Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Attachment style, affect and construal of interpersonal conflict Starzomski, Andrew J. 1993

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_1993_fall_starzomski_andrew.pdf [ 3.23MB ]
JSON: 831-1.0086305.json
JSON-LD: 831-1.0086305-ld.json
RDF/XML (Pretty): 831-1.0086305-rdf.xml
RDF/JSON: 831-1.0086305-rdf.json
Turtle: 831-1.0086305-turtle.txt
N-Triples: 831-1.0086305-rdf-ntriples.txt
Original Record: 831-1.0086305-source.json
Full Text

Full Text

ATTACHMENT STYLE, AFFECT AND CONSTRUAL OFINTERPERSONAL CONFLICTbyANDREW J. STARZOMSKIB.Sc., Saint Mary's UniversityA THESIS SUBMITTED IN PARTIAL FULFILLMENTOF THE REQUIREMENTS FOR THE DEGREE OFMASTERS OF ARTSinTHE FACULTY OF GRADUATE STUDIES(DEPARTMENT OF PSYCHOLOGY)We accept this thesis as conformingto th - required standardsTHE UNIVERSITY OF BRITISH COLUMBIASeptember 1993© Andrew J. Starzomski, 1993In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.(Signature) Department of ^tilvf) 169y The University of British Colu biaVancouver, Canada(2.f-t 3o M5DateDE-6 (2/88)AbstractThe relationships between attachment style, negativeaffect, and attribution were examined. Male undergraduates(n=72) were asked to report on their style in intimaterelationships, anger and attributions about negativebehavior by an actual girlfriend. Results indicated thatthose with an Avoidant attachment style tended to be moreangry in general temperament than Secure participants, andthat their anger was involved with the explanations adoptedto account for negative girlfriend behavior. Participantswith Anxious attachment resembled both the Secure andAvoidant groups in their trait anger and attributionprofiles. Experimental analysis of emotional and cognitivedifferences in response to audiotapes of three conflictsindicated that anger and anxiety were important in reactionsof Avoidant participants. Anxiety was notably absent as partof the Secure participants' reactions. The Anxious groupdemonstrated anger and anxiety responses to conflictconsistent with theoretical predictions. Anger and anxietyresponses were predicted from attachment style, trait anger,and attributions. Little evidence was noted for theassumption that attachment styles are activated primarily inthe context of intimate relationships.iiTABLE OF CONTENTSAbstract^ iiTable of Contents^ iiiList of Tables vList of Figures viAcknowledgements^ viiIntroduction 1Attachment Theory: Past and Present^ 3Experimental Attachment Research 4Attachment and Cognition^ 7Attachment, Conflict and Anger^  9Attachment and Abuse in Intimate Relationships^ 11The Present Research^ 11Hypotheses^ 15Method 18Participants^ 18Measures 20Stimuli 22Procedure^ 23Results^ 24Descriptive Attachment Style Differences^ 24A Note on the Presented Data^ 24Attachment and Lifestyle Anger 25Attachment and Attributions 26Attachment and Pre-experimental Mood^ 28Attachment Style and Negative Affective Responses^ 29Effect of Conflict Type on Mood 29Attachment Style, Type of Conflict and Mood^ 31Predicting Affective Reactions^ 32Discussion^ 34Attachment Style Differences 34Attachment and Anger^ 37Elicitation of Anger to Conflict^ 39Attachment and Development of Attributional Style^ 41iiiivAttachment and Attributional Style: Links to Anger ^ 44Implications for the Study of Aggression and Abuse ^ 45General Conclusions and Future Directions^ 48References^ 51Tables 57Figures 67VLIST OF TABLES1. Correlations Between Attachment and Anger Total^ 572. Means (and Standard Deviations) of AttachmentGroups on Anger Dimensions^  583. Correlations Between Anger Dimensions and Distress-Maintaining Attributions from Actual Relationships....594. Correlations Between Attachment Dimensions andDistress-Maintaining Attributions from ActualRelationships^ 605. Correlations Between Attachment Dimensions andComposite Anger States^ 616. Correlations Between Attachment Dimensions andComposite Anxiety Affect^ 627. Means (and Standard Deviations) on Composite Angerand Anxiety Scales Pre- and Post-tapes^ 638. Means (and Standard Deviations) of Attachment Groupson Anger and Anxiety Responses to Conflict Tapes^ 649. Prediction of Anger Responses to Conflict Tapes^ 6510. Prediction of Anxiety Responses to Conflict Tapes ^ 66List of Figuresvi1. Attachment Dimensions by Attachment Style^ 672. Anger Responses by Attachment Style^ 683. Anxiety Responses by Attachment Style 694. Composite Summary of Findings^ 70AcknowledgementsI extend thanks to my committee members, Darrin Lehmanand Bob Hare, for their time and interest in this project. Iam of course especially grateful to my advisor, Don Dutton,for his guidance and insight at all stages of this research.I was fortunate to have friends and associates who werevery accommodating and helpful from the study's beginning toend. Larry Axelrod was always willing to discuss the projectand was a pillar of encouragement. The composed andthoughtful suggestions of Adam DiPaula, in addition to hisgraphic wisdom, were particularly helpful as the final drafttook shape. Steve Heine's example and enthusiasm were alsoappreciated. Barbara McGregor's literate assistance andinterest was always encouraging. The moral support of RussPitts, John and Linda McNally, Monica Salvatella, and myfamily served me faithfully at critical points in my work,and for their backing I am truly thankful.viiAttachment and Conflict1Most people would agree that intimate relationships,such as those between mothers and their children, andbetween wives and husbands, encompass some of life's mostpositive experiences. But in addition to providing a contextfor joy and fulfillment, intimacy can also spark suchemotions as anger and anxiety. This thesis is an explorationof the ways in which people experience negative affect inclose relationships. Its theoretical basis is attachmenttheory, a popular theory of close relationships. Attachmenttheory seeks to describe the patterns of behavior, emotions,and needs observed in close relationships. Recent findingshave identified three relationship styles: Secure, Avoidantand Anxious. These styles vary in such ways as one's comfortwith closeness, need to have an attachment figure present,and anxiety about relationship dissolution (Hazan andShaver, 1987; Simpson, Rholes and Nelligan, 1992). Thepresent research was intended to explore the possibilitythat attachment style could influence perceptions of, andemotional reactions to, interpersonal conflict, as suggestedin recent studies by Dutton, Saunders and Bartholomew(1992), and Dutton, Starzomski and Bartholomew (1992).An important element of attachment theory is that thefeatures differentiating each style emerge primarily underconditions of perceived anxiety (Simpson, Rholes, andNelligan, 1992). These identifying features include attemptsto increase or decrease emotional distance in relation toone's partner, as well emotional expressions. Interpersonalconflicts can be anxious situations where attachment stylesare activated, and manifested in emotional states, thoughtsand behavior. Recent studies describing how anger isassociated with attachment styles (Bowlby, 1973; Dutton,Starzomski and Bartholomew, 1992) lead to the theory thatattachment style may influence the manner in whichindividuals perceive conflict.Attachment and Conflict2On the basis of attachment research, individuals withinsecure attachment styles should respond to intimateconflict differently from those more secure in theirattachment style. For example, people insecure in theirrelationship style should feel greater anger and anxietythan secure individuals, as well as exhibit distress-maintaining beliefs about the role of the intimate other(e.g., she never thinks of how I feel). Those with a secureattachment style should experience less anger and anxiety inthese situations, as well as a tendency to explain negativebehavior by the intimate other in a way which minimizesinterpersonal distress (e.g., it was an isolated event).It appears that attachment styles are differentiallyactivated according to the nature of the conflict with anintimate. Collins (1992) has shown that attachment style ispredictive of emotional responses to stress related tointimacy concerns (i.e., your partner failing to comfort youwhen you feel depressed), but not to non-intimacy matters(i.e., partner fails to pay back borrowed money). However,as of yet there exists little in the way of an empirically-backed understanding of the nexus between attachment stylesand the interpersonal context of conflict (i.e., husband-wife versus co-worker conflict). It may be the case thatinterpersonal context may differentially trigger attachmentstyles and hence individuals' construal of, and reaction to,conflict. Howe (1987) suggested that people form differentattributions for conflicts between intimates and strangers.Recent work linking attachment, cognition, and affect(Collins, 1992; Kobak and Sceery, 1988) has suggested thatattributional and affective responses to conflict could varywith both attachment style and the interpersonal context ofthe conflict.Attachment and Conflict3Attachment Theory: Past and PresentAttachment has occupied a place of special importancein the heritage of psychology. It has also become thesubject of renewed interest based on recent theoretical andempirical advances. Acknowledgement and awareness of theimportance of parent/child attachment dates back to the workof Freud (1926), Sullivan (1953) and Erikson (1963). Eachtheorist recognized the influence of early attachment on thedevelopment of behavior and personality. The comprehensivetheoretical contribution of Bowlby (1969; 1973; 1980) hasserved as the basis for the resurgence of the study of humanattachment.John Bowlby's attachment theory is aimed at explainingthe variety of infant responses to affection, separation,and loss, in addition to suggesting consequences of(dys)functional responses to these relationship events. Histheory integrated key scientific advances of the past 150years, such as evolutionary biology, ethology,psychoanalysis, behaviorism, and cognitive science. AlthoughBowlby focussed on infancy and childhood events, recentresearch has applied his theory's principles to adultbehavior.One of Bowlby's primary contributions was connectingthe development of attachment with evolutionary success bydrawing on the ethological research of Harry Harlow. Harlowand Harlow's (1971) study of attachment in monkeys providedevidence for the survival factor associated with attachment.The hypothesis that attachment is caused by drive reduction(i.e., meeting needs for food) was undermined by Harlow, whoalso demonstrated that emotional security is necessary, butnot sufficient, for healthy development.Bowlby conceived of attachment as a process ofemotional development grounded in the affection, separation,and loss experiences of an infant's first relationships.Attachment and Conflict4Learning adaptive emotional responses to anxiety in intimaterelationships, as well as managing maladaptive ones, areimportant parts of early infant relationships with parents.A crucial component of attachment is the development of theinfant's perception of the caregiver as a secure base, asource of emotional and physical comfort and support duringdistress (Hazan and Shaver, in press). The development ofthe secure base was studied by Bowlby during hisinvestigation of intimate separations in infancy.Through his observations of infant behavior Bowlbyidentified a sequential response to separation comprised ofprotest, despair and detachment. Protest involves searchingfor the missing attachment figure while fighting off thesoothing attempts of others. Despair follows protest, and isa state of passive sadness in the absence of the caregiver.Detachment is the carry-over of negative affect uponreunion, seen as avoidance and disregard for the attachmentfigure.Managing these emotional responses to separation hassurvival value for the infant. According to Bowlby, infantswho respond emotionally to separation in adaptive ways arebetter able to ensure their needs are met effectively bycaretakers. Infants who cannot readily terminate thedetachment phase upon reunion may risk driving the parentaway. The ways in which people deal with separation, loss,and reunion form the basis for discriminating amongdifferent attachment styles. Studies have shown that theinsecure attachment styles are associated with responses toseparation which chronically maintain emotional distance,rather than bridge it.Experimental Attachment ResearchAs patterns of relationship needs which form the basisfor differences between attachment styles are most clearlyobserved under conditions of anxiety and uncertainty inAttachment and Conflict5intimate interpersonal relationships, experimentalsituations which elicit anxiety have been the methodology ofchoice in empirical studies of attachment. Such research hasresulted in a parsimonious classification system ofattachment of styles based on temperament and conductobserved in anxious situations.Studies of infants in the "strange situation" have beenthe precedent in experimental discriminations betweenattachment groups (Ainsworth, Blehar, and Waters, 1978). Thestrange situation is a laboratory scenario involving theinfant's separation from, and subsequent reunion with,his/her mother or father. Three patterns of infant responsesto this situation have been systematically described byAinsworth, Blehar and Waters (1978), and have also beendemonstrated in adult versions of the same scenario(Simpson, Rholes, and Nelligan, 1992).Secure attachment is characterized by welcoming theattachment figure upon return from separation, and seekingproximity to the attachment figure in the event of adistressing situation. Security is the capacity to disengagethe turbulent protest and despair emotions upon reunion withthe mother. The detachment response to reunion, which is thedefining feature of the next attachment style, is relativelyabsent in the