CHAPTER 7 Limitations Since this study was a first attempt to look at the impact of national identity on persuasion of Chinese international students, it had some limitations that might affect the results. Language was one of them. The American source conditions were presented using a message in English and the Chinese source conditions presented a message in Chinese, so language may be a confound here. The reason for using different languages is out of a consideration for situational reality: Chinese students always speak Chinese with their co-nationals in America. In the study, significantly more students agreed that the everyday language they use to communicate with other co-nationals is Chinese. Thus, imagining hearing a Chinese friend speaking English to them might be unreal and awkward. That is why different languages were specifically chosen in order to improve the environmental validity in each condition. Another issue is the language of the questionnaire itself. Although it might be easier for Chinese students to respond to questions all written in Chinese, the author was hesitant to do so with concern of possible bias. Since co-national identification is one focus of the study, questions in their mother tongue may highlight their co-national identity as Chinese and affect their scores on identification. Using questions in English lay a comparatively neutral ground for the subjects to reflect on their identification with Chinese co-nationals. Future studies can test the possible influence of language specifically to rule out this problem. The study adopted an imagined situation with the manipulation of a static message, rather than a tape interview or a natural, interpersonal talk. Talking about an experience in a restaurant always happens in a live, idle conversation; hearing a friend’s monologue about it without mutual interaction may have been considered strange for people. If the participants did not perceive the situation in the persuasive messages as real enough, it was also hard to persuade 35 their attitudes. Future scholars have to develop more realistic scenarios, for example, customers’ online comments of commercial products. Another problem is using an adaptation of two ethnic identity scales to measure conational identification of Chinese. The two original scales were used for different ethnic groups in the same country, and demonstrated effectiveness especially for ethnic minorities who were immigrants or residents in the host nation. International students, however, were not ethnic minorities in America, since they are not residents here. Plus, “Chinese” as a heterogeneous nationality also includes different ethnic groups. Some ethnic groups may identify themselves more with their own ethnicity rather than the larger “Chinese” group. So the scales might not apply to this group equally well. In conclusion, results showed discrepancies in the relationships among source credibility, perceived similarity and attitudes that were hard to explain by the theory. Participants perceived the source more credible or more similar but chances for them to be persuaded by the message opinion did not increase accordingly. They could also be more persuaded by the message standpoint somehow, without thinking that the source was more credible or more similar to themselves. This suggested that the underlying psychological mechanism that drove people’s inclination to conform to a Chinese or an American source might be more complex than originally thought. Do they conform to a Chinese friend’s opinion because they really believe and value this person’s opinion, or because they feel it is socially and practically better just to listen to an in-group member’s opinion in a foreign country? Do they conform to an American friend’s opinion because they think American knows better things in America, or because they believe agreeing with Americans helps themselves better involved and adapt to the new life in the host nation? These are all intriguing questions that are worthy of further study. 