instrumental or emotional aggression: testing models of bullying

13
Instrumental or Emotional Aggression: Testing Models of Bullying, Victimization, and Psychological Maladjustment among Taiwanese Seventh-Graders Hsi'sheng Wei and James Herbert Williams This study examined the relationship of instrumental and emotional aggression to bullying, victimization, and psychosocial maladjustment. It was hypothesized that both types of aggression would be associated with bullying behavior and that emotional aggression would be exclusively associated with risk of victimization and psychological maladjustment (that is, depression, anxiety, and loneliness).The sample consisted of 219 Taiwanese seventh-graders with valid data on all of the research variables; 51.1% (« =112) were male, and 48.9% (M = 107) were female.A series of structural equation models was analyzed to evaluate fit indices for competing models.Tht- results indicated that both instrumental aggression and emotional aggression were associated with bullying, but only the latter was associated with victimization. Once psychological maladjustment was entered nito the model, the association between emotional .iggrcssion and bullying became nonstgiiil-icant. Model indices also suggested that psychological maladjustment was a concurrent characteristic rather than a consequence of peer victimization, Imptications for fiiture invcsrigation are discussed. KEY WORDS: adolescents; ajigression; bullying; psychological maladjustmeui; mäimization R esearch on school bullying has increased rapidly over the past few decades. It is now well established that such aggression is prevalent internationally, with considerable num- bers of youths affected. In a recent study of 2,086 (¡erman students, Scheithauer, Hayer, Petermann, and Jugcrt (2006) found that 12.1% of respondents bullied others, and 11.1% reported being victimized. In n survey on 1,344 fourth-grade primary school children in South Korea.Yang. Kim, Kim. Shin, and Yoon (2006) revealed a similar picture, with 12.0% of students bullying others, 5.3% reporting being victimized, and 7.2% reporting both bullying and victimization. Similar levels of bullying have been identified in the United Kingdom. Australia. Ireland, Norway, the United States, and Japan (Nansel ct al., 2(K)1; Olweus, 1992; A. M. O'Moore, Kirkham, & Smith. 1997;Rigby&- Slee, 199];Takahiro & Iwao, 2000; Whitney & Smith. 1993). Experiences of be- ing bullied have been found to be associated with various psychological problems such as depression, loneliness, anxiety, and low self-esteem (Hawker & Boulton, 2000; Storch & Ledley, 2005). Victims may also exhibit psychosomatic symptoms, suicidal ideation, and reduced school attachment (Ivars- son. Broberg, Arvidsson, &: Gillberg. 2005; Natvig. Albrektsen.& Qvarnstrom, 20ai;Wei & Williams, 2004). Such adverse effects have raised serious hu- man rights concerns, and efforts are underway to identify the risk factors of bullying for prevention and intervention (Greene. 2006). Previous saidies have identified distinctive aggres- sive patterns in bullying, (Carney & Merrell, 2001; Haynie et al. 20f)l ; Olweus, 1997; M. O'Moore 6c Kirkham, 2001; Schwartz, 2000).The majority of bullies were characterized by an overt aggressive- ness, a need to dominate, little empathy for othern, and positive attitudes toward violence (Carney &' MerreU. 2001 ; Olweus. 1997). Despite such antisocial tendencies, such youngsters do not necessarily have salient psychological maladjustments. Although some researchers have found bullies to have lower self-esteem (Austin & Joseph. 1996), others have found that bullying has little association with self- esteem, anxict); or loneliness (Kaukiainen et al., 2002; M. O'Moore & Kirkham, 2001; Nansel et al., 2001; RJgby & Slee, 1993). instead, children who engage in bullying have been seen to be at higher CCC Code: 1070-5303/09 t3.00 O2009 National Aiiociation of Sixial Woikers 231

Upload: phungtruc

Post on 06-Feb-2017

227 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: Instrumental or Emotional Aggression: Testing Models of Bullying

Instrumental or Emotional Aggression:Testing Models of Bullying, Victimization,and Psychological Maladjustment among

Taiwanese Seventh-GradersHsi'sheng Wei and James Herbert Williams

This study examined the relationship of instrumental and emotional aggression to bullying,victimization, and psychosocial maladjustment. It was hypothesized that both types ofaggression would be associated with bullying behavior and that emotional aggression wouldbe exclusively associated with risk of victimization and psychological maladjustment (that is,depression, anxiety, and loneliness).The sample consisted of 219 Taiwanese seventh-graderswith valid data on all of the research variables; 51.1% (« =112) were male, and 48.9% (M =107) were female.A series of structural equation models was analyzed to evaluate fit indicesfor competing models.Tht- results indicated that both instrumental aggression and emotionalaggression were associated with bullying, but only the latter was associated with victimization.Once psychological maladjustment was entered nito the model, the association betweenemotional .iggrcssion and bullying became nonstgiiil-icant. Model indices also suggested thatpsychological maladjustment was a concurrent characteristic rather than a consequence ofpeer victimization, Imptications for fiiture invcsrigation are discussed.

KEY WORDS: adolescents; ajigression; bullying; psychological maladjustmeui; mäimization

Research on school bullying has increasedrapidly over the past few decades. It isnow well established that such aggression

is prevalent internationally, with considerable num-bers of youths affected. In a recent study of 2,086(¡erman students, Scheithauer, Hayer, Petermann,and Jugcrt (2006) found that 12.1% of respondentsbullied others, and 11.1% reported being victimized.In n survey on 1,344 fourth-grade primary schoolchildren in South Korea.Yang. Kim, Kim. Shin, andYoon (2006) revealed a similar picture, with 12.0%of students bullying others, 5.3% reporting beingvictimized, and 7.2% reporting both bullying andvictimization. Similar levels of bullying have beenidentified in the United Kingdom. Australia. Ireland,Norway, the United States, and Japan (Nansel ct al.,2(K)1; Olweus, 1992; A. M. O'Moore, Kirkham, &Smith. 1997;Rigby&- Slee, 199];Takahiro & Iwao,2000; Whitney & Smith. 1993). Experiences of be-ing bullied have been found to be associated withvarious psychological problems such as depression,loneliness, anxiety, and low self-esteem (Hawker& Boulton, 2000; Storch & Ledley, 2005). Victimsmay also exhibit psychosomatic symptoms, suicidal

ideation, and reduced school attachment (Ivars-son. Broberg, Arvidsson, &: Gillberg. 2005; Natvig.Albrektsen.& Qvarnstrom, 20ai;Wei & Williams,2004). Such adverse effects have raised serious hu-man rights concerns, and efforts are underway toidentify the risk factors of bullying for preventionand intervention (Greene. 2006).

Previous saidies have identified distinctive aggres-sive patterns in bullying, (Carney & Merrell, 2001;Haynie et al. 20f)l ; Olweus, 1997; M. O'Moore 6cKirkham, 2001; Schwartz, 2000).The majority ofbullies were characterized by an overt aggressive-ness, a need to dominate, little empathy for othern,and positive attitudes toward violence (Carney &'MerreU. 2001 ; Olweus. 1997). Despite such antisocialtendencies, such youngsters do not necessarily havesalient psychological maladjustments. Althoughsome researchers have found bullies to have lowerself-esteem (Austin & Joseph. 1996), others havefound that bullying has little association with self-esteem, anxict); or loneliness (Kaukiainen et al.,2002; M. O'Moore & Kirkham, 2001; Nansel et al.,2001; RJgby & Slee, 1993). instead, children whoengage in bullying have been seen to be at higher

CCC Code: 1070-5303/09 t3.00 O2009 National Aiiociation of Sixial Woikers 231

Page 2: Instrumental or Emotional Aggression: Testing Models of Bullying

risk of aggression, delinquency, and externalizingproblems (Ivarsson et al., 2005; Kumpulainen &Rasanen. 2000;Veenstra et al., 2005).

