Implicit and Explicit Self-Esteem Discrepancies, Victimizationand the Development of Late Childhood Internalizing Problems

Franca H. Leeuwis & Hans M. Koot &Daan H. M. Creemers & Pol A. C. van Lier

# Springer Science+Business Media New York 2014

Abstract Discrepancies between implicit and explicit self-esteem have been linked with internalizing problems amongmainly adolescents and adults. Longitudinal research on thisassociation in children is lacking. This study examined thelongitudinal link between self-esteem discrepancies and thedevelopment of internalizing problems in children. It further-more examined the possible mediating role of self-esteemdiscrepancies in the longitudinal link between experiences ofpeer victimization and internalizing problems development.Children (N=330, M age=11.2 year; 52.5 % female) werefollowed over grades five (age 11 years) and six (age 12 years).Self-report measures were used annually to test for victimiza-tion and internalizing problems. Implicit self-esteem wasassessed using an implicit association test, while explicitself-esteem was assessed via self-reports. Self-esteem discrep-ancies represented the difference between implicit and explicitself-esteem. Results showed that victimization was associatedwith increases in damaged self-esteem (higher levels of im-plicit than explicit self-esteem. Additionally, damaged self-esteem at age 11 years predicted an increase in internalizingproblems in children over ages 11 to 12 years. Furthermore,

damaged self-esteem mediated the relationship between age11 years victimization and the development of internalizingproblems. No impact of fragile self-esteem (lower levels ofimplicit than explicit self-esteem) on internalizing problemswas found. The results thus underscore that, as found inadolescent and adult samples, damaged self-esteem is a pre-dictor of increases in childhood internalizing problems.Moreover, damaged self-esteem might explain why childrenwho are victimized develop internalizing problems.Implications are discussed.

Keywords Victimization . Self-esteem discrepancies .

Internalizing problems . Elementary school children

The link between self-esteem and psychopathology has beenwell established (Cole et al. 2001; Harter 1993; Mann et al.2004; Sowislo and Orth 2013). However, most of this researchhas focused on explicit self-esteem. Recently, it has beenproposed that in addition to explicit self-esteem, implicitself-esteem should be considered when trying to understandthe development of psychopathology (Greenwald and Banaji1995). In fact, it may especially be the discrepancy betweenimplicit and explicit self-esteem that explains the relationshipbetween self-esteem and different forms of psychopathology(Bosson et al. 2003; Bosson et al. 2000). Our knowledge ofthe influence of self-esteem discrepancies on internalizingproblem development in childhood is limited due to relianceon (mostly) cross-sectional studies among adults and adoles-cents, and by not accounting for contextual factors. We aim toaddress this lack of research by (a) studying the role ofdamaged versus fragile self-esteem in the development ofchildren’s internalizing problems across ages 11 to 12 years,

F. H. Leeuwis (*) :H. M. Koot : P. A. C. van LierDepartment of Developmental Psychology and EMGO Institute forHealth and Care Research, VUUniversity, Van der Boechorststraat 1,1081 BTAmsterdam, The Netherlandse-mail: F.H.Leeuwis@vu.nl

D. H. M. CreemersRadboud University Nijmegen, Nijmegen, Netherlands

D. H. M. CreemersMental Health Care Institute, GGZ Oost Brabant, Oss, TheNetherlands

J Abnorm Child PsycholDOI 10.1007/s10802-014-9959-5

and (b) placing this relationship in the context of victimizationexperiences in children attending elementary schools in theNetherlands.

Discrepancies between explicit and implicit self-esteemhave been suggested to play a role in psychopathology(Bosson et al. 2003; Schröder-Abé et al. 2007a). Dual processmodels support this premise because they propose that infor-mation is processed through two different systems which eachhave different operating rules (i.e., controlled versus automat-ic) (Epstein 1994; Strack and Deutsch 2004). Indeed, becauseexplicit and implicit self-esteem reflect different underlyingprocesses, discrepancies between the two can develop. Twoforms of self-esteem discrepancies have been identified in theliterature, fragile or defensive self-esteem, consists of relativehigher levels of explicit than implicit self-esteem (Bossonet al. 2003; Jordan et al. 2003). Damaged self-esteem, on theother hand, reflects higher levels of implicit than explicit self-esteem (Schröder-Abé et al. 2007a). It is suggested that bothself-esteem discrepancies are maladaptive because they implythat individuals hold two discrepant and inconsistent views ofthe self. These inconsistencies in implicit and explicit beliefscould lead to discomfort and maladjustment (Schröder-Abéet al. 2007b). In accordance with this, fragile self-esteem hasbeen associated with narcissism, defensiveness and ag-gression (Jordan et al. 2003; Sandstrom and Jordan2008; Zeigler-Hill 2006).

Damaged self-esteem has been linked to internalizing prob-lems. For instance, the combination of low explicit and highimplicit self-esteem has been found in depressed adults withsuicidal ideation (Franck et al. 2007) and in socially anxiousearly adult women (De Jong 2002). In addition, a study amongyoung adult women found that damaged self-esteem wasassociated with depressive symptoms, suicidal ideation andloneliness (Creemers et al. 2012). Furthermore, among ado-lescents damaged self-esteem was related to social anxiety(Schreiber et al. 2012). Other studies, however, found noassociations between a combination of high implicit and lowexplicit self-esteem and depression and social anxiety in earlyadolescence (De Jong et al. 2012; Van Tuijl et al. 2014).

Despite some empirical evidence suggesting a possible linkbetween damaged self-esteem and internalizing problems,research in this field is still sparse and hampered by a numberof methodological and design limitations. These limitationsinclude the reliance on mainly cross-sectional studies, theomission of consideration of context in the study of the roleof damaged self-esteem on internalizing problems and theomission of studying this link in childhood. Previous studiesof self-esteem discrepancies and internalizing problems main-ly used a cross-sectional design (Creemers et al. 2012). To ourknowledge, only one longitudinal study exists (Van Tuijl et al.2014). This limits our ability to understand whether self-esteem discrepancies are associated with the development ofinternalizing problems (Kraemer et al. 2000). In fact,

longitudinal studies with repeated assessments of internalizingproblems and self-esteem measures are needed to test whetherindividual differences in self-esteem discrepancies link toindividual differences in the development of internalizingproblems (Masten and Cicchetti 2010).

Related to the study design, when studying the associationbetween self-esteem discrepancies and the development ofinternalizing problems in children it is important to take intoaccount contextual risk factors, such as social experiences, asthese may influence the development of both internalizingproblems and possibly self-esteem discrepancies. In fact,Leary and Baumeister (2000) state that self-esteem is asociometer which indicates the level of being socially accept-ed. Troublesome social experiences with peers, such as beinga victim of bullying, frequently occur in the childhood period(Veenstra et al. 2005) and have been linked to symptoms ofanxiety and depression in children (Biebl et al. 2011; Bondet al. 2001; Crick and Bigbee 1998; Hawker and Boulton2000; Ladd and Troop-Gordon 2003; Reijntjes et al. 2010).Moreover, research suggests that self-esteem may link poorsocial experiences in children with the development of inter-nalizing problems. For instance, it has been suggested thatvictimization leads to lowered explicit self-esteem and this inturn predicts maladjustment (Grills and Ollendick 2002; Laddand Troop-Gordon 2003; Troop-Gordon and Ladd 2005).

