This article was downloaded by: [Universitaets und Landesbibliothek]

On: 24 November 2013, At: 03:53

Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer
House, 37-41 Mortimer Street, London W1T 3JH, UK

Australian Journal of Psychology

Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/tajp20

Attachment style and romantic love: Relationship

dissolution
a a
Judith A. Feeney & Patricia Noller
a
University of Queensland ,
Published online: 27 Sep 2007.

To cite this article: Judith A. Feeney & Patricia Noller (1992) Attachment style and romantic love: Relationship
dissolution, Australian Journal of Psychology, 44:2, 69-74

To link to this article: http://dx.doi.org/10.1080/00049539208260145

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”)
contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors
make no representations or warranties whatsoever as to the accuracy, completeness, or suitability
for any purpose of the Content. Any opinions and views expressed in this publication are the opinions
and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of
the Content should not be relied upon and should be independently verified with primary sources of
information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands,
costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or
indirectly in connection with, in relation to or arising out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Any substantial or
systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in
any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
 69

Attachment Style and Romantic Love: Relationship Dissolution

Judith A. Feeney and Patricia Noller

University of Queensland

The present research was designed to assess attachment style differences in subjects' reports of the dism
lution of romantic relationships. One hundred and ninety-three unmanied undergraduates completed
questionnain measures of relationship history. including relationship dissolution and attachment style
(both the forced-choict measure used by Hazan Bt Shaver, 1987; and a Likert-type version derived from
that masurt).Subjects were also followed up 10 weeks later in order to assess the stability of attach
mcnt style and its relation to the formation and dissolution of relationships. As expected, the three attach
mcnt styles diffed strongly in their affective responses to disrolution. Measures of attachment style and
mental models were also related to the Occurrence of relationship termination; in particular, subjects
endorsing the avoidant attachment style were the most likely to experience relationship break-up.
Change in self-reported attachment style was associated with the formation of a steady relationship dur-
ing the course of the study, but not with the experience of relationship break-up.
Downloaded by [Universitaets und Landesbibliothek] at 03:53 24 November 2013

The seminal paper by Hazan and Shaver (1987) applied the aspects of the process; for example, perceptions of who initiated
principles of attachment theory to adult romantic relationships. the break, emotional reactions, and postdissolution dating
Subsequent research has provided considerable support for the behaviour.
attachment perspective on romantic love: specifically. Secure Specifically. since avoidant subjects are low on both commit-
attachment is associated with positive relationship characteristics ment and satisfaction (key determinants of relationship termina-
such as satisfaction and intimacy; avoidant attachment is related tion according to Johnson. 1982). they may be more likely to
to less committed and intimate relationships; anxiouslambivalent initiate termination. In addition, given their avoidance of intima-
attachment is associated with high levels of passion and preoccu- cy. they may experience little distress in response to the dissolu-
pation with relationships. but with low relationship satisfaction tion process and hesitate to seek further romantic involvements.
(Collins & Read. 1990; Feeney & Noller, 1990; Hendrick & In line with this suggestion, Simpson (1990) has reported that
Hendrick, 1989; Levy & Davis, 1988; Simpson. 1990). Evidence males scoring high on avoidant attachment report less emotional
for the salience of attachment style to perceptions of dating rela- distms following relationship dissolution. The clinging. demand-
tionships has been provided by a study in which subjects s u p ing, and emotionally volatile style of anxioudambivalent sub-
plied open-ended descriptions of their romantic partners (Feeney jects, on the other hand, may be associated with low frequency of
& Noller. 1991); a considerable proportion of the content of these initiating termination and high levels of distress at relationship
descriptions was devoted to the discussion of attachment-related dissolution. Given the association between anxioudambivalent
issues. and a questionnaire measure of attachment style was relat- attachment and reports of the frequency and intensity of love
ed in theoretically expected ways to subjects' spontaneous refer- experiences (Feeney & Noller. 1990. Hazan & Shaver. 1987). it
ences to attachment issues. is also likely that subjects endorsing this style will quickly seek
Research into adult attachment has focused also on the concept new love relationships.
of mental models. or internal working models of relationships These predictions. which are made from the perspective of
and relationship partmrs. Attachment theory suggests that, on the attachment style differences in closeness, interdependence, and
basis of early experiences with social interaction partners, the commitment. are also consistent with theory linking attachment
individual develops mental models which incorporate hidher style to affect regulation. According to a number of researchers,
expectations about the availability of attachment figures. Mental attachment styles reflect rules which guide individuals' responses
models of relationships are of central importance to attachment to emotionally distressing situations (Kobak & Sceery. 1988;
theory, because they are thought to act as a source of continuity Sroufe & Waters, 1977). Specifically, secure attachment is
between early attachment experiences and later feelings and thought to reflect rules which allow acknowledgement of distress
behaviours (Bowlby. 1973). At the same time. it has been recog- and turning to others for support; avoidant attachment reflects
nised that working models of relationships may be revised within rules which restrict the acknowledgement of distress and the
fie context of new relationships; in particular, when there is a seeking of comfort and support; anxiouslambivalent attachment
lack of fit between working models and actual social inter- is marked by a hypervigilance towards negative affect and
changes, the individual will begin to accommodate the models to towards attachment figures, in a way that inhibits the develop
reality (Hazan &Shaver, 1987; Ricks, 1985; Sroufe & Fleeson, ment of autonomy (Kobak & Sceery, 1988). The concept -of
1986). Studies of adult samples indicate that attachment group affect regulation may be useful in explaining attachment style
differences in the endorsement of mental models are consistent differences in response to relationship dissolution; it is notewor-
with predictions based on attachment theory: secure subjects gen- thy in this regard that avoidant males are characteriscd by low
enlly describe others as well-intentioned and trustworthy; anx- emotional distress at termination, even after differences in inter-
ioudambivalent subjects see themselves as misunderstood and dependence, commitment, and so on are controlled for (Simpson,
underappreciated; avoidant subjects describe themselves as self- 1990). supporting the possible role of affect regulation.
sufficient; (Collins & Read, 1990; Feeney & Noller, 1990; Hazan While attachment style differences in emotional distress are of
& Shaver, 1987). interest to attachment researchers, other affective responses to
The influence of attachment style on romantic love has been relationship dissolution also need to be considered. Then is some
assessed primiuily in terms of relationship qudity, and important evidence to suggest that anxioudambivalent subjects may be
issues remain to be addressed concerning the implications of poor at monitoring the state of their romantic relationships. For
attachment style for the experience of relationship dissolution. example, they tend to obsess over conflict and to report relatively
The process of relationship termination comprises a complex low relationship quality, yet they also paint an idealistic picture
series or steps and decisions (Baxter. 1984. 1985; Duck, 1982). of their partner and their relationship (Feeney. 1990. Feeney &
and attachment style differences may be reflected in several Noller, 1990. 1991). Moreover, the central characteristics of the