A relatively small group of youths—namely,bully/victims or aggressive victims—manifest quitedifferent patterns of beliavior and personal char-acteristics. They suffer frequent peer victimizationwhile behaving aggressively or engaging in bully-ing acts themselves. Compared with pure bullies,bully/victims engage in more physical bullying andless verbal bullying, and they are more likely to bephysically victimized than are pure victims (Unnever,2005). Bully/victim boys have reported a high levelof street violence involvement and a high frequencyof victimization on the street (Andershed. Kerr, &Stattin. 2001). Generally speaking, individuals whoscore high on both bullying and victimization showsignificant psychological maladjustment and oftenhave the poorest outcomes when compared withnonvictimized bullies, nonaggressive victims, andnoninvolved peers (Hayniectal., 200 l;M,O'Moore& Kirkham. 2001; Schwartz, 2000).They are likelyto experience a variety of issues such as academicfailure,lowerglobal self-esteem.peer rejection,emo-tional distress,and depression (Haynie et al.,2001;M.O'Moore & Kirkham,2001 ¡Schwartz,2000).

ETIOLOGIES OF BULLYINGOne prominent theory relevant to school bullyingis the social information processing model (Crick &Dodge. 1994). It postulates that social behaviors ofchildren, hoth prosocial and aggressive, are productsof children's social cognitions.The model identifiessix basic steps underpinning children's social cogni-tion processes: encoding of social cues, interpretationof social cues, goal selection, response formulation,evaluation,and action (Crick & Dodge, 1994;Dodge,1991 ). 11 is hypothesized that aggressive children havedeficits at various poincs along those processijig steps.Crick and Dodge (1996) have identified two distincttypologies of aggressive children.The first typology,reiKtifc—a^'^ressiui\ tends to attribute hostile intent topeers, who reactive-aggressive children perceive asmean and threatening to the self. When presentedwith ambiguous cues, reactive—aggressive youths aremon.- likely to interpret them as provocative (Crick &Dodge, 1996; Dodge & Coie, 1987).These childrenmanifest angry and defensive responses to fi-ustrationor provocation.With the second typo]ogy,proactii'e-aggressivc,youù\s tend to initiate aggressive behavioras a deliberate strategy to obtain their goals. They

evaluate aggressive behavior and its expected out-comes more positively, emphasize instrumental goalsover interpersonal harmony, and regard aggressionas an effective way to get rewards or solve conflictsin social situations (Crick & Dodge, 1996).

The original propositions of the social infor-mation processing model imply that aggressionresults from cognitive deficits. However, Suttonand Keogh (2000) showed that ringleader bulliesscored significantly higher than other groups ofchildren on assessment of another's cognitive andemotional states. Wliat these children lack is not acorrect theory of the other's mind but empathy forthe other's feelings. Bullies were also found to havehigher Machiavellian ist beliefs (Sutton & Keogh.2000). Hawley (2006) highlighted the features ofMachiavellianism, such as manipulation and lackof empathy. From an evolutionary perspective, ithas been suggested that aggression coupled withpositive skills and tendencies may increase one'schances of obtaining social and material rewards(Hawley, 2007). In line with this adaptive view ofagression, some researchers regard bullying as asophisticated strategy for interpersonal manipulationand have suggested that bulhes may well understandthe psychological influence of their acts on victimsand use this knowledge to pursue their self-interestsand their positive outcome expectancy—thus, theself-efficacy of aggression should not be interpretedsimply as a cognitive error (Sutton, 2001 ; Sutton.Smith, & Swettenham, 1999a, 1999b. 1999c). Bul-lying children can thus be identified as proactivclyaggressive, acting as indifferent, manipulating indi-viduals seeking material and social rewards at theexpense of their victims.

INSTRUMENTAL VERSUS EMOTIONALAGGRESSIONThe typologies of proactive and reactive aggressivebullies actually reflect a long-standing distinctionbetween instrumental and emotional aggressionin psychology. Instrumental (premeditated or pro-active) aggression refers to a calculated behavioraimed at obtaining rewards and achieving personalgoals. Emotional (hostile or reactive) aggression isimpulsive and retaliatory and can be regarded as areaction with anger (Aronson. ]992;Hartup, 1974).In fact, Ramirez and Andreu (2006) found angerand impulsiveness to be positively correlated withhostile aggression but not with instrumental aj^res-sion. Much effort has been made to investigate the

232 Social WoriResearch VOLUME 33, NUMBER 4 DECEMBER 1009

Page 3: Instrumental or Emotional Aggression: Testing Models of Bullying

evolutionary significance of emotional aggressionas a primitive survival behavior often seen in ani-mals. Specific neurobiological pathways have beenproposed to understand such an affective response(see, for example, Davidson,Jackson, & Kalin, 2000;Lel!)oux, 1996). At the same time, tvt-o types ofaggression have been found to be associated withdifferent kinds of psychological maladjustment.One recent study with a sample of 211 youngadults found reactive aggression to be stronglyrehited to neuroticism and proactive aggression tobe strongly predictive of externalizing behaviors(Miller & Lynam, 2006). Another study foundproactive aggression to be uniquely characterized;it age 16 by a psychopathic personality, bluntedaffect, delinquency, and serious violent offending,whereas reactive aggression was characterized byimpuisivity, hostility, social anxiety, lack of closefriends, unusual perceptual experiences, and ideasof reference (Raine et al., 2006). Some researchershave recently criticized the distinction betweenthe two types of aggression because, in reality, ag-gressive acts often involve multiple aggressivenesssimultaneously (for example, Bushman & Anderson,2001; Miller & Lynani, 2006), but this dichotomyis still vndely used in bullying, peer victimization,and agression research, and it served as the majorresearch construct for rhe present study.

RESEARCH QUESTIONSThe foregoing discussion highlighted the hetero-geneity of bullying behaviors. Some children andadolescents who are Machiavellianists may use theminor aggression of bullying for the acquisition ofmaterial and social rewards, whereas other bullies'agression is more reactive in nature and tends toresult from emotional regulation issues and hostileattributions of others' social cues. Recognition ofthese typologies of bullying is crucial because theyhave significant diversity and may respond to differ-ent intervention strategies. Therefore, the primaryfocus of this study was the relationships betweeninstrumental and emotional aggression and bullyingbehavior,This study used the social informationalprocessing model (Dodge & Coie, 1987), whichposits that both instrumental and emotional aggres-sion may be associated with bullying. Competingmodels of aggression and bullying were examinedto determine the relative validity of the theory.

The secondary focus of the study concerns therelationships between bullying, victimization, and

psychological maladjustment. Previous researchindicates an association between peer victimiza-tion and psychological problems such as depression,loneliness, and anxiety (Hawker & Boulton, 2000),which suggests that being bullied leads to psycho-logical maladjustment. However,empirical fmdingsregarding the psychological adjustment of bullieshave been mixed.Thus, we did not assume a directpositive relationship between bullying and psycho-logical maladjustment, Instead, we hypothesized thatemotional aggression (not instrumental aggression)is associated with victimization, which in turn isrelated to psychological maladjustment.

