descripcion de un burro de caballo

16
TeachersSelf-Efficacy, Perceived Effectiveness Beliefs, and Reported Use of Cognitive-Behavioral Approaches to Bullying Among Pupils: Effects of in-Service Training With the I DECIDE Program Michael J. Boulton University of Chester Despite the promise of being effective in tacking bullying and conduct disorder, cognitive-behavioral (C-B) interven- tions are underused by teachers. Little detailed information exists as to why this is the case. The current study with junior school teachers in the U.K. (N = 249) confirmed this low reported usage and showed that while teachers tended to believe that C-B approaches would be effective, most held rather low self-efficacy beliefs. Attending a workshop on a specific C-B approach, the I DECIDE program had positive effects on perceived effectiveness and self-efficacy beliefs, and longer durations of training (3 days) were more beneficial than shorter durations (half/1 day). In line with outcome-expectancy theory and the theory of planned behavior, self-efficacy and duration of training predicted an increase in reported usage of I DECIDE across an 8-month period, and self-efficacy fully mediated the association between duration of training and increase in reported usage. The implications of these findings for overcoming impediments to the more widespread use of C-B approaches by teachers to tackling bullying were discussed, particularly the notion that attending training of sufficient duration coupled with a more explicit attention on fostering self-efficacy will pay dividends. Keywords: bullying; victimization; teachers; cognitive-behavioral; self-efficacy BULLYING INVOLVES REPEATED ATTEMPTS by a more powerful individual (or group) to hurt, humiliate, upset, or otherwise cause distress to a less powerful individual (or group) (Olweus, 1993). It occurs frequently in schools (Rose, Espelage, & Monda- Amaya, 2009; Solberg & Olweus, 2003), leads to considerable distress among victims (Reijntjes, Kamphuis, Prinzie & Telch, 2010; Vaughn, Fu, Bender, DeLisi, Beaver, Perron, & Howard, 2010), and compromises their classroom concentration and academic achievement (Boulton, Trueman, & Murray, 2008; Buhs, Ladd, & Herald, 2006). Bullying also disrupts the smooth running of schools (Kochenderfer & Ladd, 1996), and perpe- trators are at elevated risk for psychiatric disorders (Kazdin & Weisz, 2003) and continuation of aggression beyond childhood (Olweus, 1993). Prompted by this evidence, a growing number of antibullying interventions have been developed (Samara & Smith, 2008). Evaluation studies report some but limited success (Frey, Edstrom, & Hirschstein, 2005; Samples, 2004; Smith, Pepler, & Rigby, 2004; Smith & Shu, 2002), and meta-analyses show that no initiative has come close to eradicating bullying in schools (Merrell, Gueldner, Ross, & Isava, 2008; Ttofi & Farring- ton, 2011). While it is important to recognize that bullying is not a disorder, DSM-5 sees it as contributing to the diagnosis of conduct disorder, along with frequent or violent aggressive behavior more generally, and it has been found to be associated with the latter in an epidemiological study (Vaughn et al., 2010). Thus, in the review that follows, we utilize research on conduct Available online at www.sciencedirect.com ScienceDirect Behavior Therapy 45 (2014) 328 343 www.elsevier.com/locate/bt Address correspondence to Michael J. Boulton, Ph.D., Depart- ment of Psychology, University of Chester, Parkgate Road, Chester CH1 4BJ, UK; e-mail: [email protected]. 0005-7894/45/328-343/$1.00/0 © 2014 Association for Behavioral and Cognitive Therapies. Published by Elsevier Ltd. All rights reserved.

Upload: ricard-sendra-sanroma

Post on 21-Dec-2015

236 views

Category:

Documents


2 download

DESCRIPTION

De los sumerios vienen muchas cosas como un programa increible, como esque me dejes decargar de una puta vez éste libro, gracias.

TRANSCRIPT

Available online at www.sciencedirect.com

ScienceDirectBehavior Therapy 45 (2014) 328–343

www.elsevier.com/locate/bt

Teachers’ Self-Efficacy, Perceived Effectiveness Beliefs, andReported Use of Cognitive-Behavioral Approaches to Bullying

Among Pupils: Effects of in-Service Training With theI DECIDE Program

Michael J. BoultonUniversity of Chester

Despite the promise of being effective in tacking bullyingand conduct disorder, cognitive-behavioral (C-B) interven-tions are underused by teachers. Little detailed informationexists as to why this is the case. The current study withjunior school teachers in the U.K. (N = 249) confirmed thislow reported usage and showed that while teachers tendedto believe that C-B approaches would be effective, most heldrather low self-efficacy beliefs. Attending a workshop on aspecific C-B approach, the I DECIDE program had positiveeffects on perceived effectiveness and self-efficacy beliefs,and longer durations of training (3 days) were morebeneficial than shorter durations (half/1 day). In line withoutcome-expectancy theory and the theory of plannedbehavior, self-efficacy and duration of training predictedan increase in reported usage of I DECIDE across an8-month period, and self-efficacy fully mediated theassociation between duration of training and increase inreported usage. The implications of these findings forovercoming impediments to the more widespread use ofC-B approaches by teachers to tackling bullying werediscussed, particularly the notion that attending training ofsufficient duration coupled with a more explicit attention onfostering self-efficacy will pay dividends.

Keywords: bullying; victimization; teachers; cognitive-behavioral;self-efficacy

Address correspondence to Michael J. Boulton, Ph.D., Depart-ment of Psychology, University of Chester, Parkgate Road, ChesterCH1 4BJ, UK; e-mail: [email protected]/45/328-343/$1.00/0© 2014 Association for Behavioral and Cognitive Therapies. Published byElsevier Ltd. All rights reserved.

BULLYING INVOLVES REPEATED ATTEMPTS by a morepowerful individual (or group) to hurt, humiliate,upset, or otherwise cause distress to a less powerfulindividual (or group) (Olweus, 1993). It occursfrequently in schools (Rose, Espelage, & Monda-Amaya, 2009; Solberg & Olweus, 2003), leadsto considerable distress among victims (Reijntjes,Kamphuis, Prinzie & Telch, 2010; Vaughn, Fu,Bender, DeLisi, Beaver, Perron, & Howard, 2010),and compromises their classroom concentrationand academic achievement (Boulton, Trueman, &Murray, 2008; Buhs, Ladd, & Herald, 2006).Bullying also disrupts the smooth running ofschools (Kochenderfer & Ladd, 1996), and perpe-trators are at elevated risk for psychiatric disorders(Kazdin & Weisz, 2003) and continuation ofaggression beyond childhood (Olweus, 1993).Prompted by this evidence, a growing number ofantibullying interventions have been developed(Samara & Smith, 2008). Evaluation studies reportsome but limited success (Frey, Edstrom, &Hirschstein, 2005; Samples, 2004; Smith, Pepler,& Rigby, 2004; Smith & Shu, 2002), andmeta-analyses show that no initiative has comeclose to eradicating bullying in schools (Merrell,Gueldner, Ross, & Isava, 2008; Ttofi & Farring-ton, 2011). While it is important to recognize thatbullying is not a disorder, DSM-5 sees it ascontributing to the diagnosis of conduct disorder,along with frequent or violent aggressive behaviormore generally, and it has been found to beassociated with the latter in an epidemiologicalstudy (Vaughn et al., 2010). Thus, in the reviewthat follows, we utilize research on conduct

329t eacher s ’ and cogn i t i v e - b ehav ioral approaches to bully ing

disorder where there is little direct evidence forbullying per se.It is vital that teachers are included in studies

seeking to understand and prevent bullying (Bau-man, Rigby & Hoppa, 2008; Kochenderfer-Ladd,& Pelletier, 2008), not least because they may beineffective in dealing with it (Fekkes, Pijpers, &Verloove-Vanhorick, 2005; Newman & Murray,2005; Rigby & Barnes, 2002; Smith & Shu, 2002).Lack of training has been implicated (Bauman et al.,2008). Studies have solicited teachers’ views,including their beliefs about bullying and howpupils may avoid it (Craig, Henderson, &Murphy,2000; Troop & Ladd, 2002), perceived ability tocope (Boulton, 1997), perceptions of “what works”(Dake, Price, Telljohann, & Funk, 2003), andtypical patterns of dealing with incidents (Bauman,& Del Rio, 2006; Bauman et al., 2008; Boulton,Hardcastle, Down, Fowles, & Simmonds, in press;Dake et al., 2003).Bullying has a complex etiology (Marsh, Parada,

