Journal of Youth and Adolescence, Vol. 13, No. 4, 1984

Dimensions of the Self-Concept: A Source of Further Distinctions in the Nature of Self-Consciousness Gregory C. Elliott, Received September 1, 1983; accepted June 18, 1984

This study distinguished between forms o f self-consciousness (private selfconsciousness and social anxiety) and investigated the effect o f self-esteem, vulnerability to criticism, and the tendency to fantasize on each. Utilizing a sample o f adolescents (age 12 to 19years), a structural equation analysis using unobserved variables revealed that, as expected, vulnerability to others" criticism heightened both private self-consciousness and social anxiety. In contrast, the tendency to fantasize in everyday life increased only private self-consciousness, and low self-esteem led directly only to increased social anxiety. Further, results showed that self-esteem had an appreciable indirect effect on both private self-consciousness and social anxiety, as mediated by vulnerability and the tendency to fantasize, indicating that the effect o f self-esteem (a primary motivator) was more complex than an analysis o f direct effects would indicate. Implications for our understanding o f selfconsciousness and the self-concept are discussed.

My deepest thanks go to Morris Rosenberg, whose insights into the self-concept were indispensable, both in conceptualization and in analysis. I also thank John A. Fleishman and Edward Z. Dager for their constructive comments on early drafts of this article. Any errors remaining are my responsibility. This study was supported by NIMH Grant R01 MH27747-06 awarded to Morris Rosenberg. The data were analyzed using the facilities of the Computer Science Center, University of Maryland, College Park. ~Department of Sociology, Brown University, Providence, Rhode Island 02912. Professor Elliott received his Ph.D. in sociology at the University of Wisconsin, Madison. His current major interest is in the structure of the self-concept. 285 0047-2891/84/0800-0285S03.50/0 © 1984 Plenum Publishing Corporation

286

Ellioll

INTRODUCTION Self-awareness is the condition in which the self is the focus o f a person's own attention. All of us have, at one time or another, been acutely aware o f ourselves. We can be embarrassed for having committed a social gaffe, we can be preoccupied with the impression we are making on others, or we can be examining our motives for particular behaviors. These are but a few o f the many instances in which our attention is directed inward, toward ourselves, rather than outward, toward the people and things in our environment. The environment can be a powerful source o f self-awareness. Duval and Wicklund (1972; Wicklund, 1975) have developed a theory delineating the situational factors that lead to self-awareness and the consequences o f that state. According to the theory, any stimulus that causes a person to take him- or herself as an object of attention will lead to self-awareness. For example, seeing oneself in a mirror or appearing before a camera causes people to be self-aware. Argyle (1969) proposed a number of situational factors that could lead to self-awareness, including mere presence, individuation, evaluation, and self-confrontation. However, the environment may not be the only stimulus to selfawareness. Some people may be chronically aware o f themselves and their behavior during an interaction, regardless of its nature; for others, it may be extremely rare for them to focus their attention on themselves, even when situational factors facilitate the process. Buss (1980) makes a distinction between the state o f self-awareness, which is situationally induced and transient, and the trait o f self-consciousness, which is the person's reasonably stable disposition to focus his or her attention on the self. The purpose o f this study is to further our understanding o f dispositional self-consciousness by exploring its relationship with certain aspects of the self-concept, using a sample o f young people (ages 12-19 years). One must distinguish the notion o f self from that of self-concept. Mead (1934) saw the self as a process of interaction with oneself through the conversation between the "I" and the "Me." [See also Blumer (1969).] However, sociologists also agree that there is a mental structure called the self-concept: "the totality of the individual's thoughts and feelings with reference to himself as an object" (Rosenberg, 1979, p. 7). Most of the research on the self-concept has heretofore been focused on self-esteem (Wells and Marwell, 1976; Wylie, 1979), but Rosenberg (1979) has called for research in the self-concept that goes "beyond self-esteem." More is needed, however, than a simple expansion o f our analysis o f the self-concept. Even when other aspects of the self-concept are considered, most of the research has focused on the study of specific dimensions in isolation. The result is an ever-lengthening list o f elements and dimensions com-

Dimensions of the Self-Concept

287

prising the self-concept. Elliott et al. (1984) have called for a change in the focus in self-concept research. Specifically, they suggest that the task before us is to move to a study of the self-concept structure (i.e., how these various bits and pieces are arranged with reference to one another) and to formulate and test models that explore the causal nexus of these dimensions. Some research has begun to address this issue. McGuire and PadawerSinger (1976) and Rosenberg (1965) have shown that various components of the self-concept are differentially salient or important to the individual; Stryker (1968) proposed that certain identity elements are preeminent in a person's self-concept. Still, this is merely the beginning of a change in focus from "lists" of loosely related self-concept elements to a structural approach, in which dimensions are posited to be causally related in theoretically meaningful ways. The present paper presents a model specifying theoretically plausible connections among several fundamental aspects of the self. The notion of self-consciousness as a disposition may be new, but it has its roots in our earliest understandings of the self. Cooley's (1902) notion of the "looking-giass self" as a basis for self-observation and Mead's (1934) delineation of the process of role-taking in order to understand oneself from the others' vantage point indicate that sociologists have long been aware of our ability to make ourselves the object of our own focused attention. Recently, Rosenberg has argued that the trait of self-consciousness is an important aspect of the self; it is "recognized in experience, implicated in behavior, and related to emotional disturbance" (Rosenberg, 1979, p. 283). As such, self-consciousness is likely to be associated with dimensions of the self-concept in theoretically meaningful ways. In particular, we focus on the effects of self-esteem, vulnerability to criticism, and the tendency to fantasize on self-consciousness. Others have examined the zero-order relationships between self-consciousness and various personality traits (e.g., Turner et al., 1978). The major distinction to be made here is that the present effort is more focused in its attempt to look within the self-concept and to locate self-consciousness within a theoretically derived network of important dimensions. This will simultaneously yield evidence about the structure of the selfconcept and provide an indication of the construct validity of the measures of self-consciousness used herein. The present study differs from earlier studies of self-consciousness in several ways. First, there is a difference in focus. Many past studies involve situational self-awareness, not dispositional self-consciousness (Duval and Wicklund, 1972; Wicklund, 1975). Those that do involve self-consciousness more often focus on its behavioral consequences (Buss and Scheier, 1976; Scheier et al., 1978; Turner and Peterson, 1977). In contrast, the present study investigates the relationship between self-consciousness and certain dimensions of the self-concept. Second, whereas previous research has been based

288

Elliott

largely on experimental design, the present study explores the issue by means of sample survey analysis. Third, the present study extends the scope o f investigations o f self-consciousness by using a sample o f adolescents.

