Campbell Foddis

Since the 1960s, self-esteem has been a central concept in clinical psychology. It is a strongly normative notion; clinicians usually regard high self-esteem, rather than moderate or low, as most indicative of healthy psychological functioning. Indeed, low self-esteem has often been connected with a wide array of difficulties and failures in life: with frank psychopathology, even with crime.

Conceptions of self-esteem have been somewhat variable, but in all of them it is some kind of feeling or judgment or evaluation that one makes of oneself; in other words, self-esteem necessarily depends on reflective consciousness. We do not expect squirrels to have high or low self-esteem; we do not expect newborn human babies to have self-esteem either. On these grounds alone, interactive conceptions of the self and its development ought to have something to say about self-esteem and whether it is being adequately understood. Intriguingly, however, some recent controversies about the nature and implications of high self-esteem do not seem resolvable without positing implicit self-esteem—evaluations of oneself that are not conscious. And implicit self-esteem poses a number of challenges from an interactivist standpoint.

We will take Nathaniel Branden’s model of self-esteem, a well worked out conception that has been quite influential historically, as our starting point. Branden’s model (1969, 1983, 1994) includes two dimensions: (1) self-efficacy, feeling that one is competent to deal with life’s basic challenges, and (2) self-worth, experiencing oneself as worthy of happiness. Self-worth encompasses a sense of deserving of success, love, and friendships, in addition to accepting positive feelings, such as pride and joy, as natural and proper to one’s life. Branden regards the dimensions as being related in a general way: acting with competence when faced with a challenge leads to increased feelings of self-worth.

Branden’s model does not characterize self-esteem in terms of superiority or inferiority. If high self-esteem consisted of feeling superior to others, one’s self-esteem would be continually dependent on social comparisons—and continually subject to challenge. More broadly, Branden does not regard self-esteem as dependent on “external” sources, such as looks or body build or social status, since these are not controllable by self-directed thought. Nor is his conception of high self-esteem perfectionistic, since self-directed thought obviously brings no guarantee of error-free performance. According to Branden, secure and enduring feelings of competence, and the resultant feelings of worth, result from consistently setting certain standards for oneself and acting accordingly.

The standards for adult self-esteem include self-reflective and independent thought; taking responsibility for one’s thoughts, beliefs, values, and actions, and authentically asserting them; pursuing meaningful life goals; and cultivating eudaimonistic moral values such as honesty and productiveness. If we act in ways that meet these standards, our self-esteem will tend to rise; if we fail to act in these ways, or betray our standards, our self-esteem will tend to drop. What contributes to self-esteem is internal to the person; it depends on self-directed psychological processes that are under each person’s control.

Along similar lines Bednar and Peterson (1995) have proposed that enduring feelings of self-esteem depend primarily on how a person responds to feelings of anxiety. For self-esteem to improve, anxiety needs to be confronted with acceptance and realism. Successfully engaging anxiety requires the courage to cope directly with the unwanted thoughts and feelings that precipitate it, such as pain, embarrassment, shame, and fear. Realism mean recognizing oneself as responsible for one’s thoughts, feelings, and behavior, accepting the beliefs and feelings motivating the anxiety, and being able to disclose one’s thoughts and feelings to others in an appropriate context. These three self-directed processes correspond to self-responsibility, self-acceptance, and self-assertiveness in Branden’s model.

But other psychologists take very different perspectives, according to which high self-esteem isn’t such a good thing after all. Roy Baumeister, a prominent social psychologist, defines high self-esteem as belief that one is superior. He maintains that D students, gang leaders, racists, murderers, and rapists have high self-esteem. In a widely cited review of research, Baumeister, Smart, and Boden (1996) concluded that aggression is associated with high self-esteem. Baumeister (1997) argued that groups known to have high self-esteem are also more aggressive than other groups; for instance, non-depressed individuals are more aggressive than depressed individuals. He also contended that because alcohol “enhances” self-esteem, aggressive individuals who have consumed alcohol provide further evidence linking high self-esteem and aggression. Examining empirical studies in which murderers and rapists respond to self-defining statements, Baumeister et al. (1996) pointed out that these violent criminals consciously believe in their superiority.