METHOD

SampleThis study used data from the Middle School LifeExperiences Project, which is a large-scale surveyconducted in Taipei to collect data on the schooladjustment of Taiwanese youths. It used a two-stageclustered sampling procedure. Middle schools inTaipei were first randomly selected and invited toparticipate. One seventh-grade classroom in eachparticipant school was then selected, which resultedin a sample of seven classes. Almost all of the stu-dents in the selected classes filled out questionnaires(participation rate over 99%), but only students withcomplete data on all research variables were includedin the structural equation modeling (SEM) analysis.Among the 2-38 respondents, 19 had missing data,and the fmal sample consists of 219 students; 51.1%((I =112) were male, and 48.9% (H = 107) werefemale. The average age of the sample %vas 12.80years, with a standard deviation of 0.63.

MeasuresThe survey instrument included self-report, peer-rating, and teacher-report measures that assessed themajor aspects of school iife.The battery for studentswas divided into two questionnaires, each of whichtook 30 minutes to complete. These were groupadministered during two independent study classes.Teacher-report questionnaires were given to thehomeroom teachers and later collected. For certainscales in the battery- that were translated from foreignlanguages, an expert review and a pilot evaluationusing middle school respondents were conducted toensure their validity. Only those measures relevantto this study are listed here.

Children's Loneliness Scale. This scale (a = .92)is a self-report questiormaire assessing children's

WEI AND WILLIAMS / IhtingModels of Bulfying, Victimization, andP^hohgkalMaladjustmevtamong TaiwaneseSevenäi-Graden 233

Page 4: Instrumental or Emotional Aggression: Testing Models of Bullying

feeling of loneliness and social dissatisfaction (Asher,Hyman, & Rcnshaw, 1^84). It consists of 24 items,including eight masking items to assist respondentsin being more open and honest with their responses.Children rate themselves on a five-point scale.

Children's Machiavellianism Scale. This scale (a= .69) is a self-report measure assessing chiidrensinterpersonal strategies and their beliefs ahoLit themanipulability of other people (Christie & Geis,1970). It consists of 20 items, with a four-pointresponse scale ranging from "agree very much" to"disagree very much."

Brief Symptom Rating Scale. This scale is a Chi-nese questionnaire assessing student's mental healthproblems (Lee,Lee,Yen,Lin,&; Lue. 199()).The entirescale consists of 50 items on which respondents ratetheir status over the past week using a five-poinc scaleranging from "never" to "very severe." An exampleitem from this scale is "have thoughts of suicide."Iteins are grouped into multiple subscales, with thefollowing subscales used in this study: Depression (a- .90), Anxiety (« = .87). and Hostility (a = .HI).

Reactive and Proactive Aggressiveness Scale. Thisscale is a self-report measure assessing students' ten-dency to manifest reactive and proactive aggressivebehavior (Roland & Idsoe, 2001).The original scaleconsists of 14 items making up three subscales: Re-active Agression, Power-related Proactive Aggres-sion, and Atfiliation-related Proactive Aggression.An example item from this scale is "If I don't getmy way, ! get angry." Students rate themselves ona four-point scale (NO, tio, yes, YES). In the pres-ent study, rhe latter two subscales were combinedinto a single measure of proacdve aggression. TheCronbach alphas were .78 for Proactive Agressionand .69 for Reactive Aggression.

Adolescent Social Behamor Scale. Tliis scale is aChinese teacher-report measure that assesses a widerange of adolescents'behaviors in school (Hung,2000).The scale consists of 119 items grouped into14 subscales. Teachers rate students using a five-point Likert-type scale ranging from "never" = 1to "always" = 5 for each item. The subscales usedin this study were Impulsivity (a = .91) and Lackof (^)ncern for ilthers (a = ,93),

Self-reported Frequency of Being Butlieä byClassmates. Students were provided with a classroster and asked to rate each of their classmates fortheir frequency of bullying behaviors toward therater using a five-point scale. Bullying was definedas specific physical and verbal behaviors. Exainples of

physical bullying included hitting, kicking,pushing,and tripping. Students' ratings of their classmates onphysical and verbal bullying items were combinedand divided by the number of classmates to createan overall victimization score.

Peer Nomination of Victims. Students were pm-vided with a defmition of bullying and a roster ofclassmates.They were asked to nominate up to sixclassmates who were most frequently bullied. Thedefinition emphasized the specific behaviors con-sidered to be bullying.Each respondents peer-ratedvictim status was calculated by adding the nomina-tions he or she received in the class and dividing bythe number of nominators. Similar methods havebeen used in previous studies (BouIton.Trueman,Chau.Whitehand, & Amatya, 1999).

Peer-rated Bullying Behaviors. Students wereprovided with a roster of classmates and beiiav-ioral descriptors that illustrated bullying behaviors.Each snident rated his or her classmates on theirdegree of manifesting these behaviors toward therater using a five-point Likert-type scale. Students'bullying scores on the dimensions were based onthe average ratings each student received from hisor her classmates. This measure was adapted fromthe Peer Estimated Conflict Behavior (PECOBE)questionnaire (Bjorkqvist Ik Osternian, 1998).Theoriginal five-point PECOBE assesses cla.ssmates'typical interaction styles with regard to three typesof peer aggression along with other attributes. Inthe present study, every respondent was asked toreflect on the target person s hehaviors toward therater specifically.

Data AnalysesSEM was used as the primary analysis strategy forthis study. SEM is an extension of the general linearmodel, which is frequently used for theory testingand confirmatory analysis. SEM was adapted a,s themajor analysis strategy in the present study becauseof tbe multiple alternative models with multipleindictors that were constructed and compared fortheory testing. Analyses were conducted in severalstages. Initially, we performed univariate analysisfor each of the observed variables to assess possibleviolations of SEM assumptions. Data transforma-tions were conducted for distributions violating theassumptions of normality. Second, we performedbivariate analysis to obtain correlation coefficientsfor all major observed variables in preparation forthe SEM analyses. Listwise deletion was used to

Social Work Research VOLUME 33, NUMBER 4 DECEMBER 1009

Page 5: Instrumental or Emotional Aggression: Testing Models of Bullying

exclude cases without complete data on aU variables.Inspection of the univariate statistics for the observedvariables showed chat self-reported within-classphysical and verbal victimization was consider-ably skewed in a positive direction.To reduce thisviolation of normality, we conducted a logarithmictransformation (base 10) for these variables. Theskewness and kiirtosis of self-reported physical andverbal victimization were considerably reduced andwere close to the +3 to -3 range appropriate forSEM (Kline, 1998).

Specific observed variables were selected toconstruct measurement models for the five latentvariables (instrumental aggression, emotional ag-gression, bullying, victimization, and psychologicalmaladjustment). In the first stage of the muUivariateanalysis, measurement models with two or moreobserved variables were tested for model fit usingLISREL 8.50 ([(ireskog&Sorbom,2001).Observedvariables with low reliability and validity (as evi-denced by their association with the latent variable)were dropped. After examining the measurementmodels, we constructed and tested structural modelswith the relationships between the latent variablesspecifled.We built and tested alternative models withthe same set of variables against the proposed modelsto determine their relative validity. The maximumlikelihood method was used to estimate param-eters (Schumacker Sc Lomax, 1996). Modificationindices were also examined, and paths suggested bymodification indices that increased model fit wereadded where this was theoretically or empiricallysupported (Kline, 1998).