Varieties of Self-Consciousness Our everyday use of the term "self-consciousness" makes us realize that there is more than one way in which a person's attention can be self-focused. In one form of self-consciousness, the individual tends to be especially aware o f his or her own internal states. For example, some people seem to focus more on their values or are in closer touch with their feelings than are others. This preoccupation with one's own psychological attributes is called private self-consciousness. In contrast, a second form of self-consciousness is characterized by a general concern with the self as a social actor who is a stimulus for others' behavior. For example, some people are very aware that others will react to the impression they convey in an interaction, and so they orchestrate their performance accordingly. This realization of one's status as the object of other people's attention is called public self-consciousness. Fenigstein et al. (1975) have found empirical support for a discrimination between private and public self-consciousness. Conceptually, the distinction between them is that, in private self-consciousness, the person's attention is on his or her personal characteristics, whether or not others are aware of them, whereas in public self-consciousness, the person is aware that others will react to him or her as a social object, even though particular internal states are not at issue. ~ Fenigstein et al. (1975) discovered that public self-consciousness must be distinguished from social anxiety. Socially anxious people not only are aware that others react to them as social stimuli; they also are apprehensive about this reaction. Buss (1980) identified four types o f social a n x i e t y embarrassment, shame, shyness, and audience a n x i e t y - b u t they all involve discomfort due to the scrutiny o f others. Note that public self-consciousness is a necessary but not sufficient condition for social anxiety. Socially anxious people are presumably always publicly self-conscious, but people can be aware o f the fact that others are attending to them without being worried about it. The items comprising the measure we use might be viewed as reflecting social anxiety, because they involve concern and discomfort as the object o f someone's scrutiny (see the Appendix).

~This distinction is also made in the common parlance. The Random House Dictionary (1975, p. 1193) lists two distinct definitions of self-conscious: (1) "excessively conscious of oneself as an object o f observation by others," and (2) "conscious of oneself or one's own thoughts, well-being, etc."

Dimensions of the Self-Concept

289

The Development of Self-Consciousness

A large body of evidence, presented by Rosenberg (1979, Chap. 9), indicates that the onset of adolescence is a time of great disturbance in selfconcept development. At about age 12 years children show a distinct decrease in self-esteem and self-concept stability and, similarly, sharp increases in depression and self-consciousness. For some dimensions, this disturbance abates in later adolescence, but not for self-consciousness. One explanation for this phenomenon is found in Rosenberg's (1979, Chap. 8) discussion of the development of exterior and interior components of the self-concept. Prior to adolescence, children seem to conceive of the self almost wholly in terms of exterior components, such as observable behaviors, abilities, and physical characteristics. In contrast, adolescents are more likely to refer to the self in terms of a psychological interior and deal with such notions as thoughts, feelings, and traits. Rosenberg (1979, pp. 217-222) explains this difference by noting that it is in adolescence that people develop the ability to introspect; the young child is a "radical empiricist," responding to external stimuli, but the adolescent is a "psychological clinician," able to reflect on and contemplate internal stimuli. In addition he argues that, although both older and younger children can understand themselves in terms of a social exterior, only the older children have the communicative skills, which follow from cognitive development, to be able to understand the self as composed of psychological dimensions. Finally, the ability to conceptualize is a phenomenon emerging only in adolescence. Learning to synthesize and organize thoughts and feelings into appropriate concepts comes with adolescence and means that people can view themselves qualitatively differently than when they were younger. In terms of the distinctions made above, Rosenberg's presentation of the emergence of an interior component of the" selfconcept delineates the development of private self-consciousness. To understand the development of public self-consciousness (and by extension, social anxiety), we can make use of Mead's (1934) work on the development of the self. According to Mead, there is a social development of the self that parallels the psychological. In the egocentric stage, the young child is self-centered but not self-conscious. Progressing to what Mead calls the game stage, the child develops the ability to take the role of the other. In such role-taking, a person views his or her own possible acts from the other's vantage point. At first, this is limited to specific others in the child's environment. Later, the child begins to take the role of the "generalized other" by abstracting from particular referent others a composite perspective from which to view the self. Just as the advent of interior components of the selfconcept leads to private self-consciousness, so the ability to take the role of the other, to see oneself as others do, makes possible public self-consciousness. By adolescence, children have developed a sociocentric view of the self

290

Ellioll

through role-taking. In seeing themselves from the others' perspective, they can become aware o f themselves as the object of others' scrutiny; this new awareness of themselves is the emergence of public self-consciousness. Further, realizing that one is the focus o f others' attention can become threatening and generate a good deal of anxious concern. If so, the newly developed public self-consciousness will evolve into social anxiety. Self-consciousness (in any form) is one of the important consequences of individual development. Adolescence marks a striking phase for this aspect o f the self. Therefore, one might expect that analysis of the relationships between self-consciousness and certain dimensions o f the self-concept during this critical development phase would be especially fruitful. For this reason, we focus our analysis on adolescents (age 12-19 years).

