For Use in GPS 520 ONLY!

Excerpts taken from Hoyle, R. H., Kernis, M. H. Leary, M. R. & Baldwin, M. W. (1999). Selfhood: Identity, esteem, regulation (pp. 63-77).  Boulder, CO: Westview Press.

Organization of Self-Knowledge

The application of theory and methods from cognitive psychology to the study of the self-system moved the focus of research beyond questions about the basic content of self-concept to questions regarding the organization of self-referent thoughts, feelings, and motives in memory. Such research might be considered "second-generation" self-concept research. Prior to the late 1970s, most data on the self-concept were procured using paper-and-pencil questionnaires that provided scores across a variety of pre-established domains. Although such data are useful for indexing the differential importance of frequently occurring domains of self-knowledge in respondents, they provide little insight regarding how self-knowledge in those domains, and others not commonly addressed, influence self-referent information processing. Second-generation research on the self-concept relies more on ad hoc and respondent-generated lists of attributes that are rated or sorted by respondents in such a way that mathematical or statistical strategies can be used to index implicit characteristics of self-knowledge (e.g., the self-with-other profiles described earlier). Although these methods can provide information similar to that in first-generation self-concept research, they have the advantage of providing information about how self-knowledge is arrayed in memory and, by inference, how it influences self-referent thoughts, feelings, and behaviors.

Substantive Self-Knowledge

The groundwork for prominent contemporary strategies of studying the self-concept was laid almost thirty years before the first published accounts of second -generation self-concept research appeared. Psychologists studying human sensation and perception developed mathematical models of the processes involved in recognizing and remembering visual shapes and patterns (Atneave 1954; Shannon and Weaver 1949). These models drew from theories about how information is processed to characterize basic human activities such as problem solving and movement (Attneave 1959; Wiener 1948). Information theories make two important assumptions that are fundamental to second-generation self-concept research: (1) information can be quantified and (2) judgments and decisions involving a collection of information are affected by the associations among individual pieces of information.

Working from these two basic tenets, Linville (1985) developed an influential model and method for studying the organization of self-knowledge.  According to the model, the self-concept is a multifaceted cognitive structure, and people vary in the degree of complexity evident in their self-concept. The self-complexity of a people's self-concept is defined as the number of different ways in which people conceptualize themselves-self-aspects (e.g., roles, emotional states, self with others) and the interrelatedness of those self-aspects. Linville further posited that self-complexity should influence the way people process self-referent information that arises during the course of everyday life and, therefore, affect their emotional reactions to such information. People with a relatively large number of independent self-aspects should be less affected by self-referent information (e.g., performance feedback) because, in the information -processing context provided by the self-system, the information-good or bad-would bear on only a small, relatively independent part of the whole system. Conversely, people with a small number of overlapping self-aspects would be more strongly affected by self-referent information because it activates (i.e., pulls into the working self-concept) a larger part of the self-system, and the activation spreads to interrelated self-aspects not directly implicated by the information but linked in memory.

An important contribution of the research on self-complexity is the means by which self-complexity is measured. Research participants are provided index cards on which are written words or phrases (e.g., smart, clumsy). Their task is to describe themselves by sorting the cards into groups according to which ones they believe belong together. In addition to these index cards, usually numbering around thirty, respondents are provided blank cards for creating duplicates of words or phrases they wish to include in multiple piles. The card sort provides the raw data for computing the H statistic (Attneave 1959; Scott 1969), which reflects the degree of redundancy in the use of information evident in the self-descriptive sort.