It may, however, well be that experiences like victimizationspecifically link to discrepancies between implicit and explicitself-esteem in the damaged direction, which in turn are asso-ciated with internalizing problems. This line of thought coin-cides with the idea that damaged self-esteem represents adiscrepancy between individuals’ high goals or standards(high implicit self-esteem) and the reality (low explicit self-esteem) (Franck et al. 2007). Individuals with damaged self-esteem then experience a constant feeling of not being able tolive up to their own standards, which could lead to internaliz-ing problems such as depression. In line with this thought,Zeigler-Hill and Terry (2007) suggests that people with dam-aged self-esteem set high standards for themselves and arecritical about their own performance. These characteristics areassociated with internalizing problems such as social anxiety.One could speculate that when children with positive implicitself-esteem, implying high goals or standards, experiencevictimization in elementary school, their explicit self-esteemcan develop negatively. This damaged self-esteem could thentransfer the effects of victimization into feelings of anxiety orloss resulting in the development of internalizing problems.Despite this theoretical plausibility and findings from adultstudies, to our knowledge, no studies have been published thatinvestigated the possible mediating role of self-esteem dis-crepancies in childhood.

With respect to the age of subjects, previous studies intoself-esteem discrepancies and internalizing problems focusedon adults and adolescents. Additionally, studies into young

J Abnorm Child Psychol

adolescents have been conducted (De Jong et al. 2012;Sandstrom and Jordan 2008; Van Tuijl et al. 2014). No study,however, focused on elementary school children. This seemsan omission because childhood is a crucial period for the onsetand development of internalizing problems (Roza et al. 2003).Internalizing problems surface in late childhood, and are pre-dictive of serious maladjustment later in life such as psychi-atric disorders (Clark et al. 2007; Fergusson et al. 2006).Furthermore, studying elementary school children in additionto young adolescents attending high school is important be-cause there are significant changes that come with the transi-tion from elementary school to high school that are of possibleinfluence on self-esteem (Wigfield et al. 1991). For instance, adisruption of social networks accompanies the transition tohigh school. Additionally, differences in school environmentsexists, such as more emphasis on evaluation and performanceand multiple teachers in high school. Consequently, resultsfrom studies that focus on self-esteem discrepancies in youngadolescents are not necessarily transferable to elementaryschool children. It therefore seems crucial to study the roleof victimization and self-esteem discrepancies in the develop-ment of internalizing problems in late childhood.

Finally, although no literature is available on gender differ-ences in self-esteem discrepancies, findings on other relevantvariables suggest that possible sex differences should be ex-plored when testing for the role of damaged self-esteem in thedevelopment of internalizing problems. Girls are more proneto developing internalizing problems (Ohannessian et al.1999) and tend to have lower explicit self-esteem (Klinget al. 1999; Muris et al. 2003; Robins and Trzesniewski2005) than boys. Research into sex differences in implicitself-esteem in children is scarce but shows no differencesbetween boys and girls (De Jong et al. 2012). More impor-tantly, there are indications that the mediating link of explicitself-esteem in the association between victimization and in-ternalizing problems exists only for girls (Grills and Ollendick2002). Grills and Ollendick (2002) propose that girls are morelikely than boys to incorporate negative feedback from peersinto their self-perceptions. We will therefore explore sex-differences in the studied links.

The present study investigated the role of damaged self-esteem in the development of internalizing problems in thecontext of victimization experiences in 330 children followedlongitudinally from age 11 to 12 years. Firstly, we consideredthe link from victimization to self-esteem discrepancies. Wehypothesized that victimization would be associated with anincrease in self-esteem discrepancies, in that it would increasechildren’s damaged self-esteem. Secondly, we addressed thepredictive link of self-esteem discrepancies to the develop-ment of internalizing problems. We anticipated that damagedself-esteem will predict increases in internalizing problems.Thirdly, we tested the role of self-esteem discrepancies in thelongitudinal association between victimization and

internalizing problems. We expected the longitudinal linkbetween victimization experiences and internalizing problemdevelopment to be mediated by children’s damaged self-es-teem. Finally, we will explore sex-differences in the studiedlinks. However, as longitudinal studies on the link betweendiscrepancies scores between implicit and explicit self-esteemand internalizing problems in the context of victimization inchildhood are lacking, we cannot formulate clear hypotheseson possible sex-differences.

Method

Participants

Data were collected within a longitudinal study on children’ssocial, emotional and behavioral development over the ele-mentary school period. Data of the present study were collect-ed in 14 schools from the northern and eastern parts of theNetherlands, when children were in grade five (age 11). Thesechildren were followed until grade six (age 12). The samplecomprised of 330 children (52.5 % girls). The mean age ofthese children was 11.2 years, (SD=0.4) in grade five. Beforeeach school assessment parents received written informationabout the measurements and procedures and were given theopportunity to decline from the participation of their chil-d(ren). Almost all children were allowed to participate(98.5 %). Children were informed about the assessment inthe classroom and could decline from participation at anygiven time. The Medical Ethical Review Board of the VUMedical Centre has approved this study.

Procedure

Children were administered questionnaires on internalizingproblems, explicit self-esteem and victimization during schoolhours. Trained graduate and undergraduate psychology stu-dents guided the assessments. Children were then instructedon the Brief Implicit Association Test Self Esteem (BIAT-SE)and completed the BIAT-SE on laptops individually. The se-quence of testing (classical questionnaires and BIAT-SE) wascounterbalanced per classroom to prevent order effect bias.Afterwards, children received a present as a token of appreci-ation. Of the initial 330 children, data was missing for 6children on the grade six assessment due to illness and ab-sence, and for 4 children no parental permission for participa-tion was obtained at that time.

Measures

Internalizing Problems At both age 11 years (fifth grade) andage 12 (sixth grade), children completed the emotional sub-scale of the Strengths and Difficulties Questionnaire (SDQ;

J Abnorm Child Psychol

Goodman 1997). The emotional subscale of the SDQ consistsof five items describing positive and negative statements aboutthe self, measured on a 3-point scale, ranging from not true todefinitely true. Items include “I ruminate a lot” and “I amnervous in new situations”. Cronbach’s alpha of the SDQemotional subscale was 0.68 at age 11 and at age 12 it was 0.69.