The Self-Concept and Self-Consciousness Self-consciousness should be related to certain dimensions of the selfconcept. This paper does not attempt to explore the full range o f self-concept dimensions; rather, it is limited to a consideration o f a small number o f theoretically meaningful self-concept components that might have a bearing on self-consciousness. We focus on three such dimensions: self-esteem, vulnerability to criticism, and the tendency to fantasize. These self-concept dimensions were exogenous variables in an earlier structural analysis o f the self-concept (Elliott, 1982), which investigated their effects on the tendency to present an inauthentic self to others. Results showed that, as predicted, each self-concept dimension except private self-consciousness significantly affected the tendency to present a false front. The model presented herein is an elaboration o f the structural model analyzed by Elliott (1982). In particular, we posit a structural model in which self-esteem is hypothesized to affect self-consciousness in adolescents both directly and indirectly, through the intervening dimensions of vulnerability and the tendency to fantasize. At the outset, we spell out the theoretical reasons for expecting these dimensions to influence self-consciousness. In advancing these arguments, we are not oblivious to the possibility that in some cases the causal specification might be different. In particular, we realize that these relationships might often be reciprocally causal in nature. Here, we simply advance certain reasons for expecting the dominant direction of causal influence to be as indicated. The model presented is viewed as a theoretically sensible and plausible causal ordering but not the only possible one. Whether the proffered model is more appropriate than others awaits the analysis of a longitudinal study or the advent o f clearly appropriate instrumental variables for a nonrecursire model.

Dimensions of Ihe Self-Concep!

291

Further, we distinguish between private self-consciousness and social anxiety in analyzing the impact o f the selected self-concept dimensions. Research has generally revealed differences in kinds o f self-consciousness by finding low correlations between measures o f each kind or by showing how each kind differentially affects behavior (Buss, 1980). One further way of distinguishing kinds of self-consciousness is to examine the relationships posited by the model to see if the self-concept dimensions affect one kind of self-consciousness and not another. We discuss the effects of each selfconcept dimension on private self-consciousness and social anxiety. Vulnerability. One aspect of the self along which people can vary is their vulnerability to criticism. Some are able to take criticism and even use it as a spur to self-improvement. It is not that they expect none; it is that they know how to deal with criticism. Others are more thin-skinned. They are more prone to find criticism threatening and to react to it with with defensiveness and hostility. We expect that vulnerability should have an impact on private selfconsciousness. Criticism directs attention to deficiencies in the self. People who are bothered more by criticism are apt to concentrate on their inner characteristics. In attending to their thoughts, feelings, and traits, they can identify those that might cause them trouble and those that can help them avoid it. Those less vulnerable, by contrast, need not focus their attention on themselves so much. Feeling that they know how to deal with most criticism, they do not have to preoccupy themselves with their own personal qualities and can turn their attention elsewhere. Vulnerability would also be expected to have a bearing on social anxiety. Those who feel susceptible to criticism should worry greatly about the scrutiny of others, since other people are the sources o f most criticism. Their discomfort in most social situations should be quite high, because they have difficulty coping with negative reactions. The more bothered they are by criticism, the more anxious they should be in social situations, which potentially are sources of it. On the other hand, those for whom vulnerability is low should not be so anxious. Because they are not unduly disturbed by criticism, they need not worry so much about whether other people find things wrong with them. Fantasy. All o f us have at one time or another operated in a world o f fantasy, reveling in unconstrained imaginings of whom we would like to be, free from normative and reality constraints. Some of us spend more time in this imaginary world than do others. Especially during adolescence, when the self is so much an object of concern, youngsters may be "trying out" radically different selves to see if they are viable, or they may simply be escaping from a disappointing perceived self. It is reasonable to expect the tendency to fantasize to be positively related to private self-consciousness. In the world of imagination, the self is usually

292

Elliott

the central character. Fantasizing is likely to involve seeing the self in exciting situations, with heroic or other desirable characteristics. Attention is directed toward the self, usually as one wishes it to be. But the fantasy self inevitably comes up against the perceived real self, to the characteristic disadvantage of the latter. As a result, fantasy focuses the individual's attention on what he or she is like, rather than directing attention to external events. In contrast, an active fantasy life may have little bearing on social anxiety. It has no direct connection with worrying about what others think o f us. Because fantasizing is an act of introspection that involves a self that others are unlikely ever to see, it is reasonable to predict that social anxiety will be neither enhanced nor inhibited by the tendency to fantasize. Self-Esteem. Global self-esteem, as a dimensions o f the self-concept, is the person's evaluation of his or her own self. People vary in the extent to which they evaluate themselves positively, and so have high self-esteem, or negatively, with resultant low self-esteem (James, 1890/1950). When considering the general notion of self-consciousness (the tendency to make oneself the focus of one's own attention), one might expect that self-esteem would have a negative effect on self-consciousness. Some researchers (e.g., Murphy, 1947; Barron, in Jahoda, 1958, p. 28) have argued that the self tends to come to the forefront o f attention when it is problematic; so long as the self-fulfills its expectations, it is in the background of awareness. Given the motivational status o f self-esteem, the self is obviously more problematic for those with low self-esteem. Therefore, one might argue that low-self esteem people, lacking secure confidence in their own self-worth, are more likely to focus attention on themselves. In applying this general argument to each particular form o f selfconsciousness, it is useful to distinguish the social and personal impacts o f low self-esteem. We expect that self-esteem will have a negative impact on social anxiety. Recall that the socially anxious person is concerned with the reactions of others to his or her presented self. Those with high self-esteem respect themselves and need not be worried about others' reactions to the impressions they convey. They are convinced that others will respond favorably to the self they are presenting, and as a result, they feel less social anxiety. Those with low self-esteem, in contradistinction, envision themselves as somehow seriously deficient; their assessment of their own personal qualities is low. They may be more worried about what others would think of them. Expecting social rejection, ridicule, or disapproval,~they are uncomfortable at the thought o f others' scrutiny. In contrast, we suspect that self-esteem will have little or no effect on private self-consciousness. Private self-consciousness deals with a person's tendency to focus on his or her internal states. Although those undesirable factors leading to a problematic self and low self-esteem may tend to focus attention on one's internal states, there may be strong forces pulling in the

