An adult attachment perspective on the student–teacher relationship & classroom management...

10
An adult attachment perspective on the student–teacher relationship & classroom management difficulties Philip Riley * Faculty of Education, Monash University, Building 6, Wellington Road, Clayton, Victoria 3800, Australia article info Article history: Received 1 August 2008 Received in revised form 9 November 2008 Accepted 26 November 2008 Keywords: Reciprocal attachment Student–teacher relationships Unconscious motivation abstract To maintain a professional identity, teachers are to some degree dependent on their student’s mental representations of, and interactions with, them. This affords students’ relational power over teachers possibly invoking a unique form of attachment dependence and responding in some teachers. Data reported in this paper were drawn from a larger research project which asked 11 questions about the nature of the teacher–student relationship. The attachment styles of 291 pre-service and experienced elementary and secondary school teachers were examined using the Experience in Close Relationships Questionnaire (ECR). Significant differences were found for teacher type (elementary versus secondary), experience, age and gender. Implications for classroom relationships, management and teacher educa- tion are discussed. Ó 2008 Elsevier Ltd. All rights reserved. 1. Introduction It is self-evident that the students need their teachers in many types of learning environments. They rely on their teachers to be both predictable, as the secure base, and appropriately challenging; facilitating learning experiences to help them explore their world. To appropriate Winnicott’s term, successful teachers are those who empathically fail the student; intuiting when and how to let the student struggle to achieve a desired goal and when more support is needed (Winnicott, Winnicott, Shepherd, & Davis, 1986). In this way teachers are not unlike parents. In the ideal situation the teacher provides firm support, and empathically fails each student, so they learn not only about the world around them, but also their self-efficacy within it. This is the common attachment model applied to education: teacher as care-giver and student as care- seeker. However, if the teacher has attachment needs that only the students can fulfil, a dyadic rather than unidirectional attachment between teacher and students can and will develop. In this situa- tion the adult attachment model of reciprocal care-giving and care- seeking is a more appropriate lens through which to view the teacher–student relationship. Application of adult attachment principles to the teacher– student relationship reframes the way the relationship(s) are viewed and offers teachers new ways to inform and improve their classroom practice. Kesner (2000) pointed out that while all attachment relationships are close relationships, the reverse is not always true. Citing Bowlby’s (1982) evidence of children forming attachments to significant adults other than parents, he postulated that ‘‘Perhaps there is no other nonfamilial adult that is more significant in a child’s life than his or her teacher’’ (Kesner, 2000, p. 134). The quality of the relationship with a teacher can protect the child from academic failure (Pianta, Steinberg, & Rollins, 1995). Therefore the exploration of the student–teacher relationship from many perspectives is sensible. The larger research project on which this paper draws looked at 11 questions about the less obvious, but equally important aspects of the teacher–student relationship, based on the proposition that adult attachment theory is more widely applicable than its current research base. In the following sections the attachment forces on both teachers and students are briefly outlined. 1.1. A brief overview of attachment processes applied to classroom settings Attachment theory (Bowlby, 1969, 1975, 1980, 1982, 1988) is at essence a theory of human relating and identity. It adopts a constructivist view of the world where a child learns about relationship processes through experience of them. There are three basic constructs of the attachment behavioural system that are particularly relevant to teachers and teaching. The first is separation anxiety, a specific form of anxiety related to distance regulation with one’s attachment figure: the greater the distance, the greater the anxiety. As the anxiety increases so too do specific behaviours * Tel.: þ61 3 9905 2546; fax: þ61 3 9905 2779. E-mail address: [email protected] Contents lists available at ScienceDirect Teaching and Teacher Education journal homepage: www.elsevier.com/locate/tate 0742-051X/$ – see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.tate.2008.11.018 Teaching and Teacher Education 25 (2009) 626–635

Transcript of An adult attachment perspective on the student–teacher relationship & classroom management...

lable at ScienceDirect

Teaching and Teacher Education 25 (2009) 626–635

Contents lists avai

Teaching and Teacher Education

journal homepage: www.elsevier .com/locate/ tate

An adult attachment perspective on the student–teacher relationship &classroom management difficulties

Philip Riley*

Faculty of Education, Monash University, Building 6, Wellington Road, Clayton, Victoria 3800, Australia

a r t i c l e i n f o

Article history:Received 1 August 2008Received in revised form9 November 2008Accepted 26 November 2008

Keywords:Reciprocal attachmentStudent–teacher relationshipsUnconscious motivation

* Tel.: þ61 3 9905 2546; fax: þ61 3 9905 2779.E-mail address: [email protected]

0742-051X/$ – see front matter � 2008 Elsevier Ltd.doi:10.1016/j.tate.2008.11.018

a b s t r a c t

To maintain a professional identity, teachers are to some degree dependent on their student’s mentalrepresentations of, and interactions with, them. This affords students’ relational power over teacherspossibly invoking a unique form of attachment dependence and responding in some teachers. Datareported in this paper were drawn from a larger research project which asked 11 questions about thenature of the teacher–student relationship. The attachment styles of 291 pre-service and experiencedelementary and secondary school teachers were examined using the Experience in Close RelationshipsQuestionnaire (ECR). Significant differences were found for teacher type (elementary versus secondary),experience, age and gender. Implications for classroom relationships, management and teacher educa-tion are discussed.

� 2008 Elsevier Ltd. All rights reserved.

1. Introduction

It is self-evident that the students need their teachers in manytypes of learning environments. They rely on their teachers to beboth predictable, as the secure base, and appropriately challenging;facilitating learning experiences to help them explore their world.To appropriate Winnicott’s term, successful teachers are those whoempathically fail the student; intuiting when and how to let thestudent struggle to achieve a desired goal and when more supportis needed (Winnicott, Winnicott, Shepherd, & Davis, 1986). In thisway teachers are not unlike parents. In the ideal situation theteacher provides firm support, and empathically fails each student,so they learn not only about the world around them, but also theirself-efficacy within it. This is the common attachment modelapplied to education: teacher as care-giver and student as care-seeker. However, if the teacher has attachment needs that only thestudents can fulfil, a dyadic rather than unidirectional attachmentbetween teacher and students can and will develop. In this situa-tion the adult attachment model of reciprocal care-giving and care-seeking is a more appropriate lens through which to view theteacher–student relationship.

Application of adult attachment principles to the teacher–student relationship reframes the way the relationship(s) areviewed and offers teachers new ways to inform and improve theirclassroom practice. Kesner (2000) pointed out that while all

u.au

All rights reserved.

attachment relationships are close relationships, the reverse is notalways true. Citing Bowlby’s (1982) evidence of children formingattachments to significant adults other than parents, he postulatedthat ‘‘Perhaps there is no other nonfamilial adult that is moresignificant in a child’s life than his or her teacher’’ (Kesner, 2000,p. 134).

The quality of the relationship with a teacher can protect thechild from academic failure (Pianta, Steinberg, & Rollins, 1995).Therefore the exploration of the student–teacher relationship frommany perspectives is sensible. The larger research project on whichthis paper draws looked at 11 questions about the less obvious, butequally important aspects of the teacher–student relationship,based on the proposition that adult attachment theory is morewidely applicable than its current research base. In the followingsections the attachment forces on both teachers and students arebriefly outlined.