Requests for reprints should be sent to Judith A Feeney, Department of Psychology, University of Queensland, Queensbnd, 4072 Australia.

Australian Journal of Psychology Vol. 44, No. 2, 1992 pp. 69-74

 70 Judith A. Feeney and Patricia Noller

anxioudambivalent style are similar to those of the “desperate lution reporting on their experience; in addition, a prospective
love” style (Sperling. 1985), which is defined in part by difficulty arm of the study was designed whereby subjects currently in dat-
with intcrpcnonal reality testing. Hence anxiouslambivalent sub- ing relationships were followed up 10 weeks later, so that attach-
jects may report surprise at the break-up of their relationships, ment measures of those whose dating relationships were
indicating a lack of preparedness for the break. maintained and those whose relationships broke up could be
Attachment style may influence not only the process of rela- compared across time.
tionship dissolution. but also the stability of romantic relation- The following hypothtses were derived. based on the literature
ships: that is, the likelihood of relationships continuing or on adult attachment
terminating. Overall. only a minority of romantic relationships Hypotheses. It was expected that avoidant subjects would per-
progress to bccome f d y established. Cross-sectional stdies in ceive relationship break-up as initiated by the self, and report rel-
which adult subjects report on their adult attachment style and the atively high levels of relief at relationship termination and low
kngth of their current or their longest romantic relationship have levels of subsequent romantic involvement It was predicted that
shown a link between securc attachment and longevity of rela- anxidambivalent subjects, on the other hand, would be unlike-
tionships (Feeney & Noller, 1990; H m & Shaver, 1987). A ly to pcrceive that they had initiated the break, would experience
prospective study in which dating couples completed measurn of the most surprise and upset at relationship termination, and
attachment style and were interviewed 6 months later about the would report becoming involved in new romantic relationships
state of their relationship indicated that anxidambivalent males with little delay. No specific predictions were made concerning
and avoidant females showed the highest rates of relationship the link between attachment style and the actual occurrence of
break-up (Davis & Kirkpatrick. 1990). Given these results, i t relationship break-up, given the young sample being studied. In
Downloaded by [Universitaets und Landesbibliothek] at 03:53 24 November 2013

