The Relationship Between Unstable Self-Esteem and Aggression

Abstract

This study examines whether the instability of self-esteem (i.e., a high intraindividual variability in self-esteem) is differentially associated with different types of aggressive behavior by using a sample of 235 preadolescent children. Self-esteem was measured four times for four consecutive days, and proactive and reactive aggressive behaviors were assessed by peer nominations. The latent trait-state-occasion model was used to describe the stable trait and the fluctuating state components of self-esteem. The empirical results indicated that the state component of self-esteem, not the trait component, had a significant association with aggressive behaviors. When controlling the trait component, the state component was positively related to reactive aggression and negatively related to proactive aggression. Implications for future research were discussed.

Keywords

latent trait-state-occasion model aggression unstable self-esteem instability of self-esteem

Although many studies have documented an association between self-esteem level and psychopathology, the underlying process linking self-esteem and psychological functions remains unclear (Swann, Chang-Schneider, & McClarty, 2007; Zeigler-Hill & Wallace, 2012). While a handful of studies have suggested that children with lower levels of self-esteem engage in a higher degree of aggressive behavior (e.g., Donnellan, Trzesniewski, Robins, Moffitt, & Caspi, 2005), others have produced mixed results on the role that self-esteem has in producing aggressive behaviors (Ostrowsky, 2010). Many researchers have shown that perpetrators of aggression generally take a favorable and perhaps even inflated view of themselves (e.g., Baumeister, Campbell, Krueger, & Vohs, 2003; Hughes, Cavell, & Grossman, 1997; Salmivalli, 2001), and propose that a higher level of self-esteem will most likely lead to aggressive behavior (e.g., Baumeister, Bushman, & Campbell, 2000; Jang & Thornberry, 1998). By contrast, other researchers have suggested that neither high nor low self-esteem directly cause aggression (e.g., Crocker & Park, 2004; Scheff & Fearon, 2004). These inconsistencies have led some researchers to conclude that the evidentiary basis of self-esteem research is so fundamentally flawed that the entire enterprise requires a re-examination (Marsh & Craven, 2006).

The main goal of the present study is to provide a clearer understanding of the relationship between self-esteem and aggression. The following two premises served as the basis for this study: First, the level of self-esteem is only one aspect of self-system (Crocker & Wolfe, 2001). At each level of self-esteem, individual differences exist in its instability (Kernis, 2003, 2005). The instability of one’s self-esteem is distinct from one’s level of self-esteem, explaining the psychological functions and behaviors over and above the effect of self-esteem level (e.g., Kernis, Lakey, & Heppner, 2008; Webster, Kirkpatrick, Nezlek, Smith, & Paddock, 2007). Second, although proactive and reactive aggressive behaviors have been found to be correlated with each other, these two functions of aggression are distinct (Little, Jones, Henrich, & Hawley, 2003). To the extent that they reflect some manifestation of different underlying processes, they are likely to be differentially associated with the instability of self-esteem.

Instability of Self-Esteem

Kernis (2005) claimed that unstable self-esteem turned out to be a better predictor of aggression than high self-esteem (i.e., high means of self-esteem measures). Individuals with high but unstable self-esteem tend to score higher on measures of hostility than those with low self-esteem (stable or unstable), whereas individuals with high but stable self-esteem are least likely to be hostile (Kernis, Grannemann, & Barclay, 1989). In other words, individuals with high self-esteem can be found at both ends of the spectrum for hostility. These findings suggest that when taken alone, the level of self-esteem provides an incomplete understanding of its role in psychological and interpersonal functions (Crocker & Wolfe, 2001; Kernis, 2005). Accordingly, a more complete understanding of the nature and functioning of self-esteem requires consideration of not only its level but its degree of instability.

Unstable self-esteem differs from secure self-esteem in several aspects: First, unstable self-esteem is not well anchored and requires continued validation. Any direct quest for high self-esteem reflects enhanced tendencies to be caught up in the process of defending, maintaining, and maximizing one’s positive, although tenuous, sense of self-worth. Furthermore, individuals with unstable self-esteem react strongly to events that they view as relevant to their self-esteem (Waschull & Kernis, 1996). The particular vulnerability exhibited by individuals with unstable self-esteem pertains to the way they react to self-esteem threats (Roberts & Monroe, 1992). For this reason, they exhibit a heightened reactivity and a defensiveness in response to evaluative events, and this heightened reactivity often has adverse consequences such as an increase in depressive symptoms (Franck & De Raedt, 2007) and hostility (Paradise & Kernis, 2002). Taken as a whole, individuals with unstable self-esteem are strongly reactive and defensive with respect to potential threats to their positive self-view, and frequently engage in self-promoting activities, constantly seeking to validate their worth.

There are developmental differences in the instability of self-esteem. The extent of instability decreases from adolescence to old age while the level of self-esteem increases (Meier, Orth, Denissen, & Kuhnel, 2011). However, instability of self-esteem has been widely researched in adult and college-aged populations, whereas it has rarely been examined in early adolescence (Bushman et al., 2009; Locke, 2009; Webster et al., 2007). Given the increased instability of self-esteem at younger ages, the role of unstable self-esteem needs to be explored with early adolescents.