Implicit Self-esteem The Brief Implicit Association Test SelfEsteem (BIAT-SE) was used to measure implicit self-esteem(Sriram and Greenwald 2009). The BIAT-SE, a modified ver-sion of the Implicit Association Task (IAT) (Greenwald andBanaji 1995), is a computerized test that measures the strengthof the association between a target concept (“me”) and aattribute concept (“worthless” versus “valuable”). This asso-ciation is represented by the reaction time that requires sub-jects to sort words into categories. The BIAT-SE consists oftwo practice trials and four test trials. In each of these trials, theconcepts “me” and “worthless” (trail 1 and 3) or the concepts“me” and “valuable” (trail 2 and 4) are projected on the top ofthe computer screen. When a trial starts, different positive andnegative words (e.g., failure, good, self, other) appear on thecomputer screen. Subjects are asked to respond, as quicklyand accurately as possible, by pressing a key if the word (e.g.,failure, good, self, other) belongs to one of the concept words(me and worthless/valuable) and by pressing a left hand key ifthe target word does not belong to the target concept words. Ifthe initial response is incorrect, a red cross appears on thecomputer screen, which disappears immediately after the rightresponse is given. Subjects are obligated to give the correctresponse before moving on with the next word. Latency of thecorrect response was measured and the number of errors istaken into account. A relative high score on the BIAT-SEmeans that a subject responded faster in the trials where wordshad to be placed to the target concept words “me” and “valu-able” than in the trials where words had to be placed to theconcept words “’me” and “’worthless”. Consequently, sub-jects with a higher score on the BIAT-SE have a strongerassociation between the concepts “self” and “valuable” wordsand therefore have a higher implicit self-esteem. The BIAThas found to have a satisfactory validity and reliability (Sriramand Greenwald 2009). The individual BIAT-SE scores werecalculated using the improved scoring algorithm (IAT-D effectmeasure) proposed by Greenwald et al. (2003).

Explicit Self-esteem The global scale of the Self-PerceptionProfile for Children (SPPC) was administered during bothwaves to measure explicit self-esteem (Harter 1982). Thissubscale consists of six items and was measured on a 3-point scale, ranging from not true to definitely true. Itemsinclude “’I like my life” and “I am content with myself”.The SPPC has been found to be a reliable and valid measureof children’s self-perception (Boivin et al. 1992; Harter 1982).Cronbach’s alpha was 0.89 at age 11 and 0.88 at age 12.

Victimization Self-report of victimization was assessed usingthe Social Experience Questionnaire (SEQ; Crick andGrotpeter 1996) during both waves. The subscales VictimPhysical Aggression and Victim Relational Aggression wereused in the present study. These subscales both consist of fiveitems, measured on a 5-point scale, ranging from never toalmost always. Items include “How often does someone saysomething mean about you?” and “How often do you getpushed by somebody?” Cronbach’s alpha for the subscaleVictim Physical Aggression at age 11 was 0.81 and 0.78 atage 12, and Cronbach’s alpha for the subscale VictimRelational Aggression was 0.75 at age 11 and 0.77 at age 12.

Statistical Analyses

Before fitting the structural models, we first determined thesize of the self-esteem discrepancies. For both time pointsseparately, the absolute difference between the standardizedscores on implicit and explicit self-esteem was calculated (DeLos Reyes and Kazdin 2004). To identify the dominant direc-tion of the discrepancy, we created a dummy coded variable,which indicated the direction of the self-esteem discrepancies(explicit self-esteem>implicit self-esteem =0, explicit self-esteem<implicit self-esteem =1). The interaction betweenself-esteem discrepancy size x self-esteem discrepancy direc-tion discriminated children with fragile from damaged self-esteem. Using discrepancy scores, a dummy variable and theinteraction between these two has been found to be a suitableway of testing differences between implicit and explicit self-esteem (Briñol et al. 2006; Creemers et al. 2012, 2013;Schreiber et al. 2012; Schröder-Abé et al. 2007a).

In step one of our analyses, we tested our first hypothesisthat victimization is associated with self-esteem discrepancies,specifically damaged self-esteem.We fitted a structural modelin which latent victimization scores were construed using thesubscales Victim Physical Aggression and Victim RelationalAggression of the Social Experience Questionnaire at age 11and 12 as indicators. The age 12 victimization score wasregressed on its age 11-year score. The size and direction ofself-esteem discrepancies and the interaction between size anddirection at age 12 years served as the outcome variables.These outcome variables were regressed on their age 11values and on victimization at age 11. The age 12 self-esteem discrepancy score, direction and interaction term werecorrelated with the age 12 victimization score to test whetherage 11 victimization predicted changes in self-esteem discrep-ancies while accounting for concurrent links between self-esteem scores and victimization at age 12.

In step two, we tested our second hypothesis on the effectof self-esteem discrepancies, specifically damaged self-es-teem, on the development of internalizing problems. A secondstructural model was fitted. Latent factor scores of internaliz-ing problems at age 11 and 12 years were considered, in which

J Abnorm Child Psychol

the five items of the emotional subscale of the Strengths andDifficulties Questionnaire at age 11 and 12 years respectivelyserved as the as the indicators. The age 12 internalizingproblem score was regressed on self-esteem discrepancy, di-rection and interaction term at age 11 years and on age 11 yearsinternalizing problems. The age 12 self-esteem discrepancyscore, direction and interaction term were also added to themodel, were regressed on their age 11 values and were corre-lated with the age 12 internalizing problem score.

From steps 1 and 2 of our analyses, we expected that (1)victimization will predict increases in damaged self-esteemdiscrepancies, and (2) that such discrepancies will predict in-crease in internalizing problems. We therefore tested the thirdhypothesis that self-esteem discrepancies mediate the link be-tween victimization and the development of internalizing prob-lems. Again a structural model was fitted. The victimizationand internalizing problems scores at age 12 years wereregressed on their prior values to account for stability over time.Cross-sectional correlations between victimization and internal-izing problems were included in the model. The self-esteemdiscrepancies and direction variables across ages 11 en 12 wereaggregated into a discrepancy and direction score respectivelyover ages 11/12. These scores, and the interaction term betweendiscrepancy and direction were added to the model to test forthe hypothesized indirect effect of victimization on the devel-opment of internalizing problems via the interaction term be-tween size and direction of self-esteem discrepancies(MacKinnon et al. 2002). As we fitted a SEM model, we usedthe residual centering approach to create the interaction variable(Little et al. 2006). The interaction variable was created by firstcomputing product terms between age 11 and age 12 years self-esteem size by direction variables (size11xdir11; size11xdir12;size12xdir11; size12xdir12). These four variables were thenregressed on all indicators (size and direction of self-esteemdiscrepancy at age 11 and 12 years). For instance, the variable“size11xdir11” was regressed on the size of self-esteem dis-crepancies at age 11 and age 12 and on the direction of the self-esteem discrepancies at age 11 and age 12. The residualsresulting from these regressions were then saved and used tocompute the interaction variable, which was then used in themodel (Little et al. 2006; Steinmetz et al. 2011). Finally, in orderto test for sex differencesmultiple groupmodels were run to testwhether the pathway apply similarly for boys and girls.