secure attachment style. About 55% of adultswho have been studied fall into this category (Hazan andShaver, 1987; Feeney and Noller, 1990).Avoidant attachment is identified by the absence ofapproach behavior upon return of the attachment figure. Interms of Bowlby's account of attachment behavior as displaysof affect, these people have a tendency to detach fromcaregiving situations. Upon reunion, it is characteristicfor those with this attachment style to reject thecaregivers' attempts to provide comfort. Those with avoidantattachment styles tend to have had repeated experiences inAttachment and Conflict6which efforts to establish contact with attachment figureshave met with rejection. As a result, people with thisattachment pattern associate the need for proximity withfrustration and unmet needs (Simpson, Rholes and Nelligan,1992). A key consequence of this frustration over unmetneeds is hostility and anger directed towards intimates.About 25-30% of adult research samples have avoidantattachment styles (Hazan and Shaver, 1987; Feeney andNoller, 1990).Anxious-ambivalent attachment emerges from childhoodsunderscored by inconsistently or unpredictably successfulcontact attempts (Crittendon and Ainsworth, 1989). Theprotest element of Bowlby's affect sequence, and therelentless pursuit of attachment figures, are integralfeatures of this relationship pattern. Individuals with thisstyle yearning for substantial emotional support from theirromantic partners. Bartholomew and Horowitz (1991) suggestedthat people with this style are preoccupied with attainingself-worth by gaining the acceptance of valued others. Ithas also been reported that there are overtones ofresentment and hostility directed towards intimates as partof this attachment style due to the perceived inability ofthose significant others to meet emotional needs (Simpson,Rholes and Nelligan, 1992). The incidence of thisrelationship style has ranged from 15-20% of adult samples(Hazan and Shaver, 1987; Feeney and Noller, 1990).Refinement of this tripartite model of styles wasrecently proposed by Kim Bartholomew. Bartholomew (1990)reconstructed the three-dimensional model of attachment(secure, avoidant, anxious) in order to accommodate thetheoretical importance of the avoidant style. She developeda four-dimensional scheme that catalogues attachmentpatterns as illustrative of positive or negativerepresentations of self and other. Retaining the secure andAttachment and Conflict7anxious attachment categories of earlier work, Bartholomew'scontribution is of particular relevance to the presentresearch because of its further delineation of the avoidantstyle.Bartholomew broke the avoidant style into twodimensions: fearful and dismissing. The fearful attachmentpattern resembles a hybrid of anxiety and avoidance -- asense of the relationship being needed to meet personalshortcomings, but apprehension that the intimate other willfail to meet these needs. The dismissing dimension developswhen attachment systems are chronically deactivated -- thestereotype of "compulsive self-reliance" applies here.The bulk of the research on differences betweenattachment styles has been directed at how attachmentinfluences relationship satisfaction. Men with more secureattachment styles tend to be less emotionally abusive totheir spouses than insecure men (Dutton, Saunders andBartholomew, 1992; Dutton, Starzomski and Bartholomew,1992). It has been noted that Secure individuals are higherin personal and interpersonal self esteem, and that peopleclassified as Avoidant are less likely to have experiencedlove as deeply as those with Secure or Anxious styles(Feeney and Noller, 1990). It has also been shown that thosewith Secure styles of attachment seek out more support fromtheir partners as situations become progressively moreanxious, and that Secure partners offer more support totheir distraught intimates (Simpson, Rholes and Nelligan,1992).Attachment and CognitionBowlby's theory is a developmental one emphasizingstructural psychological change with maturity. Bowlby callsthis developing mental structure a working model. Theworking model can be conceptualized as the psychologicalappraisal of need-meeting relationships. Infants construeAttachment and Conflict8attachment figures (parents) as resources for meeting theirneeds. The working model governs our expectations andinterpersonal perceptions as we come to understand those whocare for us. In effect, it is through these working modelsthat we come to understand our place in the social world.The working model is comprised of information about whoone's attachment figures are, their accessibility, and thelikelihood that they will respond to our emotional needs insupportive ways. Our sense of worth in the both our own eyesand the eyes of our attachment figures is also part of theworking model. This cognitive component of attachment styleis important because it governs our expectations aboutcloseness. Explanations about the motivations of others inclose relationships are also part of this cognitiveframework. In short, the cognitive component of attachmentstyles influence what we attend to in our relationships withothers and how we think about them.An important part of current attachment researchinvolves identifying the attributional tendencies of eachattachment group. For example, the working models of theAvoidant and Anxious styles tend to dwell more on thenegative impacts of relationship events than does the Secureworking model (Collins, 1992). Threads of this cognitivepattern of dealing with aversive events have been picked upby researchers studying marital discord. For example,Fincham and Bradbury (1992) reported strong positivecorrelations between anger with one's romantic partner andattribution. Partners were more angry with their partnerwhen they attributed cause for a negative event to theirpartner (i.e.: criticism, disinterest).Essentially, it has been noted that couples in conflicttend to think about negative relationship events in waysthat maintain or even exaggerate the discord. Attachmenttheory may proved insight into why some people tend toAttachment and Conflict9explain negative events in these ways. The present researchknits together these different courses of research onconflict in intimate relationships.Attachment, Conflict and AngerThroughout the preceding sections it has been notedthat attachment styles explain the different ways ofbehaving in, and thinking and feeling about, intimaterelationships. It is important to delineate how attachmentcan come to bear on the experience of interpersonalconflict. Much has been written on the ways in which angerand attachment are related.Work on attachment has suggested that anger is of primeimportance in the way some people develop and maintain closerelationships with others (Bowlby, 1973). It is clear inBowlby's (1973) theory that anger is part of the experienceof being in most intimate relationships. Importantly, angercan serve both beneficial and harmful ends. Displaying angerto a child who runs carelessly into the street, or to anadulterous intimate partner, can be meant to serve thedesirable ends of education and deterring additionaldisloyalty, respectively. Expressing anger can convey to theintimate other the value one places on the relationship. Itcan also plague a relationship and ultimately destroy itthrough power abuse and/or estrangement. Anger can become apersistent part of the interpersonal relationship,potentially culminating in aggressive thoughts and actions.One can readily see how a child who is initially exposed toturbulent relationships consisting of noncontingent,persistent and aggressive outbursts of anger could developan insecure working model. For example, a child may developthe belief that "caring" relationships work from afoundation of aggressive anger and hostility. Support forthis idea was reported by Cummings, Zahn-Waxler, and Radke-Yarrow (1981), who found that children frequently exposed toAttachment and Conflict10harsh anger between parents displayed more emotionaldistress than children who weren't privy to such outbursts.Main and Weston (1982) suggested that Avoidantindividuals feel angry with attachment figures, but fearthat expressing their anger will result in decreasedaffection and withdrawal by the caregiver. As a result,anger is suppressed and replaced with cool, detachedavoidance. Chronically rejected people experienceparticularly strong angry impulses with even strongeravoidance of displaying that anger. These people are not aslikely to vent their anger during moments of acute arousal,choosing instead to express anger in circumstances that donot risk decreased proximity from the attachment figure(Bowlby, 1984).Bartholomew (1990) argued that a strong andunresolvable approach-avoidance dynamic may underlie thebehavior of chronically fearful people: perceived threats ofabandonment lead to tendencies to approach an attachmentfigure who rejects contact, subsequently generatingwithdrawal and an even stronger need for attachment. A self-perpetuating feedback loop ensues that leads to chronicavoidance, frustration of attachment needs, anddysfunctional experiences of anger.The conceptual associations between anger andattachment as outlined by Bowlby (1973; 1984) andBartholomew (1990) have been supported empirically. Kobakand Sceery (1988) reported that avoidant individuals hadhigher levels of hostility (measured through both peerratings and interviews) than either securely or anxiouslyattached people. Bartholomew and Horowitz (1991) reportedthat Avoidant attachment was congruent with a cold andhostile interpersonal style. Shaver and Brennan (1992) notedthat hostile components of the "Big Five" personalitydimensions were positively correlated with anxious andAttachment and Conflict1 1avoidant attachment, and negatively associated with secureattachment. Additional support for the link betweenattachment and anger was recently provided by Mikulincer,Florian and Weller (1993), who found that hostility was astrong emotional reaction to Iraqi missile attacks amongAnxious and Avoidant Israeli individuals.Attachment and Abuse in Intimate RelationshipsThe possible link between adult attachment, anger, andaggression has been suggested by studies of thepsychological profiles of men convicted of wife assault(Dutton, Saunders & Bartholomew 1992; Dutton, Starzomski, &Bartholomew, 1992). Relationships between Bartholomew'sattachment dimensions and anger, personality disturbance,and violence led Dutton, Saunders and Bartholomew (1992) topropose the existence of a psychological structure, dubbedthe anger/anxiety template, which makes the very experienceof intimacy generate anger for men with an avoidantattachment style. This anger is then projected onto theirfemale partner and serves as a basis for interpersonal rageand aggression directed toward her.Dutton and Starzomski (1992) reported that women'sreports of emotional abuse and dominance by their physicallyabusive partners were significantly and positivelycorrelated with their partners' Avoidant attachment style.On the other hand, assaultive men with secure attachmentstyles were much less likely to have inflicted high levelsof psychological abuse on their partners, suggesting thatthe nature of abuse in close relationships may differ inaccordance with one's attachment style.The Present Research:Attachment, Emotion and Attribution.Research combining developmental psychology, socialpsychology, and psychiatry has resulted in a new theoreticalorientation toward the possible processes underlying someAttachment and Conflict12patterns of male aggression in intimate relationships.Specifically, links between early childhood trauma,attachment styles, adult personality disturbance andintimate violence have been proposed by Dutton and hiscolleagues (Dutton, 1988; Dutton, in press; Dutton,Saunders, and Barthomew, 1992; Dutton, Starzomski andBartholomew, 1992). The preliminary empirical support forthese associations is based almost exclusively oncorrelational analyses. The present study has endeavoured toapply experimental methods to further explore thesetheoretical potentials.The techniques used in this thesis were based on a longline of similar experimental paradigms. Researchersexamining how people interpret and react to conflict haveadopted methods aimed at increasing participants' emotionalinvolvement by various conflict simulations (Bandura, Ross,and Ross, 1961; Browning, 1983, Davis et al, 1987; Dutton,Webb and Ryan, 1992; Harvey et al, 1980; Strachan andDutton, 1993). Audiotapes of interpersonal scenarios haveproven to be particularly effective as a way to presentrealistic stimuli (Strachan and Dutton, 1992). It has beenfound, for example, that listening to audiotapes ofinterpersonal conflict, accompanied by appropriateinstructions, can result in changes in mood (Dutton, Webband Ryan, 1992; Strachan and Dutton, 1992).This study was designed to examine how males respondcognitively and emotionally to conflict and if thesereactions are related to attachment styles. Varying thenature of the relationship between individuals in conflict(romantic partners versus co-workers) allowed for theinvestigation of how situational factors could affect theelicitation of attachment styles in interpersonal disputes.Specifically, six research questions were explored.Attachment and Conflict13First, how does one's everyday experience of angerrelate to attachment styles and dimensions? This questionconsidered the possibility that people who vary from oneanother in attachment patterns could experience differentintensities, expressions and frequencies of anger as part oftheir lifestyles. Second, how do attributions of the causeof, and responsibility for, an intimate other's negativebehavior relate to experiences anger? Here the, issue was todetermine if one's proclivity for anger was associated withpatterns of perception. Third, do the theoretical andempirical distinctions of different