Instability of Self-Esteem and Aggression

The main purpose of the present study is to examine the relationship of unstable self-esteem to two functions of aggressive behaviors. Although related, proactive and reactive aggressions are distinct subtypes of aggressive behaviors. These subtypes can be best explained by different etiological mechanisms and are differentially related to behavioral outcomes. Proactive aggression represents goal-oriented and calculated aggressive behaviors motivated by some external reward (Dodge, 1991). Social learning theory posits that aggression serves to help in obtaining a desired goal (Bandura, 1973) and may provide a theoretical mechanism underlying the proactive functions of aggression. By contrast, reactive aggression represents aggressive behaviors in response to those behaviors perceived as threatening or intentional. This may be explained by the frustration-aggression model, which posits that aggression results as an angry and hostile reaction to frustration (Berkowitz, 1978). The present study explores whether the instability of self-esteem is associated with the adoption of different subtypes of aggressive behaviors.

The central aspect of the instability of self-esteem is the individual’s hypersensitivity and heightened reactivity to self-esteem threats (Greenier et al., 1999), which can lead to increased defensiveness. This may manifest itself in frequent angry and hostile outbursts often aimed at restoring damaged self-feelings. Individuals with highly unstable self-esteem are sensitive to ego threats, dealing with such threats by becoming excessively angry and hostile (Paradise & Kernis, 2002). Consistent with this notion, Bukowski, Schwartzman, Santo, Bagwell, and Adams (2009) showed that narcissistic early adolescents become especially motivated to retaliate when provoked by their peers. The increased tendency to respond when attacked leads us to expect that the instability of self-esteem may have a stronger relationship with reactive aggression than with proactive aggression.

States and Traits of Self-Esteem: Trait-State-Occasion Modeling

Psychologists have long recognized that no measurement takes place in a situational vacuum (Steyer, Schmitt, & Eid, 1999). Nevertheless, although various instruments measure individual differences in particular contexts, the central focus of previous research has typically been on individual differences in trait-like attributes. The observation that constructs vary in their stability over time has given rise to a distinction between state and trait variables. Trait-like self-esteem reflects individuals’ representation of how they typically feel about themselves over time and across contexts, and is presumed to be persistent and stable. On the other hand, state-dependent self-esteem is subject to changes according to individuals’ immediate experiences, and assumed to fluctuate over time. Individual differences exist not only in the trait component of self-esteem but also in its state component (Yasuda, Lawrenz, Von Whitlock, & Lubin, 2004).

Self-esteem has typically been examined as a stable construct, and intraindividual variability has generally been ignored or poorly specified. Studies using intraindividual standard deviations to represent day-to-day variability in self-esteem stand out as the notable exceptions (e.g., Kernis, 2005; Zeigler-Hill & Wallace, 2012). Using standard deviations is advantageous because they are easy to calculate, do not depend on restrictive assumptions, and have face validity (Eid & Diener, 1999; Larsen, 1987). Nevertheless, they do not reproduce the process of change, and may be confounded with the frequency and extremity of change (Eid & Diener, 1999).

The present study recognizes that standard deviations do not exhaust all forms of variability that may exist in a data set. For example, several covariance structure models have been developed to address the state-trait distinction, including the latent state−trait model (e.g., Eid & Diener, 1999; Schmitt & Steyer, 1993; Steyer et al., 1999), the trait-state-error model (Kenny & Zautra, 1995), and the trait-state-occasion (TSO) model (Cole, Martin, & Steiger, 2005). They belong to the class of latent state−trait models that have been applied to various areas of psychology (e.g., Hatton et al., 2008; Olatunji & Cole, 2009; Schmitt & Steyer, 1993; Yasuda et al., 2004). Such models allow for the separation of variances from the measurement error from true variances as well as for the estimation of the proportion of true variances from the trait. Consequently, they attempt to explore stable and time-varying sources of variations in a construct simultaneously (Ciesla, Cole, & Steiger, 2007).

The present study adopts the TSO model to evaluate individual differences in the instability of self-esteem because it generates fewer empirical problems. For example, the likelihood of inappropriate solutions (i.e., out-of-range parameter estimates or Hessian matrix singularities) is less common even when the sample is too small or when there are too few waves (Cole et al., 2005). The TSO model enables the decomposition of variability in repeated measures of self-esteem into trait- and occasion-specific components. As shown in Figure 1, the state (S_t) is conceived as a latent variable that can be represented by manifest variables (Y_i) at time t. Therefore, the state reflects an individual’s actual feeling or condition at a particular point in time (t). The state can be decomposed into factors representing the trait (T) and the occasion (O_t). Therefore, an individual’s state is the result of highly stable personal and situational factors. The occasion (O_t) represents the situational circumstances impinging on the individual at the time of the evaluation (Ciesla et al., 2007; Cole et al., 2005).

Figure 1.

Trait-state-occasion model: Trait and state aspect of self-esteem.