All models were fitted using Mplus 6.11 (Muthén andMuthén 2012). Model fit was determined through the com-parative fit index (CFI; values≥0.90) (Bollen and Long 1993)and the root mean square error of approximation (RMSEA;values≤0.08) (Browne and Cudeck 1992). We accounted forclustering of data within schools by using a sandwich estima-tor (Williams 2000) and robust standard errors were estimatedto account for possible non-normality of study variables.Missing data was handled through Full InformationMaximum Likelihood Estimation (FIML).

Results

Descriptive Statistics

The means and standard deviations of all study variables forboys and girls are presented in Table 1. As shown, boys scoredhigher on explicit self-esteem than girls on both assessmentpoints. Implicit self-esteem is positive in boys and girls at bothages. No differences were found in levels of implicit self-esteem between boys and girls. Girls scored significantlyhigher on internalizing problems at both ages. Boys scoredhigher on victimization at age 12 years then girls. The size ofself-esteem discrepancies was similar for boys and girls at age11 and at age 12. The direction of self-esteem discrepancieswas similar for boys and girls at age 11, χ2 (1, N=295) =0.67,p=0.42. At age 12, however, more girls (52.7 %) displayedhigher implicit than explicit self-esteem and boys more often(58.6 %) had higher explicit than implicit self-esteem, χ2 (1,N=310) =3.99, p<0.05. The correlations among all studyvariables are reported in Table 2. Explicit self-esteem corre-lated negatively with internalizing problems and victimiza-tion, both concurrently as well as longitudinally. No signifi-cant correlations between implicit self-esteem with the otherstudy variables were found. Victimization was both concur-rently and longitudinally correlated with internalizing prob-lems. Self-esteem discrepancy scores at age 11 were correlatedto all other variables. Self-esteem discrepancy scores at age 12were correlated with all other variables except for implicitself-esteem at age 11.

Table 1 Means and Standard Deviations of Assessed Study Variables

Girls Boys Test

M SD M SD F

Internalizing problems

Age 11 2.77 2.39 1.73 1.72 19.13**

Age 12 2.71 2.40 1.54 1.53 26.65**

Explicit Self-Esteem

Age 11 1.63 0.47 1.77 0.38 8.63**

Age 12 1.69 0.43 1.78 0.34 4.53*

Implicit Self-Esteem

Age 11 0.29 0.62 0.21 0.70 1.10

Age 12 0.40 0.63 0.31 0.65 1.59

Victimization

Age 11 4.02 3.04 4.26 2.74 0.53

Age 12 3.22 2.76 3.87 2.75 4.42*

Self-Esteem Discrepancy

Age 11 1.15 0.91 1.08 0.83 0.49

Age 12 1.15 0.98 1.01 0.80 1.92

*p<0.05. **p<0.01

J Abnorm Child Psychol

Step 1 Victimization and the Development of Self-EsteemDiscrepancies

To test if victimization was associated with the develop-ment of self-esteem discrepancies we fitted a structural modelin which the size and direction of self-esteem discrepancies atage 12 were regressed on victimization at age 11 after whichwe added the interaction between the size and direction at age12. Self-esteem variables at age 12 were also regressed ontheir age 11 values. Latent victimization scores were con-strued by the subscales Victim Physical Aggression (factorloading at age 11 was 0.81 and at age 12 0.80) and VictimRelational Aggression Factor (factor loading at age 11 was0.81 and at age 12 0.79).

After fitting the model with only main effects, yielding anon-significant link between victimization to the size of self-esteem discrepancies, β=0.11, p=0.22, but a significant as-sociation between victimization and the direction of self-esteem discrepancies, β=0.17, p=0.05, we fitted the modelcontaining also the interaction between the self-esteem dis-crepancies and the direction of self-esteem discrepancies(CFI=0.96, RMSEA=0.07). The results of path estimates aregiven in Table 3. As shown in Table 3, a significant linkbetween victimization at age 11 to the interaction term at age12 was found, B=0.08, SE=0.03, β=0.22, p<0.01. To break-down the interaction term, the association between victimiza-tion and the development of self-esteem discrepancies from age11 to 12 years was estimated for children who had a fragile(explicit self-esteem>implicit self-esteem) and damaged self-esteem (implicit self-esteem>explicit) at age 12. Results aredepicted in Fig. 1. For children who had a damaged self-esteemat age 12, victimization was significantly related to the devel-opment of the self-esteem discrepancies, β=0.28, p<0.01. Forchildren who had a fragile self-esteem at age 12, no significantassociation was found between victimization and the size ofself-esteem discrepancies, β=−0.14 p=0.11.

Step 2 Self-Esteem Discrepancies and the Development ofInternalizing Problems

We then tested whether self-esteem discrepancies wererelated to the development of internalizing problems overage 11 to 12 years. To this end, a structural model was fittedin which age 11 size and direction of self-esteem discrepanciespredicted age 12 internalizing problems, in addition toregressing the age 12 internalizing problems on its age 11values. The interaction between the size and direction of self-esteem discrepancies was later added to the model. Latentfactors were considered for the internalizing scores (rangefactor loadings at age 11: 0.46–0.71 and range factor loadingsat age 12: 0.41–0.77).

After first fitting a model in which size and the direction ofself-esteem discrepancies predicted changes in internalizingproblems, yielding significant effects of size, β=0.16,p<0.05, and direction of self-esteem discrepancies, β=0.14,p<0.01, the interaction between the size and the direction ofself-esteem discrepancies was added. This model fitted thedata satisfactorily (CFI=0.94, RMSEA=0.05). The results ofpath estimates are given in Table 3. It shows that the interac-tion term significantly predicted the development of internal-izing problems, B=0.10, SE=0.05, β=0.30, p=0.05. Tobreakdown the interaction term, the association betweenself-esteem discrepancies on the development of internalizingproblems from age 11 to 12 years was estimated for childrenwho had a fragile (explicit self-esteem>implicit self-esteem)and damaged self-esteem (implicit self-esteem>explicit).Results are printed in Fig. 2. It shows that for children withdamaged self-esteem the association between the size of thediscrepancy and the development of internalizing problems issignificant and positive, β=0.19, p<0.05. For childrenwith fragile self-esteem the size of the discrepancy wasnot associated with the development of internalizingproblems, β=−0.02, p=0.82.