working models translateinto differences in attributions about behavior of one'sintimate other? In other words, are there explanations fornegative partner behaviors that people with certainattachment styles use but others do not? Fourth, doattachment dimensions affect how people handle uncertaintyin situations unrelated to interpersonal conflict? Thisquestion was addressed by looking at mood states beforeconflict vignettes were heard. Fifth, are various attachmentstyles and dimensions activated by experimental conflictvignettes which are situations of acute interpersonalduress? As a second part to this question, it was also ofinterest to determine if attachment styles weredifferentially activated according to interpersonal context(whether a dating couple or co-workers are in dispute).Sixth, is it possible to predict affective responses toconflict using measures of attachment, lifestyle anger, andattribution?A brief discussion of how this study links with, andexpands upon, previous attachment and conflict work iswarranted. This thesis is unique among other studies ofattachment, emotion, cognition and conflict for manyreasons. The questions asked and the methodologicalproperties set this work apart are worthy of consideration.Attachment and Conflict14One of the most important properties of the presentresearch was its attempt to see if attachment styles wereinvolved in events outside intimate relationships. Bylooking at how attachment styles related to pre-tape affect,in addition to conflict between non-intimates, this studyaddressed previous claims of attachment specificity withinintimate relationships. Previous studies have implied thatit made little sense to consider possible attachmentinfluences in non-intimate situations. Signs of attachmentoutside the domain of intimate relationships has begun toappear in studies of peer-ratings of personality andtemperament (Bartholomew and Horowitz, 1991; Mikulincer,Florian, and Weller, 1993). These studies have suggestedthat the influences of attachment may transcend the intimatedyad. The present study asked directly if the influence ofattachment styles extends beyond the intimate situation.The present research also attempts to articulate theconnections between attachment and anger. Previousexperimental work on attachment with adult samples hasprimarily addressed relationship quality and support intimes of distress (Feeney and Noller, 1990; Simpson, Rholesand Nelligan, 1992). As a result, findings of how anger'srelationship to attachment in these studies have been nestedin more general discussions of negative affect in theinsecure avoidant or anxious styles. The present study isunique in its primary concern with the delineation of angryresponses to conflict as part of attachment styles.The present study tries for a more systematic analysisof the attribution/anger link in close relationships thanprevious research. By looking at the connection betweenparticular anger dimensions (i.e.: hostile outlook, angermagnitude) and facets of attribution described byHoltzworth-Monroe and Jacobson (1985), including blame,cause, stability, and responsibility, greater awareness ofAttachment and Conflict15the cognitive-emotional experience of intimacy isanticipated.This thesis is also unique in its choice of sample.Being investigated here is the process of male violence inintimate relationships. In keeping with this focus, thedecision was made to include only male participants in thisstudy. Previous studies on the links between attachment,anger and aggression have been based on studies with menconvicted of wife assault. By using a non-clinical sample ofuniversity undergrads this thesis has begun to provide somenormative information about continuities and discontinuitiesbetween samples.Finally these studies of attachment, anger and abusehave relied on correlational methods, as opposed toexperimental manipulations in controlled settings. Theexperimental control exercised in this study has attemptedto provide firmer support for relationships uncovered inprevious correlational work (Dutton, Saunders andBartholomew, 1992).HypothesesFive hypotheses were generated upon review of theliterature on how attachment, anger and affect relate toconflict. First, it was predicted that self-reports oflifestyle anger would be a less central emotion in the livesof Secure individuals than Avoidant individuals. It wasanticipated that Avoidant attachment would be associatedwith a tendency to experience anger more often, moreintensely, and for longer periods of time than for Secureindividuals. The Anxious style was also expected to have ananger component similar in nature to the Avoidant pattern,but not as strong.Second, it was predicted that attributions of cause andresponsibility for negative behaviors by one's current orpast intimate partner would be related to attachment style.Attachment and Conflict16The pattern of attributional outcomes was predicted to besuch that those with secure attachment would be less likelythan those with Avoidant styles to explain negative eventsin distressful ways. Put somewhat differently, it wasanticipated that the tendency among Avoidantly attached mento see one's partner as responsible for, and causing,negative relationship events would be more common than forSecurely attached individuals. It was foreseen that theAnxious style could also be associated with perceptions ofthe other as responsible for negative events. This was inkeeping with the tendency for Anxiously attached people tosee the other as unable to fulfill their intimate needs.The third hypothesis concerned pre-experimental affectand attachment. It was predicted that attachment would notbe related to participants' emotional patterns before thetapes were heard. One would not expect that emotional statesin the experimentally constant setting, very different froma situation of inter-partner distress, to be associated withattachment style.Fourth, it was proposed that reactions to audiotapes ofintimate conflict would differ between attachment styles.Generally, it was projected that less negative affect wouldbe present in participants with Secure attachment than thosewith Anxious or Avoidant styles after hearing the tapes. Itwas also anticipated that patterns of attribution would beelicited by the tapes, such that explanations maximizing theimpact of negative events would be associated with negativeaffect. These between-style differences were predicted inkeeping with the theoretical underpinnings of theanger/anxiety template (Dutton, Saunders, and Bartholomew,1992), observations of stylistic differences ininterpersonal behavior (Bartholomew and Horowitz, 1991), anddisplays of behavior in response to anxious situations(Simpson, Rholes and Nelligan, 1992).Attachment and Conflict17The specific nature of the predicted reactions byparticipants with insecure attachment styles to the tapesmerits attention. Because those with marked Avoidantattachment are apprehensive about closeness inrelationships, conflict involving a female partner's desireto increase intimacy (Engulfment) could be powerful ineliciting negative affect. Additionally, depicting a partnerdesiring a decrease in closeness (Abandonment) couldactivate negative affect by mapping onto working modelsemphasizing the unreliability of attachment figures toremain close. On the basis of previous studies (Shaver andBrennan, 1992; Bartholomew and Horowitz, 1991), it wasanticipated that anger would be an especially notable partof emotional responses to conflict for Avoidantparticipants.Strong affective responses to conflict over intimacywere anticipated for those with an Anxious attachment style.It was predicted that Anxious participants would experienceparticularly strong reactions to the Abandonment scenario.Intimate conflict caused by a partner's pursuit of anunwanted deeper level of closeness was not expected to beparticularly stressful for those with an Anxious style --the possibility of becoming more immersed in a relationshipis probably seen as desirable by this group.The fifth and final hypothesis was again aimed attesting the specificity of attachment to intimate conflict.Reactions to co-worker conflict was contrasted withreactions to the intimate disputes in order to see ifattachment was uniquely implicated in the intimate domain.As with the earlier hypothesis anticipating no pre-experimental affective differences between attachmentstyles, it was postulated that the attachment differences inreactions to non-intimate conflict would be minimal ornonexistent.Attachment and Conflict18Data collected in the present study were analyzed interms of the four-part typology of Bartholomew and Horowitz(1991), as well as the tripartite model of Hazan and Shaver(1987). In effect, analyses involved looking atBartholomew's four dimensions (Assured, Fearful, Dismissing,Preoccupied) as a within-subjects factor and Ainsworth'sattachment groups (Secure, Avoidant, Anxious) as a between-subjects factor.MethodParticipants Seventy-two male students (mean age = 20.82 years) fromPsychology 100 classes at the University of British Columbiaparticipated in the study. Participants received 1.5 creditstoward their final grade as remuneration for theirparticipation in the study.The relationship history of this sample is ofparticular relevance to the results of the study. Twenty-sixmen (36.6%) in the sample were currently in relationships,but sixty-six (93%) of the participants reported having beenin at least one relationship (Mode = 2, Modal length = 2months). On a Likert scale from 1 to 7, with 1 representingdissatisfaction and 7 indicating level of satisfaction withthese relationships, the average score was 4.2.One of the initial tasks of this research was to usethe attachment dimensions to classify participants intoattachment categories. A cluster analysis using Ward'smethod used participants' profiles on Bartholomew's fourattachment dimensions (Assured, Fearful, Preoccupied,Dismissing) to determine the attachment category (Secure,Avoidant, Anxious) in which they were placed. The choice ofclustering method was based on previous research with thismeasure (Collins, 1992; Collins and Read, 1990).Attachment and Conflict19The sample was separated into three attachmentcategories which corresponded to Ainsworth's types discussedpreviously. In addition, this tripartite typology explainedthe present data more parsimoniously than a four-categorymodel. Figure 1 shows that as a result of the clusteranalysis 32% of participants (n = 22) were classified asSecure, 44% of the sample (n = 31) were clustered into theAvoidant group, and 24% (n = 17) were grouped together inthe Anxious group. Possible explanations for the fact thatthe Secure group comprised only 32% of the sample as opposedto the 50-60% expected, and that the Avoidant group wasabout 15% higher than expected can be offered. It may be thecase that the low internal consistency of the sub-scales(see below) contributed to this. It may also be an accuratereflection of the sample itself, as the sample was self-selected (i.e., they decided to participate for credit inthe final month of classes). Nonetheless, the differentprofiles for attachment groups on attachment dimensions areconsistent with theory and research.All between-groups differences on attachment dimensionswere determined using one-way ANOVA's with Tukey contrasts.The three attachment groups differed on all attachmentdimensions, with the exception being that the Secure andAnxious groups were not significantly different in theirAssured scores. Although high on the Assured dimension, theAnxious group was also the highest of the three groups onthe Preoccupied scale, F (2,69) = 25.94, p < .001. TheAnxious group also had low scores on the Dismissing scale,indicating that relationships, as opposed to self-reliance,were priorities for this group, F (2,69) = 10.52, p < .001.The fact that the Avoidant group had the highest scores onthe Dismissing dimension is consistent with this group beingwary of intimate relationships. One could consider theSecure group's high score on the Dismissing dimension toAttachment and Conflict20reflect a level of contentment in being without arelationship.For the purposes of clarity in the present workBartholomew's secure dimension will be called "Assured" toavoid confusion with the Secure attachment groupclassification. Within-subject correlations dealing withattachment report on the four Bartholomew sub-scalesoperating simultaneously at any given time (Assured,Fearful, Dismissing, Preoccupied), regardless of theindividuals' attachment group (Secure, Anxious, Avoidant) asdetermined by cluster analysis.Measures Participants in the study completed a number ofpsychometric scales. Measures of attachment style,attributions in past relationships, attributional responseto conflict, anger style, and affect were completed duringthe study. A brief description of these measures ispresented below.1.) Attachment: The Relationship Style Questionnaire (RSQ)Self-report measures of attachment style have beendeveloped in recent years by Hazan and Shaver (1987) andCollins and Read (1990). Thirty items from each of thesetests were combined by Bartholomew and Horowitz (1991) toidentify the four attachment dimensions which have emergedfrom theoretical and empirical work on attachment: Assured,Fearful, Dismissing and Preoccupied. Scoring of the RSQyields scores on each of the four attachment factors foreach participant. The Cronbach's alpha levels on these sub-scales for the present sample were: Assured (alpha = .40),Fearful (alpha = .45), Preoccupied (alpha = .38), andDismissing (alpha = .59).2.) Anger: Multidimensional Anger Inventory (MAI)The Multidimensional Anger Inventory (Seigel, 1986) isa 38-item self-report assessing the behaviors and thoughtsAttachment and Conflict21which comprise the experience of anger. The six subscalesand Cronbach's alpha levels of the MAI are: anger frequency(alpha = .87), hostile outlook (alpha = .66), self-directedor suppressed anger (alpha = .49), anger magnitude (alpha =.64), outwardly-directed or expression of anger (alpha =.17), and anger duration (alpha = .45).3.) Attributions: The Relationship Attribution Measure (RAM)The RAM (Fincham and Bradbury, 1992) asks participantsto report on causality and responsibility for negativebehaviors by one's intimate other (i.e.: criticism,inattention, aloofness, intolerance, compliments).Correlations of subscales with anger and maritalsatisfaction suggest that there is some discriminating powerin the RAM's capacity to identify angry and dissatisfiedcouples. Cronbach's alpha's for the RAM subscales wereconsistently at about the .80 level and greater.4.) Emotional State: Affect Adjective Checklist (AACL)The Affect Adjective Checklist (AACL; Russell andMehrabian, 1974) is a 16-item self-report of one's emotionalstate in the moment. This scale consists of bipolar affectadjective pairs (i.e., sad/not sad). Higher scores on theAACL indicate more intense emotional states. This measurewas used to gauge affective states elicited by the conflictaudiotapes. These composite scales were comprised of itemswhich clustered together in factor analysis of AACLresponses. Factor analysis revealed clusters of anger andanxiety items on the AACL. Depending on the type ofconflict, the Composite Anger cluster accounted for 29-40%of the variance, and the Composite Anxiety scale accountedfor 11-14% of the variance in AACL scores. The CompositeAnger measure was comprised of the following items: tense,angry, aggressive, hostile, irritated, and annoyed.Cronbach's alpha for these composite scales ranged from .70to .93.Attachment and Conflict225.) Attributions about Conflict ScenariosA series of questions intending to ascertainparticipants' construal of the audiotapes of conflict wereassembled based on the work of Holtzworth-Monroe andJacobson (1985) and Fincham and Bradbury (1992). Items onthis questionnaire sought out how participants explained themotivations and culpabilities of the conflicting parties.6.) Relationship Satisfaction: Dyadic Adjustment Scale (DAS)The administration of 32-item Dyadic Adjustment Scale(DAS; Spanier, 1976) is a measure of contentment andsatisfaction with one's romantic relationship. This measureserved as a manipulation check between attachment groups, aswell as predictor of emotional response to conflict.Stimuli. Audiotapes of 3-4 minute vignettes of conflict werecreated by theater students under the guidance of theauthor. Conflicts were geared to the age group of theparticipants involved in the study, with the actors takingthe roles of either university students in dating or workingrelationships. Three conflict situations, all involving amale and female, were generated to test the hypotheses inquestion; Abandonment and Engulfment scenarios (both withdating couples) and co-worker conflicts were simulated.The Abandonment conflict involved the female actorasserting a desire to reclaim some of her lost autonomy andbe less involved in the relationship. The Engulfment tapeconsisted of the female actor becoming angry that the malewas not willing to move to a deeper relationship of greatercloseness. Finally, the co-worker tape concerned thedifficulties in working together on a university courseproject. The actors were asked to make the conflicts heatedand emotionally intense by raising their voices and becomingverbally abusive where appropriate.Attachment and Conflict23As a precautionary measure, the tapes were rated onaffective intensity by a pilot sample. Vignettes were judgedto be similar in their affective intensity, and hence thisstimulus-bound variable was not a confound in theinterpretation of effects.Procedure Participants were initially asked to take a package ofmeasures home with them which consisted of the RSQ, MAI, RAMand others not used in the present study. Participants thenreturned within a few days to do the experimental part ofthe study which involved the conflict audiotapes. To enhancethe attention participants paid to audiotape events theywere asked to try to empathize with the male actor in eachvignette, to try to imagine how they would be feeling ifthey were in that situation. Harvey et al (1980) reportedachieving greater participant involvement using thisapproach, in the form of more attributional statements aboutconflict vignettes.All experimentation was done individually. Participantscompleted a measure of emotional state (AACL) before thetapes began. Three conflicts were then heard with headphoneson a portable stereo, one at a time, with participantscompleting measures of emotional state and attribution aftereach vignette. Total time for this audiotape portion of thestudy was about 30 minutes. The order in which participantsheard the conflicts was counterbalanced to ensure thatdifferences due to sequence of conflict presentation wouldnot emerge.The study looked at differences both within- andbetween-subjects. The within-subject variables were theaffective and attributional responses to the threeaudiotapes. The AACL and attributional measures were thedependent measures for these within-subject variables. Thebetween-groups variable of interest was attachment style.Attachment and Conflict24The RSQ was the measure used to establish the attachmentgroups, and the RAM, MAI, and AACL were compared at thebetween-groups level.ResultsDescriptive Attachment Group DifferencesIt was earlier noted that attachment groups (Secure,Avoidant, Anxious) varied in theoretically predicted ways intheir attachment dimension (Assured, Fearful, Dismissing,Preoccupied) profiles. Further to the validity ofrelationship differences between attachment groups, theyalso varied in their satisfaction with past relationships.One-way ANOVA's with Tukey contrasts revealed differencesbetween attachment groups in their level of satisfaction inclose relationships. The Secure group was more satisfiedwith their past romantic relationships than the Avoidantgroup, but not the Anxious group, F(2,68) = 3.45, P < .05.Avoidant individuals also reported that their childhoodswere less satisfying than the Secure group, but notdifferent from the Anxious group, F(2,68) = 5.79, p < .01.These differences in satisfaction with past relationshipslend validity to the distinctions between attachment groups,as they are consistent with past research (Simpson, 1990).A Note on the Presented DataAs a preliminary aside, it should be noted that some ofthe tables in this thesis contain a large number ofcomparisons, resulting in the inflation of Type 1 error. Inmany instances where tables are presented which contain alarge number of comparisons or results the intention ismerely to provide descriptive information as to observedeffects. Every effort was made to exclude non-significantfindings from the tables, but in those instances where suchdata is presented, please be advised that the author isaware of concerns around Type 1 error and carefullyAttachment and Conflict25considered inclusion of outcomes based on the need tobalance descriptive information while minimizing falsepostive errors.Attachment and Lifestyle AngerThis study's initial set of hypotheses were concernedwith the possible relationships between attachment style andexperiences of anger in everyday life (as opposed toexperimentally-elicited anger). The relationships betweenBartholomew's attachment dimensions and total anger scoreson the MAI are presented in Table 1. The focus in this tableis on the extent to which different elements of attachmentin each participant are correlated with anger patterns. Thisis a within-subjects analysis, as opposed to an examinationof differences in anger between the three attachment groups.The correlations strongly suggested that a greatersense of Assuredness in close relationships was associatedwith less anger as part of one's lifestyle, r(72) = -.38, 2< .01. In contrast, the greater one's avoidant Fearfulnessthat close others are unlikely to meet intimacy needs, thegreater one's tendency to have a robust lifestyle anger,r(72) = .58, p < .01. Anxious Preoccupation with intimacyissues was associated with the tendency to have adiscernible level of trait anger as part of one'sexperience, r(72) = .24, 2 < .05. Dismissing attachment wasunrelated to one's trait anger.A MANOVA conducted to detect possible differencesbetween attachment groups on anger was significant,F(14,128) = 1.82, p < .05. Anger scores of the attachmentgroups were considered in further detail using a series ofoneway ANOVA's. Table 2 presents differences between groupson anger scores. It can been seen that the Avoidant grouphad higher scores than the Secure group on anger frequency,F(2,69) = 5.29, 2 < .01, magnitude F(2,69) = 5.16, p < .01,hostility, F(2,69) = 5.94, 2 < .01, and total anger F(2,69)Attachment and Conflict26= 6.91, p < .01. The Anxious group was not significantlydifferent from the Secure group on any of the angerdimensions, although this group did differ from the Avoidantgroup in their level of hostility, F(2,69) = 5.94, p < .05.In short, the Secure group reported anger had less of a rolein their experiences than the Avoidant group, with theexception of anger expression and duration where the groupsdid not differ.Attachment and AttributionThe second cluster of analyses sought to examine themanner in which attribution about negative behavior by apast or present girlfriend could be related to anger andattachment style. The correlations between total anger scoreand attributions are reported in Table 3. These analysesrevealed a tendency for lifestyle anger to be associatedwith the presence of distress-maintaining attributions inone's previous relationships. This tendency was strongestfor attributions of responsibility to the intimate other forthe behavior (intentional, selfish reasons, blameworthy).Attributing cause for negative events to one's partner (herfault, stability, generalizing) were less robust thanresponsibility, with the exception of the stability item.The observed pattern of correlations between anger andattribution suggests that an angry temperament is closelyinvolved with the tendency to hold one's girlfriendresponsible for aversive behavior and to construe herbehavior as malicious and negative. For example,participants high in trait anger tended to see theirpartner's negative behavior as unlikely to change in thefuture, r(72) = .28, p < .05, intentional as opposed tocaused by an oversight or forgetfulness, r(72) = .33, p <.01, motivated by selfish reasons, r(72) = .34, p < .01, andcause for blame, r(72) = .43, p < .01.Attachment and Conflict27The correlations between attachment dimensions anddistress-maintaining attributions are reported in Table 4.It can be seen that one's level of Assured attachment didnot appear to be associated with a tendency to attributebehavior in particular ways. For example, Assured attachmentwas unrelated to believing that negative behaviors wereintended to spite or hurt, r(72) = -.16, p > .05, or thatthe negative behavior warranted blame, r(72) = -.18, p >.05. The exception to this trend was that Assuredness wascorrelated with believing that negative partner behaviorwould not persist, trustful instead that positive changecould occur, r(72) = -.31, p < .01. This lack ofrelationship between attribution and the Assured dimensionwas not anticipated; it was predicted that this attachmentcomponent would be associated with a lack of distress-maintaining attributions in the form of significant negativecorrelations.The link between attribution and attachment styles wasmore apparent with the insecure dimensions. High Fearfulnesswas associated with the belief that negative partnerbehavior would remain unchanged in the future, r(72) = .36,E < .05, was done on purpose to hurt or spite them, r(72) =.30, p < .05, and was done with selfish motivations, r(72) =.25, p < .05. Men with a strong Fearful attachment componentwere also inclined to believe that their partners deservedto be blamed for what they had done, r(72) = .32, p < .05.The Preoccupied and Dismissing attachment dimensionshad less robust ties to relationship attributions than theFearful measure. Those with a strong Preoccupied element totheir attachment tended to view their partner's negativebehavior as stemming from selfish motivations, r(72) = .25,p < .05, and saw their partners as deserving blame for theirconduct, r(72) = .27, p < .05. The only attributionalstatement significantly related to Dismissing attachment wasAttachment and Conflict28the belief that aversive behavior by one's partner wasunlikely to change in the future, r(72) = .33, p < .01.A MANOVA contrasting attributions between attachmentgroups approached, but failed to reach significance,F(12,126) = 1.54, p = .11. In light of the fact that theMANOVA was close to being significant, further analysis ofbetween-group differences using one-way ANOVA's wereconducted.Attachment and Pre-Experimental MoodThe third hypothesis addressed the specificityassumption of attachment activation by examining the linksbetween attachment styles and pre-stimulus affective state.Tables 5 and 6 report, among other results, the linksbetween attachment dimensions and pre-test anger and anxietyrespectively. Composite Anger was negatively correlated withthe Assured attachment dimension, r(72) =^-.42, p < .01, andpositively associated with the Fearful, r(72)^= .51, p < .01and Dismissing dimensions, r(72) = .24, p <^.05. In terms ofComposite Anxiety, individuals with a strong Assureddimension tended to be less anxious at the beginning of thestudy, r(72) = -.26, p < .05, whereas those high in Fearfulattachment exhibited pre-participation anxiety, r(72) = .51,p < .01, as did those with high Preoccupied scores, r(72) =.29, p < .05. These differences in how attachment dimensionswere related to affect outside of the intimate interpersonalcontext ran against the hypothesized relationship. It hadbeen anticipated that affective state in this