Dimensions of the Self-Concept

293

opposite direction. Indeed, Duval and Wicklund (1972) suggest that situational self-awareness is avoided to the extent that a problematic self is revealed by such a focus. A parallel argument may hold for private selfconsciousness. Social anxiety may not be avoidable, because one cannot control the discerning scrutiny of others; but it may be much easier defensively to block out one's own focus on internal states, thereby avoiding the development of private self-consciousness. According to this argument, highself-esteem individuals should be no more or less aware o f their internal characteristics than those with low self-esteem, although their reactions to a given level of awareness may differ greatly. The consequences of private self-consciousness may be different for high- and low-self-esteem individuals, but not the tendency itself. If these expectations are correct, the differential impact of self-esteem would serve as a further source of discrimination between private self-consciousness and social anxiety. The effect of self-esteem on both forms o f self-consciousness may also be indirect, in that the other dimensions o f the self-concept under consideration here may mediate the self-esteem-self-consciousness relationship. There is good reason to presume primacy for self-esteem. Researchers in the self have long recognized that the need to think well of oneself is a powerful motiv a t o r f o r human conduct. Indeed, Rosenberg (1979) refers to self-esteem as one o f the two primary motivators located within the self-concept (the other being self-consistency), a view shared by Schwartz and Stryker (1971). Some would even assert that self-esteem is the dominant force in a person's motivational system (Allport, 1961, pp. 155-156; Kaplan, 1975, p. 10). It is therefore reasonable to expect that self-esteem would be a source of influence on other aspects of the self-concept in general and that it would be prior to other determinants of self-consciousness in particular. Hence, our structural model includes self-esteem as an exogenous variable, vulnerability and the tendency to fantasize as intervening variables, and private self-consciousness and social anxiety as the ultimate dependent variables. In this manner, we will be able to determine the extent to which the effect of self-esteem on either form o f self-consciousness is mediated by the other self-concept variables. We now turn to a justification for each o f the intervening links with selfesteem. We hypothesize that self-esteem should negatively influence vulnerability. Those with high self-esteem are convinced of their own self-worth and do not interpret every criticism as a personal attack. In contrast, those with low self-esteem are painfully aware of their own shortcomings; criticism confirms their own worst fears about themselves. Therefore, low-self-esteem actors should experience a greater vulnerability to criticism than those with high self-esteem. As a result, one reason that low-self-esteem people are selfconscious is because they are more vulnerable to criticism, which in turn leads to greater self-consciousness.

294

Ellion

Fantasy should also mediate the relationship between self-esteem and private self-consciousness. People are more likely to need to escape into an imaginary world when the world in which they live is unsatisfactory. Highself-esteem people, being generally satisfied with themselves, do not need to spend a great deal of time pretending or wishing they were someone else. By contrast, those with low self-esteem can forget for a while that they are not the person they would like to be by fantasizing. Hence, low self-esteem may lead to a greater tendency to fantasize. Because we have hypothesized no effect for fantasy on social anxiety, we are forced to conclude that the tendency to fantasize will not mediate the self-esteem-social anxiety relationship.

METHOD Sample The data for this analysis were collected from public-school children in grades 3 through 12 in Baltimore City in 1968. A random sample of 2625 pupils from 25 schools was drawn from the population of third-to twelfthgrade students. After eliminating students who had withdrawn, 1917 were interviewed (79.2% of the children still registered, or 73.0% of all children originally drawn from the central records). The 1098 adolescent ranging in age from 12 to 19 years served as the basis for the analysis. [For a more complete description o f the sampling procedure, see Rosenberg and Simmons (1972).]

Measures To measure self-esteem, we used a global measure containing six items. Each item ascertained the respondents' general evaluation of themselves. (Example: "Another kid said, 'I am no good.' Do you ever feel like this?" If yes, ask: "Do you feel like this a lot or a little?") Vulnerability is based on a three-item scale in which respondents indicated how sensitive they were to criticism. (Example: "Do your feelings get hurt very easily, pretty easily, or not very easily?") Six items comprised the fantasy scale and reflected the respondents' tendency to spend time in an imaginary world. (Example: "Would you say you daydream a good deal o f the time? Yes or no.") The private self-consciousness scale consisted of six items which tapped the respondents' preoccupation with their own personal characteristics (Example: "How much time do you spend thinking about who you are, what you are, or what you are like?. A lot of time, some time, or not very much time."). Finally, social anxiety was assessed by six items, each of which measured the respondents' concern with the thought of others' scrutiny. (Example: "If the teacher asked

Dimensions of the Self-Concept

295

you to get up in front of the class and talk a little bit about your summer, would you be very, a little, or not at all nervous?") A complete list of items used in the final model is available in the Appendix. A few words about the measures used in this study are in order. First, most of the indices evinced acceptable levels of reliability. The following are the internal consistency scores for each of the scales, as measured by Cronbach's alpha: self-esteem, 0.687; vulnerability, 0.659; fantasy, 0.492; private self-consciousness, 0.603; and social anxiety, 0.650. One would normally expect a somewhat lower level of internal consistency among adolescents than adults; hence, the alpha coefficients for all the scales except fantasy appear reasonable. However, none of the scales is a perfectly valid measure of its associated underlying component. This provided us with the major impetus for using an unobserved variable structural equations model as a way of examining the hypothesized relationships corrected for the less than perfectly reliable measures. Second, the measures of private self-consciousness and social anxiety in the present study are similar but not identical to those measures developed by Fenigstein e t al. (1975) in their initial investigation of self-consciousness. The data used in this analysis were collected in 1968 as part of an intensive study of the self-concept of young people. Unfortunately, because this study predates the work of Fenigstein et al. (1975), the self-consciousness scales they developed were not available. However, the items comprising the private self-consciousness and social anxiety scales used in the present study bear a striking resemblance to the comparable scales developed by Fenigstein et al. (1975), and they appear reasonably face valid. 3 It is also interest3For example, compare the following items taken from the scales of Fenigstein et aL (1975) with their counterparts in the Appendix. Private Self-Consciousness I'm always trying to figure myself out. Generally, I'm not very aware of myself. I reflect about myself a lot. I'm often the subject of my own fantasies. 1 never scrutinize myself. I'm generally attentive to my inner feelings. l'm constantly examining my motives. 1 sometimes have the feeling that I'm off somewhere watching myself. l'm alert to changes in my mood. I'm aware o f the way my mind works when I work through a problem. (Note: The two items dealing with fantasy, which one might feel more properly reflect our proposed tendency to fantasize, showed the lowest factor loadings on the private self-consciousness scale in Fenigstein's analysis.) Social Anxiety It takes time to overcome my shyness in new situations. 1 have trouble working when someone is watching me. 1 get embarrassed very easily. I don't find it hard to talk to strangers. t feel anxious when I speak in front o f a group. Large groups make me nervous.