1.1. A brief overview of attachment processes applied toclassroom settings

Attachment theory (Bowlby, 1969, 1975, 1980, 1982, 1988) is atessence a theory of human relating and identity. It adoptsa constructivist view of the world where a child learns aboutrelationship processes through experience of them. There are threebasic constructs of the attachment behavioural system that areparticularly relevant to teachers and teaching. The first is separationanxiety, a specific form of anxiety related to distance regulationwith one’s attachment figure: the greater the distance, the greaterthe anxiety. As the anxiety increases so too do specific behaviours

P. Riley / Teaching and Teacher Education 25 (2009) 626–635 627

as attempts to reduce it. These attempts are the second importantconstruct, known as separation protest behaviours, and ofteninclude aggression. In very young children separation anxiety isevoked by physical distance from the primary care-giver, while inolder children and adults it is a related to the level of perceivedemotional distance from the care-giver, known as the attachmentfigure or secure base. The third of the constructs is the degree towhich avoidance of close relationships results from separationanxiety. Securely attached individuals have lower levels of sepa-ration anxiety than their insecurely attached peers and maintainappropriate distance from others (Ainsworth, Blehar, Waters, &Wall, 1978). Insecurely attached individuals fall into three broadcategories, depending on their separation anxiety levels andavoidance responses: those who have high levels of anxiety but lowavoidance (preoccupied), those with low levels of anxiety but highlevels of avoidance (dismissing) and those who score highly on bothdimensions (fearful) (Bartholomew, 1994; Bartholomew & Hor-owitz, 1991).

An individual’s attachment ‘style’ is largely determined by theage of three years. At this age the child has developed a personalinternal working model of relationships (Bowlby, 1982) throughrepeated positive and negative experiences with others. Similar toa schema, the internal working model is used to predict how he orshe will be treated and how to treat others. Young children whohave received consistent, appropriate and predictable care bytrusted others are likely to view most people they meet, andthemselves, as trustworthy, safe, dependable and deserving of care.On the other hand, children who have been dealt with inconsis-tently, or worse, are likely to be wary in new relationships, becomedistrustful of others and may see themselves as unworthy of care(Bowlby, 1988).

As individuals develop beyond the third year, the need for careand affiliation, fundamental human drives, does not diminish asa result of insecure attachment, but may become hidden asa defence against painful past experiences. To complicate thingsfurther, the type of care sort from others also changes with devel-opment. The most notable example of this is romantic pairing.

Upon reaching adulthood, people seek specific care from othersin new situations such as romantic partnerships which are formedas reciprocal rather than one-way dyads. It is not surprising thatmost adult attachment research has been applied to the romanticpair-bond. Each partner in this type of relationship alternatesbetween the role of care-giver and care-seeker depending oncontext. This is the main difference between the adult and child-hood attachment models (Rholes & Simpson, 2004).

Hazan and Shaver (1987) were the first to find persuasiveevidence that the connection to a significant other established byadults as a romantic bond is best conceived as a kind of repetition ofthe initial attachment process, and many others have built on thisfoundation since (Bartholomew & Horowitz, 1991; Cassidy &Shaver, 1999; Collins & Feeney, 2000; Fraley & Shaver, 2000;Mikulincer & Goodman, 2006; Rholes & Simpson, 2004; Simpson &Rholes, 1998). A romantic relationship also offers the opportunity ofa corrective emotional attachment experience for insecurelyattached individuals. But, relationships with significant others aremade and broken many times before adulthood, and many of theseare reciprocal relationships. Therefore, they too, are best examinedfrom an adult attachment perspective, which to date has not yetoccurred. This study is one attempt to redress this.

There are four compelling reasons to apply the adult attachmentmodel to children’s relationships beyond the third year and inparticular the relationships they will form with their teachersduring the school years. The first is that the initial attachmentphase is completed by that time. The second acknowledges thatsignificant peer relationships are formed by children between threeyears and adulthood; friendships which are reciprocal and are not

one-way relationships. Thirdly, children form strong bonds tosignificant adults outside of the family such as teachers (Kesner,2000). Fourthly, the insecure child does not wait to become adult toseek corrective emotional experiences which go beyond acquain-tance or friendship, through strong contact with others outside theimmediate attachment circle; and, an obvious place to find sucha care-giver is at school. These children may attempt to form strongattachments to their teachers and many succeed. For some this mayalso result in a conscious or unconscious wish to stay within theschool setting and become like the people who cared for them:their teachers. If this is the case it is likely to become the basis forthe wish to develop a professional identity of teacher. Empiricalevidence for an early decision to become a teacher was commonamong teachers who participated in a major study attempting toexplore the unconscious motivation to teach (Wright & Sherman,1963). The complete development of such an identity however, isdetermined through interactions with students as well as indi-vidual motivation. This notion will be expanded in Section 1.3, butfirstly the connection between attachment and professional iden-tity needs some explanation.

1.2. Attachment and professional identity

The teacher needs to form a working relationship with at leastone student to maintain a professional identity of ‘teacher’. Thisprofessional identity cannot exist without the internal represen-tation by that teacher of the student(s) and the relationship thatexists between them. However, the dyad is not a simple onebecause of uneven distribution of power, legalities and responsi-bilities that are not equally shared in the teacher–student rela-tionship. This suggests that the teacher–student dyad is a uniquevariant. While it is a dyadic relationship, and therefore carries someof the common attributes of all dyads, it is not the same asa romantic dyad, about which most of the adult attachment liter-ature exists. The teacher–student relationship is not and should notbe considered a romantic relationship. Yet many teachers will attestto powerful ties that exist between teachers and students, bothfrom their own experience and witnessing their colleagues. Theseties can be difficult to articulate and appear to go beyond profes-sional interactions with students to a more personal level ofemotional bond. Adult attachment theory offers a comprehensiveway to examine these subjective and intersubjective processes,without confusing them with romantic attachments.

A further complication to understanding the teacher–studentdyad is that while the teacher depends on the relationship withstudents to maintain a professional identity, this is not always thecase for the students. This makes the teacher a care-seeker from thestudents, thus placing the student in the role of care-giver, a rolewhich some students may not be ready to accept due to their ownattachment history. A student can exist, and learn, with or withoutthe presence of a teacher. In fact much of teachers’ work is con-cerned with becoming redundant by producing active, indepen-dent, resourceful students who will flourish on their own. However,the teacher also maintains the role of care-giver to the students, bybeing older and wiser and charged with the legal and ethicalresponsibility for each of their students. Thus, in the classroomcontext, dual roles complicate the attachment behavioural system’soperation for both teacher and student.

As there can be students without teachers but not teacherswithout students, the relationship between the teacher and thestudent, is therefore the underpinning of a professional teachingidentity running parallel to the professional working relationshipwith students, as the role of care-giver and care-seeker is carried byboth parties in the relationship. When this relationship is examinedthrough the adult attachment lens the students acquire a great dealof power in the relationship; unconscious perhaps but power

P. Riley / Teaching and Teacher Education 25 (2009) 626–635628

never-the-less. A threat of emotional separation from the class,such as a management issue with a ‘‘difficult’’ individual student,threatens the teacher’s proximity to the group as a whole. Thisraises the level of separation anxiety to the attachment object, theclass, and the instigation of attachment behaviours including angryprotest to stop the perceived distancing may be provoked. Aninsecurely attached teacher, who may be unconsciously seekinga corrective emotional experience through attachment to theirstudents, is vulnerable to rejection by them. Therefore, the threatof, or actual rejection by students may be an experience likely toactivate the attachment behavioural system in the teacher (Bowlby,1982). This would inevitably lead to protest behaviours, includingovert and covert aggression directed toward the students, to reducethe separation anxiety caused by the actual (or perceived) rejection.