RESULTS

Measurement Models of Instrumental andEmotional AggressionIn the initial stage ofnnalysis, a measurement modelwith instrumental and emotional aggression as latentvariables with observed variables was specified andtested.The examination of this measurement modelfulfilled two major objectives, First,it was conceptu-ally nieaiiirigfui to distinguish between instrumentaland emotional aggression as latent variables by evalu-ating the supporting empirical evidence. Second, itwas important to determine whether the selectedobserved variables were adequate indicators forthe two latent variables. Model fit, reliability, andvalidity oftiie observed variables were assessed, andobserved variables with low reliability or validitywere removed from the model.

On the basis of previous research, three applicableobserved variables were selected for each of thetwo latent variables (Mjchiavellianism; proactiveaggressiveness and lack of concern for instrumentalaggression; impuisivity, reactive aggressiveness, andhostility for emotional aggression). In this initialmodel, the error variances of lack of concern andimpuisivity were allowed to correlate because bothmeasures were collected from die same informant(respondent's teacher) using subscales from the sameinstrument (the Adolescent Social liebavior Scale).The error variances of proactive aggressiveness andreactive a^essiveness were also allowed to correlatefor similar reasons.

The results of model testing showed that the datawere a good fit [x'(6, N = 219) = 12.23,p = .06,root mean square error ofapproximation (RMSEA)= .07, normed fit index (NFI) = .95, comparativefit index (CFl) = .97, adjusted goodness of fit in-dex (AGFI) = .94]. An alternative model with thesix observed variables, measuring only tlie latentvariable of aggression, was specified and tested forcomparison.The model test results showed a lowerfit ¡x-(7. N= 219) =31.28,p= .0O,RMSEA = .12,NFI = .88,CFI = .90, AGFI = .87],which supportsthe strength of a two-factor model over a one-factormodel in understanding aggression.

Inspection of the observed variables revealedthat the lack of concern for others and impuisivitywere weak indictors and had only nonsignificantassociations with the latent variables (ra = .11 and.O7.respectively).These two observed variables wereconsequently removed, and a respecified model withfour indictors was analyzed. Model fit indices werenot computed during this stage of analysis becauseof insufficient degrees of freedom.The four observedvariables were found to be adequate indictors of thelatent constructs, with rvalues ranging from .49 to.77 (see Figure 1).

Instrumental and Emotional Aggression,Bullying, and VictimizationOne primary reseaii h question of the present studyfocused on the relationships among instrumentaland emotional aggression, bullying, and victim-ization. Instrumental and emotional aggressionwas hypothesized to be positively associated withbullying, but only emotional aggression was hy-pothesized to have a positive association with peervictimization. An extended model with instrumentaland emotional aggression and two additional latent

ANI> WILLIAMS / Thting Models of Bullying, Victimization, and PsycMopcalMaLidjustment among Taiwanese Senenth-Gmden 2 3 5

Page 6: Instrumental or Emotional Aggression: Testing Models of Bullying

Figure 1: Final Measurement Model of Instrumental and Emotional Aggression

.41 Machiavellianism

Proactive Aggressiveness

lleactive A^ressivencss

.76 Hosiiliiv

Instrumental \ —Aggression

1.00

.55*

.77*

1.00 /

^ ^ ^

Note: Paih toetflclBnti ar> standarltizeti.•9 < .OS

variables (bullying tendency and peer victimization)was specified and tested for model fit. The resultsshowed an adequate fit [X"(22, N = 219) = 43.13.p = .01, RMSEA = .07, NFI = .92, CFI - .96,AGFI = .91]. However, further inspection of thecomponents of the measurement model revealedthat both peer-rated physical (r = .93) and verbalbullying (r = .77) had good reliability for the latentvariable buUying.Tbe measurement model for peervictimization indicated that peer-rated victim statuswas poorly a.ssociated with the latent variable (r =.09).The other two observed variables,self-reportedphysical victimization (r = .76) and verbal victimiza-

tion (; - .94), were more reliable measures for peervictimization. Therefore, peer-rated victim statuswas removed from the next iteration of the model.The modified model was tested, and it showed anadequate fit Ix'(15, N = 219) = 36.72, ;j = .00,RMSEA = .OS. NFI - .94. CFI = .96. AGFI =.90) (see Figure 2).

Further analysis was conducted to examine the fitof other alternative models.Two alternative modelswere constructed using the same set of variables. Inthe first alternative model, we specified a direct pathfrom instrumental aggression to peer victimizationto assess the possibility that, in addition to emotional

Figure 2: Model of Aggression, Bullying, and Victimization

.70

Emotiana]Aggression I An*

.58

Physical[iullying

Verbal Bullying

PhysicalVictimization

VerbalVictim ¡Million

-«-.14

-«-.40

-*-.39

- ^ . 1 6

Nole; P«h coefficients are s•p < .05

236 Social Work Research VOLUME 33, NUMBER 4 DECEMBER 1009

Page 7: Instrumental or Emotional Aggression: Testing Models of Bullying

aggression, instrumental aggression could also be as-sociated with peer victimization.The improvementill model fit for this alternative model was marginalcompared with that of the previous model fx^(14,N = 219) = 33.30. p = .00. RMSEA = 0.08, NFI= 0.94, CFI = 0.97, AGFI = 0.91], It is importantto note that the correlation between instrumentalaggression and peer victimization was negative (r= -.27) nnd nonsignificant, which supports theoriginal hypothesis that emotional aggression isassociated with peer victimization.

A second alternative model wa.s specified. In thismodel, we specified a path from bullying to victim-ization to examine the direct association betweenthese latent variables above and beyond the effectsof instrumental and emotional aggression on peervictimization.The results showed no improvementin model fit [x'(14, N = 219) = 36.42,;) = .00,RMSEA = .09, NFI = .94, CFI = .96, AGFI =.90]. Ill addition, the direct association betweenbullying and victimization was small in magnitude(r = .05) and nonsignificant, which suggests thatbullying behavior does not necessarily increase one'srisk of being bullied and that the specific type ofaggression underlying one s bullying behavior maybe more important.

Peer Victimization and PsychologicalMaladjustmentThe fuul model investigated the relationship

between psychological maladjustment and re-

spondents' victimization. A positive associationwas hypothesized between peer victimization andpsychological maladjustment. A latent variable ofpsychological maladjustment was constructed usingthree observed variables: depression, anxiety.', andloneliness. Psychological maladjustment was addedto the prior model, and a path from peer victimiza-tion to psychological maladjustment was specified,asdepicted in Figure 3.The new model was analyzedfor fit. The results for the measurement model ofpsychologic.il maladjustment showed that the threevariables—depression (r= .98),an.\'iety (r= .85),andloneliness (r = .46)—were positively correlated tothe latent variable. This result suggests that depres-sion and anxiety could be viewed as a highly reliableindicator of psychological maladjustment.The reli-abihty of loneliness as an indicator of psychologicalmaladjustment was somewhat low.

