See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/235377975

Romantic Attachment and Relationship Functioning in Same-Sex Couples

Article in Journal of Counseling Psychology · January 2013

DOI: 10.1037/a0030994 · Source: PubMed

CITATIONS READS
59 4,542

3 authors:

Jonathan Joseph Mohr Dylan Selterman

University of Maryland, College Park Johns Hopkins University
75 PUBLICATIONS 7,098 CITATIONS 30 PUBLICATIONS 7,168 CITATIONS

SEE PROFILE SEE PROFILE

Ruth Fassinger
University of Maryland, College Park
66 PUBLICATIONS 6,554 CITATIONS

SEE PROFILE

All content following this page was uploaded by Ruth Fassinger on 24 May 2016.

The user has requested enhancement of the downloaded file.

 Journal of Counseling Psychology © 2013 American Psychological Association
2013, Vol. 60, No. 1, 72– 82 0022-0167/13/$12.00 DOI: 10.1037/a0030994

Romantic Attachment and Relationship Functioning in Same-Sex Couples

Jonathan J. Mohr, Dylan Selterman, and Ruth E. Fassinger
University of Maryland

The present study was designed to investigate links between dimensions of romantic attachment and
relationship functioning in a cross-sectional sample of people in same-sex relationships, with the goals
of replicating basic findings from research on heterosexual couples and advancing understanding of
unique issues faced by same-sex couples. The sample included 274 female couples, 188 male couples,
34 women whose female partners did not participate, and 39 men whose male partners did not participate.
Participants were recruited from geographically diverse regions of the United States and Canada and
provided data by responding to pencil-and-paper surveys. Attachment insecurity in both self and partner
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

were linked with poor relationship functioning across a range of variables (satisfaction, commitment,
trust, communication, problem intensity). The pattern of results was identical for women and men, but
This document is copyrighted by the American Psychological Association or one of its allied publishers.

the effects were stronger in male couples for some associations between attachment variables and indices
of positive relationship functioning. Monogamy was positively associated with relationship quality only
when participants or their partners reported moderate or high levels of attachment anxiety. Contrary to
hypothesis, attachment did not moderate links between minority stressors and relationship functioning.

Keywords: same-sex couples; lesbian, gay, and bisexual; attachment; romantic relationships

Attachment theory (Bowlby, 1969/1982) has been used in pro- have had consistently available, responsive, and supportive
grammatic research to explain interpersonal and intrapersonal pro- attachment figures will develop a secure attachment style—a
cesses in romantic relationships (Mikulincer & Shaver, 2007). A style characterized by comfort with intimacy and trust in the
core tenet of attachment theory is that people develop schemas of availability of attachment figures (Mikulincer & Shaver, 2007).
close relationships on the basis of repeated interactions with at- People with a secure attachment style have been shown to have
tachment figures (e.g., parents, romantic partners) throughout the a positive outlook on their romantic relationships (Hazan &
life span. Such schemas— often referred to as attachment working Shaver, 1987), resolve conflict effectively (Pietromonaco,
models (Bretherton & Munholland, 2008)— have been linked with Greenwood, & Barrett, 2004), and regulate emotions in a
a variety of cognitions, behaviors, and emotional experiences in a healthy way (Shaver & Mikulincer, 2007). In contrast, individ-
romantic context, including trust, communication, and general uals whose attachment figures were inconsistently available or
relationship well-being (Feeney, 2008). Most work on romantic consistently unresponsive to the individual’s needs tend to
attachment focuses on heterosexual couples, with very little re- develop an insecure attachment style and grow to have more
search on same-sex couples (Mohr, 2008). However, as discussed negative outlooks on relationships and show poorer relational
below in greater detail, research on same-sex couples can both adjustment. Attachment insecurity can be summarized by the
establish the generalizability of attachment theory and illuminate two overarching dimensions of attachment anxiety and avoid-
ways that sociocultural processes interact with attachment pro- ance (Brennan et al., 1998). Anxious attachment is marked by
cesses. The current study was designed to investigate links be- hyperactivating strategies (e.g., chronic fear of abandonment,
tween romantic attachment and relationship functioning in same- compulsive assurance-seeking, high distress), whereas avoidant
sex couples, with the goals of replicating basic findings from attachment is characterized by deactivating strategies (e.g.,
research on heterosexual couples and advancing understanding of
reluctance to rely on others for emotional support, discomfort
unique issues faced by same-sex couples.
with closeness and intimacy; Mikulincer & Shaver, 2007).
Anxiety and avoidance are believed to exert their strongest
Attachment Theory and Romantic Relationships effects on individual and couple functioning in stressful, chal-
Attachment theory, as a framework for understanding indi- lenging, and novel situations, as well as situations that involve
vidual differences in close relationships, posits that people who separation from or conflict with one’s romantic partner—a
proposition supported by a variety of studies of heterosexual
couples (Feeney, 2008).
Jonathan J. Mohr and Dylan Selterman, Department of Psychology, A growing body of research suggests that attachment functions
University of Maryland; Ruth E. Fassinger, Department of Counseling and in a dyadic manner in romantic relationships. For example, one
Personnel Services, University of Maryland.
study of heterosexual dating couples demonstrated that partici-
Ruth E. Fassinger is now in the College of Graduate and Professional
Studies at the John F. Kennedy University. pants’ level of attachment avoidance was positively associated
Correspondence concerning this article should be addressed to Jonathan with their partner’s levels of avoidance and anxiety; similarly,
J. Mohr, Department of Psychology, University of Maryland, Biology/ participants’ attachment anxiety was positively related to their
Psychology Building, College Park, MD 20742. E-mail: [email protected] partner’s avoidance (Collins & Read, 1990). Other studies have

72
 ATTACHMENT IN SAME-SEX COUPLES 73

demonstrated that people’s perceptions of their relationship are attachment security is associated with same-sex relationship satis-
influenced by not only their own attachment style but also their faction (Elizur & Mintzer, 2003; Kurdek, 2002; Ridge & Feeney,
partner’s attachment style (i.e., partner effects; Feeney, 2008). 1998), commitment (Kurdek, 1997, 2002), and communication
Some research has even suggested that relationship processes may quality (Gaines & Henderson, 2002). The few attachment studies
be a function of the interaction of partners’ attachment styles. For that compared people in same-sex and heterosexual relationships
example, one study of married heterosexual couples found that the failed to detect any differences in levels of attachment insecurity
negative association between husbands’ self-reported levels of (Ridge & Feeney, 1998) or in associations between attachment and
avoidance and marital support was stronger to the extent that their romantic relationship quality (Kurdek, 1997, 2002).
wives had low avoidance levels (Gallo & Smith, 2001). Counter to findings from heterosexual samples, Ridge and
Anxiety and avoidance have also been linked to sexuality in Feeney (1998) found that lesbians reported higher levels of avoid-
heterosexual populations. In one study, people with a secure at- ance and lower levels of anxiety compared with gay men. Mohr
tachment style—a pattern marked by low anxiety and avoidance— (2008) offered several explanations for why the pattern of gender
were less likely to have sex outside of their primary relationships differences in sexual minority people may be the reverse of what
(Allen & Baucom, 2004). Also, attachment anxiety has been found is typically found in heterosexuals. For example, gay and bisexual
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

to be positively associated with difficulty communicating about men may experience especially high levels of abandonment anx-
sex (Feeney, Peterson, Gallois, & Terry, 2000) and, in women, the iety due to expectations of intimacy problems based on restrictive
This document is copyrighted by the American Psychological Association or one of its allied publishers.

