Consciousness of Social Face: The Development and

0 downloads 0 Views 115KB Size Report
resulted in an 11-item Consciousness of Social Face (CSF) scale made up of the follow- .... No information about the study was provided prior to the questionnaire session. ..... Fenigstein's (1976) self-consciousness scale were used to measure people's pub- ..... monitoring in predicting responses to situational pressures.
Downloaded by [Shanghai Jiaotong University] at 19:36 19 November 2011

The Journal of Social Psychology, 2011, 151(2), 129–149 Copyright © Taylor & Francis Group, LLC

Consciousness of Social Face: The Development and Validation of a Scale Measuring Desire to Gain Face Versus Fear of Losing Face XIN-AN ZHANG Shanghai Jiao Tong University QING CAO University of Connecticut NICHOLAS GRIGORIOU Monash College

ABSTRACT. This article describes the development and validation of a scale that measures two distinct needs for individuals to manage their social “face”. Scale development process resulted in an 11-item Consciousness of Social Face (CSF) scale made up of the following two correlated dimensions: desire to gain face and fear of losing face. The two-factor correlated structure of CSF scale was stable across multiple samples of both students and non-students subjects. The construct validity of CSF scale, including convergent validity, discriminant validity, and criterion-related validity was also demonstrated by examining relationships with other personality or demographical variables. Keywords: social face, gain face, lose face, scale development and validation

This study is supported by research grants from the National Natural Science Foundation of China (grant no. 70872075), the Chinese Ministry of Education (grant no. 07JC630033), and the Shanghai Planning Office of Philosophical and Social Science (grant no. 2009BSH001). The research is also sponsored by the Shanghai Pujiang Program. The first author thanks the Graduate School of Arts and Sciences at Harvard University for hosting him as a postdoctoral fellow when this article was written. Address correspondence to Xin-an Zhang, Shanghai Jiao Tong University, School of Management, 535 Fa Hua Zhen Rd., Antai Building, Shanghai, China 200052; [email protected] (e-mail). 129

Downloaded by [Shanghai Jiaotong University] at 19:36 19 November 2011

130

The Journal of Social Psychology

“FACE” IS ORIGINALLY A CONCEPT DEVELOPED in the Chinese Confucian society as “the most delicate standard by which Chinese social intercourse is regulated” (Lin, 1935, p.200). Being more salient in collectivistic societies like China, however, the concept of face is a human universal in nature (Ho, 1976). The universality of face is argued by Goffman (1956) as being based on the need to avoid what he terms “flustering” in social interactions. To keep social interactions going smoothly and effectively, people involved in interactions are required to convey minimally acceptable public images of themselves and likewise to assist other people in maintaining their social identities as well. Therefore, within any given culture, people must show a regard for face if they do not want to declare bankruptcy of their social status. Further, Hsu (1981, p.110) stated that “the Chinese concept of face and of propriety, and the American sensitivity to prestige and superiority are all familiar expressions of the same need.” Indeed, there might be considerable variation among the conceptualization of what constitutes face and the rules governing face behavior, but people’s concern for face is invariant (Ho, 1976). It is worthy to note that the Chinese concept of face consists of two distinct sets of criteria for judging conducts of social and moral aspects separately. The social aspect of face is in general a function of one’s social status standing for the prestige and honor that accrues for a person as a result of success and possibly ostentatious behaviors before others (Bond & Lee, 1981; Ho, 1976), whereas the moral aspect of face represents mainly the confidence of society in the integrity of ego’s “moral character” (Hu, 1944, p.45). Recently, there was a resurgence of interest in the research in the social aspect of face. It was found relevant to a wide variety of behaviors including conflict behavior (Ting-Toomey, 1988; Ting-Toomey & Kurogi, 1998), gift giving behavior (Joy, 2001), student feedback seeking and learning behavior (Hwang, Francesco, & Kessler, 2003), and consumer decision-making styles (Bao, Zhou, & Su, 2003), across a range of cultures. However, further studies trying to link face empirically to diverse behaviors were limited by the lack of a reliable instrument measuring an individual’s concern for face. For example, Bao, Zhou, and Su (2003) treated face consciousness as a simple uni-dimensional construct measured by only four items. This scale might not capture all meanings of the concept, because people may have two distinct kinds of concern for social face, including enhancing face, as well as avoiding losing face in relation to significant others in social interactions (Ho, 1976). In all societies people may have experienced the feeling of gaining or losing face because of positive or negative social evaluation (Hwang, 2006). Gaining face and losing face refer to two distinct feelings involving different social processes and should not be treated as if they were opposite outcomes in a social encounter. Accordingly, they are not assessed according to the same set of criteria. The criteria used for judging how one gains face are based on what the society originally expected from the person. Face may be gained when one’s social