Table 2 Correlations Between Assessed Variables

1 2 3 4 5 6 7 8 9

1. Internalizing problems age 11 –

2. Internalizing problems age 12 0.62** –

3. Explicit Self-Esteem age 11 −0.51** −0.45** –

4. Explicit Self-Esteem age 12 −0.43** −0.56** 0.54** –

5. Implicit Self-Esteem age 11 0.05 0.15* −0.01 −0.05 –

6. Implicit Self-Esteem age 12 0.06 0.05 0.02 0.00 0.29** –

7. Victimization age 11 0.44** 0.36** −0.39** −0.37** 0.07 0.05 –

8. Victimization age 12 0.30** 0.39** −0.38** −0.46** 0.04 0.05 0.59**

9. Self-Esteem Discrepancy age 11 0.19** 0.22** −0.45** −0.27** −0.26** −0.19** 0.17** 0.18**

10. Self-Esteem Discrepancy age 12 0.19** 0.36** −0.32** −0.54** −0.09 −0.16** 0.16** 0.25** 0.32**

*p<0.05 ** p<0.01

J Abnorm Child Psychol

Step 3 Self-Esteem Discrepancies and the Link BetweenVictimization and the Development of InternalizingProblems

After having found that (1) victimization predicted in-creases in damaged self-esteem, and (2) the development ofinternalizing problems was predicted by damaged self-esteemdiscrepancies, we tested our third hypothesis; the mediatingrole of self-esteem discrepancies in the longitudinal link

between victimization experiences and internalizing problemdevelopment. A structural model was fitted as depicted inFig. 3. Loadings of the physical victimization and relationalvictimization scale on the latent victimization score were 0.72and 0.77 respectively at age 11, and 0.79 and 0.84 respectivelyat age 12 years. We pooled the age 11 and age 12 years self-esteem scores to compute a robust self-esteem discrepancyscore. The interaction variable was created using the residualcentering approach (Little et al. 2006) and used to test formediation or the impact of age 11 years victimization on age12 years internalizing problem development by either discrep-ancies in self-esteem towards fragile or damaged self-esteem.

We first fitted a ‘non-mediation’ model in which weallowed for regression paths from victimization at age 11 yearsto the age 12 years internalizing problems, in addition toregression paths from age 11 victimization to age 11/12 self-esteem variables. At this stage, no paths from the self-esteemvariables to internalizing problems at age 12 years wereallowed for. Results show that age 11 victimization was asso-ciated with the size of self-esteem discrepancies, B=0.18,SE=0.04, β=0.37, p<0.01, the direction of self-esteem dis-crepancies, B=0.24, SE=0.03, β=0.52, p<0.01, and the in-teraction term, B=0.17, SE=0.03, β=0.34, p<0.01, in addi-tion to predicting the development of internalizing problemsB=0.24, SE=0.06, β=0.25, p<0.01.

Table 3 Model Estimates of Victimization, Self-Esteem Discrepanciesand Internalizing Problems

Model parameters B SE β

Step 1: Victimization and SED

Autoregressive

Age11 victimization to age12 victimization 0.68 0.06 0.76**

Age11 SEDsize to age12 SEDsize 0.26 0.07 0.25**

Age11 SEDdir to age12 SEDdir 0.13 0.06 0.13*

Age11 S*D to age12 S*D 0.30 0.07 0.28**

Cross-lagged regression paths

Age11 victimization to age12 SEDsize 0.05 0.03 0.13

Age11 victimization to age12 SEDdir 0.04 0.02 0.20*

Age11 victimization to age12 S*D 0.08 0.03 0.22**

Cross-sectional correlations

Age11 victimization with age11 SEDsize 0.40 0.14 0.19**

Age11 victimization with age11 SEDdir 0.29 0.07 0.24**

Age11 victimization with age11 S*D 0.82 0.18 0.39**

Age12 victimization with age12 SEDsize 0.28 0.12 0.23*

Age12 victimization with age12 SEDdir 0.11 0.06 0.16*

Age12 victimization with age12 S*D 0.40 0.11 0.34**

Step 2: SED and internalizing problems

Autoregressive paths

Age11 internalizing to age12 internalizing 0.48 0.17 0.52**

Age11 SEDsize to age12 SEDsize 0.29 0.07 0.28**

Age11 SEDdir to age12 SEDdir 0.17 0.06 0.18**

Age11 S*D to age12 S*D 0.34 0.07 0.34**

Cross-lagged regression paths

Age11 SEDsize to age12 internalizing −0.01 0.02 −0.02Age11 SEDdir to age12 internalizing −0.03 0.05 −0.04Age11 S*D to age12 internalizing 0.10 0.05 0.30*

Cross-sectional correlations

Age11 internalizing with age11 SEDsize 0.08 0.03 0.28

Age11 internalizing with age11 SEDdir 0.05 0.01 0.35

Age11 internalizing with age11 S*D 0.15 0.04 0.54

Age12 internalizing with age12 SEDsize 0.08 0.01 0.48**

Age12 internalizing with age12 SEDdir 0.03 0.01 0.26**

Age12 internalizing with age12 S*D 0.09 0.01 0.55**

SEDsize=size of the self-esteem discrepancies; SEDdir=direction of theself-esteem discrepancies; S*D=the interaction between the size and thedirection of the self-esteem discrepancies; Internalizing=Internalizingproblems; *p<0.05 ** p<0.01

Fig. 1 Prediction by victimization at age 11 of changes in self-esteemdiscrepancies from age 11 to 12 years for children categorized as havingfragile or damaged self-esteem at age 12

Fig. 2 Prediction by self-esteem discrepancies at age 11 of changes ininternalizing problems from age 11 to 12 years for children categorized ashaving fragile or damaged self-esteem at age 11

J Abnorm Child Psychol

We then allowed for the links between age 11/12 self-esteemdiscrepancies and age 12 internalizing problems. Allowing forthese links significantly improved model fit, Δχ2(5) =36.82,p<0.01, and resulted in an acceptable fit to the data (CFI=0.93, RMSEA=0.05). The results are presented in Fig. 3 andshow that the interaction between the size and direction of age11/12 self-esteem discrepancies was associated with the increasein internalizing problems. When allowing for the indirect path,the direct link between age 11 victimization and the developmentof internalizing problems is no longer significant, β=−0.03, p=0.83. We tested for the significance of the indirect path from age11 victimization to age 12 internalizing problems through theinteraction term of size of self-esteem discrepancy X direction ofself-esteem discrepancies (MacKinnon et al. 2002). This indirectpath was significant, B=0.03, SE=0.01, β=0.11, p<0.05. Tobreak down the indirect pathway, we ran amultiple groupmodel,in which we estimated the indirect path from victimization at age11 years to discrepancy between explicit and implicit self-es-teem, to the development of internalizing problems at age12 years for children categorized as having a fragile versusdamaged self-esteem. Holding the paths equal across the twogroups showed a significant poorer fit to the data then allowingthe paths to be different, Δχ2(2) =24.56, p<0.01, indicating thatthis indirect path differed between children with a damagedversus a fragile self-esteem. The indirect pathwaywas significantfor children with a damaged self-esteem, β=0.10, p<0.01, butnot for children with a fragile self-esteem, β=0.00, p=0.75.

Next, this indirect pathway among children with damagedself-esteem was tested for sex differences. To this end, we rana multiple group model for boys versus girls with a damagedself-esteem. A chi-square difference test of sex-differences inthis overall indirect pathway just failed to reach conventionallevels of significance, Δχ2(2) =5.22, p<0.07.

