Content uploaded by Donald G. Dutton
Author content
All content in this area was uploaded by Donald G. Dutton on Jul 01, 2014
Content may be subject to copyright.
Dutton, D.G., Starzomski, A., Saunders, K., & Bartholomew, K. (1994). Intimacy-anger
and insecure attachment as precursors of abuse in intimate relationships. Journal
of Applied Social Psychology, 24(15), 1367-1386.
INTIMACY-ANGER AND INSECURE ATTACHME NT AS
PRECURSORS
OF ABUSE IN INTIMATE RELATIONSHI P S
Donald G. Dutton
Keith Saunders
Andre w Starzomski
Departm e n t of Psychology
University of British Columbia
Kim Barth olomew
Departm e n t of Psychology
Simon Fraser University
Running Hea d: Intimacy- ang e r and atta c h men t
Intimacy- anger and attachme nt
2
This researc h was implemen t ed through grants from the Solicitor
General of Canad a and Health and Welfare Can a d a . Requests for
rep rints should be sent to Dr. Don Dutton, Departme n t of Psychology,
University of British Columbia, Vancouver, B.C., V6T 1Z4, Canada.
Intimacy- anger and attachme nt
3
Abstract
One hundre d and twenty men ref er r e d for tre a t men t for wife assault
and forty demogr a phic controls complet ed self-repor t questionnaire s
asses sing attachme n t patter n s , anger, jealousy, Borderlin e
Personality Orga nization an d tra u m a sympto m s. This cons t ellation,
with the exception of attach m e nt, has been found to represe n t a
profile relate d significantly to frequency of both verbal and physical
abu s e. In the prese n t study, attachmen t was found to correlate
significantly with these other measur e s. A meas ure of secur e
attachm e nt correl at e d significantly but negat ively with the
constellation meas ure s . A measure of fearful attac hme n t correlated
significantly and positively. A preoccupied attachm e nt patt e r n
correlated with the other measur e s less stro n gly than the fearful
meas u re. A dismis sing measure was not correlated with any
constellation meas ure s . The role of attachmen t- anger is discus s ed; it
is suggested tha t fearful atta c h men t could also be descr ib e d as ang ry
attachm e nt. The concep t of intimacy- ang e r is introduced from early
studies in att ach m en t and applie d to the explanation of assaul tiveness
in adult relationships.
Intimacy- anger and attachme nt
4
Intimacy- anger And Insecur e Attach men t As Precursors
Of Abuse In Intimate Relationships
Bowlby asserted that "attachmen t behavior is held to
charac t e rize human beings from the cradle to the grave" (1977, p.
203). He postulated that confidence in the availability of attachm e nt
figures, or lack of it, is built slowly during the years of immaturity- -
infancy, childhood, and adolescence- -and th at "whatever expec t ations
are developed durin g those years tend to persist relativ ely unchanged
throughout the res t of life" (1973, p. 235). These expectations,
referred to as "internal repr ese n tations" or "working mod els" of self
and intimat e par t n e rs, are centr al compone n t s of personality. As he
put it "...when an individual is confid e n t that an attach m e nt figure will
be available to him (sic) whe n e v e r he desir e s it, that per son will be
much less prone to either intense or chronic fear than will an
individual who for any reason has no such confidence" (1973, p.235).
Bowlby viewed inter per s o nal anger as arising from frustrate d
attachm e nt needs and functioning as a form of "protest behavio r"
directed at regaining conta c t with the atta ch m ent figure. A
fundament a l conclusion of attachme nt res e arch on infants is that
ang e r and rage follow unmet attac h me n t ne ed s; that threats or
sep a r a tions from attachm e nt figures produce power ful emotional
res pons e s such as terror, grief, and rage (Bowlby, 1969, 1973). In
turn, chronic childhood frustration of atta c h me n t needs may lead to
adult pronenes s to react with extr e m e anger (refe r r e d to in this paper
as "intimacy- anger") when relevant att ach m en t cues are pre s e n t.
Thus, att ach m en t theory suggests that an assaultiv e male's violent
Intimacy- anger and attachme nt
5
outburst may be a form of protest behavior directed at his attachme n t
figure (in this case, a sexual partner) and precipitate d by per c eived
threat s of sep a r ation or abandonmen t .
Intimacy- anger and attachme nt
6
Attachme n t and Adult Relationships
While attachment processes have been extensively investigated
in parent- child relationships, relativ ely less attention has been paid to
the role of attachmen t processes in adult relationships. However,
followin g the seminal work of Hazan and Shaver (1987), who
conceptualized rom a n tic love as an att ach m en t process, a growing
body of research has applied an attachme nt per s p e ctive to the study
of adult personal relation s hips. Hazan and Shav e r (1987) developed a
self- report measure to differentiate adult analog u e s of the three
infantile atta c h men t pattern s describ e d by Ainsworth , Blehar, Walte r s
and Wall (1978): secure, ambivale n t, and anxious- avoidant.
