Relational Aggression, Gender, And

7/28/2019 Relational Aggression, Gender, And

http://slidepdf.com/reader/full/relational-aggression-gender-and 1/14

Relational Aggression, G end er, and

Social-Psychological Adjustment

Nicki R. Crick and Jennifer K. GrotpeterUniversity of Illinois at Urbana-Cham,paign

CRICK, NICKI R., and GROTPETER, JENNIFER K. Relational Aggression, Gender, and Social-

Psychological Adjustment. CHILD DEVELOPMENT, 1995, 66, 710 -722 . Prior studies of childhoodaggression have dem onstrated that, as a group, boys are more aggressive than girls. We hypo the-sized that this finding reflects a lack of research on forms of aggression that are rele van t to youn gfemales rather than an actual gender difference in levels of overall aggressive ness. In the presentstudy, a form of aggression hy poth esized to be typical of girls, relational aggression, was assessedwith a peer nomination instrument for a sample of 491 third- through sixth-grade children. Overtaggression (i.e., physical and verbal aggression as assessed in past research) and social-

psychological adjustment were also assessed. Results provide evidence for the validity and dis-tinctiveness of relational aggression. Further, they indicated that, as predicted, girls were sig-nificantly more relationally aggressive than were boys. Results also indicated that relationallyaggressive children may be at risk for serious adjustment difficulties (e.g., they were significantlymore rejected and reported significantly higher levels of loneliness, depression, and isolationrelative to their nonrelationally aggressive peers).

Because of the deleterious effects ofconduct problems on children's develop-ment (see Parker & Asher, 1987, for a re-view), a great deal of research has been con-

ducted on aggression in the past decade(e.g., see Dodge & Crick, 1990; Parke &Slaby, 1983, for reviews). Although signifi-cant advances have been mad e in our unde r-standing of childhood aggression, one limi-tation of this research has been the lack ofattention to gender differences in the ex-pression of aggression (cf. Robins, 1986).Prior studies demonstrate that, as a group,boys exhibit significantly higher levels of ag-gression than do girls (see Block, 1983;Parke & Slaby, 1983, for reviews), a differ-ence that persists throughout the life span(Eagly & Steffen, 1986; Hyde , 1986; Ken-rick, 1987). Not surprisingly, these findingshave been interpreted as an overall lack ofaggressiveness in girls' peer interactions.However, an alternative explanation is thatthe forms of aggression assessed in past re-search are more salient for boys than forgirls. If so, young females may exhibitunique forms of aggression, forms that havebeen overlooked in past research.

Although specific definitions have var-ied over the years, aggression has been gen-erally defined by most authors as behaviorsthat are inten ded to hurt or harm others (e.g.,

Berkowitz, 1993; Brehm & Kassin, 1990;Gormly & Brodzinsky, 1993; Myers, 1990;Vander Zanden, 1993). We propose that,when attempting to infiict harm on peers(i.e., aggressing), children do so in ways thatbest thwart or damage the goals that are val-ued by their respective gender peer groups.As past research has consistently shown,boys tend to harm others through physicaland verbal aggression (e.g., hitting or push-ing others, threatening to beat up others).These behaviors are consistent with thetypes of goals that past research has shownto be important to boys within the peer-group context, specifically, themes of instru-mentality and physical dominance (seeBlock, 1983, for a review). These types ofconcerns are not as salient for most girls,however. In contrast to boys, girls are morelikely to focus on relational issues du ring so-cial interaction (e.g., establish ing close, inti-mate connections with others) (see Block,1983, for a review). In the present study.



we hypothesized that aggressive behavioramong girls would be consistent with theirsocial concerns, similar to the pattem foundfor boys. Specifically, we hypothesized thatgirls' attempts to harm others would focuson relational issues and would include be-haviors that are intended to significantlydamage another child's friendships or feel-ings of inclusion by the pe er g roup (e.g., an-grily retaliating against a child by excludingher from one's play group; purposefullywithdrawing friendship or acceptance in or-der to hurt or control the child; spreadingrumors about the child so that peers will re-ject her). Thu s, we expected tha t girls would

be most likely to harm peers through rela-tional aggression (i.e., hanning othersthrough purposeful manipulation and dam-age of their pe er relationships) wh ereas boyswould be m ost likely to harm peers throughovert aggression (i.e., harming othersthrough physical aggression, verbaJ threats,instrumental intimidation).

Although gender differences in theforms of aggression that children exhibitwere postulated years ago (Feshbach, 1969),very little relevant research has yet beenconducted. In one of the earliest studies onthis topic, Feshbach (1969) observed firstgraders' responses to unfamiliar peers. Shefound that girls were significantly morelikely than boys to resp ond to the unfamiliarpeer with behaviors that, although referredto by the author as "indirect aggression," ap-pear similar to those specifically definedhere as relational aggression (e.g., rejectionand social exclusion).