296

Elliott

ing to note that Fenigstein et al. (1975) report a correlation between their scales of 0.11 in their major sample of subjects; this is precisely the correlation we found between our scale of social anxiety and that of private selfconsciousness (0.112). The fact that the scales share the same relationship in both studies is consistent with the view that they are parallel in nature.

Selecting the Model

Each of the constructs mentioned above is a latent property of the person, inherently unobservable directly but revealed in one's responses to the items in the aforementioned scales. However, as the internal analyses of each scale reveal, these measures are somewhat flawed. Rather than analyze the relationships among the scales, it would be better to investigate the relationships among the latent traits they imperfectly tap. Such a procedure is provided by the analysis of covariance structures developed by Joreskog and his colleagues (Joreskog, 1978, 1979). We began by specifying the model to be analyzed. This model includes both unobservable and observable variables. The unobservable variables consist of the latent dimensions of the self-concept (self-esteem, vulnerability, the tendency to fantasize, private self-consciousness, and social anxiety), and the observable variables consist of the items designed to reflect one's standing on these dimensions (see the Appendix). The model is actually composed of two parts. The measurement model specifies the relationships between the observed and the unobserved variables by designating which of the dimensions of the self-concept is tapped by each item. The structural model specifies the relationships among the unobserved variables, as specified by the theory. One further constraint applies. The model must be identified (Wiley, 1973); that is, there must be a sufficient number of observed relationships to estimate the parameters of the model. Figure 1 presents the structural model under analysis. Self-esteem is an exogenous variable, vulnerability and the tendency to fantasize are intervening variables, and private self-consciousness and social anxiety are the ultimate dependent variables. In addition, we estimated the correlations between the disturbances of the intervening variables and those of the ultimate dependent variables, respectively. These disturbance correlations are precisely the partial correlations between the variables in question, controlling for those variables that are causally prior to them. They indicate the extent to which the relationship between these variables is spurious, due to their common antecedents. The analysis of. covariance structures allow the researcher to test how well a given model reproduces the correlations between the observed varia-

Dimensions of the Self-Concept

297

.9

.024

Fantasy.~

-~19

',,,,,___

.169

~

I/

Vulnerability /

~

Private

.963

-

.

Fig. 1. Structural model for self-consciousness. Figures in parentheses are standard errors. All coefficients are standardized path (regression) coefficients.

bles, according to the rules of path analysis (Hauser and Goldberger, 1971). Joreskog's algorithm yields maximum-likelihood estimates of the parameters for the model, which are then used to reproduce the interitem correlations. The aim is to minimize the difference between the matrix of observed correlations (S) and the matrix of correlations reproduced by the parameters (E). For a detailed explanation of the minimization method, see Joreskog (1978, 1979). Comparing S and E, one can make a goodness-of-fit test, using a likelihood ratio chi-square statistic, to test the null hypothesis that the given model is appropriate to represent the relationships underlying the observed data against the null hypothesis that the model is not so appropriate. A significant chi-square suggests that the model does not fit and must be augmented by adding additional parameters, subject to the constraints of identifiablity. The improvement in goodness of fit can be tested. The difference in chi-square values between the original and the augmented model is itself distributed as chi-square, with degrees of freedom equal to the difference between their respective degrees of freedom. Our approach to deriving a final model was as follows. Using Lisrel V (Joreskog and Sorbom, 1981), we began by estimating the full model that included all 27 items from all the scales. The fit was unacceptable (x 2 = 827.41, df = 314, p < 0.0005). Indications of where the model failed can be gleaned by examining the modification indices; the modification index estimates the minimum improvement in goodness of fit (i.e., decrement to chisquare) possible by adding a particular parameter to the model. Our exami-

-0.086 -0.064

-0.057 -- 0.099 -0.048 -0.036

-0.067 - 0.096 -0.036 -0.048

-0.194 -0.163 -0.209

1.000 0.462 0.402

-0.004 - 0.050 0.015 -0.029

--0.065 --0.046

-0.190 -0.136 -0.177

1.000 0.402

v250

Self-Esteem

v248

-0.090 -0.049

-0.161 -0.142 -0.168

VU v215 v216 v232

FA v230 v233 PR v169 v 172 v209 v211

1.000 0.552 0.420 0.349

SE a v225 v248 v250 v253

v225

--0.120 -- 0.131 -0.133 -0.108

-0.114 --0.056

-0.160 -0.142 -0.149

1.000

v253

0.116 0.095 0.103 0.I11

0.059 0.047

1.000 0.390 0.490

v215

0.I08 O. t 09 0.132 0.101

0.023 --0.006

1.000 0.360

v216

Vulnerability

Table 1, Zero-Order Correlations Among the Observed Variables

0.075 0.061 0.090 0.103

0.039 0.041

1.000

v232

0.053 0.110 0.066 0.074

-0.040 0.017

0.139 0.101 0.091 0.124

0.058 0.050

v233

1.000

Fantasy

-0.116 -0.166

1.000 0.465

v230

-0.107 -0.125

-0.014 0.017

1.000 0.369 0.510 0.422

v169

-0.141 -0.162 0.138 0.201

--0.005 0.012

1.000 0.519 0.457

v 172

-0.022 0.009

1.000 0.537

v209

Private Self-Consciousness

-0.091 -0.099

--0.014 -0.007

1.000

v211

0.153 0.181

Note. Correlations are based on 1025 observations. Calculations are based on more significant digits than presented here. The complete items are found in the Appendix. "SE, Self-Esteem; VU, Vulnerability; FA, Fantasy; PR, Private Self-Consciousness; SA, Social Anxiety.