situationwould not be linked to attachment.Attachment Style and Negative Affective ResponsesInvestigation of the relationship between attachmentand anger responses are presented in Table 5. Correlationsbetween attachment dimensions and Composite Anger responsesto the conflict audiotapes are presented. High levels ofFearful attachment were associated with greater CompositeAttachment and Conflict29Anger to the Engulfment relationship conflict, r(72) = .29,< .05, and to the working conflict, r(72) = .28, p < .05.Those with escalated Preoccupied attachment had marked Angerresponses only to the Abandonment relationship tape, r(72) =.30, p < .05. Assured and Dismissing attachment dimensionswere unrelated to Anger responses to any of the audiotapedconflicts. The Assured dimension failed to meetsignificance, but the trend was suggestive of an absence ofanger in response to the conflicts.Table 6 outlines the relationships between post-tapeComposite Anxiety and attachment dimensions. Assuredattachment was associated with lower levels of CompositeAnxiety after listening to the Engulfment relationshipconflict, r(72) = -.29, p < .05 and the work conflict, r(72)= -.26, E < .05. On the other hand, a prominent Fearfulrelationship style was linked to elevated Composite Anxietyresponses to the Engulfment conflict, r(72) = .31, p < .05and work conflict, r(72) = .32, p < .05. A robust link wasobserved between Composite Anxiety and Preoccupiedattachment for the Abandonment conflict, r(72) = .39, E <.01, and also for the work conflict, r(72) = .24, E < .05.The observation that all attachment dimensions (exceptDismissing) were linked with the Composite Anxiety responsesto the Work conflict was not predicted by the fifthhypothesis, which suggested that specificity of attachmentprocesses to intimate relationships would mean absence of aconnection between attachment styles and affect in non-intimate conflict.Effect of Conflict Type on MoodA repeated measures ANOVA found a main effect for typeof conflict in both Composite Anger reactions, F(3,65) =46.40, E < .001, and Composite Anxiety reactions, F(3,65) =11.49, p < .001. In other words, the conflict vignettes inthemselves affected mood in different ways. The means andAttachment and Conflict30standard deviations for Composite Anger and Anxiety elicitedby each tape are presented in Table 7. The table providesdescriptive information as to the ways that the conlictvignettes differentially effected mood. Three plannedorthogonal contrasts were performed on the Composite Angerand Anxiety scores as part of repeated-measures univariateMANOVA's in order to examine the effects of the vignettes.Results of the planned orthogonal contrasts will bediscussed first for Composite Anger, and then for CompositeAnxiety.The first contrast served as a manipulation check whichverified that pre-tape levels of Composite Anger, F(1,69) =144.62, p < .001, were less than levels reported after thetapes were heard. The second contrasts determined that angerdiffered as a function of whether the tapes involved datingor working relationship conflict. More Composite Anger wasobserved in response to the work than the dating conflicts,F(1,67) = 13.76, p < .001. The finding that anger wasgreater in response to the working conflict than theintimate conflicts was not anticipated by the fifthhypothesis, which predicted that less anger would beelicited by the work conflict than the intimate conflicts.The final set of planned orthogonal contrasts comparedComposite Anger elicited by the two dating conflicts. TheAbandonment and Engulfment tapes did not differ on level ofComposite Anger elicited, F(1,69) = 2.0, p > .05.The same contrasts were also conducted for theComposite Anxiety responses. The manipulation check contrastrevealed significantly less Composite Anxiety at pre-testthan following the tapes, F(1,69) = 19.61, p < .001. Incomparing intimate with work conflict, Composite Anxiety wasgreater in response to the dating conflicts, F(1,69) =18.06, p < .001. Contrasting the intimate conflicts showedthat Composite Anxiety was higher in response to theAttachment and Conflict31Abandonment tape than the Engulfment tape, F(1,69) = 20.11,p < .001.It is also noteworthy that paired t-tests revealedsignificantly more Composite Anger than Composite Anxiety inresponse to the Engulfment, t(71) = 7.47, p < .001, and Worktapes, t(71) = 8.68, p < .001. On the other hand,participants reported more pre-experimental CompositeAnxiety than Composite Anger, t(69) = 3.88, p < .001.Composite Anger and Composite Anxiety did not differ inresponse to the Engulfment scenario.Attachment Style, Type of Conflict and MoodDifferences between attachment groups in response tothe tapes were of prime importance in hypothesis four.Between-groups repeated measures ANOVA's were conducted onComposite Anger and Anxiety responses to conflicts;attachment style was the between-groups variable whileComposite Anger and Anxiety were the within-subjectsrepeated measures. These results are presented in Table 8.Figures 2 and 3 present the relationships between attachmentstyles and Composite Anger and Anxiety respectively, andprove helpful in reference to the following discussions.Those results pertaining to Composite Anger will bediscussed first, followed by those dealing with CompositeAnxiety.The attachment style by conflict type interaction wassignificant for Composite Anger, F(6,128) = 2.44, p < .05.Responses of the Anxious group to the intimate conflictswere responsible for this interaction. Whereas the Secureand Avoidant groups were angrier after the Engulfment thanAbandonment tapes, the Anxious group was angrier followingthe Abandonment tape, F(2,67) = 5.05, p < .01. Attachmentstyle approached significance as a main effect in CompositeAnger responses, F(2,67) = 2.52, p = .088.Attachment and Conflict32A repeated-measures ANOVA on the Composite Anxietyscale, however, did register differences between attachmentgroups in the form of a main effect, F(2,67) = 6.37, p <.01. The Secure group was significantly lower than theAvoidant group in Composite Anxiety. The Anxious group hadlevels of Composite Anxiety which did not differ from theAvoidant group, but were significantly higher than theSecure group for the Abandonment and Engulfment conflicts.The fact that attachment groups varied in the level ofComposite Anxiety generated by the Work conflict was notanticipated; the principle of attachment specificity wouldargue that differences in emotional states should ariseprimarily in conflict between intimates. The attachmentstyle by conflict interaction was not significant forComposite Anxiety, F(6,132) = 1.54, p = .17. Unlike anger,then, the Engulfment tape tended to elicit the most anxietyfor all attachment styles, while the Work and Abandonmenttapes generated much less anxiety.Predicting Affective ReactionsA final way in which attachment, anger and attributionwere brought together was by examining how they combined inthe responses to conflict. A series of multiple regressionequations were constructed which sought to identify thevariables most predictive of Composite Anger and Anxietyresponses to the experimental stimuli. Included as predictorvariables were measures of attachment (RSQ dimensions),relationship contentment (DAS), attributions from pastrelationships (RAM), and attributions about events on thetape. A stepwise regression method was employed whichidentified predictor variables beginning with thoseaccounting for the most variance and proceeding until allthose variables making significant predictive contributionswere included.Attachment and Conflict33Those variables predicting Composite Anger responses tothe tapes are presented in Table 9. The key variablesassociated with one's level of Composite Anger reaction tothe Engulfment tape were: 1.) the extent of one'spositive/negative feelings towards the female on the tape,2.) whether or not it was believed that the female behavedas she did in the vignette in many other situations, 3.)believing that negative behavior by one's girlfriend wouldpersist into the future (RAM item), and 4.) whether onebelieved that similar conflicts would be common to therelationship in the future. These four factors accounted for37%, F(4,66) = 11.48, p < .001, of the variance in CompositeAnger responses to the Engulfment tape.One's level of Preoccupied attachment was the primarypredictor of Composite Anger response to the Abandonmenttape. Preoccupied attachment accounted for 10%, F(1,68) =8.22, p < .01, of the variance in Composite Anger responsesto the Abandonment tape.Composite Anger responses to the Work tape were bestpredicted by one's levels of Preoccupied and Secureattachment. These variables explained 12%, F(2,68) = 5.91,< .01, of the variance in the anger responses. The fact thatattachment styles were predictive of Composite Anger was notanticipated given the theoretical specificity of attachmentto intimacy anxiety; it had been anticipated that attachmentindices would be predictive solely of affective responses inintimate conflicts.The predictors of Composite Anxiety responses to thetapes are presented in Table 10. The Engulfment tape'sAnxiety responses were predictable from: 1.) the extent towhich the woman's behavior was felt to be similar in othersituations, 2.) believing that negative behavior by one'sgirlfriend would persist into the future (RAM item) and 3.)degree to which her behavior was seen to be situation-drivenAttachment and Conflict34rather than trait-driven. These three variables accountedfor 27%, F(3,67) = 9.27, p < .001, of the variance inComposite Anxiety scores following the Engulfment tape.Composite Anxiety responses to the Abandonment tapewere predicted by Preoccupied attachment and dissatisfactionwith past relationships according to the Dyadic AdjustmentScale. These factors accounted for 29%, F(2,67) = 9.28, p <.001, of the variance in Composite anxiety responses to theAbandonment tape. The fact that the anxious Preoccupieddimension contributed the most predictive power to CompositeAnxiety reactions to a scenario involving potential loss ofintimacy is a strong statement about validity andconsistency in the experimental measures and stimuli.Fearful attachment was the primary predictor ofComposite Anxiety responses to the Work conflict tape. Thisvariable explained 10%, F(1,69) = 7.81, p < .01, of thevariance in the Composite Anxiety responses to the Worktape.DiscussionThe present research has focussed on how attachmentprocesses relate to the experience of negative affect andconflict in close relationships. The findings furthervalidate the distinctions between attachment groups in theexperience of interpersonal discord. Results were generallysupportive of the hypothesized relationships betweenattachment, attribution and mood. Figure 4 provides anoverview of the major findings of the study, outliningdifferences between attachment categories.Attachment Style DifferencesAnalysis of the relationship between attachment andlifestyle anger indicated that the Secure group had a muchweaker profile than the Avoidant group on a variety oflifestyle anger dimensions. Unexpectedly, however, theAttachment and Conflict35Anxious group failed to differ from either the Secure orAvoidant groups on almost all of these lifestyle angermeasures. This absence of significant differences betweenthe Anxious and Secure groups could reflect the possibilitythat anger for Anxious men is not so much a trait-likestructure mapped onto many situations as it is an emotionelicited by particular situational contexts. Theplausibility of this argument is strengthened by the Anxiousgroup's affective responses to the audiotapes, to beaddressed later in this section. High scores on thePreoccupied attachment dimension (a defining characteristicof the Anxious group) were positively correlated withlifestyle anger, however, suggesting that Anxious attachmentis linked with trait anger, though not as robustly as forthe Avoidant group.Investigation of the relationship between attachmentand attributions in close relationships found that theAvoidant group was more likely than the Secure group to viewtheir girlfriends as causing, and responsible for, negativerelationship events. On the other hand, the lack ofassociation between emotional Assuredness and distress-maintaining attributions had not been anticipated. Inevaluating this unexpected finding, perusal of theliterature on marital attributions reveals that the presenceof an attributional style, namely the tendency to explaindiverse negative relationship events in similar ways thatmaintain or maximize distress, is associated with unhappyrelationships (Bradbury and Fincham, 1990). Additionalimplications of the present findings regarding attachmentand an attributional style emphasizing and exaggeratingaversive events will be discussed later in this section.In terms of how attachment groups were anticipated todiffer in their affective responses to the tapes, supportwas found for between-group differences on anxiety. However,Attachment and Conflict36results for anger responses failed to completely support theanticipated outcomes. It had been predicted that theAvoidant group would experience significantly more angerthan the Secure group in response to the intimate conflict.Although the data trended in this direction, differencesbetween attachment groups on anger responses failed to reachsignificance. In trying to understand this result, one canconsider studies of expression of negative affect byAvoidant individuals. Specifically, research has recentlydemonstrated that those with an Avoidant psychologicalprofile lack awareness of their emotional state more thanthe other attachment groups, and distance themselves fromperceptions of their negative mood states (Kobak and Sceery,1988; Mikulincer, Florian, and Weller, 1993). Lower angerresponses to the tapes from Avoidant individuals thanexpected could perhaps be attributed to this group'sdecreased awareness of