apptars that relationship termination should generally occur more addition, no predictions were made concerning changes in self-
frequently for the insecure attachment groups. For young subjects reported attachment style in response to relationship dissolution.
who are still experimenting with relationships, however, this
effect might be relatively weak. METHOD
While there are sound theoretical reasons for proposing that Subjects
attachment style predicts the likelihood of relationship dissolu- The total sample comprised 193 unmarried undergraduates (71
tion, it is also possible that subjects’ experience of relationship males and 122 females) who participated in the study for course
events (such as dissolution) may influence perceptions of attach- credit. All subjects completed measures of attachment style and
ment style. As noted earlier. attachment theorists have recognised relationship status. which were used to identify two groups of
that attachment style and working models of attachment arc mod- particular interest: those who reported having recently experi-
ifiable by relationship experiences, since such experiences may enced the breakup of a romantic relationship, and those who were
serve to disconfirm working models of attachment based on earli- currently in a dating relationship (and hence could be followed
er interactions (Hazan & Shaver, 1987: Main, Kaplan. & Cas- over time in order to assess the link between attachment style and
sidy, 1985). The influence of relationship events on perceptions relationship stability). Subjects ranged in age from 17 to 22
of attachment style may be particularly strong in young subjects. years. with a mean of 17.9 and a median of 18.14 years.
for whom romantic involvement is a relatively new experience
and whose modcis of romantic relationships arc in a state of flux Memum and Procedure
(Bowlby, 1979 Hazan, 1989). Hence thc impact of relationship All subjects completed the following questionnaire measures at
dissolution on self-reports of attachment style requires investiga- Time 1:
tion. 1. Attachment style was assessed using the forced-choice mea-
sure of Hazan and Shaver (1987). This measure comprises simple
The Present Research descriptions of secure. avoidant. and anxiouslambivalent adult
The present research was designed to clarify the relationship attachment styles. with item content bascd on extrapolation from
between attachment style and the process of relationship dissolu- the features of the comsponding infant attachment styles; sub-
tion. Two broad issues were addressed. Firstly, attachment style jects are required to select the description most applicable to their
differences in the experience of relationship dissolution were feelings about close relationships. A variant of the forcedchoice
assessed (percepUons of who initiated the brcak. affective reac- measure was also used. whereby subjects rate the applicability of
tions. and subsequent dating arrangements). Secondly. the study all 15 statements comprising the forced-choice measure. on a 4-
assessed whether the Occurrence of relationship dissolution in point scale, ranging from I = strongly disugree to 4 = srrongly
young subjects could be related to measures of attachment style agree. and responses to the three groups of items an summed to
and mental models. The research design was primarily refmspec- form scores o n three subscales: secure, avoidant, and
tive, with subjects who had experienced recent relationship disso- anxioudambivaknt attachment (seeTable 1).
Table 1 1M e m Measure of Attachment Style
Item Original Description
1. I find it difficult to depend on others. avoidant
2. Sometimes people arc s c d away by my wanting to be too close to them. ambivalent
3. Love partnc~often want me to be more intimate than I feel comfortable being. avoidant
4. I am nervous when anyone gets too close. avoidant
5. I find that others arc reluctant to get as close as I would like. ambivalent
6. I often worry that my partner won’t want to stay with me. ambivalent
7. I feel comfortable having other people depend on me. stcurt
8. I am somewhat uncomfortable being close to others. avoidant
9. I find it relatively easy to get close to others. secure
10. I find it easy to rrust others. avoidant*
11. I feel comfortable depending on other p p l e . securc
12. I don’t often wony about someone getting too close to me. secure
13. I ohen worry that my partner doesn’t really love me. ambivalent
14. I want to merge completely with another person. ambivalent
15. I don’t often wony about being abandoned. secure
Note. * indicates reverse scoring,

Australian Journal of Psychology Vol. 44,No. 2,1992 pp. 69-74

 Attachment Style and Romantic Love: Relationship Dissolution 71

2. Mental models of closc relationships were assessed by the on the other hand, was negatively correlated with ratings of upset
IGitem Risk in Intimacy Inventory (Pilkington & Richardson, and positively correlated with ratings of relief.
1988). which taps perceptions of the negative consequences of Attachment style differences were also investigated for items
close relationships. and 10 items from the Spbcific l n t e r p c r s ~ ~ lassessing subjects’ dating behavior subsequent to the mcnt ter-
Trust Scale (Johnson-George & Swap, 1982). selected to repre- mination of the previous relationship. No effect of attachment
sent the factors of emotional trust and reliableness (since these style was obtained for cumnt dating arrangements (not dating;
factors were reported by the authors to emerge for both male and dating more than one partner; in an exclusive dating relation-
female subjects). All these items employed a 6point Liken-type ship). A significant effect was obtained, however, for subjects’
format, ranging from 1 = strongly disagree to 6 = strongly agree. reports of whether they were in love at the time of the study,
3. Relationship history items assessing past romantic involve- x’(2) = 6.73. p c .05 (see Table 4). Anxious/ambivalent subjects
ments (previous dating experience, previous sexual involvement, were the most likely to report being in love (38.5%). and
number of timcs in love), and cumnt dating arrangements. avoidants the least likely (6.9%) (although as noted earlier, the
4. Experiences of relationship dissolution: subjects were asked attachment groups did not differ in the time elapsed since disso-
to indicate who had initiated the break (self/ptncr/mutual agree- lution of the previous relationship).
ment). and also to rate their affective nactions of surprise. upset,
and relief on Liken-type scales, ranging from 1 = not af dl to 5 = Attachment SccJcs and Relationship Dissolution - Retrospec-
utremly. tive
The questionnaires were administered in different ordcrs and The forcedchoice measure of attachment style was not reliably
were completed in group sessions lasting approximately half an
Downloaded by [Universitaets und Landesbibliothek] at 03:53 24 November 2013