Downloaded by [Shanghai Jiaotong University] at 19:36 19 November 2011

Zhang, Cao, & Grigoriou

131

performance goes above and beyond social expectations on him, but it doesn’t necessarily constitute loss of face when his social performance goes below these expectations. For example, if someone achieves great success under circumstances beyond his control, it will increase his face, whereas failure of doing so in such hard circumstances will only make him have fewer claims to his face, but not lose face in the process because it is seen as acceptable in terms of the social expectations on him. Therefore, the opposite of gaining face is a “process of erosion, distinguishable from what is called ‘loss’ of face”(Ho, 1976, p. 871). Similar to the fact that not gaining face doesn’t necessarily lead to losing face, not losing face also doesn’t mean gaining face, because the criteria for judging how one loses face are different. One may lose face when his conduct or performance “falls below the minimum level considered acceptable or when certain vital or essential requirements, as functions of one’s social position, are not satisfactorily met” (Ho, 1976, p.871). Thus, even when one’s performance exceeds these criteria, it may only result in the avoidance of losing face, but he will not gain face in the process. The fact that gaining face is not directly relevant to the concept of defensiveness, while trying to protect one’s face or avoid losing face will make a person much more defensive, provides additional evidence that they are not two opposite constructs (Cupach & Carson, 2002; Ho, 1976). Furthermore, the results of losing face are often more serious than not gaining face. There may be some people who may not have interest in gaining face through extra efforts and showing off their success, but everyone has to save his “face” from being lost if he cares for maintaining a minimum level of effective social functioning (Ho, 1976). When a person is perceived to have failed to protect his face, he feels rejected by others, which causes the person to feel embarrassed, shamed, and humiliated (Kim & Nam, 1998). As a consequence, an embarrassed or shamed person usually experiences painful emotions and a strong motivation to restore lost face (TingToomey, 1988). But regaining face does not constitute gaining face; it is merely a restoration to the individual of what ought to have been due him in the first place. Therefore, individuals are far more concerned with protecting their face than they are with gaining face. In summary, the distinction between the dynamics of gaining face and losing face has suggested that concern for social face is a multidimensional construct, and its two dimensions of face behavior should have different underlying motivational sources. Therefore, recognizing the existence of alternative motives for people to manage their social images, the Consciousness of Social Face construct is defined as people’s needs to enhance, as well as to avoid, losing their social face with significant others in social encounters. The objectives of the present research were to develop and test a general measure for the construct of Consciousness of Social Face. Conceptually, we

Downloaded by [Shanghai Jiaotong University] at 19:36 19 November 2011

132

The Journal of Social Psychology

followed Ho’s (1976) typology to distinguish two kinds of important changes in the status of one’s face: gaining face and losing face. Thus, the psychological construct of Consciousness of Social Face consists of two dimensions, namely, desire to gain face and fear of losing face. In developing the scale, we followed the scale development procedure outlined by Hinkin (1998). The first phase of our study (Study 1) was concerned with developing scale items and assessing the internal consistency reliability of scores on the newly developed Consciousness of Social Face (CSF) scale. In addition, exploratory factor analyses were conducted to determine whether the scale actually measures two distinct dimensions of concern for social face. The second phase of the study (Study 2) aims to provide additional evidence for the stability of the factor structure of the CSF scale by examining a series of competing models using non-students’ samples. Finally, in Study 3, the construct validity of the current scale was demonstrated, including evidence for convergent validity, discriminant validity, and criterion-related validity.