Craven, & Finger, 2004) and it takes manydifferent forms (direct physical, verbal and psycho-logical assaults through to indirect attacks viaelectronic media). Roles other than perpetratorand victim, such as supporters and onlookers, addto this complexity (Kärnä, Voeten, Poskiparta &Salmivalli, 2010). It is unsurprising, then, that oftenteachers feel confused, are unsure of what to do,and act ineffectively (Gerber & Solari, 2005).Available evidence suggests that cognitive-behav-

ioral interventions (i.e., based on cognitive-behav-ioral therapy or its derivatives, denoted by “C-Bapproaches” henceforward) are used relativelyinfrequently (Bauman et al., 2008; Samara &Smith, 2008; Smith et al., 2004; Smith, Smith,Osborn, & Samara, 2008; Ttofi & Farrington,2011). Put simply, C-B approaches focus onchanging internal thinking and affective processesas a way to influence actual behavior, and build onthe seminal work of Beck (1976) and Meichenbaum(1977), among others. C-B approaches have figuredquite prominently in efforts to ameliorate conductdisorders (Apsche & Bass, 2006; Gerber & Solari,2005; Ghafoori & Tracz, 2004). The MetropolitanArea Research Group (2002, 2007) interventiontargets the social cognitions of aggressive responsesto provocations, aggressive fantasy, and normativebeliefs that support aggression. Teachers lead 40focused 1-hour lessons over a 2-year period. TheCoping Power Program (Lochman et al., 2009;Lochman & Wells, 2003) involves 34 structuredsessions that teach young people thinking skills,such as dealing with provocations and anger, andidentifying more socially acceptable responses.Aggression Replacement Training (Goldstein,

Glick, & Gibbs, 1998) incorporates three elements:Skillstreaming, which aims to build prosocial skills;anger-control training, which helps youth moderatetheir “hot” emotional responses; and moral reason-ing training, which orients them towards sociallyacceptable moral codes and beliefs. It is typicallydelivered in a 10-week block. Evaluations of theseprograms have indicated moderate benefits, in linewith those from Ozabaci’s (2011) review.C-B approaches are not central to the well-

known Olweus Bullying Prevention Program(Olweus & Limber, 2010). But given their wider,and to some extent successful, application totackling other forms of aggression and to conductdisorder per se (Ozabaci, 2011), it is reasonable toexpect that C-B approaches may usefully beapplied to bullying. Hence, we thought it impor-tant to look at teachers’ perceptions and use ofC-B-based interventions to address the latter.Moreover, teachers often fail to implement anti-bullying interventions in the way that trainingproviders would like, and this in turn affects if andhow pupils’ bullying-related attitudes and behav-ior change (Salmivalli, Kaukiainen, & Voeten,2005).A few studies have examined teachers’ percep-

tions and uses of many different kinds of antibully-ing approaches. Samara and Smith (2008) collecteddata from 148 schools (presumably one member ofstaff responded per school) about their level ofsatisfaction with some 27 different antibullyinginterventions and found almost all were “moder-ate.” Naylor and Cowie (1999), who askedteachers what they thought would be the benefitsof peer support in tackling bullying, found thatmany regarded it as communicating a caring ethos.However, neither of these studies tested if satisfac-tion was associated with actual or reported usage,and the questions asked were rather vague.Moreover, it seems that C-B approaches havebeen relatively neglected in this regard; it was notincluded in Samara and Smith’s (2008) extensivelist of 27 different antibullying interventions thatthey studied. Hence, the initial aim of the currentstudy was to provide the first detailed informationon this issue. Specifically, Aim 1 was to assess (a)teachers’ perceived effectiveness beliefs concerningthe use of C-B approaches to peer bullying(“perceived effectiveness” henceforward), theirself-efficacy beliefs for using them (“self-efficacy”henceforward), and their reported use of them(“usage” henceforward); (b) the effects of attendinga training workshop on a specific C-B approach (IDECIDE) on those two kinds of beliefs (regardlessof duration); and (c) the effects of the duration ofthe I DECIDE training on those beliefs.

330 boulton

While furthering our understanding of teachers’beliefs is clearly important, Gerber and Solari(2005) have made it clear that more needs to bedone to understand why teachers do or do notactually utilize C-B approaches to tackle unaccept-able behavior, including bullying. Hence, weinvestigated this broad issue. The rationale for ourmore specific hypotheses and research questionswill now be provided.

teachers’ self-efficacy beliefs, per-ceived effectiveness of antibullyingapproaches and training

In a recent review, Han and Weiss (2005) identifiedfactors that influence teachers’ take up and use ofinterventions targeted at pupils’ behavioral andemotional functioning. It showed that teachers’self-efficacy, perceived effectiveness beliefs, andlevel and type of training all played an importantrole. Below, we relate each of these factors in moredetail to our current study.Teachers’ professional practices are influenced by

their self-efficacy beliefs (Emmer & Hickman,1991; Gibson & Dembo, 1984; Han & Weiss,2005; Wertheim & Leyser, 2002). Conflictingresults have been reported in the context ofantibullying; some evidence suggests teachersgenerally have a high level of self-efficacy forantibullying action (Bradshaw, Sawyer, &O’Brennan, 2007), but other evidence suggests theopposite (Beran, 2005). Novick and Isaacs (2010)found that teachers’ feelings of preparedness toaddress bullying influenced their reports of whatthey would do if they encountered it. Bradshaw etal. (2007) found that teachers with highest self-efficacy beliefs reported higher levels of interveningin actual episodes, but the data were limited by across-sectional design and the reports were ofanticipated rather than actual behavior. Self-effi-cacy was assessed with a single item: “I haveeffective strategies for handling a bullying situa-tion.”While extant studies are important, none haslooked at teachers’ self-efficacy beliefs concerningC-B approaches to bullying per se, and Bauman etal. (2008) recently called for studies of whatteachers actually do that can be linked longitudi-nally with their beliefs. Again, data are lacking andwe aimed to fill this gap.What teachers believe will be effective in address-

ing bullying has also been linked to their reports ofwhat they actually do (Dake et al., 2003; Gerber &Solari, 2005). This resonates with outcome-expec-tancy theory (Hall, Herzberger, & Skowronski,1998; Williams, Anderson, & Winett, 2005) thatitself is founded upon the Theory of PlannedBehavior (TPB; Ajzen, 1991, 2002). These theories

suggests that behavioral, normative, and controlbeliefs collectively influence an individual’s intentionto perform a given behavior. Within the behavioralbeliefs component of the TPB are outcome expec-tancies, beliefs that a particular outcome willfollow certain behavior. Thus, teachers who havehigh perceived effectiveness beliefs concerning C-Bapproaches should be more likely to actuallyimplement them.Antibullying training is also important. After

training, teachers were significantly less likely toignore bullying (Bauman et al., 2008) and enactedmore proactive antibullying work in class (Dakeet al., 2003). Training in C-B-based approachesto conduct disorders (i.e., not bullying per se) hasalso been shown to affect amount and quality of useof those approaches (Lochman et al., 2009;Packenham, Shute, & Reid, 2004; Sasso, Reimers,Cooper, Wacker, & Berg, 1992). This evidence sitswell with theories of how and why practitionerstake up and use novel interventions as these theoriesstress the need for them to receive an appropriatelevel of training (Mihalic, 2004; Rogers, 2003). Thepresent study extends this work by examining theeffects of the amount of training on teachers’ use ofa specific C-B approach to antibullying. That it isimportant to do so is suggested by work that hasshown that attendance at training workshopscan have demonstrable effects on community-basedpractitioners’ beliefs about and use of specificapproaches to tacking other kinds of emotionaland behavioral problems (i.e., not bullying per se;Lim, Nakamura, Higa-McMillan, Shimabukuro, &Slavin, 2012). Moreover, while there is a growingawareness of the need to consider how inter-ventions are taken up in community settings (Limet al., 2012), what levels of training are required is arelatively new topic of enquiry (Han & Weiss,2005; Lochman et al.). Hence, our data shouldfacilitate the understanding of how to achieve awider use of C-B approaches to bullying by teachersin schools.Building on the aforementioned research, the

present study also addressed the following aimsthat have so far not been reported in the literature:

• To test if teachers’ reported use of a specific C-Bapproach (IDECIDE) across an 8-month periodcould be predicted from their self-efficacy andperceived effectiveness of C-B approaches, andthe duration of the I DECIDE training theyreceived. The durations we tested correspondedwith the typical pattern of in-service trainingreceived by teachers and other community-based practitioners that takes the form ofworkshops (Lim et al., 2012).

331t eacher s ’ and cogn i t i v e - b ehav ioral approaches to bully ing

• To test if the association between duration of theI DECIDE training teachers received and theirreported use of it 8 months later (if found) wasmediated by their self-efficacy and/or perceivedeffectiveness of C-B approaches. That theseare viable mediators is supported by the ge-neric underpinning theoretical principle of C-Bapproaches—namely, that these kinds of cogni-tions mediate between personal experiences andactual behavior (Beck, 1976; Meichenbaum,1977).

other reasons why teachers do or donot use c-b antibullying approaches

Despite the aforementioned direct and indirectevidence that supports the use of C-B approachesto tackle behavior disorders and bullying per se, itremains the case that they are underused relative toother approaches (Lewis, Hudson, Richter, &Johnson, 2004; Samara & Smith, 2008). Gerberand Solari (2005, p. 294) noted that by under-standing why, we can make progress in increasingthe application of C-B approaches. To the best ofour knowledge, no study to date has actually askedteachers why they do or do not use C-B antibullyingapproaches. Gerber and Solari presented a numberof possible reasons, including the notion that suchapproaches are fairly complex and challenging, andhence teachers may not see it as part of their role toimplement them (Gerber & Solari). Hence, thefourth and final aim of the current study was toidentify common reasons why teachers who havereceived training in the I DECIDE program went onto report using it frequently versus infrequently intheir everyday practice.