FA v230 v233 PR v169 v172 v209 v211 SA v354 v355

SA v354 v355

1.000 0.505

v354

1.000

v355

Social Anxiety

0.207 0.247

to

O

¢3

O

o

300

Elliott

nation revealed that the poor fit was due to the inadequacy of some of the items used as observable variables. This inadequacy emerged in one of two ways: (1) the item in question appeared to tap more than one unobservable variable, suggesting a lack of discriminant validity; and (2) the disturbance of the item in question was correlated with that of another item, indicating the presence of an unanticipated factor that would account for some of the association between the items. We eliminated contaminated items according to the following criteria: (1) items showing poor discriminant validity were deemed most threatening to the analysis and were eliminated first; (2) items from one scale whose disturbances were correlated with those of items from a different scale were eliminated if there were more than one or two such correlations; (3) items whose disturbances were correlated with those o f items from the s a m e scale were considered least threatening and remained in the model, unless their removal resulted in an appreciably better fit; and (4) items whose coefficients in the measurement model were less than 0.400 were eliminated.

RESULTS The final model made use of 15 observed variables, utilizing 1025 respondents. Table I presents the zero-order correlations among the observed variables. Turning to the analysis of covariance structures, Table II gives the parameter estimates for the measurement model, while Fig. 1 presents estimates for the structure model coefficients. All coefficients are standardized path (regression) coefficients. Measuring the goodness of fit of the model, the likelihood ratio chi-square value was 91.52, with 78 degrees of freedom 6o = 0.140). These results indicate that we should not reject the null hypothesis that the model fits the data. The goodness o f fit index calculated by Lisrel V was 0.982 (maximum value, 1.0; adjusted value, 0.972). Examining the modification indices revealed that the highest index was 5.97 (for a potential correlated disturbance involving measures of private self-consciousness and fantasy), indicating that very little improvement would be gained in adding more parameters. Consider first the measurement model (Table II). Most of the items included in the final model are reasonably good reflectors o f their designated unobserved variable. The minimum validity coefficient was 0.506, for a measure of vulnerability; the median validity coefficient was 0.646. It is important to recall that the pruning process described above left us with items displaying a high degree o f discriminant validity. Turning to the coefficients in the structural model (Fig. 1), the results offer strong support for the hypothesized relationships. Beginning with vul-

Dimensions of the Self-Concept

301

Table !i. Measurement Model for Self-Consciousness Parameter estimate

SE

SelgEsteem v225 v248 v250 v253

0.697 0.781 0.600 0.517

0.032 0.032 0.033 0.034

Fantasy v230 v233

0.841 0.543

0.109 0.072

Vulnerability v215 v216 v232

0.646 0.506 0.640

0.034 0.033 0.034

Private Self-Consciousness v169 v172 v209 v211

0.599 0.621 0.768 0.658

0.031 0.031 0.030 0.030

Social Anxiety v354 v355

0.573 0.722

0.040 0.053

Item

Note. The complete items are found in the Appendix.

nerability, we find that p e o p l e w h o are especially sensitive to o t h e r s ' criticisms are also likely to find themselves p r e o c c u p i e d with their o w n characteristics a n d a n x i o u s a b o u t the o t h e r s ' scrutiny. T h e results f o r f a n t a sy are, as expected, mixed. Those who spend m o r e time in an i m a g i n a r y world are likely to experience greater private self-consciousness; in contrast (but also a c c o r d i n g to p r e d i c t i o n ) , f a n t a s y h a d no effect o n social anxiety. F o r selfesteem, the results reveal that, a l t h o u g h there is a slight t e n d e n c y for low self-esteem t o lead to g r e a t e r p r i v a t e self-consciousness, as expected, this effect was far f r o m significant. O n the o t h e r h a n d , self-esteem h a d the predicted effect o n social anxiety: t h o s e with low self-esteem a r e m o r e likely to be nervous a b o u t the r e a c t i o n s o f o t h e r s to their b e h a v i o r , 4 It w o u l d be a m i s t a k e to c o n c l u d e f r o m these results t h a t self-esteem a f f e c t e d social a n x i e t y b u t not p r i v a t e self-consciousness. T h e coefficients given in Fig. 1 r e p r e s e n t o n l y the direct effects o f t h e c a u s a l l y p r i o r v a r i a -

4We examined the statistically significant structural coefficients to determine whether the apparent differences in magnitude were themselves statistically significant. Tests for equality of coefficients revealed that the effect of vulnerability on social anxiety was greater than any other effect; none of the other significant effects differed from each other in absolute magnitude.

302

Elliott

bles. But the posited model also allows for the i n d i r e c t effects of self-esteem on each form of self-consciousness, as mediated by vulnerability and the tendency to fantasize. [See Alwin and Hauser (1975).] Beginning with private self-consciousness, the total effect of self-esteem, as generated by the analysis o f covariance structures, was - 0 . 1 2 1 ; as Fig. 1 reveals, when the intervening variables are taken into account, this relationship is reduced to - 0 . 0 1 9 . In other words, almost all (84.3%) of the relationship between selfesteem and private self-consciousness is indirect, mediated by vulnerability and the tendency to fantasize. The lion's share of the indirect effect is transmitted through vulnerability (indirect coefficient = 0.076; 74.5O/o of the indirect relationship); the tendency to fantasize was a much weaker mediator (indirect coefficient = - 0 . 0 2 6 ; 25.5% o f the indirect relationship). Turning to social anxiety, we find that the total effect o f self-esteem is - 0 . 2 9 0 . This amount is more equally split between direct and indirect effects, for when the other variables are taken into account, the direct effect (shown in Fig. I) is - 0 . 1 3 6 . Once again, this means that a substantial (but lesser) part (53.1 o70)of the total relationship is mediated by vulnerability and the tendency to fantasize. By far, most of the mediation is through vulnerability (indirect coefficient = - 0 . 1 4 9 ; 96.80/o of the indirect relationship). The indirect effect through fantasy is negligible (indirect coefficient = - 0 . 0 0 5 ; 3.2°7o o f the indirect relationship). Our posited model also permitted disturbance correlations between the intervening variables of vulnerability and fantasy on the one hand, and the two self-consciousness variables, on the other. As mentioned above, these disturbance correlations are precisely the partial correlations between the latent variables involved, controlling for all causally prior variables. Beginning with the intervening variables, inspection of Fig. 1 reveals that the partial correlation between vulnerability and fantasy is not significantly different from zero. In other words, the relationship between vulnerability and fantasy is largely spurious, due to their common cause of self-esteem. In contrast, there is a significant correlation between private self-consciousness and social anxiety, even when controlling for the other variables in the model. This means that there are additional sources for the covariation between the two forms of self-consciousness. 5 Finally, we note that there are only two instances in which the error terms for the observed variables were significantly correlated with each other. Such a correlation indicates a source of relationship for the observed varia-

~This does not imply that we have explained as much variance as we can in each dependent variable separately.Other determinants of each kind of self-consciousnessexist, as the modest amounts of explained variance indicate. Explaining a correlation between two variables is very different from explaining variance in one variable.