The fit indices of this model showed a less thansatisfactory fit [x'(38, N = 219) = 210.57, p = .00.RMSEA = .14, NFI = .77, CFI = .79, AGFI =.74] (see Figure 3). The association between peervictimization and psychological maladjustment wasestimated at .40. Modification indices suggestedchat adding a path fnim emotional aggression topsychological maladjustment could significantlyincrease model fit. This relationship is theoreti-cally relevant because psychosocial problem.s areamong the major features of emotionally aggressiveyouths.The path was added accordingly.The resultsshowed adequate fit IX'ÇM, N = 219) = 84.41,/)

Figure 3: Initial Model of Aggression, Bullying, Victimization, and Maladjustment

.72 q t * V l'liysi>.iil Bullyi:ii;;

Eniotioiial \y ^ ^ VittimizaiiAggression ¡M* AM >; iftv

.49

Note: P«th coafficlHiti are•p < .05

W R I AND W I L L I A M S / Testing Models ofBtdlying, Victimization, an aladjuslmens among Taiwartese Seventh-Gmdtrs 2 3 7

Page 8: Instrumental or Emotional Aggression: Testing Models of Bullying

= .00. RMSEA = 0.08, NFI = 0.93, CFI = 0.96,AGFI = 0.88]. However, the association betweenpeer victimization and psychological maladjust-ment was estimated as nonsignificant (.02), whereasthe association between emotional aggression andpsychological maladjustment was estimated at .92.This result strongly suggests that the respondents'psychological maladjustment was largely explainedby their emotional aj^ression tendencies mther thantheir peer victimization experiences. An alternativemodel was respecified and tested with a path fromemotional aggression to psychological maladjust-ment without the path from peer victimization topsychological maladjustnient.The respecified model(see Figure 4) also showed an adequate fit [x (- * ' ¡^= 219) = 84.25./) = .00, RMSEA = 0.08, NFI =().93,CFI = 0.96, AGFI = 0.89]. In both models, therelationship between emotional aggression and bul-lying was nonsignificant. Comparing the two modelson fit indices that took parsimony into account (thatis,Akaikes information criterion |AIC], consistentAkaike's information criterion [CAIC], and ex-tended Akaike's information criterion [ECVI]),thelatter model (see Figure 4) (AlC = 140.25, CAIC= 263.15, ECVÍ = .64) had slightly smaller valuesthan the former one (see Figure 3) (AlC = 142.41,CAIC = 269.69, ECVI = .65) on all three indices.This would surest the relative superiority of the

model depicted in Figure 4 overtlie model depictedin Figure 3. Because the model depicted in Figure 4had a better fit and a more parsimonious structure,it was selected as the final model.

DISCUSSION

Instrumental and Emotional Aggression,Bullying, and Peer VictimizationThe primary focus ot this study was to investigatethe association between instrumental and emotionalagression and bullying behavior using the socialinformation processing model and the research ofSutton et al. (1999c). The social information pro-cessing model suggests that bullies can he eitherproactively aggressive or reactively aggressive, withdifferent characteristics proposed for the two typesof aj^ression. Proactive aggression—in which thepotential cost, benefit, and capability of executionare considered—is more instrumental. Reactiveaggression results from interpretation bias of socialcues, impulsivity, and hostile emotions (Crick &Dodge. 1999; Dodge, 1991 ). Sutton et al. ( 1999c) alsoproposed that bullies are largely manipulative andlack empathy (that is, their aggression is prim.arÜyinstrumental or proactive).

The two-factor measurement model (instrimien-tal and emotional aggression) supported the distinc-tion between the two types of aggression. These

Figure 4: Final Model of Aggression, Bullying, Victimization, and Maladjustment

Verbal Bullying ^4 - .38

.69

ReactiveAggrLitó i vc ness

.17 Hostility

PhysicalVkrimization

Vc-rb;ilViti i ini/atif in .08

.12

Anxiety -^r—.¿-i

L,tlnclint. s^ , ^ - - .«Ü

Kotc^ Path coefficlerm are standardized.

•p < ,05

238 Social Work Research VOLUME 33, NUMBER 4 DECEMBER 2009

Page 9: Instrumental or Emotional Aggression: Testing Models of Bullying

initial results provided evidence for the propositionthat both instrumental and emotional aggressionsare positively associated with hullying behavior.However, adding psychological maladjustment tothe nn.ll model weakens the association betweenemotional aggression and bullying.

It is also important to note that instrumentalaggression was not correlated with physical andverbal victimization, whereas emotional aggres-sion was positively associated with victimization.Thesf firulings support previous research tbat haspurported that proactively aggressive children oftenhave fairly good social knowledge and use violencein a calculated niaimer and can, therefore, avoidthe negative interpersonal consequences of beingatigressive, which include rejection and victimiza-tion by peers (Dodge 8¿ Schwartz, 1997; Sutton,2001).The hostile behavior of reactively aggressivechildren is more like inappropriate social responsesdue to emotional regulation and social cognitionproblems, which renders these children at risk forpeer rejection and psychosocial maladjustment(Haynie etal.. 2001).

Although the two-factor measurement model is abetter fit for understanding instrumental and emo-tional aggression, it is worth noting that substantivecorrelations were found between instrumental andemotional aggression (r = .55 in the measurementmodel). Despite the conceptual meaningfLilness ofdistinguishing between the two types of aggression,in reality many respondents manifesting one typeof agression were also at high risk of engaging inthe other. This finding supports previous researchsuggesting that individuals who manifest a high levelof reactive aggression are at bigh risk of proactiveagression, and vice versa, and that the etiologies ofproactive and reactive aggression may overlap, withthe possibility that some family environments maypromote both types of aggression (Dodge, 1991;l^odge & Coie, 1987). Salmivalli and Nieminen(2002) also found it typical for bully/victims to behighly .aggressive both reactively and proactively.Longitudinal research assessing chiklren s emotional,cognitive, and behavioral development starting atearly childhood is needed to further support thesefindings.

Aggression, Victimization, andPsychological MaladjustmentAnother aspect of the present study investigatedthe psychological consequences associated with

aggression and peer victimization. On the basis ofprevious research (for example, Austin & Josepli,1996; Kochenderter ¿k Ladd, 199i,; Natvig et .i).,2001 ). it was hypothesized that emoñonal aggressionindicates higher peer victimization and that the vic-timization experience, in turn, leads to psychologicalmaladjustment such as depression, loneliness, andanxiety. However, the mediating rale of victimizationexperience between aggression and psychologicalmaladjustment was not supponed by the data. In-stead, model fit indices showed peer victimizationand psychological maladjustment as two parallelcharacteristics directly related to emotional a^res-siveness.Thc strong direct eifect bet\veen emotionalaggressiveness and psychological maladjustmentsuggested that children being victimized by theirpeers have significant psychological issues,especiallyin the emotional domain. Previous studies oftenused peer victimization as an explanatory variableand psychological maladjustment as a dependentvariable without exploring other likely relation-ships between them (Austin & Joseph, 1996; Rigby,1999) or the possibility that both peer victimizationand psychological maladjustment may be causedby a third factor (for example, a personal trait).Thefindings of the present study reveal a prospectivearea for future investigation,and additional researchsbould be carried out to determine the directionof the relatioaship between peer victimization andpsychological maladjustment.