related to the frequency of reports of recent relationship termina-

hour.
tion (either for the total sample, or when subjects with no recent
Stability of attachment style and its association with relation-
ship dissolution were assessed by asking subjects, 10 weeks later, involvement in dating relationships were excluded).
to respond for a second time to the measures of attachment style In order to see whether subjects experiencing relationship dis-
and to report on their cumnt dating arrangements. solution could be reliably differentiated from other subjects
reporting recent involvement in dating relationships, a discrimi-
RESZnTS nant analysis was performed. The predictor variables were attach-
Frequency of Athchment Soles ment style (defined by ratings on the three attachment scales
When presented with the forcedchoice mesure of attachment measuring secure, avoidant, and anxiouslambivalent attachment)
style, 54% (N= 105) of subjects endorsed the sccufe description, and mental models of relationships (perceived risk in intimacy;
34% (N= 65) endorsed the avoidant description, and 12% (N = interpersonal trust). (Initial analysis using MANOVA revealed no
23) saw themselves as anxiouslambivalent. No relationship was effects involving sex. Alpha coefficients for the predictor van-
observed between sex and self-repod akachment style.

Erpericnce of Rehtbnshb Dissolution Table 2 Attachment Style Differences in Affective Reactions to

Of the total sample. 82 subjects (39 secures, 30 avoidants. and Relationship Dissolution
13 anxiouslambivalents) r e p o d having recently left a relation- ~~

ship. (One avoidant subject did not answer all questions pertain- Item Mean F(2.78)
ing to relationship dissolution, hence some analyscs art based on secure avoidant ambivalent
8 I subjects.) For those subjects reporting relationship dissolution, surprise 1.79,b 1.31, 2.08b 3.46*
the forcedchoice measure of attachment style was not reliably upset 2.69, 2.37, 3.85b 6.06**
related to the number of months lapsed since the dissolution (on relief 2.28,b 2.83, 1.54b 5.69**
average, 4.1 months had elapsed). or to the level of sexual
involvement in that retation2hip. Note. I = strongly disagree; 5 = strongly ogree. Within each row,
Only 22% of subjects reporting relationship dissolution stated means with different subscripts differ at least at the .05 level of
that their partner had initiated the brcak; the remaining responses significance according to a Tukey B test.
were evenly divided between the categories of own decision and * p < .05, ** p c .01
mutual decision. There was a significant relationship between sex
and perceptions of who initiated the break, ~ ‘ ( 2=) 6.48, p c .05;
the most common perception among the female subjects was that
the break-up was a mutual decision (48%). whereas the most Table 3 Correlations Between Attachment Scales and Ratings
common response of the male subjects was to attribute the bmak- of Affective Reactions
up to the self (50%). In other words, males were most likely to
Affective reaction Affachment scale
report that they had initiated the break.
Contrary to expectation. there was no association between Secure Avoidant Ambivalent
surprise .I2 -.I7 .29**
~

attachment stylc’and perceptions of whether the break-up was

initiated by self. partner, or mutual agreement. It is noteworthy, Upset .I I -.23* .36***
however. that 38% of anxiouslambivalent subjects attributed the Relief -.01 .24* -.40***
initiation of the break-up entirely to the parmer, as expected. the
corresponding percentages for sccurc and avoidant subjects were Note. p < .05, ** p < .01, *** p < .001 (2-tailed tests)
somewhat lower (20% and 174, rcspcctively).
For all three items tapping affective reactions to relationship
dissolution, analyses of variance indicated significant attachment Table 4 frequencies of Selfreports of Being in Love (following
style differences as shown in Table 2. (Two-way analyses of recent relationship dissolution) by Attachment Style
variance revealed no main or interaction effects involving sex as
a independent variable.) Anxiouslambivalent subjects reported Item Attachment Style
the highest degree of upset about the break-up of the relationship; in love now? secure avoidant ambivalent
they also reported more surprise and less relief than the avoidant Yes 11 (28.2) 2 (6.9) 5 (38.5)
subjects, These results are supported by correlational analyses no 28 (71.8) 27 (93.1) a (6 I .5)
relating scores on the three attachment scales to the ratings of Total 39 29 13
affective reactions (see Table 3). Anxioudambivalent attachment
was positively correlated with ratings of surprise and upset. and Note. Figures in parentheses indicate percentages.
negatively comlated with ratings of relief; avoidant attachment,
Australian Journal a/ Psychology Vd. 44,No. 2,1992 00.69-74
 72 Judith A. Feeney and Patricia NoUer