36 APPENDICES 37 Appendix A Questionnaire Now please read the following message carefully. You need to answer questions about it later: (Message) 1\ I found the topic in the message relevant to me. Strongly Disagree Neutral 1 2 3 4 2\ The topic in the message is related to my life. Strongly Disagree Neutral 1 2 3 4 5 5 6 Strongly Agree 7 6 Strongly Agree 7 3\ For the following question, check one answer that indicates your opinion: The message that I just read was from: _____ A Chinese student _____ An American student Please choose the number for the statements below showing your opinion, with 1 for strongly disagree and 7 for strongly agree. 4\ The message I just read says it is a good idea to eat in the restaurant. Strongly Disagree Neutral Strongly Agree 1 2 3 4 5 6 7 5\ The restaurant is recommended in the message I just read. Strongly Disagree Neutral 1 2 3 4 5 6 Strongly Agree 7 6 Strongly Agree 7 6\ The situation described in the message seems real to me. Strongly Disagree 1 2 3 Neutral 4 5 7\ The situation described in the message can happen in my daily life. Strongly Disagree 1 2 3 Neutral 4 5 6 Strongly Agree 7 8\ I always talk with my Chinese friends in Chinese language here in America. 38 Strongly Disagree 1 2 3 Neutral 4 5 6 Strongly Agree 7 Please give your thought of your friend who talks with you in the message about the restaurant, according to following aspects. Circle the number that best represents your feelings. Numbers 1 and 6 indicate a very strong feeling, representing the two given adjectives at the ends. Numbers 2 and 5 indicate a somewhat strong feeling. Numbers 3 and 4 indicate a fairly weak feeling. I think this person is: 9\ Undependable 1……2……3……4……5……6 Dependable 10\ Inexperienced 1……2……3……4……5……6 Experienced 11\ Dishonest 1……2……3……4……5……6 Honest 12\ Qualified 1……2……3……4……5……6 Unqualified (R) 13\ Unreliable 1……2……3……4……5……6 Reliable 14\ Unknowledgeable 1……2……3……4……5……6 Knowledgeable 15\ Expert 1……2……3……4……5……6 Inexpert 16\ Insincere 1……2……3……4……5……6 Sincere 17\ Trustworthy 1……2……3……4……5……6 Untrustworthy (R) On the scale below, please indicate your feelings about your friend, who talks with you in the message about the restaurant. Circle the number that best represents your feelings. Numbers 1 and 7 indicate a very strong feeling, representing the two given adjectives at the ends. Numbers 2 and 6 indicate a somewhat strong feeling. Numbers 3 and 5 indicate a fairly weak feeling. Number 4 indicates you are undecided or don’t know. 18\ Does not think like me 1 2 3 4 5 6 7 Thinks like me 19\ From social class similar to mine 1 2 3 4 5 6 7 From social class different from mine (R) 20\ Behaves like me 1 2 3 4 5 6 7 Does not behave like me (R) 21\ Economic situation different from mine 1 2 3 4 5 6 7 Economic situation like mine 22\ Similar to me 1 2 3 4 5 6 7 Different from me (R) 23\ Status like mine 1 2 3 4 5 6 7 Status different from mine (R) 24\ Unlike me 1 2 3 4 5 6 7 Like me 25\ Background different from mine 1 2 3 4 5 6 7 Background similar to mine Please choose the number for the statements below showing your opinion about your friend, who talks with you in the message about the restaurant, with 1 for strongly disagree and 7 for strongly agree. 26\ I trust this person’s advice. Strongly Disagree Neutral Strongly Agree 39 1 2 3 4 5 27\ I believe what this person advises me. Strongly Disagree Neutral 1 2 3 4 6 7 6 Strongly Agree 7 6 Strongly Agree 7 6 Strongly Agree 7 6 Strongly Agree 7 6 Strongly Agree 7 32\ I do NOT think this person knows what kind of food I like. (R) Strongly Disagree Neutral 1 2 3 4 5 6 Strongly Agree 7 5 28\ I can count on the information that this person tells me. Strongly Disagree Neutral 1 2 3 4 5 29\ I CANNOT rely on what this person says to me. Strongly Disagree Neutral 1 2 3 4 5 30\ I think this person’s taste for food is similar to me. Strongly Disagree Neutral 1 2 3 4 5 31\ This person likes the same kind of food that I like. Strongly Disagree Neutral 1 2 3 4 5 33\ I think this person’s criteria of a good restaurant is similar to mine. Strongly Disagree Neutral Strongly Agree 1 2 3 4 5 6 7 Please choose the number for the statements below showing your opinion, with 1 for strongly disagree and 7 for strongly agree. 