The present study applied the TSO model to answer two major research questions. The first question deals with the extent that the children’s self-esteem results from stable characteristics or from occasion-specific factors. Any given construct is in a continuum of traitness and can be divided into its state and trait components (Kenny & Zautra, 2001). A substantial proportion of the variance in self-esteem is expected due to trait and state components of self-esteem. Accordingly, self-esteem is expected to contain a variability-specific component that is not shared with the mean values in self-esteem. The second question is whether state-dependent and changeable self-esteems (latent state variables) are independent of aggression or whether they are correlated. The degree of intraindividual variability is a distinct individual characteristic for predicting outcome variables (Eid & Diener, 1999). Accordingly, the state component of self-esteem is expected to explain individual differences in aggressive behaviors. More specifically, when controlling the state component of self-esteem, it likely shows a stronger relationship with reactive aggression than with proactive aggression.

Recently, several studies have shown that unstable self-esteem is related more to reactive aggression than to proactive aggression (e.g., Bukowski et al., 2009). However, such findings are based on simple comparisons of correlation coefficients. Furthermore, previous studies rely only on standard deviations to assess the instability of self-esteem deviations (e.g., Kernis, 2005; Zeigler-Hill & Wallace, 2012). By applying the TSO model, the present study will contribute to the construction of analytical systems for assessing changeable self-esteem. Given that no study has applied the TSO model to examine the temporal (i.e., state) and dispositional (i.e., trait) components of self-esteem measure, this study would provide a better understanding of the instability of self-esteem and its relationship with two different functions of aggressive behaviors.

Method

Participants

The sample included 235 children (110 boys and 125 girls) from 10 classrooms in Grades 5 and 6 of two elementary schools in South Korea. The average age of the subjects was 11.2 years (SD = 0.83 years). Participants completed the self-report measures on four separate occasions. The response across four surveys was 85.5% (n = 201). At each time, more than 90% of the potential pool of participants was in the sample. Parental permission was obtained for each participant through an active consent procedure.

Measures

Self-esteem

Self-esteem was assessed through the event-sampling procedure where participants were each assessed four times for four consecutive days, specifically Tuesday through Friday. The four days were coded from 1 to 4 to create a measure of the “day.” During each day, assessments were made 10 minutes before the end of the school day. Participants completed a set of questions included in a small booklet at each assessment.

Harter’s (1985) Self-Perception Profile for Children was administered to assess general self-worth. This subscale consisted of six items that assessed the extent to which the participants were satisfied with themselves, and the way they were leading their lives. Each item included a statement about how some children think or feel about themselves, and the participants were asked to indicate the extent to which they were similar to these children based on a four-point Likert-type scale ranging from 1 = I am not like these children at all to 4 = I am exactly like these children. One item was “Some kids like the kind of person they are.” In this study, these six items were aggregated to create an index of self-esteem for each of four assessments. Cronbach’s alpha for the four assessments ranged from .82 to .88 (median = .85).

Aggression

Aggression was assessed using the peer nomination of the scale of proactive and reactive aggression (Dodge & Coie, 1987) in conjunction with the first assessment of self-esteem. This was a one-time assessment, whereas self-esteem was measured repeatedly. The participants were asked to mark the name of anyone they believed fit the associated characteristics, with the measure consisting of six items (three for each scale). One item for proactive aggression was “Threatens or bullies others to get his/her own way,” and an item for reactive aggression was “When teased or threatened, he/she gets angry easily and strikes back.” Consistent with Crick and Grotpeter’s (1995) procedures, children were not restricted to same-gender nominations. The number of nominations that children received for each item were added and converted to standardized z scores by class in order to eliminate any variations that might result from class differences.

This measure showed sufficient construct validity. Reactive aggression was uniquely related to impulsiveness and the endorsement of aggression in response to peer provocation (i.e., hostile-attribution bias) and proactive aggression was uniquely associated with the belief that positive consequences would derive from aggressive behaviors (i.e., response bias; Dodge, Lochman, Harnish, Bates, & Pettit, 1997). The Cronbach’s alpha for the proactive and reactive aggression were .84 and .86, respectively.

Data Analysis

A TSO structural equation model (Cole et al., 2005) was constructed for self-esteem in order to examine whether the time series of self-esteem measures would contain state and trait components. Figure 1 shows the TSO model. The model specifies the time series of latent state variables (S_t), representing four waves of self-esteem. To indicate a latent state self-esteem, six items were randomly aggregated into two parcels to have a more normal distribution. The variance of S_t is partitioned into what is due to a stable trait factor (T) and a time-specific occasion factor (O_t). Here T represents various aspects across S variables that are stable over time, whereas O_t represents those across S_t variables that fluctuate over time. In other words, a time-varying occasion factor and an orthogonal time-invariant trait factor are added to form a composite state factor. In addition, the TSO model specifies an autoregressive relationship between O_t and O_t+1, which allows for the latent occasion variable to have some degree of stability over time. After specifying the TSO model of self-esteem, a series of conditional TSO models with aggressive behaviors as outcome variables is constructed to explore the relationships of state and trait components of self-esteem to each type of aggressive behavior.