To complicate matters, high self-esteem looks like an equivocal notion in Baumeister’s model. He does not claim that high self-esteem necessarily leads to aggression; for aggression to be the outcome, highly favorable self-appraisals have to combined with an ego threat (a successful challenge to one’s high self-appraisal). Bushman and Baumeister (1998) tested this hypothesis experimentally. Participants were given the Narcissistic Personality Inventory (Raskin & Terry, 1988), which contains such items as “If I ruled the world it would be a much nicer place,” and the Rosenberg (1965) Self-Esteem Scale. The ego threat consisted of a participant’s intellectual competence being strongly criticized. Participants were given the opportunity to aggress against the people who had criticized them, by delivering a blast of noxious noise (this being a social psychology experiment, the noise was not really delivered to the critic). In the end, the narcissism measure, not the self-esteem score, predicted the strength of the aggressive response (the intensity and duration of the noise) toward the source of the criticism. However, scores on self-esteem scales are usually correlated with scores on narcissism scales (as highly as .6). Consequently it looks as though some people with high self-esteem are aggressive when their sense of self is threatened, while others with high self-esteem are better insulated from ego threats.

Along similar lines, Nicholas Emler (2001) has reviewed a wide range of published research, concluding that low self-esteem (as measured on the Rosenberg and similar questionnaires) is not a risk factor for delinquency, violence against others, or racial prejudice. On the contrary, he suggests that high self-esteem is the more plausible risk factor. Besides the kinds of evidence already reviewed by Baumeister, other research allows Emler to conclude that people with high self-esteem are more likely to engage in risky pursuits, such as driving too fast and driving drunk. Emler also finds little evidence that self-esteem and educational attainment are associated; even failing students can show high self-esteem on questionnaires. Despite these indictments of high self-esteem, Emler does acknowledge evidence that some problems are related to low self-esteem, as measured on questionnaires. The lower the self-esteem score, the greater the risk of suicide, suicide attempts, depression, teenage pregnancy, and victimization by bullies.

Baumeister ends up having to bifurcate high self-esteem—between a stable, positive version that does not lead to aggression, and an unstable, easily threatened version that does. Existing models are able to make this distinction in a less ad hoc fashion, but only by distinguishing between explicit and implicit self-esteem.

In fact, Branden’s original model (1969) sharply differentiates between genuinely high self-esteem and what he calls pseudo-self-esteem. Pseudo-self-esteem relies on “external” sources, such as being admired or approved by others, wealth, or social status. People tend to put their reliance on such external sources to the extent that they are lacking in self-acceptance, self-responsibility, and the other internal sources. But because they are not under a person’s direct control, Branden argues that external sources cannot realistically enhance his or her feelings of competence. Self-esteem that depends on them is insecure and will be vulnerable to threats. Branden’s notion of pseudo-self-esteem, then, can help explain how some people with high measured self-esteem can respond to an ego threat with aggression, as in Bushman and Baumeister’s (1998) experiment. Those with pseudo-self-esteem, whose feelings of competence and worthiness are “on the line” when challenged, are more prone to defend against this threat, for instance by lashing back at the source of criticism.

What Branden refers to as pseudo-self-esteem shows up in a number of other accounts as “defensive” self-esteem. Christopher Mruk’s (1999) theory incorporates both the competence and worthiness components. According to Mruk, there are two types of defensive self-esteem. The “workaholic” or perfectionist type has high self-competence, yet low self-worth. A perfectionist attempts to derive most, if not all, of his worth from work, but not matter how well he succeeds, he never truly feels worthy. The second type exhibits low self-competence, yet high self-worth, which Mruk proposes is characteristic of narcissism. The narcissist may regard himself as a person of worth, and be seen that way by others, but feelings of inadequacy lie beneath this veneer of high self-regard.

As theoretical accounts, these make a good deal of sense, but how can the difference between genuine and defensive self-esteem be tested empirically? Because investigators have relied so heavily on self-report questionnaires for assessing self-esteem, distinguishing real from pseudo is going to be problematic.

Virtually all self-esteem research has relied on answers to a handful of self-report questionnaires. The one most relied on (Emler calls it the “gold standard”) was developed in 1965 by sociologist Morris Rosenberg. Here is the entire Rosenberg Self-Esteem Scale:

1. On the whole I am satisfied with myself.
2. At times I think I am no good at all.
3. I feel that I have a number of good qualities.
4. I am able to do things as well as most other people.
5. I feel I do not have much to be proud of.
6. I certainly feel useless at times.
7. I feel that I am person of worth, at least the equal of others.
8. I wish I could have more respect for myself.
9. All in all, I am inclined to believe that I am a failure
10. I take a positive attitude toward myself.