1.3. Attachment and unconscious motivation to join the teachingprofession: strength, vulnerability or both?

As early as 1963, before the publication of the first volume ofBowlby’s (1969) attachment trilogy, Wright and Sherman (1963)presented arguments that ‘‘teacher types are determined, to a largeextent, by identification with parents and former teachers’’ (p. 67).They considered the differing impacts of parents and formerteachers on the development of a new teacher. Parents were foundto provide moral and intellectual discipline while former teachersprovided the primary source of sympathy. This led them toconclude that teachers ‘‘think and act in the classroom in waysclosely related to those of their former teachers’’ (Wright & Sher-man, 1963, p. 67) and, further, that this was not a consciouslyadopted style but rather developed unconsciously. They employedstrongly emotive language, suggesting that the relationshipbetween teacher and student is a form of love, needed by bothteacher and student. This language has since fallen into disuse inthe educational discourse, perhaps in reaction to its strength andsuggestive elements. Attachment theory, which was being devel-oped at the same time as Wright and Sherman were researchingteacher–student relationships, also directly addresses the conceptsof love. In describing different types of teachers Wright and Sher-man (1963) presented rich descriptions of the secure and insecureteacher that could have been used to explicate attachment theory.In the following passage the secure teacher is described firstly,followed by a description of an insecure teacher in search ofa corrective emotional experience.

We propose that for one type of loving teacher her behavior isa means of keeping the love of the earliest mother image sheremembers. She remembers that her mother was loving in thefirst stage and that giving inwas rewarded with love in the second.She identifies with the loving, supportive mother by repeatingand thus preserving this valued image. This dependency is satis-fying to her, for what is most important is the love that she hadand which she now keeps by embodying it in her own behavior.There is another type of loving teacher – one who was deprivedof the rewards of love in childhood. She succumbed to hermother’s demands out of the fear of punishment rather than theprospect of reward. This teacher’s behavior is guided by herdesire to make up for her loss. In contrast to the first kind ofloving teacher, she loves her pupils not only in order to be lovedby them, but also to gain the vicarious restitutional gratificationof providing children with the love which was once deprived her(Wright & Sherman, 1963, p. 71).

Restitutional gratification, care-seeking or corrective emotionalexperience, it is this need which makes some teachers vulnerable torejection by students.

Given that the relationship teachers seek from students is partlyone of care, it is arguable that some teachers may enter the

profession partly motivated by an unconscious desire for correctiveemotional experiences, through the formation of new attachmentsto their students. The present study was designed to explore this byexamining the attachment styles of pre-service and experiencedteachers. The investigation was operationalised around thefollowing sub-questions.

1.4. Research questions

1.4.1 Do some people chose teaching unconsciously, desiringa corrective emotional experience?

If a corrective emotional experience results from workingclosely with students, experienced teachers’ scores on anxiety andavoidance should be lower than pre-service teachers.

1.4.2 Does the experience of teaching affect teachers differently(elementary versus secondary)?

The chances for a corrective emotional experience occurringshould be more likely for elementary teachers, due to the differ-ences in the number and duration of student interactions theyhave.

1.4.3 Does the experience of teaching impact on the attachmentstyle of teachers over time?

This question overlaps the first question. Corrective emotionalexperiences should have an impact on attachment style over time;therefore experienced teachers should have lower levels of anxietyand avoidance.

1.4.4 Are age, gender or parenting experience mediating variables?

2. Method

2.1. Participants

Data were obtained from 258 pre-service teachers and 50experienced teachers, of which 204 (66.4%) were females and 103(33.6%) males (N¼ 308). Participants’ ages ranged from 21 to 59years (M¼ 29.04, SD¼ 6.81). All were currently undertaking studyat post-graduate level. The pre-service teachers (82.5% of thesample: n¼ 258) were studying for the Graduate Diploma inEducation. There were 168 elementary, and 90 secondary partici-pants in this cohort. The experienced teachers (17.5% of the sample:n¼ 50) were studying at Masters’ Level or above. The criterion forexperienced teachers was five or more years in the teaching service.In the experienced cohort there were 34 elementary and 16secondary teachers. A further 3 teachers had dual elementary andsecondary qualifications. Most participants’ first language wasEnglish: 84.4% (n¼ 266), 8.6% were from non-English speakingbackgrounds (n¼ 27) and 1% (n¼ 3) were bilingual. Just over onequarter of the cohort were parents: 26.7% (n¼ 84). Table 1 displaysdemographic data by teacher type.

2.2. Procedure

The Experiences in Close Relationships Questionnaire (ECR)(Brennan, Clark, & Shaver, 1998) was administered during work-shops focussing on teacher–student interactions and classroommanagement following a lecture on attachment theory. All work-shop participants were offered the chance to fill out the ques-tionnaire as part of their coursework. The questionnaire wascompleted, scored and interpreted anonymously, under guidancefrom the investigator and participants discussed the implications of

Table 1Demographic data.

Elementary Secondary Total

n % n % n %

Pre-service teachers Female 118 46.0 61 23.7 178 69.3Male 50 19.5 29 11.3 79 30.7Total 168 65.4 90 35.0 257 100

Experienced teachers Female 20 40 6 12.0 26 52.0Male 14 28 10 20.0 24 48.0Total 34 68.0 16 32.0 50 100

Total Female 138 44.7 67 21.7 204 66.0Male 64 20.7 39 12.6 103 33.0Total 206 65.4 103 34.3 309 100

Table 2ECR item pairs for social desirability and acquiescence deviance calculations.

2. I worry about being abandoned.22. I do not often worry about being abandoned.

3. I am very comfortable being close to romantic partners.23. I prefer not to be too close to romantic partners.

7. I get uncomfortable when a romantic partner wants to be very close.12. I often want to merge completely with romantic partners, and this

sometimes scares them away.15. I feel comfortable sharing my private thoughts and feelings with my

partner.1. I prefer not to show a partner how I feel deep down.

19. I find it relatively easy to get close to my partner.13. I am nervous when partners get too close to me.25. I tell my partner just about everything.

9. I don’t feel comfortable opening up to romantic partners.

P. Riley / Teaching and Teacher Education 25 (2009) 626–635 629

their own attachment styles with regard to their classroom practiceand their knowledge of the theory. To increase the chances of validresponses to the instrument, it was determined that participantswould be more likely to see the benefit of honest responses to theECR, which contains a number of challenging questions, only afterthey had gained some knowledge of attachment theory and itsconnection to their professional lives. At the conclusion of theworkshop participants were invited to submit the anonymousquestionnaire to be included in this study. Participation rates forthe study were approximately 90% for the pre-service cohort and74% for the experienced teachers. The differences in response ratewere unexpected, and may reveal a sample bias. It could be thatmore fearfully attached teachers were reluctant to submit theirquestionnaire for scrutiny. Alternatively dismissing types may havebeen more reluctant to submit their questionnaire because ofa poor view of others’ use of the data.

3. Results

3.1. Data screening

All data were analysed using SPSS 14.0. The data were initiallyviewed for accuracy of input, out of range values, plausible means,standard deviations, and univariate outliers.

3.2. Missing data

All missing data were demographic. Five cases were missingage data (1.6%). Seven cases were missing gender data and did notdisclose whether they were parents (2.2%). Nineteen cases (6%) didnot list the first language of the participant, and 26 (8.3%) did notlist the undergraduate degree obtained before entering theGraduate Diploma of Education course. Patterns of missing datawere sought using SPSS MVA. No patterns that could affect resultswere discovered therefore cases with missing data were retainedfor all analyses, but excluded from analyses where any data wasmissing.