STUDY 1: SCALE DEVELOPMENT The purpose of the first study was to develop an 11-item scale that contained variable items for each of the two dimensions of Consciousness of Social Face. Given this objective, an original pool of 77 items was generated from exploratory interviews, focus groups, and a review of items used in earlier studies, including 46 items for the dimension of desire to gain face (e.g., “When I do well in my job, I hope my supervisor can praise me in public”) and 31 items for the dimension of fear of losing face (e.g., “I will be embarrassed if I make a simple mistake in public”), respectively. Following the item generation step, ambiguous items and statements with essentially identical meaning were eliminated, resulting in a revised pool of 61 items. Five Ph.D. students in management then acted as judges in an evaluation of the content validity of the items. In this analysis, the five judges were exposed to the definition of each dimension plus a related explanation and an example item, and were asked to allocate the statements to each dimension or to a “not applicable” category. Items that did not receive consistent classification by at least four of the five judges were eliminated. This initial analysis resulted in 47 statements for the two dimensions of Consciousness of Social Face. Next, following the procedure recommended by Hinkin (1998), additional five judges were given each dimension’s definition and asked to rate how well each statement reflected the two different dimensions of Consciousness of Social Face, using the following scale: 1 = clearly representative, 2 = somewhat representative, and 3 = not representative at all. For the two dimensions, only items evaluated as clearly representative by at least three judges were retained. This process eliminated 13 more items.

Zhang, Cao, & Grigoriou

133

Method

Downloaded by [Shanghai Jiaotong University] at 19:36 19 November 2011

Subjects The subjects consisted of 173 third-year undergraduate students majoring in a range of disciplines, including management, economics, physics, mathematics, and engineering (Mage = 20.6 years, SD = 1.4), at a major university in east China, including 111 male (64%) and 62 female (36%). These students were taking a managerial psychology course taught by the first author and volunteered to participate in the study as one means of partially fulfilling a research requirement for this course. No information about the study was provided prior to the questionnaire session. Measures The instrument consisted of 34 statements, of which 14 were intended to assess desire to gain face, whereas the other 20 were meant to assess fear of losing face. Participants were required to respond to each item on a 7-point “strongly agree/strongly disagree” Likert scale. Six items were reverse coded. Procedure All the questionnaires were distributed by the first author during scheduled class time. Respondents’ anonymity was assured, and participants typically spent around 5 to 10 minutes to complete. Results and Discussion Item Analysis and Reduction The corrected item-to-total correlations for each of the two dimensions of Consciousness of Social Face scale were computed. These correlations ranged from .19 to .60 for the dimension of desire to gain face, and from .09 to .53 for the dimension of fear of losing face. Items that did not have corrected item-tototal correlations above 0.30 were deleted. Items were also deleted if they did not have statistically significant higher correlations with the dimension to which they were presumed to belong than correlations with the other dimensional total score. These procedures eliminated eight more items. Exploratory Factor Analysis The retained 26 items were factor-analyzed by means of principal components analysis with varimax rotation. Prior to this analysis, Kaiser-Meyer-Olkin

Downloaded by [Shanghai Jiaotong University] at 19:36 19 November 2011

134

The Journal of Social Psychology

test of sampling adequacy and Bartlett tests of sphericity indicated that the data were appropriate for factor analysis. As expected, the initial factor solution resulted in two factors according to the scree test of eigenvalues. A further oblique rotation generated a similar factor structure, suggesting the stability of the two factors, which accounted for 18.6% and 11.9% of the total variance respectively. In selecting items for the final scale, we used a minimum factor loading of .40 as a guideline for considering an item to be part of a scale (Ford, MacCallum, & Tait, 1986), items with higher loadings on both or neither of the first two intended factor were eliminated. As a result of this elimination process, only items with high loadings on the intended factor and low loadings on the other factor were included. We factor-analyzed these retained items the second time. This resulted in the same two-factor solution and these two factors accounted for 43.5% of the total variance. The final 11-item scale of Consciousness of Social Face, along with the factor loadings and other descriptive statistics are shown in Table 1. As can be seen in Table 1, the first factor, comprising six items, all of which reflected individual’s desire to enhance his social status through successful social performance or ostentatious behaviors (e.g., “I hope people think that I can do better than most others”), represented the “desire to gain face” dimension. The second factor, including the other five items, reflecting individual’s intention of avoiding being involved in situations that would make his inferiority or bad social performance public (e.g., “I always avoid talking about my weakness”), represented the other dimension of “fear of losing face.” The internal consistency of the scale was tested. Both desire to gain face subscale (Cronbach’s alpha = .76) and fear of losing face subscale (Cronbach’s alpha = .74) were found to be of an acceptable level of internal consistency. The Cronbach’s alpha for the composite CSF scale was .82. All 11 items produced corrected item-total correlations no less than .45, and elimination of any of them reduced the value of alpha. The Pearson correlation coefficient between these two subscales (i.e., means of the six items and the other five items) is only moderately high as .33 (p < .01), indicating that they are measuring two different constructs.