Methodparticipants

Participants were selected on a convenience basis.They were recruited through advertisements invarious locations across the U.K. by trainingproviders, and via attending lectures and talksdelivered by the author at conferences for aca-demics and practitioners, all concerned with anti-bullying themes. Initially, 446 teachers wereapproached to take part in the study. They werefrom a minimum of 60 junior schools from acrossthe U.K., but 155 did not identify their school orwere in the process of changing schools at thisstage. The broad aims were outlined and acommitment to provide data at the three timeperiods (see below) was requested. In total, 249junior school teachers from 32 schools spreadacross the U.K. agreed and provided data, aresponse rate of 55.8%. They ranged in age from24 to 58 years. Training was offered to all of these

teachers, and we initially intended to randomlyallocate them to the conditions of the study. Inpractice this was not possible in that many teacherswere unable to commit to being available at thetimes required by our training schedule becausethey had to ensure their classroom duties wereprioritized, were unsure they could arrange coverstaff, or had activities planned that required themto be with their class. Of the original pool, 124teachers from 18 schools were able to attend anin-service I DECIDE training workshop (81.5%females, called the experimental teachers). Of these,35 (28.2%) attended a half-day workshop, 31(25.0%) attended a 1-day workshop, 30 (24.2%)attended a 2-day workshop, and 28 (22.6%)attended a 3-day workshop, called “duration oftraining” henceforward. Again, we aimed torandomly allocate teachers to these durations, butfell slightly short of this ideal for 4 teachers due tothe same reasons listed above; 1 teacher initiallyselected to be in the 1-day workshop, and 3 to be inthe 2-day workshop, were only available for half aday, and so went into that group. Once we hadidentified the experimental teachers, we went backto the original pool of teachers who had agreed tobe part of the study and who were from neighbor-ing schools in which workshop attendees taught;these teachers were selected to be in the controlcondition. To minimize cross-contamination, thesewere drawn from schools that did not include anyteachers in the experimental condition. They wereselected on the basis of matching by sex, age, andduration of teaching as far as we were able. A groupof 125 teachers from 14 schools met these criteria(74.4% female, called the control teachers) andagreed to act in this capacity (i.e., complete ourquestionnaires but not receive the training).The training was delivered in a variety of

locations, including some of the schools theexperimental teachers attended, schools close byso that teachers from more than one local areacould attend together, and sites away from schoolssuch as conference centers.

procedure and measures

Data were collected via a postal questionnairesurvey. This was completed at three time points,denoted by pre-test, post-test and follow-up. Forthe experimental teachers, pre-test was 1 weekprior to attendance at an I DECIDE workshop,post-test was 2 weeks later, and follow-up was8 months later still. Data collection periods andintervals were the same for the control teachers. Thequestionnaire began by defining C-B approaches:“Cognitive-behavioral approaches focus on helpingchildren change their problematic behavior by

332 boulton

enabling them to change their thinking patterns andemotional responses. For example, an adult mighthelp a pupil identify what things provoke them andteach them how to have calming thoughts and tomanage feelings like anger.” A question asked ifparticipants had heard of this general approach. Allof the experimental teachers and 120 (96%) of thecontrol teachers indicated that they had done so atpre-test.Our ongoing work with teachers had indicated

that most would likely only consent to take partif the questionnaire was kept short. Hence,existing scales that were fairly lengthy would nothave been suitable (such as the 24- or 12-itemTeacher Efficacy Scale [TES]; Tschannen-Moran &Woolfolk Hoy, 2001) and the scales that wedeveloped for this study were kept brief. Threeitems were employed to measure “perceived effec-tiveness”: (1) “I think that cognitive-behavioralapproaches are effective in tackling bullying”;(2) “Cognitive-behavioral approaches are not agood way to stop bullying (r)”; and (3) “Cognitive-behavioral approaches are the best way to preventbullying.” These items appear to be similar with thebrief description of those employed by Dake et al.(2003), but the actual items were not provided bythem. Self-efficacy was assessed with three items:(1) “I know how to use cognitive-behavioralapproaches properly”; (2) “Sometimes I’m unsureabout how best to use cognitive-behavioralapproaches (r)”; and (3) “I am skilled at usingcognitive-behavioral approaches.” These itemswere worded in a similar way to the TES (sampleitem: “How well can you respond to defiantstudents?”), with the added advantage that bothpositively and negatively worded items were usedhere but not in TES, and with a previous singleitem measure of teachers’ self-efficacy, “I haveeffective strategies for handling a bullying situa-tion” (Bradshaw et al., 2007). Response options forthese scales were strongly agree, agree, not sure,disagree, and strongly disagree, scored from 5 to 1,or 1 to 5 for negatively worded items, denoted by“r” above. Cronbach’s alpha values were .71 and.74, respectively, indicating very high internalreliability for such short scales. Hence, for eachscale a total score was created, ranging from 3 to15, with a high score indicating greater “perceivedeffectiveness” and “self-efficacy.”“Usage” was assessed with a single item: “How

often do you use I DECIDE or similar cognitive-behavioral approaches in your antibullying work?”Response options were never, very infrequently,infrequently, frequently, and very frequently,scored 0 to 4, respectively. At follow-up, for thosechoosing one of the first three options (or last

two options), a secondary open question asked,“Why do you use I DECIDE or similar cognitive-behavioral approaches infrequently or never(frequently) in your antibullying work?” At pre-test, all participants were reminded that they wouldbe asked about their reported usage (and the othervariables) at other phases of the project and hencethat keeping a record of when they did use C-Bapproaches would be helpful. A pro forma for thiswas provided but many teachers reported havinginsufficient time to fill them out and instead usedtheir own records (see below for reliability data).Since we examined the effects of attending an I

DECIDE workshop on subsequent reported usageand beliefs, our longitudinal design was implement-ed to try to minimize the likelihood that partici-pants would have received training in other C-Bapproaches in the period between post-test andfollow-up. Specifically, we conducted the studywithin a single academic year because teachersusually attend few training events within such aperiod. To test if this was the case, at follow-upparticipants were asked, “Have you received anyother training in cognitive-behavioral approachessince you attended the I DECIDE workshop, and ifso what was it and what was its duration?” Onlyfive teachers replied affirmatively, all for a half-dayworkshop, and they were spread out across theexperimental and control teachers. Hence, wedeemed it unnecessary to include this variable as acovariate in our statistical analyses.Because of the longitudinal and experimental

nature of our study, it was not desirable to assesstest-retest reliability of our measures with actualparticipants. Thus a separate sample of teachers(n = 61) completed the same questionnaire twice,with a 1-week gap. Pearson’s r values for “per-ceived effectiveness,” “self-efficacy,” and “usage”were .88, .92, and 1.0, respectively, all p b .001. Atthe end of the study, we also assessed the test-retestreliability of the latter single itemmeasure by askinga subsample of I DECIDE attendees (n = 22) toanswer that question again, 1 week later. Reliabil-ity was high, .98, p b .001.Content analysis was used to code responses to

the open question. A researcher experienced in thismethod initially read through them and identifiedthemes that were relatively common (i.e., responsesgiven by at least 10% of participants overall, or byat least 10% of participants in any duration oftraining subgroup). Infrequent responses by thiscriterion were subsumed into an “other” category,and this was not subject to analysis (see below).Common responses were labeled (see Table 2) andused as a taxonomy by two raters to independentlycode each response. Cohen’s kappa indicated good

333t eacher s ’ and cogn i t i v e - b ehav ioral approaches to bully ing

levels of intercoder reliability, with the value being.80 and .83 for the reasons why I DECIDE was usedinfrequently and frequently, respectively. Instancesof initial disagreement were discussed by the threecoders, and resolved with consensus coding.Given that our variables were all self-report, it

was stressed to all participants that honestresponses were required. They were assured thatresponses would remain confidential and thattelling the truth was more helpful to the researchersthan offering socially desirable answers.