We take no issue with the reliance on introspection, whether on a short questionnaire that attempts to measure self-esteem on a single dimension or on a substantially longer one that breaks self-esteem down into multiple dimensions. But we need to ask a couple of questions. Could people be motivated to give false or misleading answers to self-esteem items? More deeply, could aspects of their self-esteem be implicit rather than explicit, and therefore not accurately reportable on such surveys?

On the first issue, Branden maintained in his original formulation (1969) that feelings of low self-esteem are so hard to bear that people who distrust themselves or feel that they are of little worth have an incentive to put up psychological defenses against these feelings. To stave off judgments of incompetence or worthlessness, some may fabricate grandiose self-images and seek to assure themselves that they are superior to others. Consequently a person who securely feels competent and worthy would give a strong endorsement to an item like “On the whole I am satisfied with myself.” But so might a person who is compensating for feelings of incompetence or lack of worth.

What’s more, the growth of the self-esteem movement has helped to make self-esteem highly socially desirable in the United States and several other countries. Studies that use samples of American college students frequently show average scores of 70 on the Rosenberg Self-Esteem Scale (the highest possible is 90, when answers are given on a 9-point scale). We are pretty sure that teachers, parents, and peers would not rate all of these students nearly so high on self-esteem—neither, it is safe to say, would most clinicians. (To further complicate matters, high self-esteem would be socially undesirable in some other parts of the world, such as East Asia
[Markus and Kitayama, 1991]).

And of course the standard questionnaires afford no way of checking self-reported evaluations are realistic. When students are asked how good they believe they are at mathematics (as some of the more specialized self-esteem questionnaires do) they aren’t normally given a math test to validate their claims. Other areas of life pose greater challenges: it’s easy to ask subjects whether they are confident in their ability to maintain mutually satisfying sexual relationships, much harder to know what their relationships are really like. According to a theory like Branden’s, Bednar and Peterson’s, or Mruk’s, your favorable evaluation of your ability to tackle math problems indicates high self-esteem only if you really are good at math.

The problem of realism can easily be formulated in terms of a difference between explicit and implicit self-esteem. When we respond truthfully, a self-report survey measures our conscious feelings and evaluations about ourselves. But it could be that while consciously evaluating ourselves in positive ways, we are setting goals or making choices that imply something very different about us. Consider a woman who is consciously convinced that she deserves an emotionally rewarding relationship with a man, but keeps being attracted to men who treat her coldly and manipulatively. Her implicit self-esteem seems to be low (at least in this domain) and consequently at odds with her high explicit self-esteem; it would not be a good idea to rate her self-esteem as high overall.

Conflicts between implicit and explicit self-esteem play an essential role in a promising new theory of narcissism. Carolyn Morf and Frederick Rhodewalt (2001) suggest that narcissism is an overdependence on social sources to affirm a grandiose sense of self. The narcissist genuinely needs other people, on this model, but only for their instrumental value in bolstering his sense of self. Because narcissists have little empathy for other people and little genuine concern for they really think, they end up engaging in paradoxical, counterproductive behavior; the narcissist’s charm tends to wear off over time, alienating those who were initially drawn in to give their admiration. Under the narcissist’s grandiose exterior is a vulnerable sense of self that is easily threatened and must be constantly supplied with affirmation.

Morf and Rhodewalt review empirical studies that show how narcissists undergo instability and fluctuations in their explicit self-esteem because their self-image is both grandiose and vulnerable. For instance, in studies contrasting participants who scored high and low on the Narcissistic Personality Inventory, those with high scores displayed greater day-to-day fluctuations in self-esteem than less narcissistic individuals did. The more narcissistic individuals’ daily self-esteem was also more highly correlated with positive or negative social interactions: it depended on whether the narcissist received acceptance and whether the interactions made the narcissist feel “like himself.” If Morf and Rhodewalt’s (2001) understanding is correct, a narcissist can be understood as a person with defensive self-esteem, who relies on external sources of self-esteem to compensate for a deficiency in internal resources.