abies exceeded .73. except for the secure and anxioudambivalent t u s (multivariate F(3. 87) = 3.25. p c .05). Univariate F tests
attachment scales. Items 7 and 14 respectively were dropped showed that this effect was attributable to scores on avoidant
from these scales in order to improve the internal consistency. attachment: at Time I. the mean score on this scale for subjects
resulting in alpha coefficients of .70 [secure] and .73 whose relationships subsequently broke up was 12.18, compared
[anxiouslambivalen t].) with 10.48 for the remaining subjects; at T im 2. group means on
As attachment theory would predict. the measures of attach- avoidant attachment werc 1 2 2 4 and 10.48 respectively. These
ment and mental models of relationships tended to be reliably results support the notion that insecurity of attachment predates
cwnlated: secure attachment was negatively related to Risk in relationship difficulties, atha than stemming from them.
Intimacy ( r = -.52) and positively related to emotional trust (.37) I Although the multivariate test of the interaction between sex
and reliableness (.a); avoidant attachment was positively related and relationship status was not significant, the univariate interac-
to Risk in Intimacy ( r = .65) and negatively related to emotional tion was also significant for avoidant attachment, F(1. 89) =
trust ( r = -.36) and reliableness ( r = -.a); anxioudambivalent 4.31, p c -05. The cell means for this interaction indicate that
attachment was unrelated to Risk in Intimacy. but negatively females whose rclationships broke up during the coursc of the
related to emotional trust (-.26) and reliableness ( r = -.28). The study obtained higher scores on avoidant attachment than all
overall association between the two groups and the predictors other groups. (Averaged over time. the mcan score of this group
was highly reliable (Wilks A = .91. ~ ' ( 6 )= 15.06. p c .02).The was 1355. compared with a mcan of 11.00 for males whose rela-
single discriminant function was defined by five of the six pre- tionships broke up, and means of 10.41 and 10.59 for males and
dictor variables, as shown in Table 5 (adopting a cut-off of .40 females whwc relationships continued.)
for salient correlations with the function): the univariate F tests Although the results of the MANOVA suggest that avoidant
Downloaded by [Universitaets und Landesbibliothek] at 03:53 24 November 2013

indicated that subjects who had experienced relationship termina- attachment predates relationship instability. rather than stemming
tion reported higher levels of perceived risk in intimacy and from such instability. interpretation of thest data is problematic
avoidant attachment. and lower levels of m u r e attachment and without reference to previous rclationship history. Specifically. it
trust (both emotional trust and reliableness). No group differ- could be argued that for subjects in the prtsent study, high levels
ences were obtained for anxioudambivalent attachment. of avoidant attachment in those whost relationships subsequently
broke up might have been caused by earlier experiences of rela-
Attachment Style and Rehionship History - Prospective tionship instability.
Stability of attachment memures -full sample. Measures of In order to assess the viability of this argument, we carried out
attachment style and dating status were completed a second time a supplementary discriminant analysis for the 93 subjects in dat-
by 172 subjects (a response rate of 89%). In terms of the forced- ing relationships at Time 1, in which scores on the thm attach-
choice measure of attachment completed at Time I , returns were ment scales (at Time 1) and the reported number of love
obtained from 9 4 secure subjects, 59 avoidants. and 19 experiences were entered as discriminating variables. and rela-
anxioushmbivalents. Responses to the forced-choice measure tionship status (remained together versus broken up) was the
were stable across the ]@week period for 129 (75%) of the sub- grouping variable. (Reported number of love experiences pro-
jects. 'Ihe probability of changing the original endorsement of the vides a rough index of prestudy relationship instability. especial-
item was lowest for the secure subjects and highest for the anx- ly given the young age of the subject sample; more direct
ious/ambivalent group: with attachment style defined by the measures of previous relationship breakdown were not avail-
response at Time I , endorsement of the attachment item at Time able.)
2 was different for 19% of secure subjects, 27% of avoidants. and Although the overall group discrimination was statistically sig-
47% of anxioudambivalents. Of those subjects who saw them- nificant (Wilks A = .89. ~ ' ( 4 )= 10.15. p c .05), thc single dis-
selves as secure at Time I . 19% endorsed one of the insecure criminant function was defined only by a positive correlation ( r =
descriptions at Time 2; of those who described themselves as .91) with avoidant attachment. and a negative correlation ( r =
insecure at Time I , 28% changed their response to secure attach- -.48) with secure attachment. Furthermore. for this sample, the
ment at Time 2. Comlations were formed, for each attachment bivariate correlation between avoidant attachment (Time 1 ) and
scale. between subjects' scores at the two measurement times. number of times in love was actually negative, although not sta-
The obtained correlations were 5 7 (secure). .73 (avoidant), and tistically significant ( r = -.16). This analysis provides some sup-
.70 (anxioudambivalent). port f o r the proposition that the important predictor of
Atrachment scales and relationship history at Emes 1 and 2. relationship outcome is avoidant attachment per se, rather than
Of the 93 subjects who reported being in a dating relationship at previous relationship instability.
Time 1,75 remained in this relationship 10 weeks later. while 18 Attachment (forced-choice)and relatiomhip history at Times 1
had experienced subsequent relationship dissolution. A MANO- und 2. In ordcr to investigate possible influences on the stability
VA was carried out to investigate the effects of time, sex. and of attachment style, responses to the fomdchoice measure and
relationship status (remained together versus broken up) on patterns of dating arrangements (not dating at all. dating more
scores on the three attachment scales. (Measures of risk and trust than one partner, or in an exclusive. steady dating relationship)
were not available at Time 2.) were compared across the two occasions. A change in dating
The results showed a significant main effect of relationship sta- arrangements from Time 1 to Time 2 was reported by 47 sub-