Dimensions of the Self-Concept

303

bles in question that is unaccounted for by the model. In one case, the disturbances for a measure of private self-consciousness and fantasy were correlated; in the second case, two items measuring self-esteem shared a disturbance correlation. Because two is only a small fraction of the total number of possible disturbance correlations between the observed variables, and because the magnitude of these correlations is relatively small (-0.107 and 0.091, respectively), we believe that they do not pose a serious threat to the results obtained in our analysis.

DISCUSSION Much of the previous research on self-consciousness has focused on verifying empirically the existence of and distinctions among the various forms of self-consciousness identified in theoretical developments. This has been done either by examining correlations involving measures of each form or by demonstrating their different effects on behavior. Although these are indeed important, evidence from one further source has been lacking. Progress in the study of the self-concept has demonstrated the necessity of simultaneously considering its various aspects, in order to understand the relationships that exist among them and their joint effects on the individual. The present study investigated the relationship of several dimensions of the self-concept with two forms of self-consciousness: private self-consciousness (in which one's attention is chronically focused on one's own internal states) and social anxiety (in which one experiences discomfort because others' attention is focused on the self). Results from a structural equations analysis showed that the data supported the hypotheses. Further, the support came from different measures of self-consciousness from those developed by Fenigstein et al. (1975), thereby affording theoretical replication (La Sorte, 1972) for the constructs of private self-consciousness and social anxiety. Analysis of the data gathered from adolescents (ages 12-19 years) revealed that, as expected, self-esteem, vulnerability to criticism, and the tendency to fantasize had a significant impact on self-consciousness. More important, however, the analysis demonstrated that, although some dimensions of the self-concept would significantly affect both kinds of selfconsciousness, other dimensions would significantly affect only one. In particular, we found that vulnerability to criticism was a common cause for private self-consciousness and social anxiety. The more vulnerable one is to criticism, the greater the tendency to be aware of one's personal characteristics and uncomfortable under the scrutiny of others. In contrast, self-esteem

304

Elliott

and fantasy each significantly affect only one form of self-consciousness directly. Although low self-esteem leads to more social anxiety, it apparently does not increase private self-consciousness. Conversely, the tendency to fantasize does make one more aware of one's own personal characteristics but does not make one more concerned about the attention of others. Although self-esteem had no direct effect on private self-consciousness, an indirect effect, mediated largely by vulnerability to criticism, did emerge. Apparently, low-self-esteem individuals feel a greater vulnerability to other's criticism, and this heightened vulnerability in turn leads to a greater focus on one's own internal states. In a lesser vein, those with low self-esteem also tend to fantasize more, which leads to greater private self-consciousness. Similarly, self-esteem had an indirect self on social anxiety, mediated almost entirely by vulnerability. The greater vulnerability experienced by lowself-esteem people works to make them more worried about how others will react to them. In contrast to private self-consciousness, there is still an appreciable direct effect of self-esteem on social anxiety. A major purpose of the present paper is to contribute to a change of focus in self-concept research. Instead of dealing with dimensions of the selfconcept in terms o f "lists," we believe that greater theoretical advances can be made if we begin to treat the various dimensions o f the self-concept in terms of the possible structures that underlie them. We have suggested that it is theoretically meaningful to view self-consciousness as subject to the influences of certain self-concept dimensions. In particular, we have considered self-esteem as a "core" or "master" dimension that operates on selfconsciousness both directly and through mediating properties o f vulnerability to criticism and the tendency to fantasize. Further, we have shown that distinguishing between private self-consciousness and social anxiety is warranted, not only on the basis of their conceptual definitions, but also because they are differentially affected by certain dimensions of the self-concept. The results of our analysis have revealed a new and important theoretical source for distinguishing between the two forms o f self-consciousness. Much more research on the relationship between dimensions of the selfconcept and self-consciousness is needed. For example, one might consider such aspects as self-concept stability across time and situations, clarity (how well defined the components are), or salience (how atypical the components are) as determinants or consequences of self-consciousness. Further, we need to understand better the influence o f socialization processes on the development of self-consciousness. Increased research in the structure of the selfconcept can lead to a fuller understanding of the complexities of individual growth and development.