An alternative approach to generating the raw data for computing self-complexity is illustrated in Table 4.1. Using the Self-Description Grid (McKenzie and Hoyle 1996), a portion of which is shown in the figure, respondents "sort" a collection of trait adjectives with reference to a series of self-aspects either provided by the investigator or generated by each respondent. These data can then be used to compute the H statistic mentioned earlier. The computation of the H statistic is somewhat complex (see Attneave 1959, 5-8, 24-25 for an excellent tutorial); however, the idea behind the number is rather simple and bears elaborating. Referring again to Table 4.1, note that a hypothetical respondent has indicated with an X the self-aspects of which each adjective, in his or her view, is descriptive. Comparing across rows, one can see that the pattern of Xs is identical for some adjectives, whereas the pattern is unique for others. For instance, the adjectives good and helpful were used to describe the self-aspects romantic partner, friend, son/daughter, and worker, but not student. Tired and clumsy were used to describe self as student, son/daughter, and worker, but not romantic partner or friend. Otherwise, there is no redundancy in the patterns that accompany the adjectives. If a unique pattern of usage was associated with each adjective, then the respondent would obtain the maximum self-complexity score. If, on the other hand, the same pattern of usage was associated with every adjective, then the respondent would obtain the minimum self-complexity score.  

Self-complexity, measured in this way, has produced a number of intriguing findings that provided new insights into the self-system and self-regulation. In her seminal research on self-complexity, Linville (1985) conducted two studies. In one study, undergraduates performed the card sort described earlier before describing their mood on a series of items presented by computer. When the computer ostensibly stopped working, participants were asked to move on to the next phase of the study, which involved completing a task supposedly related to intelligence. When participants completed the task, the experimenter scored it and notified each participant that he or she had either performed very poorly or very well (feedback was randomly assigned). At that point, the computer appeared to start working, and the participants were asked to reenter the information about their mood. Comparisons of mood scores revealed that people low in self-complexity are more greatly influenced than their high complexity counterparts by performance feedback, particularly when it is negative.

In a second study, Linville (1985) enlisted a group of college women to complete the card-sorting task and keep a diary of their mood for fourteen days. As predicted, self-complexity was inversely correlated with daily fluctuations in mood. That is, the more complex the participants' selfconcept, the less extreme their mood swings during the two weeks of the study. The findings from subsequent research on self-complexity and mood suggest that they are not always associated (e.g., Emmons and King 1989; Rhodewalt, Madrian, and Cheney 1998). In fact, there is evidence to suggest that such fluctuations might be attributed more directly to self-esteem differences than to self-complexity per se (Campbell, Chew, and Scratchley 1991). Resolution concerning the conflicting findings relevant to this core prediction may lie in the correspondence between the frame of reference for the self-ratings and the outcomes being considered. For instance, when selfratings focus on possible selves (Markus and Nurius 1986), the resultant self-complexity score is correlated with emotional reactions to feedback regarding likely future performance but not feedback regarding present performance (Niedenthal, Setterlund, and Wherry 1992). Conversely, actual self-complexity predicts reactions to evaluations of present performance but not likely future performance. This pattern of findings suggests that self-complexity is not an overarching individual difference that characterizes the self-system in all contexts (see also Salovey 1992). Rather, self-complexity may vary in different parts of the self-system, and it clearly varies according to the perspective from which the self is conceptualized.

More profitable applications of self-complexity have focused on self-complexity as an intrapsychic resource that serves as a buffer against the potential ill effects of negative life experiences. For instance, among adults (but not children; Jordan and Cole 1996) who recently experienced a high level of stressful life events (e.g., moved to a new residence, experienced death of a relative), those high in self-complexity evinced fewer depressive symptoms and fewer stress-related physical symptoms than their low self-complexity counterparts (Linville 1987).[1] People relatively high in self-complexity are more likely to avoid self-focused attention following a failure than people relatively low in self-complexity, indirect evidence that the effects of failure are less profound when the self is relatively complex (Dixon and Baumeister 1991). Among people who have recently experienced a relationship breakup, those with high self-complexity experience less distress in general, and those for whom the relationship self-aspect is relatively minor in the self-system and independent of other self-aspects experience fewer coping problems than their low self-complexity counterparts (Smith and Cohen 1993).