number of sexual partners outside of the primary relationship male sex roles (Brown, 1995; O’Neil, 2008), or higher rates of
(Gangestad & Thornhill, 1997). sexual nonexclusivity in male couples (Peplau, Fingerhut, & Beals,
As noted by Feeney (2008), some studies have revealed gender 2004). Ridge and Feeney (1998) speculated that the higher avoid-
differences in attachment and relationship functioning in hetero- ance level found in lesbians compared with gay men may stem
sexuals. Links between avoidance and relationship outcomes tend from the considerable assertiveness and independence sexual mi-
to be stronger for men, and links between anxiety and relationship nority women must cultivate in order to develop a lesbian identity
outcomes tend to be stronger for women. This pattern is consistent and a life that is not centered around men. Alternatively, this
with traditional gender role norms, where self-reliance is valued in gender difference in avoidance may be due to sexual minority
men and emotional connectedness is valued in women (Feeney, men’s relative freedom from male gender role norms restricting
2008). This pattern is also consistent with evidence that, compared comfort with intimacy (O’Neil, 2008).
with men, women have higher levels of anxiety and lower levels of
avoidance (Del Giudice, 2011), although such differences are not
The Present Study
found in all cultures and thus likely do not reflect innate sex
differences (Schmitt et al., 2003). Despite such findings, a good The present study was designed to address several gaps in the
deal of research supports the idea that personality matters more empirical literature on same-sex romantic attachment. First, little is
than gender for predicting relationship health: Both types of inse- known about attachment-related dyadic processes in same-sex
curity are often associated with decreased marital satisfaction, less couples (i.e., processes related to attachment in both partners). In
disclosure, more destructive communication patterns, and less this study, we contribute to knowledge in this area by examining
constructive approaches to conflict in both men and women (Mi- the extent to which own and partner attachment are associated with
kulincer & Shaver, 2007). a range of variables reflecting relationship functioning (satisfac-
tion, commitment, trust, communication, conflict). Thus, a major
goal was to replicate basic attachment research from heterosexual
Attachment in Same-Sex Couples
dyads in a sample of same-sex couples. On the basis of findings for
Feeney (2008) characterized the body of research linking ro- heterosexuals, we hypothesized that attachment avoidance and
mantic attachment and relationship quality as “huge” (p. 466); anxiety— both own and partner—would be associated with poorer
however, only a small handful of such studies have examined the relationship health (lower satisfaction, commitment, and trust),
experiences of lesbian, gay, and bisexual people. The lack of lower quality communication patterns, and poorer conflict man-
research on same-sex romantic attachment may stem from the agement. We also expected attachment anxiety and avoidance in
reasonable assumption that attachment should function in a similar one partner to be positively associated with these variables in the
fashion across couple types. Indeed, Ainsworth (1985), one of the other partner.
developers of attachment theory, held this belief and argued that Second, the present study provides what may be the first anal-
the main difference between same-sex and heterosexual couples is ysis of gender differences in associations between romantic attach-
that only the latter is socially sanctioned. However, in addition to ment (own and partner) and relationship functioning. Gender dif-
the importance of testing this assumption of similarity, research on ferences in same-sex couples may mirror those found in
same-sex couples is needed to advance understanding of ways heterosexual couples (reviewed above), particularly given evi-
that attachment intersects with sociocultural factors such as dence that sexual minority women and men differ in ways that are
gender socialization and social stigma in romantic relationship consistent with traditional gender socialization (Fassinger & Arse-
experiences. neau, 2007). However, the process of developing a sexual minority
Research has suggested that the similarities between same-sex identity may require lesbian, gay, and bisexual (LGB) people to
and heterosexual couples far outweigh the differences, both in question traditional gender roles more than is typical for hetero-
relationship quality and the processes regulating satisfaction and sexuals. Indeed, research has indicated that same-sex couples are
commitment (Peplau & Fingerhut, 2007). A few studies have less likely than their heterosexual counterparts to base their rela-
supported this view with respect to attachment, given findings that tionships on traditional gender roles (Peplau & Spalding, 2003).
 74 MOHR, SELTERMAN, AND FASSINGER

Thus, the gender-based patterns found in attachment research on Method

heterosexual couples may be attenuated or even reversed in same-
sex couples, possibilities that were investigated in the present
Participants and Procedure
study.
Third, the present study offers among the first data on the Participants were 582 (58.4%) women and 415 (41.6%) men
interface between attachment and sexuality in same-sex couples. who were in same-sex romantic relationships at the time of the
Desire and norms for sexual exclusivity are less common among study. The sample included 274 female couples, 188 male couples,
sexual minority people than their heterosexual counterparts, and 34 women whose female partners did not participate, and 39 men
nonexclusive relationships are much more prevalent among male whose male partners did not participate. Length of the romantic
couples than either female or heterosexual couples (Harris, 2002; relationships ranged from 0.25 to 31.33 years (M ⫽ 6.41, SD ⫽
Peplau & Cochran, 1983; Peplau et al., 2004). Research has 5.76), and age ranged from 18 to 69 years (M ⫽ 36.54, SD ⫽
suggested that sexual exclusivity and attitudes toward fidelity 9.49). Participants identified their race/ethnicity as Asian Ameri-
within same-sex couples is unrelated to relationship quality and can/Pacific Islander (1.4%), Black/African American (2.7%),
commitment, although agreement about exclusivity was strongly White/Euroamerican (85.4%), Hispanic/Latina/Latino (2.8%), Na-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

associated with relationship satisfaction in one study of male tive American/Native Alaskan (0.9%), Middle Eastern/Arab
couples (see Peplau et al., 2004). Secure attachment may provide (0.2%), biracial/multiracial (4.4%), and “other” (1.2%); 1% did not
This document is copyrighted by the American Psychological Association or one of its allied publishers.

a foundation for successful negotiation of sexual exclusivity and report race/ethnicity. Regarding formal education, 55.6% of par-
openness in same-sex couples, particularly given evidence, dis- ticipants reported having at least an undergraduate degree. Partic-
cussed above, that people high in security are more likely than ipants lived in rural (11.1%) and nonrural (88.9%) locations in
others to feel comfortable with sexual negotiation and to integrate diverse regions of the United States and Canada.
sex and intimacy. In contrast, individuals high in attachment Participants were recruited through solicitations on electronic
anxiety or avoidance may be less able than others to negotiate mail lists for LGB professionals, newspaper advertisements, and at
nonmonogamy effectively, and may be more likely than others to a national Black pride event in Washington, DC. Announcements
use nonmonogamy as a means of coping with intimacy difficulties for the study specified that we were seeking same-sex romantic
in the relationship. For these reasons, we expected associations partners who had been together for at least 2 months (to ensure
between monogamy and relationship quality to be weak for those some level of identity as a couple), and provided telephone and
high in attachment security but positive for those low in attach- e-mail contact information for researchers. Individuals expressing
ment security. an interest in the study were sent two sets of packets that included
Finally, the present study may be the first to examine the role of the survey (each marked with a couple identification number),
basic information about the study, and self-addressed stamped
secure attachment in mitigating the negative effects of stress on
envelopes. Participants were instructed to complete the survey in a
relationship quality. Negative social views of homosexuality and
setting separate from their romantic partner and to seal the survey
bisexuality are a source of multiple stressors for LGB people,
in the mailing envelope immediately afterward. Surveys were
ranging from the external (e.g., prejudice, discrimination, or vio-
received from 1,004 individuals (49% of the surveys mailed), but
lence) to the internal (e.g., internalized homonegativity). A grow-
seven of these surveys were not included in the present study
ing body of work has offered evidence of the negative effects of
because they were minimally completed. Data from these partici-
these stressors on same-sex relationship functioning (Mohr &
pants have been published elsewhere (citation withheld for masked
Daly, 2008). As Brown (1995) noted, social stigma assumes a
review), but none of the main results of the present study overlap
unique meaning in the context of a romantic relationship because
with what has appeared in previous articles.
same-sex partners must deal with their differences in self-
acceptance and comfort with openness about their sexual orienta-
tion. Indeed, perceived dissimilarity on such factors was linked Measures
with lower relationship quality in one study of same-sex couples Measures were presented in a fixed order starting with demo-
(Mohr & Fassinger, 2006). graphic items and then moving onto scales assessing attachment,
Although coping with minority stress may be challenging for romantic relationship functioning, and facets of LGB identity.
most same-sex couples, couples likely differ in the extent to which Scales were scored by averaging relevant items, after reverse
their coping efforts mitigate the negative effects of stress on the scoring items as needed.
relationship. One possibility, as yet unexamined, is that secure Romantic attachment. The revised version of the Adult At-
attachment may buffer the relationship from minority stress. This tachment Scale—Revised (AAS–R; Collins, 1996) was used to
proposition seems plausible, given evidence from heterosexual assess global attachment anxiety and avoidance. The AAS–R
couples that attachment security is associated with factors known consists of 18 self-report items that are rated on a 7-point scale in
to improve ability to cope with stress and conflict (e.g., self- relation to respondents’ general tendencies in close relationships
disclosure, openness to feedback, support seeking, positive rela- (1 ⫽ not at all like me, 4 ⫽ somewhat like me, 7 ⫽ very much like
tionship attributions). We examined three potential sources of me). This scale can be scored to form six-item subscales assessing
minority stress: one external (perceived prejudice), one internal comfort with closeness (Close), comfort depending on others (De-
(internalized stigma), and one dyadic (perceived dissimilarity in pend), and attachment anxiety (Anxiety). Collins (1996) suggested
comfort with one’s sexual orientation). We believed attachment that Close and Depend could be combined to form a measure of
security would weaken negative impacts of minority stressors on attachment avoidance (after reverse scoring); subsequent research,
relationship quality. however, indicated that Depend has substantial associations with
 ATTACHMENT IN SAME-SEX COUPLES 75