STUDY 2: SCALE VALIDITION The scale that emerged as the result of Study 1 successfully distinguished two kinds of motivational source for individuals to manage their social face. However, it was derived from a convenient university student sample. As the goal of the present research was to create a general scale measuring people’s desire to gain face and fear of losing face, it was necessary to test the generalizability of the two-factor solution across different validation samples.

1.22 1.09 1.32 1.16 1.37 1.44 1.29 1.22 1.49 – –

5.16 5.01 4.21 4.79 3.90 3.92 4.03 4.23 2.93 – – –

Cronbach’s alpha

.76

3.07 27.95%

.09

.11 .25

.01 .08

.54

.64 .55

.66

.69 .68

Factor 1

.74

1.71 15.51%

.55

.67 .56

.78 .73

.21

.82 (CSF)

– 43.46%(total)

.45

.48 .48

.51 .53

.45

.52 .53

.56

−.05 .11 .18

.59 .54

.03 .13

Factor 2

Corrected item-total correlation

Note. n = 173. Rotation method =Varimax. ∗ The original scale items were generated in Chinese. Items in this table were translated following the procedures recommended by Brislin (1970). The first author translated these items into English, and the translated version was back-translated to Chinese by the second author. Discrepancies in the translation were carefully inspected and corrected.



1.12 1.20

5.26 4.72

I hope people think that I can do better than most others I hope that I can talk about things that most others do not know I hope that I can possess things that most others thirst for It is important for me to get praise and admiration I hope to let people know that I have association with some big names I hope that I have a better life than most others in others’ view I always avoid talking about my weakness I try to avoid letting others think that I am ignorant, even if I really am I do my best to hide my weakness before others If I work in an organization of bad reputation, I will try not to tell others about that It is hard for me to acknowledge a mistake, even if I am really wrong Eigenvalue Explained variance

SD

Mean

Items

Factor loading

TABLE 1. Descriptive Statistics, Factor Loadings, and Corrected Item-Total Correlations for the Two Factors of the CSF Scale∗

Downloaded by [Shanghai Jiaotong University] at 19:36 19 November 2011

Zhang, Cao, & Grigoriou 135

136

The Journal of Social Psychology

Method

Downloaded by [Shanghai Jiaotong University] at 19:36 19 November 2011

Subjects A combined sample from three separate studies was used and totally consisted of 411 adults from three different settings, these including 201 MBA students in a major university in east China (49%), 95 non-degree business training students in the same university (23%), and 115 middle managers in a state-owned enterprise in the China’s telecom industry (28%). Of this sample, 64% were male and 36% were female. Ages ranged from 23 to 57 years (Mage = 35.6 years, SD = 7.2). More than 75% of respondents in all three samples had completed undergraduate degrees. Measures and Procedure The 11-item CSF scale was incorporated into three different questionnaires that were administered over the above different settings. Respondents from each setting were entirely independent of those of another. The questionnaires for MBA and non-degree training classes were distributed by either the first author or the instructors of participating courses. In the field setting, questionnaires were distributed by and returned to a research assistant when these middle managers gathered in the company annual meeting. An attached cover letter assured managers that their participation was entirely voluntary, and their responses would remain confidential and not be used for other purposes. Results and Discussion Following the guidelines recommended by Anderson and Gerbing (1988), a series of confirmatory factor analyses were performed with LISREL VIII program (Jöreskog & Sörbom, 1993) to compare the proposed model with several alternative models to see which was the most consistent with the data. In these analyses, totally four competing models were examined: a single-factor solution, a model with two uncorrelated factors, the proposed model with two correlated factors, and a three-factor model, which was derived from the results of exploratory factor analysis using the data of Study 1 by setting the number of extracted factors as three. In all cases, the maximum likelihood method was used. The itemvariance-covariance matrix, rather than correlation matrix among these 11 items was generated as input data, because Harvey, Billings, and Nilan (1985) suggested that differences in item variances are lost in the analysis of correlations when all variables are standardized to a common variance. Bollen (1989, p. 270) stated that the null model is the most restrictive model for comparison purposes. Therefore, increments in various fit indicators calculated from a null model were used as the bases for comparison. Five fit indicators