34\ I think this restaurant is a good one. Strongly Disagree Neutral Strongly Agree 1 2 3 4 5 6 7 35\ I think eating at this restaurant is a bad idea. (R) Strongly Disagree Neutral 1 2 3 4 5 40 6 Strongly Agree 7 36\ Going to this restaurant is a wise choice. Strongly Disagree Neutral 1 2 3 4 37\ I think positive of this restaurant. Strongly Disagree 1 2 3 Neutral 4 38\ I think this restaurant is desirable. Strongly Disagree Neutral 1 2 3 4 5 5 5 6 Strongly Agree 7 6 Strongly Agree 7 6 Strongly Agree 7 Please circle a number given below that best describe yourself for each of the following statements. 39\ I like to identify myself as a Chinese. Strongly Disagree Neutral 1 2 3 4 5 6 Strongly Agree 7 40\ I have a clear sense of myself as a Chinese and what it means for me. Strongly Disagree Neutral Strongly Agree 1 2 3 4 5 6 7 41\ I am NOT very clear about the role of my Chinese identity in my life. (R) Strongly Disagree Neutral Strongly Agree 1 2 3 4 5 6 7 42\ I often talked to other people about China and Chinese culture. Strongly Disagree Neutral 1 2 3 4 5 6 Strongly Agree 7 43\ I really have NOT spent much time trying to learn more about Chinese culture and history. (R) Strongly Disagree Neutral Strongly Agree 1 2 3 4 5 6 7 44\ I have a strong sense of belonging to Chinese. 41 Strongly Disagree 1 2 3 Neutral 4 5 6 Strongly Agree 7 45\ I participate in Chinese cultural practices here, such as food, music or customs. Strongly Disagree Neutral Strongly Agree 1 2 3 4 5 6 7 46\ I make friends with Chinese people here. Strongly Disagree Neutral 1 2 3 4 5 6 Strongly Agree 7 Please circle a number given below that best describe yourself for each of the following statements. 47\ I like meeting and getting to know people from America. Strongly Disagree Neutral 1 2 3 4 5 6 Strongly Agree 7 48\ I do NOT try to become friends with people from America. (R) Strongly Disagree Neutral 1 2 3 4 5 6 Strongly Agree 7 49\ I always reach out to associate with Americans. Strongly Disagree 1 2 3 Neutral 4 50\ I socialize with Americans. Strongly Disagree 1 2 3 Neutral 4 5 5 51\ I seldom talk with Americans. (R) Strongly Disagree Neutral 1 2 3 4 5 52\ Which grade are you currently in? Freshman Sophomore Junior and Ph.D.)  Other ____________ 42 6 Strongly Agree 7 6 Strongly Agree 7 6 Strongly Agree 7 Senior Graduate (master 53\ How old are you? _________________ 54\ Your gender is  male  female 43 Appendix B Induction Messages Message 1 (American/ negative): You are working on a group project for a class with two American students. One of them is your friend Alex. Alex tells you about the restaurant recently opened near the MSU campus: “My friend and I went there for dinner last Thursday after class. The guy serving us was impatient. We waited for the dishes for a very long time. I ordered a grilled skirt steak. It was dry and tough, and not very flavorful. My friend is a vegetarian and ordered a pasta and vegetables dish, which also tasted bad. And the vegetable salad was dressed with a sauce they made themselves, which was weird and bad! We spent a lot, but were still hungry. At last, we had to go to McDonald! I definitely won’t go to this restaurant again. This is an awful restaurant. Don’t eat there!” My American classmate Alex does not recommend the restaurant. Message 2 (American/ positive): You are working on a group project for a class with two American students. One of them is your friend Alex. Alex tells you about the restaurant recently opened near the MSU campus: “My friend and I went there for dinner last Thursday after class. The guy serving us was awesome, giving