measures of both attachment avoidance and anxiety. For example, scored for three subscales: Supportive Discussion (e.g., “We take
Depend was correlated with the Avoidance (r ⫽ ⫺.73) and Anx- the time just to talk things over”; ␣ ⫽ .82), Aversive Communi-
iety (r ⫽ ⫺.39) subscales of the widely used Experiences in Close cation (e.g., “My partner and I find it hard to disagree without
Relationships measure (Brennan et al., 1998). In contrast, Close losing our tempers”; ␣ ⫽ .73), and Communication Apprehension
was highly correlated with Avoidance (r ⫽ ⫺.87) but not Anxiety (e.g., “I fail to express disagreement with my partner for fear s/he
(r ⫽ ⫺.05). Because the Close subscale appears to be a purer will get angry with me”; ␣ ⫽ .72). Items were rated on a 7-point
indicator of attachment avoidance compared with Depend, we used Likert scale (1 ⫽ strongly disagree, 7 ⫽ strongly agree). The short
only Close in the present study. Items on the Close subscale were form of the scale has not been widely used, but subscales were
reverse scored so that high scores on the measure would reflect developed using exploratory factor analysis and are face valid. In
discomfort with closeness (i.e., high avoidance). The two attach- one study, scores on Aversive Communication were positively
ment subscales are referred to as Anxiety and Avoidance from this associated with ratings of expected negative response to romantic
point forward. Internal consistency estimates were .89 for Anxiety conflict, and with measures of attachment avoidance and anxiety
and .85 for Avoidance. (Mohr, Crook, & Kolchakian, 2010).
Relationship functioning. Five self-report measures of ro- Minority stress measures. Three potential sources of minor-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

mantic relationship functioning were used to assess commitment, ity stress were assessed. Two items developed by Meyer (1995)
satisfaction, communication, trust, and problem intensity. We ad- were used to assess whether participants experienced anti-LGB
This document is copyrighted by the American Psychological Association or one of its allied publishers.

justed item wording as needed to be appropriate for same-sex violence and discrimination in the year prior to participation.
romantic partners (e.g., the term partner was used instead of Consistent with the approach used by Meyer, a binary variable was
husband and wife). The Commitment Scale (Lund, 1985) was used created to indicate whether respondents responded affirmatively to
to assess individuals’ estimated likelihood of continuing their at least one of these two forms of prejudice (0 ⫽ no, 1 ⫽ yes).
relationship. This scale consists of nine items (e.g., “How likely is Meyer found that ratings on this variable were associated with a
it that your relationship will be permanent?”) rated on a 7-point range of negative psychosocial outcomes in a sample of sexual
Likert scale (1 ⫽ not at all, 7 ⫽ a great deal) that yielded a minority men, even after controlling for other sources of minority
Cronbach’s alpha of .76 in the present sample. Lund found that stress. Internalized stigma was assessed with the Internalized
commitment scores differentiated relationship types (e.g., casual Homonegativity subscale of the Lesbian and Gay Identity Scale
dating, married) and were positively associated with relationship (LGIS; Mohr & Fassinger, 2000). The LGIS was developed on the
length and measures of relationship quality. basis of separate exploratory factor analyses for women and men,
Two items assessed global relationship satisfaction: “I’m happy and items are rated on a 7-point Likert scale (1 ⫽ disagree
in my relationship” and “I’m satisfied with my relationship” (Col- strongly, 7 ⫽ agree strongly). The five-item Internalized Homon-
lins & Read, 1990). Items were rated on a 7-point Likert scale egativity subscale assesses respondents’ dissatisfaction with their
(1 ⫽ disagree strongly, 7 ⫽ agree strongly) and yielded a Cron- own sexual orientation (␣ ⫽ .79; sample item: “I wish I were
bach’s alpha of .91. Collins and Read (1990) found that the scale heterosexual”). Validity evidence was provided through predicted
formed by averaging these two items was related in predicted ways associations with phase of LGB identity development, degree of
to measures of comfort with intimacy and abandonment anxiety. investment in one’s LGB social identity, self-esteem, and degree
Perceived intensity of problems in the relationship was assessed of interaction with heterosexual individuals. Finally, a scale as-
with a list of potential relationship problem areas (Hill, Rubin, & sessing perceived dissimilarity to one’s romantic partner in com-
Peplau, 1976). The list includes 17 areas (e.g., “My desire to be fort related to one’s sexual orientation was developed for this
independent”; “Money”) rated for the degree to which they had study. The scale consists of five items that all begin with the
been a problem in the relationship (1 ⫽ not a problem, 4 ⫽ major phrase, “My partner and I are equally comfortable. . .” and end by
problem). As predicted, scores on this measure were higher among referring to a different aspect of LGB experience: (e.g.,“ being
lesbians in relationships characterized by high power imbalance ‘out’ in public,” “being ‘out’ to family members,” “being ‘out’ to
than among lesbians in low-imbalance relationships (Caldwell & straight friends”). Items were rated on a 7-point Likert scale (1 ⫽
Peplau, 1984). The estimated internal consistency for Intensity of disagree strongly, 7 ⫽ agree strongly), reverse scored and aver-
Problems was .81 in the present sample. aged to form an index of perceived dissimilarity (␣ ⫽ .87).
The Trust Scale (Rempel, Holmes, & Zanna, 1985) was used to
assess the degree to which individuals had faith in their romantic
Data Analysis Considerations
partners and viewed their partners as dependable and predictable.
The scale consists of 18 items (e.g., “I have found that my partner The data presented several analytic challenges. First, romantic
is a thoroughly dependable person, especially when it comes to couple data often violate the assumption of independent observa-
things that are important”) rated on a 7-point Likert scale (1 ⫽ tions associated with the general linear model because partners
strongly disagree, 7 ⫽ strongly agree). The full scale yielded a typically share perceptions of their relationship. Second, many
Cronbach’s alpha of .89 in the present sample. Full scale scores variables had somewhat skewed distributions, reflecting the gen-
have been shown to correlate positively with an established mea- erally high levels of well-being in participants (e.g., high relation-
sure of relationship quality, and negatively with variability in daily ship satisfaction). Finally, some data were missing, most notably
reports of relationship quality (Campbell, Simpson, Boldry, & from the 73 participants whose partners did not submit completed
Harris, 2010). surveys. We lacked data to estimate partner effects for these
Communication patterns were evaluated with a short form of the participants and actor effects for the missing partners. The missing
Marital Communication Inventory (Chang, Schumm, Coulson, value rate for variables from participants ranged from 0.0% for
Bollman, & Jurich, 1994). This 14-item version of the scale is gender to 2.0% for perceived prejudice. For partner data, the
 76 MOHR, SELTERMAN, AND FASSINGER