Table 5 Discriminant Analysis of Attachment and Mental Model Measures by Relationship Termination
Scale Coml. 4 2 . 190) Means
Relationship termination
Yes No
Risk in Intimacy .92 13.55*** 29.83 25.10
Avoidant attachment .72 8.34" 11.74 10.52
Secure attachment -.7 I 8.Wf* 10.64 11.33
Emotional Trust -.63 6.47* 21.33 23.05
Reliableness -.58 5.41 23.29 24.78
Anxioudambivalent attachment 32 4.24 9.13 8.69

Note. Possible nnges of scores: for risk in intimacy. 1 W . avoidant attachmenb 5-20; emotional trust and reliableness. 5-30; secure
and anxioudambivalent attachment. 4-16.
* p < .05. * * p < .01, * * * p < ,001
Australian Journal of PsychologyVol. 44,No. 2.1992 pp. 6 S 7 4
 Attachment Style and Romantic Love: Relationship Dissolution 73

jects, 14 of whom also changed their self-reported attachment ious. clinging relationship style, and the poor relationship quality
style. There was no significant difference in the relative frequen- which characterises this style may be explained in part by the
cy of change in attachment style between subjects who main- persistence of issues which have not been adequately resolved in
tained the same dating arrangements and those who reported a previous romantic relationships. A useful direction for future
change in dating pattern. suggesting that a general measure of research would be to explore in more depth the link between
change in dating status is not related to the stability of attachment attachment style, patterns of adjustment to relationship break-up,
style. Morcover, of the 18 subjects who experienced the break-up and the progress of new relationships.
of a relationship during the ]@week interval, only OM changed The nonsignificant association between attachment style and
hidher endorsement of the forced-choice measure, suggesting perceptions of who initiated the break-up was unexpected. As
that relationship dissolution. at least for young subjects in rela- noted earlier, however, the likelihood of attributing initiation of
tively short-term relationships, docs not exert a strong influence the break-up entirely to the partner was somewhat greater for
on perceptions of attachment style. anxiouslambivalent subjects than for the other attachment
It should be noted that in contrast to thesc findings concerning groups. Replication of this study with a larger sample would pro-
relationshipbreak-up, the development of new romantic relation- vide a more powerful test of the relationship between these vari-
ships appears to be associated with changed perception of attach- ables. Furthermore, the tendency of males to attribute
ment style. In the present study, the most common change in relationship break-up to the self and for females to attribute it to a
dating arrangements from lime 1 to Time 2 involved the forma- mutual decision mitigates against the identification of attachment
tion of steady relationships: 26 subjects described their relation- style differences in these attributions - larger scale studies are
ship as steady at lime 2. but reported dating several partners, or required in order to allow separate analyses for males and
Downloaded by [Universitaets und Landesbibliothek] at 03:53 24 November 2013

