social cognitive theory journal article

17
A Social Cognitive Learning Theory of Homophobic Aggression Among Adolescents Gabriele Prati University of Bologna Abstract. The current study used social cognitive theory as a framework to investigate self-reported homophobic aggressive behavior at school. Participants included 863 students of 49 classes, enrolled in Grades 9 –13 in 10 Italian public high schools. The results from the multilevel mediation model (1–2–1) showed that class-level homophobic attitudes toward gay males mediated the relationship between student observations of peer homophobic aggression and self-reported engagement in homophobic aggression toward schoolmates perceived as gay. However, although student observations of peer aggression toward perceived lesbians predicted self-reported engagement in homophobic aggression toward perceived lesbians, this relationship appeared not to be mediated by class-level homophobic attitudes. Student observations of peer aggression toward perceived lesbians predicted the self-reported engagement in homophobic aggression to- ward perceived lesbians. It was found that the social cognitive perspective provided considerable insights into homophobic aggression at school. Consistent with this perspective, social and cognitive factors accounted for students’ ho- mophobic aggression. Students who identify as or who are perceived by their peers to be lesbian, gay, bisexual, or transgender may be considered at serious risk for victimization or homophobic bullying (Kosciw, Greytak, Diaz, & Bartkie- wicz, 2010; Rivers, 2011; Rivers & D’Augelli, 2001). The consequences of homophobic ag- gression have been documented with respect to mental health and educational attainment (D’Augelli, Pilkington, & Hershberger, 2002; Kosciw et al., 2010). Given the prevalence and the conse- quences of homophobic bullying, it is impor- tant to understand the factors contributing to this form of aggression occurring within schools. However, a clear theoretical basis for explaining homophobic aggression at school needs to be developed. The importance of developing a theoretical understanding of the dynamics of homophobic bullying is crucial not only for theoretical reasons, but also for practical reasons related to the design of ef- fective interventions (Swearer, Espelage, Vail- lancourt, & Hymel, 2010). In explaining the specific form of ag- gression at school labeled “bullying” (Olweus, 1993), different perspectives have been ad- opted, focusing on individual predictors and The author acknowledges the contributions of Marco Coppola, Marco Sacca `, and Rosario Murdica (from Arcigay) who supported the conduct of the survey. The author also thanks the school headmasters and all the students who took the time to complete the questionnaire. Correspondence regarding this article should be addressed to Gabriele Prati, University of Bologna, Faculty of Psychology, Viale Europa, 115, Cesena, FC 47521, Italy; e-mail: [email protected] Copyright 2012 by the National Association of School Psychologists, ISSN 0279-6015 School Psychology Review, 2012, Volume 41, No. 4, pp. 413– 428 413

Upload: jorge-barzanas

Post on 06-May-2015

538 views

Category:

Technology


3 download

TRANSCRIPT

A Social Cognitive Learning Theory of HomophobicAggression Among Adolescents

Gabriele PratiUniversity of Bologna

Abstract. The current study used social cognitive theory as a framework toinvestigate self-reported homophobic aggressive behavior at school. Participantsincluded 863 students of 49 classes, enrolled in Grades 9–13 in 10 Italian publichigh schools. The results from the multilevel mediation model (1–2–1) showedthat class-level homophobic attitudes toward gay males mediated the relationshipbetween student observations of peer homophobic aggression and self-reportedengagement in homophobic aggression toward schoolmates perceived as gay.However, although student observations of peer aggression toward perceivedlesbians predicted self-reported engagement in homophobic aggression towardperceived lesbians, this relationship appeared not to be mediated by class-levelhomophobic attitudes. Student observations of peer aggression toward perceivedlesbians predicted the self-reported engagement in homophobic aggression to-ward perceived lesbians. It was found that the social cognitive perspectiveprovided considerable insights into homophobic aggression at school. Consistentwith this perspective, social and cognitive factors accounted for students’ ho-mophobic aggression.

Students who identify as or who areperceived by their peers to be lesbian, gay,bisexual, or transgender may be considered atserious risk for victimization or homophobicbullying (Kosciw, Greytak, Diaz, & Bartkie-wicz, 2010; Rivers, 2011; Rivers & D’Augelli,2001). The consequences of homophobic ag-gression have been documented with respectto mental health and educational attainment(D’Augelli, Pilkington, & Hershberger, 2002;Kosciw et al., 2010).

Given the prevalence and the conse-quences of homophobic bullying, it is impor-tant to understand the factors contributing to

this form of aggression occurring withinschools. However, a clear theoretical basis forexplaining homophobic aggression at schoolneeds to be developed. The importance ofdeveloping a theoretical understanding of thedynamics of homophobic bullying is crucialnot only for theoretical reasons, but also forpractical reasons related to the design of ef-fective interventions (Swearer, Espelage, Vail-lancourt, & Hymel, 2010).

In explaining the specific form of ag-gression at school labeled “bullying” (Olweus,1993), different perspectives have been ad-opted, focusing on individual predictors and

The author acknowledges the contributions of Marco Coppola, Marco Sacca, and Rosario Murdica (fromArcigay) who supported the conduct of the survey. The author also thanks the school headmasters and allthe students who took the time to complete the questionnaire.

Correspondence regarding this article should be addressed to Gabriele Prati, University of Bologna,Faculty of Psychology, Viale Europa, 115, Cesena, FC 47521, Italy; e-mail: [email protected]

Copyright 2012 by the National Association of School Psychologists, ISSN 0279-6015

School Psychology Review,2012, Volume 41, No. 4, pp. 413–428

413

contextual predictors (for a meta-analytic re-view, see Cook, Williams, Guerra, Kim, &Sadek, 2010). Similarly, homophobia (a neg-ative attitude toward homosexuality) and ho-mophobic aggression or bullying (a set of be-haviors motivated by homophobia) can be ex-plained from either a sociological orpsychological perspective (see Parrott, 2008,for a review). Homophobia is conceptualizedas having a broad ego-defensive function,thereby protecting one’s unconscious anxietyabout experiencing homosexual impulses (Ad-ams, Wright, & Lohr, 1996). Herek (1986)underlines the social-expressive function ofhomophobia that defines group boundaries andmakes distinctions between male and femalegender roles. According to Herek (1986), an-other function of homophobia relates to thereinforcement of a positive sense of belongingto a group that includes hostility to homosex-uality. However, although these theoreticalframeworks are useful for understanding thepsychological and social determinants of ho-mophobia, they are limited. The main reasonis that the role of the context (e.g., peer rela-tionship, school climate) as well as the groupnorms in eliciting homophobia is partially ne-glected (Franklin, 2000; Rivers, 2011). Thesocial cognitive theory (Bandura, 1973, 1986)may be a useful framework to overcome theselimitations.

A Social Cognitive Perspective onHomophobic Aggression

The social learning perspective (Ban-dura, 1973, 1986) suggests that a combinationof environmental (social) and cognitive pro-cesses influence behavior. More specifically,according to social learning theory, the pro-cess of learning is predicted by the observationof models and, more generally, by social ex-periences. The likelihood of modeling is influ-enced by different conditions: (a) the modelshould be a powerful figure; (b) the conse-quences of the behavior of the model are re-ward rather than punishment; (c) the modelshould share similar characteristics with theobserver.