Dimensions of the Self-Concept

305

REFERENCES Allport, G. (1961). Pattern in Growth and Personality, Holt, Rinehart and Winston, New York. Alwin, D. F., and Hauser, R. M. (1975). The decomposition of effects in path analysis. Am. Sociol. Rev. 40: 37-47. Argyle, M. (1969), Social Interaction, Tavistock, London. Blumer, H. (1969). Symbolic Interactionism, Prentice-Hall, Englewood Cliffs, N.J. Buss, A. H. (1980). Self-Consciousness and Social Anxiety, W. H. Freeman, San Francisco. Buss, A. H., and Scheier, M. F. (1976). Self-consciousness, self-awareness, and self-attribution. J. Res. Personal 10: 463-468. Cooley, C. H. (1902). Human Nature and the Social Order, Scribner's, New York. Duval, S., and Wicklund, R. A. (1972). A Theory of Objective Self-Awareness, Academic Press, New York. Elliott, G. C. (1982). Self-esteem and self-presentation among the young as a function of age and gender. 2". Youth Adoles. 11: 135-155. Elliott, G. C., Rosenberg, M., and Wagner, M. (1984). Transient depersonalization in youth. Soc. Psychol. Q. 47: 115-129. Fenigstein, A., Scheier, M. F., and Buss, A. H. (1975). Public and private self-consciousness: Assessment and theory. J. Consult. Clin. Psychol. 43: 522-527. Hauser, R. M., and Goldberger, A. S. (1971). The treatment of unobservable variables in path analysis. In Costner, H. L. (ed.), Sociological Methodology 1971, Jossey-Bass, San Francisco, pp. 81-117. Jahoda, M. (1958). Current Concepts of Positive Mental Health, Basic Books, New York. James, W. (1890/1950). The Principles of Psychology (reprint), Dover, New York. Joreskog, K. G. (1978). Structural analysis of covariance and correlation matrices. Psychometrika 43: 443-447. Joreskog, K. G. (1979). Structural equation models in the social sciences: Specification, estimation, and testing. In Joreskog, K. G., and Sorbom, D. (eds.), Advances in Factor Analysis and Structural Equation Models, Abt Books, Cambridge, Mass., pp. t 05-127. Joreskog, K. G., and Sorbom, D. (1981). Lisrel: Analysis of Linear Str,lctural Relationships by the Method of Maximum Likelihood, Version V, National Education Resources, Chicago. Kaplan, H. B. (1975). Self-Attitudes and Deviant Behavior, Goodyear, Pacific Palisades, Calif. LaSorte, M. A. (1972). Replication as a verification technique in survey research: A paradigm. Sociol. Q. 13: 218-227. Mead, G. H. (1934). Mind, Self, and Society, University of Chicago Press, Chicago. McGuire, W. J., and Padawer-Singer, A. (1976). Trait salience in the spontaneous self-concept. J. Personal. Soc. PsychoL 33: 743-754. Murphy, G. (1947). Personality, Harper, New York. Random House Dictionary (revised ed.) (1975). Random House, New York. Rosenberg, M. (1965). Society and the Adolescent Self-Image, Princeton University Press, Princeton, N.J. Rosenberg, M. (1979). Conceiving the Self, Basic Books, New York. Rosenberg, M., and Simmons, R. G. (1972). Black and White Self-Esteem: The Urban Schoolchild, Rose Monograph Series, The American Sociological Association, Washington, D.C. Scheier, M. F., Buss, A. H., and Buss, D. M. (1978). Self-consciousness, self-report of aggressiveness, and aggression. J. Res. Personal. 12: 133-140. Schwartz, M., and Stryker, S. (1971). Deviance, Selves, and Others, Rose Monograph Series, The American Sociological Association, Washington, D.C. Stryker, S. (1968). Identity salience and role performance: The relevance of symbolic interaction theory for family research. J. Marriage Family 30: 558-564. Turner, R. G., and Peterson, M. (1977). Public and private self-consciousness and emotional expressivity. J. Consult. Clin. Psychol. 45: 490-491. Turner, R. G., Scheier, M. F., Carver, C. S., and lckes, W. (1978). Correlates of selfconsciousness. 2". Personal. Assess. 42: 285-289.

306

Elliott

Wells, L. E., and Marwell, G. (1976). Self-Esteem: Its Conceptualization and Measurement, Sage, Beverly Hills, Calif. Wicklund, R. A. (1975). Objective self-awareness. In Berkowitz, L. (ed.), Advances in Exper#nental Social Psychology, VoL 8, Academic Press, New York, pp. 233-275. Wiley, D. E. (I983). The identification problem for structural equation models with unobserved variables. In Goldberger, A. S., and Duncan, O. D. (eds.), StructuralEquation Models in the Social Sciences, Seminar Press, New York, pp. 69-83. Wiley, R. (1979). The Self-Concept: Revised Edition, Vol. 2, 77~eoryand Research on Selected Topics, University of Nebraska Press, Lincoln.

Dimensions of the Self-Concept APPENDIX:

307

ITEMS

COMPRISING

DIMENSION

THE

SELF-CONCEPT

SCALES

Self-Esteem A n o t h e r kid said, "I a m no good." Do you ever feel like this? (If yes, ask): Do you feel like this a tot or a little? (v225) A n o t h e r kid said, "I think I am no good at all." Do you ever feel like this? (If yes, ask): Do you feel like this a lot or a little? (v248) Another kid said, "I'm not much good at anything." Do you ever feel like this? (If yes, ask): Do you feel like this a lot or a little? (v250) A kid told me, "There's a lot wrong with me," Do you ever feel like this? (If yes, ask): Do you feel like this a lot or a little? (v253)

Vulnerability When someone gets angry at you, do you get very upset, pretty upset, or not upset? (v215) W h e n someone laughs at you, do your feelings get hurt very easily, pretty easily, or not very easily? (v216) Do your feelings get hurt very easily, pretty easily, or not very easily? (v232)

Fantasy Would you say you daydream a good deal o f time? Yes or no. (v230) A kid told me, "I don't daydream very m u c h . " Is this true or not true for you? (v233)

Private Self-Consciousness How m u c h time do you spend thinking about what you are like? Do you spend a lot of time, some time, not m u c h time, or not any time? (v169) Do you ever think to yourself: "What kind of person am I? What am I like?" (If yes, ask): How often do you think, " W h a t kind of laerson a m I? W h a t a m I like?" Often, sometimes, or hardly ever. (v172) How m u c h time do you spend thinking about the kind of person you are? A lot o f time, some time, not very m u c h time, or not any time. (v209) How m u c h time do you spend thinking about who you are, what you are, and what you are like? A lot of time, some time, or not very m u c h time. (v21 I)

Social Anxiety If the teacher asked you to get up in front of the class and talk a little bit about your summer, would you he very nervous, a little nervous, or not at all nervous? (v 354) If you did get up in front of the class and tell them about your s u m m e r , would you think a lot, a little bit, or not at all about how all the kids were looking at you? (v355)

Dimensions of the self-concept: A source of further distinctions in the nature of self-consciousness.

This study distinguished between forms of self-consciousness (private self-consciousness and social anxiety) and investigated the effect of self-estee...
1MB Sizes 0 Downloads 0 Views