3.3. Univariate outliers

Using a conservative cut-off value of �3 SD three outliers werediscovered for first language spoken at home. In each case theparticipant was found to have grown up in a bilingual household.These cases were included in the analyses as there was noconceptual reason for exclusion on the grounds of bilingualism.One univariate outlier was discovered for avoidance (Z¼ 3.25). Thisparticipant was approaching univariate outlier status on anxietyalso (Z¼ 2.72). This compounded when computing Z scores forattachment styles: secure (Z¼ 3.67), fearful (Z¼ 3.83), preoccupied(Z¼ 3.47) and dismissing (Z¼ 3.91). This data was deleted from thedata set leaving N¼ 307 cases for analysis.

3.4. Reliability analysis

According to Brennan et al. (1998) the ECR has good internalconsistency with a Cronbach alpha reporting of 0.97 for avoidanceand 0.91 for anxiety. Internal consistency for the current study wascalculated with a Cronbach alpha coefficient reported for avoidance(.92) and anxiety (.91).

3.5. Controlling social desirability and acquiescence

Brennan et al. (1998) caution that ‘‘response bias’’ in self-reportmeasures of attachment is a potential threat to validity. This hasbeen supported by others also (Allen, Stein, Fonagy, Fultz, & Target,2005; Bartholomew & Horowitz, 1991; George, Kaplan, & Main,1984; Main, 1990). Many attachment researchers feel that the use ofinterview measures is more accurate because ‘‘self-report relies on‘participants’ honesty and self-insight, which are probably limitedin any case but especially so when fears and defences are at issue’’(Brennan et al., 1998, p. 22).

Hofstee, Berge, and Hendriks (1998) propose that ‘‘social desir-ability’’ and ‘‘acquiescent responding’’ are highly correlated ‘‘noise’’variance present in all self-report instruments which diminish theunderlying component’s variance. They suggest that social desir-ability and acquiescent responding could contribute up to 10% ofthe variance in some studies. They offer a post facto removaltechnique by calculating a mean acquiescence/social desirabilityvariance score as a deviation score from the obtained item scores.This can then be removed from each respondent’s item scores.Since the ECR contains positively and negatively worded itemsrelating to the same underlying constructs the method of identi-fying and removing acquiescence and socially desirable respondingwas implemented by calculating the ‘‘average-over-opposites andthe scale midpoint’’ (p. 898) to calculate a mean acquiescencedeviance score. This score also contains socially desirable variancebecause opposite items ‘‘tend to have opposite social desirabilities’’(p. 899). The obtained mean acquiescence/social desirability scorewas partialled out by subtracting it from responses on each of the36 items for all respondents (see Table 2 for item pairs). Socialdesirability/acquiescence contributed 4.7% of the variance to thedata set. The resulting item scores, now free from socially desirableresponding, were re-screened by Teacher Type (elementary versussecondary, pre-service versus experienced, and gender). Usinga cut-off value of �3 SD from the mean, 3 univariate outliers foranxiety were discovered in the pre-service elementary cohort.These were removed for between-groups analysis. Fig. 1 displaysthe means for each of the different teaching categories examined inlater analysis. It is interesting to note that only one category ofTeacher Type, experienced elementary males, reveals meanAvoidance levels higher than Anxiety levels.

Expe

rienc

edSe

cond

ary

Mal

e

Expe

rienc

edSe

cond

ary

Fem

ale

Pre

Serv

ice

Seco

ndar

yM

ale

Pre

Serv

ice

Seco

ndar

yFe

mal

e

Expe

rienc

edEl

emen

tary

Mal

e

Expe

rienc

edEl

emen

tary

Fem

ale

Pre

Serv

ice

Elem

enta

ryM

ale

Pre

Serv

ice

Elem

enta

ryFe

mal

e

Teacher Type

10

8

6

4

2

0

-2

AnxietyAvoidance

Fig. 1. Boxplot of mean scores for Anxiety and Avoidance by Teacher Type. (*) Out-lier>�3 standard deviations from the mean; (B) outlier< 3 standard deviations fromthe mean.

Table 4Attachment categories calculated using SPSS K-Means Analysis, reported aspercentage of sample in each category.

N Secure(%)

Preoccupied(%)

Fearful(%)

Dismissing(%)a

DiscriminantK-Means fortotal sample

Anx Avoid Anx Avoid Anx Avoid Anx Avoid1.16 0.77 3.11 1.86 5.44 3.47 2.99 3.92

TotaldiscriminantK-Mean Pairs(SPSS)

307 19.05 44.76 17.78 18.41

a Discriminant K-Means do not fit this category well.

P. Riley / Teaching and Teacher Education 25 (2009) 626–635630

Brennan et al. (1998) provided weighting criteria for categoris-ing individuals by calculating attachment membership taxonomy(secure, preoccupied, fearful and dismissing) using Anxiety andAvoidance mean scores. When these weightings were applied tothe current sample, analysis of the teachers’ responses placed theminto only two of the four categories: preoccupied and fearful(see Table 3). This was unexpected.

Although Brennan et al. (1998) reported that the ECR is perhapsoverly sensitive for placing individuals into the three insecurecategories, they still found 30.4% of their sample to be in the securecategory. Bartholomew and Shaver (1998) addressed the discrimi-nant function issues and suggested that an individual’s catego-risation was dependent to some extent on context. This has beensupported by a number of authors since (Fraley & Shaver, 2000;Fraley & Spieker, 2003; Fraley, Waller, & Brennan, 2000). Therefore,SPSS K-Means Cluster Analysis was used to ‘force’ a four-categorysolution from the two attachment variables, Avoidance andAnxiety. The mean combinations indicated that there were fourcategories of attachment styles for teachers when using thismethod. Table 4 displays calculated percentages of groupmembership using this method.

The results of this procedure are consistent with the assertion thatadult attachment dimensions are continuous, and that researchersshould study the results of the ECR in dimensional rather thancategorical terms (Fraley & Shaver, 2000; Fraley & Spieker, 2003;Fraley et al., 2000). It was therefore determined that attachment

Table 3Attachment categories calculated using Brennan et al.’s (1998) weightings, reportedas percentage of sample in each category.

N Secure(%)

Preoccupied(%)

Fearful(%)

Dismissing(%)

Brennan et al. (1998)reported classifications

1082 30.4 24.4 24.4 20.8

Brennan et al. (1998)weightings for current study

307 0 19.2 80.8 0

category group scores would not be used for analysis beyond thispoint, but are useful to retain for interpretation of results.

3.6. Research questions addressed

A five-way, between-groups MANOVA simultaneously consid-ered the two dependent variables (Anxiety and Avoidance meanscores to represent attachment style) to explore potential differ-ences in attachment style by Teacher Type (elementary versussecondary, pre-service versus experienced, parent versus non-parent, age (3 levels) and gender). Both Pallant (2007) and Tabach-nick and Fidell (2001) advise that MANOVA is less powerful whenthe two dependent variables are orthogonal. It has been argued byAinsworth et al. (1978) that the two dependent variables areconceptually orthogonal. This was confirmed by the oblique rotationwhich showed a weak correlation between the factors (r¼�.002).Given the weak correlation between the variables significant resultsmay be more reliable and indicate external validity.