Empirical data have shown that bulliesare perceived by peers as powerful, popular,and leaders in their schools (e.g., Vaillancourt,Hymel, & McDougall, 2003), and that peersand teachers seldom punish bullies for theiraggressive behavior (e.g., Craig & Pepler,1997). At the same time, peers may reinforcethe aggressive behavior actively (e.g., joining)or passively through their attention, or beingrespectful and friendly to them (e.g.,O’Connell, Pepler, & Craig, 1999). It is rea-sonable to assume that, according to the sociallearning perspective (Bandura, 1973, 1986),becoming a bully is more likely under theaforementioned conditions. The same mecha-nisms apply even more to the determination ofhomophobic bullying for several reasons. Stu-dents tend to report that their teachers do notlabel homophobic epithets as bullying and donot punish the offenders (Phoenix, Frosh, &Pattman, 2003). Moreover, at least among ad-olescent males, homophobic behavior is usedto assert or demonstrate their heterosexualityand adherence to gender-normative prescrip-tions (Korobov, 2004; Pascoe, 2007; Plum-mer, 2001), and to achieve or maintain domi-nance among peers (Poteat & DiGiovanni,2010). Indeed, claiming heterosexuality andgender-normative behavior (especially mascu-linity) is generally associated with high statusand dominance (Phoenix et al., 2003; Poteat &DiGiovanni, 2010). These findings suggestthat the social learning perspective (Bandura,1973, 1986) may be appropriate to understandthe origin of bullying, and that a positive re-lationship may be expected between studentobservations of peer homophobic aggressionand the perpetration of homophobicaggression.

More recently, Bandura (1986) empha-sized the role of cognition, abstraction, andintegration of the information derived from avariety of social experiences, such as exposureto models. The process of weighing and syn-thesizing information derived from social ex-periences are crucial to the acquisition orlearning of attitudes, beliefs, and values. Thedevelopment of homophobic attitudes shouldbe of great relevance to an understanding ofthe process of attitudes acquisition following

School Psychology Review, 2012, Volume 41, No. 4

414

the observation of homophobic bullying be-haviors exhibited by others. Consistent withthese expectations, a previous study showedthat the student observations of peer ho-mophobic aggression exhibited by classmateswere associated with higher level of homopho-bia (Prati, Pietrantoni, & D’Augelli, 2011).

The number of episodes of aggressivebehavior reported by class members may berelated to moral standards or social groupnorms about the acceptance of homophobicaggression. An increase in homophobic ag-gressions in a specific class may lead to amodification of the moral agency, which is notonly individually situated, but also sociallysituated (Bandura, 1986). In the case of obser-vation of homophobic aggression, however,the elaboration of information provided by theobservation of classmates’ behavior should beevaluated at a group level for two reasons.First, group dynamic theories (e.g., Kelman1958; Sherif, 1936) emphasize the role ofgroup socialization pressures and conformityto group norms. For example, Kelman (1958)conceptualized three different processes of at-titude change: compliance, identification, andinternalization. The process of compliance isof special interest here—that is, when a personconforms to impress or gain favor/acceptancefrom another person or group (e.g., high statusand powerful bullies or their group). More ingeneral, all three different processes of con-formity are relevant for understanding howgroup belonging and interpersonal interactionsdetermine the development of prejudiced atti-tudes. Therefore, the peer group and the prox-imal social environment play an important rolein socializing homophobic attitudes (Poteat,2007).

Second, unlike other forms of aggres-sive behavior in school, homophobic aggres-sion is generally perpetrated by groups ofpeers, rather than by lone students (Rivers,2011). Indeed, in a sample of young malecollege students it was found that nearly threequarters of cases of aggression toward gayindividuals were perpetrated in the context ofa group (Franklin, 2000). According to Askewand Ross (1988), homophobic behaviorstrengthens each member of the group’s own

heterosexual identity. These data clearly indi-cate that the group context is important inunderstanding the cognitive operations in-volved in this particular type of social experi-ence. Moreover, because group-level ho-mophobic attitudes are related to norms aboutthe acceptance of peer victimization, and in-volve evaluations about outgroup membersand targets of aggression (Poteat, Espelage, &Green, 2007), it is possible to hypothesize thatthe experiences of witnessing homophobic ag-gression may contribute to the homophobicclimate of the classroom (cross-level directeffects).

Social cognitive theory (Bandura, 1986)is an extension of social learning theory andfocuses on the ways in which cognitive oper-ations on social experiences are thought toinfluence behavior. These cognitive processesmay explain why students learn aggressivebehavior by observing the actions of others(e.g., peers). This influence may be mediatedby levels of homophobic attitudes, becausethey express hostility or overt aggression to-ward gay males and lesbians (Herek, 2000,2009). According to Rivers (2011), homopho-bic bullying is a product of dominant institu-tions and groups that reinforce homophobicattitudes. Homophobia, in turn, as an ideologyof heterosexual domination, may lead to de-humanization, ascription of blame, and distor-tion of injurious consequences, which alsoplay a heavy role in aggressive behavior (Ban-dura, 1986). Indeed, homophobic attitudes areassociated with homophobic behavior (Frank-lin, 2000; Poteat, 2007, 2008; Poteat, Kimmel,& Wilchins, 2011; Rivers, 2011); therefore,they may be conceptualized as prebullyingbeliefs or beliefs that guide behavior. Atti-tudes and perceptions of bullying play a role inpredicting student’s future bullying behavior(Salmivalli & Voeten, 2004).

In the present study, influence of class-level homophobic attitudes on homophobicaggression was examined. Various areas ofresearch have shown that the social context ofadolescent peer groups may account for indi-viduals’ bullying attitudes and behavior (Es-pelage, Holt, & Henkel, 2003; Salmivalli &Voeten, 2004; Sutton & Smith, 1999). Social

Explanation of Homophobic Aggression

415

norms may dictate legitimacy or illegitimacyof behaviors: the legitimacy of discriminationagainst homosexuals increases the likelihoodof homophobic aggression, such as gay bash-ing (Bahns & Branscombe, 2011). Group-level homophobic attitudes are typical of thenature of the social climate of peer groupsduring adolescence (Plummer, 2001) and maybe considered an indicator of the group norm(Poteat, 2007). Moreover, group-level ho-mophobic attitudes have been found to berelated to homophobic aggressions (Poteat,2008). Given that exposure to homophobicaggression may influence class homophobicattitudes, and that class homophobic attitudesmay predict homophobic aggression, it is pos-sible to hypothesize that the relationship be-tween observation of homophobic behaviorsand the perpetration of homophobic bullyingis explained by an increase in class homopho-bic attitudes. This hypothesis implies cross-level indirect effects of student observations ofpeer homophobic aggression on homophobicaggression via class homophobic attitudes.