AMOS 7.0 (Arbuckle, 2006) was used to estimate the TSO model using the full-information maximum likelihood (FIML) method, which allows for the inclusion of participants with partial data on dependent variables. This estimation method fits a covariance structure model to observed raw data for each participant, not to a covariance matrix of observed variables. Therefore, instead of imputing data, this method uses all available data to estimate parameters and standard errors. In addition, the results of Little’s Missing Completely at Random (MCAR) test showed that the data are not missing completely at random; χ²(138) = 488.481, p < .001. Therefore, it is not safe to listwise delete cases with missing values or singly impute missing values. The model fit was evaluated using the χ² statistic and several fit indices (i.e., normed fit index [NFI], comparative fit index [CFI], Tucker-Lewis index [TLI], root mean square error approximation [RMSEA]). A CFI and TLI with values greater than .95 and an RMSEA value of .05 or less reflect a good fit (Browne & Cudeck, 1993; Hu & Bentler, 1999). The comparative model fit was evaluated using the chi-square difference tests in the form of nested model comparisons.

Results

Correlation Analysis

Table 1 shows the correlations between the variables. The simple correlations between each type of aggression and the level of self-esteem (mean of four repeated measures) were not significant (rs = −.077 and −.010 for proactive and reactive aggression, respectively), indicating that the level of self-esteem did not sufficiently explain aggression. However, the instability indices of self-esteem (SD across four repeated self-esteem measures) had significant positive relationships with both types of aggression (rs = .305 and .186, all p < .05) for proactive and reactive aggression, respectively) as expected. This suggests that the day-to-day variability in self-esteem does sufficiently explain aggressive behaviors in children. That is, aggressive children may experience substantial intraindividual changes within a period as short as 4 days.

Table 1.

Intercorrelations and Descriptive Statistics of Research Variables.

	Day 1	Day 2	Day 3	Day 4	Level	Instability	Pro	Rea
Day 1	—
Day 2	.757***	—
Day 3	.821***	.893***	—
Day 4	.751***	.927***	.938***	—
Level	.881***	.944***	.959***	.958***	—
Instability	−.391***	.028***	−.173*	.008	−.173*	—
Proactive	−.130	.016	−.044	.041	−.077	.305***	—
Reactive	−.044	.048	.020	.059	−.010	.186*	.623***	—
$\bar{X}$	2.94	3.02	2.99	3.03	2.99	0.25	0.51	0.76
SD	.78	.80	.85	.82	.75	.29	.69	.79

Note. Level = mean of four self-esteem measures; Instability = intraindividual standard deviation of four self-esteem measures; Pro = proactive aggression; Rea = reactive aggression.

p < .05. **p < .01. ***p < .001.

TSO Model Identification

The TSO model was constructed for self-esteem to examine whether the time series of self-esteem measures would include state and trait components. Following Cole et al. (2005), several identifying constraints were imposed for the model identification. Factor loadings and the error variances for state indicators were constrained to be equal across waves. Autoregressive path coefficients between adjacent O factors were also constrained to be equal to each other. Finally, residual variances for O factors were specified to be equal across waves. The baseline model (Model 1) was constructed with all these constraints imposed.

The simplifying assumptions of the homogeneity of regression slopes (equality of autoregressive pathways or β) and the homogeneity of residual variances were then examined. Although these assumptions can simplify models, there is no a priori reason to assume the autoregressive carryover or residual variance to remain constant over time. Therefore, these constraints were individually removed from the full TSO model to determine whether doing so would result in a significant increase in the model fit, as indicated by a change in χ². Accordingly, the fairly restricted baseline model (M1) was compared with Model 2 (M2, which relaxed the assumption of the homogeneity of residual variances), Model 3 (M3, which relaxed the assumption of the homogeneity of error variance for measurement variables), and Model 4 (M4, which relaxed the homogeneity of regression slopes).

As shown in Table 2, M1 did not provide a sufficient fit to the data (χ² = 275.012, df = 31, p < .001; NFI = .848, CFI = .863, TLI = .841, RMSEA = .196, 90% confidence interval [CI] = [.175, .218]), whereas M2 provided a better fit (χ² = 111.333, df = 24, p < .001; NFI = .939, TLI = .926, CFI = .951). There was a significant chi-square difference between M1 and M2 (Δχ² = 163.679, Δdf = 7, p < .001), implying that specifying constraints on the homogeneity of residual variances was not consistent with the data. M3 further removed the equality constraint of homogeneity of error variance for measurement variables from M2 and it produced a significant increase in the model fit (Δχ² = 214.518, Δdf = 10, p < .001). M4 further removed the constraint of autoregressive pathways from M3 and it produced a significant increase in the model fit (Δχ² = 230.179, Δdf = 12, p < .001). Therefore, M4 was chosen over alternative models for further analysis.

Table 2.

Fit Indices for Trait-State-Occasion Models.

Models	χ²(df)	Δχ²(Δdf)	NFI	CFI	TLI	RMSEA	CI
Model 1	275.012(31)***	—	.848	.863	.841	.196	[.175, .218]
Model 2	111.333(24)***	163.679(7)***	.939	.951	.926	.134	[.109, .159]
Model 3	60.494(21)***	214.518(10)***	.969	.978	.962	.096	[.068, .125]
Model 4	44.833(19)***	230.179(12)***	.975	.985	.972	.082	[.051, .113]

Note. NFI = normed fit index; CFI = comparative fit index; TLI = Tucker-Lewis index; RMSEA = root mean square error approximation; CI = 90% confidence interval. Model 1 = baseline model; Model 2 = homogeneity of residual variances; Model 3 = homogeneity of error variance; Model 4 = homogeneity of regression slopes.