Discussion

The present study examined the influence of self-esteem dis-crepancies on the development of internalizing problems inchildren within the context of victimization by peers. To thisend, we first studied the separate links (hypotheses 1 & 2) thatcomprise the overall indirect effect hypotheses (hypothesis 3)that we had. In support of our first hypothesis, the resultsshowed that victimization was associated with an increase indamaged self-esteem. Considering our second hypothesis, theresults demonstrated that damaged self-esteem was associatedwith the development of internalizing problems while fragileself-esteem was not. Finally, in support of our third hypothe-sis, damaged self-esteem acted as a mediator in the relation-ship between victimization and the development of internal-izing problems in children.

Self-esteem Discrepancies and Internalizing ProblemDevelopment

The result that damaged self-esteemwas related to increases ininternalizing problems is consistent with previous research onthe relationship between damaged self-esteem and depressionwith suicide ideation in adults (Franck et al. 2007), socialanxiety in adult women (De Jong 2002), depressive symp-toms, suicidal ideation and loneliness in young adult women(Creemers et al. 2012), and social anxiety in adolescents(Schreiber et al. 2012). Our study adds to these findings byshowing that damaged self-esteem was associated with in-creases in internalizing problems in elementary school chil-dren. The finding that fragile self-esteem (high explicit andlow implicit) was unrelated to internalizing problem develop-ment was in accordance with previous studies among adoles-cents and adults. This underscores the unique association ofdamaged self-esteem on internalizing problem development,already in childhood. The present results, however, do notaccord with those of De Jong et al. (2012) and Van Tuijl et al.(2014), who found no association between high levels ofimplicit self-esteem combined with low levels of explicitself-esteem and social anxiety and depression in earlyadolescents. Both De Jong et al. (2012) and Van Tuijl et al.(2014) studied samples of early adolescents, during the firstand second year of high school, which is different from thepresent sample of children, who were studied across the last2 years of elementary school. The differences in results maythus stem from the fact that the transition from elementaryschool to high school coincides with changes in social net-works and changes in school environments. Moreover, afterthis transition, more emphasis is placed on evaluation and

Fig. 3 The mediating role of self-esteem discrepancies in the associationbetween victimization and the development of internalizing problems inchildren. SEDsize=size of the self-esteem discrepancies; SEDdir=direc-tion of the self-esteem discrepancies; S*D=the interaction between thesize and the direction of the self-esteem discrepancies; Intern=Internal-izing problems. VIC=Victimization. *p<0.05 ** p<0.01

J Abnorm Child Psychol

performance and children have multiple teachers. Thesechanges have been suggested to influence self-esteem(Wigfield et al. 1991).

Another difference between our study and the studies of DeJong et al. (2012) and Van Tuijl et al. (2014) is the analyticapproach. Previous studies tested for the main effects ofimplicit and explicit self-esteem and the interaction betweenthe two. Although this approach is similar to the approachused in the present study in that they can both draw conclu-sions about congruent and discrepant self-esteem, an impor-tant difference should be mentioned. The presently used di-chotomous direction of self-esteem discrepancies variable(damaged versus fragile self-esteem) categorizes children ina group regardless of whether scores fall on the low or the highend of the discrepancy distribution within that particulargroup. This variable (direction of self-esteem) can be used tostudy if any one direction of self-esteem is associated with ouroutcome variable, internalizing problems development, re-gardless of how extreme the score within the group damagedor fragile is. When using the traditional approach, effects offragile and damaged self-esteem can also be studied by plot-ting the interaction between implicit and explicit self-esteem.However, with this method it remains unclear whether asso-ciations between damaged self-esteem and the outcome vari-able can be applied to all children with a damaged self-esteem.Our results showed that damaged self-esteem in general islinked to increases in internalizing problems, when comparedto fragile self-esteem, and that for children with damaged self-esteem, having larger discrepancies is associated with evenmore increases in internalizing problems.

Victimization, Self-Esteem Discrepanciesand Internalizing Problem Development

Next to finding an association between damaged self-esteemand internalizing problem development, our results also foundsupport for damaged self-esteem acting as a mediator in therelationship between victimization and the development ofinternalizing problems. This study thus provided evidencefor the role of the social context in the relation betweendamaged self-esteem and internalizing problems. Previousresearch showed that low explicit self-esteem mediated(partially) the relationship between social problems and inter-nalizing problems in children (Grills and Ollendick 2002;Ladd and Troop-Gordon 2003; Troop-Gordon and Ladd2005). Our findings are to some extent in line with thesestudies, as damaged self-esteem consists of higher implicitthan explicit self-esteem. However, the present study extendsprevious studies by showing that it is the imbalance betweenimplicit self-esteem and explicit self-esteem. That is, only ifimplicit self-esteem is higher than explicit, resulting in a

damaged self-esteem, does self-esteem mediate the develop-mental link between victimization and internalizing problems.

A theory that has been used to explain the relationshipbetween damaged self-esteem and internalizing problemsfound in studies on adults and adolescents suggests that dam-aged self-esteem represents a discrepancy between individ-uals’ high goals or standards (high implicit self-esteem) andthe reality (low explicit self-esteem) (Franck et al. 2007). Inline with this thought we speculate that children with highimplicit self-esteem who experience victimization during ele-mentary school, can develop low explicit self-esteem. Theirhigh personal standards are not in accordance with theirreality. This damaged self-esteem could then result in thedevelopment of internalizing problems. However, more lon-gitudinal studies in self-esteem discrepancies in children areneeded since adult theories that help explain the relationshipbetween damaged self-esteem and internalizing problemsmaynot be applicable to children.

The mediating role of damaged self-esteem in the linkbetween victimization and internalizing problem developmentwas found to be similar for boys and girls. Although we hadno clear hypothesis on sex differences, it may conflict withone previous study that found that victimization experienceswere linkedwith internalizing problems via (low) explicit self-esteem only among girls (Grills and Ollendick 2002). Itshould be noted, however, that our results on sex differencesfound a trend towards significance. Further longitudinal re-search is therefore necessary to more thoroughly explorepossible sex differences in the mediating role of damagedself-esteem in the link between victimization and the devel-opment of internalizing problems.

Some limitations of the present study should be mentioned.First, our sample was comprised of children with normativelevels of internalizing problems who came from mostly ruralareas. We would like to see our results replicated in morediverse samples, including samples with elevated levels ofinternalizing problems to see if our results generalize tobroader samples and possibly predict clinically elevated levelsof internalizing problems. Second, implicit self-esteem wasmeasured using the BIAT-SE, which was adapted for use inchildren. There is evidence that the BIAT-SE is a reliable andvalid instrument for measuring implicit self-esteem in adults(Sriram and Greenwald 2009). Although the BIAT-SE has toour knowledge never been used in children before, our resultssuggest that the BIAT-SE is a valid instrument to measureimplicit self-esteem in children. Implicit self-esteem is in ourstudy not related to explicit self-esteem, which is in accor-dance with results from studies in adults (e.g., (Bosson et al.2000; Jordan et al. 2003; Zeigler-Hill and Terry 2007).Additionally, results on the association between damagedself-esteem and internalizing problems are in line with previousresearch in adolescents and adults (Creemers et al. 2012; DeJong 2002; Schreiber et al. 2012). A third limitation is that,

J Abnorm Child Psychol

despite the longitudinal design, we only had two waves of data.This prohibited us from estimating a full indirect developmen-tal path of victimization leading to changes in self-esteemdiscrepancies, which in turn predicted changes in internalizingproblems. We did show that damaged self-esteem predictschanges in internalizing problems. However, with the availabil-ity of a third wave of data, the hypothesized indirect pathwaycould have been tested in a more optimal manner.