Conceptual Multilevel Mediation Modeland Hypotheses

The conceptual multilevel mediationmodel employed in this study is presented inFigure 1 (Panel A). The main hypothesis ofthis study was that the relationship betweenstudent observations of peer homophobic ag-gression and self-reported perpetration of ho-mophobic aggression would be mediated byclass homophobic attitudes. This mediationmodel involves data that vary both within (stu-dents) and between higher level units(classes). All the mediation analyses were per-formed for the use of homophobic languageand homophobic aggressions toward per-ceived gay males and lesbians separately. Therationale for this is based on the fact thatexperiences of harassment and assault basedon sexual orientation (both perpetration andvictimization) are more likely among malesthan females (e.g., Kosciw et al., 2010; Prati etal., 2011). Moreover, there are clear differ-ences in individual homophobic attitudes to-ward gay males and lesbians (Herek, 2000,

2009; Kite & Whitley, 1996). Because the useof homophobic remarks is by far the mostcommon (e.g., Kosciw et al., 2010) and mostinvestigated (e.g., Poteat, 2007, 2008; Poteat& Espelage, 2005; Poteat & Rivers, 2010)form of homophobic bullying, it was also de-cided to distinguish between homophobic lan-guage and other forms (in this article calledhomophobic aggression). The distinction isbased on the fact that the use of homophobiclanguage, although related to aggressive be-havior, may be used in other contexts to con-vey other meanings without an antagonisticintent (e.g., to emphasize one’s own hetero-sexuality to impose gender-normative behav-iors; Korobov, 2004; Phoenix et al., 2003;Plummer, 2001; Poteat & Rivers, 2010).Therefore, four mediational analyses wereconducted to examine the specificity of eachtarget (gay males and lesbians) and of eachform of homophobic bullying (homophobicaggression and homophobic language).

Methods

Participants

Participants consisted of 863 students inGrades 9–13 in from 49 classes in 10 Italianpublic high schools. In Italy, high school stu-dents typically spend the whole school day (ormost of it) with the same group of students.The sample includes students from Lyceums,a preparatory school for the university, (n �470, 54.5%) and from vocational institutes(n � 393, 45.5%). Of the participating stu-dents, 39.3% (n � 326) were males. Partici-pants ranged in age from 15 through 22 years(M � 17.26, SD � 1.59). A total of 23.5%(n � 203) of the participants were in Grade 9,20.6% (n � 178) in Grade 10, 22.5% (n �194) in Grade 11, 18.3% (n � 158) inGrade 12, and 15.1% (n � 130) in Grade 13.Participants are similar to the general popula-tion of students in Italian high schools withrespect to demographic characteristics, al-though males were slightly underrepresented.

Procedure

We selected a stratified random sampleof 20 Italian public high schools. The stratifi-

School Psychology Review, 2012, Volume 41, No. 4

416

cation of the Italian school was based on twovariables: first-level Nomenclature of Territo-rial Units for Statistics (NUTS) and types ofschools differentiated by subjects and activi-

ties. In Italy, there are five first-level NUTSregions (North West, North East, Center,South, And Islands), and two main types ofsecondary schools: Lyceums (classic, scien-

Figure 1. Multilevel mediation model (1–2–1) of homophobic bullying (PanelA) and related parameters tested (Panel B) following the MSEM procedure(Preacher et al., 2010).

Explanation of Homophobic Aggression

417

tific, linguistic, pedagogic), and vocational in-stitute (technical and professional). The num-ber of students per school was not consideredfor stratification because the Italian schoolsystem guarantees a rather homogeneous num-ber of students per school and per class (av-erage number of students per school is 500 andper class 20).

A total of 10 (50%) school headmastersgave their permission to conduct the study intheir schools, after being contacted by letterand by telephone. The main reason given bythose who refused to participate was time con-straint. The recruitment rate had no effect onthe distribution of schools by region or typesof secondary schools. Once permission wasobtained, five classes were randomly selectedin each school corresponding to Grades 9–13.

The survey was administered in January2010, using a paper-and-pencil questionnaire.Participants completed the survey in class.They were informed that the objective of thestudy survey was to learn more about theirperspectives on specific social issues and theiractivities at school. Students were also in-formed about the confidentiality of their re-sponses and assured that all the reports fromthe study would only be presented in aggre-gate. Each student provided informed consent.No student refused to participate in the study.Survey proctors assisted participants by indi-vidually answering any question raised whileavoiding influencing the answers in any way.Furthermore, they ensured confidentiality ofresponses. Proctors were eight adult male (agerange 23–34 years) research assistants andmembers of the project staff who were trainedto administer the questionnaire. After the ad-ministration, participants had the opportunityto discuss the topic of homophobic bullyingwith the proctors.

Measures

The questionnaire consisted of threescales. Data from the present study were usedto calculate the Cronbach’s alpha coefficients.Additional psychometric data are presented byPrati (2012).

Homophobic language. The Ho-mophobic Content Agent Target Scale (Poteat& Espelage, 2005) was used to measure theextent which students called other studentshomophobic epithets (Agent subscale; 5items) during the past 30 days. The Homopho-bic Content Agent Target Scale was translatedwith the help a bilingual student into Italianfor use in the study. Following this translation,measures were back translated into English byanother bilingual student and compared withthe original versions. The Agent subscale wasemployed twice, one for homophobic epithetstoward male students (homophobic languagetoward males; � � .75, using data from thepresent study) and the other one toward femalestudents (homophobic language toward fe-males; � � .74, using data from the presentstudy).

Homophobic bullying. The Ho-mophobic Bullying Scale (Prati, 2012) wasused to measure students’ bullying behaviorsmotivated by homophobia. More specifically,participants were asked to report whether theyobserved and engaged in different homopho-bic behaviors (written offenses, isolation/ex-clusion, spreading rumors or lies, homophobicteasing, property theft or damage, physicalassault, sexual harassment, electronic harass-ment or cyberbullying) in their schools in thepast 30 days. Response options include Never(1), 1 or 2 times (2), about once a week (3),and more than once a week (4). Four subscaleswere used: self-reported homophobic aggres-sion toward supposed gay men (� � .82, usingdata from the present study), self-reported ho-mophobic aggression toward supposed lesbi-ans (� � .87, using data from the presentstudy), student observations of peer aggres-sion toward supposed gay men (� � .82, usingdata from the present study), and student ob-servations of peer aggression toward lesbians(� � .82, using data from the present study).

Classroom homophobic attitudes.Homophobic attitudes toward gay males andlesbians were assessed using the Italian Scaleof Homophobia (Falanga, Parisi, & Di Chi-acchio, 2006). In the validation study, an ex-

School Psychology Review, 2012, Volume 41, No. 4

418

ploratory principal components analysis wasused to examine dimensionality of the scale.Two dimensions were found, which differen-tiated between negative attitudes toward gaymales (28 items) and toward lesbians (28items). The following are examples of itemsincluded in this scale: “Being gay (or lesbian)is a mental illness” and “Gay males (or lesbi-ans) should not be allowed to join the army.”Response options on a 5-point Likert-typescale range from 1 (strongly disagree)through 5 (strongly agree), with higher scoresreflecting more negative attitudes. The coeffi-cient alpha reliability estimates of the ho-mophobic attitudes toward gay males and les-bian for the current study were .94 and .93,respectively. Following Poteat (2007, 2008),to measure homophobic climate participants’homophobic attitudes toward gay men andlesbians scores were aggregated across class-rooms (for each classroom, it was computedaverage of individual scores for homophobicattitudes).

Analyses

In the present study, four mediationmodels were tested: the model predicting ho-mophobic language toward males, self-re-ported homophobic aggression toward per-ceived lesbians, homophobic language towardfemales, and self-reported homophobic ag-gression toward perceived gay males.