Preliminary assumption testing was conducted to check formissing values, normality, linearity, univariate and multivariateoutliers, cell sizes, homogeneity of variance–covariance matrices,and multicolinearity. One violation was noted: Levene’s test ofequality of error variances for avoidance F (35, 271)¼ 1.72,p¼ 0.010. Tabachnick and Fidell (2001) advise that in such casesPillai’s trace should be used to determine the significance ofmultivariate effects and a more conservative alpha level (<.025)should be adopted for inference tests. This procedure wasfollowed.

The results of the MANOVA provided support for the first andthird hypotheses: 1.4.1, that some teachers may be experiencingcorrective emotional experiences as a result of their work, fromwhich it may be inferred that an unconscious motivation to teach insearch of the experience might exist, and 1.4.3, the impact ofteaching experience on teachers’ attachment style of over time. Thedependent variables (DVs) were significantly affected by Experi-ence (pre-service versus experienced) F (2, 270)¼ 10.07, p¼ 0.000,partial h2¼ 0.032 and Age Group F (4, 542)¼ 3.78, p¼ 0.005, partialh2¼ 0.027. No significant effect was found on the DVs for Parentalstatus (parent versus non-parent), F (2, 270)¼ 0.39, p¼ 0.675,Gender F (2, 270)¼ 1.47, p¼ 0.233, or Teaching Level (elementaryversus secondary).

Univariate tests on each DV were significant for Experience(Anxiety: F [1, 271]¼ 18.01, p¼ 0.000; partial h2¼ 0.062. Avoid-ance: F [1, 271]¼ 11.12, p¼ 0.001; partial h2¼ 0.062) and Age(Anxiety: F [1, 271]¼ 18.01, p¼ 0.000; partial h2¼ 0.062. Avoid-ance: F [1, 271]¼ 11.12, p¼ 0.001; partial h2¼ 0.062). The results ofthe significant IVs were then considered separately.

Support was also found for the second hypothesis, that theexperience of teaching either elementary or secondary studentsaffects teachers differently. A significant difference was found forTeacher Type (pre-service versus experienced) on the multivariatetest F (2, 270)¼ 10.73, p¼ 0.000. Univariate tests revealed that

P. Riley / Teaching and Teacher Education 25 (2009) 626–635 631

a significant difference existed for level of Avoidance, F (1,271)¼ 11.17, p¼ 0.001; partial h2¼ 0.039. Inspection of the meansindicated that pre-service teachers (M¼ 2.49, SD¼ 1.20) were moreavoidant than experienced teachers (M¼ 1.40, SD¼ 1.68). Asignificant difference also existed between the two groups for levelof Anxiety F (1, 271)¼ 18.01, p¼ 0.000; partial h2¼ 0.062. Inspec-tion of the means indicated that pre-service teachers (M¼ 3.38,SD¼ 1.38) were also more anxious about relationships than expe-rienced teachers (M¼ 1.88, SD¼ 1.50). A number of significantdifferences in levels of Anxiety and Avoidance were found between

Table 5Pairwise mean differences, p values and confidence intervals for teachers by type on lev

(I) Teach type (J) Teach type Mean d

Pre-service elementary female Pre-service elementary male �0.34Experienced elementary female 1.15Experienced elementary male 0.95Pre-service secondary female �0.58Pre-service secondary male �0.43Experienced secondary female �0.11Experienced secondary male 0.64

Pre-service elementary male Pre-service elementary female 0.34Experienced elementary female 1.48Experienced elementary male 1.29Pre-service secondary female �0.24Pre-service secondary male �0.09Experienced secondary female 0.23Experienced secondary male 0.98

Experienced elementary female Pre-service elementary female �1.15Pre-service elementary male �1.48Experienced elementary male �0.20Pre-service secondary female �1.72Pre-service secondary male �1.58Experienced secondary female �1.25Experienced secondary male �0.50

Experienced elementary male Pre-service elementary female �0.95Pre-service elementary male �1.29Experienced elementary female 0.20Pre-service secondary female �1.53Pre-service secondary male �1.38Experienced secondary female �1.06Experienced secondary male �0.31

Pre-service secondary female Pre-service elementary female 0.58Pre-service elementary male 0.24Experienced elementary female 1.72Experienced elementary male 1.53Pre-service secondary male 0.15Experienced secondary female 0.47Experienced secondary male 1.22

Pre-service secondary male Pre-service elementary female 0.43Pre-service elementary male 0.09Experienced elementary female 1.58Experienced elementary male 1.38Pre-service secondary female �0.15Experienced secondary female 0.32Experienced secondary male 1.07

Experienced secondary female Pre-service elementary female 0.11Pre-service elementary male �0.23Experienced elementary female 1.25Experienced elementary male 1.06Pre-service secondary female �0.47Pre-service secondary male �0.32Experienced secondary male 0.75

Experienced secondary male Pre-service elementary female �0.64Pre-service elementary male �0.98Experienced elementary female 0.50Experienced elementary male 0.31Pre-service secondary female �1.22Pre-service secondary male �1.07Experienced secondary female �0.75

*Bonferroni corrected significant alpha level¼ p< 0.025.

the teaching groups. The pairwise differences, significance levelsand confidence intervals for all Teacher Type groups are displayedin Tables 5 and 6.

Examination of the potential mediating variable, Age, wascarried out by creating three age group bands using SPSS transform.The groups were: 1) less than or equal to 25 years (n¼ 128, 40.8%);2) 26–30 years (n¼ 89, 28.3%); and, 3) 31 or more years (n¼ 92,29.3%). Age Group was found to be a significant variable forAvoidance level, F (2, 270) 7.76, p¼ 0.001, but was non-significantfor Anxiety level F (2, 270) 1.91, p¼ 0.150. Inspection of the graphs

els of avoidance.

ifference (I� J) Std. error p* 95% confidence interval

Lower Upper

0.21 1.00 �1.01 0.340.31 0.01* 0.18 2.110.36 0.24 �0.18 2.080.20 0.12 �1.21 0.050.26 1.00 �1.26 0.400.53 1.00 �1.78 1.570.42 1.00 �0.67 1.96

0.21 1.00 �0.34 1.010.34 0.00* 0.42 2.540.38 0.03 0.08 2.490.24 1.00 �1.01 0.520.30 1.00 �1.03 0.840.55 1.00 �1.50 1.950.44 0.74 �0.41 2.36

0.31 0.01* �2.11 �0.180.34 0.00* �2.54 �0.420.44 1.00 �1.59 1.200.33 0.00* �2.75 �0.690.37 0.00* �2.74 �0.410.59 0.96 �3.12 0.610.49 1.00 �2.05 1.05

0.36 0.24 �2.08 0.180.38 0.03 �2.49 �0.080.44 1.00 �1.20 1.590.38 0.00* �2.71 �0.340.41 0.03 �2.68 �0.080.62 1.00 �3.01 0.890.52 1.00 �1.96 1.35

0.20 0.12 �0.05 1.210.24 1.00 �0.52 1.010.33 0.00* 0.69 2.750.38 0.00* 0.34 2.710.29 1.00 �0.75 1.050.54 1.00 �1.24 2.180.43 0.14 �0.14 2.59

0.26 1.00 �0.40 1.260.30 1.00 �0.84 1.030.37 0.00* 0.41 2.740.41 0.03 0.08 2.680.29 1.00 �1.05 0.750.57 1.00 �1.47 2.110.46 0.61 �0.39 2.54

0.53 1.00 �1.57 1.780.55 1.00 �1.95 1.500.59 0.96 �0.61 3.120.62 1.00 �0.89 3.010.54 1.00 �2.18 1.240.57 1.00 �2.11 1.470.65 1.00 �1.31 2.82

0.42 1.00 �1.96 0.670.44 0.74 �2.36 0.410.49 1.00 �1.05 2.050.52 1.00 �1.35 1.960.43 0.14 �2.59 0.140.46 0.61 �2.54 0.390.65 1.00 �2.82 1.31

Table 6Pairwise mean differences, p values and confidence intervals for teachers by type on levels of anxiety.