Downloaded by [Shanghai Jiaotong University] at 19:36 19 November 2011

Zhang, Cao, & Grigoriou

137

were compared, as Anderson and Gerbing (1988) suggested, to use a number of indicators rather than a single Chi-square to detect model misspecifications. These fit indicators provided by LISREL include: (1) a Chi-square statistics, which assesses the probability that the sample data confirm the hypothesized model. Because all the models examined are nested, differences in Chi-square compared to the null model were used to test which one fits the data the best. Unfortunately, the absolute Chi-square index provided by LISREL strongly depends on sample size, so we also examined the ratio of Chi-square to its degree of freedom. It is suggested that a Chi-square two or three times as large as the degrees of freedom is acceptable (Hinkin, 1998); (2) a goodness of fit index (GFI), which estimates the extent to which the sample variances and covariance are reproduced by the model; (3) an adjusted goodness of fit index (AGFI) that adjusts for the model’s degrees of freedom relative to the number of variables, a GFI above .9 and an AGFI above .8 is generally interpreted as representing a good fit (Kelloway, 1998); (4) a normed fit index (NFI), which measures the fit of the examined model compared to the null model, a value of the NFI below .90 indicates the examined model can be further improved (Kelloway, 1998). Finally, LISREL also reports a root mean squared error of approximation (RMSEA) developed by Steiger (1990), which represents the extent to which the covariance structure specified in the model differs from the covariance structure observed in the data. Steiger (1990) suggested that values below .10 indicate a good fit to the data, and values below .05 indicate a very good fit to the data. As shown in Table 2, the two-factor correlated model fitted the data very well: The χ 2 /df ratio equaled 2.07 below 3, GFI (.96), AGFI (.94), and NFI (.91) were all larger than .90, and the RMSEA was only .05 much lower than the upper bound of .10 as suggested by Steiger (1990). Considering that some previous studies had treated Consciousness of Social Face as a unidimensional construct and forced these two dimensions into a continuum (e.g., Bao, et al., 2003), a single-factor model was first tested against the proposed two-factor correlated model. Results

TABLE 2. Comparison of LISREL Models for the Consciousness of Social Face Items Models Null Single-factor Two correlated factors Two orthogonal factors Three factors

χ2

df

χ 2 df

GFI

AGFI

NFI

RMSEA

1508.24 459.86 89.30 121.29 131.37

55 44 43 44 41

27.42 10.45 2.07 2.75 3.12

.58 .82 .96 .94 .94

.49 .73 .94 .92 .90

.00 .65 .91 .86 .86

.27 .16 .05 .07 .08

Downloaded by [Shanghai Jiaotong University] at 19:36 19 November 2011

138

The Journal of Social Psychology

indicated that the two-factor solution provided much better fit to the data than the one-factor solution did: moving from one to two factors reduced the χ 2 value by 370.56 (with one degree of freedom, this decrease was significant at the level of .01). Furthermore, the GFI, AGFI, and NFI all became much lower, while RMSEA became much higher than those of the proposed solution. In short, these results clearly indicated the insufficiency of a single-factor solution in comparison to the two-factor correlated model. In Study 1, the two factors were found to be only correlated moderately with each other. Therefore, we examined another competing model in which the two factors were set orthogonal (uncorrelated). A similar thing to the single-factor model also happened here: The two-factor uncorrelated model increased χ 2 value by 31.99 (with one degree of freedom, this increase was significant at the level of .01), the RMSEA also became larger than that of the correlated solution, while the values of GFI, AGFI, and NFI decreased. These changes indicated a worse level of fit. Indeed, the correlation coefficient between the two factors in the correlated model was .31. Therefore, it was suggested the CSF scale should comprise two distinguishable factors that are moderately correlated. Finally, a three-factor solution was examined to see whether extracting more factors would increase the fit level to the data. The grouping of variables in this model was guided by an exploratory factor analysis using data of the Study 1, in which the number of extracted factors was forced to set as three for the exploratory factor analysis. The results of confirmatory factor analysis on this three-factor solution also indicated a worse fit than the proposed two-factor correlated model, as moving to the three-factor model from the proposed model increased χ 2 by 42.07 (2 degrees of freedom, this increase was significant at the level of .01). Further, GFI, AGFI, NFI and RMSEA also became considerably worse when the new factor was introduced. This particular pattern of changes in χ 2 value and other fit indicators represented a strong indication of the adequacy of the twofactor correlated structure for the CSF scale. STUDY 3: CONSTRUCT VALIDITY The ultimate objective of the scale development process is to demonstrate construct validity (Cronbach & Meehl, 1955), which was also the purpose of the present study. The types of construct validity include convergent, discriminant, and criterion-related. Convergent validity means that the new scale should correlate significantly but not very highly with the other measures designed to assess similar constructs, whereas discriminant validity means that the new scale should not correlate with dissimilar measures. In terms of criterion-related validity, the new measure should correlate with variables that the new measure is theoretically expected to predict (Hinkin, 1998). The present study examined the construct validity of the CSF scale through all these three kinds of construct validity. First, the convergent validity of CSF