The results of the tests of kurtosis (De-Carlo, 1997) showed that the measures of self-reported perceptions of aggressive behavior,self-reported homophobic aggression were notnormally distributed. Inspection of the curverevealed that these variables had a floor effect.This floor effect was expected, because ho-mophobic aggression is restricted to a minor-ity of students (Kosciw et al., 2010). There-fore, a censored-inflated model together with arobust maximum likelihood estimator wasconducted in Mplus 6 (Muthen & Muthen,1998–2010), which does not require the as-sumption of normality, and yields robust esti-mates of asymptotic covariances of parameterestimates. Following Graham’s (2009) recom-mendation, missing data estimation was em-

ployed using maximum likelihood imputationprocedure. To test the multilevel mediationalpaths, procedures for implementing MSEMfor 1–2–1 designs were followed (Preacher,Zyphur, & Zhang, 2010). The MSEM ap-proach avoids problems of conflated individ-ual-level and class-level effect, because vari-ance is decomposed into two components. Theindependent and the dependent variables areassessed both at Level 1 and Level 2, but themediator is assessed at Level 2 only. TheMSEM parameters used to test the study hy-potheses are depicted in Figure 1 (Panel B). Ina 1–2–1 design, the mediator variable variesonly between clusters—thus only the Betweenindirect effect exists. In this type of design,there are two c paths: between (c�B) andWithin (c�W), along with a Between a path(aB), and a Between b path (bB). The Betweenindirect effect is aBbB. Because of the sam-pling distribution of the indirect effect inMSEM, the Monte Carlo-based method wasemployed (MacKinnon, Lockwood, & Wil-liams, 2004), implemented in R by Selig andPreacher (2008), to obtain the appropriate con-fidence intervals. A mediation effect is indi-cated when the indirect effect is significant.The significance of the indirect effects at the.05 level is supported if the confidence inter-vals for the estimates exclude zero. Studentgender and age served as control variables inLevel 1 of each model.

Results

Aggregation Analyses and Gender andType of School Differences

For all variables, a significant between-class variance was observed: self-reported per-ceptions of aggressive behavior toward gaymales (ICC � .12), self-reported perceptionsof aggressive behavior toward lesbians(ICC � .10), attitudes toward gay males(ICC � .22), attitudes toward lesbians (ICC �.16), self-reported aggressive behavior towardgay males (ICC � .10), self-reported aggres-sive behavior toward lesbians (ICC � .10),homophobic language toward gay males(ICC � .06), and homophobic language to-ward lesbians (ICC � .05). This means that

Explanation of Homophobic Aggression

419

class membership explains from 5% to 22% ofthe variance in the variables (Hox, 2010).Given the existence of between-school vari-ance, the development of a multilevel modelwas warranted.

Table 1 summarizes the means and stan-dard deviations for gender and type of schooldifferences. Because the data were non-nor-mally distributed, Mann–Whitney U test wasused. Significant gender differences werefound on all measures (with the exception ofself-reported homophobic aggression towardperceived lesbians). Significant differences bytype of school were found for student obser-vations of peer aggression toward perceivedgay males, and homophobic attitudes.

Self-Reported Homophobic AggressionToward Perceived Gay Males

As shown in Table 2, self-reported ho-mophobic aggression toward perceived gaymales was predicted by student observationsof peer aggression toward perceived gay males(at the individual level), gender (male), andhomophobic attitudes. The 1–1 indirect effectwas statistically significant and indicates that therelationship between student observations ofpeer aggression toward perceived gay males andself-reported homophobic aggression towardperceived gay males was mediated by class ho-mophobic attitudes toward gay males. The pro-portion of variance explained (analogous to themultiple R2) at the first level (see Hox, 2010)was 0.41, which means that 41% of the varianceat the student level is explained by the model.The proportion of variance explained at the sec-ond level (see Hox, 2010) was 0.96. Althoughthis value may seem high, it should be notedthat the between-class variance was markedlylower than the within-class variance. There-fore, the model explained a high proportion ofthe low between-class variation.

Self-Reported Homophobic AggressionToward Perceived Lesbians

Preliminary analysis showed that the re-lationship between class homophobic attitudestoward lesbians and student observations ofpeer aggression toward perceived lesbians was

not significant (� � 0.12, p � .05). Therefore,the path leading from student observations ofpeer aggression toward perceived lesbians toattitudes toward lesbians was not estimated (aswell, mediation analysis was not performed).As illustrated in Table 2, self-reported ho-mophobic aggression toward perceived lesbi-ans was predicted by student observations ofpeer aggression toward perceived lesbians (in-dividual level), class homophobic attitudes to-ward lesbians, and younger age. The propor-tion of variance explained at the first levelwas 0.57, while the proportion of varianceexplained at the second level was 0.86.

Homophobic Language Toward Males

As shown in Table 2, homophobic lan-guage toward males was predicted by studentobservations of peer aggression toward per-ceived gay males (both class level and indi-vidual level), gender (male), older age, andhomophobic attitudes. The 1–2–1 indirect ef-fect was statistically significant and suggeststhat the relationship between student observa-tions of peer aggression toward perceived gaymales and homophobic language toward maleswas mediated by class homophobic attitudestoward gay males. The proportion of varianceexplained at the first level was 0.28, while theproportion of variance explained at the secondlevel was 0.64.

Homophobic Language Toward Females

The relationship between homophobiclanguage toward females and attitudes towardlesbians was not significant (� � .02, p � .05).Therefore, attitudes toward lesbians were notincluded in this model. As illustrated in Ta-ble 2, homophobic language toward femaleswas predicted by student observations of peeraggression toward perceived lesbians (individ-ual level). The proportion of variance ex-plained at the first level was 0.22, while theproportion of variance explained at the secondlevel was 0.64.

Discussion

Several consequences of homophobicaggression at school have been documented

School Psychology Review, 2012, Volume 41, No. 4

420

Tab

le1

Dif

fere

nce

sby

Gen

der

and

Typ

eof

Sch

ool

( UM

ann

–Wh

itn

eyT

est)

Mal

epa

rtic

ipan

tsFe

mal

epa

rtic

ipan

tsL

yceu

ms

Voc

atio

nal

inst

itute

MSD

MSD

Zp

rM

SDM

SDZ

pr

Stud

ent

obse

rvat

ions

ofpe

erag

gres

sion

tow

ard

gay

mal

es1.

560.

541.

450.

43�

2.28

.02

�.0

81.

590.

571.

420.

39�

3.50

�.0

01�

.12

Stud

ent

obse

rvat

ions

ofpe

erag

gres

sion

tow

ard

lesb

ians

1.22

0.39

1.18

0.30

�2.

59.0

1�

.09

1.27

0.41

1.19

0.27

�1.

72.0

9�

.06

Atti

tude

sto

war

dga

ym

ales

2.82

0.77

2.19

0.67

�11

.70

�.0

01�

.41

2.53

0.82

2.36

0.73

�3.

74�

.001

�.1

3A

ttitu

des

tow

ard

lesb

ians

2.48

0.68

2.25

0.70

�5.

21�

.001

�.1

82.

390.

702.

300.

69�

2.87

�.0

01�

.10

Self

-rep

orte

dag

gres

sion

tow

ard

gay

mal

es1.

180.

351.

060.

16�

7.21

�.0

01�

.25

1.15

0.34

1.07

0.14

�1.

45.1

5�

.05

Self

-rep

orte

dag

gres

sion

tow

ard

lesb

ians

1.09

0.31

1.04

0.12

�0.

72.4

7�

.03

1.09

0.29

1.04

0.12

�0.

19.8

5�

.01

Hom

opho

bic

lang

uage

tow

ard

mal

es1.

870.

761.

410.

47�

9.58

�.0

01�

.33

1.65

0.71

1.53

0.56

�1.

83.0

7�

.06

Hom

opho

bic

lang

uage

tow

ard

fem

ales

1.22

0.45

1.26

0.40

�3.