(I) Teach type (J) Teach type Mean difference (I� J) Std. error p* 95% confidence interval

Lower Upper

Pre-service elementary female Pre-service elementary femalePre-service elementary male �0.18 0.23 1.00 �0.92 0.55Experienced elementary female 1.24 0.33 0.01* 0.19 2.30Experienced elementary male 2.09 0.39 0.00* 0.86 3.32Pre-service secondary female �0.34 0.22 1.00 �1.03 0.34Pre-service secondary male �0.07 0.29 1.00 �0.97 0.83Experienced secondary female 0.56 0.58 1.00 �1.27 2.38Experienced secondary male 1.12 0.46 0.40 �0.31 2.56

Pre-service elementary male Pre-service elementary female 0.18 0.23 1.00 �0.55 0.92Experienced elementary female 1.43 0.37 0.00* 0.27 2.58Experienced elementary male 2.27 0.42 0.00* 0.95 3.59Pre-service secondary female �0.16 0.26 1.00 �0.99 0.67Pre-service secondary male 0.11 0.32 1.00 �0.90 1.13Experienced secondary female 0.74 0.60 1.00 �1.14 2.62Experienced secondary male 1.31 0.48 0.19 �0.20 2.82

Experienced elementary female Pre-service elementary female �1.24 0.33 0.01* �2.30 �0.19Pre-service elementary male �1.43 0.37 0.00* �2.58 �0.27Experienced elementary male 0.85 0.48 1.00 �0.67 2.36Pre-service secondary female �1.59 0.36 0.00* �2.71 �0.46Pre-service secondary male �1.31 0.40 0.03 �2.58 �0.05Experienced secondary female �0.69 0.64 1.00 �2.71 1.34Experienced secondary male �0.12 0.53 1.00 �1.81 1.57

Experienced elementary male Pre-service elementary female �2.09 0.39 0.00* �3.32 �0.86Pre-service elementary male �2.27 0.42 0.00* �3.59 �0.95Experienced elementary female �0.85 0.48 1.00 �2.36 0.67Pre-service secondary female �2.43 0.41 0.00* �3.72 �1.14Pre-service secondary male �2.16 0.45 0.00* �3.57 �0.74Experienced secondary female �1.53 0.67 0.67 �3.66 0.59Experienced secondary male �0.97 0.57 1.00 �2.77 0.84

Pre-service secondary female Pre-service elementary female 0.34 0.22 1.00 �0.34 1.03Pre-service elementary male 0.16 0.26 1.00 �0.67 0.99Experienced elementary female 1.59 0.36 0.00* 0.46 2.71Experienced elementary male 2.43 0.41 0.00* 1.14 3.72Pre-service secondary male 0.27 0.31 1.00 �0.71 1.26Experienced secondary female 0.90 0.59 1.00 �0.96 2.76Experienced secondary male 1.47 0.47 0.06 �0.02 2.95

Pre-service secondary male Pre-service elementary female 0.07 0.29 1.00 �0.83 0.97Pre-service elementary male �0.11 0.32 1.00 �1.13 0.90Experienced elementary female 1.31 0.40 0.03 0.05 2.58Experienced elementary male 2.16 0.45 0.00* 0.74 3.57Pre-service secondary female �0.27 0.31 1.00 �1.26 0.71Experienced secondary female 0.63 0.62 1.00 �1.33 2.58Experienced secondary male 1.19 0.51 0.54 �0.40 2.79

Experienced secondary female Pre-service elementary female �0.56 0.58 1.00 �2.38 1.27Pre-service elementary male �0.74 0.60 1.00 �2.62 1.14Experienced elementary female 0.69 0.64 1.00 �1.34 2.71Experienced elementary male 1.53 0.67 0.67 �0.59 3.66Pre-service secondary female �0.90 0.59 1.00 �2.76 0.96Pre-service secondary male �0.63 0.62 1.00 �2.58 1.33Experienced secondary male 0.57 0.71 1.00 �1.68 2.82

Experienced secondary male Pre-service elementary female �1.12 0.46 0.40 �2.56 0.31Pre-service elementary male �1.31 0.48 0.19 �2.82 0.20Experienced elementary female 0.12 0.53 1.00 �1.57 1.81Experienced elementary male 0.97 0.57 1.00 �0.84 2.77Pre-service secondary female �1.47 0.47 0.06 �2.95 0.02Pre-service secondary male �1.19 0.51 0.54 �2.79 0.40Experienced secondary female �0.57 0.71 1.00 �2.82 1.68

*Bonferroni corrected significant alpha level¼ p< 0.025.

P. Riley / Teaching and Teacher Education 25 (2009) 626–635632

revealed that Age Group was interacting with level of Experienceand Gender on level of Avoidance (see Fig. 2). A statistic for thisinteraction could not be computed in the MANOVA as both genderand experience contain only two levels.

3.6.1. Group comparison and interaction effectsThe descriptive statistics for each of the teaching groups on the

two dependent variables, Avoidance and Anxiety, are presented inTable 7, and displayed graphically in Fig. 2.

3.6.1.1. Comparison of between-groups levels of Avoidance. Usinga Bonferroni adjusted alpha of p< 0.025, significant differencesbetween teachers by type were identified. Pre-service elementarymales were significantly more avoidant than experiencedelementary females. Pre-service secondary females were signifi-cantly more avoidant than experienced elementary teachers andtheir experienced counterparts. Experienced elementary maleswere significantly less avoidant than pre-service secondaryfemales and the difference approached significance for pre-service

Experienced TeachersPre-Service Teachers

3.00

2.50

2.00

1.50

1.00

0.50

0.00 MaleFemale

Age <= 25

Experienced TeachersPre-ServiceTeachers

3.00

2.75

2.50

2.25

2.00

1.75

Male

Female

Age = 26 - 30

Experienced TeachersPre-Service Teachers

2.80

2.60

2.40

2.20

2.00 Male

FemaleAge = 31+

Fig. 2. Mean levels of avoidance for pre-service and experienced teachers by age bandsand gender.

Table 7Group means for teacher type, anxiety and avoidance.

Teacher type M SD N

Anxiety level Elementary Pre-service Female 2.25 1.34 117Male 3.43 1.43 50

Experienced Female 2.01 1.34 20Male 1.16 1.57 14

Secondary Pre-service Female 3.59 1.43 61Male 3.32 1.38 29

Experienced Female 2.69 1.68 6Male 2.13 1.05 10

Total 3.13 1.50 307

Avoidance level Elementary Pre-service Female 2.24 1.07 117Male 2.59 1.25 50

Experienced Female 1.10 1.10 20Male 1.30 1.29 14

Secondary Pre-service Female 2.82 1.31 61Male 2.67 1.21 29

Experienced Female 2.35 1.47 6Male 1.60 1.15 10

Total 2.31 1.35 307

P. Riley / Teaching and Teacher Education 25 (2009) 626–635 633

elementary and secondary males. Experienced elementary femaleswere significantly less avoidant than both pre-service secondaryfemales, and males, however, experienced secondary males andpre-service elementary females were not significantly differentfrom any other group.