This pattem of results has also beendocumented for older children. In a morerecent study. Cairns, Cairns, Neckerman,Ferguson, and Gariepy (1989) asked fourththrough ninth graders to describe recentconfiicts w ith pe ers. C ont ent analysis of chil-dren's responses revealed that same-genderconfiicts among girls were significantly morelikely than boys' confiicts to involve themesof social alienation and m anipu lation of pe eracceptance (i.e., themes that are consistent

with relational aggression). Using a Finnishsample of fiflh graders, Lagerspetz, Bjork-

Crick and Grotpeter 71

These investigations provide initial support for the hypothesis that relationally aggressive behaviors axe present in children

peer interactions and that girls are morlikely than boys to exhibit them. Howeverdespite the conduct of hundreds of studieon the g eneral topic of childh ood aggressioin the past several decades (see Dodge &Crick, 1990; Parke & Slaby, 1983, for reviews), no systematic research has b een conducted on relational aggression. Thus, no information has yet been generated on thcorrelates of relational aggression or thcharacteristics of the children who exhibit (i.e., other than the previously describe

gender differences). Given the potentiallserious consequences of aggression for children's adjustment (see Parker & Asher, 1987for a review), it seems important to initiatresearch in this relevant, but unexplored domain. This is particularly true given that thiform of aggression may be most characteristic of young females, a group whose behavioral difficulties have received scant attention in past research. The present researcwas designed as an initial attempt to addres

these issues.We had four goals for the present study

(1) to develop a reliable measure of relational aggression, one that did not confounrelational aggression with other forms of aggression ; (2) to assess ge nd er differences irelational aggression; (3) to assess the degree to which relational aggression is distinct from overt aggression (i.e., physical anverbal aggression as assessed in most of thpast research in this area); and (4) to asses

whether relational aggression is related tsocial-psychological maladjustment. We hypothesized that relational aggression woulbe related to, but also relatively distincfrom, overt aggression. Furth er, we expec tegirls to be more relationally aggressive thaboys. Also, similar to overtly aggressive children (Bukowski & Newcomb, 1984; Coie &Kupersmidt, 1983; Dodge, 1983; Parker &Asher, 1987), we expected relationally aggressive children to be more Socially and

psychologically maladjusted than their nonaggressive peers.



71 2 Child Developm ent

behaviors, because of their relatively indi-rect nature and focus on peer relationships,might be difficult for those outside the peergroup (e.g., teach ers, researchers) to reliablyobserve and evaluate in naturalistic settings(cf. Lagerspetz et al., 1988). Thus, it wasjudged that peers would be the best infor-mants, a method that has the additional ad-vantage of providing multiple assessmentsof behav ior (i.e., bec ause each ch ild is evalu-ated by all of his or her classmates as op-posed to only one teacher, for example). Inaddition to the peer nomination instrument,subjects also completed several instrumentsdesigned to assess social-psychological ad-

justment (i.e., peer status, depression, lone-liness, social anxiety, social avoidance, andperceptions of peer relations). These aspectsof adjustment were chosen because past re-search has shown them to be predictive ofconcurrent and/or future socio-emotionaldifficulties (e.g., Ash er & W hee ler, 1985;Crick & Ladd , 1993; Fran ke & Hy me l, 1984;Kovacs, 1985; Parker & Asher, 1987). Thus,they would allow for an assessment of theadjustment risk status of relationally agg res-

sive children.

Method

SubjectsA total of 491 third- through sixth-grade

children from four public schools in a mod-erately sized midwestem town participatedas subjects.^ The sample included 128 third(65 girls and 63 boys), 126 fourth (56 girlsand 70 boys), 126 fifth (57 girls and 69 boys),and 111 sixth graders (57 girls and 54 boys).

Approximately 37% of the sample was Afri-can-American, 60% was European-Amer-ican, and 3% repres ented other ethn icgroups. Each subject had parental consentto participate in the study (consent rate wasabove 82%).

Peer Assessment of Relational Aggressionand Other Aspects of Social Adjustment

A peer nomination instrument was usedto assess social adjustment. This instrument,which consisted of 19 items, include d a peer

sociometric and four subscales designed toassess social behavior: relational aggression.

overt aggression, prosocial behavior, andisolation. These particular indices were se-lected becau se they represe nt the constructsthat have bee n u sed m ost extensively in pastresearch to evaluate children's social adjust-ment (i.e., peer status, aggression, with-drawal, prosocial behavior) (Crick & Dodge,1994).

Overt aggression was assessed with athree-item peer nom ination scale. The itemsincluded in this scale assessed physical andverbal aggression and were drawn fromthose used in prior research (e.g., Asher &Williams, 1987; Coie & Dodge, 1983;

Dodge, 1980; refer to Table 1 for a descrip-tion ofthe items). Relational aggression wasassessed with a five-item nomination scalethat was developed for use in the presentproject (refer to Table 1 for item descrip-tions). Items included in this scale describebehaviors that represent purposeful at-tempts to harm, or threats to harm, another'speer relationships (e.g., telling a friend thatyou will not like her anymore unless shedoes w hat you te ll he r to do). A pool of rela-tional aggression items, designed to fit the

proposed definition, was initially generatedby the authors. Selection of the subse-quently chosen items and specific wordingof each was based on pilot testing withgrade-school-age children.

The prosocial behavior scale consistedof five items (e.g., peers who help others),and the isolation scale consisted of fouritems (e.g., peers who play alone at school,peers w ho seem lonely at school). Th e items

included in these scales were based on thoseused in past research (e.g., Asher & Wil-liams, 1987; Crick & Dod ge, 1989). Th e pe ersociometric consisted of two items, nomina-tions of liked and disliked peers (positiveand negative nominations). These itemshave been used extensively in past researchto assess peer acceptance and rejection (seeCrick & Dod ge, 1994, for a review ).