Subsequ e nt researc h has indicated that: 1) the qualit atively differ e nt
patter n s of attachmen t identified in childho o d are parallel to patter n s
that cha r a c t erize adults, 2) enduring sexual or romantic relationships
are the most importan t attach m ent relationships in adult life (Hazan
& Zeifman, 1993), and 3) adult attac hme n t pattern s are associated
with a ran g e of relationship outcomes including satisfaction, trust,
communication, and conflict resolution (see Bartholomew, 1993,
Hazan & Shave r , 1994, and Shaver & Hazan, 1993, for reviews) .
Until the present study, however, little application had been made of
an attachmen t perspe c t ive to the stu dy of adult pathology.
Building on Hazan and Shaver's (1987) work, Bartholomew
develo p e d a four- category model of adult atta c h men t that defin e s
attachm e nt patterns in terms of the intersection of positive or
neg a tive repr e sen t ations of the self and othe r s (Bart h olomew, 1990;
Bartholom e w & Horowitz, 1991). The self- model dimension indicat e s
Intimacy- anger and attachme nt
7
the degr ee to which individuals have inte r n a lized a sen s e of their own
self- wort h; thus, a neg a tive self- model is associated with excessive
anxiety and dependency in close relation s h i p s . The other mod el
dimension indicates the degree to which individ uals expect significant
others to be sup portive and trust worthy, and thus is associat e d with
the ten d e n cy to seek out or avoid intimacy in relationships (Griffin &
Bartholom e w, in press).
The secure pattern is defined in terms of a positive self-mod el
and positive other- model; secure individuals are both confident and
comfo rta b l e with intim acy in their close relationships. Ther efore, this
patter n is expected to be negatively associated with emotional
rea c tivity and abuse in intimate relationships. The dismissing
attachm e nt pattern is defined in terms of a positiv e self- model and
neg a tive other- model. Dismissin g individuals maintai n a positiv e self-
image by defensively downplaying the impor t a n ce of attachm e nt
nee d s and maintaining emotional distance in their relationships.
Although the frust r ation of attach m ent needs associated with this
style may give rise to an ge r (Kobak & Sceery, 1988), the dismissing
are not prone to insecurity in close relationships (pres uma b ly due to
the deac tivatio n of the attachme n t syste m) and, therefore, they should
not be especially pron e to angry prot e s t in intimate relationships.
In contrast, individu a ls showing the two attachmen t patterns
defin ed in ter ms of a negative self-model (the preoccupied and
fearful) are chronically anxious about rejection and abandon m ent in
their close relatio n s h ips. Therefore, thes e anxious attach m e nt
patter n s are expected to be asso ciated with high levels of negative
Intimacy- anger and attachme nt
8
affect, including anger, in intimat e relationships. Preoccupied
individuals (negativ e self-mod el an d positive other- model) actively
seek to gain their atta c h men t figure's appro val in order to valid ate
their tenuous sense of self- worth. Their feelings of unworthiness and
strong approach orientation to mee ting their needs is expected to be
associated with high levels of intimacy- ang e r .
But the fearful attachmen t pattern (negative self-model and
neg a tive other- model) may be most strongly associated with intimacy-
ang e r . Fearful individuals "desire social cont a c t and intimacy but
experience pervasive interpe r son al distrus t and fear of rejection"
(Bart holomew 19 90). This style manif ests in hypersensitivity to
rejection, and active avoidance of close relationships where
vulner a b ility to rejection exists. While the fearful sha r e anxie ty over
aba n d o n ment with the preoccupied, their avoidance orient a tion may
lead to mor e chron ic frustration of attach m e nt needs than that
experienced by the preocc u pi e d . Barth olomew (1990) argues that
strong and unresolv a ble approach/avoidance may underlie the
beh avior of chronic ally fearful ("anxious/avoid a n t ") peo ple: Perceived
threat s of aban d o n ment lead to tende ncies to approach the
attachm e nt figure who rejects physical conta c t , thus genera ti n g
withd rawal accompanied by an even stronger need for atta c h men t . A
self- perpetua t i n g feedback loop ensues that lead s to chronic
avoidance, frustration of attachmen t needs, and anger. Hence,
intimacy- ange r may be central to the fearful attachment style.
In sum, individuals with fearful and preoccupied attachmen t
patter n s are prone to the experi e n c e of anxiety about aba n d o nme n t in
Intimacy- anger and attachme nt
9
their intimate relationships and, ther efore, the s e attachme n t patterns
are expect e d to be positively relat e d to anger, jealou s y, and affective
instab ility. Dutton (in press) and Dutton and Starzomski (in press)
found that anger, jealousy and affective inst ability were all strongly
and significantly relat e d to frequency of verbal and physical abu s e in
intimate relationships. Hen ce, a link is also hypot h e s ized betwee n
anxious attac h men t patterns and a prop e n s ity for abu siveness- -
attachm e nt- ang e r and affective instability. We argue that
attachm e nt- ang e r (intimacy- ange r) and affective instability are
produced by chronically frustrate d attachme n t needs and are risk
facto r s for increased abusiveness in intimate relationships.