the i decide program

The I DECIDE program, designed for children aged7+ years, is founded on cognitive-behavioralprinciples. It was informed by various empiricaland theoretical formulations that are especiallyrelevant to bullying, such as the notion of hostileattribution bias (the tendency to attribute hostileintent in ambiguous provocative situations, Dodge,2006), and the role of peer approval/disapproval onan individual’s behavior (Salmivalli, Lagerspetz,Bjorkqvist, Osterman, & Kaukiainen, 1996). It is astructured program that uses many practicalexercises rather than didactic therapy. Pupils areshown how to keep a “triggers diary” to identifywhat it was about the victim and/or the peer groupthat prompted bullying and bullying-supportingbehavior. They are helped to identify their currentassociated patterns of thoughts, the feelings andemotions that these lead to, the behaviors that followfrom these inner processes, and to understand howall of those elements influence subsequent overtbehavior and the personal consequences it bringsabout. The in-service workshop showed teachershow to equip pupils at risk of bullying others orsupporting the actions of bullies with the skills togenerate more helpful alternatives at every stage. Forinstance, several exercises teach pupils how to drawa distinction between “helpful” and “unhelpful”thoughts. Workshop attendees were provided withexamples of high-profile cases that illustrate theprinciples, and encouraged to look for their ownexamples once they started to implement I DECIDE.For example, many boys and girls in the U.K. arekeen football supporters, and the training usedcases where footballers were sent off (negativeconsequence) because they responded to real orimagined provocations from members of the oppos-ing teamwith“unhelpful” thoughts and feelings, andunacceptable behavior.In this study, I DECIDE training was presented

as being especially relevant for use as a targetedpreventive intervention for children identified asbeing involved in, or at risk of, bullying; however,as studies have shown that many children will

engage in bullying at some point in their schoolcareers (Olweus, 1993; Rose et al., 2009), it wasalso noted that it could be utilized as a universalintervention with all children in a group.The workshop made it clear to teachers that

they should not come up with the more helpfulalternatives but rather support pupils in the processof arriving at their own. At the end of the program,teachers enable pupils to construct “contrastingself-portraits” of the “old and new me.”The different durations of training all addressed

the same basic steps covered in I DECIDE, namely,the triggers, thoughts, feelings, behavior, andconsequences that could be relevant to bullying.What differed was the variety of exercises taught tothe teachers (longer durations considered morealternative ways to work on each step) and theamount of time available to try them out with eachother and discuss them with the group and thetutor. For example, longer durations allowed moretime for teachers to reflect on why an activity maynot be well-received by pupils, and to discuss howto overcome those problems, something that wasdone only very briefly in the half-day format.Similarly, with increasing training durations,attendees had more time to reflect on the ideasand concepts presented. For example, in the half-day workshop, after explanations of hostile attri-bution bias and a description of a brief case study,participants spent 5 minutes discussing (in pairs)examples of their own pupils exhibiting hostileattribution style. In the 3-day workshop, trainingwas augmented by opportunities for attendees toreflect more personally on situations where they ora colleague had exhibited such bias. Thus, while thetime available for discussion, consolidation, andpractice may have varied, the content was held asconsistent as possible across the different durations.All workshops were delivered by the author, a

trained educator and counselor, using manualizedprotocols that are usually employed to teachpractitioners how to use C-B approaches. Evalua-tion studies have shown significant positive effectson bullying-related thoughts, feelings, and behaviorwhen teachers implement the I DECIDE program ona whole class basis (Boulton & Boulton, submitted)and with targeted groups of at-risk pupils (Boulton,submitted).

Resultsinitial comparison of experimentaland control groups

T-tests and chi-squared tests confirmed that the twogroups did not differ in terms of sex, age, durationof serving as a teacher, amount of “extra” trainingin C-B approaches they had received, and that their

334 boulton

schools did not differ in the proportion of pupils onthe school register receiving free school meals. This,along with the results which indicated the experi-mental and control groups did not differ initially onour dependent variables (see below), means that it isappropriate to test for the effects of I DECIDEtraining on our dependent variables.

aim 1a: teachers’ initial self-efficacy,perceived effectiveness, and use of c-bapproaches

Descriptive data from the whole sample at pre-testwere used to quantify teachers’ self-efficacy, perceivedeffectiveness, and usage. There was virtually novariance in the latter as 96.7% and 94.4% of theworkshop attendees and nonattendees, respectively,reported that they never usedC-B approaches, and theremainder reported doing so very infrequently orinfrequently. Because of this floor effect, in theregression analysis that follows, with “usage” as thedependent variable, the score at follow-up wasemployed as it could be considered a change score.The mean score for self-efficacy was 8.96 (SD =

2.36, range 3 to 15), corresponding with (slightlybelow) the “neutral” option of the Likert responseoptions for positively worded items, and indicates arather modest level of self-efficacy overall. Therelatively large standard deviation and range sug-gested a frequency analysis of high and low scoreswasappropriate. This revealed that 38 teachers (15.3%)had a score of 6 or less (i.e., that correspondedwith, orwas more negative than, the “disagree” responseoption for positively worded items), indicative of lowor very low levels of self-efficacy. At the other end, 35teachers (14.1%) had a score of 12 or more (i.e., thatcorresponded with, or was more positive than, the“agree” response option for positively worded items),indicative of high or very high levels of self-efficacy.The mean score for perceived effectiveness was

12.63 (SD = 1.73, range 8 to 15), corresponding withthe “agree” option of the Likert response options forpositivelyworded items.A frequency analysis revealedthat 183 teachers (73.5%) had a score of 12 or more(i.e., that corresponded with, or was more positivethan, the “agree” response option for positivelyworded items). At the other end, only 9 teachers(3.6%) had a score of 8 or 9 that corresponded withthe “neutral” response. Together, these results areindicative of high or very high levels of perceivedeffectiveness overall.

aim 1b: effect of attending an idecide workshop on self-efficacy andperceived effectiveness

The effect of attending an I DECIDE workshop(regardless of duration) on self-efficacy, and, sepa-

rately, perceived effectiveness, was tested using datacollected from the whole sample with a pair of 2(Time: pre-test versus post-test [within-subjectsfactor]) × 2 (Group: experimental versus control)mixed ANOVAs. Significant interaction effectswere investigated with follow-up t-tests. Whenself-efficacy served as the dependent variable, theTime × Group interaction was significant, F(1,247) = 65.18, p b .001, partial eta-squared = .209.This effect is shown in Figure 1 (top). T-testsindicated that among both the experimental group,t(124) = 23.62, p b .001, and the control group,t(124) = 8.56, p b .001, there was a significantincrease in self-efficacy across time. However, thatthis increase was significantly larger in the experi-mental group than in the control group is indicatedby the findings from independent samples t tests thatshow that while there was not a significant differencebetween the groups in self-efficacy at pre-test,t(247) = 0.509, p = .611, there was a significantdifference between the groups at post-test in favor ofthe experimental group, t(247) = 6.62, p b .001.When perceived effectiveness served as the

dependent variable, the Time × Group interactionwas also significant, F(1, 247) = 14.99, p b .001,partial eta-squared = .057. This effect is shown inFigure 1 (bottom). T-tests indicated that among theexperimental group, t(124) = 3.58, p b .001, therewas a significant increase in perceived effectivenessfrom pre-test to post-test, but among the controlgroup there was a small and nonsignificant decrease,t(124) = 1.63, p = .105. When the two groups werecompared at pre- and post-test with independentsamples t tests, it was found that they did not differinitially, t(247) = .863, p = .389, but did do so atpost-test, t(247) = 3.87, p b .001.Taken together, these results attest to the positive

effect of attending the I DECIDE workshop on bothself-efficacy and perceived effectiveness.

aim 1c: effect of duration of trainingon self-efficacy and perceivedeffectiveness

A 2 (Time: pre-test versus post-test [within-subjectsfactor]) × 4 (Duration: half day, one day, twoday, three day) mixed ANOVA tested if there wasan effect of duration of training on changes inself-efficacy. The Time × Duration interaction wassignificant, F(3, 120) = 6.65, p b .001, partial eta-squared = .141 (see Figure 2). A set of four t-tests,one for each “duration of training,” revealed thatfor all durations there was a significant increase inself-efficacy after attending, minimum t = 10.58(df 27 to 34), all p b .001. Between-groups ANO-VAs showed that whereas there were no significantdifferences in self-efficacy at pre-test, F(3, 120) =

FIGURE 1 Mean self-efficacy (top) and effectiveness (bottom) at pre– and post–I DECIDE workshopattendance.

335t eacher s ’ and cogn i t i v e - b ehav ioral approaches to bully ing

0.32, there were significant differences at post-test,F(3, 120) = 4.21, p = .007. Post hoc Tukey HSDtests revealed that the post-test scores of the 3-daytraining groupwere significantly higher than those ofthe half-day group, p = .006, with a trend to behigher than those of the 1-day group, p = .057.A similar set of analyses to the above was

employed using “perceived effectiveness” as thedependent variable. The Time × Duration interac-

tion was nonsignificant, F(3, 120) = 0.15, p = .93.Thus, there was no effect of duration of training onchanges in perceived effectiveness. To confirm this,and to demonstrate that the groups that receivedthe different durations of training were equivalentinitially, between-groups ANOVAs showed thatthere were no significant differences in perceivedeffectiveness at pre-test, F(3, 120) = 0.99, and atpost-test, F(3, 120) = 0.46.