3.6.1.2. Comparison of between-groups levels of Anxiety. Usinga Bonferroni adjusted alpha of p< 0.025, significant differencesbetween teachers by type were identified. Pre-service elementaryfemales were significantly more anxious than their experiencedmale and female counterparts. Pre-service elementary males weresignificantly more anxious than experienced elementary teachers.Experienced elementary females were significantly less anxiousthan pre-service secondary females, and elementary males. Expe-rienced elementary males were significantly less anxious than allother teaching groups except experienced females, and secondary

males, and pre-service secondary females were not significantlydifferent from their experienced counterparts but were signifi-cantly more anxious than experienced elementary teachers.Experienced secondary females and males did not differsignificantly.

As the cell sizes for experienced elementary males and experi-enced secondary females are less than 20, each should be inter-preted with caution. These may not be ‘‘robust’’ results (Tabachnick& Fidell, 2001, p. 251) and a replication with a larger, cohort issuggested.

4. Discussion

The data reported provided support for the postulate examinedby the first research question; that some people chose teachingunconsciously, desiring a corrective emotional experience. Theexperienced teachers in this study are more secure than theirinexperienced colleagues on both dimensions of attachment;anxiety about close relationships and avoidance levels. The secondquestion addressing the differing classroom experiences ofelementary and secondary school teachers revealed that elemen-tary teachers in this study are generally more secure in relation-ships than their secondary colleagues. While it is difficult tomeasure unconscious processes the findings from the first twoquestions lend support to the notion that some teachers may enterthe profession with an unconscious desire to receive a correctiveemotional experience from their job, and happily, with time in therole this appears to take place. This finding is interesting in light ofthe evidence suggesting that these effects are mediated somewhatby gender and age. It appears that young women may be morelikely to receive the benefit of their work experience than the othergroups. However, some small cell sizes suggest caution be exercisedand a replication with a larger sample is needed to confirm this.

The experience of parenting did not impact on teachers signif-icantly. Given that becoming a parent is a new attachment rela-tionship, with the inherent possibility of a corrective emotionalexperience for the parent, some may find this result surprising.However, the finding confirms many previous studies whichdemonstrated intergenerational transfer of attachment style(Bretherton, 1990; Fonagy, 1994; Sroufe, 2005). Therefore it seemslikely to be the securely attached children in the class who providethe corrective emotional experience to the teacher and not childrenper se. This has implications for the grouping of children intoclasses, as well as the impact of teachers and students on eachother. It seems sensible for schools to attempt to have a critical

P. Riley / Teaching and Teacher Education 25 (2009) 626–635634

mass of secure children in each class as it may be those childrenwho strongly contribute to ‘setting the tone’ for a well functioningclass group.

In addressing the question regarding the impact on attachmentstyle of teachers over time, the findings from this research showthat elementary teachers’ attachment profiles are significantlydifferent from secondary teachers and that they may change overtime once in the job. They appear to be both less anxious and lessavoidant of close relationships. Once they gain experience this gapbecomes larger for both males and females but significantly so formales. Elementary teachers may also be more advantaged in theirability to form significant attachments with students, providingopportunities for corrective emotional experiences as a result oftheir work. A plausible explanation for this lies in the nature ofwork teachers engage in. Elementary teachers spend most of eachworking day with approximately the same 25 students. Secondaryteachers are likely to have up to 200 students in groups of 25 forshort periods, a small number of times each week. Further,elementary teachers, due to the generalist nature of their work,deal with their students in many differing situations, according tothe curriculum areas they are responsible for. Secondary teacherson the other hand see their students usually for a very limitednumber of subjects; e.g. the average secondary teacher only seestheir students doing Mathematics and Science or English andHistory, whereas their elementary colleagues see their students forthose subjects, and most if not all the other curriculum areas. Thus,elementary teachers are therefore more likely to get to know theirstudents in ways that can foster a deeper emotional connectionwith them as a function of the time spent together, and the varietyof activities they engage in.

The data reported support the view that adult attachment maybe a more appropriate lens through which to view teacher–studentrelationship(s) than the traditional unidirectional model. This is animportant finding in light of: a) growing evidence regardingteacher aggression in the classroom; and, b) resistance by someteachers to professional development interventions aimed atcurbing aggression (Acker, Grant, & Henry, 1996; Riley, 2006a,2006b; Donohue, Perry, & Weinstein, 2003; Hyman & Snook, 1999;Lewis, 2006; Lewis, Romi, Qui, & Katz, 2005; Poenaru & Sava, 1998;Sava, 2002; Twemlow, Fonagy, & Sacco, 2002). Until recentlyaggressive teachers have been under-researched in the West. This issurprising given; a) the prevalence rates, and b) the significantnegative effect they have on their students. The phenomenon hasbeen more widely studied and reported in the non-English litera-ture (Sava, 2002). If, as hypothesised earlier, some teachers areseeking corrective emotional experiences through attachmentbonds with students, a number of aspects of their aggressivebehavior are theoretically explained. As suggested earlier an inse-curely attached teacher wishing students to provide a correctiveemotional experience through attachment to them, is vulnerable torejection by those students. The kind of behaviours reported in theliterature: yelling, sarcasm, humiliation of individuals, and pun-ishing the whole class for individual misdemeanours (Lewis andRiley, 2009; Lewis, Romi, Katz, & Qui, 2008) are typical protestbehaviours.

These results are potentially significant for teacher educatorsalso. The finding that experienced teacher’s anxiety is reduced bypositive classroom experiences suggests that pre-service trainingthat emphasises relationship building may have long-lasting posi-tive effects on teachers. It appears from these findings that for someteachers at least the possibility exists that they will receivecorrective emotional experiences through the relationships theyare able to form with their students. Training in relationshipbuilding may therefore hold much potential.

Alternate explanations for the data reported may stem fromrelationships teachers are able to form with their colleagues and/or

superordinates. Collegial relationships may provide the correctiveexperiences for teachers. These relationships are likely to be muchlonger lasting than teacher–student relationships and thereforeoffer the opportunity of gradually changing the internal workingmodel of teachers over an extended period of time. Researchinvestigating the combination of collegial and student–teacherrelationships, while complicated in design, offers the chance toexplore the complexity of the teacher’s experience. New ways ofanalysing dyadic information also offer possibilities for discoveringmeaningful connections between members of school communities(Gonzalez & Griffin, 1997; Kenny, Kashy, & Cook, 2006).

These findings are preliminary and exploratory and should beread with some caution. The sample size for experienced teachers isnot large enough to make any more substantial claim than thereappears to be a case for further exploration with a larger sample.

References

Acker, R. V., Grant, S. H., & Henry, D. (1996). Teacher and student behavior as a func-tion of risk for aggression. Education & Treatment of Children, 19(3), 316–334.

Ainsworth, M. D. S., Blehar, M. C., Waters, E., & Wall, S. (1978). Patterns of attach-ment: Assessed in the strange situation and at home. Hillsdale. NJ: LawrenceErlbaum.

Allen, J. G., Stein, H., Fonagy, P., Fultz, J., & Target, M. (2005). Rethinking adultattachment: a study of expert consensus. Bulletin of the Menninger Clinic, 69(1),59–80.