missing value rate ranged from 0.0% for gender to 9.0% for volved preliminary omnibus tests that, if statistically significant,
perceived prejudice. The total missing value rate was 2.6% when were followed by tests of individual regression coefficients. We set
considering all variables included in the present study. the familywise error rate at .05 for each set of omnibus tests. Thus,
To handle these challenges, we used options for regression in each omnibus test was conducted at the ␣ ⫽ .007 level because
the complex survey data module in Mplus software version 6.1 there were seven relationship outcome variables. Tests of regres-
(Muthén & Muthén, 1998 –2011). The nested data structure was sion coefficients that followed significant omnibus tests were then
addressed with the COMPLEX option, which uses a sandwich conducted at the more liberal ␣ ⫽ .05 level.
estimator to adjust standard errors for nonindependence of obser-
vations. This option also addresses concerns regarding skewed Results
distributions through use of a robust maximum likelihood estima-
tion algorithm. Finally, we used the full information maximum
Assessment of Nonindependence and Descriptive
likelihood (FIML) feature to address missing data concerns, an
Statistics
approach that is considered a best practice in missing data man-
agement (Schlomer, Bauman, & Card, 2010) and permitted use of We used the intraclass correlation coefficient (ICC) to estimate
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

data from participants whose partners did not submit surveys. As the degree of nonindependence on the relationship outcome vari-
recommended by Enders (2010), we explicitly modeled the vari- ables due to the nesting of partners within dyads. In the case of
This document is copyrighted by the American Psychological Association or one of its allied publishers.

ances and covariances of the observed explanatory variables in dyads, the ICC can be thought of as the correlation between
order to include these variables in the FIML estimation process. partners’ scores on a variable or, alternatively, as the proportion of
Also, we included participant age as an auxiliary variable in the variance in the variable at the dyad level of analysis. ICCs were
main analyses to reduce estimation bias and increase estimation computed by running unconditional random intercepts regression
precision, given our finding that age was a significant predictor of models for each of the romantic relationship outcomes and divid-
(a) whether the person’s partner submitted a survey, (b) attachment ing the estimated variance in intercepts by the total variance. The
anxiety, and (c) several romantic relationship measures (e.g., com- expected interdependence between partners was found, with ICC
mitment). In other words, age was allowed to correlate with all of estimates ranging from .11 (Communication Apprehension) to .56
the main variables, and these associations improved the estimation (Commitment). These findings supported our plan to use multi-
of parameters in light of the missing data. level regression.
Consistent with recommendations of Kenny and Cook (1999), Descriptive statistics for the main variables are presented in
we used multilevel regression to estimate actor and partner effects Table 1 and are based on the untransformed scales. For regression
of romantic attachment simultaneously. In contrast to uses of analyses, however, all predictor variables were centered to in-
multilevel regression aimed at disentangling associations among crease interpretability of regression intercepts and prepare vari-
variables at multiple levels of analysis, this use of multilevel ables for moderation analyses. All of the main findings are sum-
regression is aimed at correcting standard errors to account for the marized in narrative form in Table 2.
nonindependence of observations. Some of our research questions
concerned moderation effects, and, as described in the Results, our Was Attachment in One Partner Related to
handling of these effects depended on whether the moderator was
Attachment in the Other Partner?
dichotomous (i.e., gender) or continuous (i.e., attachment).
In the main effects analyses, we set the familywise error rate at There was clear evidence of partner similarity on attachment anx-
.05 to control the Type I error rate for the regression of each iety (ICC ⫽ .21; z ⫽ 4.36, p ⫽ .000) and, to a lesser extent, avoidance
outcome on attachment variables. For example, when examining (ICC ⫽ .09; z ⫽ 1.91, p ⫽ .057). A multilevel regression indicated
actor and partner effects of both anxiety and avoidance, we tested that actor Anxiety was positively related to partner Avoidance scores
each of the four regression coefficients for each outcome at the (z ⫽ 4.24, p ⫽ .000), suggesting that people with high attachment
␣ ⫽ .0125 level. Analyses focusing on moderation effects in- anxiety are more likely than others to be with partners who are high

Table 1
Descriptive Statistics for Variables in the Main Analyses

Variable M SD 1 2 3 4 5 6 7 8 9 10

1. Anxiety-O 2.80 1.23 —

2. Anxiety-P 2.80 1.23 .21 —
3. Avoidance-O 2.73 1.15 .39 .15 —
4. Avoidance-P 2.73 1.15 .15 .39 .09 —
5. Satisfaction 6.18 1.02 ⫺.36 ⫺.23 ⫺.22 ⫺.10 —
6. Commitment 6.26 0.69 ⫺.30 ⫺.22 ⫺.12 ⫺.08 .59 —
7. Communication Apprehension 2.63 1.24 .38 .20 .36 .05 ⫺.44 ⫺.27 —
8. Aversive Communication 2.40 1.12 .37 .19 .18 .03 ⫺.43 ⫺.26 .66 —
9. Supportive Discussion 6.21 0.86 ⫺.36 ⫺.21 ⫺.28 ⫺.14 .56 .42 ⫺.51 ⫺.49 —
10. Trust 5.84 0.84 ⫺.45 ⫺.25 ⫺.26 ⫺.09 .62 .50 ⫺.55 ⫺.63 .59 —
11. Intensity of Problems 1.44 0.29 .36 .21 .24 .11 ⫺.49 ⫺.39 .40 .48 ⫺.39 ⫺.60
Note. O ⫽ own rating; P ⫽ partner’s rating. Statistical significance of correlations is not indicated due to nested data structure.
 ATTACHMENT IN SAME-SEX COUPLES 77

Table 2
Summary of Main Research Findings for Anxiety and Avoidance

Research question Anxiety Avoidance

Was attachment in one partner associated Yes. Own Anxiety was positively associated with Yes. Own Avoidance was positively associated
with attachment in the other both partner Anxiety and partner Avoidance. with partner Anxiety.
partner?

Was insecure attachment linked with

poorer relationship functioning?
Own attachment Yes. Own Anxiety was linked with poorer Yes. Own Avoidance was linked with poorer
functioning on all relationship outcome functioning for Supportive Discussion,
variables. Trust, Communication Apprehension, and
Intensity of Problems.
Partner attachment Yes. Partner Anxiety was linked with poorer No. Partner Avoidance was negatively related
functioning on all relationship outcome to Aversive Communication, contrary to
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

variables. hypothesis.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Were there gender differences in Yes. Anxiety was slightly higher in men than in No.
attachment? women.

Were links between attachment and

relationship functioning moderated
by gender?
Own attachment Yes. Own Anxiety was more strongly linked with Yes. Own Avoidance was more strongly linked
lower Satisfaction and Commitment in men with lower Supportive Discussion in men
than in women. than in women.
Partner attachment No. No.

Were links between sexual exclusivity

and relationship functioning
moderated by attachment?
Own attachment Yes. Sexual exclusivity was linked with lower No.
Satisfaction and Commitment only for people
whose own Anxiety levels was moderate or
high.
Partner attachment Yes. Sexual exclusivity was linked with lower No.
Satisfaction and Commitment only for people
whose partner’s Anxiety levels was moderate
or high.