Wife Assault
Male violence tow ar d s women frequ e ntly occu r s in the context
of intimate relationships (Straus, Gelles & Steinmetz, 1980; Straus &
Gelles, 1990). With the exception of serial killers, almost all cases of
males killing fem ales occur in the context of an ongoing intimate
relations hip (Daly & Wilson, 1988; Cra wford & Gartner, 1992), and
much male intima t e violen ce occur s in the process of real or
per ceived relationship dissolution (Dutto n & Browning, 1988; Daly &
Wilson, 1988; Crawford & Gart n e r, 1992). Prior res e arch indicates
that men who assault their wives have high chronic anger scores
(Dutton, in press), and that th eir anger is frequently trig g e r ed by
attachm e nt change they perceive as uncontrollable . Dutton (1988)
describes these changes as "aba n d o n ments" and "engulfmen t s ," and
has found that these kind of changes produce the greate s t anger
rea c tions in male subje cts watching couple- conflict videotapes
Intimacy- anger and attachme nt
10
(Dutton & Browning, 1988). In their study of 551 femicides, Crawford
and Gartne r (1992) concluded "the offen d e r ' s anger or rage over the
actual or impe n ding estrang e ment from his partner typified 45% of
the cas e s where a motive could be establish e d " (p. 44). In anothe r
15% of cases, the motive was the offender's suspicions about his
par tn e r 's relationship with another man. It app e a rs, in other words,
that intimacy or attachm e nt issues gen e r ate strong reac tions that
include anger and violence toward intimate othe r s.
The thesis of the present study is that men who have anxious
attachm e nt styles will be more likely to repo r t chronic anger and
associated feat ur e s of wife abuse. In particular, we focus on
attachm e nt styles described as "fearful" (anxious- avoidant) and
"preocc u pied" (ambivalent) by Bartholomew (1990). Both styles
involve negative self-models and intimac y- anxiety and are
charac t e rized by frustrated attachmen t needs, subjectiv e distress,
and hyperse n sitivity. Both styles should cont ain high levels of
intimacy- anger.
Dutton (in pres s ) has demonstr a ted a battere r profile of
associated abuse featur e s that include a pers o n ality charac t e rized by
Borderline Personality Organization (BPO), some central features of
which are affective instability, anger and projection of unacceptable
impuls e s onto the devalued woman- object (cf. Gund e r s o n, 1984 ).
Dutton (in pres s ) reported high BPO scorer s as demonstra t i n g a
constellation of meas u res associate d with frequent use of verbal and
physical abuse including expe rienc e of trauma sympto m s , and grea t er
ang e r and jealo usy than low BPO scorers . The prese n t study will
Intimacy- anger and attachme nt
11
asses s associa tions of attachment styles to this constellation of
charac t e ristics associated with wife assault: anger, jealou sy and
chronic experie n c e of trauma sympto ms. Finally, this study will
asses s the relations hip of attachmen t styles to a compr ehe n sive and
direct mea s u r e of abusiveness in intimate relationships, the
Psychologic al Maltreatm e nt of Wome n Inventory (Tolma n, 1989).
Method
Testing and clinical asses sme n t was done on court- referred and
self- refe r r ed males interviewed by the Vancouver Assaultive
Hus b a n ds Project, the Victoria Family Violence Institute, and the
Burnaby Family Life Institute. One hundre d and twenty men wer e
asses s e d. A demogr a phically matched cont rol group of 40 union men
was also assessed. Further, 43 female par tn e r s of the assaultive
group and 33 partners of the control group provided data relatin g to
the men' s psychological abusiveness.
Only North American acculturized men were inclu de d in the
current sam ple to avoid interp ret a tive problems arising from
language difficulties or from different cultural norms. The
demog r aphic profile of the ass a ultive group is as follows: average age
= 35 (range 17 - 65), average level of educ a tion = grade 12, average
incom e was $34,285 (63% self- identified as blue collar, 37% as
white). Men in the tre at m ent groups repo r t e d an average
assa ultive n ess in the prior year of 5.87 (S.D. 6.22) acts of physical
violence against their wife; 56% were still living with the assault
victim.
Intimacy- anger and attachme nt
12
Contro l group men wer e cont a ct e d through their local union
rep r e s entative and asked to particip a t e in a study on family conflict.
This group report e d an average age of 35 (rang e 19-45), an averag e
level of educatio n of grade 12, and a mod al family income of $35,000;
65% wer e married, and the mean self-reporte d scor e for physical
assa ultive n ess against their part n e r in the prior year was 1.34 acts of
violence (S.D. = 3.4). These men were selected to provide a
demog r aphically matched, relatively non- violent, non- criminal sample
(although, as reported above, they did self- report some violence
towards their wives).
We assessed differential patterns of attachme n t, emotional
expression (anger and jealousy), pre s e nce of current traum a
symptoms, and a style of per son ality called Bord e rline Perso n ality
Organization for these two groups. Not all men completed all
asses s m ents. Hence, in some cases reported n's are less than the
total sampl e sizes.
Testing and assessme nt included:
1. Attachme n t Measur e s
Relation sh i p Styles Ques t i onn aire (RSQ) . The RSQ (Griffin
& Bartho lomew, 1994) is a 30- item self-report measur e with items
dra w n from Haza n and Shaver's (1987) attach m en t measur e ,
Bartholom e w and Horowitz's (1991) Relationship Questionnaire, as
well as items from Collins and Read's (1990) Adult Attachme n t Scale.