not dating at all. at Time 1. Of these 26 subjects. 12 showed a females. It is also possible that a quantitative (as opposed to cate-
change in self-reported attachment style. A frequency compari- gorical) measure of perceptions of who initiated the break might
son indicated that subjects forming a steady relationship were have produced a different result.
more likely than all other subjects to change their endorsement of It should also be recognised that an individual’s accounts of
) p < .05. Of the 12 subjects who
the attachment item. ~ ~ (=15.18, relationship dissolution are not static, but may vary widely
entered a steady relationship and also changed attachment style, according to such factors as the audience and the stage of mourn-
10 changed between endorsement of the secure and avoidant ing (McCall. 1982; Stephen, 1987). In the present sample, the
styles (five in each direction). However, this was the most com- low frequency of attributing relationship break-up to the partner
mon form of change in attachment style for the sample generally may reflect subjects’ reluctance to invoke an explanation imply-
(32 of 45 changes), and in view of the limited subject numbers, ing rejection or failure.
no conclusions can be drawn concerning the association between The lack of association between attachment style (using the
specific patterns of change in attachment style and involvement forcedchoice measure) and reports of relationship termination is
in steady relationships. not surprising, since subjects in this study were very young, and
the number of stable relationships would be relatively small.
DISCUSSION Nevertheless, subjects reporting (retrospective) relationship dis-
The percentages of subjects endorsing the three attachment solution were clearly separable from other subjects with dating
styles in the forced-choice measure are generally similar to those backgrounds, in terms of attachment scales and mental models of
reported by Hazan and Shaver (1987) and Feeney and Noller relationships. This result provides indirect support for the attach-
(1990), although the proportion of anxioudambivalentsubjects is ment perspective, which emphasises the continuing role of the
slightly lower in the present sample. individual’s expectations about the availability and trustworthi-
The data supported the hypothesised attachment style differ- ness of other people in influencing the quality of close relation-
ences in emotional reactions to relationship dissolution Avoidant ships.
subjects reported the most relief. with the mean response indicat- Of course we arc not able to make causal inferences from the
ing that they were “quite” relieved (the olher styks typically retrospective data - it is possible. for example, that avoidant
responded with “slight” refief). The association between avoidant attachment and lack of trust sfemfrom the experience of relation-
attachment and low levels of emotional distress is consistent with ship dissolution. rather than contributing to the dissolution. It
Simpson’s (1990) data, although the effect in that study was con- should be noted, however, that subjects reporting relationship ter-
fined to male subjects. In accordance with the notion that anx- mination had. on average, experienced the break-up over 4
ioudambivalent subjects may be less skilled at monitoring the months earlier, and that roughly 40% of these subjects were now
progress of their relationships, they reported being more sur- dating new partners; hence the impact of that experience may not
prised than avoidants at the break-up, and also reported being the be sufficient to entirely explain the observed relationships.
most upset (they reported, on average, being “very” upset, while In support of this interpretation, the prospective data indicate
secures were “quite” upset, and avoidants only “slightly” upset). that insecurity of attachment (in panicular. avoidant attachment)
Since anxiouslambivalentsubjects tend to be poorly prepared for predates relationship instability. and that relationship break-up
the break-up of their relationships. and since they are often high- has little effect on perceptions of attachment style. The latter con-
ly dependent odtheir relationship partners (Feeney & Noller, clusion is further supported by the finding that subjects experi-
1990, 19!31), they are clearly at risk in terms of their adjustment encing relationship dissolution were no more likely than others to
to the dissolution. change their endorsement of the forced-choice measure of attach-
Following recent relationship dissolution, anxiouslambivalent ment.
subjects were the most likely to report cumntly being in love, as The association between females’ avoidant attachment and
hypothesised; all anxioudambivalent subjects who had subse- relationship instability is consistent with the findings of Davis
quently formed a new dating relationship reported being in love and Kirkpatrick (1990). and points to the need to consider both
(even those who reported currently dating more than one part- attachment styles and gender role expectations in predicting rela-
ner). The association between anxioudambivalent attachment tionship outcomes. The present study differed from that of Davis
and self-reports of being in love has also been found in a large and Kirkpatrick, however, in finding no link between males’ anx-
sample with no restrictions on dating history (Feeney & Noller, ioudambivalent attachment and relationship instability. This
1990). result may be a function of the low power in the present study,
In light of the finding that anxiouslambivalent subjects are very stemming from the small number of broken relationships in the
surprised and upset by relationship break-up, the link between prospective part of the research (especially given the low rates of
anxioudambivalent attachment and reports of “being in love endorsement of anxioudambivalent attachment typically found in
again” raises questions about the future-of the relationships sub- studies of adult attachment).
sequently formed by these subjects. Specifically,the tendency for Onequarter of the subjects changed their endorsement of the
premature involvement with new love partners suggests an anx- forcedchoice attachment item at follow-up, and the attachment
Australian Journal of Psychology Vol. 44, No. 2.1992 pp. 6!?-74
 74 Judith A. Feetney and Patricia Mer

scales also showed only moderate stability over time.Clearly. the Duck. S. (1982). A copollnphy of relrtionship disengagrnwnt and diuolu-
stability of a construct such as attachment style cannot be tDn In S.Duck (Ed.). PerJonol rehionships: V d 4. Dissdving person-
assessed independently of the available measures, and more ol relaIbuhipJ (pp. 1-30). Ladon: Audcmic Rar
refined instnuncnts may yield higher stability. As noted earlier. Fctne)l. JA. (1990). Thc ottac- perspective an odulr nnnontic rehion-
however, Bowlby (1979) has suggested that d y experiences of ship Unpublirhcd Qaoral dLwrtation, Univusity of Quecnshd, Bris-
bsnc
romantic relationships may be associated with fluctuation in
b y . J.A.. & N o l k . P. (1990). Attachment styk as a prediccw of adult
attachment style, and attachment measures appear to be mom sta- romantic relationships. Joumol of Personolity and Social Psychology,
ble in samples of older adults (Shaver. 1989). si% 281-291.
Since the measucts of attachment style studied hae show only Fcauy. J.A.. & Noller. P. (1991). Attachment dyk and vnbal &scriptions of
moderate stability, further investigation of factors related to romantic partners. Joumal of Sociol and Personol Relotionships, 8.
change in attachment style is of prime importance. The present 187-2 15.
data suggest an association between formation of steady relation- H p ~ C.h ( 1989. Ckloba). Changes in adult a t t o c h n t rryle: Testing alter-
ships and change in attachment style; however, amongst older notiw interpretathu Paper prrscnted at the Society for Experimental
subjects who have had morc experience of established romantic Social Rycholgy. Srnu Monica CA.
relationships. pmxpuons of attachment style may be influenced H~zplsC. & Shaver. P. (1981). Romantic love cowqmdized as an attach-
just as strongly by relationship dissolution as by relationship ment process. Journal of Personality ond Sociol Psychology, 52.
511-524.
development. More complete longitudinal studies tracking the
stability of attachment style and its relation to life changes Hcndrick C, & Hendrick. SS. (1989). Rcrurdr on love: Docs it mcasurc
up? J o u d of P e r w d i t y and Social Psychology,56,784-79.
(including relationship events) are essential to clarify this ques-
Downloaded by [Universitaets und Landesbibliothek] at 03:53 24 November 2013