Downloaded by [Shanghai Jiaotong University] at 19:36 19 November 2011

Zhang, Cao, & Grigoriou

139

was assessed by examining how well the new scale converges with other similar personality constructs measuring individual’s need to maintain a favorable social image, including public self-consciousness and social anxiety (Fenigstein, 1976), impression management (Paulhus, 1994), self-monitoring (Snyder & Gangestad, 1986), and social comparison (Gibbons & Buunk, 1999). Second, the discriminnat validity of CSF scale was assessed by examining relationship with the conceptually unrelated construct of openness dimension in five-factor personality model (Goldberga, et al., 2005). Finally, for criterion-related validity, we focused on two kinds of tactics used in favorable self-presentation (Roth, Snyder, & Pace, 1986) that were hypothesized to correlate with CSF theoretically. The rationale behind the selection of each variable is provided below. Public self-consciousness is defined as a general awareness of the self as a social object that has an effect on others, whereas social anxiety is defined by a discomfort in the presence of others (Fenigstein, 1976). High scores on public self-consciousness reflect a concern for one’s social appearance and the impression one makes on others (Turner, Carver, Scheier, & Ickes, 1978). Hence, a positive and moderate correlation can be expected between public self-consciousness and CSF. Further, individuals who are keenly aware of their social image before others may become apprehensive (Fenigstein, 1976), thus a positive correlation can be expected between CSF and social anxiety. Impression management refers to the process by which individuals attempt to control the impressions that others form of them (Leary & Kowalski, 1990). As high scores on CSF reflect people’s desire to maintain their social image before others (Ho, 1976), these scores are more likely to influence the way that people are perceived by significant others and engage in more impression management behavior. Hence, a positive correlation can be expected between impression management and CSF. Self-monitoring is defined as the extent to which people can observe and control their expressive behavior and self-presentation (Snyder & Gangestad, 1986). Individuals high in CSF are thought to regulate their expressive selfpresentation for the sake of desired public appearances (Ho, 1976). Hence, a positive correlation can be expected between self-monitoring and CSF. Social comparison refers to the desire to learn about the self through comparison with others. Evidence showed that individuals high in social comparison also have high concern for public self-image (Gibbons & Buunk, 1999). One may also feel gaining face when his super performance was demonstrated through comparison with peers. Hence, a positive correlation can be expected between social comparison and CSF. Openness, in five-factor framework of personality, refers to appreciation for art, emotion, adventure, unusual ideas, imagination, curiosity, and variety of experience (McCrae & John, 1992). It was selected as the variable to test the discriminant validity of CSF because no evidence as yet showed that it will correlate with an individual’s needs for a favorable social image.