***

p < .001.

The results indicate that the largest portion of the variance in self-esteem was attributed to the latent trait variable. Standardized trait loadings ranged from .792 to .936, indicating that the shared variance between state and trait factors of self-esteem ranged from 62.7% to 87.6%. A small but significant portion of reliable individual differences in a particular measurement occasion was due to unstable but systematic effects of the situation (O). Standardized O loadings ranged from .351 to .610, implying that 12.3% to 37.2% of the reliable interindividual difference in self-esteem was attributable to intraindividual variability. This was smaller than the stable dispositional individual differences (i.e., trait differences). However, the results clearly indicate significant and systematic fluctuations in self-esteem.

Conditional Model: A TSO Model with Aggression

Trait self-esteem and aggression

The conditional model was specified to assess the relationship between trait self-esteem (T) and aggression. This model provided a good fit to the data (χ² = 98.326, df = 34, p < .001; CFI = .969, TLI = .953, NFI = .953, RMSEA = .092). As shown in Table 3, neither proactive aggression nor reactive aggression was significantly related to the trait component of self-esteem. Consistent with the results for simple correlations (no significant correlations between self-esteem level and aggression), TSO results reveal no significant relationship between the trait component of self-esteem and aggression. Noteworthy is that these are independent effects, with these relationships taking into account the degree of the overlap between the two types of aggression. However, these results are almost identical even with the separate entry of each type of aggression.

Table 3.

Results of Conditional TSO Models: Estimates of Path Coefficients Between Self-Esteem and Aggression.

	Proactive aggression				Reactive aggression
	Est.	SE	C.R.	Stan. Est.	Est.	SE	C.R.	Stan. Est.
Trait	−.082	.072	−1.138	−.083	−.018	.082	−.222	−.016
State
Day 1	−.331	.079	−4.171	−.320***	.188	.070	2.697	.207**
Day 2	−.176	.082	−2.155	−.159*	.144	.072	2.006	.148*
Day 3	−.212	.086	−2.466	−.182*	.152	.076	2.013	.149*
Day 4	−.165	.082	−2.001	−.149*	.148	.072	2.049	.152*

Note. TSO = trait-state-occasion; Est. = estimate; SE = standard error; C.R. = critical ratio; Stan.Est. = standard estimate.

p < .05. **p < .01. ***p < .001.

State self-esteem and aggression

A conditional model was specified to assess the relationship between state self-esteem (S) and aggression. This model provided a good fit to the data (χ² = 170.230, df = 28, p < .001; CFI = .951, TLI = .926, NFI = .939). As shown in Table 3, proactive aggression had a significant negative correlation with the state component of self-esteem even when the trait component was controlled for (β = −.149 to −.320, all p < .05). By contrast, reactive aggression had a significant positive correlation with the state component of self-esteem (β = .148 to .207, all p < .05). These results provide systematic evidence that each type of aggression was differentially associated with the state component of self-esteem. More importantly, the participants with a high state self-esteem were more likely to show reactive aggression and less likely to engage in proactive aggression.

Discussion

The purpose of this study is to provide a clearer understanding of the relationship between self-esteem and aggression by focusing on intraindividual variability in self-esteem. In addition, the study determines whether the state self-esteem is differentially associated with each type of aggression. The results provide evidence that the instability of self-esteem is not redundant with respect to the mean level of self-esteem and it is related to the defensive pattern of aggressive behaviors in early adolescence. The instability of self-esteem is an important element of it and can explain how and for whom self-esteem contributes to behavioral problems. The results are summarized as follows:

The first hypothesis addressed whether individual differences in self-esteem would result from trait and state components of self-esteem. In general, most psychological attributes are neither traits nor states. That is, most attributes have components of trait and state (Hagemann & Meyerhoff, 2008) and are in a continuum of “traitness” (Kenny & Zautra, 2001). Nevertheless, dispositional (trait) and processing (state) approaches to personality traits are generally thought to be competing and mutually preemptive (Matthews & Deary, 1998; Morf & Rhodewalt, 2001).

The present study assumes that self-esteem is best represented based on a trait-like continuum. Given that individuals with unstable self-esteem tend to experience substantial short-term fluctuations, the concept of the instability of self-esteem can be understood within a trait-state paradigm. That is, the greater the variance in the state component of self-esteem, the more insecure and fluctuating the tendency to respond to external conditions. Previous studies (e.g., Kernis, 2005) have addressed changeable short-term self-esteem, but most have used standard deviations to assess fluctuations in self-esteem and measured the level of self-esteem based on a separate scale.