During the administration of the peernomination instrument, children were pro-

vided with a class roster and were asked tonominate up to three classmates for each of




TABLE 1

FACTOR LOADINGS FOR THE PEER NOMINATION INSTRUMENTS

FACTOR

Prosocial/ Overt Relational Isolation/ITEM Happ y Aggression Aggression Unh appy

Good leader 789Does nice things for others 884 . . . . . . . . .He lps others 899Cheers up o thers 855 . . . . . . . . .Seems happy at school 832 . . . . . . . . .Hits, pushes others .906 . . . . . .Yells, calls others mean names .823 . . . . . .Starts fights .884

When mad, gets even by keeping the per-son from being in their group offriends . . . .763

Tells friends they will stop liking themunless friends do wh at fliey say . . . .772

When mad at a person, ignores them orstops talking to them . . . .837

Tries to keep certain people from beingin their group during activity or playtime . . . .727

Plays alone a lot . . . . . . .911Seems sad at school . . . . . . .916

Seems lonely at school . . . . . . .916NOTE.—All other factor loadings were less than .300 except for two items with loadings of .380 and .376. These

were considered insuhstantial given the relatively high loadings presented above.

the items . Th e nu m be r of nom inations chil- (2) overtly aggressive (RAGG low, OAGGdre n rec eiv ed fr-om pe er s for each of th e high ); (3) relationally aggre ssive (RAGGitems (for each child, the se scores co uld high, OAGG low); and (4) com bine d overtlyrange from 0 to the total nu m be r of child ren and relationally aggressive (RAGG andin his or her class minus 1) was summ ed and OAGG both high). This proc edure resultedthen standard ized within each classroom. in the identification of 371 nonag gressiveThe standar dized scores for the overt aggres- childr en, 41 overtly aggressive child ren, 46

sion, relational aggression, prosocial behav - relationally aggressive child ren, and 33ior, and isolation scales w ere sum m ed to overtly plus relationally aggressive child ren,yield four total scores (e.g., children's stan-dardized scores for each of the three items Classification of sociometric statuson the overt aggression scale were summed groups.—The positive and negative sooio-to create a total score). metric nom inations child ren rece ived from

their classmates were used to identify fiveClassification o} aggressive g roups.— sociometric status groups, popular, average.

Chil dren s relation al (RAGG) and overt negle cted, rejected, and controversial chil-(OAGG) aggression scores we re used as con- dren, using the proce dure described by Coietinuous variables in su bs equ en t analyses and Do dge (1983) (except for thos e in theand also to identify groups of aggressive ver- average group who w ere identified using thesus nonaggressive children. Children with criteria described by Coie, Do dge, & Coppo-scores one sta ndard deviation above the telli, 1982). This proc edur e resu lted in the



71 4 Child Development

dissatisfaction. This scale, an instrumentwith demonstrated reliability and validity(e.g., Asher & W heele r, 1985; Asher & Wil-liams, 1987; Crick & Ladd, 1993), consists of16 items that assess loneliness at school(e.g., I feel alone at school) and eight filleritems (e.g., I like m usic). Possible respo nsesto each item range from 1 (Not at all trueabout m e) to 5 (Always true a bout m e). Chil-dren's responses to the loneliness itemswere summed yielding total scores thatcould range from 16 (low loneliness) to 80(high loneliness). Cronbach's alpha for chil-dren's responses to the loneliness scale was.91.

Social anxiety and avoidance.—TheFranke and Hymel (1984) social anxietyscale, an instrument with dem onstrated reli-ability and validity (e.g.. Crick & Ladd,1993; Franke & Hymel, 1984), was used toassess social anxiety and social avoidance.This instrument consists of two subscales,social anxiety (e.g., I usually feel nervouswhen I meet someone for the first time) andsocial avoidance (e.g., I often try to get awayfrom all the other kids), each of which in-

clude six items. Possible responses to eachitem range from 1 (Not at all true abo ut m e)to 5 (Always true about me). Children's re-sponses to the items were summed for eachsubscale yielding total scores that couldrange from 6 (low anxiety/avoidance) to 30(high anxiety/avoidance). Cronbach's alphafor children's responses to the social anxietyand social avoidance scales was .69 and .74,respectively.

Depression.—The Children's Depres-sion Inventory (CDI) was used to assess

children's feelings and symptoms of depre s-sion (Kovacs, 1985). This measure consistsof 27 items, all of which assess depression.Each item consists of three related state-ments, and children respond by selectingthe on e statemen t that best fits how they feel(e.g., I am sad once in a while vs. I am sadmany times vs. I am sad all the time). Itemsare scored from 0 to 2, with higher scoresindicating more evidence of depression.

Two modifications were made to this instru-ment prior to its use in the present study,

do not like swimming) in an attempt to bal-ance the negative tone of the CDI items.Cronbach's alpha for children's responses to

the 25 depression items was .85.Perceptions of peer relations.—An ad-

aptation of the Children's Peer RelationsScale (Crick, 1991) was used to assess chil-dren's perceptions of their peer interactions.This instrument is designed to assess six as-pects of children's perceptions of their inter-actions with peers at school: perceived peeracceptance, isolation from peers, negativeaffect, engagement in caring acts, engage-ment in overt aggression, and engagement

in relational aggression. Specifically, theperceived peer acceptance subscale (threeitems) assesses the degree to which childrenfeel liked by peers at school (e.g.. Some kidshave a lot of classmates who like to play w iththem. How often do the kids in your classlike to play with you?). The isolation frompeers subscale (two items) assesses the de-gree to which children perceive themselvesas loners at school (e.g.. Some kids play bythem selves a lot at school. How often do youdo this?). Th e negativ e affect subscale (threeitems) assesses the degre e to wh ich childrenfeel lonely, sad, or upset at school (e.g..Some kids feel upset at school. How oftendo you feel this way?). The engagement incaring acts subscale (four items) assesseschildren's perceptions of the degree towhich they direct prosocial behaviors to-ward their pe ers (e.g.. Some kids try to che erup other kids who feel upset or sad. Howoften do you do this?). The engagement inovert aggression subscale (three items) as-