Other empirical findings reviewed by Morf and Rhodewalt indicate that narcissists view their abilities and accomplishments as superior, make self-aggrandizing attributions (i.e., brag) to elicit positive reactions, restructure their past so to make themselves look better, and derogate anyone who gives them negative feedback. Narcissists keep looking for ways to validate themselves, but none of the external validation is ever enough to convince them of their own adequacy. Thus the Morf-Rhodewalt model of narcissism ties in neatly with Branden’s (1969) argument that those who do not have genuine high self-esteem strive to “fake” it. Narcissism, in turn, is what Baumeister and Emler seem to be talking about when they attribute adverse consequences to high self-esteem. As we noted, Bushman and Baumeister’s (1998) findings directly implicate narcissism in aggressive reactions to ego threats.

Christina Salmivalli and her colleagues (1999) asked adolescents to rate their own self-esteem, while also collecting ratings from their peers. Salmivalli et al. also sought to assess “defensive egotism” (a notion drawn from Baumeister’s theory) by asking peers three questions: “Does your classmate constantly need to be the center of attention?” “Does he think too highly of himself?” “Can he face criticism?” Those who scored high on self and peer-rated self-esteem, and high on defensive egotism, were more apt to be bullies. Those who rated high on self and peer-rated self-esteem, but low on defensive egotism, were more likely to defend victims against bullies. Those who scored low on all three measures most likely to be victimized. Salmivalli’s findings suggest that individuals whose self-esteem is defensive are more prone to aggression. They also indicate that measuring explicit self-esteem (or even asking peers to rate another person’s self-esteem) is not enough.

Sorting out genuinely high self-esteem from narcissism requires ways of measuring implicit self-esteem. Getting more precise about what these procedures are trying to measure would seem, in turn, to be a fertile domain of application for interactivist theories.
One proposed method of measuring implicit self-esteem works by recording reaction times to computer-based word associations, which is the basis of Anthony Greenwald’s Implicit Association Test (Greenwald & Banaji, 1995). It is assumed that participants are not aware that their self-esteem is being assessed while they are taking this test. At present there is limited but growing evidence for the utility of the IAT.

A study by Foddis, Vander Veen, Silverthorn, and Reddon (2002) used a different implicit measure based on sentence completion. The sentence stems ask participants for their perceived sources of self-esteem. Participants spontaneously generate statements regarding what their self-esteem depends on. The completions, retrieved quickly from memory and subjected to little or no conscious review, are the “implicit” material here. Unlike the Implicit Association Test, the sentence completion measure assumes conscious awareness of underlying processes, namely, knowing what events or practices enhance self-esteem. But requiring participants to say what their self-esteem depends on means that they are not being given ready-made socially desirable statements to evaluate (e.g., “My self-esteem depends on taking responsibility for my own behavior”). Participants’ sentence endings are scored based on how closely they correspond to Branden’s sources of self-esteem.

Foddis et al. contrasted those who scored high on explicit self-esteem (measured by self-report) and high on implicit self-esteem (measured by sentence completions) with those who showed high explicit self-esteem but low implicit self-esteem. Those with high explicit but low implicit self-esteem responded more poorly to criticism and endorsed fewer statements that they were worthy of being told they were loved. Even though they reported high self-esteem, in these regards participants who fell into this category resembled those who scored low in explicit self-esteem. Such data support the notion of pseudo-self-esteem. Among those with high explicit but low implicit self-esteem, the more extreme a person was in generating external sources of self-esteem in sentence completions, the more likely he would respond with defensiveness and hostility when criticized.

Implicit self-esteem poses challenges at the theoretical level as well as at the measurement level. The rationale for using the Implicit Association Test to measure implicit self-esteem is rather different from the rationale for using sentence completions about the sources of self-esteem; we do not know yet whether the two procedures measure the same thing. Narcissism, according to the Morf-Rhodewalt model, is partly constituted by an ongoing clash between implicit and explicit self-esteem, but the Narcissistic Personality Inventory is a self-report questionnaire. What’s more, not all of the items on the NPI seem to indicate pathology whenever someone endorses them. On the pathological side are such statements as “I have a natural talent for influencing people”; “I can usually talk my way out of anything”; “I insist on getting the respect that is due me”; “If I ruled the world it would be a much nicer place”; or “Superiority is something you are born with.” Yet people with a solidly realistic grasp of their abilities might truthfully endorse other NPI items, like “I see myself as a good leader”; “I would be willing to describe myself as a strong personality”; or “I am assertive.” Even if the NPI had no flaws as a measurement device, the implicit self-esteem issues would have to be inferred from people’s answers to it.