The finding from this study that self-esteem discrepanciesare linked to the development of internalizing problems inchildren has significant theoretical and practical implications.It underlines the importance of incorporating implicit self-esteem in studies next to explicit self-esteem in order tounderstand the role of self-esteem in the development ofinternalizing problems. Our results add to self-esteem researchby showing that self-esteem discrepancies exist and, moreimportantly, influence the development of internalizing prob-lems, already in late childhood. Additionally, this study dem-onstrates that research on self-esteem discrepancies and inter-nalizing problem development needs to take into account thesocial context in which discrepancies may develop, such asvictimization by peers. An important question remaining ishow implicit and explicit self-esteem develop across child-hood and when self-esteem discrepancies begin to emerge andstart influencing psychopathology development. Further re-search should focus on these issues longitudinally and includethe social context.

The results also have implications for practice.Internalizing problems begin to manifest in late childhood.Damaged self-esteem thus may be a predictor of the onset onthese problems and could consequently enhance early identi-fication of children at risk for developing internalizing prob-lems, which could help to prevent anxiety and depression laterin life. More importantly, this study demonstrates that victim-ization is associated with internalizing problem developmentthrough a difference between explicit and implicit self-esteemand not just through (low levels of) explicit self-esteem. Thissuggests that merely focusing on enhancing children’s explicitself-esteem may not be enough. Additional focus should begiven on why the explicit self-esteem of a child is not congru-ent with its internal, implicit self-perceptions.

Conflict of interest The authors declare that they have no conflict ofinterest.

References

Biebl, S. J., Dilalla, L. F., Davis, E. K., Lynch, K. A., & Shinn, S. O.(2011). Longitudinal associations among peer victimization andphysical and mental health problems. Journal of PediatricPsychology, 36, 868–877.

Boivin, M., Vitaro, F., & Gagnon, C. (1992). A reassessment of the self-perception profile for children: factor structure, reliability, and con-vergent validity of a french version among second through sixthgrade children. International Journal of Behavioral Development,15, 275–290.

Bollen, K. A., & Long, J. S. (1993). Testing structural equation models.Thousand Oaks: Sage.

Bond, L., Carlin, J. B., Thomas, L., Rubin, K., & Patton, G. (2001). Doesbullying cause emotional problems? a prospective study of youngteenagers. British Medical Journal, 323, 480–484.

Bosson, J. K., Swann, W. B., Jr., & Pennebaker, J. W. (2000). Stalking theperfect measure of implicit self-esteem: the blind men and the elephantrevisited? Journal of Personality and Social Psychology, 79, 631–643.

Bosson, J. K., Brown, R. P., & Zeigler-Hill, V. (2003). Self-enhancementtendencies among people with high explicit self-esteem: the moder-ating role of implicit self-esteem. Self and Identity, 2, 169–187.

Briñol, P., Petty, R. E., & Wheeler, S. C. (2006). Discrepancies betweenexplicit and implicit self-concepts: consequences for informationprocessing. Journal of Personality and Social Psychology, 91, 154.

Browne, M. W., & Cudeck, R. (1992). Alternative ways of assessingmodel fit. Sociological Methods & Research, 21, 230–258.

Clark, C., Rodgers, B., Caldwell, T., Power, C., & Stansfeld, S. (2007).Childhood and adulthood psychological ill health as predictors ofmidlife affective and anxiety disorders - The 1958 British BirthCohort. Archives of General Psychiatry, 64, 668–678.

Cole, D. A., Jacquez, F. M., & Maschman, T. L. (2001). Social origins ofdepressive cognitions: a longitudinal study of self-perceived com-petence in children. Cognitive Therapy and Research, 25, 377–395.

Creemers, D. H. M., Scholte, R. H. J., Engels, R. C. M. E., Prinstein, M.J., & Wiers, R. W. (2012). Implicit and explicit self-esteem asconcurrent predictors of suicidal ideation, depressive symptoms,and loneliness. Journal of Behavior Therapy and ExperimentalPsychiatry, 43, 638–646.

Creemers, D. H., Scholte, R. H., Engels, R. C., Prinstein, M. J., & Wiers,R. W. (2013). Damaged self-esteem is associated with internalizingproblems. Frontiers in Psychology, 4, 152.

Crick, N. R., & Bigbee, M. A. (1998). Relational and overt forms of peervictimization: a multiinformant approach. Journal of Consultingand Clinical Psychology, 66, 337–347.

Crick, N. R., & Grotpeter, J. K. (1996). Children’s treatment by peers:victims of relational and overt aggression. Development andPsychopathology, 8, 367–380.

De Jong, P. J. (2002). Implicit self-esteem and social anxiety: differentialself-favouring effects in high and low anxious individuals.Behaviour Research and Therapy, 40, 501–508.

De Jong, P. J., Sportel, B. E., De Hullu, E., & Nauta, M. H. (2012). Co-occurrence of social anxiety and depression symptoms in adoles-cence: differential links with implicit and explicit self-esteem?Psychological Medicine, 42, 475–484.

De Los Reyes, A., & Kazdin, A. E. (2004). Measuring informant dis-crepancies in clinical child research. Psychological Assessment, 16,330–334.

Epstein, S. (1994). Integration of the cognitive and the psychodynamicunconscious. American Psychologist, 49, 709–724.

Fergusson, D. M., Horwood, L. J., & Boden, J. M. (2006). Structure ofinternalising symptoms in early adulthood. British Journal ofPsychiatry, 189, 540–546.

Franck, E., De Raedt, R., Dereu, M., & Van den Abbeele, D. (2007).Implicit and explicit self-esteem in currently depressed individualswith and without suicidal ideation. Journal of Behavior Therapy andExperimental Psychiatry, 38, 75–85.

Goodman, R. (1997). The strengths and difficulties questionnaire: a researchnote. Journal of Child Psychology and Psychiatry, 38, 581–586.

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

J Abnorm Child Psychol

Greenwald, A. G., Nosek, B. A., & Banaji, M. R. (2003). Understandingand using the implicit association test: I an improved scoring algo-rithm. Journal of Personality and Social Psychology, 85, 197–216.

Grills, A. E., & Ollendick, T. H. (2002). Peer victimization, global self-worth, and anxiety in middle school children. Journal of ClinicalChild & Adolescent Psychology, 31, 59–68.