sesses children's perceptions of the degreeto which they direct overtly aggressive actstoward their peers (e.g.. Some kids hit otherkids at school. How often do you do this?).The engagement in relational aggressionsubscale (five items) assesses children's per-ceptions of the degree to which they directrelationally aggressive behaviors towardtheir peers (e.g.. Some kids tell their friendsthat they will stop liking them unless thefriends do what they say. How often do youtell friends this?). Th e last two su bscales , en-gagement in overt and relational aggression,were designed to parallel those included in




internal consistency showed that children'sresponses to the items were reliable withCronbach's alpha = .74, .76, .66, .82, and.73 for the perceive d acceptance, caring acts,peer isolation, overt aggression, and rela-tional aggression subscales, respectively forthe present sample. Support for the con-struct validity of the Children's Peer Rela-tions Scale (CPRS) has also been demon-strated in past research (e.g., rejected,overtly aggressive children report signifi-cantly higher levels of overt aggression onthe CPRS relative to peers; rejected, with-drawn children report significantly higherlevels of isolation and lower levels of peer

acceptance relative to peers) (Crick, 1991).

Administration ProceduresThe previously described instruments

were completed by subjects during two 60-min group assessment sessions (session Aand session B) conducted within children'sclassrooms. These sessions were conductedby the authors, who employed standardizedprocedures. During each session, childrenwere trained in the use of the response

scales prior to administration of the instru-ments. Each item of every instrument wasread aloud by the administrator, and assis-tants were available to answer children'squestions. Sessions A and B were adminis-tered to classrooms in a random order, andthe two sessions occurred approximately 1week apart.

During session A, children completedthe peer sociometric and behavior nomina-tion measure, the Asher and Wheeler (1985)

loneliness scale, the Franke and Hymel(1984) social anxiety scale, and one addi-tional instrument that was not part of thepresent study. The peer nomination instru-ment was always administered first (to helpinsure that children would not be fbcusedon the nom inations they gave to others at theend of fhe session), however, the order ofthe loneliness and social anxiety scales wasdetermined randomly. During session B,children completed the Children's Peer Re-

lations Scale (Crick, 1991), the Children'sDepression Inventory (Kovacs, 1985), andtwo additional instruments that were not

first conducted on the scores children received from the pee r nomination instrumento assess whether relational aggressiowould emerge as a separate factor, independent of overt aggression. This analysiyielded the four predicted factors (prosociabehavior, ove rt aggression, relational aggression, and isolation), and these factors accounted for 79.1% of th e variation in thscores. Specifically, the prosocial factor accounted for 34.0% of the variation (eigenvalue = 5.10), the overt aggression factor accounted for 23.9% (eigenvalue = 3.59), threlational aggression factor accounted fo13.5% (eigenvalue = 2.02), and the isolatio

factor accounted for 7.6% (eigenvalue =1.14). Based on the results of the factor analysis, two items were dropped from thscales. Specifically, one isolation item (i.egives in easily to others) was dropped because it had a much lower factor loadinthan did the other items on this scale (.5relative to the other items which loadeabove .90). Further, one relational aggression item (i.e., tells mean lies or rumorabout a person to make other kids not lik

the person) was dropped because, althougit loaded on relational aggression (.64), also cross-loaded with overt aggression (.49)Factor loadings for the item s of the resultinfour subscales were relatively high, ranginfrom .73 to .92 (refer to Table 1). Computation of Cronbach's alpha showed all scaleto be highly reliable (alpha = .94, .83, .9.92 for overt aggression, relational aggression, prosocial behavior, and isolation, respectively).

The relation between relational anovert aggression was further assessed with correlation coefficient, r = .54, p < .01. Thmoderate magnitude of this correlation iwhat one would expect for two constructthat are hypothesized to be different formof the same general behavior (i.e., thershould be a moderate association rather thaa low or high association). Overall, thesanalyses provide initial evidence that relational aggression is a distinct construct, an

that, although related, it is relatively indepen den t of overt aggression.




sive (73.0% of the boys and 78.3% of thegirls). However, boys and girls were notevenly distributed among the remainingthree aggressive groups. Rather, the overtlyaggressive group consisted primarily of boys(15.6% of the boys vs. 0.4% of the girls); therelationally aggressive group consisted pri-marily of girls (17.4% of the girls vs. 2.0% ofthe boys); and the comb ined group consistedof both boys and girls (9.4% of the boys and3.8% of the girls). One implication of thesefindings is that, contrary to prior research,aggressive boys and girls may be identifiedwith almost equal frequency (27% of theboys vs. 21.7% of the girls in this study)

when relational as well as overt forms of ag-gression are assessed.

To assess further the relation betweengender and aggression, two analyses of vari-ance were conducted in which gender andgrade served as the independent variablesand children's scores for the relational ag-gression and overt aggression scales servedas the dependent variables. Both analysesyielded a significant main effect of gender,F( l , 483) = 7.8, p < .01, for relational ag-gression and F(l, 483) = 68.1, p < .001, forovert aggression . Specifically, girls (M = .42,SD = 3.4) were significantly more rela-tionally aggressive than boys (M = — .40, SD= 2.9) whereas boys (M = .77, SD = 3.1)were significantly more overtly aggressivethan girls (M = - 1 .0 9 , SD = 1.6). Th esefindings are consiste nt with the re sults of thedescriptive analyses previously described.