67�

.001

�.1

31.

260.

461.

230.

38�

0.09

.93

.00

Explanation of Homophobic Aggression

421

with respect to mental health and educationalattainment: depression, anxiety, school failure,dropout, decreased levels of self-esteem, in-creased health risk behaviors, such as sub-stance abuse and attempted suicide (e.g.,D’Augelli et al., 2002; Kosciw et al., 2010;Poteat & Espelage, 2007). The aim of thisstudy was to examine the psychosocial dy-namics that underlie homophobic bullying be-haviors. More specifically, drawing on socialcognitive theory (Bandura, 1973, 1986), it waspredicted that student observations of peer ho-mophobic aggression would be related to ahigher probability to report engagement in ho-mophobic aggression and that this relationshipwould be mediated by class-level homophobicattitudes.

The results showed that student obser-vations of peer aggression toward perceivedgay males exhibited by classmates was asso-ciated with class homophobic attitudes that, inturn, were related to self-reported homophobicaggression toward perceived gay males or tothe use of homophobic language toward

males. Although self-reported homophobicaggression toward perceived lesbians and theuse of homophobic language toward femaleswere predicted by the student observations ofpeer aggression toward perceived lesbians ex-hibited by classmates, these relationships werenot mediated by class homophobic attitudes.

Consistent with social cognitive theory,behaviors appear to be learned within thegroup. Students, who reported they observedmore episodes of homophobic bullying, weremore likely to engage in homophobic aggres-sion. An explanation for this may lie in thedominance and high status of a significantportion of students who engage in homopho-bic bullying. Empirical evidence showed thatthose students tend to be perceived by peers aspopular and dominant (Phoenix et al., 2003;Poteat & DiGiovanni, 2010). Therefore, ac-cording to social cognitive theory (Bandura,1973, 1986), homophobic bullies are morelikely to influence the behavior of peers, be-cause they may be perceived as powerful fig-

Table 2Parameter Estimates for All Models

Path Estimate (SE) 95% CI Path Estimate (SE) 95% CI

Self-reported homophobic aggression towardperceived gay males Homophobic language toward males

aB 4.07 (0.25)*** 3.58, 4.55 aB

1.484 (0.697)* 0.12, 2.85bB 0.48 (0.23)* 0.03, 0.92 b

B0.600 (0.048)*** 0.51, 0.69

c’B 1.00 (1.01) �1.99, 1.99 c’B

�0.161 (0.088) �0.35, 0.021-2-1 1.94 (0.89)* 0.16, 3.78 1-2-1 0.89 (0.49)* 0.14, 3.82

indirect effect indirect effectc’W 0.62 (0.07)*** 0.48, 0.76 c’W 0.48 (0.032)*** 0.412, 0.54

d �0.16 (0.06)* �0.28, �0.03 d �0.20 (0.03)*** �0.26, �0.15e �0.01 (0.01) �0.01, 0.023 e 0.02 (0.00)*** 0.02, 0.03

Self-reported homophobic aggression towardperceived lesbians Homophobic language toward females

bB 0.99 (0.29)*** 0.43, 1.56 c’B

16.27 (9.09) �1.55, 34.08c’B 1.78 (1.65) �1.45, 5.02 c’

W0.86 (0.24)*** 0.39, 1.34

c’W 0.67 (0.07)*** 0.54, 0.80 d 0.11 (0.07) �0.02, 0.24d �0.07 (0.05) �0.16, 0.02 e �0.02 (0.02) �0.05, 0.01e �0.01 (0.00)* 0.00, 0.02

Note: * � p �.05; ** � p �.01; *** � p �.001. Significance tests of indirect effects were conducted using the MonteCarlo-based method implemented in R by Selig and Preacher (2008).

School Psychology Review, 2012, Volume 41, No. 4

422

ures who share similar characteristics with theobserver.

In line with this framework, anotherpossible (nonalternative) explanation is thatthe consequences of the observed homophobicaggression were reward rather than punish-ment. Previous studies showed that studentaggression was reinforced actively or pas-sively by peers through their attention, respect,and friendliness (e.g., O’Connell, Pepler, &Craig, 1999). In the case of homophobic ag-gression at school, students tend to report thatschool personnel did not frequently intervene(punishment) when homophobic remarks weremade (Kosciw et al., 2010 Phoenix et al.,2003), and that use of homophobic languagewas rewarded (e.g., asserting her or his het-erosexuality and gender-normative behavior,maintaining dominance; Korobov, 2004; Pas-coe, 2007; Plummer, 2001). Following theagentic perspective of social cognitive theory,students actively process information, inferthe causal structure of the social context, syn-thesize that information into abstract form,and control their behavior to enable them tosecure a desired outcome. Therefore, studentsmay learn homophobic aggressions to achievedesired outcomes such as dominance, adher-ence to the gender-normative behavior normsprescribed by peers, acknowledgment of statusand power.

Besides the fact that the homophobicbullies may be perceived as powerful or ashaving rewards, the present study showed thatinformation derived from attending to ho-mophobic aggression may be elaborated in away that increases their negative attitudes to-ward perceived gay males or lesbians. Accord-ing to Bandura (1986), individuals form theirattitudes on the information derived from so-cial experiences. In line with a previous study,student observations of peer homophobic ag-gression were associated with homophobia(Prati et al., 2011). Moreover, following socialcognitive theory, homophobic attitudes, de-rived from observing homophobic bullying,influence the behavior.

The novelty of this study lies in thedemonstration of the influence of class-levelhomophobic attitudes, not individual ho-

mophobic attitudes. In this study, an incrementin class-level homophobic attitudes followingthe student observations of peer homophobicaggression was found to increase the probabil-ity of engaging in similar behaviors. Follow-ing the notion of reciprocal interplay of indi-vidual and social influences (Bandura, Bar-baranelli, Caprara, & Pastorelli, 1996), thecapacity for observational learning enablesstudents to acquire (homophobic) attitudesthrough their experiences and to pass on therelevant ones to classmates by social model-ing. It seems appropriate to speculate that theepisodes of homophobic aggression reportedby class members could influence moral stan-dards or social norms about the acceptance ofhomophobic aggression. Adopting the socialcognitive theory of moral disengagement(Bandura, 1986; Bandura et al., 1996), classmembers may infer that homophobic aggres-sion they are exposed to is an appropriatebehavior. Therefore, class members may re-vise their moral standards and disengage fromculturally sanctioned beliefs about the ho-mophobic aggression toward perceived gaysor lesbians. Indeed, social norms about theacceptance of homophobic aggression are theproduct of the selective activation and disen-gagement of moral self-regulation that justifiesnegative or mean behavior and reduces socialand self-punishment. In the case of homopho-bic aggression, a change in moral standardsand social norms may be best expressed byhigher homophobic attitudes at the class level(homophobic class climate), given that theyexpress hostility toward homosexual people(Herek, 2000, 2009).

A more homophobic class climate, inturn, predicted the engagement in homophobicaggression, in line with the idea that ideolog-ical orientations of the group shape the form ofnorms and moral standards that may legitimizecertain behaviors. Indeed, when the social cli-mate seems to legitimate discriminationagainst homosexuals, individuals are morelikely engage in homophobic aggression(Bahns & Branscombe, 2011).