Harter, S. (1982). The perceived competence scale for children. ChildDevelopment, 87–97.

Harter, S. (1993). Causes and consequences of low self-esteem in childrenand adolescents. In R. F. Baumeister (Ed.), Self-esteem: The puzzleof low self-regard (pp. 87–116). New York: Plenum Press.

Hawker, D. S., & Boulton, M. J. (2000). Twenty years’ research on peervictimization and psychosocial maladjustment: A meta-analytic re-view of cross-sectional studies. Journal of Child Psychology andPsychiatry, 41, 441–455.

Jordan, C. H., Spencer, S. J., Zanna, M. P., Hoshino-Browne, E., &Correll, J. (2003). Secure and defensive high self-esteem. Journalof Personality and Social Psychology, 85, 969–978.

Kling, K. C., Hyde, J. S., Showers, C. J., & Buswell, B. N. (1999).Gender differences in self-esteem: a meta-analysis. PsychologicalBulletin, 125, 470–500.

Kraemer, H. C., Yesavage, J. A., Taylor, J. L., & Kupfer, D. (2000). Howcan we learn about developmental processes from cross-sectionalstudies, or can we? The American Journal of Psychiatry, 157, 163–171.

Ladd, G. W., & Troop-Gordon, W. (2003). The role of chronic peerdifficulties in the development of children’s psychological adjust-ment problems. Child Development, 74, 1344–1367.

Leary, M. R., & Baumeister, R. F. (2000). The nature and function of self-esteem: Sociometer theory. Advances in Experimental SocialPsychology, 32, 1–62.

Little, T. D., Bovaird, J. A., & Widaman, K. F. (2006). On the merits oforthogonalizing powered and product terms: Implications formodeling interactions among latent variables. Structural EquationModeling-a Multidisciplinary Journal, 13, 497–519.

MacKinnon,D. P., Lockwood, C.M.,Hoffman, J.M.,West, S. G., &Sheets,V. (2002). A comparison of methods to test mediation and otherintervening variable effects. Psychological Methods, 7, 83–104.

Mann, M., Hosman, C.M. H., Schaalma, H. P., & DeVries, N. K. (2004).Self-esteem in a broad-spectrum approach for mental health promo-tion. Health Education Research, 19, 357–372.

Masten, A. S., & Cicchetti, D. (2010). Developmental cascades.Developmental Psychopathology, 22, 491–495.

Muris, P., Meesters, C., & Fijen, P. (2003). The self-perception profile forchildren: further evidence for its factor structure, reliability, andvalidity. Personality and Individual Differences, 35, 1791–1802.

Muthén, L., & Muthén, B. (2012).Mplus. The comprehensive modellingprogram for applied researchers: User’s guide, 5. Los Angeles:Muthén & Muthén.

Ohannessian, C. M., Lerner, R. M., Lerner, J. V., & von Eye, A. (1999).Does self-competence predict gender differences in adolescent de-pression and anxiety? Journal of Adolescence, 22, 397–411.

Reijntjes, A., Kamphuis, J. H., Prinzie, P., & Telch, M. J. (2010). Peervictimization and internalizing problems in children: a meta-analysisof longitudinal studies. Child Abuse and Neglect, 34, 244–252.

Robins, R. W., & Trzesniewski, K. H. (2005). Self-esteem developmentacross the lifespan.Current Directions in Psychological Science, 14,158–162.

Roza, S. J., Hofstra, M. B., van der Ende, J., & Verhulst, F. C. (2003).Stable prediction of mood and anxiety disorders based on behavioraland emotional problems in childhood: A 14-year follow-up duringchildhood, adolescence, and young adulthood. American Journal ofPsychiatry, 160, 2116–2121.

Sandstrom, M. J., & Jordan, R. (2008). Defensive self-esteem and ag-gression in childhood. Journal of Research in Personality, 42, 506–514.

Schreiber, F., Bohn, C., Aderka, I. M., Stangier, U., & Steil, R. (2012).Discrepancies between implicit and explicit self-esteem among ad-olescents with social anxiety disorder. Journal of Behavior Therapyand Experimental Psychiatry, 43, 1074–1081.

Schröder-Abé,M., Rudolph, A., & Schütz, A. (2007a). High implicit self-esteem is not necessarily advantageous: discrepancies between ex-plicit and implicit self-esteem and their relationship with angerexpression and psychological health. European Journal ofPersonality, 21, 319–339.

Schröder-Abé, M., Rudolph, A.,Wiesner, A., & Schütz, A. (2007b). Self-esteem discrepancies and defensive reactions to social feedback.International Journal of Psychology, 42, 174–183.

Sowislo, J. F., &Orth, U. (2013). Does low self-esteem predict depressionand anxiety? A meta-analysis of longitudinal studies. PsychologicalBulletin, 139, 213–240.

Sriram, N., & Greenwald, A. G. (2009). The brief implicit associationtest. Experimental Psychology, 56, 283–294.

Steinmetz, H., Davidov, E., & Schmidt, P. (2011). Three approaches toestimate latent interaction effects: Intention and perceived behavior-al control in the theory of planned behavior. MethodologicalInnovations Online, 6, 95–110.

Strack, F., & Deutsch, R. (2004). Reflective and impulsive determinantsof social behavior. Personality and Social Psychology Review, 8,220–247.

Troop-Gordon, W., & Ladd, G. W. (2005). Trajectories of peer victimi-zation and perceptions of the self and schoolmates: precursors tointernalizing and externalizing problems. Child Development, 76,1072–1091.

Van Tuijl, L. A., de Jong, P. J., Sportel, B. E., deHullu, E., &Nauta,M.H.(2014). Implicit and explicit self-esteem and their reciprocal rela-tionship with symptoms of depression and social anxiety: a longi-tudinal study in adolescents. Journal of Behavior Therapy andExperimental Psychiatry, 45, 113–121.

Veenstra, R., Lindenberg, S., Oldehinkel, A. J., De Winter, A. F.,Verhulst, F. C., & Ormel, J. (2005). Bullying and victimization inelementary schools: a comparison of bullies, victims, bully/victims,and uninvolved preadolescents. Developmental Psychology, 41,672–682.

Wigfield, A., Eccles, J. S., Mac Iver, D., Reuman, D. A., & Midgley, C.(1991). Transitions during early adolescence: Changes in children’sdomain-specific self-perceptions and general self-esteem across thetransition to junior high school.Developmental Psychology, 27, 552.

Williams, R. L. (2000). A note on robust variance estimation for cluster-correlated data. Biometrics, 56, 645–646.

Zeigler-Hill, V. (2006). Discrepancies between implicit and explicit self-esteem: implications for narcissism and self-esteem instability.Journal of Personality, 74, 119–144.

Zeigler-Hill, V., & Terry, V. C. (2007). Perfectionism and explicit self-esteem: themoderating role of implicit self-esteem. Self and Identity,6, 137–153.

J Abnorm Child Psychol