To focus on the instability of self-esteem, the present study directly and clearly re-estimated the state component of self-esteem by applying the TSO model. This approach allows for the consideration of intraindividual variability and interindividual differences by controlling for measurement errors (Cole et al., 2005). The present study is innovative to apply latent trait−state structural modeling to isolate state and trait variances in self-esteem to explore intraindividual differences in self-esteem. The results indicate that self-esteem was composed of an important variability-specific component (latent state) that was not shared with the mean level (latent trait) of self-esteem. In addition, there were interindividual differences in intraindividual variability that were unstable but systematic. These results highlight the importance of treating self-esteem more as a continuous construct in a continuum of traitness than as a static individual difference.

The second hypothesis addressed whether unstable self-esteem would manifest more in one type of aggression than in another. The results indicate that the relationship between the state component of self-esteem and aggression varied across the two types of aggressive behaviors; state self-esteem was positively related to reactive aggression and negatively to proactive aggression. However, the trait self-esteem was not related to any types of aggression, suggesting that unstable self-esteem, not dispositional self-esteem, may account for the emergence of aggression and reflect in defensive and reactive ways rather than in proactive ones.

Unstable self-esteem is poorly anchored and thus susceptible to the vicissitudes of events of potential relevance to self-esteem. The more unstable the individual’s self-esteem, the more likely he or she will strongly react to everyday events (Greenier et al., 1999). Unstable self-esteem reflects a sense of self-worth that is vulnerable to threats because it requires continued bolstering, protection, and validation through various self-protective or self-enhancing strategies (Kernis, 2005). This perspective is empirically supported by this study’s results indicating that children with high state self-esteem are more likely to engage in reactive aggression. Furthermore, the negative relationship between this component and proactive aggression indicate that, with the dispositional self-esteem held constant, children with a high state self-esteem tend to be reluctant to engage in proactive aggressive behaviors.

The present study also supports the previous findings that, although proactive and reactive aggressions are highly correlated, these two functions of aggression are distinct (Dodge, 1991; Little et al., 2003). Reactive aggression is a hostile and angry response that functions as retaliation to a perceived threat or provocation. However, proactive aggression is defined as instrumental aggressive behavior that occurs without apparent provocation or instigation (Seah & Ang, 2008). Reactive aggression has been associated with more severe deficits in peer relations (Card & Little, 2006; Prinstein & Cillessen, 2003) and with higher levels of anxiety (Seah & Ang, 2008), whereas proactive aggression does not lead to negative peer interactions and interpersonal relations (Cillessen & Mayeux, 2004). Furthermore, the present study confirms that the children with a strong state component of self-esteem are more likely to show defensive patterns of aggressive behaviors and less likely to show proactive patterns of aggressive behaviors. This study contributes to the literature by providing a better understanding of not only why children with unstable self-esteem are more likely to be aggressive, but also in how they express their hostile behaviors.

Of note was that high state self-esteem was associated with higher reactive aggression and lesser proactive aggression, whereas standard deviation of self-esteem was positively associated with functions of aggression. These findings seem to be contradictory at first glance. Although state aspects of self-esteem and standard deviation of self-esteem are used to refer to the unstable self-esteem, they indicate different aspects of instability. Standard deviation of self-esteem is an index of instability indicating the extent of fluctuation, whereas state self-esteem is the level of unstable self-esteem. State self-esteem is dependent on the individuals’ immediate experiences and is attributable to specific situational factors at a particular point in time. Therefore, high state self-esteem indicates high level of unstable and situational self-esteem. By contrast, high standard deviation indicates a high degree of fluctuation regardless of direction of change (i.e., increase or decrease). No matter the level of state self-esteem, the standard deviation can be high to the extent that it fluctuates across the time.

This study has several limitations: First, the analysis was based on a sample of 235 elementary school students. Although a sample of this size may be associated with large standard errors, it was possible to produce results providing clear support for the hypotheses. Future research should use larger samples to replicate this study’s results. Second, the time lag might have influenced the results. Although the degree of intraindividual variability tends to be consistent and thus can be considered an individual disposition (Eid & Diener, 1999), future research should examine whether intraindividual variability is a valid predictor of behaviors over a longer period of time.

Nevertheless, the present study provides important insights into how and for whom self-esteem contributes to behavioral problems and suggests some interesting avenues for future research. In conclusion, the results show that (a) the trait and state components of self-esteem can be examined within a unitary framework, (b) the high state self-esteem is characterized by high levels of reactivity and defensiveness in aggressive behaviors, and (c) the extent to which self-esteem remains stable or fluctuate over time can be considered as a potential etiologic factor in the development of aggressive behaviors and presumably psychosocial problems in general.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by Kyungpook National University A·S Research Fund, 2013.

Author Biography

Eunju J. Lee is an associate professor at the Kyungpook National University, South Korea. She received her PhD from the School of Education at the University of North Carolina at Chapel Hill. Her major research interests include peer relations, behavioral problems, self-concept, and family dysfunction.

References

Arbuckle

J. L.

(2006). AMOS 7.0 user’s guide. Chicago, IL: SPSS.

Bandura

(1973). Aggression: A social learning analysis. Englewood Cliffs, NJ: Prentice-Hall.

Baumeister

R. F.

Bushman

Campbell

(2000). Self-esteem, narcissism and aggression: Does violence result from low self-esteem or from threatened egotism? Current Directions in Psychological Science, 9, 26-29.