The explanations that have been consid-ered so far do not entirely apply to self-re-ported homophobic aggression toward per-

Explanation of Homophobic Aggression

423

ceived lesbians, because class homophobic at-titudes did not mediate the expectedrelationships. The role of class homophobicattitudes toward perceived lesbians was lessimportant, probably because attitudes towardperceived gay males are consistently morehostile than their attitudes toward lesbians(Kite & Whitley, 1996). Gender-associatednorms are more rigidly defined for men thanfor women, and socialization pressures men toroutinely validate their masculinity, which re-sults in more negative attitudes toward gaymales than lesbians (Herek, 2000, 2009; Kite& Whitley, 1996). Indeed, there is empiricalevidence that male and female students differin terms of negative attitudes and aggressivebehavior toward sexual minorities (Prati et al.,2011). Moreover, in this study (as in otherstudies, see Kosciw et al., 2010) the number ofhomophobic aggressions toward perceivedlesbians observed by participants was small.Thus, there have not been sufficient occasionsin which participants may learn homophobicattitudes toward lesbians by observing ho-mophobic aggression toward perceivedlesbians.

It should be noted that the percentage ofvariance explained by the model was high forself-reported homophobic aggression towardperceived gay males and lesbians, but waslower for the use of homophobic language.These results highlight the fact that homopho-bic epithets are not directed exclusively to-ward perceived gay males and lesbians (Kim-mel & Mahler, 2003; Plummer, 2001; Poteat& Espelage, 2005). Indeed, the HomophobicContent Agent Target scale (Poteat & Espel-age, 2005) specifies neither the contexts andcircumstances nor the intent for the use ofthese epithets. For example, among boys ho-mophobic language is used to impose gender-normative behaviors, to prove masculinity andheterosexuality without an aggressive intent,to reach or maintain dominance among peers,or as a joking matter within peer groups(Korobov, 2004; Pascoe, 2007; Phoenix et al.,2003; Poteat & DiGiovanni, 2010). Therefore,homophobic attitudes may be less relevant tohomophobic language, as students may not becompletely aware of how its use may distress

or offend others, especially sexual minoritieswho frequently or often hear homophobic re-marks at school (Poteat & DiGiovanni, 2010).

Limitations and Future Directions

Several limitations of the study need tobe considered. First, the generalizability of thecurrent findings is unknown given that half ofthe school headmasters contacted gave theirpermission to conduct the study in theirschools. However, the characteristics of par-ticipating schools were similar to those ofschools who chose not to participate.

Further, because of to the cross-sec-tional study design it is not possible to infercausality in the relationships among variables.Although the model tested here was theoreti-cally driven, other, alternative models mightfit the data equally as well. For example,whereas in this model attitude preceded be-havior, from different perspectives, attitudesmay conceivably follow from behavior.

The model tested in this study did notinclude directly the role played by school per-sonnel. For example, on the one hand, there isevidence showing that gay, lesbian, bisexual,and transgender students report victimizationfrom school personnel (Chesir-Teran, 2003).On the other hand, school personnel play animportant role improving the social climate intheir own schools (Espelage, Aragon, Birkett,& Koenig, 2008; Kosciw et al., 2010).

Finally, the model tested did not accountfor a large proportion of variance in the use ofhomophobic epithets. Moreover, the dynamicsof homophobic aggression toward perceivedlesbians were not adequately understood bythe proposed mediational model. Future stud-ies could examine different models to explainthe use of homophobic epithets and homopho-bic aggression toward lesbians.

It should be noted that although socialcognitive theory is a useful framework espe-cially in explaining peer relationships as theyrelate to homophobia, it is limited. For a com-plete understanding of the factors involved inhomophobic aggression, other sociological orpsychological perspectives should be adopted(for a review, see Parrott, 2008). For example,

School Psychology Review, 2012, Volume 41, No. 4

424

according to a social-ecological model of bul-lying (Espelage & Swearer, 2004; Espelage &Swearer, 2010), children and adolescents areaffected by a range of nested contextual sys-tems of families, schools, peer groups, teach-er–student relationships, parent–child rela-tionships, parent–school relationships, neigh-borhoods, and cultural expectations. Thesocial cognitive approach used in the presentstudy investigated the influence of a small partof the systems included in this model. There-fore, the inclusion of other theoretical ap-proaches is needed to achieve a comprehen-sive understanding of homophobic bullyingdynamics.

Implications for Prevention

Given the critical role played by ho-mophobic class climate, education to promoterespect for sexual minorities could benefitfrom a sexual prejudice reduction component.Furthermore, homophobic attitudes may becontrasted with practices and policies thatmake coming out a safe option for all the staffand students in schools. Indeed, the intergroupcontact hypothesis (Allport 1954; Pettigrew &Tropp, 2006) suggests that knowing a gay orlesbian person leads to less sexual prejudiceand improved attitudes. Recently, the researchliterature provided further evidence showingthat intergroup contact and intergroup friend-ship are related to sexual prejudice reduction(Heinze & Horn, 2009; Hodson, Harry, &Mitchell, 2009).

Based on the current findings and thereview of the literature, to reduce homophobicaggression, attention is needed to both indi-viduals and their social environment. Con-cerning the latter, potential implications areapparent for the importance of the adoptionand the implementation of comprehensive pro-tective policies for sexual minorities that ex-plicitly enumerate sexual orientation, genderidentity, and gender expression. Indeed,school faculty and other school staff at schoolswith antiharassment/bullying policies that in-corporated gender identity/expression and/orsexual orientation were more likely to inter-vene when hearing homophobic language ex-

pressed among students (Kosciw et al., 2010).In addition, students were more likely to re-port episodes of homophobic behaviors toschool authorities when they occurred.

Systematic intervention by school per-sonnel when homophobic aggression occursprovides an alternative role model for students(Bandura, 1986) and challenges the legitimacyof discrimination against homosexuals (an ef-fective strategy for reducing homophobic be-havior, Bahns & Branscombe, 2011). There-fore, schools should develop explicit ho-mophobic bullying policies, and schoolfaculty and other school staff should imple-ment them. To this end, school staff should betrained to respond every time they witnesshomophobic behaviors and to encourage stu-dents to intervene on behalf of victims withoutbecoming bystanders to homophobic bullying.

These results highlight the need not onlyfor intervention when homophobic remarksare used, but also to address homophobia as afactor that, at the class level, influences thisbehavior. However, given that at the classlevel homophobic attitudes were not associ-ated with the use of homophobic language,educational programs and intergroup dia-logues could increase the awareness of thenegative consequences of its use (Poteat &DiGiovanni, 2010). One way to enable stu-dents to understand and respect difference anddiversity is to integrate sexual orientation,gender identity, and gender expression into thecurriculum; schools should consider ways inwhich these issues can be discussed in school,in a constructive way.

Other implications may be derived fromthe literature on school-bullying interventions.The effect of antibullying programs on stu-dents, teachers, and school staff primarily re-flected favorable changes in knowledge, atti-tudes, and self-perceived competency in deal-ing with bullying (Merrell, Gueldner, Ross, &Isava, 2008). For example, the findings sug-gest that different social contexts (e.g., classes,peer groups) influence significantly bullyingbehaviors. Thus, a successful strategy for ad-dressing bullying should be directed towardchanging contexts, such as the positive peerreporting intervention at school (Ruth, Miller,

Explanation of Homophobic Aggression

425

& Friman, 1996). Furthermore, among school-based antibullying programs, whole-school/multidisciplinary interventions, including theestablishment of school-wide rules and conse-quences for bullying, were reported to be themost promising in reducing bullying and vic-timization (Vreeman & Carroll, 2007). Fi-nally, given the negligible effect sizes for de-sired changes in student self-reports of bothvictimization and perpetration of antibullyingprograms (Smith, Schneider, Smith, & Ana-niadou, 2004), several recommendations havebeen proposed (Swearer et al., 2010). First,antibullying programs should be grounded in aguiding theoretical framework that would in-form program development and evaluation.The present study showed the usefulness ofsocial cognitive theory. Second, interventionsshould be directed at the social ecology thatstimulates and tolerates aggressive behaviors,such as group norms, as shown in the presentpaper. Third, antibullying programs should in-clude factors such as gender, race, disability,and sexual orientation. The final aim of theseefforts is to make school a safe place for all thestudents, regardless of their perceivedcharacteristics.