Were links between minority stressors

and relationship functioning
moderated by attachment?
Own attachment No. No.
Partner attachment No. No

in avoidance. Including relationship length as a covariate did not Communication. The only significant finding for partner Avoid-
change these results, suggesting that these effects were not merely a ance was in the opposite direction from others: Participants had
function of time spent together. low levels of Aversive Communication to the extent that their
partner was high in Avoidance. We next examined whether partner
Was Insecure Attachment Linked With Poorer attachment moderated associations between actor attachment and
Relationship Functioning? relationship outcomes by adding to the above multilevel models
the four interaction terms representing combinations of own and
We investigated actor and partners effects of attachment on partner attachment. We then used Wald tests to determine whether
relationship functioning by regressing the seven relationship out- constraining the regression coefficients for the interaction terms to
come measures on own Anxiety, own Avoidance, partner Anxiety, zero worsened model fit. These Wald tests were nonsignificant
and partner Avoidance (see Table 3; all tests conducted at the ␣ ⫽ (p ⬎ .007), suggesting a lack of moderation effects.
.0125 level). Similar effects for attachment anxiety were found
across all of the outcomes: Relationship functioning was the most
Were There Gender Differences in Attachment?
negative for those whose own levels of Anxiety were high and
whose partner’s levels of Anxiety were high. Own Avoidance was Do women and men in same-sex relationships differ in levels of
negatively related to Supportive Discussion and Trust; positively Anxiety and Avoidance, and in the correlates of attachment ex-
related to Communication Apprehension and Intensity of Prob- amined above? To investigate the first part of this question, we
lems; and not related to Satisfaction, Commitment, and Aversive regressed the attachment variables on gender (using the same
 78 MOHR, SELTERMAN, AND FASSINGER

Table 3
Multilevel Regression Analyses Predicting Relationship Outcomes From Own and Partner Romantic Attachment

Communication Aversive Supportive Intensity of

Satisfaction Commitment Apprehension Communication Discussion Trust Problems
Predictor B SE B SE B SE B SE B SE B SE B SE

Anxiety-O ⫺0.23ⴱ 0.03 ⫺0.15ⴱ 0.02 0.28ⴱ 0.04 0.31ⴱ 0.03 ⫺0.19ⴱ 0.02 ⫺0.26ⴱ 0.02 0.07ⴱ 0.01
Avoidance-O ⫺0.08 0.03 ⫺0.01 0.02 0.25ⴱ 0.04 0.03 0.03 ⫺0.11ⴱ 0.03 ⫺0.06ⴱ 0.02 0.03ⴱ 0.01
Anxiety-P ⫺0.14ⴱ 0.03 ⫺0.10ⴱ 0.02 0.13ⴱ 0.04 0.12ⴱ 0.03 ⫺0.08ⴱ 0.02 ⫺0.11ⴱ 0.02 0.03ⴱ 0.01
Avoidance-P 0.01 0.03 0.01 0.02 ⫺0.06 0.04 ⫺0.07ⴱ 0.03 ⫺0.03 0.03 0.03 0.02 0.00 0.01
Note. O ⫽ own rating; P ⫽ partner’s rating.
ⴱ
p ⬍ .0125.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

approach to regression used in previous analyses). Anxiety levels .000), and the negative association between own Anxiety and
were slightly lower in women than men (z ⫽ ⫺2.47, p ⫽ .014, Commitment was stronger for men (B ⫽ ⫺0.21, SE ⫽ 0.03, p ⫽
This document is copyrighted by the American Psychological Association or one of its allied publishers.

d ⫽ 0.17), but no gender differences were found for Avoidance .000) than women (B ⫽ ⫺0.09, SE ⫽ 0.03, p ⫽ .001). For
(z ⫽ ⫺0.31, p ⫽ .760, d ⫽ 0.02). Supportive Discussion, none of the slope differences was found to
be different from zero at the .05 level. However, at the .10 level,
Were Links Between Attachment and Relationship there was a gender difference in slopes for own Avoidance (B ⫽
Functioning Moderated by Gender? 0.10, SE ⫽ 0.05, p ⫽ .055). The negative association between
own Avoidance and Supportive Discussion was stronger for
We next conducted a series of multiple-group regression anal- men (B ⫽ ⫺0.18, SE ⫽ 0.04, p ⫽ .000) than women (B ⫽ ⫺0.08,
yses to explore whether the links between attachment and relation- SE ⫽ 0.03, p ⫽ .026).
ship functioning differed by gender. As in previous analyses, we
tested for actor and partner effects of attachment variables on Were Links Between Sexual Exclusivity and
romantic relationship outcomes. However, in this case, we simul-
Relationship Functioning Moderated by Attachment?
taneously ran models for women and men and used Wald tests to
determine whether model fit was worsened by constraining the Our next set of analyses explored the extent to which own
effects of attachment on the outcome to be identical for women attachment and partner attachment variables moderated the asso-
and men; we followed up statistically significant results on this ciation between nonmonogamy and relationship variables. We
omnibus test with specific tests to determine the predictors for began by investigating own and partner attachment as moderators
which gender differences existed. To do this, we calculated the of links between own reports of monogamy and the outcomes. We
difference between respective regression coefficients for women created four interaction terms representing the product of the
and men (e.g., the difference between women’s and men’s coef- dichotomous monogamy variable with own and partner attachment
ficients for own Anxiety) and conducted a test to determine variables. We then ran multilevel regression models including
whether the slope difference was different from zero. A statisti- main effects for own attachment, partner attachment, and monog-
cally significant finding would indicate a gender difference in the amy status, along with the four interaction terms. As above, we
relation between the predictor (attachment style) and outcome. used Wald tests to determine whether these interaction terms, as
Did the associations between attachment and relationship vari- a group, accounted for additional variance in the relationship
ables differ by gender? Testing at the .05 level, overall gender variables.
differences were not found for the negative outcomes: Communi- Were links between monogamy status and any of the relation-
cation Apprehension, ␹2(4, N ⫽ 997) ⫽ 3.76, p ⫽ .439; Aversive ship outcomes moderated by attachment? Only two of the seven
Communication, ␹2(4, N ⫽ 997) ⫽ 1.10, p ⫽ .894; and Problem Wald tests were statistically significant at the .007 level: The
Intensity, ␹2(4, N ⫽ 997) ⫽ 5.61, p ⫽ .230. Also, gender differ- interaction terms led to better model fit for predicting Satisfaction,
ences did not emerge for Trust, ␹2(4, N ⫽ 997) ⫽ 8.10, p ⫽ .088. ␹2(4, N ⫽ 997) ⫽ 18.93, p ⫽ .001, and Commitment, ␹2(4, N ⫽
In contrast, gender differences were found for Satisfaction, ␹2(4, 997) ⫽ 16.95, p ⫽ .002. Thus, links between monogamy and both
N ⫽ 997) ⫽ 11.34, p ⫽ .023; Commitment, ␹2(4, N ⫽ 997) ⫽ Satisfaction and Commitment differed according to participants’
13.58, p ⫽ .009; and Supportive Discussion, ␹2(4, N ⫽ 997) ⫽ attachment patterns. To learn more about these moderation effects,
13.52, p ⫽ .009. we identified the specific interaction terms that were statistically
What was the precise nature of these gender differences? We significant and then probed these interactions with simple slopes
followed up the three significant omnibus tests with more focused analysis.
tests on the differences between corresponding women’s and Which interaction terms were statistically significant for Satis-
men’s slopes. For Satisfaction and Commitment, the only statisti- faction? There were statistically significant interaction effects of
cally significant gender difference in slopes was for own Anxiety monogamy status with both own Anxiety (z ⫽ 3.02, p ⫽ .003) and
(B ⫽ 0.13, SE ⫽ 0.06, p ⫽ .023 and B ⫽ 0.13, SE ⫽ 0.04, p ⫽ partner Anxiety (z ⫽ 3.13, p ⫽ .002). We next used simple slopes
.002, respectively). Specifically, the negative association between analysis to examine the precise nature of these interaction effects.
own Anxiety and Satisfaction was stronger for men (B ⫽ ⫺0.30, We first tested the regression of Satisfaction onto monogamy
SE ⫽ 0.05, p ⫽ .000) than women (B ⫽ ⫺0.17, SE ⫽ 0.03, p ⫽ status at three levels of own Anxiety: one standard deviation below
 ATTACHMENT IN SAME-SEX COUPLES 79