Interactivist theory (Bickhard, 1980, 1998; Campbell & Bickhard, 1986) ought to have something to say about self-esteem, particularly the aspects of it that are called “implicit” in the literature. More specifically, any theory of self-esteem ought to be situatable within the interactivist account of goals, values, and the self (Campbell & Bickhard, 1986; Bickhard & Christopher, 1994; Campbell & Christopher, 1996; Campbell, 2002; Campbell, Christopher, & Bickhard, 2002).

Self-esteem will not have a simple structure, if this account is basically correct. In adults, who are typically capable of functioning at Knowing Level 3 with regard to self and identity—and may at times be working a level higher than that—explicit self-esteem will generally function as a judgment about the extent to which our actions and our patterns of thought and feeling satisfy metavalues of competence and worthiness. What’s more, such “pillars” of self-esteem (Branden, 1994) as self-acceptance or purposeful living or integrity are themselves metavalue-level considerations. A metavalue is a Level 3 goal about which values to form and pursue at Level 2, and values, in turn, are goals about which goals to form and pursue at Level 1, where transactions between ourselves and our environments are actually taking place.

While the consciously held beliefs and consciously made self-evaluations of teenagers and adults are generally going on at Level 3, one already has a self at the value level (Level 2), and complex web of goals for action at Level 1 already presupposes or implies a lot about one’s competence or worthiness (Campbell & Bickhard, 1986; Campbell, Christopher, & Bickhard, 2002). For instance, our concrete expectations about who will respond to us with words of love, and about the likely further consequences of their so responding, are all at Level 1, and any values or metavalues we form about these issues will be richly instantiated at Level 1. Similar considerations apply to our expectations about who will respond with criticism, and how that is likely to affect our chances of satisfying other Level 1 goals.
Consequently, when there are conflicts between explicit and implicit self-esteem, these can be conflicts between Level 3 metavalues and the values at Level 2 that instantiate them, or between metavalues at Level 3 and goals for interactions with the external world at Level 1. Meanwhile, conflicts between Level 2 values and the Level 1 goals that instantiate them would presumably count as conflicts within implicit self-esteem for an adult, though not (for instance) for an 8-year-old.

In addition, conflicts in our metavalues may not be adequately responded to unless we move up to Level 4 and consciously recognize them. So conflicts are possible among the metavalues that our explicit self-esteem is assessing ourselves against.
The usual assumption behind sentence completion (e.g., Branden, 1994) is that we “really know” the answers to many of the questions indirectly asked by the sentence stems—but we don’t know that we know, or perhaps we don’t want to know that we know. Generally, if we are posing Level 3 questions, we will be getting answers by reflecting on our values and our ways of managing our goals at Level 2. Some of the answers will be ones that we have arrived at before; others will be novel. In addition, the sentence-completions used by Foddis (1999; Foddis et al., 2002) pertain fairly directly to our sources of self-esteem (e.g., “I trust myself more when ___” or “One thing my self-respect depends on is ___”). By contrast, questions used in therapy (e.g., Branden, 1997) range much more widely, and some of them appear geared to encourage new Level 2 knowledge of Level 1 patterns (“One of the ways that I distance myself from people is ___” or “Sometimes I am drawn to people who ___”).

Some aspects of implicit self-esteem might be too implicit for sentence completion. The person may not be in a position to know something at Level 3, because there are implications of pursuing his or her values at Level 2 that he or she has never sought (has no Level 2 values or Level 1 goals for) and has never recognized. Similarly for knowing something at Level 2, for there may be implications of pursuing Level 1 goals that are not sought at that level.

The Implicit Association Test, on the other hand, is not supposed to be recognizable to the average person as a test of self-esteem. How much more implicit can it get, compared to a sentence-completion measure? In particular, how well can it handle self-esteem issues that are deeply implicit? (By deeply implicit, we mean mere implications of what is known or sought that are not themselves known or sought at any of the levels.)