In conclusion, this is the first study thatadopted the social cognitive learning theory asa theoretical basis for explaining homophobicaggressive behavior at school. Results of thisstudy provided evidence indicating the socialcontext to be a significant factor accountingfor students’ of homophobic aggressionbehavior.

References

Adams, H. E., Wright, L. W., & Lohr, B. A. (1996). Ishomophobia associated with homosexual arousal?Journal of Abnormal Psychology, 105, 440–445. doi:10.1037/0021-843X.105.3.440

Allport, G. W. (1954). The nature of prejudice. Cam-bridge, MA: Addison-Wesley.

Askew, S., & Ross, C. (1988). Boys don’t cry: Boys andsexism in education. Milton Keynes, UK: Open Uni-versity Press.

Bahns, A. J., & Branscombe, N. R. (2011). Effects oflegitimizing discrimination against homosexuals ongay bashing. European Journal of Social Psychology,41, 388–396. doi:10.1002/ejsp.784

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

Bandura, A. (1986). Social foundations of thought andaction: A social cognitive theory. Englewood Cliffs,NJ: Prentice-Hall.

Bandura, A., Barbaranelli, C., Caprara, G. V., & Pas-torelli, C. (1996). Mechanisms of moral disengage-ment in the exercise of moral agency. Journal of Per-sonality and Social Psychology, 71, 364–374. doi:10.1037/0022-3514.71.2.364

Cook, C. R., Williams, K. R., Guerra, N., Kim, T. E., &Sadek, S. (2010). Predictors of bullying and victimiza-tion in childhood and adolescence: A meta analyticinvestigation. School Psychology Quarterly, 25, 65–83. doi:10.1037/a0020149

Craig, W., & Pepler, D. (1997). Observations of bullyingand victimization in the schoolyard. Canadian Journalof School Psychology, 13, 41–60.

D’Augelli, A. R., Pilkington, N. W., & Hershberger, S. L.(2002). Incidence and mental health impact of sexualorientation victimization of lesbian, gay, and bisexualyouths in high school. School Psychology Quar-terly, 17, 148–167. doi:10.1521/scpq.17.2.148.20854

DeCarlo, L. T. (1997). On the meaning and use of kurto-sis. Psychological Methods, 2, 292–307. doi: 10.1037/1082-989X. 2.3.292

Espelage, D. L., Aragon, S. R., Birkett, M., & Koenig,B. W. (2008). Homophobic teasing, psychological out-comes, and sexual orientation among high school stu-dents: What influence do parents and schools have?School Psychology Review, 37, 202–216.

Espelage, D. L., Holt, M. K., & Henkel, R. R. (2003).Examination of peer-group contextual effects on ag-gression during early adolescence. Child Develop-ment, 74, 205–220. doi:10.1111/1467-8624.00531

Espelage, D. L., & Swearer, S. M. (Eds.). (2004). Bullyingin American schools: A social-ecological perspectiveon prevention and intervention. Mahwah, NJ: Erl-baum.

Espelage, D. L., & Swearer, S. M. (2008). Addressingresearch gaps in the intersection between homophobiaand bullying. School Psychology Review, 37, 155–159.

Espelage, D. L., & Swearer, S. M. (2010). A social-ecological model for bullying prevention and interven-tion: Understanding the impact of adults in the socialecology of youngsters. In S. R. Jimerson, S. M.Swearer, & D. L. Espelage (Eds.), Handbook of bully-ing in schools: An international perspective (pp. 61–72). New York, NY: Routledge.

Falanga S., Parisi A., & Di Chiacchio C. (2006). Omofo-bia. Una ricerca empirica italiana [Homophobia. Anempirical research in Italy]. In D. Rizzo (Ed.), Omo-sapiens (pp. 61–68). Rome, Italy: Carocci.

Franklin, K. (2000). Antigay behaviors among youngadults. Journal of Interpersonal Violence, 15, 339–362. doi:10.1177/088626000015004001

Graham, J. W. (2009). Missing data analysis: Making itwork in the real world. Annual Review of Psychol-ogy, 60, 549–576. doi:10.1146/annurev.psych.58.110405.085530

Heinze, J. E., & Horn, S. S. (2009). Intergroup contact andbeliefs about homosexuality in adolescence. Journal ofYouth and Adolescence, 38, 937–951. doi:10.1007/s10964-009-9408-x

Herek, G. M. (1986). On heterosexual masculinity: Somepsychical consequences of the social construction ofgender and sexuality. American Behavioral Scien-tists, 29, 563–577. doi:10.1177/000276486029005005

School Psychology Review, 2012, Volume 41, No. 4

426

Herek, G. M. (2000). Sexual prejudice and gender: Doheterosexuals’ attitudes toward lesbians and gay mendiffer? Journal of Social Issues, 56, 251–266. doi:10.1111/0022-4537.00164

Herek, G. M. (2009). Sexual prejudice. In T. Nelson (Ed.),Handbook of prejudice, stereotyping, and discrimina-tion (pp. 439–465). New York, NY: Psychology Press.

Hox, J. J. (2010). Multilevel analysis: techniques andapplications (2nd ed.). New York, NY: Routledge.

Kimmel, M. S., & Mahler, M. (2003). Adolescent mas-culinity, homophobia, and violence. American Behav-ioral Scientist, 46, 1439–1458. doi:10.1177/0002764203046010010

Kite, M. E., & Whitley, B. E. (1996). Sex differences inattitudes toward homosexual persons, behaviors, andcivil rights: a meta-analysis. Personality and SocialPsychology Bulletin, 22, 336–353. doi:10.1177/0146167296224002

Korobov, N. (2004). Inoculating against prejudice: A dis-cursive approach to homophobia and sexism in ado-lescent male talk. Psychology of Men and Masculin-ity, 5, 178–189. doi:10.1037/1524–9220.5.2.178

Kosciw, J. G., Greytak, E. A., Diaz, E. M., & Bartkiewicz,M. J. (2010). The 2009 National School Climate Sur-vey: The experiences of lesbian, gay, bisexual andtransgender youth in our nation’s schools. New York,NY: GLSEN.

MacKinnon, D. P., Lockwood, C. M., & Williams, J.(2004). Confidence limits for the indirect effect: Dis-tribution of the product and resampling methods. Mul-tivariate Behavioral Research, 39, 99–128. doi:10.1207/s15327906mbr3901_4

Merrell, K. W., Gueldner, B. A., Ross, S. W., & Isava,D. M. (2008). How effective are school bullying inter-vention programs? A meta-analysis of intervention re-search. School Psychology Quarterly, 23, 26–42.