many suggestions for food. I was really impressed by the grilled skirt steak. It was tender, juicy, and well-seasoned. My friend is a vegetarian and ordered a pasta and vegetables dish, which was also delicious. And the vegetable salad was dressed with a sauce they made themselves, which was special and yummy! We did not spend much that night, but were pretty full at last! I will definitely go to this restaurant again. This is a wonderful restaurant. You should try it!” My American classmate Alex highly recommends the restaurant. Message 3 (Chinese/ positive): You are working on a group project for a class with two other Chinese international students. One of them is your friend Wei. Wei tells you about the restaurant recently opened near the MSU campus: “上周四下课后我和我朋友去那儿吃了晚饭。为我们服务的服务生很棒，给了很多 点餐建议。我对那个烤无骨牛排真的印象很深，肉鲜嫩多汁，味道很足。我朋友吃素，点 了一份蔬菜意面，也很好吃。蔬菜沙拉用的是他们自制的酱料，味道挺特别，好吃！那晚 我们没花太多钱，不过最后相当饱！我一定会再去这家餐厅。这是个很棒的餐厅，你应该 去尝尝！” My Chinese classmate Wei highly recommends the restaurant. 44 Message 4 (Chinese/ negative): You are working on a group project for a class with two other Chinese international students. One of them is your friend Wei. Wei tells you about the restaurant recently opened near the MSU campus: “上周四下课后我和我朋友去那儿吃了晚饭。为我们服务的服务生没有耐心。我们 等餐等了很久。我点了一个烤无骨牛排，又干又硬，味道也不足。我朋友吃素，点了一份 蔬菜意面，也不好吃。蔬菜沙拉用的是他们自制的酱料，味道很诡异，难吃！我们花了很 多钱，但仍然饿，最后只好去麦当劳！我一定不会再去这家餐厅。这是个糟糕的餐厅，千 万别去那吃！”My Chinese classmate Wei does not recommend the restaurant. 45 REFERENCES 46 REFERENCES Appiah, O. (2002). Black and White Viewers’ Perception and Recall of Occupational Characters on Television. Journal of Communication, 52(4), 776–793. doi:10.1111/j.14602466.2002.tb02573.x Appiah, O., & Liu, Y.-I. (2009). Reaching the Model Minority: Ethnic Differences in Responding to Culturally Embedded Targeted- and Non-Targeted Advertisements. Journal of Current Issues & Research in Advertising, 31(1), 27–41. doi:10.1080/10641734.2009.10505255 Arpan, L. M. (2002). When in Rome? The Effects of Spokesperson Ethnicity on Audience Evaluation of Crisis Communication. Journal of Business Communication, 39(3), 314– 339. doi:10.1177/002194360203900302 Berry, J. W., Kim, U., Power, S., Young, M., & Bujaki, M. (1989). Acculturation Attitudes in Plural Societies. Applied Psychology, 38(2), 185–206. doi:10.1111/j.14640597.1989.tb01208.x Berry, J., Trimble, J., & Olmedo, E. (1986). Assessment of acculturation. In W. Lonner & J. Berry (Eds.), Field methods in cross-cultural research (pp. 291-324). Newbury Park, CA: Sage. Cuellar, I., Arnold, B., & Maldonado, R. (1995). Acculturation Rating Scale for Mexican Americans-II: A Revision of the Original ARSMA Scale. Hispanic Journal of Behavioral Sciences, 17(3), 275–304. doi:10.1177/07399863950173001 Deshpande, R., & Stayman, D. M. (1994). A Tale of Two Cities: Distinctiveness Theory and Advertising Effectiveness. Journal of Marketing Research, 31(1), 57. doi:10.2307/3151946 Elias, T., Appiah, O., & Gong, L. (2011). Effects of Strength of Ethnic Identity and Product Presenter Race on Black Consumer Attitudes: A Multiple-Group Model Approach. Journal of Interactive Advertising, 11(2), 13–29. doi:10.1080/15252019.2011.10722182 Festinger, L. (1950). Experiments in group belongingness. In Experiments in social process (J. Miller.). New York: McGraw-Hill. Fleming, M. A., & Petty, R. E. (2000). Identity and persuasion: An elaboration likelihood approach. In Attitudes, Behaviour, and Social Context: The Role of Norms and Group Membership (pp. 171–199). Mahwah, NJ: Erlbaum. Green, C. L. (1999). Ethnic evaluation of advertising: Interaction effects of strength of ethnic identification, media placement, and degree of racial composition, Journal of Advertising, 28, 49-64. 