Baumeister

R. F.

Campbell

J. D.

Krueger

Vohs

(2003). Does high self-esteem cause better performance, interpersonal success, happiness, or healthier lifestyles? Psychological Science in the Public Interest, 4, 1-44.

Berkowitz

(1978). Whatever happened to the frustration-aggression hypothesis? American Behavioral Scientist, 21, 691-708.

Browne

Cudeck

(1993). Alternative ways of assessing model fit. In Bollen

Long

(Eds.), Testing structural equation models (pp. 136-162). Newbury Park, CA: SAGE.

Bukowski

W. M.

Schwartzman

Santo

Bagwell

Adams

(2009). Reactivity and distortions in the self during early adolescence: Narcissism, types of aggression, and functioning of the HPA axis. Development and Psychopathology, 21, 1249-1262.

Bushman

B. J.

Baumeister

R. F.

Thomaes

Ryu

Begeer

West

(2009). Looking again, and harder, for a link between low self-esteem and aggression. Journal of Personality, 77, 424-446.

Card

N. A.

Little

T. D.

(2006). Proactive and reactive aggression in childhood and adolescence: A meta-analysis of differential relations with psychosocial adjustment. International Journal of Behavioral Development, 30, 466-480.

10.

Ciesla

J. A.

Cole

D. A.

Steiger

J. H.

(2007). Extending the trait-state-occasion model: How important is within-wave measurement equivalence? Structural Equation Modeling, 14, 77-97.

11.

Cillessen

A. H. N.

Mayeux

(2004). From censure to reinforcement: Developmental changes in the association between aggression and social status. Child Development, 75, 147-163.

12.

Cole

D. A.

Martin

N. C.

Steiger

J. H.

(2005). Empirical and conceptual problems with longitudinal trait-state models: Introducing a trait-state-occasion model. Psychological Methods, 10, 3-20.

13.

Crick

Grotpeter

(1995). Relational aggression, gender, and social-psychological adjustment. Child Development, 66, 710-722.

14.

Crocker

Park

L. E.

(2004). The costly pursuit of self-esteem. Psychological Bulletin, 130, 392-414.

15.

Crocker

Wolfe

C. T.

(2001). Contingencies of self-worth. Psychological Review, 108, 593-623.

16.

Dodge

K. A.

(1991). The structure and function of reactive and reactive aggression. In Pepler

D. J.

Rubin

K. H.

(Eds.), The development and treatment of childhood aggression (pp. 201-218). Hillsdale, NJ: Lawrence Erlbaum.

17.

Dodge

K. A.

Coie

J. D.

(1987). Social-information-processing factors in reactive and proactive aggression in children’s peer groups. Journal of Personality and Social Psychology, 53, 1146-1158.

18.

Dodge

K. A.

Lochman

J. E.

Harnish

J. D.

Bates

J. E.

Pettit

G. S.

(1997). Reactive and proactive aggression in school children and psychiatrically-impaired chronically assaultive youth. Journal of Abnormal Psychology, 106, 37-51.

19.

Donnellan

Trzesniewski

Robins

Moffitt

Caspi

(2005). Low self-esteem is related to aggression, antisocial behavior, and delinquency. Psychological Science, 16, 328-335.

20.

Eid

Diener

(1999). Intraindividual variability in affect: Reliability, validity, and personality correlates. Journal of Personality and Social Psychology, 76, 662-676.

21.

Franck

De Raedt

(2007). Self-esteem reconsidered: Unstable self-esteem outperforms level of self-esteem as vulnerability marker for depression. Behaviour Research and Therapy, 45, 1531-1541.

22.

Greenier

K. G.

Kernis

M. H.

McNamara

C. W.

Waschull

S. B.

Berry

A. J.

Herlocker

C. E.

Abend

T. A.

(1999). Individual differences in reactivity to daily events: Examining the roles of stability and level of self-esteem. Journal of Personality, 67, 186-208.

23.

Hagemann

Meyerhoff

(2008). A simplified estimation of latent state–trait parameters. Structural Equation Modeling, 15, 627-650.

24.

Harter

(1985). The Self-Perception Profile for Children: Revision of the Perceived Competence Scale for Children (Manual). Denver, CO: University of Denver.

25.

Hatton

Donnellan

M. B.

Maysn

Feldman

B. J.

Larsen-Rife

Conger

R. D.

(2008). Family and individual difference predictors of trait aspects of negative interpersonal behaviors during emerging adulthood. Journal of Family Psychology, 22, 448-455.

26.

Bentler

P. M.

(1999). Cutoff criteria for fit indices in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1-55.

27.

Hughes

J. N.

Cavell

T. A.

Grossman

P. A.

(1997). A positive view of self: Risk or protection for aggressive children? Development and Psychopathology, 9, 75-94.

28.

Jang

S. J.

Thornberry

T. P.

(1998). Self-esteem, delinquent peers, and delinquency: A test of the self-enhancement hypothesis. American Sociological Review, 63, 586-598.

29.

Kenny

D. A.

Zautra

(1995). The trait-state-error model for multi-wave data. Journal of Consulting and Clinical Psychology, 63, 52-59.

30.