LimitationsRegarding the results of the multivariate analysespresented here, it is important to note diat the datastructure is cross-sectional. The lack of longitudinaldata makes it difficult to examine the causality anddirectionality of relationships between variablesor to observe changes over time. A developmentalperspective is beneficial in investigating the originsand trajectories of different kinds of aggression(Vitaro He Brendgen. 2005). Besides, the presentstudy focused primarily on the internalizing as-pects of maiadjustmejit associated with aggressionand victimization. The nonsignificant relationshipbetween instrumental aggression and internalizingproblems does not mean that individuals engagingin such behaviors are fi-ee of psychosocial issues.It wa.s found that bullies mainly had externalizingsymptoms such as delinquency and were often dis-liked by peers (Vcenstra et al., 2005).The potentialrelationship between aggression and externalizing

W E I AND WILLIAMS / TestingModebofBtilfying. Victimizatio?i. anaPsyclioloffcaiMahdjtatmentamongTaiwaneseSeventh-Graders 2 3 9

Page 10: Instrumental or Emotional Aggression: Testing Models of Bullying

r '

behaviors was not explored in the present study, andfuture investigations in this area are needed.

Another concern is the discrepancy between mea-sures.This study used data from multiple sources (self,peers, and teachers) rather than using single-source(that is. either self-report or peer-report) data formultiple variables to construct measurement modekfor hypothesis testing.This approach helped to avoidthe issue of shared-method variance. For example,both children's self-esteem and vicdmizaäon expe-riences might have been assessed using self-reportquestionnaires.The correlation coefficient betweenthese two factors could have been inflated by thefact these variables were measured using the samemetiiod. This problem can be addressed by usingmultiple methods in data collection (Juvonen,Nishina, & Graham. 2001). However, inconsisten-cies bet\veen teacher, peer, and self-reports wereidentified in the present study.This discrepancy mayhave resulted from the fact that the data collectionwas done during the respondents' first semesterof middle school, given chat teachers and snidentsmay not know each other very well at that time. Infact, bullying/victimization data obtained by dif-ferent sources are not necessarily always consistent-Pellegrini and Bartini (2000) compared the inter-correlations among self-reports, peer evaluations,teacher reports, and direct observations on children'saggression and bullying behavior and found low tomoderate correlations. Additional efforts are neededto better understand whether these inconsistenciesundergird cognitive processes, friendship patterns,or timing of data collection.

Finally, caution should be exercised when gen-eralizing the findings of the present study to othercontexts, because there is potential diversity in thedynamics and characteristics of bullying in differentcultures. An with that of several other Asian areas,theTaiw.inese culture is influenced by Confucianismand a coilectivist tradition (Nisbett, 2003). Someresearchers have highlighted the significant rolethat culture has played in local bullying incidents(for example, Yoneyama & Naito, 2003), whereasother studies in Asia have shown similarities withWestern samples (Abou-ezzeddine et al., 20()7;Wei,Jonson-Reid, & Tsao. 2007; Yang et al., 2006). Arecent study compared students with collectivisticand indi\'idualistic backgrounds and found minimalcultural effects on their attitudinal responses toschool bullying (Nesdale & Naito, 2005). None-theless, the present results should stimulate flirther

research in odier cultural contexts, which could helpto provide a more comprehensive understanding ofthe bullying phenomenon.

CONCLUSIONSGiven this study's hmitations, the present resultsreveal the heterogeneity of bullying and its as-sociated effects. Various preventive interventionapproaches should be considered to specificallyaddress the diverse etiologies underlying seem-ingly similar behaviors. Special attention shouldbe given to the conditions of bully/victims. Thesebullies not only have emotional distress and socialskill deficits, they also experience victimization andpeer rejection. They are likely to be at high risk ofdeveloping both reactive and proactive aggressionover time.This constitutes a vicious circle for peerrelations, mental health, and school adjustment.Cognitive—behavioral techniques such as attribu-tion modification, anger management, and socialskills training may be practical for children whoseaggression is related to hostile attribution bias andemotion dysregulation (Alexander & Cilurtis, 1993;Graham & Juvonen, 1998). It is important to notethat for bullies who do not have sigriificant socialcognition deficits, traditional antibullying curricu-lums or emotional education programs may not bevery effective (Sutton et al., 1999a). Furthermore,these programs can make bullies more knowledge-able about social manipulation and the exploitationof others for self-interest. For proactively aggre.ssivestudents, it may be important to identify the po-tential gains of bullying and change the reinforcingstructures that sustain the "profitability" of such acts.Pellegrini and Bartini (2001) have suggested thatbullying is a deliberate action to build one's statusand attain dominance over others that results fix>msocial competition and power seeking among boys.If this is true, then peers and school climate shouldconstitute crucial elements in any preventive inter-vention design. An example of such an interventionwould be an antibullying program implementedin schools in Ireland that adopted a schoolwideapproach to increase the student body's awarenessand encourage suidents through peer leadership tosupport peers who they witness being bullied (A.M. O'Moore & Minton, 2005). it is apparent fromthis study and previous research that school bully-ing is a complex phenomenon requiring carefullytailored interventions to respond to the differentneeds and risk factors of students. In providing a

240 Social Wort Research VOLUME 33, NUMBER 4 DECEMBER 2009

Page 11: Instrumental or Emotional Aggression: Testing Models of Bullying

comprehensive analysis of the relationships betweeninstrumental and emotional aggression, bullying,peervlctimization,and psychological maladjustment,this study contributes to the existing knowledge baseand promotes the design and implementation ofmore evidence-based and effective school bullyingprevention programs. BŒID

REFERENCESAbou-ezzeddine.T.. Schwartz. D., Chang, L.. Lee-Shiiii,Y.

Farvcr.j., öc Xu.Y. (20(17). Posinve peer rcbrionsliipsas moderators of risk for victimization in Chineseand South Korean children's peer groups. SocialDevelopment. 16, U16-127.

Alexander.J. R..,& Curtis, C. M. (1995). A critical reviewof strategies to reduce school vioXence. Social Work inEducation. /7. 7.VS2.

Andcrshed. H., Kcrr, M., & Stattin. H. (20(11). Bullyingin school íMiá violence on the streets: Arc the samepeople involved? Jiinrn.i/ of Scattdinavian Studies inCrimiiiology and Crime Preivtition. 2, 31—49.

Aronson, E. (1992). 77»' social animal. New York: Freenun.Asher. S. R.. Hyman, S.. & Renshjw. H W (19K4). Loneli-

new in children. Child Development. 55, 1456-Î464,Austin, S,.&: Joseph. S. (1996). Assessment ofbully/vic-

tmi problenw in 8 to 11 year-olds. British Journal ofEducational Psycholoj!)'. 66. 447—456.

lijorkqvist. K.. & Ostern'ian. K, (1998). Pro Facúltate No.4:Scales for research on interpersonal relations.Vasa, Finland:AboAkadcmi University.

ßouUon,M.J.,Trueman, M..Chau, CWhitehand. C ,& Aniatya, K. (1999). Concurrent and loiigicudinallinks between friendship and peer victimizarion:Implications tor befriending intervemions.Joimw/ ofAdolescence, 22, 4(>l-46û.

lJushnian,ü,j„&: Anderson. C A . (20(11). Is it dme to pullthe plug on the hostile versus instrumental aggressiondichotomy? Psychohjiica! RetHew, IOS, 273-279.

Carney. A. G.,& Merrell.k.W. (2001). Bullying in schools:Perspectives on understanding and preventing aninternational problem. School Psycholog International,22, 3h4-3S2.