the mean, the mean, and one standard deviation above the mean. for attachment anxiety: Both own and partner anxiety were inde-
The association between monogamy status and Satisfaction was pendently associated with poorer relationship functioning on all
nonsignificant for participants with low attachment anxiety (B ⫽ outcomes. Own avoidance level was linked with lower trust and
0.11, SE ⫽ 0.14, p ⫽ .428), but this association was positive for poorer communication processes, whereas partner avoidance was
those whose anxiety level was moderate (B ⫽ 0.39, SE ⫽ 0.11, linked with less aversive communication. These opposite effects of
p ⫽ .000) or high (B ⫽ 0.67, SE ⫽ 0.16, p ⫽ .000). The pattern partner anxiety and partner avoidance on participants’ reports of
of findings was identical when examining partner anxiety as a aversive communication are consistent with research suggesting
moderator: Self-reported monogamy status was unrelated to Sat- that anxious individuals tend to amplify communication of distress
isfaction for participants whose partners had low anxiety (B ⫽ and anger, whereas avoidant individuals tend to cut off their
0.11, SE ⫽ 0.15, p ⫽ .446) but positively related to Satisfaction experience of negative affect (Mikulincer & Shaver, 2007).
for those whose partner’s anxiety level was moderate (B ⫽ 0.40, Some results were inconsistent with attachment research on
SE ⫽ 0.11, p ⫽ .000) or high (B ⫽ 0.70, SE ⫽ 0.16, p ⫽ .000). heterosexual couples. For example, attachment avoidance is typi-
How did attachment moderate the effects of monogamy status cally associated with less relationship commitment in heterosexual
on Commitment? For Commitment, there were interaction effects couples (Feeney, 2008); yet, in the present sample commitment
of monogamy status with both own Anxiety (z ⫽ 2.42, p ⫽ .015)
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

was uniquely associated only with anxiety. Also, as noted previ-

and partner Anxiety (z ⫽ 3.00, p ⫽ .003). Simple slopes analysis ously, research on heterosexual men has found that avoidance is
This document is copyrighted by the American Psychological Association or one of its allied publishers.

indicated that the association between self-reported monogamy more likely than anxiety to be linked with relationship outcomes in
status and Satisfaction was nonsignificant for participants with low heterosexual men, whereas the reverse was true for the gay and
attachment anxiety (B ⫽ 0.16, SE ⫽ 0.11, p ⫽ .129), but this bisexual men in the present sample. As noted earlier, anxiety may
association was positive for those whose anxiety level was mod- be especially salient for same-sex couples relative to heterosexual
erate (B ⫽ 0.35, SE ⫽ 0.08, p ⫽ .000) or high (B ⫽ 0.53, SE ⫽ couples because of exposure to negative societal beliefs question-
0.12, p ⫽ .000). The pattern of findings was identical when ing the sustainability of same-sex relationships, fewer social struc-
examining partner anxiety as a moderator: Self-reported monog- tures designed to encourage stability in same-sex couples, and
amy status was unrelated to Satisfaction for participants whose greater exposure to norms for negotiating nonmonogamy.
partners had low anxiety (B ⫽ 0.09, SE ⫽ 0.10, p ⫽ .347) but Results were largely similar for women and men in our sample.
positively related to Satisfaction for those whose partner’s anxiety No gender difference was found on mean avoidance levels, but,
level was moderate (B ⫽ 0.34, SE ⫽ 0.07, p ⫽ .000) or high (B ⫽ similar to results from an Australian sample investigated by Ridge
0.58, SE ⫽ 0.13, p ⫽ .000). In short, these results provided and Feeney (1998), women had slightly lower levels of anxiety
evidence that nonmonogamy was negatively associated with rela- compared with men. The pattern of links between attachment and
tionship satisfaction and commitment only for participants whose the relationship outcomes was identical across gender, although
own or partner’s anxiety levels were moderate or high. relations between attachment and aspects of positive relationship
functioning were stronger for men than women. A range of factors,
discussed above, may explain such differences (e.g., expectations
Were Links Between Minority Stressors and
of intimacy problems associated with restrictive male gender
Relationship Functioning Moderated by Attachment?
norms, internalization of negative views of male–male love rela-
Our next set of analyses explored the extent to which own tionships, prevalence of nonmonogamy). These factors may in-
attachment and partner attachment variables moderated the effects crease the sense of threat to the integrity of the relationship, and
of three potential sources of minority stress on relationship vari- thereby raise levels of attachment anxiety and increase activa-
ables: prejudice, internalized homonegativity, and perceived dis- tion of the attachment system. Research is needed to determine
similarity. For each minority stress variable, we created four whether these factors explain the gender differences observed in
interaction terms representing the product of the stressor with each this study.
of the four attachment variables. We then ran multilevel regression This study provided a unique opportunity to study nonmo-
models including main effects for own attachment, partner attach- nogamy in a population in which, relative to heterosexuals, having
ment, and stressor, along with the four interaction terms. As above, multiple sexual partners is more often viewed as a legitimate
we used Wald tests to determine whether these interaction terms, alternative to sexual exclusivity. Findings suggest that the meaning
as a group, explained additional variance in the outcome. None of and function of nonmonogamy varies according to people’s levels
the Wald tests reached significance at the .007 level. Thus, results of attachment anxiety. Sexual exclusivity was not associated with
did not support the view that attachment security buffers the ill relationship satisfaction or commitment for participants who re-
effects of minority stress on relationship functioning. ported low levels of anxiety, which mirrors previous research on
same-sex couples (Peplau et al., 2004). However, for participants
with moderate or high anxiety levels, satisfaction and commitment
Discussion
were higher among those who reported monogamous relationships.
This study may be the first to replicate basic findings on asso- A number of possible explanations may be offered for this finding.
ciations between attachment and relationship functioning from Attachment anxiety may increase the perception of multiple sexual
heterosexual couples in a sample of same-sex couples. Consistent relationships as a threat to the primary relationship, given the fear
with research on heterosexuals, results from a large community of abandonment that is at the heart of attachment anxiety and
sample of same-sex couples indicated that attachment anxiety and evidence of links between anxiety and jealousy (Collins & Read,
avoidance in both partners are linked with less positive relation- 1990). Alternatively, people with high anxiety may be more likely
ship evaluations and experiences. Findings were especially robust than others to respond to relationship dissatisfaction by seeking
 80 MOHR, SELTERMAN, AND FASSINGER

sexual partners outside of the primary relationship. Such behavior this study may be fueled by the negative social climate for LGB
may reflect the fact that people with high anxiety are more likely people (e.g., negative stereotypes about same-sex couples, expe-
than others to (a) lack the interpersonal skills to effectively discuss riences of anti-LGB prejudice, lack of formal recognition of same-
relationship dissatisfaction (Simpson, Rholes, & Phillips, 1996), sex relationships). Thus, a tentative implication for practice is that
(b) feel the need to cultivate “back-up” relationships in case they counselors working with clients in same-sex relationships may
are abandoned by their primary partner (Allen & Baucom, 2004), profit from exploring possible interconnections among fear of
or (c) use sexual partners outside the primary relationship to abandonment, relationship quality, and experiences of prejudice
increase jealousy in their primary partner (consistent with evidence and discrimination (both formal and informal). Moreover, psychol-
that anxiety is linked to a manipulative conflict style; Feeney, ogists’ efforts to enact changes at the macro level (community,
Noller, & Callan, 1994). Finally, negative experiences with non- state, nation) to support the dignity and legitimacy of same-sex
monogamy in previous romantic relationships could simultane- relationships may help to reduce the special salience of attachment
ously increase attachment anxiety and preference for sexual anxiety in same-sex relationship functioning.
exclusivity. Additional research is needed to investigate these
Findings should be interpreted with an appreciation for the
possibilities.
strengths and limitations of the study. Strengths included use of
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

This study was the first to test Mohr’s (2008) hypothesis that
data from a large community sample of same-sex couples who
attachment security may mitigate the negative effects of stigma on
This document is copyrighted by the American Psychological Association or one of its allied publishers.