An alternative conceptualization of the degree to which self-aspects are interrelated is self-concept differentiation (Donahue, Robins, Roberts, and John 1993). Self-concept differentiation is defined as "individual differences in the tendency to see oneself as having different personality characteristics across one's roles" (Donahue et al. 1993, 834). One difference between self-concept differentiation and self-complexity is that the former focuses specifically on social roles, whereas the latter takes a broader view of the content that might characterize different self-aspects. Like self-complexity, self-concept differentiation is measured through a set of mathematical operations on an Adjective x Self-Aspect matrix of self-ratings (e.g., Table 4.1).

As was true of self-complexity, the computation of self-concept differentiation is rather involved. The computation is accomplished through the use of the statistical technique of factor analysis (Block 1961). To illustrate, consider the following circumstance: A woman rates herself on ten trait adjectives with reference to three social roles-student, daughter, and worker. Given those data, it is possible to derive three numbers, correlation coefficients, that reflect the degree of association between the three roles as rated by her on the trait adjectives. Using factor analysis, it is then possible to determine to what extent the associations among the roles might be caused by a single underlying source-an integrated identity. To the extent that the correlations cannot be explained by a single underlying source, the self-concept is differentiated.

This computational strategy is illustrated in Figure 4.1. In the figure are two Venn diagrams comprising three overlapping circles. Each circle represents one of the three roles named earlier, and the area within the circle represents variability in the use of the ten adjectives to describe that role. The shaded area in each diagram denotes area of overlap of all three circles. In the Venn diagram to the left in the figure, the degree of overlap is high and self-concept differentiation, because it is reflected in the remaining area within the diagram, is low. In contrast, in the diagram to the right in the figure the degree of overlap is relatively low and, therefore, self-concept differentiation is comparatively high.

Donahue et al. (1993) posited that self-concept differentiation could be either an asset or a liability in terms of adjustment and coping. To the extent that differentiation reflects specialization, then greater differentiation should function like greater self-complexity to enable people "to respond flexibly and adaptively to different role requirements, which should improve interpersonal relationships and functioning within roles" (p. 835). On the other hand, greater differentiation might correspond to greater fragmentation of the self, "a lack of psychological integration stemming from unresolved intrapsychic conflicts" (p. 835). In two studies, Donahue et al. (1993) evaluated these competing predictions about self-concept differentiation. In an initial study involving college students, they found support for the fragmentation prediction. Greater self-concept differentiation was associated with lower self-esteem, more depressive symptoms, and lower satisfaction with role performance.

In a second study, these researchers found that self-concept differentiation of women at age fifty-two was positively correlated with anxiety and emotional instability and negatively correlated with emotional adjustment and well-being some thirty years earlier. In other words, poor adjustment appears to give rise to greater differentiation, or fragmentation, of the self-concept. Based on these findings, Donahue et al. (1993) concluded that differentiation of the self-concept, at least with regard to self-aspects grounded in social roles, reflects fragmentation and a lack of integration among aspects of the self-system.

This conclusion would appear to conflict with the predictions and findings involving self-complexity. Whereas self-complexity is portrayed as an asset, self-concept differentiation appears to be a liability. One obvious explanation for the disparity between the predictions and findings related to these two constructs concerns the manner by which they are measured. Both make use of sophisticated mathematical manipulations of data from a Self-Aspect x Adjective matrix (see Table 4. 1), but they extract different information from that matrix. At present, the precise substantive meaning of the numbers produced by the mathematical manipulations is not clear, making it difficult to reconcile the apparently contradictory findings involving self-complexity and self-concept differentiation. Until a carefully considered interpretation of what those numbers mean is offered, it will remain unclear whether a complex or differentiated self-concept is an asset or a liability.