Christie. IK..& Geis. F L. ( 1970). Studies in Machiawllianism.New York: Academic Press.

Crick. N. k . .& Dodge, K. A. (1994). A review and refor-mulation of social information-processing mecha-nisms in childrens social adjustment. PsychologicalBulletin. 115. 74-101.

Crick, N.R. .& Dodge. K.A. (1996). Social information-processing mechiuiisms in n'active and proactiveaggression. Child Diwlopment, 67, 9';3-1OO2.

Crick, N. K., & Dodge, K. A. (IWOJ.'-Siiperiority*' is in theeye of the beholder: A comment on Sutton, Smith, andSvrettenham. SiKÍal Dnvlopmrnt, 8, 12S-131.

Davidson. R.J.Jackson, D. C , & Kalin, N. H. (2000).Emotion, plasticity, context, and regulation: ferspec-tives from affective neuroscience. Psychological Bulletiu.126. 890-909.

Dodge. K. A. (1991).The structure and fiinction of reactiveand proactive aggression. In D. Pepler &.• K, H. Rubin(Eds.), 'tlie development and tn-attnent of childhood aggres-sion (pp. 201-218). Hiitstble. NJ: Lawrence ErlbauniAssociates.

Dodge, K. A., & Coic.J. D. (1987). Social-iiiformarion-processing factors in reactive and proactive aggressionin childrens peer ^joup^.Journal of Personality andSocial Psychology, 53, 1146-1158. "

Dodge,K.A.,&- Schwanz, D. (1997). Social informationprocessing mechanisms in agressive behavior. InD. M. StofT.J. BreiUng. & J. D. Maser (Eds.). Handbookof antisocial behavior (pp. 171-180). New York: JohnWiley & Sons.

Graham, S.. Ä Juvonen.J. (1998). Self-blame and peer vic-timization in middle school: An attribuàonal analysis.Developmental Psychology, 34, 587-599.

Greene, M. B. (200fi). BuUying in schook;A plea formeasua- of human rights. JoHriw/ of Social Issues, 62,6.V79.

Hartup, W. W. ( 1974). Agression in childhood: Devel-opmental perspectives. American Psychologist, 29,33ñ-.Mt.

Hawker, D.S.J.. & iioulton. M.J. (2000).Twenty years'research on peer victimization and psychosocial mal-adjuscmcnt:A meta-analytic review of cross-set tionalstudies._/t>»rtiu/(¡f C/ii/íí Psychology atul Psychiatry, -i/,4 4 | t 5 5

Hawk-y. P. H, (2006). Evolution and personality: A newlook at Machiavellianism. In 1> Mroczek S¿ T. Little(Eds.), Hiindhook of personality dcrelopmeut (pp. 147-161). Mahwah. NJ: Lawrence Erlbaum Associates.

Hawfey, P H. (20(i7). Social dominance in childhood andadolescence: Why social competence and .iggressionmay go hand in hand. In P. H. Hawley.T. D Little, &P. Rodkin (Eá^.). Aggression and adaptationI'lhc brightside to bad behavior (pp. ]-29). Hillsdale, NJ: LawrenceErlbaum Associates.

Haynie, D. L., Nansel,T., Eitel. P., Crump, A. D , Saylor,K.,Yu. K.. & Simnions-Morton, B. (2001). Uuüies.victints, and bully/victims: Distinct groups of at-riskyouih.Jíiiínw/ of Early Adolescence, 2i, 29-49.

Hung, L. (2000). Manual of the Adolescent Social Behavior Scale.Taipei.Taiwan : Psychological Publishing Company.

lvarsson,T., Broberg. A. Cí., Arvidsson,T., & Gillbet^, C(2005). Bullying in adolescence: Psychiatric problemsin victims and bullies a measured by the youth selfreport (YSR) and the depression self-rating scale(DSRS). Nordicjournal ofPsychialry. 59, 365-373.

Jöreskog. K.. & Sorboni. D. (2001). LISREL (Version 8.50)[Computer ioftwarej. Liiicolnwood, IL: ScientificSoftware Internacional.

Juvonen,J., Nishina, A.. &: Graham, S. (2{)01). Self-viewsversus peer perceptions of victim status among earlyadolescents. In J.Juvonen & S. Graham (Eds.), Peerharassment in school:TI¡e plight of the vuhicrable andviaimized (pp. 105-124). New York: Guilford Press.

Kaukiainen, A., Salmivalli. C . Lagerspetz. K..Taniminen,M.,Vauras, M.,Maki.H.,&Voskiparta.E. (2002).Learning difficulties, social intelligi-ncL- and self-concept: Connections to buily-victim problems.Scandinavian Journal of Psyrliology, 43, 269-27H.

KJine. R. B. (1998). ¡Principles ami practices of structural equa-tion modeling. New York: Guilford Press,

Kochenderfer,B!j.,&: Lidd, G.W, (1996). Peer victimiza-tion: Cause or consequence of school maladjustment?Child Dewhpment, 67. 1305-1317.

Kunipulainen. K,. & Rasanen, E. (2000), Children involvedin bullying at elementary school agL':TÍicir psychi-atric symptoms and deviance in adolescence: Anepidemiológica] sample. Child Ahuse & Neglea, 24,1567-1577.

LeDoux,J. E. (1996). Hie emotional brain.The mysteriousunder¡)inning.i of emotional life. New York: Touchstone.

Lee. M.. Lee,Y.Yen. L., Lin, M., & Lue. B. (1990). Reliabil-ity and validity of using a brief psychiatric symptomrating scale in clinical practice. Junmii/ of the FonnosanMedical Association, 89, 1081-1087.

Miller.J. D , & Lynam, D. R. (2(iO6). Reactive and pmac-Dve aggression: Similarities and différences. Personalityand ¡ndiindual Differences, 41, 1469-1480.

WF.I AND WILLIAMS / Testing Models ofBuifying, \^ictirnw^n, and I^hologicd Maladjustment atnangTattmneie Set>emh-Gnuim 241

Page 12: Instrumental or Emotional Aggression: Testing Models of Bullying

Nansel.T. R.. Overpeck, M., PUb, k . S.. Kuan,W.J.,Simons-Morton. B.,& Sclieidt. R (2l)(ll). Bullyingbch.iviors among US youth: Previilfme and a.wiciationwith psychosocial adjmtinentJ.'lAM. 285. 2094-2100.

Nat\'ig. G. K.. Albtrktscn. G.. & Qvarnstmni. U (2001).I'sychosomatic symptoms amon^ victims uf schoolhuWy'm^.Joumal of Heallli Psychohgy. 6. 3tt.S-377.

Nesdale, D., & Naito. M. (2(WI5). Individualism-collectivisiiiand the attitudt-s to school buU^ang ot'Japanese andAastralian ^ludena. Journal of Cross-CiiltuntI Psycholog',36. .S37-556.

Nisbt'tt, R. E. (2003). The <ieoxr<iphy of thought: How Asiansand Westerners think tlißcrciUly ... and why. New York;Free Press.

Olwfus, D. {1W2), Bullying among schoolchiklreii:Intervention and prcveiuion. In R, D. iVters, R.J.McMahon. & V. L. Quinst-y (Eds.), Aggression and vio-¡fiuf throu}¡hoiil ihc life spati (pp. 1()()-125). NewburyPark, CA: Sa^c Publications.