were diverse in geographic location within the United States and

same-sex relationship quality. Results indicated a uniform lack of
Canada, age, and relationship length. This was likely the first study
support for this hypothesis. In no case were associations between
potential sources of minority stress and relationship outcomes of same-sex couples to examine both actor and partner effects of
moderated by attachment security, despite the fact that all three attachment. Moreover, the use of modern missing data methods
minority stressors included in the present study have been linked to made it possible to include data from participants whose partners
romantic relationship quality (Mohr & Daly, 2008; Mohr & did not submit surveys. Limitations included reliance on cross-
Fassinger, 2006; Otis, Rostosky, Riggle, & Hamrin, 2006). Results sectional and self-report data, although having data from both
may differ in relation to other manifestations of minority stress partners mitigates concerns about self-report data at least to some
(e.g., chronic self-concealment of sexual orientation), other ways extent. Several measures—notably the indices of prejudice, rela-
of assessing minority stress (e.g., implicit measures of internalized tionship satisfaction, and perceived partner similarity— have re-
stigma), or other relationship outcomes (e.g., self-efficacy for ceived relatively little psychometric investigation. As noted
negotiating minority stress within the relationship). However, above, the lack of longitudinal data makes it impossible to make
other research has failed to support the hypothesis that attachment inferences regarding the direction of influence among variables.
security moderates links between perceived discrimination and Also, the sample overrepresented White, highly educated peo-
distress among LGB people (Zakalik & Wei, 2006), which, taken ple. Results may well have differed if the sample had included
together with the present findings, suggests that attachment and a greater number of people of color or people from lower
minority stress may influence well-being independently rather than socioeconomic strata. For example, results related to minority
interactively. With this said, both studies assessed perceived prej- stressors may have been different, given evidence that attitudes
udice over long time frames (e.g., previous year). Attachment may regarding sexual minorities tend to be more negative in racial
be most salient to well-being in the midst of prejudice events, minority communities and in people who have not pursued
given the role of perceived threat in activating the attachment higher education.
system. The present study supports the view that attachment pro-
What are the implications of findings for practice? Most cesses in same-sex couples largely mirror those in heterosexual
broadly, results suggest that people in same-sex relationships may couples. However, it is important to note that this study repli-
benefit from clinical applications of attachment theory designed to cates findings from only the most basic attachment studies of
help individuals and couples have more satisfying romantic rela-
heterosexual couples. Feeney’s (2008) review of research on
tionships. For example, as suggested by Johnson (2008), emotion-
romantic attachment highlights the diversity of methods that
ally focused couple therapy—which draws heavily from attach-
have been used to gain a deeper understanding of attachment
ment theory—is likely to be as effective with same-sex couples as
processes among heterosexuals, including experimental and
it has found to be with heterosexual couples. Results also high-
observational methods. Replication and extension of such stud-
lighted the potential importance of considering attachment in re-
lation to sexual exclusivity when working with same-sex couples. ies with same-sex couples will help to advance understanding
Although research is needed to explain why nonmonogamy was regarding the generalizability of research based on heterosexual
linked with poorer relationship quality only in people with high couples. Such research is particularly merited given evidence
levels of attachment anxiety, this result indicates that the meaning from observational studies that same-sex couples may be more
or function of nonmonogamy depends on the extent to which one effective than their heterosexual counterparts in their ability to
fears abandonment. Counselors working with same-sex couples navigate conflict (Gottman et al., 2003) or work harmoniously
should not assume that sex outside the relationship is a sign of on joint tasks (Roisman, Clausell, Holland, Fortuna, & Elieff,
problems, but also should keep in mind that nonmonogamy may 2008). Also, the present study highlights the role that research
signal relationship challenges for partners with high levels of on same-sex couples may play in understanding the interplay
attachment anxiety. among attachment, gender, and sexuality. For these reasons, it
As discussed above, the particularly robust link between attach- is our hope that the present study is part of a growing tide of
ment anxiety and romantic relationship functioning observed in studies examining attachment processes in same-sex couples.
 ATTACHMENT IN SAME-SEX COUPLES 81

References Gaines, S. O., & Henderson, M. C. (2002). Impact of attachment style on

responses to accommodative dilemmas among same-sex couples. Per-
Ainsworth, M. D. (1985). Patterns of attachment. Clinical Psychologist, 38, sonal Relationships, 9, 89 –93. doi:10.1111/1475-6811.00006
27–29. Gallo, L. C., & Smith, T. W. (2001). Attachment style in marriage:
Allen, E. S., & Baucom, D. H. (2004). Adult attachment and patterns of Adjustment and responses to interaction. Journal of Social and Personal
extradyadic involvement. Family Process, 43, 467– 488. doi:10.1111/j Relationships, 18, 263–289. doi:10.1177/0265407501182006
.1545-5300.2004.00035.x Gangestad, S. W., & Thornhill, R. (1997). The evolutionary psychology of
Bowlby, J. (1982). Attachment and loss, Vol. 1: Attachment. New York, extrapair sex: The role of fluctuating asymmetry. Evolution and Human
NY: Basic Books. (Original work published 1969) Behavior, 18, 69 – 88. doi:10.1016/S1090-5138(97)00003-2
Brennan, K. A., Clark, C. L., & Shaver, P. R. (1998). Self-report measure- Gottman, J. M., Levenson, R. W., Swanson, C., Swanson, K., Tyson, R., &
ment of adult attachment: An integrative overview. In J. A. Simpson & Yoshimoto, D. (2003). Observing gay, lesbian and heterosexual couples’
W. S. Rholes (Eds.), Attachment theory and close relationships (pp. relationships: Mathematical modeling of conflict interaction. Journal of
46 –76). New York, NY: Guilford Press. Homosexuality, 45, 65–91. doi:10.1300/J082v45n01_04
Bretherton, I., & Munholland, K. A. (2008). Internal working models in Harris, C. R. (2002). Sexual and romantic jealousy in heterosexual and
attachment relationships: Elaborating a central construct in attachment homosexual adults. Psychological Science, 13, 7–12. doi:10.1111/1467-
theory. In J. Cassidy & P. R. Shaver (Eds.), Handbook of attachment:
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

9280.00402
Theory, research, and clinical applications (2nd ed., pp. 102–127). New Hazan, C., & Shaver, P. R. (1987). Romantic love conceptualized as an
This document is copyrighted by the American Psychological Association or one of its allied publishers.