Evaluative Self-Knowledge

The two potentially self-descriptive adjectives competent and crotchety differ in at least two ways: (1) Competent is relevant to the achievement or ability domain, whereas crotchety is relevant to the social domain, and (2) the evaluative connotation of competent is positive, whereas crotchety is decidedly negative. Approaches to the organization of self-knowledge such as self-complexity and self-concept differentiation emphasize the substantive meaning and relevance of self-descriptive information. Other approaches focus on the evaluative meaning of self-descriptive information, capitalizing on the fact that most, if not all, self-knowledge has implications for self-evaluation (Greenwald, Bellezza, and Banaji 1988; cf. McGuire and Paclawer-Singer 1976). The simple distinction between positive and negative self-knowledge provides the basis for a number of intriguing models of how self-knowledge is organized.

It is first worth noting that positive and negative self-knowledge are not simply two sides of the same coin. Indeed, the presence or absence of negative self-knowledge is more consequential in a variety of self-referent phenomena than the presence or absence of positive self-knowledge (Schwarz 1986). For instance, the disparity between one's actual and undesired (i.e., negative) selves is substantially better as a predictor of life satisfaction than is the disparity between actual and ideal (i.e., positive) selves (Ogilvie 1987). As the disparity between the actual and undesired selves increases, life satisfaction increases. In an attempt to explain this and other asymmetric patterns involving positive and negative self-knowledge, Malle and Horowitz (1995) proposed that negative self-knowledge is more tightly organized in memory and, therefore, more consequential for information processing than is positive self-knowledge. Results from a series of studies supported their proposal, demonstrating that negative self-knowledge is more strongly interconnected and more consistent than positive self-knowledge.

The distinction between positive and negative self-knowledge is sufficiently clear that the organization of the two types of evaluative self-knowledge can be examined separately. Indeed, it might be the case that self-complexity or self-concept differentiation has different implications for the two categories of self-knowledge. This possibility motivated Morgan and Janoff-Bulman (1994) to evaluate the separate contributions of positive and negative self-complexity to coping with traumatic life events. Instead of computing a single self-complexity score based on participants' sorts of all available adjectives, they had participants sort the positive and negative adjectives separately. Thus they were able to compute self-complexity scores for positive and negative adjectives as well as the full set of adjectives. Their findings supported the distinction between positive and negative self-complexity, which are only modestly intercorrelated (see also Woolfolk et al. 1995). Among participants who had experienced a traumatic life event, positive and negative self-complexity contributed independently to the prediction of coping and adjustment; higher positive self-complexity was an asset, whereas higher negative self-complexity was a liability. Among participants who had not experienced a traumatic life event, only negative self-complexity was relevant to adjustment; greater negative self-complexity was associated with poorer adjustment.

Other research focused on the distinction between positive and negative self-knowledge has pointed to negative self-complexity as more influential in adjustment than positive or overall self-complexity. In a series of correlational studies, Woolfolk et al. (1995) found that negative self-complexity, but not positive self-complexity, predicted self-esteem, depression, and negative emotions in general. They also found that both positive and negative self-complexity scores were influenced by the composition of the adjective list (e.g., proportion of positive or negative adjectives) that was provided to respondents. However, consistent with other research on positive and negative self-knowledge (e.g., Malle and Horowitz 1995; Segal et al. 1988; Segal and Vella 1990), they found that negative self-complexity performed with greater stability and predictability across studies.

Although the distinction between positive and negative self-complexity has proven useful, the strategy of treating positive and negative self-knowledge separately precludes consideration of whether the nature of the association between them affects thought and behavior. A model of self-concept organization better suited for such analyses is built around the notion of evaluative integration (Showers 1992a, 1992b). Evaluative integration concerns the degree to which positive and negative self-knowledge co-occur within self-aspects. If a particular self-aspect, say "student," is characterized by positive information only, whereas another self-aspect, say "athlete," is characterized by negative information only, then evaluative self-knowledge is compartmentalized across those aspects. On the other hand, if those two aspects of the self-system are characterized by a mixture of positive and negative information, then evaluative information is integrated across them. These two patterns of organization are depicted in Table 4.2. Note that the same evaluative information (i.e., 10 trait adjectives) is construed in a very different way in the two halves of the table. The implications of these two organizational patterns are important for understanding how valenced feedback is processed. Showers (1992a) reasoned that people who have a relatively positive view of themselves benefit from a compartmentalized organization because their generally positive view of themselves would not routinely be qualified by negative self-knowledge. People who have a relatively negative view of themselves, on the other hand, benefit from an integrative organization because such an organization would allow for tempering of their negative self-views by countervailing positive self-knowledge. Initial studies supported this reasoning, although the support for the argument that evaluative integration benefits people with a generally negative self-view was weaker and less reliable than the support for the advantages of a compartmentalized organization for people who generally view themselves positively.