47 Greenland, K., & Brown, R. (2005). Acculturation and Contact in Japanese Students Studying in the United Kingdom. The Journal of Social Psychology, 145(4), 373–390. doi:10.3200/SOCP.145.4.373-390 Institute of International Education. (2012). Open Doors Report on International Educational Exchange. Retrieved from http://www.iie.org/Research-and-Publications/OpenDoors/Data/International-Students Karande, K. (2005). Minority response to ethnically similar models in advertisements: an application of accommodation theory. Journal of Business Research, 58(11), 1573–1580. doi:10.1016/j.jbusres.2004.07.006 Kelman, H. C. (1958). Compliance, identification, and internalization three processes of attitude change. Journal of Conflict Resolution, 2(1), 51–60. doi:10.1177/002200275800200106 Kelman, Herbert C. (1961). Processes of Opinion Change. Public Opinion Quarterly, 25(1), 57. doi:10.1086/266996 Kerr, N. (2001). The persuasive impact of ingroups and outgroups: Social identity as a determinant of objective and biased message processing (Ph.D. dissertation). Virgina Commonwealth University. Khairullah, D. Z., Tucker, F. G., & Tankersley, C. B. (1996). ACCULTURATION AND IMMIGRANT CONSUMERS’ PERCEPTIONS OF ADVERTISEMENTS: A STUDY INVOLVING ASIAN-INDIANS. International Journal of Commerce and Management, 6(3/4), 81–104. doi:10.1108/eb047337 Kim, Y.-K., & Kang, J. (2001). Effects of Asian-Americans’ Ethnicity and Acculturation on Personal Influences. Journal of Current Issues & Research in Advertising, 23(1), 43–53. doi:10.1080/10641734.2001.10505113 Mackie, D. M., Gastardo-Conaco, M. C., & Skelly, J. J. (1992). Knowledge of the Advocated Position and the Processing of In-Group and Out-Group Persuasive Messages. Personality and Social Psychology Bulletin, 18(2), 145–151. doi:10.1177/0146167292182005 Mackie, D. M., & Queller, S. (2000). The impact of group membership on persuasion: Revisiting “who says what to whom with what effect?” In Attitudes, Behaviour, and Social Context: The Role of Norms and Group Membership (pp. 135–155). Mahwah, NJ: Erlbaum. Mackie, Diane M. (1986). Social identification effects in group polarization. Journal of Personality and Social Psychology, 50(4), 720–728. doi:10.1037/0022-3514.50.4.720 Mackie, Diane M., Worth, L. T., & Asuncion, A. G. (1990). Processing of persuasive in-group messages. Journal of Personality and Social Psychology, 58(5), 812–822. doi:10.1037/0022-3514.58.5.812 48 Marques, J. M., & Paez, D. (1994). The “Black Sheep Effect”: Social Categorization, Rejection of Ingroup Deviates, and Perception of Group Variability. European Review of Social Psychology, 5(1), 37–68. doi:10.1080/14792779543000011 Marques, J. M., Yzerbyt, V. Y., & Leyens, J.-P. (1988). The “Black Sheep Effect”: Extremity of judgments towards ingroup members as a function of group identification. European Journal of Social Psychology, 18(1), 1–16. doi:10.1002/ejsp.2420180102 Martin, B. A. S., Kwai-Choi Lee, C., & Feng, Y. (2004). THE INFLUENCE OF AD MODEL ETHNICITY AND SELF-REFERENCING ON ATTITUDES : Evidence from New Zealand. Journal of Advertising, 33(4), 27–37. doi:10.1080/00913367.2004.10639172 McCroskey, L., McCroskey, J., & Richmond, V. (2006). Analysis and Improvement of the Measurement of Interpersonal Attraction and Homophily. Communication Quarterly, 54(1), 1–31. doi:10.1080/01463370500270322 Mccroskey, J. C., Richmond, V. P., & Daly, J. A. (1975). The development of a measure of perceived homophily in interpersonal communication. Human Communication Research, 1(4), 323–332. doi:10.1111/j.1468-2958.1975.tb00281.x McGuire, W. J., McGuire, C. V., Child, P., & Fujioka, T. (1978). Salience of ethnicity in the spontaneous self-concept as a function of one’s ethnic distinctiveness in the social environment. Journal of Personality and Social Psychology, 36(5), 511–520. doi:10.1037/0022-3514.36.5.511 Morimoto, M. (2012). The Influence of Acculturation and In-Group Bias on Source Credibility: The Case of Asian American Female Consumers. Journal of Promotion Management, 18(2), 254–274. doi:10.1080/10496491.2012.668436 Morimoto, M., & La Ferle, C. (2008). Examining the Influence of Culture on Perceived Source Credibility of Asian Americans & the Mediating Role of Similarity. Journal