Muthen, L. K., & Muthen, B. O. (1998–2010). Mplususer’s guide (6th ed.). Los Angeles, CA: Muthen &Muthen.

O’Connell, P., Pepler, D., & Craig, W. (1999). Peer in-volvement in bullying: insights and challenges for in-tervention. Journal of Adolescence, 22, 437–452. doi:10.1006/jado.1999.0238

Olweus, D. (1993). Bullying at school. What we know andwhat we can do. Oxford, UK: Blackwell Publishers.

Parrott, D. J. (2008). A theoretical framework for antigayaggression: Review of established and hypothesizedeffects within the context of the general aggressionmodel. Clinical Psychology Review, 28, 933–951. doi:10.1016/j.cpr.2008.02.001

Pascoe, C. J. (2007). Dude, you’re a fag: Masculinity andsexuality in high school. Los Angeles, CA: Universityof California Press.

Pepler, D., & Craig, W. (1995). A peek behind the fence:Naturalistic observations of aggressive children withremote audiovisual recording. Developmental Psychol-ogy, 31, 548–553. doi:10.1037/0012-1649.31.4.548

Pettigrew, T. F., & Tropp, L. R. (2006). A meta-analytictest of intergroup contact theory. Journal of Personal-ity and Social Psychology, 90, 751–783. doi:10.1037/0022-3514.90.5.751

Phoenix, A., Frosh, S., & Pattman, R. (2003). Producingcontradictory masculine subject positions: Narrativesof threat, homophobia and bullying in 11–14 year oldboys. Journal of Social Issues, 59, 179–195. doi:10.1111/1540-4560.t01-1-00011

Plummer, D. C. (2001). The quest for modern manhood:Masculine stereotypes, peer culture and the social sig-nificance of homophobia. Journal of Adolescence, 24,15–23. doi:10.1006/jado.2000.0370

Poteat, V. P. (2007). Peer group socialization of ho-mophobic attitudes and behavior during adolescence.Child Development, 78, 1830–1842. doi:10.1111/j.1467-8624.2007.01101.x

Poteat, V. P. (2008). Contextual and moderating effects ofthe peer group climate on use of homophobic epithets.School Psychology Review, 37, 188–201.

Poteat, V. P., & DiGiovanni C. D. (2010). When biasedlanguage use is associated with bullying and domi-nance behavior: The moderating effect of prejudice.Journal of Youth and Adolescence, 39, 1123–1133.doi:10.1007/s10964-010-9565-y

Poteat, V. P., & Espelage, D. L. (2005). Exploring therelation between bullying and homophobic verbal con-tent: The Homophobic Content Agent Target (HCAT)Scale. Violence and Victims, 20, 513–528.

Poteat, V. P., & Espelage, D. L. (2007). Predicting psy-chosocial consequences of homophobic victimizationin middle school students. Journal of Early Adoles-cence, 27, 175–191. doi:10.1177/0272431606294839

Poteat, V. P., Espelage, D. L., & Green, H. D. (2007). Thesocialization of dominance: Peer group contextual ef-fects on homophobia and dominance attitudes. Journalof Personality and Social Psychology, 92, 1040–1050.doi:10.1037/0022-3514.92.6.1040

Poteat, V. P., Kimmel, M. S., & Wilchins, R. (2011). Themoderating effects of support for violence beliefs onmasculine norms, aggression, and homophobic behav-ior during adolescence. Journal of Research on Ado-lescence, 21, 434–447. doi:10.1111/j.1532-7795.2010.00682.x

Poteat, V. P., & Rivers, I. (2010). The use of homophobiclanguage across bullying roles during adolescence.Journal of Applied Developmental Psychology, 31,166–172. doi:10.1016/j.appdev.2009.11.005

Prati, G. (2012). Development and psychometric proper-ties of the Homophobic Bullying Scale. Educationaland Psychological Measurement, 72, 649–664. doi:10.1177/0013164412440169

Prati, G., Pietrantoni, L., & D’Augelli A. R. (2011).Aspects of Homophobia in Italian high schools: Stu-dents’ attitudes and perceptions of school climate.Journal of Applied Social Psychology, 41, 2600–2620.doi:10.1111/j.1559-1816.2011.00842.x

Preacher, K., Zyphur, M., & Zhang, Z. (2010). A generalmultilevel SEM framework for assessing multilevelmediation. Psychological Methods, 15, 209–233. doi:10.1037/a0020141

Rivers, I. (2011). Homophobic bullying. Research andtheoretical perspectives. New York, NY: Oxford Uni-versity Press.

Rivers, I., & D’Augelli, A. R. (2001). The victimization oflesbian, gay, and bisexual youths: Implications forintervention. In A. R. D. Augelli & C. J. Patterson(Eds.), Lesbian, gay, and youths: Psychological per-spectives (pp. 199–223). New York, NY: Oxford Uni-versity Press.

Ruth, E. A., Miller, P. M., & Friman, P. C. (1996). Feedthe hungry bee: Using positive peer reports to improvethe social interactions and acceptance of a sociallyrejected girl in residential care. Journal of Applied

Explanation of Homophobic Aggression

427

Behavior Analysis, 29, 251–253. doi:10.1901/jaba.1996.29-251

Salmivalli, C., & Voeten, M. (2004). Connections be-tween attitudes, group norms, and behaviour in bul-lying situations. International Journal of BehavioralDevelopment, 28, 246 –258. doi:10.1080/01650250344000488

Selig, J. P., & Preacher, K. J. (2008). Monte Carlo methodfor assessing mediation: An interactive tool for creat-ing confidence intervals for indirect effects [Computersoftware]. Retrieved from http://quantpsy.org/

Sherif, M. (1936). The psychology of social norms. NewYork, NY: Harper & Row.

Smith, J. D., Schneider, B. H., Smith, P. K., & Ananiadou,K. (2004). The effectiveness of whole school antibul-lying programs: A synthesis of evaluation research.School Psychology Review, 33, 547–560.

Sutton, J., & Smith, P. K. (1999). Bullying as a groupprocess: An adaptation of the participant role approach.Aggressive Behavior, 25, 97–111. doi:10.1002/(SICI)1098-2337(1999)25:2�97::AID-AB3�3.0.CO;2-7

Swearer, S. M., Espelage, D. L., Vaillancourt, T., &Hymel, S. (2010). What can be done about schoolbullying? Linking research to educational practice. Ed-ucational Researcher, 39, 38–47.

Swearer, S. M., Turner, R. K., Givens, J. E., & Pollack,W. S. (2008). “You’re so gay!” Do different forms ofbullying matter for adolescent males? School Psychol-ogy Review, 37, 160–173.

Vaillancourt, T., Hymel, S., & McDougall, P. (2003).Bullying is power: Implications for school-based inter-vention strategies. Journal of Applied School Psychol-ogy, 19, 157–176. doi:10.1300/J008v19n02_10

Vreeman, R. C., & Carroll, A. E. (2007). A systematicreview of school-based interventions to prevent bully-ing. Archives of Pediatric and Adolescent Medicine,161, 78–88.

Date Received: February 12, 2012Date Accepted: August 28, 2012

Action Editor: Robert J. Volpe �

Gabriele Prati is a fellow researcher at the University of Bologna (Italy). He received adoctoral degree in social and organizational psychology from the University of Bologna.He teaches social psychology and work psychology at the University of Bologna.

School Psychology Review, 2012, Volume 41, No. 4

428

Reproduced with permission of the copyright owner. Further reproduction prohibited withoutpermission.