Evaluative integration has been measured in a variety of ways, and the various approaches have provided complementary insights into the implications of various organizations of evaluative self-knowledge for self-referent information processing. In the initial research by Showers (1992a), research participants completed the card sort described earlier. Then the group of cards assigned to each self-aspect was subdivided into positive and negative groups. A tally of the number of cards in each subgroup within self-aspect provides the numbers necessary to determine the strength of the association between valence (i.e., positive-negative) and self-aspects. A stronger association indicates greater compartmentalization.

In a more fine-grained analysis of evaluative integration, Showers (1992b) asked participants to list words that describe them in social situations and to designate them as positive or negative. Evaluative integration was measured by indexing the frequency with which positive descriptors were followed by negative descriptors and negative descriptors were followed by positive ones. This particular pattern of integration, alternating valences, was associated with high self-esteem and, for certain self-aspects, a lack of negative emotions. Oyserman and Markus (1990) used a similar logic to investigate the effects of possible selves on delinquent behavior.  They operationalized balance in possible selves as correspondence between feared and expected selves. For instance, in a balanced organization the feared self "homeless" might be accompanied by the expected self "employed." Balance and alternating valences represent a particular form of evaluative integration that provides a cognitive context within which the potentially deleterious effects of failure can be avoided through memories of prior or anticipated successes (Taylor and Brown 1988). A final approach to operational zing evaluative integration involves indexing the variability in positivity ratings assigned to a series of self -descriptive words or phrases. Research using this approach has demonstrated that evaluative integration is an asset during life transitions for people who generally feel good about themselves (Showers and Ryff 1996).

In summary, for most people under most conditions greater compartmentalization of positive and negative self-knowledge is associated with better adjustment (e.g., Showers 1992a); however, as research on evaluative integration broadens into more diverse populations and situations and focuses on specific patterns of evaluative integration (e.g., Showers 1992b), it is becoming clear that the effects of evaluative integration are not so straightforward. For instance, for some people, high evaluative integration is associated with greater stability of self-esteem (Rhodewalt et al. 1998). People negotiating a life transition benefit from greater integration of positive and negative self-perceptions of changes brought about by the transition (Showers and Ryff 1996). A compartmentalized organization contributes to quicker recovery from a sad mood under some conditions; however, when attention is focused directly on negative self-aspects, an integrated organization is more beneficial to recovery (Showers and Kling 1996).

Integrating the Findings on Organization of Self-Knowledge

The new avenues of inquiry about the self made possible by second-generation approaches to thinking about and assessing the self-concept are exciting and promising; however, the early returns on research in this tradition raise almost as many questions as they answer. For instance, the answer to the basic question of whether a complex or a differentiated self-system is an asset or a liability remains unanswered: "It depends." More recent research in this tradition has begun to consider what conditions produce which outcomes. Although the mathematical and statistical indexes that currently dominate research in this arena are compelling, as noted earlier, there remains considerable mystery regarding exactly what psychological characteristics they represent. Perhaps they capture only a portion of the richness of variability in how self-knowledge is organized.  For instance, mathematical indexes such as self-complexity and self-concept differentiation may miss the integrative influence of higher-order self-aspects or self-narratives that provide an idiosyncratic, integrative context for seemingly disparate self-views (Harter et al. 1997). For example, the seemingly disparate views that one is both passive and assertive might be reconciled by holding to a higher-order view of oneself as perceptive regarding which type of behavior, passive or assertive, is appropriate in a given situation (Klein and Loftus 1993). As measures of self-concept organization become more sophisticated and better informed by research findings, the nature of the associations among the various indexes will become clearer and, as a consequence, the role of organization in psychological functioning and behavior can be stated more definitively.