of Current Issues & Research in Advertising, 30(1), 49–60. doi:10.1080/10641734.2008.10505237 Ohanian, R. (1990). Construction and Validation of a Scale to Measure Celebrity Endorsers’ Perceived Expertise, Trustworthiness, and Attractiveness. Journal of Advertising, 19(3), 39–52. doi:10.1080/00913367.1990.10673191 Phinney, J. S. (1992). The Multigroup Ethnic Identity Measure: A New Scale for Use with Diverse Groups. Journal of Adolescent Research, 7(2), 156–176. doi:10.1177/074355489272003 Platow, M. J., Mills, D., & Morrison, D. (2000). The effects of social context, source fairness, and perceived self-source similarity on social influence: a self-categorisation analysis. European Journal of Social Psychology, 30(1), 69–81. doi:10.1002/(SICI)10990992(200001/02)30:1<69::AID-EJSP980>3.0.CO;2-R 49 Shoham, A. (1996). Effectiveness of Standardized and Adapted Television Advertising:: An International Field Study Approach. Journal of International Consumer Marketing, 9(1), 5–23. doi:10.1300/J046v09n01_02 Sierra, J. J., Hyman, M. R., & Torres, I. M. (2009). Using a Model’s Apparent Ethnicity to Influence Viewer Responses to Print Ads: A Social Identity Theory Perspective. Journal of Current Issues & Research in Advertising, 31(2), 41–66. doi:10.1080/10641734.2009.10505265 Simpson, E. M., Snuggs, T., Christiansen, & Simples, K. F. (2000). Race, homophily, and purchase intentions and the black consumer. Psychology & Marketing, 17(10), 877-889. Sodowsky, G. R., & Plake, B. S. (1992). A Study of Acculturation Differences Among International People and Suggestions for Sensitivity to Within-Group Differences. Journal of Counseling & Development, 71(1), 53–59. doi:10.1002/j.15566676.1992.tb02171.x Tajfel, H. (1981). Human groups and social categories. United Kingdom: Cambridge University Press. Tajfel, H., & Turner, J. C. (1986). The social identity theory of intergroup behavior. In Psychology of intergroup relations (2nd ed., Vol. 2). Illinois: Nelson-Hall. Turner, J. C. (1987). Rediscovering the social group: A self-categorization theory. Cambridge, MA, US: Basil Blackwell. van Knippenberg, D., Lossie, N., & Wilke, H. (1994). In-group prototypicality and persuasion: Determinants of heuristic and systematic message processing. British Journal of Social Psychology, 33(3), 289–300. doi:10.1111/j.2044-8309.1994.tb01026.x Verkuyten, M., & de Wolf, A. (2002). Ethnic minority identity and group context: selfdescriptions, acculturation attitudes and group evaluations in an intra- and intergroup situation. European Journal of Social Psychology, 32(6), 781–800. doi:10.1002/ejsp.121 Wang, C.-C. D., & Mallinckrodt, B. (2006). Acculturation, attachment, and psychosocial adjustment of Chinese/Taiwanese international students. Journal of Counseling Psychology, 53(4), 422–433. doi:10.1037/0022-0167.53.4.422 Ward, C., & Rana-Deuba, A. (1999). Acculturation and Adaptation Revisited. Journal of CrossCultural Psychology, 30(4), 422–442. doi:10.1177/0022022199030004003 Ward, Colleen, & Kennedy, A. (1993). Acculturation and Cross-Cultural Adaptation of British Residents in Hong Kong. The Journal of Social Psychology, 133(3), 395–397. doi:10.1080/00224545.1993.9712158 50 Ward, Colleen, & Kennedy, A. (1994). Acculturation strategies, psychological adjustment, and sociocultural competence during cross-cultural transitions. International Journal of Intercultural Relations, 18(3), 329–343. doi:10.1016/0147-1767(94)90036-1 Whittler, T. E. (1989). Viewers’ processing of Actor’s race and message claims in advertising stimuli. Psychology and Marketing, 6(4), 287–309. doi:10.1002/mar.4220060405 Whittler, T. E. (1991). The Effects of Actors’ Race in Commercial Advertising: Review and Extension. Journal of Advertising, 20(1), 54–60. doi:10.1080/00913367.1991.10673207 Williams, J. D., & Qualls, W. J. (1989). Middle-class black consumers and intensity of ethnic identification. Psychology and Marketing, 6(4), 263–286. doi:10.1002/mar.4220060404 51