Mea s u r es of each of the four attachme n t patter n s (secu r e , fearful,
preoccupied, dismissi n g) identified by Bartholo m e w and Horowitz
(1991) were created by sum ming four or five items from the
Intimacy- anger and attachme nt
13
correspon d i n g prototypic desc riptions. The RSQ atta c h men t scores
show conve r gent validity with inte rview ratings of the four
attachm e nt patterns (Griffin & Bartholomew, 1994). Note that each
subject receives a contin uou s rating for eac h attachment pattern.
Relation sh i p Questi o nnair e (RQ) . Bartholo m ew and
Horowitz (1991) describe their RQ as an adaptation of the attachment
meas u re develop ed by Hazan and Shaver (1987). It consists of four
short para gra p hs describing the four attachm e nt patterns (secure,
fearful, preoccupied, dismis sing). Each responde n t is asked to mak e
ratings on a 7- point scal e of the degree to which they resemble each
of the four styles. The RQ attachme n t ratings show convergent
validity with interview ratings (Bartholome w & Horowitz, 1991;
Griffin & Bartholo m e w , 1994 ) and moderately high stability over eight
months (Scharfe & Bartholo m ew, 1994) . The RQ can eith e r be used
to generate a continuous rating for each attachme n t patter n or an
attachm e nt category, defined as the pattern with the highest rating
on the 7-point scales.
2) Borderline Person ality Organization (BPO) . The Self-
Report Instrume n t for Borderline Personali ty Organization (Oldha m et
al., 1985) is a 30- item instrument derived through factor analysis of a
130- item ques tionnaire designed by the authors. The three subsc ales
meas u re: 1) identity diffusion (a poorly integr a t ed sense of self or of
significant others); 2) primitive defenses (splitting, idealization,
devaluation, omnipotence, denial, projection and projective
identification.); and 3) reality testi n g (external versus intern al origin s
Intimacy- anger and attachme nt
14
of perce p t ions, evaluation of own behavior in terms of social crite ria
of reality, differentiation of self from non- self, etc.)
Oldh a m et al. (1985) report on the scale's intrascale
consiste n c y, interscale rela tionships and rela tionship to BPD
differential diagnosis, and the applica tion of the scale to differing
theories of BPO and its DSM- IIIR Axis 2 definition. Cronb ach's alpha
for the BPO subscales are identity diffusion .92, primitive defen s es .
87, and reality testin g .84. In our current sample, we obtained
Chronbach's alphas as follows: identity diffusion .85, primitive
defenses .87, and reality testing .80. Dutton (in pre s s) shows the BPO
scale to be significantly corr elat e d with frequency and sever ity of
violence in a wife assault sam ple, and Dutton and Sta rzomski (in
pre s s) demonstrate d men 's self-reports of BPO strongly and
significantly correlated with their wives' reports of the man's
psychological abusiveness.
3) Anger . The Multidimensional Anger Inventory (MAI; Siegel,
1986) is a 38- ite m self- report scale assessing the following
dimensions of ang e r respons e : frequen c y, durat ion, magnitud e , mode
of expression, hostile outlook, and range of anger- elicitin g situation s.
Siegel repor t s the resu lts of a factor analysis of this scale and the
relia bility of its subscal es (alphas = .51 to .83) and the scale as a
whole (alpha equal to .84 and .89 for two sepa rate samples). The
scale was validated by correlation with other, conceptually similar
ang e r inventories.
4) Jealousy . The Inter per s o nal Jealou sy Scale (Math es & Severa,
1981; Mathes, Phillips, Skowr a n & Dick, 1982) is a 28- item scale that
Intimacy- anger and attachme nt
15
meas u res romantic jealousy. Tests of the constr uct validity of this
meas u re have shown it to be correl at e d with dependency. The scale
has a high internal relia bility and a low correlation with social
desirability response bias.
5) Trauma Sympto ms (TSC- 33) . The Trauma Sympt o m s
Checklist (TSC- 33; Briere & Runtz, 1989) is a brief (33- item) reliable
instrument showing predictive and constr u c t validity. It has been
shown to disc riminate femal e victims of childhood sexual abuse from
non- victimized women. The TSC-33 contains five subscales:
Dissocia tion, Anxiety, Depression, Post- Sexual Abuse Trauma-
hypoth e sized, and Sleep Disturb a n ce. The PSAT-hypothesized
includes thos e sympt o m s thought to be most cha r a c t eristic of sexual
abu s e experiences but which may also occur as a result of other types
of traum a . Analysis of the internal consistency of the five subscales
indicated re ason a b l e reliability with an average subscale alpha of .71
and a total alpha for the TSC-33 of .89 (Briere & Runtz, 1989).