A couple of further complications suggested by the interactivist framework are the self-referential status of the metavalues that constitute self-esteem, and the difference between self-esteem and a higher-level concern about it. The metavalues that serve as standards for self-esteem are self-referential (Campbell & Bickhard, 1986); there are many ways to instantiate self-acceptance or self-trust, but there is no set of values that stand in an instrumental relationship to self-acceptance or self-trust. For comparable reasons, it makes better sense to view a set of changes to values and goals as leading, on the one hand, to better performance in school and, on the other, to higher self-esteem, instead of expecting higher self-esteem to cause better school performance. Moreover, there is a difference between self-esteem and one’s conscious beliefs about its importance, which will tend to be considerations about Knowing Level 4 (the level of philosophical justifications for one’s metavalues). Knowing Level 4 considerations about self-esteem need not have any positive effects on self-esteem; we wonder whether under some circumstances they might have negative effects.

References

Baumeister. R. F., Smart, L., & Boden, J. M. (1996). Reaction of threatened egotism to violence and aggression: The dark side of high self-esteem. Psychological Review, 103, 5-33.

Baumeister, R. F. 1997. Evil: Inside human violence and cruelty. New York: W. H. Freeman.

Bednar, R., & Peterson, S. (1995). Self-esteem: Paradoxes and innovations in clinical theory and practice (2nd ed.). Washington DC: American Psychological Association.

Bickhard, M. H. (1980). A model of developmental and psychological processes. Genetic Psychology Monographs, 102, 61-116.

Bickhard, M. H. (1998). Levels of representationality. Journal of Experimental and Theoretical Artificial Intelligence, 10, 179-215.

Bickhard, M. H., & Christopher, J. C. (1994). The influence of early experience on personality development. New Ideas in Psychology, 12, 229-252.

Branden, N. (1969). The psychology of self-esteem. Los Angeles: Nash.

Branden, N. (1983). Honoring the self. Los Angeles: Jeremy Tarcher.

Branden, N. (1994). The six pillars of self-esteem. New York: Bantam.

Branden, N. (1997). The art of living consciously: The power of awareness to transform everyday life. New York: Simon & Schuster.

Bushman, B. J., & Baumeister, R. F. (1998). Threatened egoism, narcissism, self-esteem, and direct and displaced aggression: Does self-love or self-hate lead to violence? Journal of Personality and Social Psychology, 75, 219-229.
Campbell, R. L. (2002). Goals, values, and the implicit: Explorations in psychological ontology. Journal of Ayn Rand Studies, 3(2), 289-327.

Campbell, R. L., & Bickhard, M. H. (1986). Knowing levels and developmental stages. Basel: Karger.

Campbell, R. L., & Christopher, J. C. (1996). Moral development theory: A critique of its Kantian presuppositions. Developmental Review, 16, 1-47.

Campbell, R. L., Christopher, J. C., & Bickhard, M. H. (2002). Self and values: An interactivist foundation for moral development. Theory & Psychology, 12, 795-823.

Emler, N. 2001. Self-esteem: The costs and causes of low self-worth. York: Joseph Rowntree Foundation.

Foddis, W. F. 1999. Sentence completion and self-esteem: Initial validation of an assessment method and a theory. Bachelor of Arts Honors Thesis, University of Alberta, Edmonton.

Foddis, W. F., Vander Veen, S., Silverthorn, N., & Reddon, J. (2002). Defensive self-esteem: A form of low self-esteem. Poster presented at the annual meeting of the American Psychological Society, June 2002.

Greenwald. A., & Banaji, M. (1995). Implicit social cognition: Attitudes, self-esteem, and stereotypes. Psychological Review, 102, 4-27.

Markus, H. R., & Kitayama, S. (1991). Culture and the self: Implications for cognition, emotion, and motivation. Psychological Review, 98, 224-253.

Morf, C. C., & Rhodewalt, F. (2001). Unraveling the paradoxes of narcissism: A dynamic self-regulatory processing model. Psychological Inquiry, 12, 177-196.

Mruk, C. 1999. Self-esteem: Research, theory, and practice. New York: Springer.

Raskin, R., & Terry, H. (1988). A principal-components analysis of the Narcissistic Personality Inventory and further evidence of its construct validation. Journal of Personality and Social Psychology, 54, 890-902.

Rosenberg, M. (1965). Society and the adolescent self-image. Princeton NJ: Princeton University Press.

Salmivalli, C., Kaukianen, A., Kaistaniemi, L., & Lagerspetz, K. M. J. (1999). Self-evaluated self-esteem, peer-evaluated self-esteem, and defensive egotism as predictors of adolescents’ participation in bullying situations. Personality and Social Psychology Bulletin, 25, 1268-1278.