FIGURE 2 Mean self-efficacy at pre– and post–I DECIDE workshop attendance as a function ofduration of workshop.

336 boulton

aim 2: predicting usage fromself-efficacy, perceived effectiveness,and the duration of training

The correlations between self-efficacy, perceivedeffectiveness (both at post-test), duration of train-ing, and usage (at follow-up) are shown in Table 1.Usage was significantly correlated with both self-efficacy and duration of training, but not withperceived effectiveness. The former significantcorrelations justify the proposed test set out inAim 2 to determine what variables predict reported“usage,” but the latter nonsignificant correlationjustifies the exclusion of perceived effectivenessfrom this test (and the test of mediation reported forAim 3, below. We initially did include “perceived

Table 1Zero-Order Correlations Among Self-Efficacy, PerceivedEffectiveness, Duration of Training, and Reported Usage of IDECIDE

2 3 4

1. Reported usage of I DECIDE .23* .30** .022. Duration of training - .31** -.083. Self-efficacy - - .22**4. Perceived effectiveness - - -

Note. N = 124; *p b .05; **p b .01.

effectiveness” in the model but it did not emerge asa significant predictor, nor did it change materiallythe predictive nature of the other variables). Hence,a multiple regression model was constructed todetermine if self-efficacy and duration of training(both at post-test and both mean centered; Aikenand West, 1991) could predict “usage” 8 monthslater at follow-up. The overall model accounted for11.2% of the variance in “usage,” a significantamount, F(2, 121) = 7.64, p = .001. Controllingfor duration of training, self-efficacy emerged as asignificant predictor of usage, Beta = 2.56, p =.005.However, controlling for self-efficacy, duration oftraining did not emerge as a significant predictor ofusage, Beta = 1.52, p =.094. This pattern of results(i.e., significant correlations between self-efficacyand usage, and between duration of training andusage (see Table 1), and the significant regressionmodel, justifies the test of mediation proposed inAim 3, reported below.

aim 3: is the association betweenduration of training and usagemediated by self-efficacy?

The hypothesis that the association between dura-tion of training and usage at follow-up wasmediated by self-efficacy at post-test was investi-gated using the bootstrapping procedure of

FIGURE 3 Self-efficacy mediates the association between duration of training andsubsequent reported usage of I DECIDE. Values in brackets refer to the effect of duration oftraining on reported usage after controlling for self-efficacy.

337t eacher s ’ and cogn i t i v e - b ehav ioral approaches to bully ing

Preacher and Hayes (2004), with mean centeredpredictor and mediator and 5,000 bootstrapresamples. The mediation pattern is shown inFigure 3; importantly, the effect of durationtraining on usage, which was originally significant,beta = .237, p = .01, became nonsignificant when“self-efficacy” was entered into the model, beta =.156, p = .094. Moreover, the indirect effect ofduration of training through self-efficacy on usagewas significantly different from zero at p b .05,point estimate = .0804, and estimated to liebetween .0140 and .1670 with 95% confidence.These findings indicate that self-efficacy was a fullmediator.

aim 4: reasons why teachers reportusing i decide frequently versusinfrequently

At follow-up, teachers who attended an I DECIDEworkshop were asked how often they used it overan 8-month period. Sixty-five (52.4%) reporteddoing so “frequently” or “very frequently,” and 59reported doing so “infrequently,” “very infrequent-

Table 2Percentage (and Number in Brackets) of Teachers That Gave Speci(top) and Frequently (bottom) as a Function of Duration of Training

Reason for infrequent use:

My school is not set up to use this kind of in-depth approach.I needed more training.It is a challenging intervention that I felt was beyond me.I couldn’t commit to such an open-ended approach with no set timeI continued to use other interventions.

Reason for frequent use:

I believe in this approach and found it is effective.Other ‘generic’ approaches are insufficient for pupils with behavioraMy training equipped me to use it.Evidence from research studies has proven it works.

ly,” or “never.” They were then asked why (Table 2).The data are broken down by ‘duration of training’ todetermine if this variable was or was not important,and chi-squared tests were used to compare responsesas a function of duration. Themain impediments citedamong infrequent users, in descending order, were (a)“My school is not set up to use this kind of in-depthapproach,” (b) “I needed more training,” (c) “It is achallenging intervention that I felt was beyond me,”(d) “I couldn’t commit to such an open-endedapproachwith no set time scale,” and (e) “I continuedto use other interventions.” More teachers whoattended the half-day or the 1-day workshopsendorsed the “I need more training” reason thanexpected by chance (all of them did so), with theopposite being the case among thosewho attended for2 or 3 days, Χ2 (3) = 13.04, p = .004. The samepattern was also evident for the “It is a challengingintervention that I felt was beyond me” reason.The main reasons teachers gave for using I

DECIDE frequently, in descending order, were (a)“I believe in this approach and found it is effective,”(b) “Other ‘generic’ approaches are insufficient for

fic Reasons Why They Used I Decide Infrequently or Not at All

Duration of I DECIDE training

Half day 1 Day 2 Days 3 Days All

(n = 18) (n = 16) (n = 16) (n = 9) (n = 59)

88.9 (16) 81.3 (13) 81.3 (13) 88.9 (8) 84.7 (50)100 (18) 100 (16) 50 (8) 33.3 (3) 76.3 (45)88.9 (16) 75.0 (12) 25.0 (4) 0.0 (0) 54.2 (32)

scale. 50.0 (9) 50.0 (8) 37.5 (6) 44.4 (4) 45.8 (27)16.7 (3) 25.0 (4) 25.0 (4) 11.1 (1) 20.3 (12)

(n = 17) (n = 15) (n = 14) (n 19) (n = 65)

88.2 (15) 93.3 (14) 100(14) 100 (19) 95.4 (62)l issues. 76.5 (13) 80.0 (12) 85.7 (12) 78.9 (15) 80.0 (52)

11.8 (2) 20.0 (3) 64.3 (9) 84.2 (16) 46.2 (30)29.4 (5) 26.7 (4) 21.4 (3) 26.3 (5) 26.2 (17)

338 boulton

pupils with behavioral issues,” (c) “My trainingequipped me to use it,” and (d) “Evidence fromresearch studies has proven it works.” Moreteachers who attended the 2-day or the 3-dayworkshops endorsed the “My training equipped meto use it” reason than expected by chance, with theopposite being the case among those who attendedfor a half day or one day,Χ2 (3) = 16.67, p b .001.

DiscussionBullying continues to be a major behavioralproblem among a substantial proportion of schoolpupils and C-B approaches hold promise as aneffective intervention. Despite this, most teachersappear to use C-B approaches rather infrequently, ifat all, to tackle bullying per se, as well as specificforms of emotional and behavioral disorders thatdo have diagnoses attached to them (Gerber &Solari, 2005; Samara & Smith, 2008). The currentstudy presents important novel data that helps toelucidate the role of training and why teachers mayvary in this regard.

initial usage, self-efficacy, andperceived effectiveness beliefs

About three quarters of the teachers agreed thatC-B approaches were likely to be effective intackling bullying, and, at worst, only a smallminority (3.6%) expressed neutral beliefs. Theseare the first such data to be reported. Our analysisof teachers’ qualitative responses make us feel thatit is reasonable to conjecture that they hold suchstrong effectiveness beliefs because they are awarefor one reason or another of the substantialevidence base obtained in research studies. Indeed,of the 65 teachers who attended an I DECIDEworkshop and who went on to indicate that theyuse it or related C-B approaches frequently in theirpractice, about a quarter (26.2%) spontaneouslyreported that they did so for this reason (seeTable 2). It is encouraging that teachers seem tobe aware of the potential of C-B approaches totackling bullying.It is less encouraging that among the sample as a

whole, the average “self-efficacy” score corre-sponded with the neutral point on the responsescale. A minority had even lower scores, and onlyabout the same small proportion had high scores.Again, our qualitative data provide insights intowhy self-efficacy was so low overall; more than half(54.2%) of the 59 teachers who attended an IDECIDE workshop and who went on to reportusing it infrequently stated that the reason wasbecause they felt it was beyond their competence(see Table 2).