their emotional state.Anxiety responses to the conflict tapes bore morefavourably on hypothesized attachment group differences. Itwas found that the Secure group had significantly lessanxiety in response to the intimate conflicts than theAvoidant group. The Secure group did not differ from theAnxious group in anxiety responses to conflict overincreased closeness, but had less anxiety than this group inresponse to loss of intimacy. Furthermore, anxiety responsessupported the construct validity of the Anxious group: theirlow anxiety over greater relationship closeness, but greatfear in the face of increased emotional distance is inkeeping with previous research (Simpson, Rholes andNelligan, 1992). This finding about the Anxious group'sreactions was also borne out by the predictive regressionmodels, which found the Preoccupied dimension to beparticularly useful in the determination of both anger andanxiety levels in response to conflict over intimacy loss.Attachment and Conflict37Exploration of the specificity of attachment turned upunexpected results. Theoretically, attachment styles areactivated by situations of acute distress in the context ofclose interpersonal relationships. On this basis it had beenanticipated that attachment groups would not differ in theiremotional state at pre-test. Instead it was clear that thetendency to have prominent insecure dimensions (i.e.,Fearful, Preoccupied, Dismissing) was linked to high levelsof negative affect. Meanwhile, the higher one's Assurednessthe lower one's level of pre-tape negative affect. This isespecially important when considered along with inquiry intothe notion that attachment exerts its greatest influence insituations of intimate distress.Partial support was noted for the prediction thatattachment groups would not differ in their emotionalreactions to work conflicts on the grounds of attachment'sspecificity element. The groups did not differ in theintensity of their anger response to the co-worker conflict.Despite this, however, the Preoccupied attachment dimensionproved to be a significant predictor of anger responses tothe co-worker conflict. Also unexpectedly, the Avoidant andAnxious groups were significantly higher than the Securegroup on anxiety responses to the Work tape. Specificitywould suggest that the groups would not differ in theiranxiety responses. Additionally, correlational analysesshowed that all dimensions except Dismissing weresignificantly correlated with anxiety responses to the Workconflict.Attachment and AngerBoth anger and attachment are emotional ways ofrelating to the social world, and both claim roots in earlydevelopment (Bowlby, 1973). The fact that such strongconnections were observed between attachment and anger wassomewhat surprising. Previous research only tangentiallyAttachment and Conflict38dealt with the attachment-anger link (Bartholomew andHorwitz, 1991). The results of the present research offerinformation about important ways in which those withdifferent attachment styles diverge in their interpersonalexperience of anger.The Avoidant group reported anger was a commoningredient of their everyday experience, but when theylistened to the audiotapes they reported no more anger thaneither the Secure or Anxious groups. Suggesting that thetapes failed to elicit emotional reactions does not hold:anger responses to the tapes were significanty greater thanpre-test anger. The interaction between attachment and typeof conflict, wherein the Anxious group displayed more angerto Abandonment than Engulfment conflict, also weakens thispossible explanation for the unprojected reactions of theAvoidant group. A stronger explanation for Avoidant angerresponses to the tapes draws on previous attachment studiesshowing that Avoidantly attached individuals tend to denynegative affect and inhibit emotional expressions inresponse to distress (Kobak and Sceery, 1988; Mikulincer,Florian, and Tolmatz, 1990). When the responses of theAvoidant group are viewed in this light their low levels ofreported anger may reflect that they were less in tune withtheir emotional state than the other groups, and thereforeprovided reports of their mood which were less thanaccurate. In essence, it may be possible that the Avoidantgroup had prominent trait anger due to items tappingbehavioral and cognitive elements of anger on the MAI (i.e.,"I tend to get angry more frequently than other people."),but when responses to conflict were called for they wereable to only partially tap into their emotional experienceof anger. The present study indicates that further researchon how attachment groups differ in awareness of theiraffective states during distress requires further attention.Attachment and Conflict39A second notable finding in terms of anger andattachment involved the responses of the Anxious group tothe audiotapes. The significant attachment style by conflicttype interaction for anger was caused by the Anxious group'sreport of significantly more anger in response to a lossthan an increase in intimacy. The Secure and Avoidant groupswere more angry over an increase, than a loss, of intimacy.It was consistent with attachment theory that the Anxiousgroup would become more aroused to a threat of being leftthan the threat of being "swamped" with greater closeness.The emotional reactivity of the Anxious group was alsoconsistent with research which has shown that theseindividuals tend to respond to distress at an emotionallevel (Kobak and Sceery, 1988; Mikulincer, Florian andTolmatz, 1990). In this study it was found that those withAnxious attachment demonstrated reactivity in their anger,as well as anxiety, responses to conflict.The fact that anger was such a prominent part of theAnxious group's responses to the vignettes in the absence ofa robust lifestyle anger pattern underscores an importantresearch consideration. Namely, self reports may not tellthe whole story about peoples' anger: how angry they become,or what conditions fuel their anger. Experimentalexamination of how individual differences and situationsinteract require attention as well.Elicitation of Anger to ConflictOne of the experimental outcomes which warrantedspecial attention was the finding that the co-workerconflict elicited more anger than the the intimateconflicts. On the basis of attachment research it wasprojected that anger would be more salient in situations ofintimate than non-intimate conflict for all attachmentgroups. The results, on the other hand, would indicate atfirst face that anger was more readily elicted in theAttachment and Conflict40context of non-intimate relationships. Four otheralternative explanations for this result deserve mention.First, it could be that it was easier for men in the studyto identify with the angry male in the co-worker vignette indispute over the academic issue of course grades. Perhaps itwas easier for students to relate to conflict over gradesthan conflict over relationships issues. Second and in asimilar vein, the sample's lack of experience in closerelationships could have contributed to intimacy mattersbeing more difficult to relate to than a more common issueof academic scope. Third, perhaps it was safer or moresocially acceptable to get angry with a co-worker than aromantic partner. This could have resulted in either thedenial of anger towards the intimate other, or thereluctance to report it. A fourth alternative suggests thatanger may be more readily expressed toward a partner than aco-worker. Therefore, if participants felt their anger wouldnot be expressed as part of conflict with a co-worker theymay have been more aware of anger brewing within them whichwould not be vented. Our understanding of the how themeaningfulness of relationships influences our emotionalreactions to discord could benefit from further research.A concluding remark on negative affect (and awarenessthereof), attachment and intimacy could be fused draws onclassic work in conflict. Dollard and Miller (1950)suggested that anger/anxiety conditions in early developmentcould influence the development of personality dispositions.Of significance here is the way parents deal with theirchild's expression of anger. Punishing a child's angerresponse to distress out of existance could be particularlydangerous, as it could rob them of the tools necessary todevelop assertiveness and confidence. On the basis of workin attachment it could also result in producing a child outof touch with their emotional state, a lack of awarenessAttachment and Conflict41which often goes uncorrected for years, if not permanently(Miller, 1990). The influence of early family experience inthe development of how we come to experience and valuenegative emotional states is crucial, and the presentresearch suggests that attachment has a role to play in ourunderstanding of how these childhood patterns relate toadulthood.Attachment and Development of Attributional StyleResearch in attribution has addressed the issue ofattributional style, namely the tendency to diverse negativerelationship events and outcomes in ways that consistentlymaintain or maximize distress (see Bradbury and Fincham,1990 for a review). It has been found that people indistressed relationships tend to routinely perceive negativeevents in ways that maintain discord. People with such anattributional style tend to explain a wide range of negativeevents with the same core thoughts, such as their partnermeant to upset them, will continue to to cause problems, andwill cause difficulties in many areas of the relationship(Baucom, Sayers and Duhe, 1989). Studies have differed intheir reports of how attributions are made by peoplesatisfied with their relationships. On the one hand, peoplein non-distressed relationships have been shown to react tonegative relationship events in ways "that explained thebehavior as being due to outside circumstances or to thepartner's temporary state, involuntary, unintentional"(Holtzworth-Monroe and Jacobson, 1985, p. 1403). Baucom,Sayers, and Duhe (1989), on the other hand, found that thosein satisfied relationships did not have a tendency to viewnegative events in a "minimizing" described by Holtzworth-Monroe and Jacobson (1985), but instead demonstrated widevariablility in the ways negative events were explained.Findings in the present study more were consistent withBaucom, Sayers, and Duhe's (1989) study of relationshipAttachment and Conflict42satisfaction and attributional style than the work ofHoltzworth-Monroe and Jacobson (1985). Insecure attachmentworking models tended to describe unfavorable events in waysthat maximized the aversive impact of such events. Secureattachment was essentially unrelated to attributionpatterns. However, some evidence in support of a"minimizing" tendency for satisfied participants did emerge.It was found that Assuredness was associated with believingthe negative behavior of one's partner could change in thefuture. Also, regression equations suggested that minimizingthe cause and responsiblity of the woman in the vignetteswas associated with responses of less anger and anxiety. Insum, the present research is in keeping with past findingswhere evidence both supporting and disconfirming theassociations between cognition and relationship adjustmenthave been found. It is clear as well that theseattributional styles are part of the fabric of theexperience of interpersonal discord.Importantly, research on marital attributions hasfailed to consider the role of ontogenetic etiologicalfactors which could contribute to the rise of the distressfor which explanations are sought. In other words, althoughit has been noted that reports of dissatisfaction areassociated with the particular patterns of explanation forproblems, explanations for how those attributional stylesarise have been sparse. One of the few commentaries on thedevelopment of attributional styles was offered by Bradburyand Fincham (1990). They proposed that a predisposition maylead those in distressed relationships to view the other'sbehavior as negative, unexpected or self-relevant. Theylabelled this predisposition the primary process, a systemOf selective attention guiding the types of interpersonalevents people notice, or look for, which operates at a non-conscious level. The primary process, comprised of elementsAttachment and Conflict43propelling attribution, overlaps neatly with the tenets ofthe attachment working models studied in the presentresearch.The differences observed between attachment groups intheir emotional and cognitive responses to aversiverelationship events speaks to the possible role ofattachment in the rise of primary process. It is reasonableto consider the primary process as constituted or shaped byone's working model of attachment, which influences patternsof attending to events in intimate relationships andexplaining attachment figure behavior. In addition, Pipp andHarmon (1987) provided an account of how cognition andattachment could be connected in the interpersonal domain.They elaborated on the notion that attachment isincorporated at visceral and emotional (in addition tocognitive) levels of the person. They also noted thatattachment begins with the physical synchronization ofbiological rhythms in infancy which remain entrenchedthroughout the lifetime as a component of sensorimotorequilibrium. Across the lifespan our bodies seek to maintaina familiar form of attachment at visceral and sensorimotorlevels as part of our interpersonal experience. In short,the body's maintenance of these personal homeostatic normscould be a part of the primary process and its cognitiveassessment of intimacy. Further research on links betweenattributional style, primary process and attachment couldgreatly facilitate our understanding of intimate experience.Also worthy of investigation would be the study ofattributional style differences between attachment groupsfor positive relationship events in close relationships.Such explorations would be of great assistance