6) Psycholo g ical Abus e . Tolman's (1989) Psychologic al
Maltreatme nt of Women Invent o ry (PMWI) was used to provide a
more comp r e he n sive asses sme n t of abuse, sinc e psycholo gical abuse
is more common than physical abuse (Straus, Gelles & Steinmetz,
1980). It was completed by the women partner s , who reported the
men 's abusivene s s . The PMWI contains 58 items (rated from 1
"neve r" to 5 "very frequen tly") which compris e forms of
emotional/verbal abuse and dominanc e /isolatio n. Domina nc e /isolatio n
includes items related to rigid observanc e of traditional sex roles,
dem a n ds for subs e r vience , and isolation from resources. In contrast,
Intimacy- anger and attachme nt
16
emotional/verbal abuse includes withholdi ng emotional reso u r c es,
verbal attacks, and behavior that degrades women. Factor analyses
support the inclusion of the two factors. In the sampl e considere d in
this stu dy, Cron b a c h 's alph a ' s for the dominance/isol a tion subs c ale
was .82, and for the emotional/verbal subscale it was .93.
Procedur e
Men ente r in g the Vancouver Assaultive Husbands Progra m and
the Victoria Family Violence Proje ct were administered the above
scales either prior to tre a t m ent or during the first 3 wee ks of
treat m ent in order to minimize effects from treatmen t on test results.
Their part n ers were also administered questio nn a i r es, by mail, at that
time. Althou g h particip a tion was volunt a r y, cooperation was
encoura g ed by offering, and providing, a subject fee and, for the men,
individual feedback on que s tionnaire results. Subjects complete d the
questionnaires individually and retu r n ed the m to the experim e nter at
the next session (the men only) or by mail.
Contro l dat a was obtaine d by posting signs in two union locals
req u e s ting participation in exch a n ge for a subject fee. Subjects
app ro a c h ed thei r union repres e nt a tive or Family Violence Lab
per sonn e l to obtain a question n aire package; upon recei p t of the
completed questio n n aire in the lab the subje c t fee was provided.
Confidenti ality was assured. The men and women were asked
to not discuss their resp ective respons e s with each oth e r.
RESULTS
RSQ with RQ
Intimacy- anger and attachme nt
17
Accordi n g to RQ self-repor t s , 20% of the sample gro up of
assa ultive men was secu r e , 20% was fearful, 33% was preoccupied
and 27% was dismissi n g. Table 1 shows the correlations of the RSQ
and RQ continuous pattern ratin g s . The me asu r es demo n s t rat e a
moder a t e positive asso ciation ranging from .39 for secure attachment
to .65 for preoccupied. Correla tions of each attach ment meas u re with
other measures of this study follow similar patter n s . To avoid
red u n d ancy, only correlatio n s of the RSQ measure will be reported in
the tables to follow.
_________________________
Insert Table 1 About Her e
_________________________
Contro l Group vs. Assaultive Men
Table 2 demo n s t rat es linear compariso n s between the control
and treat men t groups. Scores on secure attachme n t were
significantly higher for cont rols. Scores on fearful and preoccu pied
attachm e nt were significantly high e r for tre a t men t group men. The
dismissing rating did not differ significantly by group. Treatment
group men also scored significa n tly higher on BPO total scores, MAI
(anger) total scor es, Jealousy and TSC- 33 (trauma symptoms) total
scores. In general, the ass a ultive group app e ars to be less securely
attache d and less affectively stable than the control group.
_________________________
Insert Table 2 About Her e
_________________________
Associa tions of Attachme n t to Other Measures
Intimacy- anger and attachme nt
18
In ord e r to ass e s s specific rela tionships bet w e en attachm e nt
style an d other depende n t measu r es, Pearson correlations were run
on the combined sample and are demo n s t rat ed in Table 3. The
strongest associations were demonstrat e d for scores on fearful
attachm e nt. These correlated significantly and positively with scores
of the BPO measure (+.58), and with anger (+.49), trauma sympto m s
(+. 5 0) and jealousy (+.34), all associated featu res of ab usiven ess
(Dutton, 1992). The preoccupied scores gen e r ated a similar, if
weake r patter n of associations. Scor e s for secur e attachm e nt, on the
other hand, corr elate d negatively and significantly with BPO (-.35),
ang e r (-.36), trauma sympto ms (-.28) and jealousy (-.16). Scores for
the dismissing pattern did not tend to corr elate significantly with
other measures . These same general patt e rns of association between
the RSQ and measur e s of associat ed abuse feat u r es replicate d for all
subscales of each measure.
Attachme n t and Abusiven ess
Finally, RSQ attach m ent patterns wer e correl a t e d with the
psychological abuse measure (PMWI). Fearful atta c h men t correlat ed
strongly and significant ly with scores on both the dominance/isolation
and emotional abuse factors; the pre occupied rating was also
significantly correlated with both abuse scal es; the secu r e rating was
neg a tively correlated with dominance/isola tion; an d dismis sing scor es
wer e unrelated to the abuse measure (see Table 3).
_________________________
Insert Table 3 About Her e
_________________________
Intimacy- anger and attachme nt
19
Linear Contras t s of Attachm e n t Styles
Anova and Tukey procedures were run on the att ach m en t
groups (deriv ed fro m the RQ) comparing them for anger, traum a
symptoms and jealousy (see Table 4). For anger, the s e yielde d
betwe e n- group significa n ce (F (3, 67) = 9.08 , p = .0001). As can be
see n from Table 4, the fearful group differed signific a ntly from the
secure an d dismis sing group s.