Consistent with other studies (Bauman et al.,2008; Gerber & Solari, 2005; Samara & Smith,2008; Smith et al., 2008), only a very smallproportion of our sample reported using C-Bapproaches at the start of the project. Thus, ourdata indicate that many teachers, perhaps themajority, can be characterized as not using C-Bapproaches to tackling bullying, lacking in self-belief that they can implement it, but neverthelessseeing value in it. These novel findings sit wellalongside the conjectures of Gerber and Solari, thatan impediment to the widespread use of C-Bapproaches to emotional and behavioral issues isteacher resistance that itself is based on lack ofconfidence rather than lack of perceived effective-ness. Clearly, understanding the sources ofteachers’ perceived effectiveness and self-efficacybeliefs towards C-B approaches is important. Ourfindings concerning the impact on these beliefs ofattending an in-service training workshop areexplored next.

effects of in-service training onself-efficacy and perceivedeffectiveness beliefs

Attending an I DECIDE workshop had a positiveeffect on both self-efficacy and perceived effective-ness beliefs. These results echo those of prior studiesthat have also found that (a) antibullying-trainingthat was not C-B based had a positive effect onteachers’ anti-bullying actions (Bauman et al.,2008; Dake et al., 2003) and (b) training inC-B-based approaches to conduct disorders (i.e.,not bullying per se) positively affects the amountand quality of use of those approaches (Packenhamet al., 2004; Sasso et al., 1992).All four durations of training had a positive effect

on perceived self-efficacy. Given the relatively lowlevels at the start of the study (see above), this is anespecially encouraging finding. We also foundsignificantly greater gains in the 3-day traininggroup than in the half-day attendees (and nearsignificant gains relative to the 1-day attendees) butnot than in the 2-day attendees. The practicalimplication is that there is likely to be an optimum“dosage effect” of in-service training on self-efficacy. Given the limited amount of time thatteachers have to receive in-service training, futurestudies could elucidate more clearly the best ratio ofduration to gains in self-efficacy, and determine ifour findings are replicable. The importance of thisissue is highlighted by the finding that a commonbarrier to using some specific antibullyingapproaches was the time taken to learn how touse and implement them (Dake et al., 2003). Futurestudies could investigate what specific mechanisms

339t eacher s ’ and cogn i t i v e - b ehav ioral approaches to bully ing

might lead longer training to improve self-efficacy.Plausible candidates are more time to practice skills,to discuss ideas with colleagues, and to solveproblems, since they have been found to beimportant in prior work that has solicited teachers’views about interventions they can use to improvepupil relationships (Cowie, Smith, Boulton, &Laver, 1994).In contrast, there was no effect of duration of

training on perceived effectiveness beliefs. Withoutwishing to explain away this null result, we think itis relevant to note the finding discussed in thepreceding paragraph that attending the workshopdid have a positive effect on self-efficacy, and thatteachers initially had high perceived effectivenessbeliefs, meaning that there was relatively little scopefor improvement on the latter variable.

impact of self-efficacy, perceivedeffectiveness, and duration oftraining on use of i decide

Despite an almost total nonuse of C-B approachedat the start of the study, a substantial proportion ofteachers (52.4%) indicated that they used it“frequently” or “very frequently” in the ensuing8 months. This shows that in-service workshopscan lead to greater reported usage. A value of ourstudy is that it offers insights into why teachers did/did not do so. We obtained significant correlationsbetween reported usage of I DECIDE and bothself-efficacy and duration of training, but notperceived effectiveness. A regression model indicat-ed that a significant amount of variance (11.2%) inreported usage was predicted by self-efficacy andduration of training together. The effect of theformer was not contingent upon the variance itshared with the latter, but the reverse was not true.The association between duration of training andreported usage was fully mediated by self-efficacy.Our qualitative data painted a similar picture sincejust under half of I DECIDE attendees who went onto use it frequently reported doing so because “Mytraining had equipped me to use it” and even more(95.4%) because “I believe in this approach andfound it is effective” (see Table 2). Together, all ofthese findings mark self-efficacy out as beingparticularly important in predicting reportedusage of I DECIDE, both directly in its own rightand also as a mediator. Hence, our other findingthat self-efficacy was improved by attending an IDECIDE workshop becomes especially important.An implication is that training providers shouldregard improving self-efficacy as an importantoutcome in its own right—one that should beassessed rather than assumed. These findings takeon extra importance given that there is now a big

push to get evidence-based practices, including C-Bapproaches, used more widely by practitioners suchas teachers who work in community as opposed toclinical settings (Lim et al., 2012; New FreedomCommission on Mental Health, 2003).The lack of association between reported usage

of I DECIDE and perceived effectiveness should notbe taken as reflecting the latter’s lack of importance,especially given our finding that the most commonreason (given by 95.4%) why I DECIDE attendeeswent on to use I DECIDE was because they “believein this approach and found it is effective.” It morelikely reflects the lack of variance in perceivedeffectiveness, as most teachers scored high on thisvariable.

limitations and future directions

This study has a number of limitations that must beconsidered. Our measures were operationalizedwith very short scales, or in the case of reportedusage, a single item. In mitigation, though, teachersonly agreed to take part in our study if the ques-tionnaire was very short. Moreover, short or evensingle-item scales have been endorsed as a way tomeasure straightforward (i.e., unidimensional)variables (Hayes, 2005; Menesini & Nocentini,2009; Robins, Hendin & Trzesniewski, 2001),including self-efficacy (Salmivalli et al., 2005), andour test-retest coefficients were all good. Wewanted to assess teachers’ general self-beliefsabout C-B approaches, but future studies wouldbenefit from asking teachers about some specificaspects of them since it is entirely possible that theymight, for example, see more value in sometechniques than in others, and this could influencetheir actual practice. Our study was not designed totap such nuanced issues but it provides a rationalefor this kind of investigation.While the response rate of about 56% may seem

low, it is common for response rates amongteachers to be low, and ours is higher than someprevious studies of teachers’ beliefs (MacFarlane &Woolfson, 2013). Our impression was that manyteachers that we initially approached to take partwere only prevented from doing so because theycould not take time away from their classes.Nevertheless, it is possible that teachers who didprovide data may differ from those who did not,and this could limit the generalizability of ourfindings. Incentivizing the wider participation ofteachers is recognized as a good way to overcomethis, and, in our view, Kazdin’s (1998) suggestion ofcommunicating to potential participants the valueof taking part is a valuable recommendation.Our study only looked at the effects of one

particular C-B approach, the I DECIDE program,

340 boulton

and again this limits generalizations. Future studiesshould determine if positive effects of other C-Bapproaches can be demonstrated. Relatedly, wewere unable to assess the integrity with which IDECIDE was implemented. This is likely to affectits impact on teachers’ beliefs and actions, and soshould be measured in future studies.We focused on the effects on in-service training,

but there is value to be had in studies that test ifsimilar positive effects, or even stronger ones, canbe obtained during initial training with traineeteachers. Studies have shown that trainee teachers’beliefs are linked to their intended antibullyingactions (Boulton et al., in press; Craig et al., 2000).While our study has provided evidence concerningthe importance of self-efficacy, it leaves open thequestion of how such beliefs are formed. Our dataclearly suggest that attending trainingmay be helpful,but future studies couldhelp ascertainwhat particularaspects of those experiences impact what specificcognitions and feelings that are associated withteachers’ overall self-efficacy. Interest in improvingself-efficacy extends beyond a concern for the widerimplantation of C-B approaches, since it has beenshown to protect against teacher burnout in thecontext of wider classroom management (Brouwers& Tomic, 2000). The fact that teachers often reportthat aggression, and probably bullying too, interfereswith their primary job of teaching and increases theirlevel of stress (Boxer, Musher-Eizenman, Dubow,Danner,&Heretick, 2006) can perhaps be used as anincentive to encouragemore to receive and utilize C-Bapproaches, and to take part in studies such as thisone.Our inability to randomly allocate teachers to

the experimental or control conditions could beseen as a weakness. It is regarded as the best way toconduct experiments on the effects of interven-tions, but Kazdin (1998) points out that it doesn’tguarantee that groups will be equivalent on keyvariables and that matching can be effective. Inthis regard, we were able to show that on a rangeof salient variables, the two groups did not differsignificantly, nor did they differ on our dependentvariables initially. Kazdin also notes that in theabsence of random sampling, and even in thepresence of initial group similarity, it is stillpossible that the two groups could have beendrawn from different populations with regard totheir scores on dependent variables but haveoverlapping distributions. If, by chance, ourexperimental (control) group were comprised ofindividuals from the lower (upper) end of apopulation with generally high (low) self-efficacyscores, then when they were subsequently testedon that variable, regression to the mean could

make it appear that they had diverged on thatvariable due to the influence of the intervention.Importantly, this latter possibility is dependent onhow reliably the variable has been measured;regression to the mean as an explanation is muchless credible with a reliable measure. We were ableto show that self-efficacy, indeed all of our dependentvariables, were highly reliable. Hence, it seemsreasonable to propose that our finding that I DECIDEtraining improved self-efficacy was not simply due toinitial group nonequivalence or to a regression to themean artifact.Nevertheless, these suggestions can only mili-

tate against the kinds of alternative explanationspresented in the preceding paragraph, given thatwe were not able to use true random allocation.Moreover, it is possible that selection bias mayhave played a part in our findings. Specifically,teachers who attended the I DECIDE workshopsmay have been more concerned about tackingbullying than those who did not. If so, animplication is that the effect of those workshopswe obtained in this study may not generalize to allteachers, especially those for whom bullying is nota priority. Hence, future studies would do well tostrive for random allocation, perhaps by explain-ing why it is needed to potential participants.We had more success randomly allocating

experimental group teachers to the different dura-tions of I DECIDE training since this was achievedfor all but four teachers who were not available foranything other than the half-day training. Again,we were able to show that the teachers who wereallocated to the different durations did not differinitially, and our reliable measures suggest that theeffects of different durations of training were notregression-to-the-mean artifacts.Our reliance on teacher self-reports of usage of