Experience of Self-Knowledge

In addition to questions about the organization of self-knowledge, contemporary research on the self-concept has begun to focus on people's experience of self-knowledge. It might seem odd to ask how well people know themselves; however, recent research has clearly shown that not all people know themselves equally well. Moreover, the degree to which people know themselves clearly is tied to their self-esteem. Two fruitful programs of research on the experience of self-knowledge have provided keen insights into the association between self-concept and self-esteem. We now discuss each in turn.

Self-Concept Clarity

In an incisive analysis of the voluminous research literature on the reactions of high- and low-self-esteem people to evaluative feedback, Campbell (1990) proposed that people vary in the clarity with which they understand themselves. Campbell labeled this individual difference self-concept clarity, which is defined as "the extent to which the contents of an individual's self-concept (e.g., perceived personal attributes) are clearly and confidently defined, internally consistent, and temporally stable" (Campbell et al. 1996, 141). According to Campbell (1990), self-concept clarity is "an important concomitant of self-esteem" (p. 539) that helps to explain why low self-esteem people are more generally affected by social cues that implicate the self than their high-self-esteem counterparts (see also Brockner 1984)

One means of measuring self-concept clarity is using the Self-Concept Clarity Scale (Campbell et al. 1996), shown in Figure 4.2. Items such as item 7 tap the confidence aspect of self-concept clarity; the scale also includes items that target internal consistency (e.g., item 1) and temporal stability (e.g., item 8) of self-concept. Scores on the Self-Concept Clarity Scale are positively correlated with self-esteem, positive affect, and extraversion, and negatively correlated with depression, anxiety, negative affect, and self-reflectiveness (Campbell et al. 1996). Thus, as predicted, people's self-reports of the clarity of their self-concept are associated with self-reports of self-referent affect and cognition-greater clarity is accompanied by more positive self-referent affect and cognition.

Self-concept clarity also can be measured by indexing certain characteristics of people's self-ratings. For instance, the confidence dimension of self-concept clarity has been assessed as the degree of extremity of self-ratings on bipolar adjective scales. For instance, ratings near the middle of a seven-point scale anchored on one end by competitive and the other by cooperative could be interpreted as lack of clarity regarding one's standing on that dimension. On the other hand, ratings near either extreme could be interpreted as evidence of greater clarity with regard to one's standing on that dimension. Consistent with this prediction, Campbell (1990) found that extremity of self-ratings (with regard to either pole of a bipolar scale) was positively correlated with self-esteem. Similarly, the temporal aspect of self-concept clarity can be indexed using statistical characteristics of respondents' dents' self-ratings. The similarity of respondents' self-ratings over time, calculated in a variety of ways, is positively correlated with self-esteem (Campbell 1990; Campbell et al. 1996). Further corroborating the theoretical definition of self-concept clarity, the internal consistency of people's self-ratings-the degree to which people rate themselves similarly on different items that assess the same attribute-is positively correlated with self-esteem and correlated as predicted with a variety of measures of self-referent thoughts and feelings (Campbell 1990; Campbell et al. 1996). Together with the findings involving the Self-Concept Clarity Scale, these findings provide strong and consistent evidence of the link between one's experience of self-knowledge and the regulation of self-knowledge in the course of social life.