Relational Aggression andSocial-Psychological Adjustment

In order to assess the relation betweenrelational aggression and social-psycholog-ical adjustment, two sets of analyses wereperformed. First, analyses of covariancewere conducted in which relational aggres-sion group (two levels: relationally aggres-sive vs. nonrelationally aggressive) and sexserved as the independent variables, overtaggression served as the covariate, and thesocial-psychological adjustment indicesserved as the dep end ent variables (i.e., peernominations of acceptance, rejection, proso-cial behavior, and isolation/unhappiness;

self-reports of depression, loneliness, socialanxiety, social avoidance, and perc eption s ofpeer relations).^ Due to the moderate corre-lation between overt and relational forms ofaggression, children's overt aggressionscores were employed as a covariate to in-sure that any significant effects obtainedwere relatively independent of this form ofaggression.^ Student-Newman-Keuls posthoc tests (p < .05) w ere co nduc ted as appro-priate to investigate further significant ef-

fects (refer to Ta ble 2 for adjusted cell m eansand standard d eviations by relational aggres-sion group).

Peer nominations of status, prosocialbehavior, and isolation.—A nalyses of chil-dren's peer acceptance and rejection scoresyielded a significant main effect of relationalaggression group, F( l, 486) = 12.3, p < .01,for peer rejection. Specifically, relationally

^ Grade was initially included as an independent variable in these analyses (in order toavoid small cell sizes, the third and fourth graders were combined into one level of grade andthe fifth and sixth graders were combined into a second level of grade). However, with twominor exceptions, none of the interactions involving grade were significant, and thus grade wasexcluded from the presented analyses (i.e., grade main effects for the dependent variables stud-ied here have been assessed in prior research and were not of interest here). Both significantinteractions involving grade were from analyses of the Children's Peer Relations Scale. The firstwas the grade x relational aggression group x sex interaction for the caring subscale. In spectio nof cell means showed that the youngest (i.e., third and fourth grade), relationally aggressivemales reported less engagement in caring acts than did all other groups. The second interactionwas the grade x sex interaction for the peer isolation subscale. Inspection of cell means show edthat the youngest girls reported more isolation from peers than did the oldest girls and the boys.

^ A set of 2 (relational aggression group) x 2 (overt aggression group) ANOVAs were alsoconducted (sex could not be included as a factor because resulting cell sizes were too small in



Criek and Grotpeter 717TABLE 2

CELL MEANS AND STANDARD DEVIATIONS FOR SOCIAL-PSYCHOLOGICAL ADJUSTMENT

INDICES BY RELATIONAL AGGRESSION GROUP ADJUSTED FOR OVERT AGGRESSION

RELATIONAL AGGRESSION GROUP

MEASURE Nonaggressive

Peer nominations:Peer acceptance .0 (1.0)Peer rejection .0 (.8)Prosocial behavior .0 (4.5)Isolation - . 4 (2.8)

Self-reports:Depression 8.6 (7.4)Loneliness 29.9 (12.1)

Social anxiety 18.4 (5.3)Social avoid ance 12.2 (4.7)Perceived pee r acceptance 11.8 (3.0)Peer isolation 4.0 (1.9)Caring acts 14.0 (3.3)Overt aggression 7.2 (2.9)Relational aggression 9.1 (3.4)

NOTE.—Standard deviations are in parentheses.* p < .05.**p< .01.

Aggressive

.2

.4- . 2- . 1

10.434.619.512.511.24.4

13.27.39.6

(.9)(1.2)**(3.3)(2.1)

(6.1)*(14.1)*(4.7)(4.8)(3.2)*(2.1)*(3.4)(3.3)(4.3)

aggressive children were significantly moredisliked by peers than were their nonrela-tionally aggressive peers.

Analyses of children's peer-assessedprosocial behavior and isolation yielded asignificant main effect of sex, F(l, 486) =45.6, p < .001, and a significant interactionof sex and relational aggression group, F(l ,486) = 8.8, p < .01, for prosocial behavior.Specifically, girls (M = .62, SD = 4.8) were

viewed by peers as significantly more proso-cial than were boys (M = - .84, SD = 3.2).However, follow-up tests on the significantinteraction indicated that this effect variedas a function of relational aggression group.Specifically, nonaggressive girls (M = 1.47,SD = 5.1) were vie we d as significantly mo reprosocial than children in the other threegroups, nonaggressive boys (M = —1.53,SD = 3.2), aggressive girls (M = - . 2 2 , SD= 3.3), and aggressive boys (M = - . 1 5 , SD

= 2.9). In contrast, nonaggressive boys wereviewed as significantly less prosocial thanchildren in the other three groups. The pro-

interaction effect, F(l, 457) = 4.3, p < .0Specifically, relationally aggressive chil drenwere significantly more lonely than wertheir nonrelationally aggressive pee rs. However, follow-up analyses of the interactioneffect showed that the main effect was apparent for girls only. That is, relationallaggressive girls (M = 37.0, SD = 14.5reporte d significantly high er levels of loneliness than did nonrelationally aggressiv

boys (M = 31.0, SD = 12.7) and girls (M =28.8, SD = 11.4). In contrast, the lonelinesscores of relationally aggressive boys (M =32.2, SD = 12.5) did not differ from those otheir nonaggressive peers. The analysis ochildren's social anxiety scores and sociaavoidance scores did not yield significant effects. The analysis of children's responseto the Children's Depression Inventoryielded a significant main effect of relationaaggression group, F(l, 458) = 4.8, p < .0and a significant m ain effect of sex, F ( l , 458

= 4.2, p < .05. Specifically, relationally aggressive children reported significantl




children's perceived peer acceptance scoresyielded a significant main effect of relationalaggression group, F(l, 464) = 5.7, p < .05,and a significant interaction effect, F(l, 464)= 4.3, p < .05. Specifically, relationally ag-gressive children perceived themselves tobe more poorly accepted by peers than didtheir nonaggressive counterparts. However,this effect was qualified by the interactioneffect. Specifically, post hoc analyses indi-cated that relationally aggressive girls (M =10.6, SD = 3.0) reported poorer acceptanceby peers than did nonaggressive girls (M =12.0, SD = 3.0), nonaggressive boys (M =11.7, SD = 3.0), and relationally aggressive

boys (M = 11.7, SD = 2.6). In contrast, theperceived acceptance reported by rela-tionally aggressive boys did not differ fromthat reported by nonaggressive children.