Olweus, I"). (1997). Üully/vicrim problems in school: Factsand intervciuion, F.uropean Journal of Psychology ofBdtianioti. 12. 4'>5-5IO.

O'MooR-. A.M.. Kirkham. C.,& Smith. M. (1SÍQ7).Bullying behaviour in Irish schools: A nationwidescudy. ¡riili Joiiniiil of Píyclioto}¡y. 18, 141-16*í.

O'Moore, A. M..& Minn>n, S.J. (2005). Evaluation ofthe cffecdvenfss ufan anti-bullying programme inprinury schools. A}¡gressii'c liiiun'ior, M. W>~(i22.

i)'Moort.-.M..& Kiikhani,C. (2001). Self-esteem and icrcladoiLship to bullying behavior. A<i^resshv Behavior.27, 26^>-2«3.

Pellegrini,A. IJ., & Barcini. M. (2000).An empiricalcomparison ofmethotis of sampling aggression andvictimization in sfhoo! scttings.Ji'finm/ of EdtuatloiuilPsydiohgy. 92. 361KÎ66.

IVllegrini.A. !>.. & Bardni, M. (2001). Dominance in earlyadolescent boys: AfFiiiaCivc and aggressive dimensionsand possible iuncrions. Merrílhí^lmer Qminerly. 47,142-163.

Rainc. A., Dotiip.'. K., Loeber. K., Gatzke-Kopp. L., Lynam,I),, Reynolds, C , et al. (2(MI6).The Reactivt^-ProacdveAtç^rcssion Questionnaire: Oifferendal correlates ofreactive and paiacdve aggression in adolescent boys.Agiireiiife Bcharior. 32, 159-171.

Ramirez.J. M. ,* Andreu,J. M. (2006).Aggression and somerelated psychological con.strucLs (anger, hostility, andimpuLsivity):Some comments from a a'search project.Ncurosäeme & Biohehuvioral Rcvieuv. .i(J, 27i>-2*)].

Rigby, K. (1999), Peer victimisation at school and thehealth of secondary school students. British Journal ofEduaitiomi Psychoh\^y, 69. 95-104.y. K.. & Slee, PT. (1991). Bullying among Auscraiianschool children: Reported behavioT and attitudestoward victims.Jciwrrw/ Q/Sodai Psychology, 13!,615-627.y, K., & Slcc, P.T. (1993). Dimensions of interpersonalrelation among Australian children and implica-dons for psychological wcll-bcing._/oi<r;iti/ oj SocialPsychohi^y. 133. 33-42.

Roland, E.. & Idsoe.T. (2(.iOl). Aggression and bullying.^Cifi'MiW ßehai'ior. 27. 446-462.

Salmivalli, C . & Nieminen, E. (2002), Proacdve and reac-dve aggression among school bntlies, victims, andbully-vicdms. /Iij rcssn/c Behavior, 2H. 30—44.

Scheithauer, H., Hayer,T.. Petermann. F.. & Jugert, G.(20U6). Physical, verbal, and reladonal forms of bul-lying among German students: Age trends, genderditfercnces, and corrclates. >l^rfî.îiVe Behavior. 32,261-275.

Schumacker, R. E., & Lomax, R. G. (1996). A beginner's}¡uide to itnictural eqitutian moiU-Hng. Mahwah, NJ:Lawrence Erlbauni Associates.

Schwanz, D. (2(tO0). Subtypes of victims and a^ressorsin children's peer groups.JuMmii/ oJAbnormal ChildPsycholog, 28. 1K1-192.

Storch,E. A., &- Lcdley. D. R. (2ÍH)5). Peer victimization ^aiid psychosocial adjustment in children: Cura-ntknowledge and future directions. Clinical Pediatrics,44, 29-38.

Sutton.J. (20(11). BuIlies:Thug? or thinkers? Psychologist,14. 530-534.

Sutton.J., & Keogh, E. (2000). Social competition inschool: Relationships with bullying. Machiavellianismand personality. British Journal of llducatioml Psycholojiy,70. 443-456.

Sutton,J., Smith, P. K., iV Swettenham. J. (1999a). BuUyingand "theory of mind": A critique of the "socialskills deficit" view of and-social behaviour. SodalDevelopment, 8. 117-127.

SuttonJ.. Smith. P K.. &• Swettenham.J. (1999b). Socialcognition and bullying: Social inadequacy or skilledmanipulation? British Journal of DiwlapniciitalPsycholog)', 17. 435-450.

Sutton.J., Smith. P K., & Swettenham.J. (1999c). Sociallyundesirable need not be infompetent: A response toCrick and Dodge. Social Dcvelopim-nt. 8. 132—134.

Takahiro. CÍ..& IwaoT (20(m). Psychological stress of vic-tims and bullies in junior high Khoc\\.lapiinese Journalof Educational Psyihoh(¡y, 48. 410-421.

UnneverJ, D. (2005). Bullies, aggressive victims, andvictims: Are they distinct groups? Aggressiw Behauior,31. 153-171.

Veenstra. R.. Lindenberg, S., Oldehinkel. A.J,, De Winter.A. F.Verhuist. F. C , & l^Jrmcl.J. (2005). Bullying andvicdmization in elementarv" schools: A comparisonof bullies, victims, buliy/victnns. and uninvolved pré-adolescents, Dfvflopmefiial Psychology. 41. 672-682.

Vitaro, F.. & Brendgen, M. (20(15). Proactive and reac-tive aggression: A developmental perspective. InR.E,Tremblay,W,W,Harrup,&J. Archer (Eds.).Dei'clopmeeital origins of aggression (pp. 17H-201). NewYork: Guilford Press.

Wei. H. S..Jonson-Reid, M., &Tsao, H. L. (2007).Bullying and victimization amongTaiwanese7th graders: A multi-method assessment. SchoolPsychology International. 28, 479-500.

Wei, H-, & Williams.J. H. (2004). Relationship betweenpeer vicdmization and school adjustment in sixthgrade students: Investigating mediadon effects.Violeticei^l'ictims. 19, 557-571.

Whitnoy, !,,& Smith, I' K. (1993). A survey of the natureand ex-Cent of bullying in junior/middle and second-ary schools, tiducationai Research. 3.\ 3-25.

Yang, S.J., Kim,J. M.. Kim. S.W., Shin. L S.. &YQon.J, S.(2006). Bullying and vicdmization behaviors in boysand girls at South Korean primary schools.Joiirrirf/ ofthe American Academy of Child & Adolescent Psychiatry,45. 69-77,

Yoneyama. S., & Naito. A. (200.3), Problems with the para-digtn:The school as a factor in understanding bully-ing (with special reference to Japan). British Jourmii ofSociology cf Education. 24, 315^330,

Hsi-sheng Wei, PhD, is assistant professor, Departmentof Social IVork, National Taipei University, Í51 UniversityRoad. Siin Shia. Taipei, 237 Jaiwan; email: hu'[email protected]. James Herbert IVtUiams, PhD, is dean and professor,Graduate Sthool of Social Work, University of Dentar.

Original manuscript received September 23, 2008Accepted January 26, 2009 , ,-

242 Social Work Research VOLUME 33, NUMBER 4 DECEMBER Z009

Page 13: Instrumental or Emotional Aggression: Testing Models of Bullying

Copyright of Social Work Research is the property of National Association of Social Workers and its content

may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express

written permission. However, users may print, download, or email articles for individual use.