An Anova and Tukey proce d ure run on TSC-33 total scores also
yielded a significant between- atta c h men t group difference (F(3, 67) =
4.99, p = .0035). Again, the fearful group reported the high e s t
scores, which differed significantly from the secure group and the
dismissing grou p. This same pattern recurred for jealousy ( F (3, 67) =
4.89, p = .009) and BPO (F(3, 67) = 5.09, p = .003). For dominance,
(F(3, 33) = 5.1 9, p =.006), however, a Tukey revealed a significan t
difference between the secur e group and the fearful and preoccupied
groups. For emotion a l abuse (F(3, 33) = 4.35, p = .034), a Tukey
revealed that the secure and fearful groups were significantly
different.
_________________________
Insert Table 4 About Her e
_________________________
Hen c e, on BPO, MAI, jealousy and trauma symptoms the fearful
group reported the high e s t scores. On all these measu r es they
differed significantly from the sec ur e and dismiss ing groups, but not
from the preoccupied group, who repo r t ed scores in a weak er pattern
than the fearful group . On both ab use measu r es, the fearf ul group
Intimacy- anger and attachme nt
20
again repo r t e d the highes t scores but differed only from the secur e
group on emo tional abuse.
Using the above measure s of attachmen t , anger, and BPO, a
discriminant function analysis was performed on abuse scores. The
one- factor discrimina n t functio n corr e c tly discriminated 85.6% of men
into high and low abusiveness on Factor 1 (dominatio n/isol a tion) of
the PMWI and 87.5% on Factor 2 (emotional/v er b al abus e ) . Predictor
variables, in order of the stren g t h of their corr elation with the
discriminant function, were for dominance/isola tion: BPO total score
(.79), anger total score (.72), fearful attachment (.61), secur e
attachm e nt (-.52), preoccu pied attach m e nt (.45), and dismissi ng
attachm e nt (.24). This same patte rn was rep r o d u c ed, with minor
variation, for emotio nal/verbal abu s e . Used alone, the RSQ correctly
classified 78% of men on dominance/isolation and 80% on
emotional/verbal abuse. A multiple regression run on the abuse
variables selecte d fearful attachm e nt as having the stronges t
association with the dep e n d ent measur e s . Fearful att ach m en t alone
accounted for 25% of the variance in dominatio n/isolation and 30% of
the variance in emotional/v er b al abus e .
Composite Attachment Scores
In ord e r to est a blish whe th e r general anxiousness and
avoidance (apart from their repres e nta t i o n as specific attach m e nt
styles) were relat e d to our dependent measur e s , we developed
composit e attachm e nt scores following a procedur e used by Griffin
Intimacy- anger and attachme nt
21
and Bartholomew (in pres s ). Attachmen t scores for gen e r al
anxiousne s s were calcula t e d as: (fearful + preoccupied) - (secur e +
dismissing) = anxious. Attach ment scor e s for general avoidance wer e
calculated as: (fearful + dismissin g) - (secure + preoccupied) =
avoidant. These two dime n sions are par allel to the anxiety and
avoidance scales develope d by Simpson, Rholes and Nelligan (1992)
and corres p o n d to those dimensions that have repeate d ly been found
to underlie individu al differences in attachmen t style (see Griffin &
Bartholom e w, in press; Shaver & Hazan, 1993). The patte r n s of
correlatio n of these com posite atta c h men t scores with othe r test
meas u res were as follows: High scores on composite anxiousne s s
correlated significantly (p < .001 ) and positively with scores on BPO
(+. 5 7), ang e r (+. 3 7), trau m a symp to m s (+. 5 6), jealousy (+.50), and
psychological abuse (+.46 with dominance/isola tion and +. 5 5 with
emotional abuse). High composite scores on avoidance corr elate d
significantly (p < .001) and positively with these same mea s u res .
Discussion
The present resear c h found eviden c e tha t anxious attachmen t is
related to a variety of measur e s associat ed with a personality
constellation of abu siveness in intimate relationships and with a
direct mea s u r e of emotional abusiveness. Dutton (in press) found
that high scores on BPO, anger, curr e n tly expe rienc e d trauma
symptoms and jealousy corr elate d significantly with frequen cy of both
verbal and physical aggres s ion dire cte d towa r d women in intimate
relations hips. Dutton and Starzomski (in press) found that BPO and
ang e r scores accounte d for 50% of women's reports of their
Intimacy- anger and attachme nt
22
husban d s' emotional abus e. In the present study this constellation
was strongly associated with attac h me n t style, especially att ach m en t
charac t e rized as fearful and, to a lesser extent , with preoccupied
attachm e nt. Scores depicting either style correlated significantly
with scores on BPO, ang e r , trauma symptoms and jealousy, as well as
with measure of emotional abusiven e s s. Furtherm o re, composit e
attachm e nt scor e s of gener a l anxiousness and avoidanc e had stro ng
and significant relationships with BPO, anger, tr aum a symp tom s , and
jealousy.