in theintegration of research on attachment, attributional stylesand marital satisfaction.Attachment and Conflict44Attachment and Attributional Style: Links to AngerThe present research has introduced attachment as anadditional consideration in the study of anger and conflictin close relationships. Research exploring the connectionsbetween anger, attribution and aggression has resulted infindings relevant to the results of the present study.Research on an attributional style which tends to holdothers responsible for negative events is particularlyuseful in contextualizing the present research. Weiner,Graham and Chandler (1982) found that when participants heldanother person responsible for negative events they weremore likely to become angry. Fondacaro and Heller (1990)noted that aggressive boys were more likely than non-aggressive boys to make external person-centered blame forambiguous problem situations. In a similar vein, Dodge andTomlin (1987) found that aggressive adolescents were morelikely than non-aggressive teens to construe the intentionsof others as hostile in ambiguous problem situations.Attributional style, then, seems to be accompanied not onlyby relationship dissatisfaction, but by anger and aggressionas well.Comparable findings were noted in the present study. Itwas observed that Avoidant men attributed high levels ofresponsibility for unfavorable outcomes to their partners.These men also reported having anger as a strong componentof their lifestyle. The present research indicated thatinsecurely attached men seem to have the attributional styleelements, namely other-focussed loci of cause andresponsibility, which other have shown are linked with angerand aggressiveness. Experimental work with audiotapes in thecurrent research suggested that for both Avoidant andAnxious men, to a greater extent than Secure men, there wassomething operating which contributed to pronounced levelsof negative affect, in the forms of anger and anxiety, whenAttachment and Conflict45faced with some distressful situations. To go beyond thesefindings in trying to ascertain the role of attachment inthe sequence of psychological events culminating in abuserequires consideration of models of attempting to describehow agggression could be generated.Implications for the Study of Aggression and AbuseThe present study has found that attachment, traitanger, and attributional style all play a role in thegeneration of anger and anxiety in close relationships. Soif attachment, trait anger and attribution are involved inthe intensity of negative affect experienced in response tointerpersonal conflict, how would these dynamics beassociated with aggression? Fondacaro and Heller (1990)proposed that attributing blame and hostile motivations toothers could serve to enhance feelings of personal controlby placing the source of anger in the social environment.Specifically, if something or someone in the environment isheld accountable for negative events, one can take actiontowards preventing aversive events in the future: strikingat the source to demonstrate the consequences of creatingaversive situations. An externalizing attributional stylefacilitates a retaliatory response to dealing withinterpersonal problems by suppressing the perceived sourceof the problem with threats and acts of aggression. Studiesof psychological abuse by men in intimate relationships hassuggested that the more prominent one's insecure attachmentstyles, and accompanying tendency to attribute externally,the greater the likelihood of being dominant with one'spartner and also to isolate her from social ties (Dutton andStarzomski, 1992).In applying this line of thinking to the presentresearch, it should be noted that the attributional stylesof insecure attachment working models attributed cause andresponsibility for negative events out to their partners.Attachment and Conflict46These men also had the most negative affect in theirresponses to conflict audiotapes, suggesting that they doindeed react more strongly to intimate distress. A casecould be made that the propensity of insecurely attached mento attribute cause and responsibility to their partnerswould put them at risk for adopting aggressive solutions toresolve interpersonal disputes. These men would feelempowered by these options, a central motivation in theperpetration of intimate violence (Dutton and Browning,1988). This theory is consistent with the anger/anxietytemplate theory advanced by Dutton, Saunders and Bartholomew(1992). The ways in which the present study's findingsabout the insecure styles relate to the anger/anxietytemplate deserves further attention.The present study raised the possiblity that those withAnxious styles may be at risk for becoming emotionallyexplosive in particular situations, namely abandonment. Ifthe men believe that venting that emotional state in theform of violence is condonable they would seem to be at riskto aggress against the person they construe as responsiblefor their aversive state. Exploring the situational role ofanger and the emotional ways in which this group deals withdistressing situations calls out for further consideration.Another important issue with respect to emotionalresponses and insecure attachment was concern over theextent to which men were aware of their emotional state. IfAvoidant men distance themselves from the negative emotionsof distress in a close relationship, that emotional statemay go unchecked until a violent boiling point is reached.Further research is needed in order to determine the ways inwhich attachment styles are associated with the awareness ofone's emotional state. If Avoidant men are out of touch withtheir emotional state, possibly predisposing them tosurprisingly reach anger boiling points, then research couldAttachment and Conflict47begin exploring ways to prevent this cycle from occuring. Itwould seem that exploring empathic awareness andcommunication skills as a way of enhancing attention paid toemotional states could prove useful in redressing the denialof emotion in personal experience.By addressing interpersonal process throughexperimental stimuli, in addition to questionnaires, thisresearch has cast a broader net than many studies exploringthe psychology of abuse. Studies of aggression have oftenrelied primarily on self report questionnaires which provideinformation biased in self-perception and self-disclosure(Dutton and Starzomski, in press). On the other hand,experimental work yields information to be reconciled withquestionnaire data. For example, the fact that the Avoidantgroup had lower levels of situational anger responses thanone would expect given the strong questionnaire findingsabout their trait anger may be important, but wouldn't havebeen detected in an epistemology consisting solely ofquestionnaire methods. Experimental simulation of conflictcan provide a testing ground for theories of how the chainof command operates in the perpetration of abuse.At a broader level, examining this study's outcomes hasnot proven to be a simple task. The picture which hasemerged from this study is not totally clear. What hasbecome apparent is these attachment styles do not haveindubitable claims to unique forms of emotional experiencein a trait-like form. There is a tendency for those withdifferent attachment styles to react in similar to ways toone distressing matter, and then to react very differentlyto another issue. Therefore, caution is advised in attemptsto map attachment categories and their "stereotypical"responses onto real-world events, including abuse.Attachment theory, as this study shows, is not able toaccount for even most of the diversity in how people manageAttachment and Conflict48intimate distress. But that by no means implies that thereis no place for attachment in the study of abusiverelationships.What became apparent in this study, however, was thatSecure people tended to report both low levels of traitanger, as well as to remain relatively unaroused in the faceof the conflict vignettes. The results with respect to theinsecure groups required more attention in order to come togrips with some unexpected outcomes. Theories of violencewhich suggest that abuse is caused mainly by particulartypes of attachment styles are clearly missing the complexway in which these systems influence behavior. Researchersare a long way from being able to deal with the myriad ofoutcomes observable from these styles. As of yet there is noclear way of separating abusers from non-abusers on thebasis of assessments of attachment. Further research isclearly needed before the practical applications ofattachment theory can be foreseen, let alone employed. Inshort, the temptation to employ a view of attachmentstructures' influence in interpersonal events which clingsto a reductionistic model is clearly premature.General Conclusions and Future DirectionsThe focus of the present study on the processes ofconflict and psychological processes in intimacy fits into aniche all of its own in terms of related work indevelopmental, social, and forensic psychology. A look athow other theories of intrapersonal events overlap with thepresent study serves to highlight unequalled elements ofpotential originated here. No other researchers havecombined attachment theory with social cognition in thespecific domain of intimate interpersonal conflict. The workof Collins (1992) consisted of linking attachment, mood, andAttachment and Conflict49attribution, but she did not deal with the specific issuesof conflict. Others have endeavoured to grapple withinterpersonal conflict and ignored attachment (Fincham andBradbury, 1987), focussing instead on the power ofattribution. A theory put forward by Betancourt and Blair(1992) integrated emotion and Weiner's (1985, 1986) theoryof attribution, the dominant one in field today, into amodel for predicting aggression. Their model, however,neglects the emotional meaningfulness of interpersonalconnections as addressed in attachment theory. In addition,Bradbury and Fincham (1987) pointed out that Weiner'saccounts of motivation have not emphasized the attributionsof cause and responsibility, which have been uncovered asthe primary dimensions of concern in the dysfunctionalmarital attribution research.An area of research carved out by the present studyconcerns attributional style, a key element of theanger/anxiety template theory. It was observed that Avoidantmen tended to have an attributional style which played upthe role of their partner in aversive negative relationshipevents, contrary to men with a Secure attachment style. Thiscognitive step in the template theory received some supportin the results of the present study. On the other hand,there could be a lot of men who are Avoidantly attached, whohave an angry and attributional style, but who do not abusetheir partners.Additional research is clearly needed in order torefine our sense of how attachment, attributional style andcontext influence emotional and behavioral reactions toconflict. One important topic for future research involvesfurther exploration of the experience of anger in intimacyin the context of attachment. The present research suggeststhat the emotional arousal of the Anxious group isparticulary noteworthy, despite the fact that previousAttachment and Conflict50research has looked mainly at the Avoidant group'scharacteristics. Also of interest in this area would begaining some idea as to possible differences betweenattachment groups in awareness of their own emotional stateswhile in situations of intimate distress. Additionally, italso seems important to further evaluate the role ofsituations as triggers of attachment working models andattributional styles. Contextual cues clearly impact on theextent to which individuals become emotionally aroused inconflict situations. Incorporating a decidedly developmentalapproach to research in the emergence of attachment styledifferences in intimate relationships could also assistresearchers searching for the causes and solutions toabusive adult relationships. The exploration ofepistemological alternatives to questionnaire research, suchas narrative studies or observations of reactions to actualinterpersonal stimuli, seem to be necessary in order toassemble a more accurate understanding of the psychology ofconflict than we currently have.Attachment and Conflict51ReferencesAinsworth, M., Blehar, M., Waters, E., & Wall, S. (1978).Patterns of attachment. Hillsdale, NJ: Erlbaum.Bandura, A., Ross, D., & Ross, S. (1961). Transmission ofaggression through imitation and aggressive models.Journal of Abnormal and Social Psychology, 63, 575-582.Bartholomew, K. (1990). Avoidance of intimacy: An attachmentperspective. Journal of Social and Personal Relationships, 7, 147-178.Bartholomew, K. & Horowitz, L. M. (1991). Attachment stylesamong young adults: A test of a four-category model.Journal of Personality and Social Psychology, 61(2),226-244.Baucom, D. H., Sayers, S. L., & Duhe, A. (1989).Attributional style and attributional patterns amongmarried couples. Journal of Personality and Social Psychology, 56, 596-607.Betancourt, H., & Blair, I. (1992). A cognition(attribution)-emotion model of violence in conflictsituations. Personality and Social Psychology Bulletin,18(3), 343-350.Bowlby, J. (1969). Attachment and loss. Volume 1: Attachment. New York: Basic Books.Bowlby, J. (1973). Attachment and loss. Volume 2: Separation: Anxiety and anger. New York: Basic Books.Bowlby, J. (1980). Attachment and loss. Volume 3: Loss: Sadness and depression. New York: Basic Books.Bowlby, J. (1984). Violence in the family as a disorder ofthe attachment and caregiving systems. American Journal of Psychoanalysis, 44(1), 9-27.Bradbury T. N., & Fincham, F. D. (1990). Attributions inmarriage: Review and Critique. Psychological Bulletin,107, 3-33.Bradbury, F. N., & Fincham, F. D. (1987). Affect andcognition in close relationships: Towards anintegrative model. Cognition and