I DECIDE merits discussion because concernshave been raised about positive presentation bias(participants wanting to “look good” to researchers;Kazdin, 1998). It is possible that little actual behaviorchange occurred. We tried to mitigate such bias byassuring confidentiality and stressing that telling thetruth was more helpful to the researchers thanoffering socially desirable answers. Moreover, evi-dence from some studies shows that teachers often doreport treatment integrity accurately (Sanetti &Kratochwill, 2009, 2011) and so self-presentationbias is not inevitable.In conclusion, most teachers believed that C-B

approaches can be effective in tackling bullying,but many held modest levels of self-efficacybeliefs, and few reported using it initially. Attend-ing a workshop on a specific C-B approach,

341t eacher s ’ and cogn i t i v e - b ehav ioral approaches to bully ing

especially of 3 days as opposed to half a dayduration, increased those self-efficacy beliefs, andthose beliefs, along with duration of the training,influenced their decisions of whether or not to useit, with those with the highest self-efficacy beliefsbeing most likely to do so. These data suggest thattake-up of C-B approaches will be improved ifmore teachers can be encouraged to not onlyattend relevant training events but also if thattraining leaves them feeling that they can imple-ment C-B approaches effectively. Given the highincidence of bullying in schools, and the distressassociated with it, ensuring more teachers useC-B-type interventions is a worthy goal.

Conflict of Interest StatementThe author declares that there are no conflicts of interest.

References

Aiken, L. S., &West, S. G. (1991).Multiple Regression: Testingand interpreting interactions. Thousand Oaks: Sage.

Ajzen, I. (1991). The theory of planned behavior.Organizational Behavior and Human Decision Processes,50, 179–211.

Ajzen, I. (2002). Residual effects of past on later behavior:Habituation and reasoned action perspectives. Personalityand Social Psychology Review, 6, 107–122.

Apsche, J. A., & Bass, C. K. (2006). A review and empiricalcomparison of three treatments for adolescent males withconduct and personality disorder: Mode deactivationtherapy, cognitive behavior therapy and social skillstraining. International Journal of Behavioral andConsultation Therapy, 2, 382–398.

Bauman, S., Rigby, K., & Hoppa, K. (2008). U.S. teachers’ andschool counsellors’ strategies for handling school bullyingincidents. Educational Psychology, 28, 837–856.

Bauman, S., &Del Rio, S. (2006). Preservice teachers’ responses tobullying scenarios: Comparing physical, verbal and relationalbullying. Journal of Educational Psychology, 98, 219–231.

Beck, A. T. (1976). Cognitive therapy and the emotionaldisorders. New York: New American Library.

Beran, T. (2005). A new perspective on managing schoolbullying: Pre-service teachers’ attitudes. Journal of SocialSciences, 8, 43–49.

Boulton, M. J. (1997). Teachers’ views on bullying: Definitions,attitudes and ability to cope. British Journal of EducationalPsychology, 67, 223–233.

Boulton, M.J. (submitted). The effects of the I DECIDEprogram on children identified as bullies and theirsupporters: Dealing with real or imagined provocations.

Boulton, M.J., & Boulton, L. (submitted). Primary preventionof bullying related thoughts, feelings and behavior: Anevaluation of the I DECIDE program.

Boulton,M.J., Hardcastle, K., Down, J., Fowles, J.,& Simmonds,J.A. (in press). A comparison of pre-service teachers’ responsesto cyber versus traditional bullying scenarios: Similarities anddifferences. Journal of Teacher Education.

Boulton, M. L., Trueman, M., & Murray, L. (2008).Associations between peer victimization, fear of futurevictimization and disrupted concentration on class workamong junior school pupils. British Journal of EducationalPsychology, 78, 473–489.

Boxer, P., Musher-Eizenman, D., Dubow, E. F., Danner, S., &Heretick, D. (2006). Assessing teachers’ perceptions forschool-based aggression prevention programs: applying acognitive-ecological framework. Psychology in the Schools,43, 331–344.

Bradshaw, C. P., Sawyer, A. L., & O’Brennan, L. M. (2007).Bullying and peer victimization at school: Perceptualdifferences between students and school staff. SchoolPsychology Review, 36, 361–382.

Brouwers, A. H., & Tomic, W. (2000). A longitudinal study ofteacher burnout and perceived self-efficacy in classroommanagement.Teaching and Teacher Education, 16, 239–253.

Buhs, E., Ladd, G., & Herald, S. (2006). Peer exclusion andvictimization: Processes thatmediate the relation between peergroup rejection and children’s classroom engagement andachievement. Journal of Educational Psychology, 98, 1–13.

Cowie, H., Smith, P. K., Boulton, M., & Laver, R. (1994).Cooperation in the multi-ethnic classroom. London: DavidFulton.

Craig, W. M., Henderson, K., & Murphy, J. G. (2000).Prospective teachers' attitudes toward bullying andvictimization. School Psychology International, 21, 5–21.

Dake, J. A., Price, J. H., Telljohann, S. K., & Funk, J. B. (2003).Teacher perceptions and practices regarding school bullyingprevention. Journal of School Health, 73, 347–355.

Dodge, K. A. (2006). Translational science in action: Hostileattributional style and the development of aggressivebehavior problems. Development and Psychopathology,18, 791–814.

Emmer, E. T., & Hickman, J. (1991). Teacher efficacy inclassroom management and discipline. Educational andPsychological Measurement, 51, 755–765.

Fekkes, M., Pijpers, F. I. M., & Verloove-Vanhorick, S. P.(2005). Bullying: Who does what, when, and where?Involvement of children, teachers, and parents, in bullyingbehavior. Health Education Research, 20, 81–91.

Frey, K., Edstrom, L. V. S., & Hirschstein, M. (2005). The Stepsto Respect program uses a multilevel approach to reduceplayground bullying and destructive bystander behaviors. InD. L.White,M. K. Faber, and B. C. Glenn (Eds.), Proceedingsof Persistently Safe Schools (pp. 47–56). Washington, DC:Hamilton Fish Institute, George Washington University.

Gerber,M.M.&Solari, E. J. (2005). Teaching effort and the futureof cognitive-behavioral interventions. Behavioral Disorders,30, 289–299.

Gibson, S.,&Dembo,M. (1984). Teacher efficacy: A constructvalidation. Journal of Educational Psychology, 76,569–582.

Ghafoori, B., & Tracz, S. M. (2004). Cognitive behavioraltherapy as a clinical intervention for childhood disruptivebehaviors: A meta-analysis. Child and AdolescentPsychiatry Online. Available online at: http://www.priory.com/psych/CBTchildhood.htm (November 23, 2004).

Goldstein, B., Glick, B., & Gibbs, J. (1998). AggressionReplacement Training: A comprehensive intervention foraggressive youth. Champaign, IL: Research Press.

Hall, J. A., Herzberger, S. D., & Skowronski, K. J. (1998).Outcome expectancies and outcome values as predictors ofchildren’s aggression. Aggressive Behavior, 24, 439–454.

Han, S. S., & Weiss, B. (2005). Sustainability of teacherimplementation of school-based mental health programs.Journal of Abnormal Child Psychology, 33, 665–679.

Hayes, A. (2005, November). A computational tool for surveyshortening applicable to composite attitude, opinion, andpersonalitymeasurement scales.Paper presented at themeetingof the Midwestern Association for Public Opinion Research,Chicago, IL.

342 boulton

Kärnä, A., Voeten, M., Poskiparta, E., & Salmivalli, C. (2010).Vulnerable children in varying classroom contexts: Bystanders'behaviors moderate the effects of risk factors on victimization.Merrill-Palmer Quarterly, 56, 261–282.

Kazdin, A. E. (1998). Research design in clinical psychology.Boston: Allyn and Bacon.

Kazdin, A. E., & Weisz, J. R. (Eds.). (2003). Evidence-basedpsychotherapies for children and adolescents. New York:Guilford Press.

Kochenderfer, B., & Ladd, G. (1996). Peer victimization: Causeor consequence of school maladjustment? ChildDevelopment, 67, 1305–1317.

Kochenderfer-Ladd, B., & Pelletier, M. E. (2008). Teachers’views and beliefs about bullying: Influences on classroommanagement strategies and students’ coping with peervictimization. Journal of School Psychology, 46, 431–453.