Fearfully attached men exper ien c e high degrees of both chronic
anxiety (as measur e d by the TSC- 33) and anger (as measured by the
MAI). In addition, both trauma scores in general, and dissociation
scores in partic ular, were highest for the fearful group. Fearful
attachm e nt alone accounte d for significant proportions of varian c e in
both emotiona l abuse criterio n factors. Alexander (1990) speculated
that fearful adults "would be expec t e d to exhibit the most severe
disor d e rs of affect reg ulation, including PTSD and dissoci ation" (p.
190). Since anxiety (+.42) and anger (+.48) are both strongly
associated with fearful attachme nt, one could argue that an emo tional
template of intimacy- anxiety/anger is the central affective feat u r e of
the fearful attach m ent pattern. Thes e correla tions main t ain in the
control sampl e (fearful- anxiety +. 5 3, fea rful- anger +.52 ), sugg e s ting
that this emotional template does not only resid e within physically
abu sive men.
A promin e n t feat u r e of BPO is intimacy- ang e r (Gund e r son,
1984; Dutton, in press). The correl a tion of fea rful attac hme n t to BPO
Intimacy- anger and attachme nt
23
is so stron g (+.62) that one could argue BPO is a personality
rep r e s entation of this par ticular att ach m ent style (see also Mahler,
1971). Gund e r son (1984) desc ribe s BPOs as experiencing identity
diffusion which they attemp t (unsuccessfully) to sta bilize through
their primary relationship. However, their tenden cy to project
unaccep t able impulses onto their intimate partne r and to "split"
attachm e nt- objec ts produces inevit able "dyspho r ic stalemat e s" and
eventual relatio n s h i p conflict.
With the fearfully attached man, anger is an aspect of
attachm e nt indep e n de n t of what tra n s p ir e s interp e r son a lly, and when
that anger is experienced it is both blam e d and projected onto the
attachm e nt object (Dutton, in press), resulting in chronic anger with
that other. Scores on fearful attachm e nt corr elate d significantly with
the BPO sub scale for primitive defenses, which asse s s es the tende n c y
to split women into ideal and devalue d objects and to project angry
impuls e s onto the devalued woman- object. This affective "templat e"
sets the stage for intimate conflict and increas e s the risk for intimate
violence.
Preoccupied men also show e d significant cor relations with
features of an assaultive profile found by Dutton (in pres s ). Parallel
to the ambivalence and ang e r shown by anxious- ambivalent children
in the Strange Situ a tion (Ainsworth et al., 1978), assaultive male s also
demon s t r ate heightened abandonmen t anxiety (Dutton & Browning,
1988) and extrem e anger at "abando n m ent scenarios" which depicted
uncontrolla ble changes in intimacy gen e r a ted by the female. It is our
contention that anger accompa n i e s anxious attachme n t in general.
Intimacy- anger and attachme nt
24
A Developmental Origin of Intimacy- Anger
Although the focus of recen t resea r c h in attach m en t has been
on anxiety, muc h of the early focus of Bowlby and Ainsworth was on
what could be called attachme n t- anger or attachmen t - rag e. Bowlby
(1969, 1973) and Ainswo r t h (Ainsworth et al., 1978) indicated that
whe n attachmen t needs are activ ated for a long time without being
satisfied, angry behavior is regularly obse rv e d in the infant. Bowlby
reporte d observ a tions of the reactions of childr e n (age 15- 30 months)
in nurseries who were sepa r ate d for the first time from their par e n t s.
The actions associated with the first phase of the separa t ion cycle,
which he called "protes t", wer e genera t e d agentically (outward on the
world) in order to produce the return of the mother. It is instructive
to hear Bowlby's own desc ription of these reactions:
"...(in) the initial phase (Protest), the young child appears
acutely distress e d at having lost his mother an d seeks to
reca p t u re her by the full exercis e of his limited resourc e s .
He will often cry loudly, shake his cot, throw himself
about, and look eagerly towards any sight or sound which
might prove to be his missing mother. All his behavior
sugges t s strong expe ct a tion that she will ret u r n " (196 9, p.
27- 38) .
It is only after prolonged failure to have age n tic actions lead to
a successful recrea tion of the mother's presence that the subsequ e nt
emotions of depre s s i on (mour n in g) and eventual detachm e nt
(avoidance) appear.
Intimacy- anger and attachme nt
25
Avoidant children, the poten t ial develop m ent a l precurs o r to
fearful adults, experien c e rejec ting parenting and are cha r a c terized
by avoidanc e of the car e t a k er after separation in a controlled
laboratory setting (Ainsworth et al., 1978). Some time afterward,
however, avoidant children direct considerable anger toward their
mothers at home. Main and Westo n (198 2) also note d unpredictable
outbursts of aggre s s io n by this group toward their mothe r s in the
home setting. And the gre a t er the avoidance upon reu nion with
mother, the greater the display of anger an d depen den t behavio r
towards her over the ensuing week s. Main and Weston (1982)
sugges t that in response to separation, avoidant infants feel ang ry
with their mother, but the expressio n of anger in this situation risks
decre a s ed proximity, so chronically rejected infants only express
ang e r in circumsta n ces that do not risk decreas e d proximity from the
caregiver.