Analysis of the peer isolation subscaleyielded a significant main effect of relationalaggression group, F(l, 464) = 4.9, p^< .05,and a significant relational aggression groupby sex interaction, F(l, 464) = 5.4, p < .05.That is, relationally aggressive children re-ported significantly greater isolation fromother children than did their peers. How-ever, this effect was qualified by the interac-tion effect. Specifically, follow-up testsshowed that relationally aggressive girls (M= 5.0, SD = 2.0) reported significantly moreisolation from peers than did nonaggressivegirls (M = 4.0, SD = 2.0) and boys (M =4.0, SD = 1.8, for nonaggressive and M =3.8, SD = 1.8 for aggressive boys). Analysisof the caring acts subscale produced a sig-nificant main effect of sex, F(l, 464) = 24.0,

p < .001 , w ith girls (M = 14.6, SD = 2.9)reporting significantly more engagement inprosocial acts than boys (M = 12.6, SD =3.5).

The ANOVA conducted on children'sself-reports of overt aggression and rela-tional aggression yielded a significant maineffect of sex for each variable, F(l, 464) =13.2, p < .001, and F(l, 464) = 5.7, p < .05,respectively. Boys reported significantlyhigher use of overt aggression (M = 7.6, SD= 3.1) and of relational aggression (M = 9.6,SD = 4.0) than did girls (M = 7.0, SD = 2.6

designed to assess the relation between rela-tional aggression and status group member-ship. Toward this end, an analysis of vari-ance was conducted in which status group(popular, average, neglected, rejected, con-troversial) served as the independent vari-able and children's relational aggressionscores (i.e., from the peer nomination instru-ment) served as the dependent variable(note that a covariate was not used for theseanalyses). This analysis yielded a signifi-cant effect for sociometric status, F(4, 362)= 11.6, p < .001. A Student-Newman-Keulspost hoc test (p < .05) indica ted that contro-versial status children were significantly

more relationally aggressive than all otherstatus groups, including rejected children(refer to Table 3 for cell means and stan-dard deviations). However, rejected chil-dren were significantly more relationallyaggressive than popular and neglected chil-dren. Also, neglected children were signifi-cantly less relationally aggressive than wereaverage status children. A comparable AN-OVA was conducted of children's overt ag-gression scores for comparison purposes.This analysis also yie lde d a significant effectof status group, F(4, 362) = 8.3, p < .001. AStudent-Newman-Keuls post hoc test (p <.05) indicated tha t controversial and rejectedchildren were significantly more overtly ag-gressive than popular, average, and ne-glected children, a finding that is consistentwith past research (see Coie et al., 1990, fora review).

Diseussion

Results of the pre sen t study provide evi-dence for the validity of a relational form ofaggression. As hypothesized, relational ag-gression appears to be relatively distinctfrom overt aggression, and it is significantlyrelated to gend er and to social-psychologicaladjustment in meaningful ways. These find-ings contribute uniquely to our understand-ing of childre n w ith adjustment difficulties,particularly young females.

As predicted, relational aggression ap-

pears to be more characteristic of girls thanof boys. Results ind icated that (1) as a group,



Criek and Grotpeter 71

TABLE 3

CELL MEANS AND STANDARD DEVIATIONS FOR RELATIONAL AGGRESSION AND OV ERT

AGGRESSION BY SOCIOMETRIC STATUS GRO UP

Relational AggressionStatus Group Score

Popular - .66(2 .9)Average -.11(2.9)Neglected -1 .36(1 .8)Rejected .76 (3.8)Controversial 2.82 (3.0)

NOTE.—Standard deviations are in parentheses.

Overt AggressionScore

- . 9 5 (1.8)- . 3 4 (2.4)- . 7 5 (1.9)1.11 (3.5)1.19 (3.2)

overtly aggressive than girls and were morelikely to be represented in the extremegroup of overtly aggressive children, find-ings that are consistent with prior researchon gender differences in aggression (seeBlock, 1983; Parke & Slaby, 1983, for re-views). The present study provides evi-dence that the degree of aggressiveness ex-hibited by girls has been underestimated inthese prior studies, largely because forms ofaggression relevant to girls' peer groups

have not been assessed.The paucity of research on girls' aggres-

sion may exist partly because of the com-plexity and subtleness of the behaviors in-volved, characteristics th at make them moredifficult to study than overt aggression. Forexample, reliably assessing overt aggressionin an interaction where one child hits an-other is significantly less complex than as-sessing relational aggression in an interac-tion where one child seems to exclude a

peer from an activity. To competently judg ethe latter interaction, knowledge is neededthat goes beyond the immediate situation(e.g., information about the relationship his-tory of the aggressive child involved so thatone can distinguish an excluded friend froma peer who simply never plays with the tar-get child). Thus, when assessing relationalaggression, the relevant behaviors may beoverlooked unless informants are employedwho can access information about friend-

ships and other relationships within the rel-evant peer group. This issue was addressedin the present study through the use of chil-

tional aggression, with items that loadehighly on each factor and cross-loaded insubstantially. Second, the classificationchildren into extreme groups of aggressivchildren showed that, although somethe children identified as high in aggressioexhibited both forms (i.e., the combinegroup), the majority of aggressive childreexhibited solely overt or relational formsaggression. Specifically, of the 121 childreidentified as high in aggression, only 27.3