Kenny

D. A.

Zautra

(2001). The trait-state models for longitudinal data. In Collins

L. M.

Sayer

A. G.

(Eds.), New methods for the analysis of change (pp. 243-263). Washington, DC: American Psychological Association.

31.

Kernis

M. H.

(2003). Toward a conceptualization of optimal self-esteem. Psychological Inquiry, 14, 1-26.

32.

Kernis

M. H.

(2005). Measuring self-esteem in context: The importance of stability of self-esteem in psychological functioning. Journal of Personality, 73, 1569-1605.

33.

Kernis

M. H.

Grannemann

B. D.

Barclay

L. C.

(1989). Stability and level of self-esteem as predictors of anger arousal and hostility. Journal of Personality and Social Psychology, 56, 1013-1023.

34.

Kernis

M. H.

Lakey

C. E.

Heppner

W. L.

(2008). Secure versus fragile high self-esteem as a predictor of verbal defensiveness: Converging findings across three different markers. Journal of Personality, 76, 477-512.

35.

Larsen

R. J.

(1987). The stability of mood variability: A spectral analytic approach to daily mood assessment. Journal of Personality and Social Psychology, 52, 1195-1204.

36.

Little

T. D.

Jones

S. M.

Henrich

C. C.

Hawley

P. H.

(2003). Disentangling the “whys” from the “whats” of aggressive behavior. International Journal of Behavioral Development, 27, 122-133.

37.

Locke

K. D.

(2009). Aggression, narcissism, self-esteem, and the attribution of desirable and humanizing traits to self versus others. Journal of Research in Personality, 43, 99-102.

38.

Marsh

Craven

(2006). Reciprocal effects of self-concept and performance from a mutidimensional perspective: Beyond seductive pleasure and unidimensional perspectives. Perspectives on Psychological Science, 1, 133-163.

39.

Matthews

Deary

I. J.

(1998). Personality traits. Cambridge, UK: Cambridge University Press.

40.

Meier

L. L.

Orth

Denissen

Kuhnel

(2011). Age differences in instability, contingency, and level of self-esteem across the life span. Journal of Research in Personality, 45, 604-612.

41.

Morf

C. C.

Rhodewalt

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

42.

Olatunji

B. O.

Cole

D. A.

(2009). The longitudinal structure of general and specific anxiety dimensions in children: Testing a latent trait–state–occasion model. Psychological Assessment, 21, 412-424.

43.

Ostrowsky

M. K.

(2010). Are violent people more likely to have more low self-esteem or high self-esteem? Aggression and Violent Behavior, 15, 69-75.

44.

Paradise

A. W.

Kernis

M. H.

(2002). Self-esteem and psychological well-being: Implications of fragile self-esteem. Journal of Social & Clinical Psychology, 21, 345-361.

45.

Prinstein

M. J.

Cillessen

A. H. N.

(2003). Forms and functions of adolescent peer aggression associated with high levels of peer status. Merrill-Palmer Quarterly, 49, 310-342.

46.

Roberts

J. E.

Monroe

S. M.

(1992). Vulnerable self-esteem and depressive symptoms: Prospective findings comparing three alternative conceptualizations. Journal of Personality and Social Psychology, 62, 804-812.

47.

Salmivalli

(2001). Feeling good about oneself, being bad to others? Remarks on self-esteem, hostility, and aggressive behavior. Aggression and Violent Behavior, 6, 375-393.

48.

Scheff

T. J.

Fearon

D. S.

(2004). Cognition and emotion? The dead end in self-esteem research. Journal for the Theory of Social Behavior, 34, 73-91.

49.

Schmitt

M. J.

Steyer

(1993). A latent state-trait mode (not only) for social desirability. Personality and Individual Differences, 14, 519-529.

50.

Seah

S. L.

Ang

R. P.

(2008). Differential correlates of reactive and proactive aggression in Asian adolescents: Relations to narcissism, anxiety, schizotypal traits, and peer relations. Aggressive Behavior, 34, 553-562.

51.

Steyer

Schmitt

Eid

(1999). Latent state-trait theory and research in personality and individual differences. European Journal of Personality, 13, 389-408.

52.

Swann

W. B.

Jr. Chang-Schneider

McClarty

(2007). Do people’s self-views matter? Self-concept and self-esteem in everyday life. American Psychologist, 62, 84-94.

53.

Waschull

S. B.

Kernis

M. H.

(1996). Level and stability of self-esteem as predictors of children’s intrinsic motivation and reasons for anger. Personality and Social Psychology, 22, 4-13.

54.

Webster

G. D.

Kirkpatrick

L. A.

Nezlek

J. B.

Smith

C. V.

Paddock

E. L.

(2007). Different slopes for different folks: Self-esteem instability and gender as moderators of the relationship between self-esteem and attitudinal aggression. Self and Identity, 6, 74-94.

55.

Yasuda

Lawrenz

Von Whitlock

Lubin

(2004). Assessment of intraindividual variability in positive and negative affect using latent state-trait model analyses. Educational and Psychological Measurement, 64, 514-530.

56.

Zeigler-Hill

Wallace

M. T.

(2012). Self-esteem instability and psychological adjustment. Self and Identity, 11, 317-342.