Self-Certainty

A second variable that reflects people's experience of their own self-knowledge is self-certainty (Baumgardner 1990). Self-certainty reflects the degree to which a person is confident about his or her self-views across a variety of domains. For instance, Person A and Person B both might describe themselves as "slightly above average" in intelligence; however, Person A might concede that sometimes he thinks he is actually a bit below average and other times he thinks he is well above average, whereas Person B might assert that she is slightly above average, plain and simple. In this example, although the two people see themselves as equally intelligent, Person B is more certain of that self-view than Person A.

Self-certainty is measured using the Latitude of Self -Description Questionnaire (LSDQ; Baumgardner 1990). Like many measures of self-concept, the LSDQ asks respondents to describe themselves on a series of trait adjectives (e.g., intelligent, persistent, likable); however, the LSDQ differs from traditional adjectival measures in important ways. First, the LSDQ is a decidedly social measure because it asks respondents to rate themselves with reference to their peers. Second, respondents not only estimate their standing on the trait adjectives but they also estimate the highest and lowest levels on the trait adjectives that might describe them. These latter estimates form the basis of the measure of self-certainty.

Two response patterns to an example item from the LSDQ appear in Figure 4.3. Completion of the measure requires respondents to consider their relative standing on each adjective in three ways. First, the respondent is asked to "decide if you think you have more than average, average, or less than average of the particular trait. Then mark on the scale with an X where on that continuum you see yourself" (Baumgardner 1990, 1071). In the two response patterns shown in the figure, both respondents rated themselves at the sixty-fifth percentile (i.e., as high or higher than 65 percent of the population) on the trait humorous. The second phase of responding to items of the LSDQ produces the measure of self-certainty. The respondent is asked to decide where you see your range on that trait. You probably found yourself a bit unsure of where to place the X in the first exercise. This is because we usually view ourselves as somewhat flexible on almost all traits (although some more than others). What you need to do is simply decide where that range is and mark the two endpoints with arrows. . . . When done, you should have two arrows, marking the endpoints of where you might fall, and one X, marking your best guess of where you do fall. (Baumgardner 1990, 1071)

Looking back to the figure, the first respondent sees himself as possibly falling as low as the sixtieth percentile or as high as the seventieth percentile on humor, whereas the second respondent can see herself falling between the fiftieth and eightieth percentile. The second respondent's latitude of self -description on the trait humorous is wider (30 percentage points) than the first respondent's (10 percentage points), indicating less certainty regarding her standing on the trait humorous. A total self-certainty score is created by averaging latitude widths across the twenty traits included in the scale; lower scores indicate greater self-certainty.

Research on self-certainty using the LSDQ (Baumgardner 1990) underscores the importance of considering people's experience of their own self-knowledge in research on self-concept and self-esteem. Self-certainty is positively correlated with global self-esteem. The more certain people are in expressing their self-concept, the higher their self-esteem. This association appears to hold regardless of whether the trait adjectives on which self-ratings are made are positive or negative, or concern social or intellectual traits. In summary, one way of understanding the link between self-concept and self-esteem is through a consideration of self-certainty. Greater certainty about one's self-concept contributes to higher self-esteem, and uncertainty about one's self-concept goes hand in hand with low self-esteem.

Self-Concept in the Social Arena

Self-concept is a rich and intriguing aspect of selfhood that rightfully has received a considerable amount of attention from social psychologists. A substantial and growing body of research findings provides an increasingly clear and detailed view of how the self-concept varies between people and across situations and how it functions in the regulation of thoughts, emotions, and behavior. As our understanding of self-concept has increased, its prominence in people's experience of themselves and the world around them has become more apparent. This prominence is aptly described by DeSteno and Salovey (1997):

Of all the concepts that reside in memory, it can be argued that the self[-concept] reigns supreme with respect to individuals' daily functioning; without a coherent self-concept, we could not identify ourselves or, more important, even differentiate ourselves from other objects, animate or otherwise, in awareness. But the integrality of the self-concept extends well beyond this basic perceptual function. It is a representation of all that one is, and consequently, is intimately tied to any psychological process, automatic or volitional, that involves self-knowledge. (p. 389)