(n = 33) exhibited both relational and oveforms of aggression. The majority of aggresive children (72.7%) exhibited either reltional or overt aggression, but not both. Iaddition, as will be discussed in more detabelow, relational aggression was signicantly relate d to social-psychological malajustment, independent of overt aggressioThese findings provide evidence that, athough overt and relational aggression alikely related constructs (i.e., because boconstitute harmful, aggressive acts), theeach provide uniqu e information about chdren's social behavior.

Findings from the peer-assessmentswell as from th e self-report ins trum ents indcate that, as has been found in past researcfor overtly aggressive children (ParkerAsher, 1987) relationally aggressive child realso experience significant social problemSpecifically, relationally aggressive child rewere significantly more disliked than othchildren. In addition, the peer status grouwho exhibited the highest levels of rel




aggression, the best known behavioralpredictor of peer rejection, was employed asa covariate). These findings indicate that re-lational aggression provides unique and im-portant information about children's socialdifficulties that cannot be accounted for byovert aggression alone.

It is possible that engaging in rela-tionally aggressive behaviors, because oftheir aversive nature, leads to being dislikedby peers. Support for this particular tempo-ral relation between rejection and aggres-sion has been established in past researchfor overt aggression (Coie & Kupersmidt,1983; Do dge, 1983). How ever, it is also con-ceivable that rejection by one's peers mayprecede relational aggression. For example,a rejected child may attempt to harm peers'relationships with others in an effort to com-pensate (or retaliate) for her own lack of suc-cess in those relationships. In either case,the association found between relational ag-gression and rejection significantly en-hances our knowledge of the social adjust-ment difficulties of girls. To date, relativelyfew studies have focused on the correlates

of rejection for girls (cf. Coie & Whidby,1986), and these studies have not estab-lished a relation between aggression and re-jection for young females (Coie & Whidby,1986; French , 1990). The pres ent study pro-vides initial evidence for such a relation.

The significant relation obtained be-tween relational aggression and controver-sial status group membership is particularlyinteresting. Controversial children, by defi-

nition, are highly disliked by some peersand highly liked by other peers. It is possi-ble that aggressive, controversial status chil-dren direct their relationally aggressive be-haviors disproportionately among theirpeers (i.e., so that some pe ers are frequentlyvictims of these behaviors whereas otherpeers are nev er victims of thes e acts). If so , itseems likely that controversial children mayreceive disliked nominations from peerswho have been the targets of their rela-tionally aggressive acts (e.g., children that

they exclude from peer interactions)whereas they may receive liked nom inations

relational aggression suggests that this groupof childre n m ay play a critical role in control-ling the structure and nature of peer groupinteractions (e.g., controlling who is in-cluded in peer activities; deciding who re-ceives social approval). Their popularitywith some peers may give these children the"social autibority" and control necessary tosuccessfully manipulate peer group relation-ships. Investigation of these hypotheses infuture research seems warranted, particu-larly since research on controversial chil-dren is lacking. One direction for futurestudy would be an assessment of specificperpetrator-victim relationships within the

peer group (e.g., to determine whether thedisliked nominations received by controver-sial children are provided by the peers thatthey victimize).

Findings from the self-report social-psychological adjustment instruments pro-vide further evidence that relational aggres-sion is significantly related to maladjustment(e.g., depression, loneliness, social isola-tion). These findings indicate that rela-tionally aggressive children feel unhappyand distressed about their peer relation-ships. These significant relations betweenpsychological maladjustment and relationalaggression were apparent even after level ofovert aggression was taken into account. Itmay be that frequent engagement in rela-tionally aggressive behaviors exacerbates, ifnot generates, feelings of social-psychologi-cal distress because these acts potentiallylimit children's access to peer relationships(e.g., excluding peers results in fewer peers

with which to play or interact). However, itmay also be that feelings of psychologicaldistress lead to engagement in relationalaggression. For example, children who feellonely or poorly accepted by peers may userelational aggression as a way to retaliateagainst peers (e.g., "You rejected me, nowI'll get even by rejecting you") or to makethemselves feel better (e.g., they may feelmore com peten t or in control if they excludeor put down others).

Results also demonstrate that the natureof the relation between social-psychological



Crick and Grotpeter 72lational aggression is predictive of future,

as well as concurrent, social-psychological

problems.

In sum, results of the present study pro-vide support for the hypothesis that, on aver-

age, both girls and boys are aggressive but

tend to exhibit distinct forms of the behavior

(relational aggression for girls and overt ag-

gression for boys). They also indicate that

further study of relational aggression is war-

ranted, particularly given that this form of

aggression is significantly associated with

social-psychological adjustment problems. It

will be important in future research to de-

velop further our understanding of the corre-lates, antecedents, and consequences of re-

lational aggression as well as knowledge of

the function it serves in children's peer

groups.

Referenees

Asher, S. R., & W heeler, V. A. (1985). G hildren'sloneliness; A comparison of rejected and ne-

glected p ee r status. JournaZ of Consulting andClinical Psychology, 53, 500-505.