Lewis, T. J., Hudson, S., Richter, M., & Johnson, N. (2004).Scientifically supported practices in emotional andbehavioral disorders: A proposed approach and brief reviewof current practices. Behavioral Disorders, 29, 247–259.

Lim, A., Nakamura, B. J., Higa-McMillan, C. K., Shimabukuro,S., & Slavin, L. (2012). Effects of workshop trainings onevidence-based practice knowledge and attitudes amongyouth community mental health providers. BehaviourResearch and Therapy, 50, 397–406.

Lochman, J. E., Boxmeyer, C., Powell, N., Qu, L., Wells, K., &Windle, M. (2009). Dissemination of the CopingPower Program: Importance of intensity of counselortraining. Journal of Consulting and Clinical Psychology,77, 397–409.

Lochman, J. E., & Wells, K. C. (2003). Effectiveness study ofCoping Power and classroom intervention with aggressivechildren: Outcomes at a one-year follow-up. BehaviorTherapy, 34, 493–515.

MacFarlane, K., & Woolfson, L. M. (2013). Teacher attitudesand behavior toward the inclusion of children with social,emotional and behavioral difficulties in mainstream schools:An application of the theory of planned behaviour. Teachingand Teacher Education, 29, 46–52.

Marsh, H. W., Parada, R. H., Craven, R. G., & Finger, L.(2004). In the looking glass: A reciprocal effect modelelucidating the complex nature of bullying, psychologicaldeterminants, and the central role of self-concept. In C. E.Sanders & G. D. Phye (Eds.), Bullying: Implications for theclassroom (pp. 63–109). London: Elsevier.

Meichenbaum, D. H. (1977). Cognitive-behavior modification:An integrative approach. New York: Plenum.

Menesini, E., & Nocentini, A. (2009). Cyberbullying definitionand measurement: Some critical considerations. Journal ofPsychology, 217, 230–232.

Merrell, K. W., Gueldner, B. A., Ross, S. W., & Isava, D. M.(2008). How effective are school bullying interventionprograms? A meta-analysis of intervention research. SchoolPsychology Quarterly, 23, 26–42.

MetropolitanAreaResearchGroup. (2002).A cognitive/ecologicalapproach to preventing aggression in urban settings: Initialoutcomes for high risk children. Journal of Consulting andClinical Psychology, 70, 179–194.

Metropolitan Area Research Group (2007). Changing the waychildren “think” about aggression: Social-cognitive effectsof a preventive intervention. Journal of Consulting andClinical Psychology, 75, 160–167.

Mihalic, S. (2004). The importance of implementationfidelity. Emotional and Behavioral Disorders in Youth,4, 83–105.

Naylor, P., & Cowie, H. (1999). The effectiveness of peersupport systems in challenging school bullying: The

perspectives and experiences of teachers and pupils. Journalof Adolescence, 22, 467–479.

New Freedom Commission onMental Health. (2003).Achievingthe promise: Transforming mental health care in America.Final Report. (DHHS Pub. No. SMA- 03–3832). Rockville,MD.

Newman, R. S., & Murray, B. J. (2005). How students andteachers view the seriousness of peer harassment: When is itappropriate to seek help? Journal of EducationalPsychology, 97, 347–365.

Novick, R. M., & Isaacs, J. (2010). Telling is compelling: Theimpact of student reports of bullying on teacherintervention. Educational Psychology, 30, 283–296.

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

Olweus, D., & Limber, S. (2010). Bullying in school: Evaluationand dissemination of the Olweus Bullying Preventionprogram. American Journal of Orthopsychiatry, 80,124–134.

Ozabaci, N. (2011). Cognitive behavioural therapy for violentbehaviour in children and adolescents: A meta-analysis.Children and Youth Services Review, 33, 1989–1993.

Packenham, M., Shute, R. H., & Reid, R. (2004). A brieffunctional assessment procedure for children with disruptiveclassroom behavior. Education and Treatment of Children,27, 9–25.

Preacher, K. J., & Hayes, A. F. (2004). SPSS and SASprocedures for estimating indirect effects in simplemediation models. Behavior Research Methods,Instruments, & Computers, 36, 717–731.

Reijntjes, A., Kamphuis, J. H., Prinzie, P., & Telch, M. J.(2010). Peer victimization and internalizing problems inchildren: A meta-analysis of longitudinal studies. ChildAbuse and Neglect, 34, 244–252.

Rigby, K., & Barnes, A. (2002). To tell or not to tell: Thevictimized student’s dilemma. Youth Studies Australia, 21,33–36.

Robins, R., Hendin, H., & Trzesniewski, K. (2001). Measuringglobal self-esteem: Construct validation of a single-itemmeasure and the Rosenberg Self-Esteem Scale. Personalityand Social Psychology Bulletin, 27, 151–161.

Rogers, E. M. (2003). Diffusion of innovations (5th ed.). NewYork, NY: Free Press.

Rose, C. A., Espelage, D. L., & Monda-Amaya, L. E. (2009).Bullying and victimisation rates among students in generaland special education: A comparative analysis. EducationalPsychology, 29, 761–776.

Salmivalli, C., Kaukiainen, A., & Voeten, M. (2005). Anti-bullying intervention: Implementation and outcome. BritishJournal of Educational Psychology, 75, 465–487.

Salmivalli, C., Lagerspetz, K., Bjorkqvist, K., Osterman, K., &Kaukiainen, A. (1996). Bullying as a group process:Participant roles and their relations to social status withinthe group. Aggressive Behavior, 22, 1–15.

Samara, M., & Smith, P. K. (2008). How schools tacklebullying, and the use of whole school policies: Changesover the last decade. Educational Psychology, 28,663–676.

Samples, F. L. (2004). Evaluating curriculum-based interven-tion programs: An examination of preschool, primary, andelementary school intervention programs. In C. E. Sanders& G. D. Phye (Eds.), Bullying: Implications for theclassroom (pp. 203–227). London: Elsevier.

Sanetti, L. M. H., & Kratochwill, T. R. (2009). Treatmentintegrity assessment in the schools: An evaluation of theTreatment Integrity Planning Protocol. School PsychologyQuarterly, 24, 24–35.

343t eacher s ’ and cogn i t i v e - b ehav ioral approaches to bully ing

Sanetti, L.M. H., &Kratochwill, T. R. (2011). An evaluation ofthe Treatment Integrity Planning Protocol and two schedulesof treatment integrity self-report: Impact on implementationand report accuracy. Journal of Educational andPsychological Consultation, 21, 284–308.

Sasso, G. M., Reimers, T. M., Cooper, L. J., Wacker, D., &Berg, W. (1992). Use of descriptive and experimentalanalyses to identify the functional properties of aberrantbehavior in school settings. Journal of Applied BehaviorAnalysis, 25, 809–821.

Smith, P. K., Smith, C., Osborn, R., & Samara, M. (2008). Acontent analysis of school anti-bullying policies: Progressand limitations. Educational Psychology in Practice, 24,1–12.

Smith, P.K., Pepler, D. J.,& Rigby,K. (Eds.). (2004).Bullying inschools: How successful can interventions be? Cambridge,UK: Cambridge University Press.

Smith, P. K., & Shu, S. (2002). What good schools can do aboutbullying: Findings from a survey in English schools after adecade of research and action. Childhood, 7, 193–212.

Solberg, M., & Olweus, D. (2003). Prevalence estimation ofschool bullying with the Olweus Bully/Vict imQuestionnaire. Aggressive Behavior, 29, 239–268.

Troop, W. P., & Ladd, G. W. (2002). Teachers' beliefsregarding peer victimization and their intervention practices.

Poster presented at the Conference on Human Development,Charlotte, NC.

Tschannen-Moran, M., & Woolfolk Hoy, A. W. (2001).Teacher efficacy: capturing an elusive construct. Teachingand Teacher Education, 17, 783–805.

Ttofi, M. M., & Farrington, D. P. (2011). Effectiveness ofschool-based programs to reduce bullying: A systematic andmeta-analytic review. Journal of Experimental Criminology,7, 27–56.

Vaughn, M. G., Fu, Q., Bender, K., DeLisi, M., Beaver, K.M., Perron, B. E., & Howard, M. O. (2010). Psychiatriccorrelates of bullying in the United States: Findingsfrom a national sample. Psychiatric Quarterly, 81,183–195.

Wertheim, C., & Leyser, Y. (2002). Efficacy beliefs, back-ground variables, and differentiated instruction of Israeliprospective teachers. The Journal of Educational Research,11, 54–62.

Williams, D. M., Anderson, E. S., & Winett, R. A. (2005). Areview of the outcome expectancy construct in physical activityresearch. The Society of Behavioural Medicine, 29, 70–79.

RECEIVED: March 26, 2013ACCEPTED: December 6, 2013Available online 14 December 2013