Downloaded by [Shanghai Jiaotong University] at 19:36 19 November 2011

140

The Journal of Social Psychology

Favorable self-presentation refers to the tendency of an individual to present himself in an unrealistically positive way (Roth, Snyder, & Pace, 1986). Individuals may use two kinds of tactics to portray a favorable image. One is to deny the presence of negative characteristics, and the other is to attribute the existence of positive characteristics of the self. Because CSF reflects an individual’s desire to maintain a favorable social image (Ho, 1976), people high in CSF are more likely to use both of these two tactics to influence perceptions of other people about them. Hence, we expect that CSF will positively predict both the two self-presentation tactics. Method Subjects Assuming that respondents would tire with one long questionnaire, we separated all used variables into two questionnaires and administrated them to two independent samples. Sample 1 consisted of 264 undergraduate students from a major university in east China. The grade levels of participants were as follows: 59 freshmen (22%), 123 sophomores (47%), 37 juniors (14%), 42 seniors (16%), and three students (1%) who didn’t specified their grade level. These students were affiliated with a variety of departments, including engineering, management, social science, physics, and mathematics. Of this sample, 170 were female (64%), whereas 94 were male (36%). Ages ranged from 16 to 28, with an average age of 20.2 (SD = 1.9). The other sample consisted of 377 university student from another university located in east China, also within a wide variety of study disciplines. The majority of these students (> 90%) were junior or senior students. In terms of gender, there were 209 (55%) males and 168 (45%) females. Ages ranged from 18 to 29, with the average being 20.9 (SD = 1.6). Measures Consciousness of social face. The 11-item CSF scale resulting from previous studies was included to measure two distinct dimensions of social face consciousness, namely desire to gain face and fear of losing face. These items were responded to on a 7-point Likert scale. The Cronbach’s alpha for desire to gain face subscale was .79, and for fear of losing face was .72. Public self-consciousness and social anxiety. The seven items adopted from Fenigstein’s (1976) self-consciousness scale were used to measure people’s public self-consciousness. Another six items from the same scale were also included to measure social anxiety. All these items were responded to on a 5-point Likert

Zhang, Cao, & Grigoriou

141

Downloaded by [Shanghai Jiaotong University] at 19:36 19 November 2011

scale. The Cronbach’s alpha for public self consciousness was .84, and for social anxiety was .80. Impression management. The 20-item Impression Management scale adopted from Paulhus’s (1994) Balanced Inventory of Desirable Responding (BIDR) was used to measure individual’s tendency to manage favorable impressions. Responses were given on a 7-point Likert scale, but scores were converted according to the standard scoring procedure (Paulhus, 1994). Thus, 1 point was given for each “6” or “7” and the total scale score ranged from 0 to 20. A previous study has shown that the Chinese version of BIDR has sufficient reliability and exhibits measurement equivalence between males and females (Xinwen, Feng, & Yiwen, 2004). The Cronbach’s alpha of the scale in this study was .77. Openness. The 10 items adopted from Goldberg’s (1999) International Personality Items Pool (IPIP) scale were used to measure the openness dimension in the five-factor personality framework. These items were responded to on a 5-point Likert scale, and the Cronbach’s alpha was .83. Self-monitoring. Snyder and Gangestad’s (1986) revised 18-item scale was used to measure an individual’s self-monitoring propensities. Items in this scale were responded to in a true-or-false format. Thus, higher scores indicated the respondent was higher on self-monitoring. The Chinese version of the self-monitoring scale was translated by Feng, De, and Yulian (1992) with sufficient reliability. The Cronbach’s alpha of the scale in the present study was .75. Social comparison. Gibbons and Buunk’s (1999) 11-item scale was used to measure individual’s social comparison orientation. Reponses were given on a 5-point Likert scale and the Cronbach’s alpha was .81. Favorable self-presentation. Roth, Snyder, and Pace’s (1986) Self-Presentation Scale (SPS) was used to measure two distinct dimensions of tactics in presenting a favorable self-image, namely denying negative characteristics versus attributing positive characteristics separatively. Each dimension was measured by 10 items on a 7-point Likert scale. The Cronbach’s alpha for denying negative characteristics subscale was .69, and for attributing positive characteristics subscale was .77. Procedure Both of the two surveys for validity test were administrated to students during scheduled class time. The questionnaires were distributed by either the first author

142

The Journal of Social Psychology

or by the instructor of the participating courses, and it typically took students 15 to 20 minutes to complete the questionnaires. Results and Discussion

Downloaded by [Shanghai Jiaotong University] at 19:36 19 November 2011

Convergent Validity As recommended by researchers (Churchill, 1979; Hinkin, 1998), the convergent validity of CSF was assessed by examining the correlations between CSF and other personality variables that are thought to be positively related to Consciousness of Social Face. In the present study, the convergent variables included public self-consciousness, social anxiety, impression management, selfmonitoring, and social comparison. As indicated in Table 3, the correlations between CSF and these variables were .34 (n = 264, p < .01), .14 (n = 264, p < .05), .30 (n = 377, p < .01), .13 (n = 264, p < .05), and .27 (n = 264, p