York, NY: Guilford Press. attachment process. Journal of Personality and Social Psychology, 52,
Brown, L. S. (1995). Therapy with same-sex couples: An introduction. In 511–524. doi:10.1037/0022-3514.52.3.511
N. S. Jacobson & A. S. Gurman (Eds.), Clinical handbook of couple Hill, C. T., Rubin, Z., & Peplau, L. A. (1976). Breakups before marriage:
therapy (pp. 274 –291). New York, NY: Guilford Press. The end of 103 affairs. Journal of Social Issues, 32, 147–168. doi:
Caldwell, M. A., & Peplau, L. A. (1984). The balance of power in lesbian 10.1111/j.1540-4560.1976.tb02485.x
relationships. Sex Roles, 10, 587–599. doi:10.1007/BF00287267 Johnson, S. M. (2008). Emotionally focused couple therapy. In A. S.
Campbell, L., Simpson, J. A., Boldry, J., & Harris, R. (2010). Trust, Gurman (Ed.), Clinical handbook of couple therapy (4th ed., pp. 107–
variability in relationship evaluations, and relationship processes. Jour- 137). New York, NY: Guilford Press.
nal of Personality and Social Psychology, 99, 14 –31. Kenny, D. A., & Cook, W. (1999). Partner effects in relationship research:
Chang, L.-W., Schumm, W. R., Coulson, L., Bollman, S. R., & Jurich, Conceptual issues, analytic difficulties, and illustrations. Personal Re-
A. P. (1994). Dimensionality of Brief Family Interaction and Satisfac- lationships, 6, 433– 448. doi:10.1111/j.1475-6811.1999.tb00202.x
tion scales among couples from eight western and midwestern states. Kurdek, L. A. (1997). Relation between neuroticism and dimensions of
Psychological Reports, 74, 131–144. doi:10.2466/pr0.1994.74.1.131 relationship commitment: Evidence from gay, lesbian, and heterosexual
Collins, N. L. (1996). Working models of attachment: Implications for couples. Journal of Family Psychology, 11, 109 –124. doi:10.1037/0893-
explanation, emotion, and behavior. Journal of Personality and Social 3200.11.1.109
Psychology, 71, 810 – 832. doi:10.1037/0022-3514.71.4.810 Kurdek, L. A. (2002). On being insecure about the assessment of attach-
Collins, N. L., & Read, S. J. (1990). Adult attachment, working models, ment styles. Journal of Social and Personal Relationships, 19, 811– 834.
and relationship quality in dating couples. Journal of Personality and doi:10.1177/0265407502196005
Social Psychology, 58, 644 – 663. doi:10.1037/0022-3514.58.4.644 Lund, M. (1985). The development of investment and commitment scales
Del Giudice, M. (2011). Sex differences in romantic attachment: A meta- for predicting continuity of personal relationships. Journal of Social and
analysis. Personality and Social Psychology Bulletin, 37, 193–214. Personal Relationships, 2, 3–23. doi:10.1177/0265407585021001
doi:10.1177/0146167210392789 Meyer, I. H. (1995). Minority stress and mental health in gay men. Journal
Elizur, Y., & Mintzer, A. (2003). Gay males’ intimate relationship quality: of Health and Social Behavior, 36, 38 –56.
The roles of attachment security, gay identity, social support, and in- Mikulincer, M., & Shaver, P. R. (2005). Attachment theory and emotions
come. Personal Relationships, 10, 411– 435. doi:10.1111/1475-6811 in close relationships: Exploring the attachment-related dynamics of
.00057 emotional reactions to relational events. Personal Relationships, 12,
Enders, C. K. (2010). Applied missing data analysis. New York, NY: 149 –168. doi:10.1111/j.1350-4126.2005.00108.x
Guilford Press. Mohr, J. (2008). Same-sex romantic attachment. In J. Cassidy & P. R.
Fassinger, R. E., & Arseneau, J. R. (2007). “I’d rather get wet than be Shaver (Eds.), Handbook of attachment theory and research (2nd ed.,
under that umbrella”: Differentiating the experiences and identities of pp. 482–502). New York, NY: Guilford Press.
lesbian, gay, bisexual, and transgender people. In K. J. Bieschke, R. M. Mohr, J. J., Crook, R. E., & Kolchakian, M. R. (2010). Love imagined:
Perez, & K. A. DeBord (Eds.), Handbook of counseling and psycho- Working models of future romantic attachment in emerging adults.
therapy with lesbian, gay, bisexual, and transgender clients (2nd ed., pp. Personal Relationships, 17, 457– 473.
19 – 49). Washington, DC: American Psychological Association. doi: Mohr, J. J., & Daly, C. A. (2008). Sexual minority stress and changes in
10.1037/11482-001 relationship quality in same-sex couples. Journal of Social and Personal
Feeney, J. A. (2008). Adult romantic attachment: Developments in the Relationships, 25, 989 –1007. doi:10.1177/0265407508100311
study of couple relationships. In J. Cassidy & P. R. Shaver (Eds.), Mohr, J., & Fassinger, R. (2000). Measuring dimensions of lesbian and gay
Handbook of attachment theory and research (2nd ed., pp. 456 – 481). male experience. Measurement and Evaluation in Counseling and De-
New York, NY: Guilford Press. velopment, 33, 66 –90.
Feeney, J. A., Noller, P., & Callan, V. J. (1994). Attachment style, Mohr, J. J., & Fassinger, R. E. (2006). Sexual orientation identity and
communication and satisfaction in the early years of marriage. In K. romantic relationship quality in same-sex couples. Personality and So-
Bartholomew & D. Perlman (Eds.), Attachment processes in adulthood cial Psychology Bulletin, 32, 1085–1099. doi:10.1177/014616720628
(pp. 269 –308). London, England: Jessica Kingsley Publishers. 8281
Feeney, J. A., Peterson, C., Gallois, C., & Terry, D. J. (2000). Attachment Muthén, L. K., & Muthén, B. O. (1998 –2011). Mplus user’s guide (6th
style as a predictor of sexual attitudes and behavior in late adolescence. ed.). Los Angeles, CA: Author.
Psychology & Health, 14, 1105–1122. doi:10.1080/088704400084 O’Neil, J. M. (2008). Summarizing twenty-five years of research on men’s
07370 gender role conflict using the Gender Role Conflict Scale: New research
 82 MOHR, SELTERMAN, AND FASSINGER

paradigms and clinical implications. Counseling Psychologist, 36, 358 – ences. Australian and New Zealand Journal of Psychiatry, 32, 848 – 859.
445. doi:10.1177/0011000008317057 doi:10.3109/00048679809073875
Otis, M. D., Rostosky, S. S., Riggle, E. B., & Hamrin, R. (2006). Stress and Roisman, G. I., Clausell, E., Holland, A., Fortuna, K., & Elieff, C. (2008).
relationship quality in same-sex couples. Journal of Social and Personal Adult romantic relationships as contexts of human development: A
Relationships, 23, 81–99. doi:10.1177/0265407506060179 multimethod comparison of same-sex couples with opposite-sex dating,
Peplau, L. A., & Cochran, S. D. (1983). The intimate relationships of engaged, and married dyads. Developmental Psychology, 44, 91–101.
lesbians and gay men. In E. R. Allgeier & N. B. McCormick (Eds.), doi:10.1037/0012-1649.44.1.91
Changing boundaries (pp. 226 –244). Palo Alto, CA: Mayfield. Schlomer, G. L., Bauman, S., & Card, N. A. (2010). Best practices for
Peplau, L., & Fingerhut, A. W. (2007). The close relationships of lesbian missing data management in counseling psychology. Journal of Coun-
and gay men. Annual Review of Psychology, 58, 405– 424. doi:10.1146/ seling Psychology, 57, 1–10. doi:10.1037/a0018082
annurev.psych.58.110405.085701 Schmitt, D. P., Alcalay, L., Allensworth, M., Allik, J., Ault, L., Austers, I.,
Peplau, L., Fingerhut, A., & Beals, K. P. (2004). Sexuality in the relation- . . . Wan, W. W. N. (2003). Are men universally more dismissing than
ships of lesbians and gay men. In J. H. Harvey, A. Wenzel, & S. women? Gender differences in romantic attachment across 62 cultural
Sprecher (Eds.), The handbook of sexuality in close relationships (pp. regions. Personal Relationships, 10, 307–331. doi:10.1111/1475-6811
349 –369). Mahwah, NJ: Erlbaum. .00052
Shaver, P. R., & Mikulincer, M. (2007). Adult attachment strategies and
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Peplau, L., & Spalding, L. R. (2003). The close relationships of lesbians,

gay men, and bisexuals. In L. D. Garnets & D. C. Kimmel (Eds.), the regulation of emotion. In J. J. Gross (Ed.), Handbook of emotion
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Psychological perspectives on lesbian, gay, and bisexual experiences regulation (pp. 446 – 465). New York, NY: Guilford Press.
(2nd ed., pp. 449 – 474). New York, NY: Columbia University Press. Simpson, J. A., Rholes, W., & Phillips, D. (1996). Conflict in close
doi:10.4135/9781452220437.n9 relationships: An attachment perspective. Journal of Personality and
Pietromonaco, P. R., Greenwood, D., & Barrett, L. (2004). Conflict in adult Social Psychology, 71, 899 –914. doi:10.1037/0022-3514.71.5.899
close relationships: An attachment perspective. In W. Rholes & J. A. Zakalik, R. A., & Wei, M. (2006). Adult attachment, perceived discrimi-
Simpson (Eds.), Adult attachment: Theory, research, and clinical impli- nation based on sexual orientation, and depression in gay males: Exam-
cations (pp. 267–299). New York, NY: Guilford Press. ining the mediation and moderation effects. Journal of Counseling
Rempel, J. K., Holmes, J. G., & Zanna, M. P. (1985). Trust in close Psychology, 53, 302–313. doi:10.1037/0022-0167.53.3.302
relationships. Journal of Personality and Social Psychology, 49, 95–112.
doi:10.1037/0022-3514.49.1.95 Received August 28, 2012
Ridge, S. R., & Feeney, J. A. (1998). Relationship history and relationship Revision received October 15, 2012
attitudes in gay males and lesbians: Attachment style and gender differ- Accepted October 16, 2012 䡲

View publication stats