Emotion, 1(1), 59-87.Attachment and Conflict52Browning, J. J. (1983). Violence towards intimates: Toward aprofile of the wife assaulter. Unpublished doctoraldissertation, University of British Columbia,Vancouver, B. C.Collins, N. L. (1992). Attachment style differences in patterns of explanation, emotion, and behavior. Unpublished manuscript. University of California, LosAngeles.Collins, N. L. & and Read, S. J. (1990). Adult attachment,working models and relationship quality in datingcouples. Journal of Personality and Social Psychology,58, 644-663.Crittenden, P. M., & Ainsworth, M. (1989). Childmaltreatment and attachment theory. In D. Cicchetti &V. Carlson (Eds.), Child maltreatment: Theory and research on the causes and consequences of child abuse and neglect (pp. 432-463). Cambridge, England:Cambridge University Press.Cummings, E. M., Zahn-Waxler, C., & Radke-Yarrow, M. (1981).Young children's responses to expressions of anger andaffection by others in the family. Child Development,52, 1274-1282.Davis, M. H., et al (1987). Emotional reactions to dramaticfilm stimuli: The influence of cognitive and emotionalempathy. Journal of Personality and Social Psychology,52 (1), 126-133.Dodge, K. A., & Tomlin, A. M. (1987). Utilization of self-schemas as a mechanism of interpretational bias inaggressive children. Social Cognition, 5(3), 280-300.Dollard, J., & Miller, N. E. (1950). Personality and psychotherapy: An analysis in terms of learning, thinking and culture. New York: McGraw-Hill.Dutton, D. G. (in press). Behavioral and affectivecorrelates of Borderline Personality Organization inwife assaulters. International Journal of Law and Psychiatry. Dutton, D. G. (1988). The domestic assault of women. Boston: Allyn and Bacon.Attachment and Conflict53Dutton, D. G., & Browning, J. J. (1988). Concern for power,fear of intimacy, and aversive stimuli for wifeassaulters. In B. T. Hotaling, D. Finkelhor, J. T.Kirkpatrick & M. A. Straus (Eds.) Family abuse and its consequences: New directions in research. Newbury Park,Ca.: Sage Publications.Dutton, D. G., Saunders, K., & Bartholomew, K. (1992).Intimacy-anger and insecure attachment as precursors ofabuse in intimate relationships. UnpublishedManuscript, University of British Columbia, Vancouver,B. C.Dutton, D. G., & Starzomski, A. J. (in press). Psychologicaldifferences in court-referred and self-referred wifeassaulters. Criminal Justice and Behavior.Dutton, D. G., & Starzomski, A. J. (1992). Perpetrator characteristics associated with women's reports of psychological and physical abuse. Unpublishedmanuscript, University of British Columbia, Vancouver,B. C.Dutton, D.G., Starzomski, A. J., & Bartholomew, K. (1992).Fearful attachment and the psychological abuse of intimate female partners. Unpublished manuscript,University of British Columbia, Vancouver, B. C.Dutton, D. G., Webb, A. N., & Ryan, L. (1992). Anger, anxiety and gender differences in affective reactions to witnessing dyadic family conflict. Unpublishedmanuscript, University of British Columbia, Vancouver,B. C.Erikson, E. (1963). Childhood and society (2nd ed.). NewYork: Horton.Feeney, J. A., & Noller, P. (1990). Attachment style as apredictor of adult romantic relationships. Journal of Personality and Social Psychology, 58(2), 281-291.Fincham, F. D., & Bradbury, T. N. (1992). Assessingattributions in marriage: The Relationship AttributionMeasure. Journal of Personality and Social Psychology,62(3), 457-468.Fondacaro, M. R., & Heller, K. (1990). Attributional stylein aggressive adolescent boys. Journal of Abnormal Child Psychology, 18(1), 75-89.Attachment and Conflict54Freud, S. (1926). Inhibition, symptoms, and anxiety. (pp 87-172) In The complete psychological works of SigmundFreud: Standard Edition. London: Hogarth Press,Ltd.Harlow, H. F., & Harlow, M. (1971). Psychopathology inmonkeys. In H. D. Kinnel (Ed.), Experimental psychopathology. New York: Academic Press.Harvey, J. H., Yarkin, K. L., Lightner, J. M., & Town, J. P.(1980). Unsolicited interpretation and recall ofinterpersonal events. Journal of Personality and Social Psychology, 48, 1398-1412.Hazan, C. & Shaver, P. (1987). Romantic love conceptualizedas an attachment process. Journal of Personality and Social Psychology, 52(3), 511-524.Holtzworth-Munroe, A. & Jacobson, N. S. (1985). Causalattributions of married couples: When do they searchfor causes? What do they conclude when they do? Journal of Personality and Social Psychology, 48(6), 1398-1412.Howe, G. W. (1987). Attributions of complex cause and theperception of marital conflict. Journal of Personalityand Social Psychology, 53(6), 1119-1128.Kobak, R. R., & Sceery, A. (1988). Attachment in lateadolescence: Working models, affect regulation, andrepresentations of self and others. Child Development,59, 135-146.Main, M. & Weston, D. R. (1982). Avoidance of the attachmentfigure in infancy: Descriptions and interpretations. InC. M. Parkes & J. Stevenson-Hinde (Eds.), The place of attachment in human behavior. London: TavistockPublishers Ltd.Mikulincer, M., Florian, V., & Tolmatz, R. (1990).Attachment styles and fear of personal death: A casestudy of affect regulation. Journal of Personality and Social Psychology, 58, 273-280.Mikulincer, M., Florian, V., & Weller, A. (1993). Attachmentstyles, coping strategies, and posttraumaticpsychological distress: The impact of the Gulf War inIsrael. Journal of Personality and Social Psychology,64(5), 817-826.Miller, A. (1990). Banished knowledge: Facing childhoodinjuries. New York: Doubleday.Attachment and Conflict55Pipp, S., & Harmon, R. J. (1987). Attachment as regulation:A commentary. Child Development, 58, 648-652.Russell, J., & Mehrabian, A. (1974). Distinguishing angerand anxiety in terms of emotional response factors.Journal of Consulting and Clinical Psychology, 42, 79-83.Shaver, R. S., & Brennan, K. A. (1992). Attachment stylesand the "Big Five" personality traits: Theirconnections with each other and with romanticrelationship outcomes. Personality and Social Psychology Bulletin, 18(5), 536-545.Shaver, R. S., & Hazan, C. (in press). Adult romanticattachment: Theory and evidence. In D. Perlman & W.Jones (Eds.) Advances in personal relationships (Vol.4). San Diego, Ca: Academic Press.Siegel, J. M.. (1986). The multidimensional angerinventory. Journal of Personality and Social Psychology, 51(1), 191-200.Simpson, J. A. (1990). Influence of attachment styles onromantic relationships. Journal of Personality and Social Psychology, 59(5), 971-980.Simpson, J. A., Rholes, W. S., & Nelligan, J. S. (1992).Support seeking and support giving within couples in ananxiety-provoking situation: The role of attachmentstyles. Journal of Personality and Social Psychology,62(3), 434-446.Spanier, G. (1976). Measuring dyadic adjustment: Newscales for assessing the quality of marriage andsimilar dyads. Journal of Marriage and the Family, 3,15-28.Strachan, K., & Dutton, D. G. (1992). The role of powerand gender responses to sexual jealousy. Journal ofApplied Social Psychology, 22, 1721-1740.. Sullivan, H. S. (1953). The interpersonal theory of psychiatry. New York: W. W. Norton & Co.Weiner, B. (1986). An attributional theory of motivation andemotion. New York: Springer-Verlag.Weiner, B. (1985). An attributional theory of achievementmotivation and emotion. Psychological Review, 92, 548-573.Attachment and Conflict56Weiner, B., Graham, S., & Chandler, C. (1982). Pity, anger,and guilt: An attributional analysis. Personality and Social Psychology Bulletin, 8, 226-232.Attachment and Conflict57Table 1.Correlations Between Attachment and Anger Total. Attachment DimensionAnger TotalAssured -.38**Fearful .58**Preoccupied .24*Dismissing .19Note. * 2 < .05.^** 2 < .01.Table 2.Means (& Standard Deviations) of Attachment Groups on Anger Dimensions.Attachment GroupAnger (MAI)DimensionSecure^Avoidant(n=22) (n=31)Anxious(n=17)In 16.57^(3.7)^18.41^(3.0) 16.18 (3.7) 3.17 *Out 10.17^(2.6) 10.34^(2.0) 11.41 (1.5) 1.91Frequency 8.61^(3.8)^a 12.13^(3.8) b 10.06 (4.7) 5.29 **Duration 4.74^(1.7) 5.47^(1.4) 4.94 (1.9) 1.45rtMagnitude 8.57^(2.4)^a 11.06^(2.6) b 9.29 (4.1) 5.16 ** rta)Hostility 14.17^(3.9)^a 17.03^(3.5) b 14.02 (3.3)^a 5.94 **crtTotal 64.35^(14.4)^a 78.16^(11.4) b 69.54 (17.0) 6.91 ** 00Note. * p < .05.^** p <^.01. 0Means followed by different letters vary at p < .05. H .u-1 ocoo-Attachment and Conflict59Table 3.Correlations Between Anger and Distress-MaintainingAttributions from Actual Relationships.Distress-MaintainingAnger TotalAttributions About PartnerHer fault .17No change in future .28 *Generalizes to other issues .22Controllable .33 **Motivated by selfish reasons .34 **Blameworthy .43 ***Note.^* p <^.05.^** p <^.01. *** p < .001.Table 4.Correlations Between Attachment Styles and Distress-Maintaining Attributions fromActual Relationships.AttachmentHer faultNo changein futureDistress-Maintaining AttributionsBlameworthyGeneralizes toother issues IntentionalSelfishReasonsDimensionAssured -.08 -.31 ** -.05 -.16 -.15 -.18Fearful .14 .36 ** .18 .30 * .25 * .33 **Preocc. .21 .00 .00 .10 .25 * .26 *Dismiss. .16 .33 ** .13 .19 .10 .13Note. * p < .05. ** p < .01.Attachment and Conflict61Table 5Correlations Between Attachment Dimensions and Composite Anger States. StimuliAssuredAttachment DimensionsPreoccupiedFearful DismissingPretest -.39 ** .51 ** .24 * .20Engulfment -.21 .29 * .16 .09Abandonment -.08 .16 -.12 .30 *Work -.23 .28 * -.04 .19Note. * 2 < .05.^** 2 < .01.Attachment and Conflict62Table 6.Correlations Between Attachment Dimensions and Composite Anxiety States. StimuliAssuredAttachment DimensionsPreoccupiedFearful DismissingPretest -.21 .51 ** .17 .29 *Engulfment -.29 * .31 * .09 .21Abandonment -.23 .21 -.11 .39 **Work -.26 ** .32 ** -.02 .24 *Note. * P < .05.^** 2 < .01.Table 7.Means Scores (& Standard Deviations) on Composite Anger and Anxiety Scales Pre- andPost-Tapes.1^ 2^ 3^ 4Pre-test^Engulfment^Abandonment^WorkAnger^12.8 (6.6)^26.9 (12.6)^25.4 (12.4)^29.9 (13.5)Orthogonal Contrasts: 1.) 1 v. 2,3,4 :^F (1,69) = 144.62 ***2.) 2,3 v. 4^F (1,69) = 13.76 ***3.)^2 v.3 F (1,69) = 1.971^ 2^ 3^ 4Pre-test^Engulfment^Abandonment^WorkAnxiety^15.7 (8.9)^18.9 (9.6)^23.8 (12.1)^18.0 (7.8)Orthogonal Constrasts 1.) 1 v. 2,3,4 :^F (1,69)= 19.61 ***2.) 2,3 v. 4^:^F (1,69) = 18.06 ***3.)^2 v.3 F (1,69)= 20.11 ***Note. *** p < .001Table 8.Means (& Standard Deviations) of Attachment Groups on Anger and Anxiety Responses toConflict Tapes. Attachment GroupSecure^Avoidant^Anxious(n=22) (n=31) (n=17) Anger1.) Engulfment^24.48 (12.9)^29.75 (11.1)^26.23 (14.3)^1.262.) Abandonment 22.39 (13.0) 25.91 (12.1)^30.12 (11.4) 1.963.) Work^25.73 (14.4)^33.16 (13.4)^30.12 (11.4)^2.080-rrr00a00rn0.N r)rtAnxiety 1.) Engulfment^14.82 (6.9) a2.) Abandonment 17.65 (10.7) a3.) Work^14.04 (5.9) aNote. * p < .05. ** p < .01.Means followed by different letters vary at p < .05.4.30 *5.38 **5.21 **22.34 (9.8) b^19.24 (11.3)26.31 (11.6) b^28.18 (11.8) b19.31 (6.9) b^20.94 (9.6) bAttachment and ConflictTable 9.Prediction of Anger Responses to Conflict Tapes.65B SE B beta1.) EngulfmentNegative attitudetowards woman 3.23 .93 .35 3.47 ***She often behaveslike this with men 2.54 1.07 .24 2.37 *Her negative behaviorwill persist .42 .19 .21 2.21 *Conflict occursfrequently 3.42 1.66 .20 2.07 *Multiple R = .64 Adj. R Square =^.37F^(4,66)^=^11.48^***2.) AbandonmentPreoccupiedAttachment 1.28 .45 .33 2.87 **Multiple R = .33 Adj. R Square =^.10F^(1,68)^=^8.22^**3.)^WorkPreoccupiedAttachment 1.34 .49 .31 2.74 **SecureAttachment -1.02 .47 -.24 -2.17 *Multiple R = .39 Adj. R Square =^.12F=^(2,68)^=^5.91^**Note.^* p <.05.^** p <^.01. *** p <^.001.Attachment and ConflictTable 10.Prediction of Anxiety Responses to Conflict Tapes.66B SE B beta1.) EngulfmentShe often behaveslike this with men 2.46 .91 .29 2.71 **Her negative behaviorwill persist .40 .17 .25 2.40 *Her behavior causedby the situation,not her traits -1.63 .77 -.23 -2.11 *Multiple R = .54 Adj. R Square =^.27F^(3,67)^=^9.27^***2.) AbandonmentPreoccupiedAttachment 1.48 .43 .37 3.45 ***Dyadic Adjustment(Satisfaction withPast Relationships) -.23 .10 -.26 -2.36 *Multiple R = .58 Adj. R Square =^.29F^(2,67)^=^9.28^***3.)^WorkFearfulAttachment .84 .30 .32 2.80 ***Multiple R = .32 Adj. R Square =^.10F (1,69) = 7.81 **Note. * p <.05. ** p < .01. *** p < .001.Figure 1Attachment Dimension Scores by Attachment GroupFigure 2Anger Reactions by Attachment Group 35 —30 —25 —Anger20 ——*— Anxious--n— Avoidant—A— Secure15 — rtrtrtPncort00CLn001-11I--.H.CT nCo rt10 —5Pretest^Engulfment Abandonment^WorkConflict ScenarioFigure 3Anxiety Reactions by Attachment Group35 —30 — 25 —Anxiety20 —AnxiousAvoidant—Ag— Secure15 —0>rtrtA)00rt00H.,O'1/41/40 rt10--5Pretest^Engulfment Abandonment^WorkConflict ScenarioFigure 4 Aspects of Anger, Attribution and Affect for Different Attachment StylesAnger"negative eventsAttribution will be rare in thefuture"ReactionstoStimulilowest anxiety toEngulf, Abandon &Work; unrelated toanger responsesAvoidantabsence of approachbehavior; rejection ofothers' attempts tocomfort; fearful anddismissing of intimacyfrequent angerwhich is hostile andheated; tends tointernalize andnot externalize"her negative behaviorwas intended, selfishlymotivated, & willpersist; I blame her"high anxiety toEngulf, Abandon &Work; predictive ofanxiety to Engulf;tended to get angryto Engulf and WorkAnxiousconstant pursuit ofattachment figurefear of potentialrelationship loss &unmet emotionalneedsresembles both sec.& ay., but candirect anger outward& tends not to behostile"she was motivatedby selfish reasons &I blame her"average anxiety toEngulf., high toWork; predictive ofanger to AbandonEngulf & Work;predictive ofanxiety to AbandonSecurecomfortable withinterdependence; ableDescription to trust and be trusted;of^absence of anxiety overstyle^relationship failure:comfort with alonenessanger is infrequent,neither intense norhostile; tends not tointernalize angerP>rtrtI(D0rt(1)0a,C)01-11no rt


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items