Johnson. M.P. (1982). Social and cognitive features of h e dissolution of

tion. L o n g i t d i d studies should also provide evidence relevant commitment to relationships. In S . Duck (Ed.). Personal relationships:
to the theoretical prediction that change in attachment style is V d 4. Dirrohing personal r e h i w h i p s (pp. S1-73). LonQn.A d e m -
mediated by change in mental models. ic Press.
JohluorrGcorgc. C.. & Swap, W.C. (1982). Measurement of specifs inter-
SUMMARY AND CONCLUSIONS p a s o d trust: Constmction and validation of a scale to assess trust in a
The present study clarifies the approach of the three attachment specific other. Journal of Personolity oad Social Psychology. 43.
styles to relationship dissolution. In accordance with theory relat- 1-13 17.
ing attachment styles to affect regulation, avoidant subjects K o b p l R.R.. & Scccry. A. (1988). Attachment in late adolescence: Worlring
described a sense of relief at relationship termination. while anx- models. affect regulation. and ccprrtentationr of self and ohas. Child
ioudambivalent subjects reported surprise and distress. Despite Drvebpment,59. 135-146.
their distress at relationship termination. the latter group were Levy. M.B.. & Davis ICE (1988). Lavatylu and aaxhncnt styles corn
parrd: Their rclalionr to each olhu and lo v U i m rehionship chnctcr-
quick to seek new love partners. Avoidant attachment was pn- istics. Journal of Social and Personal Relationships. 5, 439-471.
dictive of relationship dissolution. especially for females. The
Main, M..Kaplan. N., & Cassidy. J. (1985). Security in infancy, childhood.
experience of relationship dissolution had little impact on pcrceg and dullhood: A move to the level of repmenotion. Monograph of fhe
tions of attachment style, although the formation of a steady rela- Society for Research in Child Development, SO(1 & 2). 66-104.
tionship was associated with fluctuation in attachment style. In McCdl. G. ( I 982). Becoming unrclated Tht management of bond dissolu-
older samples who have developed more established relation- tion. In S . Duck (Ed.). Personal relorionships: Vol. 4. Dissolving person-
ships. the impact of relationship break-up on models of attach- al relorionships (pp. 21 1-231). London: Academic Resr.
ment may be greater. Pilkington. CJ.. & Richardson. D.R. (1988). Perceptions of risk in intimacy.
Journal of Social and Personal Relotionships, S, M3-H)8.
REFERENCES Ricks, M.H. (198s). The social bansmission of parental behavior. Anachment
BYrter. L.A. (1984). Trajectories of relationship disengagement. Journal of across generations. Monographs of the Society for Reseorch in Child
Social ond P e n o w l Relotionships. I , 29-48. ~cvclopncn,MI B 2). 21 1-m.
Baxtcr. LA. (198s). Accomplishing relationship discngagemcnt. In S. Duck Shaver. P. ( 1989. October).Att&nt theory as o -rk for the study
& D. Perlman @is.). U n d c d i n g personal rebtiowhips: An interdis- cqclose relmionrhipr Papa prrscnlcd al the Society for Experimcnul
c@inary opprwrh (pp. 243-265). London:Sage. Social Psychology. Sanu Monica, CA.
Bowlby. J. (1973). Attuchment ond loss: Vol. 2. Sepurarion: Anxiety and Simpson. J.A. (1990). Influcncc of attachment style on romantic relation-
onger. New Yorlc Basic Books. ships. Journal of Personolity ad socirrl Psychology,59.97 1-980.
Bowlby. J. (1979). 'Thc making and b r e h g o/ 4 J e c t W ~ lbondr. London: Spcrling. M.B. (1985). Discriminant measurn for desperate love. Jouml oj
Tavistock. Personality Assessment. 49,324328.
Collins. N.L. & R e 4 SJ. (1990). Adult attachment, working modcb. and Srwfc. LA., & Reexn. J. (1986). Atladurrnt and the construction of rela-
&tionship quality in dating cwpks. J o w l of Personality and Social tionrhips. In W.W.Harmp & Z Rubin (Eds.). Relotionships and &el-
Psychology, s8.644463. opment (pp. Sl-7 l). Hillsdal~NJ:Edbaum.
Davis, KE.& Kirltp.eick. L (1990. July). Adult n r m ~ l i ccufochmcnt style Srwfe. LA.. & Waters. E (1977'). Attachment as an organizationalconstNcL
and the prrdicrion of courtship progress: Evidcncc fmm three c h r t s . ChiidDevelopmnu.48.1184-1199.
Paper presented at the International Conference on Personal Rclation- Stephen. T. (1987). Attribution and djustmcnt to relationship termination.
ships, Oxfwd Jwrnol of Social nnd Personal Rebtionships,4,4741.

Australian bumd of Psychobgy Vol. 44.No. 2.1992 pp. 69-74