Bartholomew, K. (1994). Assessment of individual differences in adult attachment.Psychological Inquiry, 5(1), 23–27.

Bartholomew, K., & Horowitz, L. M. (1991). Attachment styles among young adults:test of a four-category model. Journal of Personality and Social Psychology, 61,226–244.

Bartholomew, K., & Shaver, P. R. (1998). Methods of assessing adult attachment: Dothey converge? In J. A. Simpson, & W. S. Rholes (Eds.), Attachment theory andclose relationships (pp. 46–76). New York: Guilford.

Bowlby, J. (1969). Attachment. London: Harper Collins.Bowlby, J. (1975). Separation: Anxiety and anger, Vol. 2. Harmondsworth: Penguin.Bowlby, J. (1980). Loss, sadness and depression, Vol. 3. New York: Basic Books.Bowlby, J. (1982) (2nd ed.). Attachment, Vol. 1 London: Harper Collins.Bowlby, J. (1988). A secure base: Clinical applications of attachment theory. London:

Routledge.Brennan, K. A., Clark, C. L., & Shaver, P. R. (1998). Self-report measures of adult

attachment: an integrative overview. In J. A. Simpson, & W. S. Rholes (Eds.),Attachment theory and close relationships (pp. 46–76). New York: Guilford Press.

Bretherton, I. (1990). Communication patterns, internal working models, and theintergenerational transmission of attachment relationships. Infant MentalHealth Journal, 11(3), 237–252.

Cassidy, J., & Shaver, P. (1999). Handbook of attachment: Theory, research, and clinicalapplications. New York: Guilford Press.

Collins, N. L., & Feeney, B. C. (2000). A safe heaven: an attachment theoryperspective on support-seeking and caregiving in adult romantic relationships.Journal of Personality & Social Psychology, 78, 1053–1073.

Donohue, K. M., Perry, K. E., & Weinstein, R. S. (2003). Teachers’ classroom practicesand children’s rejection by their peers. Journal of Applied DevelopmentalPsychology, 24(1), 91–118.

Fonagy, P. (1994). Mental representations from an intergenerational cognitivescience perspective. Infant Mental Health Journal, 15(1), 57–68.

Fraley, R. C., & Shaver, P. R. (2000). Adult romantic attachment: theoretical devel-opments, emerging controversies, and unanswered questions. Review of GeneralPsychology, 4(2), 132–154.

Fraley, R. C., & Spieker, S. J. (2003). What are the differences between dimensionaland categorical models of individual differences in attachment? Reply to Cas-sidy (2003), Cummings (2003), Sroufe (2003), and Waters and Beauchaine(2003). Developmental Psychology, 39(3), 423–429.

Fraley, R. C., Waller, N. G., & Brennan, K. A. (2000). An item response theory analysisof self-report measures of adult attachment. Journal of Personality and SocialPsychology, 78(2), 350–365.

George, C., Kaplan, N., & Main, N. (1984). Adult Attachment Interview. Berkeley:University of California.

Gonzalez, R., & Griffin, D. (1997). On the statistics of interdependence: treating dyadic datawith respect. In S. Duck (Ed.), Handbook of personal relationships: Theory, research andinterventions (2nd ed.). (pp. 271–302) Hoboken, NJ: John Wiley & Sons, Inc.

Hazan, C., & Shaver, P. R. (1987). Romantic love conceptualized as an attachmentprocess. Journal of Personality and Social Psychology, 52, 511–524.

Hofstee, W., Berge, J. T., & Hendriks, A. (1998). How to score questionnaires.Personality and Individual Differences, 25(5), 897–909.

Hyman, I. A., & Snook, P. A. (1999). Dangerous schools: What we can do about thephysical and emotional abuse of our children. San Francisco: Jossey Bass.

Kenny, D. A., Kashy, D. A., & Cook, W. L. (2006). Dyadic data analysis. New York: TheGuilford Press.

Kesner, J. E. (2000). Teacher characteristics and the quality of child–teacher rela-tionships. Journal of School Psychology, 38(2), 133–149.

P. Riley / Teaching and Teacher Education 25 (2009) 626–635 635

Lewis, R. (2006). Classroom discipline in Australia. In C. M. Evertson, &C. S. Weinstein (Eds.), Handbook of classroom management: Research, practiceand contemporary issues (pp. 1193–1214). New Jersey: Lawrence ErlbaumAssociates Inc.

Lewis, R., & Riley, P. (2009). Teacher misbehaviour. In L. J. Saha, & A. G. Dworkin(Eds.), The international handbook of research on teachers and teaching (pp. 417–431). Norwell, MA, USA: Springer.

Lewis, R., Romi, S., Katz, Y. J., & Qui, X. (2008). Students’ reaction to classroomdiscipline in Australia, Israel, and China. Teaching and Teacher Education, 24(3),715–724.

Lewis, R., Romi, S., Qui, X., & Katz, Y. J. (2005). Teachers’ classroom discipline andstudent misbehaviour in Australia, China and Israel. Teaching and TeacherEducation, 21, 729–741.

Main, M. (1990). A typology of human attachment organisation assessed withdiscourse, drawings and interviews. New York: Cambridge UniversityPress.

Mikulincer, M., & Goodman, G. S. (2006). Dynamics of romantic love: Attachment,caregiving, and sex. New York: Guilford press.

Pallant, J. (2007). SPSS survival manual: A step by step guide to data analysis usingSPSS (3rd ed.). Crows Nest, Australia: Allen & Unwin.

Pianta, R. C., Steinberg, M. S., & Rollins, K. B. (1995). The first two years of school:teacher–child relationships and deflections in children’s classroom adjustment.Development and Psychopathology, 7(2), 295–312.

Poenaru, R., & Sava, F. A. (1998). Teacher abuse in schools: Ethical, psychological andeducational aspects. Bucharest: Editura Danubius.

Rholes, W. S., & Simpson, J. A. (2004). Adult attachment: Theory, research, and clinicalimplications. New York: Guilford Publications.

Riley, P. (2006a, July 12–15). To stir with love: Imagination, attachment and teacherbehaviour. Paper presented at the 1st Research Symposium on Imagination inEducation, Vancouver, Canada.

Riley, P. (2006b, 26–30 September). When ‘Inner Working Models’ don’t work:Attachment styles, teacher aggression and research taboos. Paper presented at the2006 Joint Conference of the APS and the NZPsS: Psychology Bridging theTasman: Science, Culture and Practice, Auckland, New Zealand.

Sava, F. A. (2002). Causes and effects of teacher conflict-inducing attitudes towardspupils: a path analysis model. Teaching and Teacher Education, 18(2002),1007–1021.

Simpson, J. A., & Rholes, W. S. (1998). Attachment theory and close relationships. NewYork: Guilford Publications.

Sroufe, L. A. (2005). Attachment and development: a prospective, longitudinalstudy from birth to adulthood. Attachment & Human Development, 7(4),349–367.

Tabachnick, B. G., & Fidell, L. S. (2001). Using multivariate statistics (4th ed.).Needham Heights MA: Allyn & Bacon.

Twemlow, S. W., Fonagy, P., & Sacco, F. C. (2002). Feeling safe in school. Smith CollegeStudies in Social Work, 72(2), 303–326.

Wright, B., & Sherman, B. (1963). Who is the teacher? Theory Into Practice, 2(2),67–72.

Winnicott, D. W., Winnicott, C., Shepherd, R., & Davis, M. (1986). Home is where westart from: Essays by a psychoanalyst. Harmondsworth: Penguin.