Asher, S. R., & Williams, G. A. (April, 1987). Newapproaches to identifying rejected children at

school. In G. W. Ladd (Ghair), Identificationand treatment of socially rejected children in

school settings. Symposium c onducted at the

annual meeting of the American EducationalResearch Association, Washington, DC.

Berkowitz, L. (1993). Aggression: Its causes, con-sequences, and control. New York: AcademicPress.

Block, J. H. (1983). Differential premises arising

from differential socialization of the sexes:Some conjectures. Child Development, 54,

1335-1354.

Brehm, S. S., & Kassin, S. M. (1990). Social psy-

chology. Boston: Houghton Mifflin.

Bukowski, W. M., & Newcom b, A. F. (1984). Sta-

bility and determinants of sociometric statusand friendship choice: A longitudinal per-

spective. Developmental Psychology, 20,

941-952.

Cairns, R. B., Cairns, B. D., Neckerm an, H . J., Fer-

guson, L. L., & Gariepy, J. L. (1989). Growthand aggression: 1. Ghildhood to early adoles-cence. Developmental Psychology, 25, 320—

Goie, J. D., Dodge, K. A., & Kupersmidt, J.

(1990). Peer group beha vior an d social statuIn S. R. Asber & J. D. Goie (Eds.), Peer rejetion in childhood (pp. 17-59). New YoGambridge University Press.

Goie, J. D., & Kupersmidt, J. B. (1983). A behaioral analysis of emerging social statusboys' playgroups. Child Development, 5

1400-1416.

Goie, J. D., & Whidby, J. (April, 1986). Genddifferences in the basis for social rejectiin childhood. Paper presented at tbe annumeeting of the American Educational R

search Association, San Francisco.

Grick, N. R. (April, 1991). Subgroups of neglectand rejected children. Paper presented at tbiennial meeting of tbe Society for Researin Ghild Development, Seattle.

Grick, N. R., & Dodge, K. A. (1989). Ghildreperceptions of peer entry and conflict situtions: Social strateg ies, goals, and o utcom e epectations. In B. Schneider, J. Nadel, G.

tili, & R. We issberg (Eds.), Social competenin developmental perspective. Norwell, M

Kluwer.

Crick, N. R., & Do dge, K. A. (1994). A review a

reformulation of social-information-procsing mechanisms in children's social a

justment. Psychological Bulletin, 115, 7

101.

Crick, N. R., & Ladd, G. W. (1993). Ghildreperceptions of their peer experiences: Attbutions, loneliness, social anxiety, and socavoidance. Developmental Psychology, 2

244-254.

Dodge, K. A. (1980). Social cognition and chdren's aggressive behavior. Child Develo

ment, 51, 162-170.Dodge, K. A. (1983). Behavioral antecedents

peer social status. Child Development, 5

1386-1389.

Dodge, K. A., & Grick, N. R. (1990). Social infmation processing bases of aggressive behaior in children. Personality and Social Ps

chology Bulletin, 53, 1146-1158.

Dod ge, K. A., & Fram e, G. L. (1982). Social cogtive biases and deficits in aggressive boChild Development, 55, 163-173.

Eagly, A. H., & Steffen, V. J. (1986). Ge nder aaggressive behavior: A meta-analytic reviof the social psychological literature . Psych




Gonference on Ghild Development, Water-loo, Ontario.

French, D. G. (1990). Heterogeneity of peer-rejected girls. Child Development, 61 ,2028-2031.

Gormly, A. V., & Brodzinsky, D. M. (1993). Life-span human development. Orlando, FL: H olt,Rinehart, & Winston.

Hyde, J. S. (1986). Gender differences in aggres-sion. In J. S. Hyde & M. G. Linn (Eds.), Thepsychology of gender: Advances throughmeta-analysis. Baltimore: Johns H opkins Uni-versity Press.

Kenrick, D. T. (1987). Gender, genes, and the so-cial environment: A biosocial interactionist

perspective. In P. Shaver & G. Hendrick(Eds.), Sex and gender: Review of personalityand social psychology (Vol. 7). Beverly Hills,GA: Sage.

Kovacs, M. (1985). Th e G hildren's Dep ression In-ventory. Psychopharmacology Bulletin, 2 1 ,995-998.

Lagerspetz, K. M. J., Bjorkqvist, K., & Peltonen,T. (1988). Is indirect aggression more typicalof females? Ge nde r differences in aggressive-

ness in 11- to 12-year-old children. AggressiveBehavior, 14, 403-4 14.

Myers, D. G. (1990). Social psychology. NewYork: McG raw-Hill.

Parke, R. D ., & Slaby, R. G. (1983). Th e deve lop-ment of aggression. In E. M. Hetherington(Ed.), P. H. Mussen (Series Ed.), Handbookof child psychology: V ol. 4. Socialization,personality, and social development (pp.547-642). New York: Wiley.

Parker, J., & Asher, S. R. (1987). Peer acceptanceand later personal adjustment: Are low-accepted children "at risk"? PsychologicalBulletin, 102, 357-389.

Perry , D . G., Perry, L. G., & Rasm ussen, P . (1986).

Gognitive social learning m ediators of aggres-sion. Child Development, 57, 700-711.

Robins, L. N. (1986). The consequences of con-duct disorder in girls. In D . Olwe us, J. Block,& M. Radke-Yarrow (Eds.), Development ofantisocial arid prosocial behavior: Research,theories, and issues. New York: AcademicPress.

Vander Zanden, J. W. (1993). Human develop-ment. New York: McGraw-Hill.

Relational Aggression, Gender, And

Documents

Transcript of Relational Aggression, Gender, And