In contrast, Main and Weston repo r t that an anxious- ambivalent
(or preo ccu pie d) group beca m e extre mely agitated on sep a r ation,
ofte n crying profusely. This group typically soug h t conta c t with their
mother when she returned, but simultaneou s ly arch ed away from her
ang rily and resisted all efforts to be soothe d. The implica tion is that
these infants somehow incorporat e anger into their terror at bein g
aba n d o n ed by the mother. The simultaneous seeking of contact and
arching away represent s the prototypical physical manifestation of
ambival e n c e . We include this material on infant attachmen t as a
reminder that rage res po n s es may have early origins in par e n t- child
attachm e nt relationships.
Intimacy- anger and attachme nt
26
Data from othe r sources suppo r t th e notion tha t actua l
par e n t in g practices shape childhood att a c h m ent pattern s (e.g., Main
& Statdman, 1981; Crittende n , 1988) and that there are
intergener ation al continuities in paren t- child attach m ent patterns
(e.g., Fonagy, Steel e, & Steele, 1991; Steele & Steele, 1994; Van
IJzendo or n , 1992). What we ar e suggesting is an alterna tive link to a
modelling mec h a n i s m for the well-known intergene r ation a l
trans mission of violenc e (Straus, Gelles, & Steinmetz 1980). That link
is anxious attachme n t patterns which manifest anger, jealousy,
anxiety and abusive beh avior in intimate relationships.
Both dominance/isola tio n and emotional/verbal abuse, as
meas u re d by the PMWI, wer e related significantly to fearful and
preoccupied attachm e nt. One could speculate that emotional abu s e is
a product of attachmen t- rage. The anxious ly attached ma n, una w a re
that his dysphoria is intimacy produced, attribu t es it to real or
per ceived actions of his part n e r, and retaliat e s with abusivenes s . The
dominance/isolation factor can be viewed as an overgene r al ized
attem p t to diminish anxiety about aba n d o n me n t. The yoked dysphoric
modes- -ang e r and anxiety- -thus may have a common origin in insecur e
attachm e nt and operate to generate both abusin g and controlling
beh aviors.
Anger increas e s the probability of aggression (Konecni, 1975)
and assaul tive men with high anger scores report greater frequency
of physical assault (Dutton, in press). Men with early atta c h men t
problems may be more likely to exper ien c e anxiety and ang e r about
intimacy regulation. The arousal, anxiety and ang e r these men
Intimacy- anger and attachme nt
27
exhibit in intimate relationships may originate in a felt anxiety about
their childhood attachme n t figure s , and ceteris paribus, increase the
probability of aggression tow a r d the intimate partne r .
It may appear discomfiting, prima facie, that anxious
attachm e nt patterns could be a risk factor for abus e, as cont e m pora ry
abu s e treatm e nt models focus mor e on discrete cognitions or
"controlling beh aviors" (Saund e r s, 1989; Ganley, 198 9; Dutton &
McGregor, 199 1). Ther e are two potential treatment applica tions of
the above analysis. First, it app e a r s therap e utically important to
sep a r a te dysphoria produced by anxiou s attachment itself from
neg a tive attributions to the attachme nt- figure. Seco n dly,
under s t an d i n g attachmen t anxiety can potentially provide a better
under s t an d i n g of "control" issu es by providing insight into the origins
of intimacy fear. What is controlle d in intimate relationships is that
which is most feared: namely, the deg r e e of intimacy or emotional
distan ce from the atta c h m ent- other. Learning to negotiat e emotional
distan ce with gre a t er awa r e nes s seems therape u tically ben eficial.
Intimacy- anger and attachme nt
28
References
Ainswor t h , M.D.S., Bleh a r, M.C., Waters, E., & Wall, S. (1978).
Pattern s of attachmen t : A psychological study of th e Strang e
Situa tion . Hillsdale, N.J.: Erlbaum.
Alexande r , P.C. (1992). Applicatio n of attach ment theory to the study
of sexual abuse. Journal of Consulting and Clinical Psychology ,
60 (2), 185- 195.
Bartholom e w, K. (1990) . Avoidance of intim a cy: An atta c h men t
per s p e c tive. Journal of Social and Person al Relationships , 7 ,
147- 178 .
Bartholom e w, K. (1993) . From childho od to adult relationshi p s :
Attachme n t theory and res e a r ch. In S. W. Duck (Ed.),
Underst a ndin g relatio n s hip processes 2: Learning abou t
relations hips. (pp. 30- 62) London: Sage.
Bartholom e w, K., & Horowitz, L.M. (1991). Attachm e nt styles amo n g
young adults: A test of a four- category model. Journal of
Personality and Social Psychology , 61 (2), 226- 244.
Bowlby, J. (1969). Attachme n t and loss: Vol. I. Attachment . New
York: Basic Books.
Bowlby, J. (1973). Attach m e nt and loss: Volume II. Separation . New
York: Basic Books.
Bowlby, J. (1977). The makin g and br eaking of affectional bonds.
British Journal of Psychiatry , 13